Principles of Behavior Modification Albert Bandura

Principles of
Behavior
Modification
Albert Bandura
to Ginny, Mary, and Carol
Copyright © 1969
e-Book 2019 International Psychotherapy

Institute
All Rights Reserved
This e-book contains material protected under

International and Federal Copyright Laws and
Treaties. This e-book is intended for personal
use only. Any unauthorized reprint or use of this
material is prohibited. No part of this book may
be used in any commercial manner without
express permission of the author. Scholarly use
of quotations must have proper attribution to
the published work. This work may not be
deconstructed, reverse engineered or
reproduced in any other format.
Created in the United States of America
www.freepsychotherapybooks.org 4
Table of Contents
Preface
1. Causal Processes
Interpretation of Causal Processes
Social Learning as a Reciprocal Influence Process
Symptom Substitution
Efficacy of Conventional Methods of Behavioral

Change
Summary
2. Value Issues and Objectives

Behavioral Specification of Objectives
Factors Impeding Specification of Objectives
Decision Processes in the Selection of Objectives
Summary
3. Modeling and Vicarious Processes

Theoretical Conceptions of Observational
Learning
Establishment of New Response Patterns

through Modeling
Vicarious Conditioning of Emotional
Responsiveness
Vicarious Extinction
Inhibitory and Disinhibitory Effects of Vicarious

Experiences
Response Facilitating Effects of Modeling

Influences
Utilization of Modeling Principles in Planned

Sociocultural Change
Summary
4. Positive Control
Theoretical Interpretations of Reinforcement
Processes
Essential Components of Reinforcement

Practices
Ethical Implications of Reinforcement Practices
Applications of Contingency Systems
Social Organizational Applications of

Reinforcement Contingencies
Summary
5. Aversive Control
Presentation of Negative Reinforcers
Applications of Aversive Contingency Systems
Removal of Positive Reinforcers
Summary
6. Extinction
Interpretations of the Extinction Process
Extinction of Positively Reinforced Behavior
Extinction of Defensive Behavior
Summary
7. Desensitization through Counterconditioning

Controlling Variables in Desensitization
Identification of the Stimulus Determinants of

Emotional Behavior
Neutralization of Threats in Symbolic or Realistic

Forms
Antagonistic Activities in Counterconditioning
Physiological Accompaniments of Emotional

Behavior
Summary
8. Aversive Counterconditioning
Development of Conditioned Aversion and
Avoidance
Sexual Deviance
Modification of Symbolic Activities
Alcoholism
Ethical Considerations in Aversion Therapy
Summary
9. Symbolic Control of Behavioral Changes

Role of Awareness of Contingencies in
Behavioral Change
Verbal Conditioning as a Function of Awareness
Interactive Effects of Cognitive and Incentive

Variables
Symbolic Control of Classical Conditioning

Phenomena
Implications of Symbolic Control for Behavioral

Modification
Discrepancy between Response Systems and the

Unconscious
Attitudinal Consequences of Behavioral and

Affective Changes
Strategies of Attitude Change
“Internalization” and Persistence of Behavioral

Changes
Stabilization of Behavioral Changes through
Development of Self-Regulatory Functions
Summary
Preface
This book presents basic psychological
principles governing human behavior within the

conceptual framework of social learning. Over the
years an impressive body of knowledge about the
mechanisms through which behavior is acquired
and modified has been accumulated. But despite
this vigorous growth of research on human

behavior, a number of psychological processes
that are highly influential in human functioning
have been overlooked or only partially
investigated. This volume reviews the recent
theoretical and experimental advances in the field

of social learning. It gives special emphasis to the
important roles played by vicarious, symbolic, and

self-regulatory processes, which receive relatively
little notice even in most contemporary theories of
behavior.
The worth of a psychological theory must be
judged not only by how well it explains laboratory

findings but also by the efficacy of the behavioral
modification procedures that it produces. Recent
years have witnessed widespread applications of

methods derived from principles of social learning
to the modification of important social phenomena
in familial, educational, clinical, and other social
settings. By requiring clear specification of

treatment conditions and objective assessment of
outcomes, the social-learning approach presented
in this book contains a self-corrective feature that
distinguishes it from change enterprises in which
interventions remain ill-defined and their

psychological effects are seldom objectively
evaluated.
New social change procedures are by tradition
enthusiastically promoted, and it is not until after
the methods have been applied for some time by a
coterie of enthusiasts that systematic tests of

efficacy are conducted. Usually the methods are
then unceremoniously retired by subsequent
controlled studies. Professional workers in this
field have, therefore, come to view any new

approach as a passing fad. However, when
laboratory tests of efficacy precede social
applications, new methods are subjected to close
scrutiny at each stage of development, and those
that evolve are likely to produce outcomes

sufficiently favorable to weather rigorous
evaluation. The successful results obtained by

social-learning procedures in carefully controlled
studies justify optimistic expectations for future

developments of this approach. The numerous
investigations reported in this book also illustrate

how understanding of major change processes can
be advanced by inventive research on socially
significant problems. Contrary to much of the
current criticism, basic research need not settle

for inconsequential dependent measures.
This book is concerned not only with the

validity of the principles set forth but also with the
conditions under which they can serve as an
instrument for human advancement. The value

issues that arise in the applications of social-
learning procedures to achieve various
psychological changes are, therefore, closely
examined, and special attention is given to the
effects of social practices on man’s self-evaluation

and self-enhancement.
While this book was being written the author
contributed chapters on modeling processes to
Volume II of Advances in Experimental Social
Psychology (Bandura, 1965) and to the Ciba
Foundation Symposium: The Role of Learning and
Psychotherapy (Bandura, 1968). Chapter 3
contains a revised and updated version of some of

the material that originally appeared in the latter
publications.
Many people contributed in one way or

another to this venture. To Ted Rosenthal and
Rogers Elliott, who read preliminary versions of
the manuscript and made many valuable

suggestions, I offer my sincere thanks. I am also
indebted to countless students and colleagues who
have helped through collaborative research and
sharing of ideas to enhance the value of what I
have written. I owe a special personal debt to my

former student and colleague, Richard Walters,
who died tragically at the height of his productive

career. Although he never read what I have
written here, our lively discussions during

collaborative projects did much to clarify some of
the theoretical issues discussed in this book.
The preparation of this volume involved
considerable work, and I wish to express my

gratitude to the people who helped lighten my
labors. I am especially grateful to Jane Crane for
deciphering illegible draft versions and for many
hours of painstaking effort in preparing the
manuscript for publication. Thanks are due to

Robert O’Connor for his assistance with drafting
and photographic matters. I should also like to pay
tribute to Darlene Lapham for her remarkably
efficient typing of the manuscript.
Finally, the dedication of this volume signifies

my profound indebtedness to my family, who
sacrificed many weekend activities and vacations
while I was absorbed in the task of writing this
book.
Albert Bandura
Stanford, California April 1969
1
Causal Processes
The development of principles and procedures
of behavioral change is largely determined by the
model of causality to which one subscribes. The
methods used to modify psychological phenomena
therefore cannot be fully understood
independently of the personality theory upon
which they are based. The major differences

between rival theoretical orientations are most
strikingly revealed in their interpretations of
grossly deviant behavior. Consequently the

systems that have been advanced to explain these
perplexing conditions will be considered in some
detail here, although this book is only partially

concerned with issues relating to deviant
behavior.
The earliest conceptions of psychopathology
viewed behavioral anomalies as external

manifestations of evil spirits that entered the
victim’s body and adversely affected his behavior.
Treatment accordingly was directed toward

exorcising demons by various methods, such as
cutting a hole in the victim’s skull, performing
various magical and religious rituals, or brutally
assaulting-—physically and socially—the bearer

of the pernicious spirits. Hippocrates was
influential in supplanting the demonological
conceptions of deviant behavior by relabeling it
disease rather than demonic manifestations.
Wholesome diets, hydrotherapy, bloodletting, and

other forms of physical intervention, some benign,
others less humane, were increasingly employed
as corrective treatments.
Although psychological methods gradually
replaced physical procedures in modifying deviant
response patterns, the analogy of physical health

and disease nevertheless continued to dominate
theories of psychopathology. In this

conceptualization, behavioral patterns that depart
widely from accepted social and ethical norms are
considered to be derivatives or symptoms of an
underlying disease. Modification of social deviance

thus became a medical specialty, with the result
that persons exhibiting atypical behavior are
labeled “patients” suffering from a “mental

illness,” and they generally are treated in
medically oriented facilities. The disease concepts

are likewise indiscriminately applied even to
social phenomena, as evidenced by the frequent
designation of cultural response patterns as

“healthy” or “sick.” Had Hippocrates represented
behavioral anomalies as products of idiosyncratic
social-learning experiences rather than as
expressions of a somatic illness, the

conceptualization and treatment of divergent
response patterns might have taken an entirely
different course.
A quasi-disease model is still widely employed
in explanations of grossly deviant behavior, but
the underlying pathology is generally considered

to be psychic rather than neurophysiological in
nature. This conceptual scheme became further
confused when the appropriateness of the disease
analogy to social behavior was increasingly
challenged (Szasz, 1961). Most personality

theorists eventually discarded the notion that
deviant behavior is a manifestation of an
underlying mental disease, but they nevertheless
unhesitatingly label anomalous behaviors as
symptoms and caution against the dangers of
symptom substitution. In these theories, the
conditions supposedly controlling behavior
continue to function analogously to toxic
substances in producing deviant responses;

however, the disturbing agents comprise a host of
inimical psychodynamic forces (for example,
repressed impulses, energized traits, psychic
complexes, latent tendencies, self-dynamisms, and
other types of energy systems) somewhat akin to

the pernicious spirits of ancient times. Many
contemporary theories of psychopathology thus
employ a quasi-medical model fashioned from an
amalgam of the disease and demonology
conceptions, which have in common the belief that

deviant behavior is a function of inimical inner
forces. Consequently, attention is generally

focused, not on the problem behavior itself, but on
the presumably influential internal agents that
must be exorcised by “catharsis,” “abreaction,” and

acquisition of insight through an extended
interpretive process. Indeed, direct modification of
so-called symptomatic behavior is considered not
only ineffective but actually dangerous, because, it

is held, removal of the symptom has no effect upon
the underlying disorder, which will manifest itself

again in a new, possibly more debilitating
symptom.
SOCIAL LABELING OF DEVIANT BEHAVIOR
Although most psychotherapists agree that
direct “symptom” removal is inadvisable and few
of them would acknowledge engaging in such
forms of treatment, remarkably little attention has

been devoted to the definition of what constitutes
a “symptom.” Categorizing a pattern of behavior as
symptomatic of an underlying disorder actually
involves a complex set of criteria, most of which
are quite arbitrary and subjective. Whether

specific actions are called normal or symptomatic
expressions will depend upon whether certain
social judges or the person himself disapproves of
the behavior being exhibited. Since symptom

labeling primarily reflects the evaluative
responses that a given behavior evokes from
others, rather than distinguishable qualities of the
behavior itself, an identical response pattern may

be viewed as a pathological derivative or as
wholesome behavior by persons whose
judgmental orientations differ. Aggressiveness in
children, for example, may be positively reinforced
and regarded as a sign of masculinity and healthy

social development by some parents, while the
same behavior is generally viewed by educational,

legal, and other societal agents as a symptom of a
personality disorder (Bandura, 1960; Bandura &

Walters, 1959).
The designation of behavior as pathological

thus involves social judgments that are influenced
by, among other factors, the normative standards
of persons making the judgments, the social
context in which the behavior is exhibited, certain

attributes of the behavior, and numerous
characteristics of the deviator himself. An
adequate theory of deviant behavior must
therefore be concerned with the factors
determining evaluative judgments. Unfortunately,

in spite of widespread use of diagnostic
classifications and the potentially serious
consequences of labeling persons as mentally
disturbed, there has been surprisingly little
systematic study of the factors governing such

judgmental behavior.
Psychopathology is characteristically inferred

from the degree of deviance from the social norms
that define how persons are expected to behave at
different times and places. Consequently, the
appropriateness of symbolic, affective, or social
responses to given situations constitutes one
major criterion in labeling “symptomatic”
behavior. Departures from normative standards

that do not inconvenience or interfere with the
wellbeing of others are usually tolerated;
deviations that produce rewarding consequences
for the members of a society, as in the case of
technological inventions and intellectual and

artistic innovations, may be actively promoted and
generously rewarded. On the other hand, deviance
that generates aversive consequences for others
elicits strong societal disapproval, is promptly
labeled abnormal, and generally is met by coercive

pressures to eliminate it.
The appropriateness criterion poses serious

problems in societies, such as our own, that are
differentiated into many subcultures whose
members subscribe to divergent behavioral norms
and therefore do not agree on what is suitable
social behavior. Members of social groups who
want rewards that are highly valued in the culture
but lack the means of obtaining them in legitimate

ways (Cloward & Ohlin, 1960; Merton, 1957) are
often forced to resort to socially unacceptable
activities. In these instances, antisocial patterns
are not only normatively sanctioned, but the social
environment provides these persons ample

opportunities, through appropriate reinforcement
contingencies and role models, to develop and to
perfect deviant modes of behavior. According to
the prevailing normative structure of these
subcultures, skillfully executed antisocial behavior

represents emulative rather than sick behavior
and is governed by the same types of variables

that control the prosocial response patterns
displayed by members of the larger society.
Other subgroups are classified as social
deviants, and therefore “sick” or “crazy,” not
because they adhere to culturally disapproved
means of gaining highly rated objectives but
because they withdraw from the dominant social

system and reject the basic cultural goals
themselves. The conforming majority within a
society may label nonconformist groups, such as
“Bohemians,” “beatniks,” and “hippies,” that refuse
to strive for the goals highly valued in the culture

as exhibiting maladaptive behavior. From the
perspective of the deviants, the life style of
conforming members is a symptomatic
manifestation of an overcommercialized, “sick”
society. Thus the same pattern of behavior may be

deemed a symptom by one social group but judged
healthy and positively reinforced by persons who

adhere to a different code of behavior. Similarly,
when a society radically alters its social and legal
norms, either the presence or absence of the same

responses may be judged inappropriate, and,
consequently, labeled symptoms of an underlying
pathology. Thus, a citizen socialized in other
respects who commits a brutal homicide will be

diagnosed as suffering from a serious mental
disorder, but a military recruit’s inability to

behave homicidally on the battlefield will likewise
be viewed as symptomatic of a “war neurosis.” The
latter example further illustrates how behavior
can come to be thought of as symptomatic because

of changes in societal norms rather than because
of a psychopathology reflected in the behavior
itself.
The discussion thus far has been concerned

with the deviant behavior of members of groups,
who mutually support and reinforce each other’s
ideologies and actions. Some individuals display
gross behavioral eccentricities that appear totally
inexplicable; persons from different subgroups
who do not share the same normative systems are
apt to view these eccentricities as pathological
manifestations. Even in these instances, when the
idiosyncratic social-learning history for the

behavior is known there is no need to assume an
underlying disease process. Lidz, Cornelison,
Terry, & Fleck (1958) report a case, for example,
in which sibling schizophrenics believed, among
other strange things, that “disagreement” meant

constipation. This clearly inappropriate
conceptual behavior was the result of exposure to
bizarre social-learning contingencies and not an
expression of a mental illness. Whenever the sons
disagreed with their mother, she informed them

that they were constipated and required an
enema. The boys were then disrobed and given

anal enemas, a procedure that dramatically
conditioned an unusual meaning to the word
“disagreement.” The cases cited by Lidz and his

associates (Lidz, Fleck, & Cornelison, 1965)
provide compelling evidence of development of
delusions, suspiciousness, grandiosity, extreme
denial of reality, and other forms of

“schizophrenic” behavior through direct
reinforcement, and of their social transmission by

parental modeling of incredibly deviant behavior
patterns.
In addition to the influence of normative

commitments in determining judgmental
responses, certain properties of behavior readily
invite one to label an emotional disorder
symptomatic. Responses of high magnitude, for
instance, often produce unpleasant experiences

for others; they are therefore more likely to be
considered pathological manifestations than are
responses of low or moderate intensities. A
youngster who is continually wrestling other
children will generally be viewed as exhibiting
youthful exuberance; in contrast, a child whose
physically aggressive behavior is more forceful
and hurtful will in all likelihood be regarded as
emotionally disturbed. Although pervasive and

intense emotional responses may be reliably
categorized, disagreements are apt to arise in the
labeling of behavior that falls at less extreme
points on the response-intensity continuum. The
line separating normality and abnormality may be

variously located depending upon the tolerance
limits for aversiveness of different judges. Even if a
high degree of consensus could be achieved in
designating the acceptable limits of amplitude for
various behaviors, no evidence exists that

emotional responses of high intensity are
mediated by psychopathological internal

processes, whereas similar responses of lesser
strength are governed by nonpathological internal
processes.
Behavioral deficits are also frequently
interpreted as symptoms of emotional disorder,
particularly when the deficits produce hardships
and aversiveness for others. Adequately endowed

children, for example, who are incontinent and
who exhibit marked deficiencies in interpersonal,
verbal, and academic skills, and adults who are
unable to meet social, marital, and vocational task
requirements tend to be labeled as emotionally

disturbed. It is generally assumed, moreover, that
the greater the deficits, the more extensive the
underlying psychopathology. The arbitrary and
relativistic nature of the deficit or competence
criterion would become readily apparent if one

were to vary the minimum standards of
competence required in any given situation. If the

standards were set at a comparatively low level,
practically all members of a society would be
judged competent and healthy, whereas the vast

majority would suddenly acquire a
psychopathology if exceedingly high standards
were adopted. In the latter case, therapists and
diagnosticians might devote much time to locating

the source of pathology within the individuals.
The intention attributed to an action will affect

its categorization by others as a symptomatic
expression. When the variables governing physical
and biological phenomena remained unknown, a

host of internal forces and deities were invoked as
causal agents. As scientific knowledge increased,
these fanciful driving forces were replaced by
explanatory concepts involving manipulable
variables. Similarly, interpretations of

psychological phenomena often assume
pathological inner agents in cases where deviance
appears unintelligible. If a person engages in
disapproved behavior to attain generally valued
material objects, his activities—being readily
understandable—are less likely to be regarded as
manifestations of emotional disease than if his
deviant behavior has no apparent utilitarian value.
Delinquents who strike victims on the head to

extract their wallets expediently are generally
labeled semiprofessional thieves exhibiting
income-producing instrumental aggression. By
contrast, delinquents who simply beat up
strangers but show no interest in their victims’

material possessions are supposedly displaying
emotional aggression of a peculiarly disturbed
sort. It is evident that in many cases of so-called
nonutilitarian aggression, the behavior is highly
instrumental in gaining the approval and

admiration of peers and in enhancing status in the
social hierarchy of the reference group. Peer-

group approval is often more powerful than
tangible rewards as an incentive for, and
reinforcer of, aggressively deviant behavior

(Buehler, Patterson, and Furniss, 1966).
The influential role of social reinforcement in
regulating dangerous, senseless behavior is clearly
revealed in a field study by Yablonsky (1962), who

found that the dominant reinforcement
contingencies in many delinquent gangs have

shifted from utilitarian antisocial activities to
destructive assaults executed in a “cool” and
apparently indifferent manner on persons and
property. The way in which aggression has taken

on status-conferring value and in which threat of
loss of “rep” may compel a person to engage in a
homicidal assault is graphically illustrated in the

following excerpt from an interview with one of
the boys studied by Yablonsky.
“Momentarily I started to thinking about it

inside; I have my mind made up I’m not
going to be in no gang. Then I go on inside.
Something comes up, then here comes all my
friends coming to me. Like I said before, I’m
intelligent and so forth. They be coming to
me—then they talk to me about what they
gonna do. Like, ‘Man, we’ll go out there and
kill this cat.’ I say, ‘Yeah.’ They kept on
talkin’. I said, ‘Man, I just gotta go with you.’
Myself, I don’t want to go, but when they
start talkin’ about what they gonna do, I say,
‘So, he isn’t gonna take over my rep. I ain’t
gonna let him be known more than me.’ And
I go ahead [p. vii].”
External contingencies of reinforcement rather
than internal emotional disease also appear to be

the major determinants of the behavior of another
youth involved in a gang killing: “If I would of got

the knife, I would have stabbed him. That would
have gave me more of a build-up. People would
have respected me for what I’ve done and things

like that. They would say, ‘There goes a cold killer’
[p. 8].” Similar reinforcement contingencies

operated in the practice of a gang apprehended
that used attacks upon people without

provocation as its main admissions requirement.
Each physical assault, which had to be observed by
a club member to be valid, was valued at 10
points; and a total of 100 points was required for
full-fledged membership (San Francisco Chronicle,

1964).
It should be noted in passing that prosocial
approval-seeking behavior like athletic
achievements or musical accomplishments, which
may likewise have no apparent utilitarian value, is

seldom labeled as emotionally disturbed behavior.
Certain subgroups simply value and reward
skillful “stomping” more highly than musical
virtuosity.
The instrumental versus emotional dichotomy,

therefore, appears primarily to reflect differences
in the types of rewards sought, and not basic
differences in the purposiveness of the behavior
itself, or in the nature of the mediating internal
events. Since some members of a society are likely
to be brought up under atypical contingencies of
social reinforcement, events which are ordinarily
neutral or aversive for others may acquire a

strong positive valence; consequently, the puzzling
behavior exhibited by these individuals may
appear to have little or no instrumental value, and
thus tend to be explained by reference to internal

psychopathological processes.
Certain behavioral requirements are

prescribed according to a person’s age, sex, social
position, occupation, race, ethnic origin, or
religion. Therefore personal attributes also enter
into social judgment of behavior that deviates
from role demands. For example, behavior

considered to be normal at an early age may be
regarded as a symptom of personality disturbance

later, as in the case of enuresis. It is very
appropriate, in this connection, to repeat

Mowrer’s (1950) query: “And when does
persisting behavior of this kind suddenly cease to
be normal and become a symptom [p. 474]?” Or
consider the attribute of sex. The differential

cultural tolerance for cross-sex behavior displayed
by males and females illustrates the role of sex
characteristics in the assignment of symptomatic
status to deviant behavioral patterns. The wearing
of female apparel by males is considered to be

indicative of a serious psychological disorder,
requiring prompt legal and psychiatric attention.
On the other hand, females may adopt masculine
garb, hair styles, and a wide range of
characteristically masculine response patterns

without being labeled as mentally disturbed. Since
masculine role behavior occupies a position of

relatively high prestige and power in our society
and often is more generously rewarded than
feminine role behavior, the emulation of

masculine tendencies by females is more
understandable and, therefore, less likely to be
interpreted by reference to disease processes.
There is another side to the influence of
personal attributes on judgmental responses. The
social-learning background and characteristics of

the person making the judgments may
significantly affect his designation of particular
behaviors as indicative of mental health or psychic

pathology. Spohn (1960) found that therapists’
social values were related to their mental health
judgments of patients’ behavior that reflected
similar value dimensions: that is, therapists
thought the patients more like themselves were

the healthier ones.
Although the presence of psychic illness is
frequently judged in terms of deviance from
particular social norms, in many cases it is
primarily based on self-definition. As Terwilliger &
Fiedler (1958) have shown, persons often label
themselves as emotionally disturbed, whereas
others may judge them to be functioning

satisfactorily within the prevailing social norms.
Evaluative discrepancies of this type typically
arise when persons impose excessive demands
upon themselves and suffer subjective distress as

a result of failure to meet self-imposed standards.
A comprehensive theory of deviance must take
into consideration self-reactions as well as societal
reactions to one’s behavior.
It is apparent from the foregoing discussion
that the categorization of behavior as symptomatic
of an underlying pathology depends upon a host of

subjective criteria, and as a consequence, the same
behavior may be characterized as “healthy” or

“sick” by different judges, in different social
contexts, and on the basis of performers’ social

characteristics. It is true, of course, that questions
of value and social judgment arise also in the
diagnosis of physical disorders. In such cases the
symptom-disease model is quite appropriate since

internal organic pathologies do in fact exist and
can be verified independently of their peripheral
manifestations. Brain tumors and dysfunctions
involving respiratory, circulatory, or digestive
organs are observable events. Where deviant

behavior is concerned, analogy with the symptom-
disease model is misleading because there are no
infected organs or psychic disease entities that can
be identified as causal agents. The psychic
conditions that are assumed to underlie

behavioral malfunctioning are merely abstractions
from the behavior. In the disease analogy these

abstractions are not only given substance and
existence independent of the behavior from which
they were inferred, but they are then invoked as

the causes of the same behavioral referents. For
these reasons, so-called symptomatic behavior can
be more adequately explained in terms of social
learning and value theory than through

inappropriate medical analogizing. An extended
account of a social-learning taxonomy of

behavioral phenomena generally subsumed under
the term “psychopathology” is presented
elsewhere (Bandura, 1968). The preceding
discussion reviewed some of the principal factors

determining the attribution of sickness to deviant
behavior. Similar social judgment processes are, of
course, involved in the attachment of descriptive

labels such as aggression, altruism, dependency, or
achievement to particular response patterns.
HYPOTHETICAL INTERNAL DETERMINANTS OF

BEHAVIOR
The questions raised concerning the utility and

validity of the concept of “symptom” apply equally
to the psychopathology presumed to underlie the
troublesome behavior. From the focusing of
attention on inner agents and forces, many fanciful
theories of deviant behavior have emerged. The

developmental history of social behavior is rarely
known, and its reconstruction from interview
material elicited by therapists or diagnosticians is
of doubtful validity. In fact, the content of

reconstruction is highly influenced by the
interviewer’s suggestive probing and selective
reinforcement of content that is in accord with his
theoretical orientation. Heine (1953), for example,
found that clients who were treated by client-

centered, Adlerian, and psychoanalytic therapists
tended to account for changes in their behavior in

terms of the explanations favored by their
respective interviewers. Even a casual survey of

interview protocols would reveal that
psychotherapists of different theoretical

affiliations tend to find evidence for their own
preferred psychodynamic agents rather than those
cited by other schools. Thus, Freudians are likely
to unearth Oedipus complexes and castration

anxieties, Adlerians discover inferiority feelings
and compensatory power strivings, Rogerians find

compelling evidence for inappropriate self-
concepts, and existentialists are likely to diagnose
existential crises and anxieties. It is equally true
that Skinnerians, predictably, will discern

defective conditions of reinforcement as
important determinants of deviant behavior. In
the latter explanatory scheme, however, the

suspected controlling conditions are amenable to
systematic variation; consequently the functional

relationships between reinforcement
contingencies and behavior are readily verifiable.
Theoretical models of dubious validity persist
largely because they are not stated in refutable
form. The lack of accurate knowledge of the
genesis of behavioral deviations further precludes
any serious evaluation of suggested determinants
that are so involved that they could never be

produced under laboratory conditions. When the
actual social-learning history of maladaptive
behavior is known, principles of learning appear
to provide a completely adequate interpretation of
psychopathological phenomena, and

psychodynamic explanations in terms of
symptom-underlying disorder become
superfluous. The spuriousness of the supposition
that psychodynamic forces produce symptomatic
behavior can be best illustrated by cases in which

the antecedents of aberrant response patterns are
known. Such examples are hard to obtain since

they require the production of deviant behavior
under controlled conditions. Ayllon, Haughton, &
Hughes (1965) furnish a graphic illustration of

how a bizarre pattern of behavior—which was
developed, maintained, and subsequently
eliminated in a schizophrenic woman simply by
altering its reinforcing consequences—was

interpreted erroneously as a symptomatic
manifestation of complex psychodynamic events

by diagnosticians who were unaware of the
specific conditions of reinforcement regulating the
patient’s behavior.
Unfortunately, the exact antecedents of deviant

behavior are rarely known, and in the absence of
powerful techniques that permit adequate control
over behavioral phenomena, clinical endeavors
have until recently lacked the self-corrective

features necessary for eliminating weak or invalid
theories of psychopathology. As a consequence,
rival interpretations of social behavior have for
decades retained a secure status with little risk
that any one type of theory might prove more
cogent than another.
In recent years there has been a fundamental
departure from conventional views regarding the
nature, causes, and treatment of behavioral

dysfunctions. According to this orientation,
behavior that is harmful to the individual or

departs widely from accepted social and ethical
norms is viewed not as symptomatic of some kind
of disease but as a way that the individual has
learned to cope with environmental and self-

imposed demands. Treatment then becomes
mainly a problem in social learning rather than
one in the medical domain. In this conceptual

scheme the remaining vestiges of the disease-
demonic model have been discarded. Response

patterns are not viewed as symptoms and their
occurrence is not attributed to internal, pernicious
forces.
Social learning and psychodynamic theories
differ not only in whether they view deviant
behavior as a quasi disease or as a by-product of
learning, but also in what they regard to be the
significant controlling factors, and in the status

assigned to internal events. As will be shown later,
social-learning approaches treat internal
processes as covert events that are manipulable
and measurable. These mediating processes are
extensively controlled by external stimulus events

and in turn regulate overt responsiveness. By
contrast, psychodynamic theories tend to regard
internal events as relatively autonomous. These
hypothetical causal agents generally bear only a
tenuous relationship to external stimuli, or even to

the “symptoms” that they supposedly produce.
Freud’s famous case of Little Hans, which has been

reinterpreted by Wolpe & Rachman (1960),
illustrates some of the major differences in
explanatory models.
Little Hans exhibited, among other things, a
phobia for horses. Freud (1955) interpreted the
phobic behavior in the following manner:
He was not only afraid of horses biting him…

but also of carts, of furniture-vans, and of
buses (their common quality being, as
presently became clear, that they were all
heavily loaded), of horses that started
moving, of horses that looked big and heavy,
and of horses that drove quickly. The
meaning of these specifications was
explained by Hans himself: he was afraid of
horses falling down, and consequently
incorporated in his phobia everything that
seemed likely to facilitate their falling
down…[p. 265],
He (father) elicited from Hans the

recollection of an event at Gmunden, the
impression of which lay concealed behind
that of the falling bus-horse. While they were
playing at horses, Fritzl, the playmate of
whom he was so fond, but at the same time,
perhaps, his rival with his many girl friends,
had hit his foot against a stone and had fallen
down, and his foot had bled. Seeing the bus-
horse fall had reminded him of this accident.
…The first person who had served Hans as a
horse must have been his father; and it was
this that had enabled him to regard Fritzl as
a substitute for his father when the accident
happened at Gmunden.…In the end his father
went into the lumf symbolism, and
recognized that there was an analogy
between a heavily loaded cart and a body
loaded with faeces, between the way in
which a cart drives out through a gateway
and the way in which faeces leave the body,
and so on…[p. 126-127],
We can now recognize that all furniture-vans

and drays and buses were only stork-box
carts, and were only of interest to Hans as
being symbolic representations of
pregnancy; and that when a heavy or heavily
loaded horse fell down he can have seen in it
only one thing—a childbirth, a delivery. Thus
the falling horse was not only his dying
father but also his mother in childbirth [p.
128].
Freud’s paper reports at least four incidents in
which horses, actual or symbolic, were associated
with fear-provoking experiences capable of
producing a conditioned phobic reaction. Hans
had been frightened at seeing horses being beaten
at a merry-go-round; he was warned to avoid
horses for they might injure him; he became

frightened when a friend was accidentally hurt
while playing horses; and, in the episode that
immediately preceded the onset of the phobic
behavior, he was terrified by a bus accident in

which he believed a horse was killed.
In the psychoanalytic schema the internal

psychic disturbance is the basic cause or instigator
of the phobic responses, while external stimuli
(horses) supposedly exert little or no controlling
influence over the deviant behavior except as a
convenient focal point for Hans’s projected

Oedipal and castration feelings.
It (the phobia) extends to horses and on to

carts, on to the fact that horses fall down and
that they bite, on to horses of a particular
character, on to carts that are heavily loaded.
I will reveal at once that all these
characteristics were derived from the
circumstance that the anxiety originally had
no reference at all to horses but was
transposed on to them secondarily [italics
added] and had now become fixed upon
those elements of the horse complex which
showed themselves well adapted for certain
transferences [p. 51].
This exposition fails to account for the
variation in both the pattern and the intensity of
Hans’s anxiety reactions under different

circumstances. In fact, the case data provide
considerable evidence that external cues served as

the primary eliciting and controlling stimuli for
Hans’s phobic responses rather than simply as

incidental targets for projected feelings.
Let us consider the major traumatic episode
which was related to the onset of Hans’s phobia.

While out walking with his mother Hans saw a
large bus-horse fall and kick with its feet. He was

terrified and thought the horse was killed in the
accident. There were three important elements in
this stimulus complex—large horse, heavily
loaded transport vehicle, and horse and vehicle

traveling at high speed. The occurrence and
intensity of Hans’s subsequent phobic reactions
varied predictably as a function of the specific
patterning of these three critical stimulus

elements. Hans was more frightened of large dray
horses than of small horses, more frightened of a
rapidly moving vehicle than of a slowly moving
one, more frightened of heavily loaded vehicles
than of empty ones, and frightened when a horse-

drawn cart made a turn:
Hans: And I’m most afraid of furniture-vans too.
Father: Why?
Hans: I think when furniture-horses are dragging a

heavy van they’ll fall down.
Father: So you’re not afraid with a small cart?
Hans: No. I’m not afraid with a small cart or with a
post-office van. I’m most afraid too when a
bus comes along.
Father: Why? Because it’s so big?
Hans: No. Because once a horse in a bus fell down.
Father: What did you think when the horse fell

down?
Hans: Now it’ll always be like this. All horses in

buses’ll fall down…[p. 49],
Father: When the horse fell down, did you think of

your daddy?
Hans: Perhaps. Yes. It’s possible…[p. 51].
Father: What carts are you still afraid of?
Hans: All of them.
Father: You know that’s not true.
Hans: I’m not afraid of carriages and pair or cabs

with one horse. I’m afraid of buses and
luggage-carts, but only when they’re loaded
up, not when they’re empty. When there’s one
horse and the cart’s loaded full up, then I’m
afraid; but when there are two horses and it’s
loaded full up, then I’m not afraid.
Father: Are you afraid of buses because there are so
many people inside?
Hans: Because there’s so much luggage on the top.
Father: When Mummy was having Hanna, was she

loaded full up too? [pp. 90-91],
The Oedipal interpretation fails not only to
account for the discriminative pattern of Hans’s
phobic behavior but also to explain satisfactorily

why he was afraid of railways and locomotives as
well, a phobia which probably generalized from
the transport vehicle stimulus complex. The
psychoanalytic interpretation would demand that

the locomotive and the railway tracks were
likewise symbolic representations of the
castrating father and the impregnated mother.
The conceptual structure of causal sequences
in psychodynamic theories of behavior is
beleaguered by serious problems. An amorphous

internal determinant cannot possibly account for
the remarkable variety of heterogeneous
behaviors as well as changes both in their
incidence and magnitude under different stimulus

conditions, toward different persons, and at
different times. How can a horse phobia be
attributed to an underlying Oedipus complex and

projected castration fears if a person responds
phobically to one horse pulling a heavy loaded
vehicle, but is relatively unafraid of two horses

drawing a loaded vehicle? When diverse stimulus
inputs produce correspondingly diverse
behavioral expressions then any internal

mediators implicated in the causal sequence must
be at least equally specific and their activation

must be closely regulated by discriminative
environmental stimuli.
The conceptual difficulties associated with
psychodynamic formulations apply equally to trait

theories of personality. These approaches assume
that people possess generalized and stable
response dispositions that determine behavior in
a variety of situations. Consequently it is

considered sufficient to sample some limited
classes of response that are regarded as
dependable indicators of how persons are likely to
behave under particular conditions. The types of
behaviors selected for measurement vary. A few of

the assessment procedures that have been
advocated at one time or another are brief
samplings of overt behavior that bear some
resemblance to the trait description,
endorsements of statements that describe

affective states, interests, or response patterns,
and farfetched responses elicited by relatively

ambiguous stimuli such as inkblots, ill-defined
pictures, doll families, and incomplete sentences.
The basic assumption of trait theories—that
persons display generalized modes of behavior
that can be predicted from a restricted sampling of
responses—finds little empirical support. For
purposes of illustration, let us consider the “trait”

of aggressiveness. Several investigations
(Bandura, 1960; Bandura & Walters, 1959) of
social-learning determinants of aggressive
behavior have shown that both adolescent and
preadolescent boys display highly discriminative

patterns of aggressive responses that vary
considerably as a function of the persons with
whom they are interacting (for example, parents,
teachers, siblings, or peers). Furthermore, the
incidence of aggression even toward the same

objects differs widely depending upon whether
physical, verbal, or more attenuated forms of

responses are measured. The boys’ discriminative
aggressive responsiveness closely reflected the
considerable amount of discrimination training

that they had undergone. The parents consistently
punished aggression directed toward themselves,
but simultaneously encouraged and positively
reinforced their sons’ aggressive behavior toward

persons outside the home.
It is evident from informal observation of

differential contingencies characteristically
applied to social response systems that,
fortunately for survival purposes, cultural

practices are much too variable to produce
generalized traits. The likelihood that a given
pattern of behavior will be rewarded, ignored, or
punished is dependent upon, among other factors,
the characteristics of the performer, the specific

form and intensity of the behavior, the objects
toward whom the actions are directed, the social
situations in which they occur, and various
temporal factors. Thus a high degree of behavioral
flexibility is required if a person is to meet the
complexities of ever changing environmental
demands. In the case of social response systems,
most of which are characterized by high
behavioral specificity, trait measurement is a

disappointing pursuit. Indeed, a comprehensive
review of the relevant empirical literature by
Mischel (1968) reveals low intercorrelations
among different measures purporting to assess
the same trait, weak relationships between

component parts of gross trait dimensions, and
little consistency of behavior patterns in different
stimulus situations. On the other hand, intellectual
performances, which are more or less uniformly
rewarded by different agents at different times

and in different settings, show substantial
consistency.
In the assessment process, behavioral data,
however obtained, are typically converted into
trait or psychodynamic constructs that are far
removed from the actual feelings and actions of
the person being evaluated. This practice rests on
the assumption that the abstractions represent
more generic systems and, therefore, possess

greater predictive power. As Mischel (1968) has
noted in a review of evidence bearing on this
issue, the transformation shifts the focus of
attention from what a person does to speculations
about what he has; from concern about the client’s

behavior to engrossment in the diagnosticians
categories of behavior. The evidence indicates that
these hypothetical constructions are better
predictors of diagnosticians’ semantic and
conceptual stereotypes than of clients’ actual

attributes and psychological reality. It therefore
comes as no surprise that assessment strategies

deriving from the dynamic trait point of view have
generally failed to match the predictive efficacy of
actuarial methods (Meehl, 1954).
The tenacious belief in generalized response
dispositions is attributed by Mischel (1968) to the
tendency to construe behavioral consistencies
even from variable performances. Hence,

generality may emerge in the inferential construct
domain, whereas a high degree of specificity may
obtain at the behavioral level. Among the factors
listed as reinforcing the impression of consistency
are included physical constancies in appearance,

linguistic characteristics, and stylistic features;
regularities in the stimulus situations in which a
person is repeatedly observed; reliance upon
broad and ambiguous trait categories that
encompass heterogeneous behaviors; utilization of

test items that require a person to rate his
behavior in “typical” social contexts rather than in

a variety of specific situations; and strong
psychological pressures to maintain a consistent,
stable view of events. Inconsistencies, therefore,

tend to be resolved by glossing over, ignoring, or
reinterpreting discrepant evidence.
The preoccupation with internal psychic

agents and energized traits has been largely
responsible for the limited progress in
development of empirically sound principles of

human behavior. The gap between stimulus inputs
and overt response events tends to be filled
readily with diverse all-powerful, animistic

constructs capable of generating and explaining
almost any psychological phenomenon. These
constructs, of course, lend themselves easily to
pseudo explanations (Skinner, 1961) in which
renaming of a behavioral phenomenon is offered

as an explanation. For example, persons who
exhibit withdrawal, delusional and hallucinatory
behavior, inappropriate emotional responses, and
behavioral deficits, will be labeled schizophrenic.
The presence of these deviant behaviors is then
attributed to an underlying schizophrenia, an
explanation that is completely circular and
contains no information whatsoever about causal
determinants. An adequate causal explanation

must specify clearly the independent variables
that produce and maintain the observed
schizophrenic behavior. In a similar manner,
traits, complexes, and dynamics, which represent
the descriptive constructs of the assessor, often

are made active entities within the client that
supposedly cause his behavior.
The major deficiencies of theories that explain
behavior primarily in terms of conjectural inner

causes would have been readily demonstrated had
they been judged, not in terms of their facility in
interpreting behavioral phenomena that have

already occurred, but rather on the basis of their
efficacy in predicting or modifying them. Because
the internal determinants propounded by these
theories (such as mental structures, Oedipal
complexes, collective unconscious) could not be
experimentally induced, and rarely possessed
unequivocal consequences, psychodynamic

formulations enjoyed an immunity to genuine
empirical verification. If progress in the
understanding of human behavior is to be
accelerated, psychological theories must be judged
by their predictive power, and by the efficacy of

the behavioral modification procedures that they
produce.
ADVERSE CONSEQUENCES OF DISEASE

INTERPRETATIONS OF DEVIANT BEHAVIOR
The conceptualization of deviant behavior as
manifestations of disease has, in several ways,
impeded development of efficacious methods of
behavioral change. In the first place, it led to heavy

reliance upon physical and chemical interventions,
unremitting search for drugs as quick remedies for

interpersonal problems, and long-term neglect of
social variables as influential determinants of
deviant response patterns. Secondly, the
mislabeling, partly by historical accident, of social

deviations as symptoms of mental illness
established medical training as the optimal
preparation for psychotherapeutic work. In fact,
such training, because of its primary concern with

somatic processes and pathologies, leaves one ill-
prepared for devising and implementing methods
that are successful in promoting favorable social
change. Had educational processes, which also
depend upon neurophysiological functioning, been

historically misconstrued as principally medical
phenomena, our society would undoubtedly be

faced with the same critical shortage of
educational facilities and well-trained

instructional personnel that characterize our
current “mental health” enterprises.
Although the designation of behavioral
eccentricities as manifestations of disease initially
resulted in more humane treatment, as Szasz
cogently points out, continued adherence to this

analogy has become a serious hindrance. Many
people who would benefit greatly from
psychological treatment avoid seeking help
because they fear being stigmatized as mentally
deranged, which often carries deleterious social

consequences. Those who are compelled by
chronic distress to seek a solution to their
interpersonal problems are typically ascribed a
sick role and are regarded as relatively helpless,
dependent, and incompetent in managing their

daily lives. By having their behavioral deviations
treated as expressions of internal psychic

pathologies they are thereby relieved of the
natural consequences of their actions. In this
connection, it is important to distinguish judicious

management of reinforcement contingencies
aimed at altering the course of future behavior
from moral judgments of personal responsibility
for past actions. There is little to be gained from

condemning delinquents for their history of
antisocial behavior, but there is much to be gained

from having them experience new response
consequences that will help them develop a more
effective way of life. When individuals are labeled
mentally ill, this often results not only in

suspension of customary response consequences
essential for change, but in substitution of
contingencies that foster maladaptive tendencies

(Ayllon & Michael, 1959). Moreover, as will be
shown later, for people who undergo long-term

institutionalization, the attendant stigmatization,
the patient-role requirements of the mental
hospital culture, the limited opportunities to

perform behaviors that are necessary in
community life, and the development of
institutional dependency produce further
impediments to successful readjustment to typical
environmental demands.
The medical orientation toward deviant
behavior has resulted also in a disinterest in, and

lack of facilities for, the modification of lesser, but
nevertheless troublesome, forms of psychological
problems. People with circumscribed behavioral

difficulties are justifiably unwilling to label
themselves mentally deranged and to enter into a
protracted expensive treatment that offers no
guarantee of success. Thus, for example, people
who suffer from snake phobias may be unable to

perform their work under certain conditions, to
participate in camping and other outdoor
activities, or to reside in locales inhabited by
harmless snakes. Treatments derived from social-
learning principles are now available that can
effectively eliminate such phobias in any person in
a few sessions (Bandura, Blanchard, & Ritter,
1969). Psychological centers that offer brief and
highly efficacious treatments for specific

behavioral dysfunctions would provide valuable
therapeutic services to many persons who would
otherwise endure unnecessary restrictions in
certain areas of their psychological functioning.
The designation of divergent beliefs and

actions as “sick” may also have an important
impact on the more general process of social
change. Improvements in the conditions of life
within a society require the continuous

modification of its institutionalized patterns of
behavior and the replacement of old standards of
conduct with new ones that are more fitting to the

altered circumstances. Proposed social reforms,
however, typically meet with strong resistance,
particularly if they represent marked departure
from established traditions and threaten vested
interests. Consequently persons often find it
necessary to violate institutionalized codes of
behavior in order to force a change in the social

system. In such instances, deviance serves a
positive function in promoting constructive
modifications. The conforming populace, despite
its protestations, eventually profits from the
nonconformists’ deviance.
Resistance to advocated social changes

sometimes takes the form of publicly labeling
those who advocate divergent practices as
emotionally disturbed. This diagnostic devaluation

is most easily applied when social deviants
attempt, as they usually do, to differentiate
themselves from the general populace by adopting

unconventional attire and hair styles or peculiar
symbols and rituals. In some totalitarian societies
it is not uncommon to silence authors who
propose certain social and political reforms by
diagnosing them as mentally deranged and
committing them to psychiatric hospitals
(Crankshaw, 1963). Although our own society

rarely imposes such legal sanctions, active
nonconformists are often discredited by
characterizing them as “perverts” and members of
the “lunatic fringe.” A society would better
preserve its potential for change by defining social

deviance as innovative rather than “sick” behavior.
Such a practice would favor evaluation of
proposed changes on the basis of their merits and
probable long-term consequences, as should be
the case.
Since social control through stigmatizing
deviance as psychic malfunction has gained

currency in our society, it would be surprising if
such mislabeling were confined to matters of
cultural norms and objects. Even the
diagnosticians themselves may yield to the
temptation to brand any dissidence as
psychopathological. In one such illustration
(Gitelson, 1962), departure from orthodoxy in

psychoanalytic theory is explained not by factual
and theoretical disagreements, but in terms of
“pathological narcissism,” “transference
neuroses,” and other psychodynamic malfunctions
in dissenting members.
Szasz (1965), who has been especially

concerned about the promotion of moral
prescriptions in the guise of psychiatric diagnoses,
has written widely on the contemporary misuse of

the notion of mental illness. He argues that, in an
effort to ensure more benevolent treatment of
persons in difficulty, they are certified as afflicted

with a mental disease. This advantage, however, is
gained at the expense of stigmatization,
degradation, and restriction of personal freedom.
Rather than the “bootlegging of humanism” on
psychiatric grounds he advocates frank
confrontation of the socio-ethical issues involved
in societal practices and active efforts to bring

about needed reforms. To take legalized abortion
as an example, Szasz (1962) contends that it
would be more honest to grant people the right to
determine for themselves whether they wish to
bring a child into the world than to invoke

psychiatric illness as a subterfuge for performing
abortions. As an analogy, if divorces were granted
only on the basis of psychiatric certification of
mental illness, the incidence of mental
derangement would suddenly rise astronomically.
Interpretation of Causal Processes
Preoccupation with internal response-
producing agents has resulted in a disregard of
external variables that have nevertheless been
shown to exercise control over behavior. An
organism that is impelled from within but is
relatively insensitive to environmental stimuli or
to the immediate consequences of its actions

would not survive for long. Human functioning, in
fact, involves interrelated control systems in
which behavior is determined by external
stimulus events, by internal information-
processing systems and regulatory codes, and by

reinforcing response-feedback processes.
Stimulus Control of Behavior
During initial phases of human development,

stimuli, except those which are inherently
aversive, exert little or no influence upon
individuals. Eventually, however, as a result of
undergoing either direct or vicarious experiences,

individuals’ behavior comes to be regulated by
antecedent stimulus events that convey
information about probable consequences of
certain actions in given situations. The
development of appropriate anticipatory reactions
to recurrent environmental cues has considerable

functional and survival value. Indeed, an
individual who did not learn to avoid physical
hazards, who did not respond appropriately to
traffic signals and other guiding cues, for example,
and who remained indifferent to important social

and symbolic stimuli, would suffer a painfully
rapid extinction.
STIMULUS CONTROL OF AUTONOMIC

RESPONSIVENESS
Many problems for which people seek relief
involve distressing autonomic overactivity
reflected in a variety of somatic complaints of a
functional nature, including chronic “tension” and

anxiety reactions, gastrointestinal disorders, and
respiratory and cardiovascular disturbances.

Conditioned emotionality is also generally
implicated, particularly during the acquisition
phase, in obsessive-compulsive reactions,
behavioral inhibitions, and phobic and other

avoidance behaviors. Depressant drugs may
provide temporary relief from intense autonomic
responses, but in cases where they are under
stimulus control, social-learning procedures that

are capable of neutralizing the emotion-arousing
properties of stimulus events offer the most direct
and effective treatment.
Autonomic responses can be most readily
brought under the control of environmental
stimuli through classical conditioning operations.
If a formerly ineffective or conditioned stimulus is

closely associated with an unconditioned stimulus
capable of eliciting a given physiological response,

the former stimulus alone gradually acquires the
power to evoke the physiological response or its

equivalent. Although some types of autonomic
responses are more difficult to condition than
others, almost every form of somatic reaction that
an organism is capable of making, including

respiratory and heart-rate changes, increases in
muscular tension, gastrointestinal secretions,
vasomotor reactions, and other indices of
emotional responsiveness (Bykov, 1957; Kimble,
1961), has been classically conditioned to

innocuous stimuli. Environmental events can
likewise acquire the capacity to control
electroencephalographic arousal through
association with either external evocative stimuli
or direct central stimulation (John, 1967).
Laboratory studies concerned with the
production of asthmatic attacks illustrate how

psychosomatic reactions can be brought under
stimulus control. Noelpp & Noelpp-Eschenhagen
(1951, 1952), for example, demonstrated that
following repeated pairing of induced asthmatic
attacks with an auditory stimulus, many of the
guinea pigs in the study exhibited respiratory
dysfunctions characteristic of bronchial asthma in

response to the conditioned auditory stimulus
alone. Stimulus control of human asthmatic
attacks is similarly demonstrated in an experiment
by Dekker, Pelser, & Groen (1957). Two patients
suffering from severe bronchial asthma inhaled

nebulized allergens to which they were
hypersensitive. After repeated inhalations of the
allergen extract that served as the unconditioned
stimulus for asthmatic attacks, inhalation of a
neutral solvent of the allergen alone, which

initially produced no respiratory changes, elicited
attacks of asthma as demonstrated by clinical

signs and vital capacity measures. In later phases
of the experiment inhalations of pure oxygen and
even the presentation of the mouthpiece, both

formerly neutral stimuli, had acquired the power
to provoke asthmatic attacks which were
indistinguishable from those induced by the
allergen itself.
In the experiment described, asthmatic
responses were conditioned to elements of the

inhalation situation and apparatus through
contiguous association. It is not surprising,
therefore, that analyses of asthmatic behavior by

Dekker & Groen (1956) produced an extremely
varied array of highly specific eliciting stimuli in
the group of patients studied; these included the
sight of dust, radio speeches by influential
politicians, children’s choirs, the national anthem,

elevators, goldfish, caged birds, the smell of
perfume, waterfalls, bicycle races, police vans, and
horses. Once the critical eliciting stimuli had been
identified in a particular case, Dekker and Groen
were able to induce attacks of asthma by
presenting the conditioned stimuli in actual or in
pictorial form. In some cases, of course, more
complex interpersonal events may serve as major
evocative stimuli.
Of particular interest is the investigators’
observation that intense emotional arousal itself
failed to produce asthmatic reactions, whereas
exposure to specific asthmatic conditioned stimuli
typically provoked marked respiratory

dysfunction. The latter observation is
corroborated by Ottenberg, Stein, Lewis, &
Hamilton (1958) in a study of the classical
conditioning and extinction of asthmatic

responses in guinea pigs. Asthma-like attacks,
which readily occurred in the presence of
conditioned stimuli, could not be induced by

means of emotion-provoking procedures involving
loud noises, painful stimulation, and electric
shock. In view of these findings, one would expect
that direct neutralization of specific eliciting
stimuli (Moore, 1965; Walton, 1960) would be
effective in modifying asthmatic responses under
the control of environmental stimuli, but that

reduction of general emotional disturbances may
have little impact on the respiratory disorder.
Both the processes and outcomes
accompanying classical conditioning operations
are considerably more complex than the general

principle might imply. Persons often display
differential susceptibility to autonomic
conditioning, which suggests that other variables
—possibly genetic, physiological, or psychological

—are contributory factors. It will also be shown
later that cognitive representation of the
contingent relationship between conditioned and

unconditioned stimuli markedly facilitates
classical conditioning. These findings call into
question peripheral theories of conditioning.
Higher-Order Conditioning. Many of the
emotional responses that persons exhibit toward
specific objects are not products of direct

associations of affective experiences with the
objects themselves. Some people, for example,

may respond anxiously toward snakes without
having had any direct aversive encounters with
them. Similarly, persons often display strong
emotional arousal at the sight or mention of

unpopular minority groups or nationalities on the
basis of little or no personal contact. These types
of reactions are frequently established on the

basis of higher-order sequences depending on the
presence or absence of particular stimuli. This

process is most clearly illustrated in simple
laboratory studies in which certain responses are
reinforced only in the presence of one stimulus

(e.g., green light), but never in a different stimulus
context (e.g., red light). After the discrimination
has been formed, a person responds only in the
presence of the green light. Thus by introducing

into the environment a discriminative stimulus
that signifies whether a particular performance is
likely to be reinforced, a considerable degree of
control over behavior can be achieved.
The following quotation presents a more
telling example of stimulus control of behavior

occurring under naturalistic conditions. In this
illustration an elaborate pattern of aggressive
behavior by an autistic boy was rarely exhibited in
the father’s presence but freely expressed in his
absence.
Whenever her husband was home, Billy was

a model youngster. He knew that his father
would punish him quickly and
dispassionately for misbehaving. But when
his father left the house, Billy would go to the
window and watch until the car pulled out.
As soon as it did, he was suddenly
transformed. . . . ‘He’d go into my closet and
tear up my evening dresses and urinate on
my clothes. He’d smash furniture and run
around biting the walls until the house was
destruction from one end to the other. He
knew that I liked to dress him in nice clothes,
so he used to rip the buttons off his shirts,
and used to go in his pants’ [Moser, 1965, p.
96].
Laboratory investigations of stimulus control
processes often involve simple situations in which
stimuli differ either on a single attribute or on a

few easily identifiable dimensions. In most real life
circumstances the cues which designate probable

consequences usually appear as part of a
bewildering variety of irrelevant events. One must,

therefore, abstract the critical feature common to
a variety of situations. Behavior can be brought
under the control of abstract stimulus properties if
responses to situations containing the critical

element are reinforced, whereas responses to all
other stimulus patterns lacking the essential
element go unreinforced. It should be noted here
that the controlling function of various social and
environmental stimuli is usually established

simply by informing people about the conditions
of reinforcement that are operative in different
situations, rather than by leaving them to discover
it for themselves through a tedious process of
selective reinforcement. However, the existence of

differential consequences is essential to maintain
stimulus control produced through instructional
means.
In discussions of stimulus control processes it

has been customary to distinguish between the
eliciting and the discriminative or response-
directing functions of stimulus events (Skinner,

1961). As noted earlier, autonomic responses are
elicited by their controlling stimuli, independently
of their subsequent consequences. An asthmatic
conditioned stimulus, for example, will induce
respiratory changes apart from the social effects
resulting from somatic reactions. On the other
hand, in the case of instrumental responses, the

discriminative stimuli simply modify the
probability that a given response will occur, but
they do not elicit it. Moreover, the stimulus control
of operant or instrumental behaviors is
established and maintained by differential

response consequences rather than through
temporal association of sets of stimulus events.
Under naturalistic conditions behavior is
generally regulated by the characteristics of

persons toward whom responses are directed, the
social setting, temporal factors, and a host of
verbal and symbolic cues that signify predictable

response consequences. Social situations,
particularly those involving a large number of
multidimensional cues, seldom recur with exactly
the same constituent elements. Because of the
constant variation in the nature and patterning of
stimuli, social learning would be an interminable
and exceedingly laborious process if responses

were entirely specific to the situation in which
they had been originally reinforced. However,
performances that have been reinforced in the
presence of certain cues are also controlled by
other stimuli which are related to them either

physically or semantically. After generalized
stimulus control has been established it can be
narrowed, if necessary, by differential
reinforcement of responses to stimuli whose
differences are progressively reduced (Terrace,

1966).
Outcome Control of Behavior
An organism that responded anticipatorily to

informative environmental cues but remained
unresponsive to the outcomes produced by its
behavior would enjoy a tragically brief life-span.
Fortunately, instrumental responses are
extensively controlled by their immediate

consequences. Responses that result in nonreward
or punishing effects are generally eliminated,
whereas those that are successful in securing
positively reinforcing outcomes are retained and
strengthened. There is some evidence (Kimmel,

1967; Miller, 1969) that autonomic responses,
which formerly were believed to be subject only to
classical conditioning, can also be modified
instrumentally to some degree by differential
consequences. Indeed, DiCara & Miller (1968)

were able to establish remarkably precise control
over vasomotor activities through differential

reinforcement.
External Reinforcement. Research conducted by
Harris, Wolf & Baer (1964), designed to modify
gross behavior disorders in nursery school
children by altering their teachers’ attentional
responses, provides impressive demonstrations of
how deviant behavior can be controlled by its

social consequences. Each case involved an
intrasubject replication design in which behavior
was successively eliminated and reinstated by
systematic variation of reinforcement
contingencies. This is a most powerful method for

isolating the controlling conditions of behavioral
phenomena. The procedure in any given case
contains four steps.
First, the child is observed for a period of time

to measure the incidence of the deviant behavior,
the contexts in which it typically occurs, and the
reactions it elicits from teachers. In one case an

extremely withdrawn boy spent approximately 80
percent of his time in solitary activities in isolated
areas of the nursery school. Observation revealed
that the teachers unwittingly reinforced his
solitariness by paying a great deal of attention to
him, reflecting his feelings of loneliness, consoling
him and encouraging him to play with other

children. When he did happen to join other
children, the teachers took no particular notice.
In the second phase of the program a new set
of reinforcement practices is substituted.
Continuing with the above example, the teachers

stopped rewarding solitary play with attention
and support. Instead, whenever the boy sought out
other children, the teacher immediately joined the
group and gave it her full attention. In a short

time, the boy’s isolation declined markedly and he
was spending about 60 percent of his time playing
with other children (Figure 1-1).
After the desired changes in behavior have
been produced, the original reinforcement
practices are reinstated to determine if the initial
Figure 1-1. Percentage of time a withdrawn boy spent in
social interaction before treatment began, during
periods when social behavior toward peers was
positively reinforced, and during periods when
teachers gave attention for solitary play. Harris, Wolf, &
Baer, 1964.
behavior was in fact maintained by its social
consequences. In this third stage, for example, the
teachers again paid no attention to the child’s

sociability but instead responded with comforting
ministrations whenever he was alone. The effect of
this traditional “mental hygiene” treatment was to
increase the child’s withdrawal to the original high

level (Figure 1-1).
In the final phase of the program the

therapeutic contingencies are reintroduced, the
deviant behavior is eliminated and the desired
behavior patterns are generously reinforced. In
the above case, after social responsivity was well
established the frequency of positive attention

from adults was gradually diminished as the boy
derived increasing enjoyment from play activities

with his peers. Follow-up observations disclosed
that the boy maintained his sociable pattern of

behavior, which contrasted markedly with his
previous isolation.
Children with a wide variety of behavior
disorders have participated in such programs, and

in each case their maladaptive behavior was
eliminated, reinstated, and removed a second time
simply by altering teachers’ social responsiveness
(Harris, Wolf, & Baer, 1964). Additional
demonstrations of reinforcement control of

grossly deviant behavior in both children and
adults are provided by Ayllon and his associates
(Ayllon & Azrin, 1965; Ayllon & Michael, 1959)
and by Wolf, Risley, & Mees (1964).
Reinforcement control of behavior is further

demonstrated by evidence that different
frequency and patterning of outcomes produce
different types of performance (Ferster & Skinner,
1957). When subjects are rewarded each time
they exhibit the desired behavior (continuous
schedule), and later the reinforcement is
completely withdrawn, they are likely to increase
responsiveness for a brief period of time and then

to display a rapid decrease in performance, often
accompanied by emotional reactions.
Sometimes behavior is reinforced only after a

specified period of time has elapsed (fixed-interval
schedule). Pay periods, eating schedules,
recreational times, and other regularly scheduled

rewarding activities illustrate the temporal cycles
of reinforcement regulating some aspects of
human behavior. When rewards are dispensed on
a fixed temporal basis the payoff is the same
regardless of the amount of behavior produced

during the intervening period. Under these
conditions, once a person develops a temporal

discrimination, the response output following
reinforcement is very low but accelerates rapidly

as the time for the next reinforcement approaches.
In naturalistic situations where temporal
reinforcement cycles may range over several
hours, days, weeks, or even months, social

approval or coercive forms of pressure are usually
brought to bear in order to maintain a steady rate
of performance. Nevertheless, even with these
added inducements, the fixed-interval schedule is
likely to generate only the minimum output

expected in a given situation, particularly if the
activity itself is somewhat unpleasant. On the
other hand, where given performances have
become intrinsically rewarding, satisfactions
derived from the activity itself may greatly

outweigh the influence of temporally occurring
rewards.
Much human behavior is sustained by ratio
schedules in which reinforcement is made
contingent upon the amount of behavior rather
than on the passage of time. In a fixed-ratio
schedule a person must complete a specified
amount of work for each reinforcement. Since
under these circumstances reinforcement depends

upon the person’s own behavior, these schedules
usually generate high and stable responsiveness.
By starting with a low ratio and gradually raising
the number of performances required per
reinforcement, very high performance rates can be

developed and maintained for a long period with
minimum reinforcement. Although ratio schedules
are exceedingly effective in generating a high
behavioral output, persons in extra-laboratory
situations, where they have considerably more

freedom of action, are likely to withdraw from
situations with schedules requiring substantial

performances for minimal returns, and to select
more beneficent reinforcing agents.
In everyday life most reinforcements are
available not only on an intermittent basis, but
also on variable schedules. The effects of variable-
interval and variable-ratio schedules on
performance have been extensively studied under

controlled laboratory conditions. In the former
case, the length of time between successive
reinforcements is varied randomly around some
mean temporal value; in the variable-ratio
schedules, the number of responses per

reinforcement is varied around a selected aver age
ratio. Since the reinforcers are dispensed
unpredictably, the usual temporal or rate
discriminations that result in cyclic responsivity
cannot develop; consequently, variable schedules

generate higher rates of response and more stable
and consistent performances than those in which

outcomes occur on a regular or fixed basis.
However, even under irregular reinforcement,
ratio schedules are more effective than interval

schedules. Research evidence in fact reveals that,
of all the variations in scheduling procedures
available, the variable-ratio schedule is most
powerful in sustaining behavior. A casual

observation of the patrons of the gambling devices
at Las Vegas attests to the generality and validity

of laboratory findings.
Evidence of schedule control of behavior has
important implications for the understanding of

behavior and for its modification. Those who have
been reared under more or less continuous
reinforcement conditions are likely to become
easily discouraged and to cease responding when
faced with frustrating nonreward or failure. By

contrast, persons whose response patterns have
been reinforced only intermittently will persist in
their behavior for a considerable time despite
setbacks and infrequent reinforcement. This, of
course, is the reinforcement history that is most
characteristic of all stable response patterns
including deviant ones. Moreover, when efforts
are made to extinguish such behavior, it is not
unusual for a parent or other persons to give in

temporarily by rewarding the behavior,
particularly if it goes on unabated or increases in
rate or intensity. Any reinforcements occurring
during the extinction process, however, will
reinstate the behavior, often at a higher level than

if extinction had not been attempted.
There are other subtle variations in the
patterning of reinforcement that significantly
influence the characteristics of behavior. As will be

shown later, differential reinforcement of behavior
that is persistent, or of high magnitude, is another
form of intermittence that establishes deviant and

obstreperous behavior of unusual resiliency.
Reinforcements can also be applied in such a way
as to produce delayed behavior. This outcome is
achieved in laboratory studies by making rewards
available after a given period of time has elapsed,
but only if the subject has refrained from
responding during the interval. Each time the

subject responds prematurely the enforced
waiting period is begun all over again. By
gradually lengthening the time interval, self-
control in the subject can be increased.
In everyday life different classes of social

behavior are controlled by multiple schedules of
reinforcement operating either concurrently or
alternately. This process is most dramatically
illustrated in an experiment conducted by Ferster

(Ferster & Skinner, 1957), in which the right-hand
responses of a subject were reinforced on a fixed-
ratio schedule, whereas responses with the left

hand were reinforced simultaneously on a
variable-ratio schedule. The subject produced two
remarkably different sets of performances, each
corresponding to the typical response-rate curves
of these types of schedules. Finally, it should be
noted that different types of positive and negative
consequential events possess differential

controlling power. The theoretical issues and
empirical findings relevant to this reinforcement
variable will be considered in subsequent chapters
of this book.
Vicarious Reinforcement. The discussion thus

far has been concerned with the extent to which
responsiveness is regulated by external outcomes
impinging directly upon a performer. There is
considerable evidence (Bandura, 1965) that the

behavior of observers can be substantially
modified as a function of witnessing other people’s
behavior and its consequences for them.

Observation of rewarding consequences generally
enhances similar performances, whereas
witnessing punishing outcomes has an inhibiting
effect on behavior. Systematic investigations of the
relative efficacy of vicarious and direct
reinforcement reveal that the changes exhibited
by observers are of the same magnitude (Kanfer,

1965) or, under certain conditions, may even
exceed those achieved by reinforced performers
(Berger, 1961; Marlatt, 1968). Moreover, vicarious
reinforcement processes are governed by
variables such as the percentage (Bisese, 1966;

Kanfer, 1965), intermittence (Rosenbaum &
Bruning, 1966), and magnitude (Bruning, 1965) of
reinforcement in essentially the same manner as
when they are applied directly to a performing
subject. Although the efficacy of vicarious

reinforcement practices is well established, the
behavioral changes displayed by observers may be

interpreted in several ways.
One possible explanation is in terms of the
discriminative or informative function of
reinforcing stimuli presented to the model.
Response consequences experienced by another
person undoubtedly convey information to the
observer about the probable reinforcement

contingencies associated with analogous
performances in similar situations. Knowledge
concerning the types of responses that are likely
to meet with approval or disapproval can later
serve a self-instructional function in facilitating or

inhibiting emulative behavior. The information
gained from witnessing outcomes experienced by
others would be particularly influential in
regulating behavior under conditions where
considerable ambiguity exists as to what actions

are permissible or punishable, and where the
observer believes that the models’ contingencies

apply to himself as well. It is highly unlikely, for
example, that witnessing social approval for
physical aggression exhibited by a person

occupying a unique role, such as a policeman,
would enhance imitative aggressiveness in
observant citizens to any great extent.
Experiments are therefore needed that test the

magnitude of vicarious reinforcement effects as a
function of comparability of social sanctions

customarily applied to models and to observers.
Typically, models’ responses are differentially
reinforced depending upon the persons toward

whom the behavior is directed and the social
settings in which it is expressed. When differential
consequences are correlated with different
stimulus conditions, observation of the
reinforcement pattern associated with the models’

responses helps the observer to identify the social
or environmental stimuli to which the modeled
behavior is most appropriate. These relevant cues
may be difficult to distinguish without the
observed informative feedback. Hence, through
repeated exposure to the outcomes of others, an
observer not only acquires knowledge of
predictable reinforcement contingencies, but he
may also discern the situations in which it is most

appropriate to exhibit a given pattern of behavior.
The resultant discrimination learning can later
facilitate the performance of matching responses
in the presence of the cues to which the model
previously had been responding with favorable

consequences (Church, 1957; McDavid, 1962;
Paschke, Simon, & Bell, 1967).
Observation of reinforcing outcomes and the
models’ concomitant reactions may also have

important activating or motivational effects on an
observer. The mere sight of highly valenced
reinforcers can produce anticipatory arousal

which, in turn, will affect the level of imitative
performance. Thus, for example, witnessing a
performer rewarded with a culinary treat for
executing a given sequence of responses will
convey the same amount of information about the
probable reinforcement contingencies to a
famished and to a satiated observer, but their

subsequent imitative performances will, in all
likelihood, differ radically because of the
differential effects of deprivation state on the
activating power of the anticipated incentive.
Similarly, variations in the magnitude of observed

reinforcers, while providing equivalent
information about the permissibility of matching
responses, have different motivational effects on
observers (Bruning, 1965). As in the case of direct
reinforcement, incentive-produced motivation in

observers is most likely to affect the speed,
intensity, and persistence with which matching

responses are executed.
A vicarious reinforcement event not only
provides information concerning probable
reinforcement contingencies, knowledge about the
types of situations in which the behavior is
appropriate, and displays of incentives possessing
activating properties, but it also includes affective

expressions of models undergoing rewarding and
punishing experiences. As was mentioned earlier,
the pleasure and pain cues emitted by a model
generally elicit corresponding affective responses
in the viewer. These vicariously aroused

emotional responses can readily become
conditioned either to the modeled responses
themselves, or to environmental stimuli that are
regularly correlated with the performer’s affective
reactions. As a consequence the subsequent

initiation of matching responses by the observer
or the presence of the correlated environmental

stimuli is likely to generate some degree of
emotional arousal. In a similar manner, witnessing
the nonoccurrence of anticipated aversive

consequences to a model can extinguish in
observers previously established emotional
responses that are vicariously aroused by modeled
displays. It is therefore possible that the

facilitative or suppressive effects of observing the
affective consequences for the model may be

partly mediated by the vicarious conditioning or
extinction of emotional responses.
Finally, reinforcements administered to

another person may have important consequences
in social evaluation. Punishment is apt to devalue
the model and his behavior, whereas models who
receive praise and admiration tend to be
attributed prestige and competence (Bandura,

Ross, & Ross, 1963; Hastorf, 1965). Changes in
model status, in turn, can significantly affect
observers’ subsequent performance of matching
responses. A particular vicarious reinforcement
event, depending upon its nature and context, may
thus produce behavioral changes in observers
through any one or more of the five processes
outlined.
The effects of observed consequences upon

performance are also likely to be influenced by the
social conditions under which the vicarious events
occur. Almost without exception, the studies
discussed above employ a paradigm in which
observers’ behavior is measured after they have

witnessed another person either rewarded or
punished by an agent with whom the observers
never have any contact and in social settings that
differ from their own. Observed consequences

may have different behavioral effects under
conditions where the reinforced performers and
the observers are members of the same group who

are present in the same setting and interacting
with the same social agents. Observers who
witness other members rewarded for a certain
pattern of behavior may temporarily increase
similar responding, but if their behavior is
consistently ignored they are apt to discontinue
the modeled behavior or even respond negatively

to .the agent’s preferential treatment.
Self-Reinforcement. Although the controlling
power of externally occurring consequences
cannot be minimized, self-reinforcement may
frequently outweigh the influence of external

outcomes in governing social behavior,
particularly in the case of older children and
adults. Until recently, self-reinforcement
phenomena have been virtually ignored in

psychological theorizing and experimentation,
perhaps as a result of preoccupation with
infrahuman learning. Unlike humans, who

continually engage in self-evaluative and self-
reinforcing behavior, rats or chimpanzees are
disinclined to pat themselves on the back for
commendable performances, or to berate
themselves for getting lost in culs-de-sac. By
contrast, people typically set themselves certain
standards of behavior and self-administer

rewarding or punishing consequences depending
on whether their performances fall short of,
match, or exceed their self-prescribed demands.
Self-reinforcing responses are to some extent
directly established through selective

reinforcements administered initially by
socialization agents. In this learning process an
agent adopts a criterion of what constitutes a
worthy performance and consistently rewards

persons for matching or exceeding the adopted
standard, while nonrewarding or punishing
performances that fall short of it. When

subsequently persons are given control over the
administration of reinforcers they are likely to
reinforce themselves in a similarly selective
manner. In a study investigating the effects of
miserly and indulgent training on rate of self-
reinforcement, Kanfer & Marston (1963)
rewarded the performances of some adults

generously with token reinforcers accompanied by
an approving attitude toward self-reward,
whereas with others the experimenter parted
grudgingly with a few tokens and cautioned
subjects against requesting rewards for

undeserving performances. Those who received
lenient training subsequently rewarded
themselves far more frequently on a different task
than subjects who were stringently trained even
though the achievements for both groups were

comparable.
There exists a substantial body of evidence

that modeling processes play a highly influential
role in the transmission of self-reinforcement
patterns. In the prototypic experiment (Bandura &
Kupers, 1964) subjects observe a model
performing a task in which he adopts either a high
performance standard or a relatively low criterion
for self-reinforcement. On trials in which the

model attains or exceeds the self-imposed demand
he rewards himself materially and expresses
positive self-evaluations, but when his attainments
fall short of the adopted behavioral requirements
he denies himself available rewards and reacts in a

self-derogatory manner. Later observers perform
the task, during which they receive a
predetermined range of scores and the
performances for which they reward themselves
are recorded. Within this general paradigm the

independent and interactive effects of a variety of
theoretically relevant variables have been studied

including, among others, prior reinforcement
history for achievement behavior and degree of
difference in ability from comparison models

(Bandura & Whalen, 1966); presence of conflicting
modeling cues (Bandura, Grusec, & Menlove, 1967;
McMains & Liebert, 1968), rewarding qualities of
the model and social reinforcement of the model’s

standard-setting behavior (Bandura, Grusec, &
Menlove, 1967); whether material self-reward is

accompanied by verbal self-evaluation (Liebert &
Allen, 1967); and the generosity with which
symbolic rewards are self- administered (Marston,
1965a).
The results of these studies show that people
generally adopt the standards for self-
reinforcement exhibited by exemplary models,
they evaluate their own performances relative to

that standard, and then they serve as their own
reinforcing agents. For instance, those who have
been exposed to models setting low standards
tend to be highly self-rewarding and self-
approving for comparatively mediocre
performances. By contrast, persons who have
observed models adhere to stringent performance
demands display considerable self-denial and self-
dissatisfaction for objectively identical

accomplishments. These findings illustrate how
self-esteem, self-concept and related self-
evaluative processes can be conceptualized within
a social-learning framework. From this
perspective, a negative self-concept is defined in

terms of a high frequency of negative self-
reinforcement and conversely, a favorable self-
concept is reflected in a relatively high incidence
of positive self-reinforcement (Marston, 1965b).
Although specific patterns of self-reinforcing

responses can be acquired observationally
without the mediation of direct external

reinforcement, undoubtedly the valuation of
performances that fall short of, match, or exceed a
reference norm results partly from past
differential reinforcements. Thus, for example,
parents who expect their children to exceed the
average performance of their group in whatever
tasks they undertake will selectively reward

superior achievements and punish or nonreward
average and lower level attainments. Differential
achievement levels thus assume positive and
negative valence and the performance standard
common to the various activities is eventually

abstracted and applied to new endeavors. That is,
a person for whom average performances have
been repeatedly devalued will come to regard
modal achievements on new tasks as inadequate
and attainments that surpass modal levels as

commendable. Once the evaluative properties of
differential accomplishments are well established,

adequate or inadequate matches are likely to elicit
similar self-reinforcing responses irrespective of
the specific performances being compared. At this

stage the whole process becomes relatively
independent of external reinforcement and the
specific contingencies of the original training
situations, but it remains dependent upon

cognitive evaluations based on the match between
self-prescribed standards, performance, and the

attainments of reference models. Social
comparison processes become involved because in
the case of most performances objective criteria of
adequacy are lacking; hence the attainments of

other persons must be utilized as the norm against
which meaningful self-evaluation can be made.
Under naturally occurring conditions modeling
and reinforcement practices often operate

concurrently in ways that either supplement or
counteract each other. Findings of research in
which both of these sources of influence are varied
simultaneously (McMains & Liebert, 1968; Mischel
& Liebert, 1966; Rosenhan, Frederick, & Burrowes,
1968) show that self-rewards are most sparingly
administered when stringent performance
standards have been consistently modeled and
imposed, whereas social- learning conditions in

which persons both model and reinforce lenient
behavioral demands produce generous self-
reward patterns of behavior. In everyday life
people frequently model the very behavior they
decry in others. Discrepant practices in which

models prescribe stringent standards for others
but impose lenient ones upon themselves, or
impose austere demands on themselves and
lenient ones on others, reduce the likelihood that
high norms will be internalized.
Of particular relevance to self-regulatory
processes is evidence that self-monitored

reinforcement can, in fact, maintain behavior. To
test the relative efficacy of self-monitored and
externally imposed systems of reinforcement,
Bandura & Perloff (1967) conducted an
experiment in the following manner: Children
worked at a task in which they could achieve
progressively higher scores by performing

increasingly more effortful responses. Children in
the self-reinforcement condition selected their
own achievement standards and rewarded
themselves whenever they attained their self-
prescribed norms. Children assigned to an

externally imposed reinforcement condition were
matched with the self-reward group so that the
same performance standard was set for them and
the reinforcers were automatically delivered
whenever they reached the predetermined level.

To ascertain whether subjects’ behavioral
productivity was due to the operation of

contingent reinforcement or to gratitude for the
rewards that were made available, children in an
incentive-control group performed the task after

they had received the supply of rewards on a
noncontingent basis. A fourth group worked
without any incentives to estimate the amount of
behavior produced by the properties of the task

itself. Because the capacity to maintain effortful
behavior over time is the most important attribute

of a reinforcement operation, the dependent
measure was the number of responses the
children performed until they no longer wished to
continue the activity.
As shown graphically in Figure 1-2, both self-
monitored and externally imposed reinforcement
systems sustained substantially more behavior
than did either the noncontingent reward or the

nonreward condition, which did not differ from
each other. Of even greater interest is the
prevalence with which children in the self-
monitored condition imposed upon themselves
highly unfavorable schedules of reinforcement.
Not a single child chose the lowest score which
Figure 1-2. Behavioral productivity of children under
conditions in which their responses were self-
reinforced or externally reinforced, or in which they
were rewarded noncontingently or not at all. Bandura
& Perloff, 1967.
required the least effort, while approximately half
of them selected the highest achievement level as
the performance meriting self-reward. Moreover,

a third of the children subsequently altered their
initial standard to a higher level, without a
commensurate increase in amount of self-reward,
thereby imposing upon themselves a more
unfavorable ratio of work to reinforcement. This

behavior is all the more striking because the self-
imposition of stringent performance demands
occurred in the absence of any social surveillance
and under high permissiveness for self-reward.
It can be reasonably assumed that most older

children have acquired standards of achievement
through modeling and differential reinforcement

and have undergone experiences in which
rewarding oneself for performances judged to be
unworthy has been socially disapproved. Hence,
under conditions where persons are provided
with ample opportunities to optimize their
material outcomes by engaging in behavior which
has low self-regard value, strongly conflicting

tendencies are likely to be aroused. On the one
hand, individuals are tempted to maximize
rewards at minimum costs of effort to themselves,
but on the other hand, low quality performances
produce negative self-evaluative consequences

which, if sufficiently strong, may inhibit
undeserving self-compensation. Indeed, many of
the children in the experiment set themselves
performance requirements that incurred high
effort costs at minimum material recompense.

These findings are at variance with what one
might expect on the basis of reward-cost theories,

unless these formulations include the self-esteem
costs of rewarding devalued behavior.
After a self-monitored reinforcement system
has been well established, a given performance
produces two sets of consequences—a self-
evaluative reaction as well as some external
outcome. In many instances self-generated and

externally occurring consequences may conflict, as
when certain courses of action are approved and
encouraged by others, but if carried out would
give rise to self-critical and negative self-
evaluative reactions. Under these circumstances,

the effects of self-reinforcement may prevail over
external influences. Conversely, response patterns
may be effectively maintained by self-
reinforcement operations under conditions of
minimal external support. It is perhaps because of

the stabilizing effects of self-reinforcement that
persons do not ordinarily behave like

weathervanes in the face of conflicting
contingencies of reinforcement which they
repeatedly encounter in their social environment.

The fact that self-reinforcement may substitute
for, supplement, or override the effects of
externally occurring outcomes (Kanfer, 1968)
complicates interpretation of behavioral changes

supposedly due to external reinforcement.
Discussions of psychopathology generally

emphasize deficit conditions, response inhibitions,
and avoidance mechanisms. However, personal
problems frequently result from dysfunctions in

self-reinforcement systems. Many of the people
who seek treatment are neither incompetent nor
anxiously inhibited, but they experience a great
deal of personal distress stemming from
excessively high standards for self-evaluation,

often supported by unfavorable comparisons with
models noted for their extraordinary
achievements. This process typically gives rise to
depressive reactions, to feelings of worthlessness
and lack of purposefulness, and to lessened
disposition to perform because of negative self-
generated consequences. In its more extreme
forms this problem is reflected in behaviors
designed to escape self-generated anguish through

alcoholism, grandiose ideation, unwillingness to
engage in activities that may have important self-
evaluative implications, and other forms of
avoidance behavior. The modification of self-
reinforcement patterns constitutes a principal

psychotherapeutic objective in conditions
involving burdensome self-demands.
Social behavior is usually regulated to some
extent by covert self-reinforcing operations which

rely upon symbolically generated consequences in
the form of self-commendation, esteem-enhancing
reactions, or self- deprecation. Persons who have

failed to develop self-monitoring reinforcement
systems, or those who make self-reward
contingent upon skillful performance of antisocial
behavior, require considerable social surveillance
to ensure that they do not transgress. Similarly,
individuals who set lax behavioral standards for
themselves are inclined to display low

achievement behavior and a liberal self-
gratification pattern of life.
Symbolic Regulation of Behavior
Some psychological theories, while
acknowledging that stimulus-response

covariations are mediated by covert events,
nevertheless adhere rigorously to causal
explanations of behavior couched almost

exclusively in terms of external manipulable
variables. The pursuit of external causes rests on
the basic assumption that covert processes are
lawfully determined by externally occurring

events and, therefore, they can be bypassed in the
prediction and control of behavior. This view has
been advocated most forcefully by Skinner (1953):
“The objection to inner states is not that they do
not exist, but that they are not relevant in a
functional analysis. We cannot account for the

behavior of any system while staying wholly
inside it; eventually we must turn to forces
operating upon the organism from without [p.
35].”
The common practice of invoking spurious

inner states or agents as determiners of behavior
has also produced justifiable wariness of
inferential variables. After a given response
pattern has been attributed to the action of a

psychic homunculus, the search for controlling
conditions promptly ceases. Although the use of
the more colorful animistic entities in explanatory

schemes is declining, the tendency to offer new
descriptive labels for behavioral phenomena in the
guise of explanations remains a flourishing
practice.
The relative neglect of experiential phenomena
results primarily from their limited accessibility.
Thought processes are directly accessible only to

the person within whom they occur and therefore
their presence, absence, and exact nature cannot

be independently verified. As a consequence, one
is forced to rely upon verbal self-reports and other
indirect indices of events occurring at a private
level. In discussions of the methodological

problems and theoretical issues regarding
symbolic processes it is customary to belabor the
limitations and inaccuracies of self-reports. It is

emphasized that, due to defective self-descriptive
facility and various distorting influences, public

and private events may be imperfectly correlated.
Not only are private events difficult to identify, but
since they cannot be directly manipulated they

have limited value in the causal analysis or
practical control of behavior. These dissuading
arguments, however, never cite the innumerable
studies demonstrating that, under many
conditions, self-described covert events have

much greater predictive power and regulatory
influence over behavior than the externally

manipulated variables typically assigned the
central explanatory role in change processes.
There exists ample evidence that one cannot

account satisfactorily for human behavior while
remaining entirely outside the organism, because
overt behavior is often governed by self-generated
stimulation that is relatively independent of
environmental stimulus events. For purposes of

illustration, let us consider an experiment
conducted by Miller (1951) to demonstrate how
emotional responding can be brought under
thought control. Students were asked to
pronounce aloud the symbols of T and 4 as they
were presented in a random sequence. The
utterance T was consistently followed by shock
stimulation, whereas the 4 was never shocked.
After the discrimination had been established,

subjects were presented with a series of dots and
instructed to think T to the first dot, 4 to the
second one, and so on in an alternating sequence.
Subjects displayed a highly discriminative pattern
of autonomic responses with thoughts of T

eliciting large autonomic responses and thoughts
of 4 producing virtually no reaction. This
discriminative responding cannot be accounted
for in terms of the properties of the external dot
stimuli, which are identical and merely signal the

occasions for self-generated cognitive activities
that produce emotional responsiveness. In fact,

the trivial function of external stimuli could be
entirely eliminated simply by instructing subjects
to generate the aversive and neutral thoughts in

an unpredictable sequence and to signal by a key
press whichever cognitive event they were about
to produce. Knowledge of the subject’s pattern of
self-generated thoughts would permit accurate

prediction of his autonomic responses. In
naturalistic situations a brief external stimulus

often initiates a long chain of cognitive activities
that is largely determined by mediational
associative linkages rather than by the temporally
remote environmental input.
Under conditions where thought processes
essentially serve as the first link in causal
sequences, one can predict behavior most
accurately on the basis of subject-defined internal

stimulation. Until instruments that can
discriminate subtle differences between symbolic
events are developed, a comprehensive approach
to the understanding of human behavior will have
to rely upon an individual both as the agent and
the object of study. Most current experimentation
simply avoids the issues of internal stimulus
control by confining research to behavioral
phenomena that can be brought under the

influence of physical properties of external stimuli.
In a paper devoted to the control of implicit
events Homme (1965) indicates that the problems
of covert response definition and detection have
been needlessly exaggerated. He rightfully

contends that under most conditions the presence
or absence of covert activities can be easily
detected by the person in whom they are
occurring. As will be shown in the concluding

chapter, persons can not only reliably discriminate
internal events, but they can manipulate them by
making self-reinforcement contingent upon their

occurrence. Furthermore, thought-induced
affective reactions may be successfully employed
for purposes of controlling one’s own overt
behavior. In the above instances implicit activities
constitute either important phenomena in their
own right or causal antecedents rather than mere
internal accompaniments of behavioral and

environmental events.
There are innumerable psychological
processes in which internal mediating events must
occur before external stimuli will exercise control
over overt performances. Verbal mediators, in the

form of self-instructions, implicit categorizing
responses, or linkages through common word
associates, are perhaps the most prevalent
symbolic regulators of behavior. Persons must

often rely on verbal self-control when external
stimuli for correct responses are absent (Bern,
1967; Luria, 1961). Also, in many forms of

conceptual behavior or in semantic generalization
persons display a common response to highly
dissimilar stimuli (e.g., artichokes, strawberries,
lobsters, onion soup, leg of lamb, rye bread, wine,
and chocolate soufflé). Performance under these
conditions is governed by a mediating rule or a
common verbally labeled attribute (healthful

edibles) rather than by the physical characteristics
of the external stimuli alone (Bourne, 1966). In a
nonmediational account of conceptual behavior,
Ferster (1968) equates conceptualization with
abstract stimulus control whereby, through

selective reinforcement, a common property of
diverse complex stimuli comes to control the
response. The view is advanced that “the term
abstract stimulus control is somewhat preferable
to concept formation because it emphasizes the

controlling properties of the stimulus rather than
an inner and unreachable process [p. 404].” The

limitations of this type of approach become readily
apparent in cases, such as the one cited above,
where different stimuli have no physical property

in common but must be categorized on the basis of
a symbolically labeled attribute.
In most higher-level functioning, the implicit

rules regulating behavior cannot be defined solely
in terms of stimulus properties or combinations of
stimulus elements. In an experiment conducted by

Sassenrath (1962), for example, students were
presented with a series of words of various
lengths, to which they were required to respond

with correct numbers that could be consistently
produced only by recourse to a complicated but
unspecified code. The principle upon which
reinforcement was administered consisted of 11
minus the number of letters in the stimulus word,

so that correct responding had to be determined
by symbolic transformations of external stimuli.
Subjects eventually made accurate symbolic
transformations, which then became inner stimuli
for accurate responsiveness. The process of self-
reinforcement, in which persons self-administer
rewarding or punishing consequences on the basis
of implicit standards of conduct, is another
phenomenon involving internal rule-regulated

behavior.
Behavior may also be governed to some extent
by imagined mediators which represent previously
observed behavioral events and environmental
situations. It is exceedingly difficult to think about

the actions of people in given settings or features
of one’s physical environment without
experiencing corresponding visual imagery. The
highly influential role of symbolic processes in

behavioral change is most evident in vicarious or
observational learning (Bandura, 1965). The
paradigm utilized to study this phenomenon

involves a nonresponse acquisition procedure in
which a person merely observes a model’s
behavior but otherwise exhibits no overt
instrumental responses; nor is any reinforcing
stimuli administered during the acquisition
period. Exposure to modeling influences is an
exceedingly effective means of transmitting and

modifying conceptual and social behavior. Since in
this mode of response acquisition observers can
acquire only perceptual and other implicit
responses resembling the modeled patterns while
they are occurring, imaginal and verbal mediators

that govern subsequent response retrieval and
reproduction clearly play a prominent role in
observational learning.
There is a growing body of evidence (Bower,

1969; Paivio, 1969) that imaginal processes serve
a mediating function in facilitating verbal
associative learning. In these studies, imaginal

mediators are manipulated experimentally by
instructing subjects to link the members of each
pair of stimulus and response terms with a
distinctive image, and by using stimulus items that
vary in their capacity to evoke vivid imagery. The
findings demonstrate that during paired
presentations subjects code stimulus and response

events into mental images for memory
representation; later, the stimuli serve as cues that
reinstate the compound image from which the
response component is decoded to its original
verbal form. Imaginally mediated associative

learning is far superior to that in which this type of
representational process is minimally operative.
Some evidence exists to suggest that arousal
mediators may also exercise a regulatory function

over emotional behavior. According to the dual
process theory of avoidance behavior, stimuli
acquire, through their temporal conjunction with

aversive experiences, the capacity to produce
arousal reactions which have both central and
autonomic components. It is further assumed that
instrumental avoidance responses become partly
conditioned to arousal-correlated stimuli. The
most direct evidence that arousal mediators
operating primarily at the central level exercise

discriminative control over avoidance behavior is
provided by Solomon & Turner (1962). Animals
first learned to make an avoidance response to a
light stimulus. They were then skeletally
immobilized by curare to prevent avoidance

responses from being conditioned directly to
external stimuli; shock was paired with one tone,
while a contrasting tone was never associated
with aversive stimulation. In subsequent tests the
animals displayed essentially the same degree of

avoidance in response to the negatively valenced
tone and the light, both of which evoked common

arousal reactions, whereas avoidance responses
rarely occurred to the neutral tone. Considering
that the light and the tones were never associated,

and assuming that the curare blocked all skeletal
activity (Black, 1967), thus precluding any
differential conditioning of avoidance responses to
the tones, the controlling power of the negatively

valenced auditory stimulus must be mediated
through either events in the central nervous

system or autonomic feedback mechanisms.
Further demonstrations of internal regulation
of behavior are furnished by studies (Bailey, 1955;

Bailey & Porter, 1955; Levine, 1953), in which
infrahuman subjects must learn to respond
differentially on the basis of internal stimulation
associated with different drive states like thirst or
hunger because the environment contains no

distinguishable guiding cues. Under these
conditions the differential cues provided by
internal drive states, or even different intensities
of the same drive, give rise to dissimilar patterns
of behavior. These findings are consistent with
those cited earlier in which internal stimuli are
endowed with controlling properties through
interoceptive conditioning.
The powerful internal control of behavior is

most vividly illustrated in grossly deviant
behavior for which the controlling contingencies
are almost entirely symbolically generated. The
passage quoted below (Bateson, 1961), was taken
from a patient’s account of his psychosis long

before it was fashionable to write about one’s
psychiatric experiences. The narrator had received
a scrupulously moralistic upbringing according to
which even most socially approved patterns of

behavior were considered deviant, sinful, and
likely to provoke the wrath of God; consequently
many innocuous acts, such as accepting

medication, elicited dreadful apprehensions,
which, in turn, motivated and maintained
exceedingly painful atonement rituals designed to
forestall the imagined disastrous consequences.
In the night I awoke under the most dreadful
impressions, I heard a voice addressing me,
and I was made to imagine that my
disobedience to the faith, in taking the
medicine overnight, had not only offended
the Lord, but had rendered the work of my
salvation extremely difficult, by its effect
upon my spirits and humours. I heard that I
could only be saved now by being changed
into a spiritual body…A spirit came upon me
and prepared to guide me in my actions. I
was lying on my back, and the spirit seemed
to light on my pillow by my right ear, and to
command my body. I was placed in a
fatiguing attitude, resting on my feet, my
knees drawn up and on my head, and made
to swing my body from side to side without
ceasing. In the meantime, I heard voices
without and within me, and sounds as of the
clanking of iron, and the breathing of great
forge bellows, and the force of flames. I
understood that I was only saved by the
mercy of Jesus, from seeing, as well as
hearing, hell around me, and that if I were
not obedient to His spirit, I should inevitably
awake in hell before the morning. After some
time I had a little rest, and then, actuated by
the same spirit, I took a like position on the
floor, where I remained, until I understood
that the work of the Lord was perfected, and
that now my salvation was secured; at the
same time the guidance of the spirit left me,
and I became in doubt what next I was to do.
I understood that this provoked the Lord, as
if I was affecting ignorance when I knew
what I was to do, and, after some hesitation, I
heard the command, to “Take your position
on the floor again then,” but I had no
guidance or no perfect guidance to do so, and
could not resume it. I was told, however, that
my salvation depended upon my maintaining
that position as well as I could until the
morning; and oh! great was my joy when I
perceived the first brightness of the dawn,
which I could scarcely believe had arrived so
early [pp. 28-29],
The above quotation provides a clear example
of how behavior can come under the complete

control of fictional contingencies and fantasied
reinforcements powerful enough to override the
influence of the reinforcement contingencies
existing within the social environment. Thus the
acceptance of medicine, an act that was later
considered a rebellion against, and the mistrust of,
the Almighty, generated extremely aversive

hallucinations of hellish torture, the cessation of
which was contingent upon the performance of
arduous bizarre behavior. The nonoccurrence of
subjectively experienced, but objectively
nonexistent threats, undoubtedly serves as an

important mechanism maintaining many other
types of psychotic behavior. Given the conjunction
of fictional contingencies and a powerful internal
reinforcing system, a person’s behavior is likely to
remain under very poor environmental control

even in the face of severe externally administered
punishments and blatant disconfirming

experiences.
When I opened the door, I found a stout man

servant on the landing, who told me that he
was placed there to forbid my going out, by
the orders of Dr. P. and my friend; on my
remonstrating, he followed me into my room
and stood before the door. I insisted on going
out; he, on preventing me. I warned him of
the danger he incurred in opposing the will
of the Holy Spirit, I prayed him to let me
pass, or otherwise an evil would befall him,
for that I was a prophet of the Lord. He was
not a whit shaken by my address, so, after
again and again adjuring him, by the desire
of the Spirit whose word I heard, I seized one
of his arms, desiring to wither it; my words
were idle, no effect followed, and I was
ashamed and astonished.
Then, thought I, I have been made a fool of!

But I did not on that account mistrust the
doctrines by which I had been exposed to
this error. The doctrines, thought I, are true;
but I am mocked at by the Almighty for my
disobedience to them, and at the same time, I
have the guilt and the grief, of bringing
discredit upon the truth, by my obedience to
a spirit of mockery, or, by my disobedience
to the Holy Spirit; for there were not wanting
voices to suggest to me, that the reason why
the miracle had failed, was, that I had not
waited for the Spirit to guide my action when
the word was spoken and that I had seized
the man’s arm with the wrong hand…[p. 33],
The voices informed me, that my conduct

was owing to a spirit of mockery and
blasphemy having possession of me…that I
must, in the power of the Holy Spirit, redeem
myself, and rid myself of the spirits of
blasphemy and mockery that had taken
possession of me.
The way in which I was tempted to do this

was by throwing myself on the top of my
head backwards, and so resting on the top of
my head and on my feet alone, to turn from
one side to the other until I had broken my
neck. I suppose by this time I was already in
a state of feverish delirium, but my good
sense and prudence still refused to
undertake this strange action. I was then
accused of faithlessness and cowardice, of
fearing man more than God.
I attempted the command, the servant

prevented me. I lay down contented to have
proved myself willing to obey in spite of his
presence, but now I was accused of not
daring to wrestle with him unto blows. I
again attempted what I was enjoined. The
man seized me, I tore myself from him,
telling him it was necessary for my salvation;
he left me and went down stairs. I then tried
to perform what I had begun; but now I
found, either that I could not so jerk myself
round on my head, or that my fear of
breaking my neck was really too strong for
my faith. In that case I then certainly
mocked, for my efforts were not sincere.
Failing in my attempts, I was directed to

expectorate violently, in order to get rid of
my two formidable enemies; and then again I
was told to drink water, and the Almighty
was satisfied; but that I was not satisfied
(neither could I be sincerely, for I knew that I
had not fulfilled his commands), I was to
take up my position again; I did so, my
attendant came up with an assistant and
they forced me into a straight waistcoat.
Even then I again tried to resume the
position to which I was again challenged.
They then tied my legs to the bed-posts, and
so secured me [pp. 34-35].
The process of behavioral change will be
conceptualized quite differently depending upon

whether one assumes that responses are regulated
predominantly by external stimulus events or
partly by mediating symbolic events. In
nonmediational interpretations, learning is

depicted as a more or less automatic process
wherein stimuli become associated with overt
responses through differential reinforcement. By
contrast, in mediational formulations the learner

plays a far more active role and his responsiveness
is subject to extensive cognitive determination. On
the basis of salience of environmental events and
past learning experiences persons select the
stimuli to which they will respond; environmental

events are coded and organized for representation
in memory; provisional hypotheses regarding the

principles governing the occurrence of
reinforcement are derived from differential

consequences accompanying overt behavior; and
after a given implicit hypothesis has been

adequately confirmed by successful corresponding
actions, the mediating rules or principles serve to
guide the performance of appropriate responses
on future occasions. Relevant empirical evidence

bearing on these two theoretical approaches will
be reviewed in the concluding chapter of this

book.
It has been customary in psychological
theorizing to construct entire explanatory

schemes around a single form of behavioral
control, to the relative neglect of other obviously
influential variables and processes. Thus, for
example, some psychologists have tended to
concentrate upon stimulus control effected

principally through classical-conditioning
operations; Skinnerians have primarily focused
upon external reinforcement control of behavior;
and researchers favoring cognitive interpretations
have been most preoccupied with mediational
processes. These resolute allegiances to partial
processes are typically accompanied by some
disdain for variables patronized by out-group
theorists. A comprehensive theory of human

behavior must encompass all three sources of
behavioral regulation, i.e., stimulus control,
internal symbolic control, and outcome control. In
many situations, of course, two or more of these
processes may operate simultaneously in

governing responsiveness.
Social Learning as a Reciprocal Influence

Process
Psychodynamic theories of personality
typically depict the deviant actions of individuals

as being impelled by powerful internal forces that
they not only are unable to control, but whose
existence they do not even recognize. On the other

hand, behavioral formulations often characterize
response patterns as depending on environmental
contingencies. The environment is presented as a
more or less fixed property that impinges upon
individuals and to which their behavior eventually
adapts. Neither view of man is particularly

heartening nor entirely accurate.
Psychological functioning, in fact, involves a
continuous reciprocal interaction between
behavior and its controlling conditions. Although
actions are regulated by their consequences, the

controlling environment is, in turn, often
significantly altered by the behavior. Examples of
the way in which behavior shapes the
environment can be found even in simple

experiments with infrahuman subjects. As a means
of studying the acquisition of avoidance responses,
Sidman (1960, 1966) devised a paradigm in which

animals could postpone the occurrence of aversive
shocks by depressing a lever. Under these
conditions some animals created for themselves
an essentially punishment-free environment,
whereas others who, for one reason or another,
were slow in acquiring the requisite coping
response produced a highly aversive milieu. When

response changes are selected as the data for
analysis, as is almost invariably the case, then the
environmental contingencies appear to be fixed,
controlling conditions; if, instead, one analyzed the
data for the amount of aversive stimulation

created by each subject, then the environment
becomes the change worthy event that may vary
considerably for different subjects and at different
times for the same subject. Within the framework
of environmental analysis, one might, for instance,

administer alcohol to one group of subjects in the
Sidman paradigm and water to another, and then

compare the types of aversive environments
produced under intoxicated and sober conditions.
Interpersonal situations, of course, provide
much greater latitude for determining the
contingencies that maintain one’s behavior. In
social interchanges the behavior of one person
exerts some degree of control over the actions of

others. To take an example, counterreactions
drawn by hostile responses are likely to be quite
different from those elicited by friendly ones.
Experimental analysis by Rausch (1965) of
sequential interchanges between children, in fact,

reveals that the immediately preceding stimulus
act on the part of one person was the major
determinant of the other person’s response. In
approximately 75 percent of the instances, hostile
behavior elicited unfriendly responses, whereas

cordial antecedent acts seldom did. Aggressive
children thus created through their actions a

hostile environment, whereas children who
displayed friendly interpersonal modes of
response generated an amicable social milieu.

These findings demonstrate that persons, far from
being ruled by an imposing environment, play an
active role in constructing their own
reinforcement contingencies through their

characteristic modes of response. The theory of
social interaction advanced by Thibaut & Kelley

(1959) relies heavily upon mutual reinforcement
contingencies. Research stimulated by this
conceptualization provides numerous
demonstrations of how outcomes in dyadic

interchanges are jointly determined by the
behavior of both participants.
It might be argued that if each person partly
creates his own environment then there is no one

remaining to be influenced. This apparent paradox
overlooks the fact that reciprocity is rarely perfect,
since one’s behavior is not the sole determinant of
subsequent events. Furthermore, controlling and
controllable events usually occur in an alternating
pattern rather than concurrently until the
interaction sequence is terminated. The reciprocal
reinforcement process involved in the unwitting
production and strengthening of tantrum behavior

in children will serve to illustrate the latter point.
On most occasions children’s mild requests go
unheeded because the parent is preoccupied with
other activities. If subsequent bids also go
unrewarded the child will generally display

progressively more intense forms of behavior
which become increasingly aversive to the parent.
At this point in the interaction sequence the child
is exercising aversive control over the parent.
Eventually the parent is forced to terminate the

troublesome behavior by attending to the child,
thereby reinforcing obstreperous responsiveness.

Such differential reinforcement practices are
highly effective in producing aversive forms of
behavior of unusual resiliency. Some of the most

vivid examples of pernicious reciprocal control are
provided in Levy’s (1943) classic study of
childhood overdependency:
Patient (4 years, 9 months) rules the

household by his screaming and imperative
voice. Mother will always comply with his
demands rather than hear him scream.…The
patient is disobedient, hyperactive,
impudent to the parents; calls them names,
kicks and scratches when not given his own
way… [pp. 361-363].
In complete command, dominating mother

and sister, who yielded in every instance
rather than endure his scenes, a fourteen-
year-old refused to go to school. He lay in
bed, ordered his sister to get his breakfast,
bring his clothes, and struck her when she
disobeyed [p. 163],…
Mother states that he (10-year-old) was

spoiled by herself and maternal
grandmother, and later she gave in to his
demands for the sake of peace…Whenever
refused, he always commanded obedience by
screaming [pp. 383-384]…After screaming
no longer availed, he used the method of
nagging, monotonously repeating his
demands [p. 163],
The above case material illustrates how certain
reinforcement practices generate particular
behavior, which, due to its aversive properties, in

turn creates the very conditions likely to
perpetuate it. Thus while nature’s programming
ensured that people’s distress would not go

unheeded for long, it also provided the basis for
the establishment of socially disturbing response
patterns. Interpersonal difficulties are most likely

to arise under conditions where a person has
developed a narrow range of social responses
which periodically force reinforcing actions from

others through aversive control (e.g., nagging
complaints, aggressive behavior, helplessness,
sick-role behavior, and emotional expressions of

rejection, suffering, and distress, and other modes
of responding that command attention). The
treatment strategies are quite different depending
upon whether one views such behavior in terms of
its functional value in controlling the
responsiveness of others or as by-products of

intrapsychic disturbances. Deleterious reciprocal
processes can be best eliminated by withdrawing
the reinforcement supporting the deviant
behavior and by hastening the development of
more constructive means of securing desired

reactions from others.
It is only because there is some degree of self-
determination of outcomes that treatment of an
individual is justifiable. To the extent that newly

established patterns of behavior create favorable
reciprocally reinforcing processes, they will be
effectively sustained over time. However, in

instances where one person’s behavior exerts little
or no control over the actions of others, perhaps
from disparities in status or power, it may become
necessary to effect changes in other people
important to him, or in the social system itself.
Symptom Substitution
It is generally assumed by therapists who

subscribe to psychodynamic models that direct
modification of deviant behavior is likely to result
in “symptom substitution.” This issue, like others

pertaining to the development and treatment of
behavioral dysfunctions, has become hopelessly
muddled by the use of an inappropriate
conceptual scheme which thoroughly obscures the

very phenomena it is designed to elucidate It is
further obfuscated by partisan claims that no such

phenomena exist (Yates, 1958), and counterclaims
that symptom substitution not only occurs, but
that the commuted forms may endanger the very
life of ill-fated clients (Bookbinder, 1962).
Relevant outcome data cited later lead one to
suspect that prognostications of dire
consequences are intended more to dissuade
therapeutic innovation than to protect clients
welfare. Indeed, as Grossberg (1964) has noted,

much more serious from a humanitarian
standpoint is the failure of “depth”
psychotherapies to effect significant changes in
behavioral conditions that produce chronic
suffering and disheartening social and vocational

incapacitation.
The dispute about symptom substitution does
involve an important psychological phenomenon,
but little headway will be made in resolving this

issue as long as it is misconstrued as one of
symptomatic versus non- symptomatic treatment,
or modification of causal versus behavioral events.

Even if the concepts of symptom and mental
disease were pertinent to behavioral dysfunctions,
which they are not, the symptom substitution
hypothesis could never be satisfactorily tested
because it fails to specify precisely what
constitutes a “symptom,” when the substitution
should occur, the social conditions under which it

is most likely to arise, and the form that the
substitute symptom will take. If consensus could
ever be attained in devising an exhaustive list of
possible symptomatic behaviors, one would be
forced, in order to prove definitively that symptom

substitution does not occur, to conduct thorough
and repeated assessments of clients’ behavior for
an indefinite period. This exhaustive toil would
still be all for naught, since there exist no reliable
criteria for determining whether the occurrence of

so-called symptomatic behaviors after completion
of treatment represents emergent substitute by-

products of a psychic pathology, the development
of new modes of maladaptive response to
environmental pressures, or the persistence of old

modes of maladaptive behavior which had gone
unnoticed until even worse behavior was
eliminated.
The symptom substitution issue would never
have been cast in its present misleading form had
it been recognized that one cannot eliminate

behavior as such, except perhaps through direct
removal of requisite neurophysiological systems.
Response patterns can be modified only by

altering the stimulus conditions that regulate their
occurrence. Hence, all forms of psychotherapy,
regardless of their self-conferred honorific titles
and virtuous aims, effect behavioral changes
through either deliberate or unwitting

manipulation of controlling variables.
Psychodynamic and social-learning
approaches to psychotherapy are, therefore,
equally concerned with modifying the
“underlying” determinants of deviant response
patterns; however, these theories differ, often
radically, in what they regard these “causes” to be,
a crucial difference which in turn influences the

types of stimulus conditions favored in the
respective treatments. To take a simple but telling
example, in an effort to gain a better
understanding of some of the factors governing

deviant behavior likely to be labeled
“symptomatic,” Ayllon, Haughton, & Hughes
(1965) induced and sustained for a time a bizarre
broom-carrying response in an adult
schizophrenic by periodic positive reinforcement

of the behavior. A psychotherapist, who was
unaware of the conditions which had established

and maintained this response pattern, invoked the
following underlying causes:
Her constant and compulsive pacing, holding

a broom in the manner she does, could be
seen as a ritualistic procedure, a magical
action.…Her broom would be then: (1) a
child that gives her love and she gives him in
return her devotion, (2) a phallic symbol, (3)
the sceptre of an omnipotent queen…this is a
magical procedure in which the patient
carries out her wishes, expressed in a way
that is far beyond our solid, rational and
conventional way of thinking and acting [p.
3],
In treating the persistent display of bizarre and
apparently purposeless behavior this therapist, on
the basis of his causal explanation, would subject

the woman to extended interpretive probing of
her sexual conflicts and delusions of omnipotence.
On the other hand, the behavioral therapist,
viewing the rewarding outcomes as the major
determinant of the so-called psychotic symptom,
would alter the reinforcement contingency

governing the behavior. Indeed, when the
occasional rewards for carrying a broom were

completely withdrawn the “symptom” promptly
vanished and, according to a two-year follow-up
study, never reappeared.
In light of the above considerations, it would
be both more accurate and advantageous to

redefine the causal versus symptomatic treatment
controversy as being primarily concerned with the
question of whether a particular form of therapy

chooses to modify conditions that, in actuality,
exercise strong or weak or no significant control
over the behavior in question.
According to the social-learning point of view,

in the course of social development a person
acquires different modes of coping with

environmental stresses and demands. These
various response strategies form a hierarchy

ordered by their probability of effecting favorable
outcomes in certain situations. A particular mode

of responding may occupy a dominant position in
various hierarchies; subordinate strategies may
differ from one situation to another and may vary
widely in their frequency of occurrence relative
both to the dominant response tendencies and

among themselves. Consequently the effects of
removing a dominant response pattern will
depend upon the number of different areas of
functioning in which it is characteristically

activated, and the nature and relative strength of
the initially weaker response dispositions.
One can distinguish several different types of
treatment approaches that are likely to produce
small, unpredictable, or unenduring changes in
deviant behavior suggestive of “symptom
substitution.” A treatment that fails to alter the

major controlling conditions of the deviant
behavior will most certainly prove ineffective.

Similarly, a poorly designed program of therapy
aimed solely at eliminating maladaptive behavior

patterns does not in itself guarantee that desired
modes of behavior will ensue. This is particularly
true when removal of deviant behavior is brought
about through withdrawal of its usual positive

consequences or by punishment or the imposition
of external restraints.
In extinction treatment, as dominant response
tendencies are eliminated through nonreward, the
person will revert to alternative courses of action,

which have proved of some value in the past. If
these initially weaker forms are nondeviant and
are adequately reinforced, then deviant patterns
are likely to be abandoned in favor of the

competing alternatives without the emergence of
any negative characteristics. If, on the other hand,
the subordinate set of responses in the client’s

repertoire is for the most part unsatisfactory, the
therapist will be faced with the task of eliminating
a long succession of ineffective patterns of
response.
Response substitution is also likely to occur
when deviant behavior is eliminated not by
removal of its maintaining conditions but by

superimposing a competing set of controlling
variables (Bandura, 1962). Thus, for example,

antisocial behavior that serves as an effective
means of securing positive reinforcement may be
temporarily suppressed through severe
punishment. However, if the offender has learned

relatively few prosocial modes of behavior,
elimination of one deviant pattern will probably
be followed by another set of deviant responses

that are more successful in avoiding detection and
subsequent punishments. Moreover, the

suppressed behavior is likely to reappear in
situations where the probability of detection is
low, or the threat of punishment is weaker.
Successive substitution of deviant behavior
likewise readily arises under conditions where
defensive responses are either punished or
physically restrained without neutralizing the
aversive properties of subjectively threatening

situations. This process is well illustrated in
Miller’s (1948) classic study of avoidance
behavior. Animals were administered shocks in a
white compartment of a shuttle box and learned to
escape the painful stimulation by running through

an open door into a black compartment. The
formerly neutral white cues rapidly acquired
aversive properties, and the animals continued to
perform the avoidant running responses even
though the shock stimulation had been completely

discontinued. The animals were then placed in the
white compartment with the door closed to block

the running behavior. However, the door could be
released by rotating a wheel. Wheel-turning was
rapidly learned and maintained by fear reduction.

When conditions were further changed so that
wheel-turning no longer released the door, but the
animal could escape from the threatening
compartment by pressing a bar, the former

response was quickly discarded while the latter
became strongly established. Thus interventions

that eliminated avoidance responses without
reducing the arousal potential of conditioned
aversive stimuli merely produced new forms of
defensive behavior.
The preceding discussion has focused on
approaches which, if used as the sole method of
treatment, may eliminate one form of deviant
behavior but lead to a different one. The problem

of deviant response substitution, however, can be
easily forestalled by including in the original
treatment program procedures that effectively
remove the reinforcing conditions which sustain
deviant behavior and concurrently foster
desirable alternative modes of behavior. Such
treatment strategies, which will be fully reviewed
in succeeding chapters, not only produce enduring
changes in the selected direction, but may also set

in motion beneficial changes in related areas of
psychological functioning.
Efficacy of Conventional Methods of

Behavioral Change
A casual survey of contemporary methods of
behavioral change would disclose a multiplicity of

“schools” of approaches, each claiming respectable
improvement rates for their particular clientele. A
closer examination of these treatment approaches,
however, reveals that the apparently multifarious

systems represent essentially a single procedure:
they all utilize a social relationship and place
heavy reliance upon verbal interpretive methods

for inducing changes in social behavior. Moreover,
only a small range of persons exhibiting
behavioral deviations are actually treated, with
varying degrees of success, by interpretive
methods.
In the first place, most antisocial personalities,

who constitute a sizable proportion of the deviant
population, simply “serve time” in penal
institutions or remain under legal surveillance.
Since such persons generally prove unresponsive
to traditional techniques, many psychotherapists

have become pessimistic about the value of
psychotherapy for modifying “psychopathic” or
antisocially deviant behavior. In the case of
younger delinquents, correctional institutions,

though often providing a more structured and
nonpunitive environment than the children have
formerly experienced, rarely offer systematic

programs that are efficacious in producing
enduring behavioral and attitudinal changes.
Similarly, most persons exhibiting gross
behavioral dysfunctions, who also derive
relatively little benefit from conventional
interview approaches, are provided mainly with
medication, “occupational therapy” in the form of

carrying out institutional routines, recreational
activities and custodial care in “mental”
institutions, where they become intermittent or
permanent residents. Indeed, the least socially
responsive psychotics are customarily assigned to

essentially custodial wards where they receive
only medication and where they mutually
extinguish one another’s limited social behaviors.
Nor have conventional methods of behavioral
change had much beneficial impact upon the

widespread problems of alcoholism, drug
addiction, and a host of other major social

problems which, in some instances, require
modification of social systems rather than the
behavior of isolated individuals.
Even in the restricted sample of persons who
consult psychotherapists and are accepted for
treatment, the dropout rates and the estimates of
behavioral change for those who remain in

treatment give little cause for complacency.
Between 30 and 60 percent of this highly selected
group (diagnosed predominantly as neurotic and
excluding grossly psychotic, alcoholic, antisocial,
and neurologically involved cases), terminate

treatment against the advice of their therapists
after several initial interviews (Frank, Gliedman,
Imber, Nash, & Stone, 1957; Garfield & Kurz, 1952;
Imber, Nash, & Stone, 1955; Kirtner & Cartwright,
1958; Knight, 1941; Kurland, 1956; Mensh &

Golden, 1951; Rickles, Klein, & Bassan, 1950). Of
those clients who continue in the therapy

programs, irrespective of the type of treatment
administered, approximately two-thirds are
usually rated as exhibiting some degree of

improvement (Appel, Lhamon, Myers, & Harvey,
1951; Eysenck, 1952; Frank et al., 1957; Kirtner &
Cartwright, 1958; Zubin, 1953). Although the
above figures are based on studies of adults, there

is little reason to believe that the picture is very
different in the case of children (Levitt, 1963).
CRITERIA OF CHANGE
The two-thirds improved figure, which has

been widely and uncritically accepted as the
typical base rate of change accompanying
interview therapies, overestimates the amount of
benefit that people actually derive from such
treatment. The criteria upon which judgments of

therapeutic efficacy are usually based leave much
to be desired. In many instances psychotherapists’
global impressions of their results serve as the
major indicants of outcome. Considering that such
ratings reflect upon therapists’ professional

competence, it is reasonable to assume that
therapists do not underrate the therapeutic value
of their methods.
Projective tests and personality questionnaires

have also been extensively employed as the
principal measures for evaluating psychotherapy.
Their widespread popularity is probably more

attributable to their availability and ease of
administration and scoring than to their direct
relevance to types of psychological changes that

clients hope to achieve by undergoing
psychotherapy. If the proverbial Martian were to
review the therapy outcome literature he would
undoubtedly conclude that earth men embark
upon expensive and time-consuming programs of

treatment to effect modifications in their
Rorschach, TAT, or MMPI responses, rather than

to overcome behavioral inhibitions, to resolve
chronic interpersonal problems, to gain control

over alcoholism, or otherwise to enhance their
level of social functioning. Since the behavioral
correlates of these personality test measures are
considerably in doubt (Mischel, 1968), evidence

that test responses have changed is of limited
value in judging the relative success of given
approaches to treatment. This is particularly true
in view of the fact that responses to personality
tests are readily amenable to response-set biases,

to implicit expectations inherent in the setting,
and to other extraneous influences.
A third course for the evaluation of
psychotherapeutic efficacy, in vogue for a long

time, focuses on changes in clients’ verbal
behavior in interview situations. Dedicated
researchers have devoted literally thousands of

arduous hours to counting the frequency of
clients’ self-reference statements, affective
verbalizations, resistive comments, self-
exploratory remarks, type-token ratios, and a host
of other verbal contents. Although this approach
yields readily quantifiable data that possess some
face validity, there is little evidence that changes

observed in clients’ verbal behavior influence
appreciably their daily interpersonal
responsiveness. These verbal indices are,
therefore, more pertinent to evaluating verbal
conditioning than fundamental behavioral change

processes.
Inasmuch as persons typically seek the help of
psychotherapists in order to modify faulty
interpersonal modes of responding and the

adverse consequences these engender, it is
remarkable that until recently behavioral changes
as a measure of success had not only been

seriously neglected, but often derogated as
superficial. Indeed, there exists no other avowedly
humanitarian enterprise in which clients’ major
concerns are so cavalierly disregarded. Whatever
personality changes a psychotherapist may choose
to promote, they should be considered of dubious
value if they are not reflected in the client’s social

behavior. To take an analogous example, medical
treatments that, on the basis of physicians’
impressions and other ambiguous indicants,
supposedly effected profound physiological
changes but, in actuality, produced no evident

changes in clients’ suffering and physical
dysfunctions, would be summarily dismissed as
both ineffectual and misleading. Clearly, objective
measures of changes in behavior constitute the
most stringent and the most important criteria of

the power of a given treatment method. Since the
areas of functioning that require modification may

differ extensively from person to person, global,
all-purpose measures of change must be replaced
by behavioral criteria that are specific and

individually tailored to the treatment objectives
selected by the client (Pascal & Zax, 1956).
Findings of comparative studies utilizing indices of
improvement based on behavioral change

(Fairweather, 1964; Lazarus, 1961; Paul, 1966)
yield success rates that are substantially below the

legendary two-thirds improved figure customarily
quoted for interview therapies.
Moreover, improvement figures usually

present a misleading picture of the effectiveness of
interview methods because dropouts have been
invariably excluded from statistical analyses.
When a particular procedure yields a relatively
high attrition rate, discarding terminators in

assessing psychotherapy becomes especially
critical. Let us assume, for instance, that of 100
persons who entered treatment, 80 withdrew
after several initial interviews, while all of the
remaining 20 cases exhibited significant
improvement. If terminators are ignored the
treatment proves to be 100 percent effective
when, in fact, only 20 percent of the cases have
been benefited. It will be recalled that a sizable

percentage of clients who enter into interview
treatments terminate after a few visits.
IMPROVEMENT RATES FOR NONTREATED CASES
In order to demonstrate that psychotherapy is
a condition that contributes to observed outcomes,

it is necessary to compare changes exhibited by
clients who have undergone treatment with those
of a comparable group of nontreated cases. Such a

comparison group is essential in order to provide
an estimate of the influence of concomitant
extratherapeutic experiences that may contribute
importantly to demonstrable changes in clients’
behavior. Assuming that the two groups are
reasonably well matched on relevant variables,
any differential change between treated and
nontreated cases can thus be regarded as
therapeutically induced. There are relatively few
studies of psychotherapeutic outcomes that meet

the minimum requirements of an adequate control
group and clear specification and objective
measurement of outcomes.
Bergin (1966) has reviewed findings of seven

studies (that met the minimal requirements of a
two-group design and some measures of change)

in which outcomes from a treated group and a
comparable nontreated group of clients were
compared. All seven studies, involving diverse
forms of therapy and diverse criteria, show that
persons who have undergone psychotherapy do

not differ significantly in average amount of
change from nontreated controls, but treatment

generally produces more variable effects. Whereas
controls do not change or improve to some extent,

those who have received treatment either remain
unchanged, benefit somewhat, attain considerable
improvement, or become worse. Lest these
variance differences temporarily revive interest in

weak behavioral change methods, it should be
noted that treatment-induced effects are less
favorable and, hence, less variable when more
stringent and socially meaningful measures are
employed. This is well illustrated by results of an

investigation conducted by Rogers (1967) and his
collaborators on the efficacy of client-centered
therapy.
Schizophrenics were administered a battery of

tests including the Rorschach, MMPI, Thematic
Apperception Test, Wechsler Intelligence Scale,
Anxiety Reaction Scales, Stroop Tests, F

Authoritarian Scale, Q-Sort, and Wittenborn
Psychiatric Rating Scales. One group of
schizophrenics participated in intensive client-
centered treatment with highly qualified
therapists, whereas matched controls received no
therapy. After completion of the treatment phase
the test battery was readministered and two

clinical psychologists made global judgments,
principally from the Rorschach and the MMPI, of
the degree of change in patients’ levels of
psychological functioning. Treated and nontreated
groups did not differ in mean improvement,

although some of the patients who received
treatment, unlike the controls, showed somewhat
larger gains while others displayed a change for
the worse. In an effort to account for this
variability, therapists’ behavior was rated from

tape-recorded samples of their interviews for
positive regard, empathy, and genuineness. Except

for scores on one scale of the MMPI test, patients
receiving high levels of the supposedly therapeutic
conditions did not differ significantly from

patients whose therapists displayed low positive
responsiveness or from nontreated controls in
self-concepts, intellectual functioning, ratings of
their behavior on the hospital ward, and global

assessments based on various personality tests. It
would seem from the overall pattern of results

that a hospitalized patient has little to gain from
undergoing client- centered treatment and may, in
fact, suffer some slight losses if his therapist
happens to be lacking in amiability.
Faced with growing evidence that interview
therapies have limited efficacy, some researchers
concluded that outcome studies should be held in
abeyance while intensified efforts are made to

elucidate the process underlying these
procedures. Outcome studies were therefore
promptly downgraded, investigators became
absorbed in minute analyses of verbal
interchanges between therapists and their clients
and, in the absence of any promising alternatives,
the traditional practices not only survived
essentially unaltered but were professionally
sanctified. The possibility that a conversational

approach to the modification of deviant behavior
is inherently too weak to justify exhaustive
process studies was rarely entertained. Under
conditions where a given treatment procedure
exercises weak behavioral control many other

extraneous variables (e.g., personality
characteristics of therapists, social attributes of
clients, minor technical variations in procedures)
singly or in combination will emerge as
determinants of change. Rather than pursue these

limiting factors, it would be far more profitable to
devise new methods that are sufficiently powerful

to override their influences. If similar errors in
research strategy are to be avoided in the
development of new treatment approaches it is

essential to establish the relative superiority of a
particular approach before undertaking intricate
process studies that might elucidate underlying
mechanisms or suggest further procedural

refinements. It is also necessary to select stringent
and unambiguous criteria of change so as to

establish precisely what a given treatment method
can or cannot accomplish.
Multiprocesses Governing Behavioral Changes.

Evaluation of psychological procedures is often
unnecessarily obscured by the use of concepts
such as “cure,” “spontaneous remission,” and
“relapse,” which may be appropriate in describing
the course of physical disease processes but are

misleading when applied to behavioral changes
that are governed by social variables. In the latter
case, the pertinent issues of concern are whether a
given set of conditions can successfully induce a
change in behavior, whether the established
changes generalize to extratherapeutic situations,
and whether the changes are maintained over
time. Since these phenomena are fundamentally
different from disease processes they require a

separate and more fitting conceptual scheme.
Thus if a primary malignant tumor has been
surgically removed, it is reasonable to speak of
cures and of possible relapses, since cancerous
cells may not have been completely extirpated. By

contrast, deviant behavior cannot be eradicated by
the removal of a global internal determinant;
rather, the occurrence of deviant behavior is
extensively controlled by its likely consequences,
and may therefore vary considerably in different

environmental settings, toward different persons,
and at different times. This would be analogous to

having malignancies appear in a given person
under one set of social circumstances and
disappear under others.
Unlike physical therapies, in the appraisal of
psychological methods it is important to
distinguish among the induction, generalization
and maintenance of behavior, because these

processes are governed by somewhat different
variables. The fact that established changes may
no longer be evident some time after treatment
has been discontinued does not necessarily mean
that the method is inadequate. On the contrary, it

may be exceedingly powerful for inducing
changes, but the gains may prove short-lived
because the proper maintaining conditions have
not been arranged. Similarly, in some cases
enduring behavioral changes are achieved, but

they do not transfer to extratherapeutic situations,
thus requiring supplementary procedures to

ensure optimal transfer effects. Outcome studies
should therefore be designed to provide
unconfounded data regarding the magnitude,

generality, and durability of outcomes associated
with given treatment approaches.
PSYCHOTHERAPY AND FRIENDSHIP EXPERIENCES
It would appear from the absence of
differential improvement rates for treated and

nontreated groups that favorable behavioral
changes, when they do occur, must be produced by
factors that are unrelated to the special methods

that are rigorously applied by psychotherapists. It
is therefore not surprising to find that intensive
specialized training and experience in traditional
psychotherapeutic procedures may not only fail to
increase the incidence of favorable outcomes but

may in some instances interfere with the
establishment of social-learning conditions likely
to foster beneficial changes. Poser (1966), in a
bold research project, compared modifications in
the psychological functioning of psychotic patients

who received either five months of group
psychotherapy by psychiatrists and psychiatric
social workers, group discussions with
undergraduate students, or no special treatment.

The undergraduates, who responded to an
advertisement for summer employment, were
selected without any additional requisites, they
had no training or experience in psychotherapy,

and they were given no suggestion as to how they
should conduct their sessions. Patients seen by the
undergraduates displayed greater gains than both
the controls or cases treated by professional
psychotherapists; the latter two groups did not

differ much from each other. Rioch and her
associates (Rioch, Elkes, Flint, Usdansky, Newman,

& Silber, 1963) likewise found that selected
married women who received part-time practical

training over a two-year period in the application
of psychotherapeutic methods performed as well

as their professional counterparts. However, in
view of Poser’s findings, it would be essential to
study the comparative efficacy of a group of
untrained therapists in order to determine

whether the protracted instruction was irrelevant
to the outcomes achieved by the trained

housewives.
The question nevertheless remains why some
persons undergo changes and others do not,

whether or not they are involved in formal
therapy. Comparative investigations of the
attributes of clients who terminate treatment
prematurely with those of clients who remain and
improve are particularly relevant in this respect.

Relative to persons who continue in treatment,
terminators typically come from lower
socioeconomic levels, are nonconforming toward
authority figures, are impulsive, relatively non-
anxious, report a history of antisocial behavior,
present deficits in verbal and emotional
responsiveness, exhibit a relative inability to
establish and to maintain social relationships, and
acknowledge little contingency between their own

behavior and the actions of others toward them.
By contrast, remainers generally come from
higher socioeconomic levels, are better educated,
are willing to explore their personal problems, are
responsive to social reinforcement, are

suggestible, introspective, relatively anxious, self-
dissatisfied, and self-condemning (Auld & Myers,
1954; Frank et al. 1957; Hiler, 1954; Imber et al.,
1955; Katz, Lorr, & Rubinstein, 1958; Lorr, Katz, &
Rubinstein, 1958; McNair, Lorr, & Callahan, 1963;

Rubinstein & Lorr, 1956). Except for
socioeconomic and educational indices—which

generally correlate significantly with continuation
in treatment but tend to be unrelated to outcome
—most of the latter personality variables are

likewise predictive of rated subsequent
improvement in psychotherapy. Thus the type of
people who continue to participate and improve in
psychotherapy have attributes similar to those of

persons who, in laboratory studies of conformity,
attitude change, and conditionability, show more

responsiveness to almost any form of social
influence procedure (Berg & Bass, 1961; Biderman
& Zimmer, 1961; Janis & Hovland, 1959).
The above findings indicate that the social

characteristics of clients, rather than the chosen
psychotherapeutic method, are the main
determinants of the successes of traditional
psychotherapy. This may explain why, in spite of

wide conceptual divergences, all “schools” of
psychotherapy achieve very similar rates of
improvement (Appel et ah, 1951; Miles, Barrabee,
& Finesinger, 1951) and, although differences may
occasionally slightly favor the treated groups
(Frank, Gliedman, Imber, Stone, & Nash, 1959;
Leary & Harvey, 1956), the magnitude of
behavioral change exhibited by nontreated cases
is not substantially less than change in clients who

have undergone some traditional form of
psychotherapy (Bergin, 1966). The types of clients
who derive some benefit from participation in
conventional forms of psychotherapy are likely to
exhibit varying degrees of favorable improvement

with little or no formal treatment (Frank et al.,
1959; Saslow & Peters, 1956; Taylor, 1955). These
demonstrable changes are probably a function of
social-learning experiences resulting from casual
or more structured interpersonal interactions

with physicians, attorneys, clergymen, teachers,
close and respected friends, and other societal

agents who possess some degree of social power,
prestige, and good judgment. All these different
sources of social influence apparently rely

primarily upon common—though not the most
reliable or potent—therapeutic elements for the
modification of social behavior.
The overall outcome data accompanying
conversational treatment approaches indicate the
necessity for distinguishing between

psychotherapy on the one hand, and friendship
experiences on the other. In a thoughtful book
entitled Psychotherapy: The Purchase of Friendship,

Schofield (1964) contends that psychotherapists
are essentially offering their clients a supportive
substitute friendship which does not require
technical professional training. He further argues
that a wide range of persons within a society, by

virtue of their superordinated social roles, their
wisdom and devotion to service, are equally
capable of providing friendships and satisfying
discussions of personal concerns. Therefore,
individuals who are in need of an understanding
and trustworthy friend with whom they can
periodically share their problems, and those who
are searching for a faith or a commitment that
would add more purpose to their lives, might do

better to seek the counsel and emotional support
of respected colleagues and enlightened societal
agents rather than to flock to psychotherapists
whose training does not ensure special expertise
in the value domain.
It should be recognized that, although

thoughtful discussions in the context of a
supportive friendship can be highly meaningful
and satisfying, they generally have little impact on

persons’ specific behavioral difficulties. Few
chronic stutterers, for example, have been cured
through amity, introspective conversation, and

wise counsel. In modifying persistent deviant
behavior and in overcoming behavioral deficits,
friendship alone is not enough. Special learning
conditions must also be arranged and skillfully
implemented over a long period if desired
psychological changes are to be consistently
achieved and adequately maintained. The latter

activities, for which the label “psychotherapy” is
appropriate, require unique skills and specialized
procedures for effecting predictable behavioral
changes.
Recent years have witnessed a marked

proliferation of psychological ventures designed to
cure all types of social maladies. These endeavors
include, among other things, meditation, massage,
sensitivity training, and marathon social

encounters in which participants from all walks of
life are provided with opportunities to analyze
each other’s interpersonal reactions. As long as

such programs are not misrepresented and people
find them personally rewarding they require no
further validation. If, on the other hand, they are
marketed as forms of psychotherapy, then
advocates of such procedures must be concerned
about the consequences of their practices and they
must assume responsibility for empirical

verification of their claims. Moreover, ethical
considerations require that clients specify the
ways in which they wish to be changed, that the
intended outcomes of the therapeutic process be
made known, and that clients be informed of the

likelihood that the treatment interventions will
enable them to deal more effectively with the life
problems for which they seek help.
While psychotherapists are promoting their

favored insights in interview approaches they may
often simultaneously (if inadvertently) reward
their clients with approval for exhibiting desired

response patterns and show disapproval of
maladaptive forms; they may reduce anxieties
through their permissive and supportive reactions
toward clients’ disturbing self-revelations; and
they inevitably model various attitudes, values,
and interpersonal modes of behavior which clients
are inclined to emulate. Many of the therapeutic

changes that occur in conventional psychotherapy
may therefore derive primarily from the unwitting
application of social-learning principles. The point
is that these beneficial outcomes are more readily
attainable when principles are applied in a more

considered and systematic manner.
Even if the traditional forms of psychotherapy
had proved highly effective, they would still have
limited social value. A method that requires

extended and highly expensive training, that can
be performed only by professional personnel, that
must be continuously applied on a one-to-one

basis over a prolonged period of time, and is most
beneficial to self-selected highly suggestible
persons cannot possibly have much impact on the
countless social problems that demand
psychological attention. Major progress will be
made in resolving these problems by
concentrating on the development of highly

efficacious principles of behavioral change and by
utilizing the large pool of nonprofessional persons
who can be trained to implement programs under
competent guidance and direction. This approach
would provide more people with more help than

they receive under current professional practices.
APPROACHES BASED ON SOCIAL-LEARNING

PRINCIPLES
In subsequent chapters of this book various
social-learning approaches to the modification of
diverse psychological phenomena will be
considered in detail. The principles underlying
each method will be reviewed along with

experimental tests of their efficacy. In addition, the
types of behavioral changes for which each

procedure is best suited will be discussed.
Although major emphasis will be given to
psychological variables that have been shown to
exercise strong control over behavior, some

attention will be devoted to pharmacological
procedures, particularly when they are employed

as adjuncts to social-learning procedures. The
psychological emphasis, however, is not meant to
minimize the genetic, biochemical, and
neurophysiological determinants of behavior. A

social-learning model does not, of course, assume
that behavior is determined exclusively by
psychological variables. Genetic endowment and

constitutional factors may set certain limits on
both the types of behavioral repertoires that can

be developed in a given person, and the rate of
response acquisition. In certain cases,
neurophysiological conditions may contribute to

the observed behavioral malfunctioning.
Moreover, biological and psychological factors
typically interact in subtle and complex ways in
producing certain patterns of social behavior.
It should also be noted in passing that
physiological variables, to the extent that they
serve as contributory factors, are most likely to be

associated with nonspecific effects as reflected in
the general tempo of responsiveness and the rate
and level of response acquisition. Such variables

do not, however, determine specific behavioral
patterns, which are due to particular social-
learning experiences. Genetic endowment cannot
account for the difference between one
schizophrenic who firmly believes that he is Jesus

Christ, and another one who entertains no
grandiose delusions. The idiosyncratic behavioral
content is obviously learned rather than
physiologically produced. Nor do capacity
variables account for gross deficits in motor,
conceptual, or affective responses that are clearly
within a person’s capabilities. Unfortunately,
deviant behavior is often prematurely attributed
to physiological determinants, an attribution

which results not only in therapeutic pessimism,
but also effectively retards further psychological
investigation of behavioral phenomena.
Summary
This chapter has presented a social-learning
interpretation of the mechanisms regulating
behavior and contrasts this approach with

theories that tend to assign causal properties to
hypothetical internal forces. The differences in

conceptual models are especially striking in
explanations of deviant behavior that have
traditionally been depicted as symptomatic by-
products of a quasi-mental disease. From a social-
learning perspective, behaviors that may be
detrimental to the individual or that depart widely
from accepted social and ethical norms are
considered not as manifestations of an underlying
pathology but as ways, which the person has

learned, of coping with environmental and self-
imposed demands.
Psychopathology is not solely a property
inherent in behavior but reflects the evaluative
responses of societal agents to actions that violate

prescribed codes of conduct. The social labeling of
a given response pattern as a pathological
expression is, in fact, influenced by numerous
subjective criteria including the aversiveness of

the behavior, the social attributes of the deviator,
the normative standards of persons making the
judgments, the social context in which the

behavior is performed, and a host of other factors.
Consequently, the same response pattern may be
diagnosed as “sick” or may be normatively
sanctioned and considered emulative by different
groups, at different times, or in different
environmental settings. Considering the arbitrary
and relativistic nature of the social judgment and

definition of deviance, the main value of the
normal versus abnormal dichotomy lies in guiding
the social and legal actions of societal agents
concerned with the maintenance of an efficiently
functioning society. This dichotomy, however, has

little theoretical significance, because no evidence
exists that the behaviors so dichotomized are
either qualitatively different or are under the
control of fundamentally different variables.
Personality theories generally assume that

energized traits and concealed motivational states
impel behavior in a variety of directions. These

hypothetical internal conditions tend to be
regarded as relatively autonomous of external
stimulation and their relationship to behavior
remains somewhat loose. In social-learning theory
both deviant and prosocial behaviors are acquired
and maintained on the basis of three distinct
regulatory systems.
Some response patterns are primarily under
external stimulus control. Autonomic
responsiveness, such as changes in cardiovascular
and gastrointestinal reactions, and emotional
behavior, can be effectively brought under the

control of environmental events through their
contiguous association with either direct or
vicarious affective experiences. Instrumental
behavior is likewise precisely regulated by

environmental stimuli that, by virtue of their
association with different contingencies of
reinforcement, signify the consequences that are

likely to accompany certain courses of action.
Some forms of deviant behavior primarily reflect
defective or inappropriate stimulus control.
A second behavioral control system involves
response feedback processes, mainly in the form
of reinforcing consequences. Both prosocial and

grossly deviant behaviors can be successively
eliminated and reinstated by varying their

immediate consequences. These influential
aftereffects may include sensory experiences that
are intrinsically produced by the activity itself,
externally arranged tangible or symbolic

outcomes, or self-evaluative reactions. The
susceptibility of behavior to reinforcement control
is further shown by the fact that even subtle

variations in the frequency and patterning of
outcomes result in distinct performance

characteristics.
The third, and in many respects the most
influential, regulatory mechanism operates
through central mediational processes. At this
higher level stimulus inputs are coded and
organized; tentative hypotheses about the
principles governing the occurrence of rewards
and punishments are developed and tested on the

basis of differential consequences accompanying
the corresponding actions; and, once established,
implicit rules and strategies serve to guide
appropriate performances in specified situations.
Symbolically generated affective arousal and

covert self-reinforcing operations may also figure
prominently in the regulation of overt
responsiveness.
In this conceptual scheme man is neither an

internally impelled system nor a passive reactor to
external stimulation. Rather, psychological
functioning involves a reciprocal interaction

between behavior and its controlling environment.
The type of behavior that a person exhibits partly
determines his environmental contingencies
which, in turn, influence his behavior. In
succeeding chapters of this book the social-
learning principles necessary to account
adequately for the development of prosocial and

deviant behavior will be further elaborated. We
shall also demonstrate how these principles may
be successfully applied to ameliorate
developmental and clinical problems, and to effect
broader social and cultural change.
REFERENCES
Appel, K. E., Lhamon, T. W., Myers, J. M., & Harvey, W.

A. Long term psychotherapy. In Psychiatric
treatment. Baltimore: Williams & Wilkins, 1951.
pp. 21-34.
Auld, F., Jr., & Myers, J. K. Contributions to a theory for

selecting psychotherapy patients. Journal of
Clinical Psychology, 1954, 10, 56-60.
Ayllon, T., & Azrin, N. H. The measurement and

reinforcement of behavior of psychotics. Journal
of the Experimental Analysis of Behavior, 1965, 8,
357-383.
Ayllon, T., Haughton, E., & Hughes, H. B. Interpretation
of symptoms: Fact or fiction. Behaviour Research
and Therapy, 1965, 3, 1-7.
Ayllon, T., & Michael, J. The psychiatric nurse as a

behavioral engineer. Journal of the Experimental
Analysis of Behavior, 1959, 2, 323-334.
Bailey, C. J. The effectiveness of drives as cues. Journal

of Comparative and Physiological Psychology,
1955, 48, 183-187.
Bailey, C. J., & Porter, L. W. Relevant cues in drive

discrimination in cats. Journal of Comparative and
Physiological Psychology, 1955, 48, 180-181.
Bandura, A. Relationship of family patterns to child

behavior disorders. Progress Report, 1960,
Stanford University, Project No. M-1734, United
States Public Health Service.
Bandura, A. Punishment revisited. Journal of

Consulting Psychology, 1962, 26, 298-301.
Bandura, A. Vicarious processes: A case of no-trial

learning. In L. Berkowitz (Ed.), Advances in
experimental social psychology, Vol. II. New York:
Academic Press, 1965. pp. 1-55.
Bandura, A. A social learning interpretation of

psychological dysfunctions. In P. London & D.
Rosenhan (Eds.), Foundations of abnormal
psychology. New York: Holt, Rinehart & Winston,
1968. pp. 293-344.
Bandura, A., Blanchard, E. B., & Ritter, B. The relative

efficacy of desensitization and modeling
approaches for inducing behavioral, affective, and
attitudinal changes, Stanford University, 1969,
13, 173-199.
Bandura, A., Grusec, J. E., & Menlove, F. L. Some social

determinants of self-monitoring reinforcement
systems. Journal of Personality and Social
Psychology, 1967, 5, 449-455.
Bandura, A., & Kupers, C. J. Transmission of patterns

of self-reinforcement through modeling. Journal
of Abnormal and Social Psychology, 1964, 69, 1-9.
Bandura, A., & Perloff, B. Relative efficacy of self-

Psychology, 1967, 7, 111-116.
Bandura, A., & Rosenthal, T. L. Vicarious classical

conditioning as a function of arousal level. Journal
of Personality and Social Psychology, 1966, 3, 54-
62.
Bandura, A., Ross, D., & Ross, S. A. Vicarious

reinforcement and imitative learning. Journal of
Abnormal and Social Psychology, 1963, 67, 601-
607.
Bandura, A., & Walters, R. H. Adolescent aggression.

New York: Ronald Press, 1959.
Bandura, A., & Whalen, C. K. The influence of

antecedent reinforcement and divergent
modeling cues on patterns of self-reward. Journal
382.
Bateson, G. (Ed.) Perceval’s narrative: A patient’s

account of his psychosis. 1830—1832. Stanford,
Calif.: Stanford University Press, 1961.
Bern, S. L. Verbal self-control: The establishment of

effective self-instruction. Journal of Experimental
Psychology, 1967, 74, 485-491.
Berg, J. A., & Bass, B. M. (Eds.) Conformity and

deviation. New York: Harper, 1961.
Berger, S. M. Incidental learning through vicarious

reinforcement. Psychological Reports, 1961, 9,
477-491.
Berger, S. M. Conditioning through vicarious

instigation. Psychological Review, 1962, 69, 450-
466.
Bergin, A. E. Some implications of psychotherapy
research for therapeutic practice. Journal of
Abnormal Psychology, 1966, 71, 235-246.
Biderman, A. D., & Zimmer, H. The manipulation of

human behavior. New York: Wiley, 1961.
Bisese, V. S. Imitation behavior as a function of direct

and vicarious reinforcement. Dissertation
Abstracts, 1966, 26, 6155.
Black, A. H. Transfer following operant conditioning in

the curarized dog. Science, 1967,155, 201-203.
Bookbinder, L. J. Simple conditioning vs. the dynamic

approach to symptoms and symptoms
substitution: A reply to Yates. Psychological
Reports, 1962, 10, 71-77.
Bourne, L. E., Jr. Human conceptual behavior. Boston:

Allyn & Bacon, 1966.
Bower, G. H. Mental imagery and associative learning.

In L. Gregg (Ed.), Cognition in learning and
memory. New York: Wiley, 1969.
Bruning, J. L. Direct and vicarious effects of a shift in

magnitude of reward on performance. Journal of
Personality and Social Psychology, 1965, 2, 278-
282.
Buehler, R. E., Patterson, G. R., & Furniss, J. M. The
reinforcement of behavior in institutional
settings. Behaviour Research and Therapy, 1966,
4, 157-167.
Bykov, K. M. The cerebral cortex and the internal

organs. Translated and edited by W. H. Gantt.
New York: Chemical Publishing Co., 1957.
Church, R. M. Transmission of learned behavior

between rats. Journal of Abnormal and Social
Psychology, 1957, 54, 163-165.
Cloward, R. A., & Ohlin, L. E. Delinquency and

opportunity: A theory of delinquent gangs. New
York: Free Press, 1960.
Crankshaw, E. Another Russian “rebel” author put in

asylum. The Observer, February 17, 1963.
Davenport, J. W. Higher-order conditioning of fear.

Psychonomic Science, 1966, 4, 27-28.
Dekker, E., & Groen, J. Reproducible psychogenic

attacks of asthma: A laboratory study. Journal of
Psychosomatic Research, 1956, 1, 58—67.
Dekker, E., Pelser, H. E., & Groen, J. Conditioning as a

cause of asthmatic attacks. Journal of
Psychosomatic Research, 1957, 2, 97-108.
DiCara, L. V. & Miller, N. E. Instrumental learning of
vasomotor responses by rats: Learning to
respond differentially in the two ears. Science,
1968, 159, 1485-1486.
Eysenck, H. J. The effects of psychotherapy: An

evaluation. Journal of Consulting Psychology,
1952, 16, 319-325.
Fairweather, G. W. Social psychology in treating

mental illness. New York: Wiley, 1964.
Ferster, C. B. Behavior principles. New York: Appleton-

Century-Crofts, 1968.
Ferster, C. B., & Skinner, B. F. Schedules of

reinforcement. New York: Appleton-Century-
Crofts, 1957.
Frank, J. D., Gliedman, L. H., Imber, S. D., Nash, E. H., &

Stone, A. R. Why patients leave psychotherapy.
American Medical Association Archives of
Neurological Psychiatry, 1957, 77, 283-299.
Frank, J. D., Gliedman, L. H., Imber, S. D., Stone, A. R., &

Nash, E. H., Jr., Patients’ expectancies and
relearning as factors determining improvement
in psychotherapy. American Journal of Psychiatry,
1959, 115, 961-968.
Freud, S. The complete psychological works of ...
Translated by James Strachey. Vol. 10. London:
Hogarth, 1955.
Garfield, S. L., & Kurz, M. Evaluation of treatment and

related procedures in 1216 cases referred to a
mental hygiene clinic. Psychiatric Quarterly, 1952,
26, 414-24.
Gitelson, M. Communication from the president about

the neoanalytic movement. Journal of the
International Psycho-Analytic Association, 1962,
43, 373-375.
Grossberg, J. M. Behavior therapy: A review.

Psychological Bulletin, 1964, 62, 73-88.
Harris, F. R., Wolf, M. N., & Baer, D. M. Effects of adult

social reinforcement on child behavior. Young
Children, 1964, 20, 8-17.
Hastorf, A. H. The reinforcement of individual actions

in a group situation. In L. Krasner & L. P. Ullmann
(Eds.), Research in behavior modification. New,
York: Holt, Rinehart & Winston, 1965. pp. 268-
284.
Heine, R. W. A comparison of patients’ reports on

psychotherapeutic experience with
psychoanalytic, nondirective and Adlerian
therapists. American Journal of Psychotherapy,
1953, 7, 16-23.
Hiler, E. W. Psychological factors associated with

premature termination in psychotherapy.
Unpublished doctoral dissertation, University of
Michigan, 1954.
Homme, L. E. Perspectives in psychology: XXIV.

Control of coverants, the operants of the mind.
Psychological Record, 1965, 15, 501-511.
Imber, S. D., Nash, E. H., Jr., & Stone, A. R. Social class

and duration of psychotherapy. Journal of Clinical
Psychology, 1955, 11, 281-284.
Janis, I. L., & Hovland, C. I. (Eds.) Personality and

persuasibility. New Haven: Yale University Press,
1959.
John, E. R. Mechanisms of memory. New York:

Academic Press, 1967.
Kanfer, F. H. Vicarious human reinforcement: A

glimpse into the black box. In L. Krasner & L. P.
Ullmann (Eds.), Research in behavior modification.
New York: Holt, Rinehart & Winston, 1965. pp.
244-267.
Kanfer, F. H. Verbal conditioning: A review of its

current status. In T. R. Dixon & D. L. Horton
(Eds.), Verbal behavior and general behavior
theory. New York: Prentice-Hall, 1968. pp. 254-
290.
Kanfer, F. H., & Marston, A. R. Human reinforcement:

vicarious and direct. Journal of Experimental
Psychology, 1963, 65, 292-296.
Kanfer, F. H., & Marston, A. R. Conditioning of self-

reinforcing responses: An analogue to self-
confidence training. Psychological Reports, 1963,
13, 63-70. (b)
Katz, M. M., Lorr, M., & Rubinstein, E. A. Remainer

patient attributes and their relation to
subsequent improvement in psychotherapy.
Journal of Consulting Psychology, 1958, 22, 411-
413.
Kimble, G. A. Hilgard and Marquis’ conditioning and

learning. New York: Appleton-Century-Crofts,
1961.
Kimmel, H. D. Instrumental conditioning of

autonomically mediated behavior. Psychological
Bulletin, 1967, 67, 337—345.
Kirtner, W. L., & Cartwright, D. S. Success and failure

in client-centered therapy as a function of client
personality variables. Journal of Consulting
Psychology, 1958, 22, 259-264.
Knight, R. O. Evaluation of the results of
psychoanalytic therapy. American Journal of
Psychiatry. 1941, 98, 434-446.
Kurland, S. H. Length of treatment in a mental hygiene

clinic. Psychiatric Quarterly Supplement, 1956, 30,
83-90.
Lazarus, A. A. Group therapy of phobic disorders by

systematic desensitization. Journal of Abnormal
and Social Psychology, 1961, 63, 504-510.
Leary, T., & Harvey, J. S. A methodology for measuring

personality changes in psychotherapy. Journal of
Levine, S. The role of irrelevant drive stimuli in

learning. Journal of Experimental Psychology,
1953, 45, 410-416.
Levitt, E. E. Psychotherapy with children: A further

evaluation. Behaviour Research and Therapy,
1963, 1, 45-51.
Levy, D. M. Maternal overprotection. New York:

Columbia University Press, 1943.
Lidz, T., Cornelison, A. R., Terry, D., & Fleck, S.

Intrafamilial environment of the schizophrenic
patient: VI. The transmission of irrationality.
American Medical Association Archives of
Neurological Psychiatry, 1958, 79, 305-316.
Lidz, T., Fleck, S., & Cornelison, A. R. Schizophrenia and

the family. New York: International Universities
Press, 1965.
Liebert, R. M., & Allen, M. K. Effects of rule structure

and reward magnitude on the acquisition and
adoption of self-reward criteria. Psychological
Reports, 1967, 21, 445-452.
Lorr, M., Katz, M. M., & Rubinstein, E. A. The prediction

of length of stay in psychotherapy. Journal of
Luria, A. R. The role of speech in the regulation of

normal and abnormal behaviour. New York:
Pergamon Press, 1961.
McDavid, J. W. Effects of ambiguity of environmental

cues upon learning to imitate. Journal of
386.
McMains, M. J., & Liebert, R. M. Influence of

discrepancies between successively modeled self-
reward criteria on the adoption of a self-imposed
standard. Journal of Personality and Social
Psychology, 1968, 8, 166-171.
McNair, D. M., Lorr, M., & Callahan, D. M. Patient and
therapist influences on quitting psychotherapy.
Journal of Consulting Psychology, 1963, 27, 10-17.
Marlatt, G. A. Vicarious and direct reinforcement

control of verbal behavior in an interview setting.
Unpublished doctoral dissertation, Indiana
University, 1968.
Marston, A. R. Imitation, self-reinforcement, and

reinforcement of another person. Journal of
261. (a)
Marston, A. R. Self-reinforcement: The relevance of a

concept in analogue research in psychotherapy.
Psychotherapy: Theory, Research and Practice,
1965, 2, 1-5. (b)
Meehl, F. E. Clinical vs. statistical prediction.

Minneapolis: University of Minnesota Press,
1954.
Mensh, I. N., & Golden, J. M. Factors in

psychotherapeutic success. Journal of the
Missouri Medical Society. 1951, 48, 180-184.
Merton, R. K. Social theory and social structure. (Rev.

ed.) New York: Free Press, 1957.
Miles, H. W. H., Barrabee, E. L., & Finesinger, J. E.

Evaluation of psychotherapy. Psychosomatic
Medicine, 1951, 12, 83-105.
Miller, N. E. Studies of fear as an acquirable drive: I.

Fear as motivation and fear-reduction as
reinforcement in the learning of new responses.
Journal of Experimental Psychology, 1948, 38, 89-
101.
Miller, N. E. Learnable drives and rewards. In S. S.

Stevens (Ed.), Handbook of experimental
psychology. New York: Wiley, 1951. pp. 435-472.
Miller, N. E. Learning of visceral and glandular

responses. Science, 1969, 163, 434-445.
Mischel, W. Personality and assessment. New York:

Wiley, 1968.
Mischel, W., & Liebert, R. M. Effects of discrepancies

between observed and imposed reward criteria
on their acquisition and transmission. Journal of
Personality and Social Psychology, 1966, 3, 45-53.
Moore, N. Behaviour therapy in bronchial asthma: A

controlled study. Journal of Psychosomatic
Research, 1965, 9, 257-276.
Moser, D. Screams, slaps and love. Life, May 7, 1965,

90A-101.
Mowrer, O. H. Learning theory and personality
dynamics. New York: Ronald Press, 1950.
Noelpp, B., & Noelpp-Eschenhagen, I. Das

experimentelle Asthma bronchiale des
Meerschweinschens: II. Mitterlung die Rolle
bedingter Reflexes in des Pathogenese des
Asthma bronchiale. International Archives of
Allergy, 1951, 2, 321-329.
Noelpp, B., & Noelpp-Eschenhagen, I. Das

experimentelle Asthma bronchiale des
Meerschweinschens: III. Studien zur Bedeutung
bedingter Reflexe, Bahnungsbereitschaft und
Heftfahigkeit unter Stress. International Archives
of Allergy, 1952, 3, 108-135.
Ottenberg, P., Stein, M., Lewis, J., & Hamilton, C.

Learned asthma in the guinea pig. Psychosomatic
Medicine, 1958, 10, 395-400.
Paivio, A. Mental imagery in associative learning and

memory. Psychological Review, 1969, 76, 241-
263.
Pascal, G. R., & Zax, M. Psychotherapeutics: Success or

failure. Journal of Consulting Psychology, 1956, 20,
325-331.
Paschke, R. E., Simon, S., & Bell, R. W. Vicarious

discriminative learning in retardates. Journal of
Paul, G. L. Insight vs. desensitization in psychotherapy.

Stanford, Calif.: Stanford University Press, 1966.
Poser, E. G. The effect of therapists’ training on group

therapeutic outcome. Journal of Consulting
Psychology, 1966, 30, 28-289.
Rausch, H. L. Interaction sequences. Journal of

499.
Razran, G. The observable unconscious and the

inferable conscious in current soviet
psychophysiology. Psychological Review, 1961,
68, 81-147.
Rickles, N. K., Klein, J. J., & Bassan, M. E. Who goes to a

psychiatrist? American Journal of Psychiatry,
1950, 106, 845-850.
Rioch, M. J., Elkes, E., Flint, A. A., Usdansky, B. C.,

Newman, R. G., & Silber, E. National Institute of
Mental Health pilot study in training mental
health counselors. American Journal of
Orthopsychiatry, 1963, 33, 678-689.
Rogers, C. R. (Ed.) The therapeutic relationship and its

impact: a study of psychotherapy with
schizophrenics. Madison, Wis.: University of
Wisconsin Press, 1967.
Rosenbaum, M. E., & Bruning, J. L. Direct and vicarious

effects of variations in percentage of
reinforcement on performance. Child
Development, 1966, 37, 959-966.
Rosenhan, D., Frederick, F., & Burrowes, A. Preaching

and practicing: Effects of channel discrepancy on
norm internalization. Child Development, 1968,
39, 291-301.
Rubinstein, E. A., & Lorr, M. A comparison of

terminators and remainers in outpatient
psychotherapy. Journal of Clinical Psychology,
1956, 12, 345-349.
San Francisco Chronicle, November 26, 1964, p. 3.
Saslow, G., & Peters, A. D. A follow-up study of

“untreated” patients with various behavior
disorders. Psychiatric Quarterly, 1956, 30, 283-
302.
Sassenrath, J. M. Transfer of learning without

awareness. Psychological Reports, 1962, 10,411-
420.
Schofield, W. Psychotherapy: The purchase of

friendship. Englewood Cliffs, N.J.: Prentice-Hall,
1964.
Sidman, M. Tactics of scientific research. New York:
Basic Books, 1960.
Sidman, M. Avoidance behavior. In W. K. Honig (Ed.),

Operant behavior. New York: Appleton-Century-
Crofts, 1966, pp. 448-498.
Skinner, B. F. Science and human behavior. New York:

Macmillan, 1953.
Skinner, B. F. Cumulative record. New York: Appleton-

Slucki, H., Adam, G., & Porter, R. W. Operant

discrimination of an interoceptive stimulus in
rhesus monkeys. Journal of the Experimental
Solomon, R. L. & Turner, L. H. Discriminative classical

conditioning in dogs paralyzed by curare can
later control discriminative avoidance responses
in the normal state. Psychological Review, 1962,
69, 202-219.
Spohn, H. E. The influence of social values upon the

clinical judgments of psychotherapists. In J. G.
Peatman & E. L. Hartley (Eds.), Festschrift for
Gardner Murphy. New York: Harper, 1960. pp.
274-290.
Szasz, T. S. The myth of mental illness. New York:
Hoeber-Harper, 1961.
Szasz, T. S. Bootlegging humanistic values through

psychiatry. The Antioch Review, 1962, 22, 341-
349.
Szasz, T. S. Psychiatric justice. New York: Macmillan,

1965.
Taylor, D. M. Changes in self concept without

psychotherapy. Journal of Consulting Psychology,
1955, 19, 205-209.
Terrace, H. S. Stimulus control. In W. K. Honig (Ed.),

Crofts, 1966. pp. 271-344.
Terwilliger, J. S., & Fiedler, F. E. An investigation of

determinants inducing individuals to seek
personal counseling. Journal of Consulting
Psychology, 1958, 22, 288.
Thibaut, J. W., & Kelley, H. H. The social psychology of

groups. New York: Wiley, 1959.
Walton, D. The application of learning theory to the

treatment of a case of bronchial asthma. In H. J.
Eysenck (Ed.), Behaviour therapy and the
neuroses. New York: Pergamon, 1960. pp. 188-
189.
Wolf, M., Risley, T., & Mees, H. Application of operant
conditioning procedures to the behaviour
problems of an autistic child. Behaviour Research
and Therapy, 1964, 1, 305-312.
Wolpe, J., & Rachman, S. Psychoanalytic “evidence”: A

critique based on Freud’s case of Little Hans.
Journal of Nervous and Mental Diseases, 1960,
131,135-148.
Yablonsky, L. The violent gang. New York: Macmillan,

1962.
Yates, A. J. Symptoms and symptom substitution.

Psychological Review, 1958, 65, 371-374.
Zubin, J. Evaluation of therapeutic outcome in mental

disorders. Journal of Nervous and Mental Diseases,
1953, 117, 95-111.
2
Value Issues and Objectives

In developing and implementing programs for
modifying behavior, the specification of goals is of
central importance. If the objectives are poorly
defined, an agent of behavioral change has no
rational basis for selecting the appropriate
treatment procedures or for evaluating the
effectiveness of his efforts. Illustrations of how the

choice of outcomes determines the selection of
procedures are provided in diverse social
practices. A physician, for example, does not

prescribe medication or surgical intervention for
his patient without first deciding what physical
changes he wants to induce; a researcher does not

choose independent variables for study in advance
of specifying the phenomena he wishes to modify;
a travel agent does not select a route for a client
before ascertaining his destination; and a teacher
does not make assignments to his students in the

absence of some type of educational objective.
Similarly, the first major task in any successful

program of behavior modification is to delineate
the changes it aims to achieve.
Often the principal aims of social change
enterprises are never clearly articulated, with the

result that programs remain directionless or offer
learning experiences that are selected fortuitously
by personal preferences of the change agents
rather than specifically for the needs of the
recipients. Even more often, however, broad

objectives are specified only in terms of ill-defined
hypothetical states (rather than behavioral
outcomes), which furnish little direction for the
selection of appropriate methods and learning
experiences. Indeed, conceptualizing psychological
abstractions as internal properties of clients
rather than as hypothetical constructs of

therapists has resulted in considerable confusion
about the types of changes effected by different
approaches to the modification of behavior.
It is widely assumed that behavioral and
psychodynamic approaches are concerned with

fundamentally different subject matters. The latter
methods supposedly treat complexes, repressed
impulses, ego strengths and mental apparatuses,
the underlying causes of behavior, whereas

behavioral approaches are believed to modify only
superficial behavior. This apparent difference in
subject matter, however, exists primarily in the

therapists’ conceptualizations, not in actual
practice.
Ego strength, to take an example, is a
hypothetical construct and not an entity within the
client. One can neither observe nor modify
hypothetical constructs. The person’s behavior—

broadly defined to include cognitive, emotional,
and motor expressions—is the only class of events
that can be altered through psychological
procedures, and therefore it is the only meaningful

subject matter of psychotherapy. Similarly,
stimulus variables are the only events that the
therapist can modify to effect behavioral change.
Psychotherapy, like any other social influence
enterprise, is thus a process in which the therapist

arranges stimulus conditions that produce desired
behavioral changes in the client. If, for instance, a

psychotherapist creates conditions that increase
the frequency of the behaviors from which ego

strength is inferred, the client will be said to have
acquired increased ego strength as a function of

treatment. On the other hand, if the frequency of
ego-strength behaviors has been reduced in the
course of psychotherapy, the client has suffered a
loss in ego strength. Clearly, ego strength is simply

a hypothetical abstraction whose presumed
behavioral referents are the only reality the

psychotherapist can modify.
In the final analysis, social-learning approaches
and all other existing forms of treatment modify

the same subject matter, namely, behavioral
phenomena. Most discussions of change-inducing
processes, however, focus on treating the
inferences made from behavioral events as though
these abstractions existed independently and

caused their behavioral referents. Philosophers of
science have cautioned against the attribution of
causal potency to described properties of
behavior. Their warnings have had little impact on
personality theorizing.
Neither traits nor types, as concepts, have
any real existence. They are merely words,
and words do not exist in the eye of the
observer nor in the people observed. A man
can not be said to have either a type or a
trait, but he can be said to fit either a type or
a trait. At present the fit will be inexact, for
dimensions of personality have not yet been
quantified well enough to permit of accurate
measurement. In the case of height, the
measurement can be precise, and little
confusion results from saying that a man has
a certain height. Observation and concept
are so closely related that the phrase is not
ordinarily understood to mean more than it
says, namely, that the extent of a given
datum of observation in one direction fits a
certain section of an ideal dimension of
distance. But if an attempt is made to fit
some mode of human conduct to the trait of
courage, the looseness of correspondence
between behavior and concepts leads to
mischievous reification. The concept parts
company with behavior, picks up undefined
notions in its flight from reality, and finally
acquires an independent real existence in its
own right, so that when it is said that a man
has courage, he will be thought of as the
fortunate owner of something considerably
more significant than a certain pattern of
behavior [Pratt, 1939, p. 115].
Similarly, a person who is plagued with “weak ego
strength” will be viewed as suffering from
something vastly more significant than the

behavioral referents from which the construct is
inferred.
For purposes of further illustration, let us

designate behaviors in which persons violate
social and legal codes of behavior and frequently
engage in assaultive activities as the external

expressions of an inferred zoognick. Based on
prevailing clinical practices, the zoognick would
come to represent an intrapsychically functioning

agent. An honorific causative power would be
conferred upon this hypothetical zoognick,

whereas the observed behavior from which its
existence is inferred would be depreciated as

superficial behavioral manifestations. Before long,
psychological tests would be constructed to
measure zoognick strength on the basis of which
diagnosticians would tautologically attribute
clients’ behavior to the action of the underlying

zoognick. Proceeding on the assumption that
“patient variables are not conceived to be

behaviors, but constructs concerning internal
constellations” (Wallerstein, 1963),
psychotherapeutic goals would be stated in terms
of removing the pernicious zoognick. On the other

hand, direct modification of the deviant behavior
would be considered not only superficial but
potentially dangerous, since elimination of the
symptomatic expressions might force the zoognick
to emerge in equally pernicious substitute forms.

A sufficiently charismatic exponent of zoognick
theory could undoubtedly develop a sizable
following with the same extraordinary conviction
in the vital importance and causative potency of
zoognicks as that shown by adherents of libidinal
forces. Oedipal complexes, collective
unconsciouses, and self-dynamisms. Finally,

humanists would embrace zoognick theory as
more befitting the complexities of human beings

than those simplistic mechanistic doctrines that
stubbornly insist that the zoognick is the deviant
behavior rechristened.
Most treatment approaches devote remarkably

little attention to the selection of objectives; when
they are specified (Mahrer, 1967), the intended
outcomes generally include a variety of abstract
virtues described in socially desirable terms, such

as reorganization of the self, restoration of
functional effectiveness, development of
individuation and self-actualization, establishment
of homeostatic equilibrium, where there is id there
shall ego be and where superego was there shall
conscious ego be, achievement of identity,
acceptance of self-consciousness, enhancement of
ego strength, or the attainment of self-awareness,
emotional maturity, and positive mental health.

While some of these objectives allude to vaguely
defined behavioral characteristics, most refer to
nebulous hypothetical states. These abstractions
convey little information unless they are further
defined in terms of specifically observable

behavior.
Behavioral Specification of Objectives
A meaningfully stated objective has at least
two basic characteristics (Mager, 1961). First, it

should identify and describe the behaviors
considered appropriate to the desired outcomes.
The term “behavior” is used in the broad sense to
include a complex of observable and potentially
measurable activities including motor, cognitive,
and physiological classes of responses.
After the intended goals have been specified in
performance, and preferably in measurable terms,
decisions can be made about the experiences that

are most likely to produce the desired outcomes.
For example, the statement, “Increase the person’s

self-confidence and self-esteem,” designates a
therapeutic intent; but it furnishes little guidance,
since it does not reveal the kinds of behaviors the
person will exhibit after he has achieved increased

self-esteem. Once self-esteem and the behaviors
that will be esteem producing for a particular
client have been delineated, one can arrange

conditions that will create the requisite behaviors
and thereby produce the condition of positive self-

evaluation. In some instances learning vocational
skills may be most relevant to acquiring self-
esteem; in some cases developing interpersonal

competencies that will secure positive responses
from others may be most appropriate; in other
cases eliminating alienating social behaviors may
be required if self-evaluation is to be altered; and
finally, in cases where a person is relatively
competent socially and vocationally, an increase in
self-esteem behavior may require the modification

of stringent, self-imposed standards of behavior
upon which self-approving and self-deprecatory

responses are contingent. Similarly, unless the
goals specify the behavior that persons will exhibit
when successfully self-actualized, internally
integrated, self-accepted, personally
reconstructed, homeostatically equilibrated, or

emotionally matured, such goals provide little
guidance.
In addition to describing the behaviors which
reflect the chosen goals, objectives must often be
further delineated by specifying the conditions
under which one may expect the behavior to
occur. Let us assume that increased assertiveness
is the goal for the treatment of an excessively
passive individual. After assertive behavior is
defined in sufficient detail that there is little

ambiguity about the interpersonal skills to be
learned, appropriate conditions can be arranged to
produce the desired changes. To demonstrate,
however, that the person has achieved the
objective, one would not require him to exhibit

assertive behavior in all social situations. Because
interpersonal demands are complex, effective
social functioning requires a well-discriminated
repertoire of behavior. Therefore a complete
statement of objectives should specify to what

degree the modified behavior is expected to be
linked to social conditions.
The emphasis on behavioral specification of
goals is not intended to encourage the selection of
inconsequential outcomes. Instead, it places
greater demands on change agents for careful
analysis of complicated objectives that cannot be
successfully attained by any method as long as
they remain couched in ill-defined, general terms.

Complex behavior is an aggregate of simpler
components which must be individually learned
and appropriately integrated. After complex
performances have been adequately analyzed,
conditions that will permit learning of the

component behaviors can be designated. Without
this type of behavioral analysis, change agents
remain at a loss how to proceed and simply fall
back on favorite routines.
Behaviorally defined objectives not only

provide guidance in selecting appropriate
procedures, but they serve an important

evaluative function as well. When desired
outcomes are designated in observable and
measurable terms, it becomes readily apparent
when the methods have succeeded, when they
have failed, and when they need further
development to increase their potency. This self-
corrective feature is a safeguard against

perpetuation of ineffective approaches, which are
difficult to retire if the changes they are supposed
to produce remain ambiguous.
SEQUENCING OF INTERMEDIATE OBJECTIVES
Establishing complex social behavior and

modifying existing response patterns can be
achieved most consistently through a gradual
process in which the person participates in an

orderly learning sequence that guides him
stepwise toward more intricate or demanding
performances. Although the specification of
ultimate objectives provides some direction and
continuity to a program of change, day-to-day
progress is most influenced by defining
intermediate objectives and the learning
experiences necessary for their attainment. A
comprehensive statement of objectives should,
therefore, contain a sequence of intermediate

goals that lead gradually to more complex modes
of behavior.
This principle of gradation is applied

extensively in the social-learning procedures
discussed in succeeding chapters. In each case,
complex behavior outcomes are analyzed into

smaller subtasks and sequenced so as to ensure
optimal progress. For example, fearful
responsiveness and defensively avoidant behavior
can be successfully eliminated by either direct
exposure to aversive events (Grossberg, 1965;

Herzberg, 1945); by exposure to models boldly
exhibiting approach behavior toward fear-

provoking situations (Bandura, Blanchard, &
Ritter, 1968; Bandura, Grusec, & Menlove, 1967);

or by symbolic reinstatement of threatening
events in the context of strong competing positive
responses (Wolpe, 1958). The therapist first
devises a ranked set of threatening situations to

which the client responds with increasing degrees
of anxiety. Initially the client is presented with the
least threatening event under favorable conditions
until his emotional responses have been
thoroughly extinguished. As treatment progresses

the fear-arousing properties of the aversive
situations are gradually increased until emotional
responsiveness to events that originally he found
most threatening is extinguished. While stimulus
gradation is not a necessary condition for

extinguishing fearful behavior, it permits greater
control over the direction and progress of

behavior changes.
Hierarchical organization of learning
experiences is even more useful in programs
designed to develop new patterns of behavior,
because the response elements that compose
complex performances may themselves be
relatively intricate compounds. Therefore,

complicated response patterns cannot be taught
without first establishing the necessary
components. In social practice, intricate modes of
behavior are best attained stepwise by modeling
progressively more complex responses Bandura,

1969; Lovaas, 1967) and reinforcing gradual
response elaborations.
Skillful sequencing of intermediate objectives
can help achieve desired goals in several ways. By

approaching a complicated learning outcome
through successive subtasks, experiences of
failure can be reduced to a minimum, because no

subtask requires constituent skills that
participants do not already possess. The degree of
positive reinforcement can therefore be
maintained at a high level by continuous progress.
If, on the other hand, people are required to
attempt complex behavior prematurely, they
experience a great many unnecessary failures.

These .experiences may jeopardize the treatment
program by decreasing positive motivation, by
inviting obstructive and avoidant responses, and
even by augmenting deviant behaviors that the
treatment was designed to modify. Graded

objectives both permit greater control over
learning outcomes and guide and focus the
behavior of participants throughout all stages of
treatment. Change programs that are poorly
organized as evidenced by isolated, haphazard,

and inadequately sequenced learning experiences
will produce discouraging results, however valid

the principles supposedly guiding the social
practices.
Factors Impeding Specification of

Objectives
In view of the importance of defining the
necessary learning conditions in terms of clearly
specified goals, it is surprising that objectives are

accorded little consideration in the theorizing and
practice of psychotherapy. Almost without

exception, treatises on psychotherapy contain
detailed prescriptions of the conditions essential
for effecting changes and admonitions about the
hazards of deviating from prescribed methods.

The outcomes that these procedures are designed
to produce and the value judgments implied by
these goals are inadequately explicated. Several

possible reasons may account for this traditional
inattention to issues of goal selection.
ADVOCACY OF NONCONTINGENT SOCIAL

REINFORCEMENT
It is widely believed that noncontingent

“relationship” experiences are the primary
determinants of behavioral change, and
consequently that the specific methods employed
are of secondary importance. In a “therapeutic”
atmosphere in which the therapist exhibits

permissive, non- judgmental and unconditionally
positive attitudes, it is contended, a variety of
methods, within certain broad limits, will produce
essentially similar changes in behavior.
This view—which is somewhat analogous to
relying on “bedside manner” rather than on

specific therapeutic interventions in the
alleviation of physical disorders—can be seriously
questioned by an example in which objectives are
clearly identified. Let us assume that two children
have been referred for treatment, one passive and

nonaggressive, the second exhibiting a
hyperaggressive pattern of behavior. Since the

goal is to increase assertiveness in the passive
child and to decrease the domineering tendencies

of the hyperaggressive child, should the therapist
employ the same methods? Clearly the answer is
in the negative. Based on established principles of
behavior change, procedures aimed at reducing

inhibitions (Wolpe, 1958), the provision of
assertive models of behavior (Bandura, 1965) and
the reinforcement of assertive response patterns
(Jack, 1934; Page, 1936; Walters & Brown, 1963)
are most appropriate and effective for promoting

increased assertiveness. These methods, however,
would be clearly inappropriate in the treatment of
the hyperaggressive child, since they would simply
strengthen the already persistent deviant
behavior. Withdrawal of rewards for aggression

(Brown & Elliott, 1965) combined with modeling
and positive reinforcement of nonaggressive

frustration responses (Chittenden, 1942) is highly
effective for decreasing aggressiveness. Although
in both of these hypothetical cases warmth,

interest, understanding, and other relationship
factors would apply equally, it is unrealistic to
expect these general factors to increase
aggressiveness in one child and to reduce it in the

other. Nevertheless therapists often adhere to a
single set of therapeutic conditions, disregarding

the nature of the client’s deviant behavior.
Maladaptive behavior may thus be strengthened
rather than weakened in cases for whom the
learning conditions are inappropriate.
The relationship view of behavioral
modification also implies that no significant
permanent changes in social behavior can be
achieved unless a social relationship is firmly

established. Until recently it has likewise been
confidently believed that a beneficent teacher-
student relationship is a necessary precondition in
the educational process. Comparative studies,
however, reveal that self-instructional programs
can equal or even surpass the efficacy of
instructors in promoting learning. The assumption
that relationship factors are requisite for the
acquisition and modification of social behavior is

refuted by countless studies of social learning. One
can, for example, acquire complex patterns of
social behavior by observing either symbolic or
real life models with whom no prior relationship
has been developed (Bandura, 1965). Moreover,

many responses that are utilized interpersonally
were originally acquired under noninterpersonal
conditions. This transfer process is demonstrated
experimentally by Walters & Brown (1963), who
found that children who were intermittently

reinforced for hitting an automated Bobo doll
subsequently displayed an increase in physically

aggressive behavior toward other children in
thwarting situations.
Relationship experiences are often designated
nonspecific influences and contrasted with various
learning procedures which are referred to as
specific influences. It is difficult to conceive of
nonspecific influences in social interchanges. Each

expression by one person elicits some type of
response from the other participant, which
inevitably creates a specific reinforcement
contingency that has a specific effect on the
immediately preceding behavior. Numerous

studies of change processes stimulated by social
reinforcement theory disclose that interpersonal
responses have specific and predictable effects on
behavior. It is possible, of course, for a change
agent to display uniformly positive or negative

responses without regard to the behavior of
another person. In such instances, however, it

might be more accurate to characterize the social
interaction as involving indiscriminate, rather
than noncontingent, reinforcement. It has been

shown by Hart, Reynolds, Baer, Brawley, & Harris
(1968) and others that abundant social
responsiveness provided on such a
“nonconditional” basis can neither create nor

maintain beneficial personality characteristics.
Guideless interest is clearly not enough.
Lest readers conclude that social-learning
approaches neglect relationship variables it
should be emphasized here that, quite the

contrary, social reinforcement processes assume a
role of major importance in the modification and
maintenance of personality patterns. Indeed, it is
research conducted within the social-learning
framework that has shown most conclusively that

relationship experiences can exert powerful
control over behavior. The central issues are,
therefore, whether a social relationship is
regarded as a facilitative or a necessary condition
for learning, and whether it is utilized
ritualistically or considerately to benefit the
recipients. Chapter 4 includes a large body of
empirical evidence demonstrating that grossly
deviant behavior in both children and adults —

including infantile behavior, self-destructive
tendencies, hypochondriacal and delusional
behavior, extreme withdrawal, chronic anorexia,
psychogenic seizures, psychotic tendencies and
other deleterious behaviors— can be eliminated,

reinstated, and substantially increased depending
upon the amount of interest, attention, and
solicitous concern such behaviors elicit from
others. A positive relationship thus has the
potentiality both to help and to harm. The well-

intentioned, benign attitudes frequently advocated
by many theories of personality may actually

foster social reinforcement contingencies that
have injurious consequences; this consideration
suggests that child-rearing, educational, and

therapeutic practices must be evaluated by their
effects upon recipients rather than by the
humanitarian intent of change agents. Many well-
meaning people who subscribe to these mental

hygiene practices, which have been widely
promulgated over the years, may at times

inadvertently support or even increase the very
problems their earnest efforts are designed to
ameliorate (Harris, Wolf, & Baer, 1964; Gelfand,
Gelfand, & Dobson, 1967; Lovaas, Freitag, Gold, &

Kassorla, 1965).
A principal assumption of most conventional
approaches to treatment is that clients will reenact
in their relationship with the psychotherapist the

maladaptive interpersonal patterns that
characterize their everyday interactions with
significant persons. Once evoked in various
strengths and guises, the inappropriate nature of
these transferred reactions can be demonstrated
and presumably modified within the therapeutic
setting. Alexander (1956), among others,
questioned these assumptions regarding
transference phenomena. He argued that the

marked dissimilarity of the therapy situation and
the social characteristics of the therapist may not
constitute a suitable stimulus for eliciting strong
generalized responses. Hence, many of the clients’
behavioral problems could not be effectively

modified solely in relation to the therapist.
Moreover, those who lead emotionally
impoverished lives often become more interested
in securing positive reinforcement from their
therapists than in solving their interpersonal

problems. Personality changes are further
obstructed if therapists, due to limited satisfaction

in their own nonprofessional relationships, use
their clients as a substitute source of gratification.
For these and other reasons, Alexander

recommended greater utilization and
extratherapeutic relationships for effecting
changes in social behavior.
It is evident from outcome studies reviewed in
Chapter 1 that, whatever clients may reenact with
their psychotherapists, relatively few beneficial

effects of these reenactments trickle down to daily
interpersonal living. Most likely the artificial
relationship provides substitute gratifications for

those lacking in the clients’ natural relationships
instead of serving as a major vehicle for
personality change. Persons would be helped
more fundamentally if their behavior patterns
were modified to enable them to derive greater

satisfactions from their everyday relationships,
thereby making the purchased relationship
unnecessary.
Many psychotherapists who do not subscribe
to the transference theory nevertheless assume
that a benign, noncontingent attitude toward
clients will produce beneficial personality
changes. Strict adherence to the position that

therapists should be unconditionally accepting is
virtually impossible, as shown in numerous
content analyses (Bandura, Lipsher, & Miller,
1960; Dittes, 1957; Goldman, 1961; Winder,

Ahmad, Bandura, & Rau, 1962). Therapists,
including those who advocate unconditional
positive regard (Murray, 1956; Truax, 1966),
display consistent patterns of approving and
disapproving responses to their clients’ behavior.

Even if unconditional social approval and
acceptance were possible, it would be no more

meaningful as a precondition for change than
noncontingent reinforcement in modifying any

form of behavior. If this principle were, in fact,
applied in child-rearing, parents would respond

approvingly and affectionately when their children
appeared with stolen goods, behaved
unmanageably in school, physically injured their
siblings and peers, refused to follow any

household routines, and behaved maliciously.
“Unconditional love” would make children

directionless, irresponsible, and completely
unpredictable. Similarly, if researchers practiced
indiscriminate positive reinforcement in
experiments in the process of social learning, they

would undoubtedly obtain meager results.
Perhaps this circumstance is relevant to the
psychotherapy outcome data discussed in the

introductory chapter.
Another corollary of the relationship view is

that psychotherapists should select the methods of
treatment that they feel most comfortable in
employing. If such reasoning guided the practice of
medicine—suppose a patient with a brain tumor
consults a surgeon who feels most comfortable in
performing appendectomies and therefore
extracts the patient’s appendix—a sizeable
portion of the patient population would have long

since departed, while an even larger number
would find themselves in short supply of
convenient anatomical structures. Successful
modification of behavior requires certain learning
conditions. Therefore, in planning a change

program, the primary focus should be on desired
objectives and their requisite conditions rather
than on the comforts of the change agents. This
does not minimize the individual differences in
psychotherapists’ capabilities for creating

different types of learning conditions. Rather it
highlights the necessity of selecting change agents

on the basis of the desired learning outcomes.
The common deemphasis of methods and
objectives also derives from the fact that most
psychotherapists are trained essentially in a single
treatment approach, which they apply with minor
variations to a wide range of deviant behavior
patterns. Rogerians offer their clients a particular

all-purpose brand of psychotherapy,
psychoanalysts provide a somewhat different
standard brand; similarly, Adlerians, Jungians,
Sullivanians, Gestaltists, existentialists, and
Rankians present still different forms of omnibus

psychotherapy. Since the client must conform to
the method offered rather than having procedures
selected for him in terms of specified objectives,
the treatment he will receive is fortuitously
determined by the school affiliation of his

psychotherapist.
School affiliations not only determine the

range of procedures that a therapist will employ in
his practice; they also define the client’s central
problems, which the techniques of the school are
designed to resolve. Psychoanalysts will uncover
and resolve Oedipal conflicts; Adlerians will
discover inadequacy problems and alter the
resultant compensatory power striving; Rogerians

will unearth and reduce self-ideal discrepancies;
Rankians will resolve separation anxieties;
existentialists will actively promote awareness of
self-consciousness. Thus in traditional approaches
therapeutic procedures and objectives tend to be

preselected with little reference to the diverse
forms of deviance exhibited by different persons.
Considering the accidental way in which
behavioral deviations are matched with learning
conditions, it is not at all surprising that clients

often terminate therapy after only a few
interviews, and that one cannot determine the

probability of improvement for those who remain.
A social-learning approach does not rely upon a
single set of conditions for effecting personality

changes, but rather it provides, within a unified
framework, diverse methods for modifying
multiform psychological phenomena.
Psychotherapists who are less strongly
committed to a particular theoretical orientation
generally attempt to vary techniques adopted

from different systems to particular problems.
However, because the literature does not provide
explicit criteria for the choice of different methods,

the range of procedures therapists do possess is
utilized more according to their intuition. These
attempts are therefore less definite, less
comprehensive, and usually less effective than a
program in which particular interventions are

used because of their demonstrated effects on
social behavior.
SELECTION OF OBJECTIVES AND ETHICAL ISSUES OF

BEHAVIORAL CONTROL
Behavioral objectives are frequently
unspecified in order to avoid acknowledging the
value judgments and social influences involved in
the modification of behavior. Psychotherapists
who subscribe to conversational methods

customarily portray their form of treatment as a
noncontingent social influence process in which

the therapist serves as an unconditionally loving,
permissive, understanding, empathizing catalyst
in the client’s efforts toward self-discovery and
self-actualization. In contrast, behaviorally

oriented psychotherapists are typically depicted
as antihumanistic, Machiavellian manipulators of
human behavior (Jourard, 1961; Patterson, 1963;
Rogers, 1956; Shoben, 1963). In truth, to the
extent that the psychotherapist—regardless of his

theoretical allegiances—has been successful in
modifying his clients’ behavior, he has either
deliberately or unwittingly manipulated the
factors that control it. It is interesting to note in
this connection that conditions that are
undesignedly imposed upon others are generally
regarded with favor, whereas identical conditions

created after thoughtful consideration of their
effects on others are often considered culpable.

There exists no other enterprise which values
incognizance so highly, often at the expense of the
client’s welfare. One suspects that this therapist-
centered value system would change rapidly if

therapeutic contracts required financial
remuneration to be made at least partially
contingent upon the amount of demonstrable

change achieved by clients in the interpersonal
problems for which they seek help.
In view of the substantial research evidence
that psychotherapists serve as models for, and
selective reinforcers of, their clients’ behavior
(Bandura, Lipsher, & Miller, 1960; Goldman, 1961;
Murray, 1956; Rosenthal, 1955; Truax, 1966;
Winder et al., 1962), it is surprising that many
therapists continue to view the psychotherapeutic
process as one that does not involve behavioral

influence and control.
In later writings, Rogers (1956), a leading
proponent of the anticontrol position, has
acknowledged that psychotherapists do in fact
manipulate and control their clients’ behavior

within the treatment setting. He contends,
however, that this benevolent external control
yields “self-actualized,” “flexible,” and “creatively
adaptive” persons whose post-therapy behavior is

under internal control and no longer subject to the
psychotherapist’s influences. The actual outcomes,
however, are considerably at variance with these

idealized pretensions. A brief comparison of
interview protocols of cases treated by Rogerian
therapists with those of clients seen by therapists
representing differing theoretical orientations
clearly reveals that, far from being individuated
and self-actualized, the clients have been
thoroughly conditioned and converted to the

belief system, vernacular, and interpretations of
reality favored by their respective
psychotherapists. Such conformity in verbal
behavior is partly achieved through selective
reinforcement. Sequential analyses of verbal

interchanges in cases treated by Rogers revealed
that the therapist consistently approved certain
behaviors and disapproved others (Murray, 1956;
Truax, 1966). As treatment progressed, approved
responses increased in frequency while

disapproved verbalizations diminished.
In the often quoted debate between Rogers

and Skinner (1956) concerning the moral
implications of behavioral control, Rogers
distinguishes among three types of control; this
provides an excellent illustration of the use of
propitious relabeling to minimize the ethical
decisions that confront therapists and other
agents of change. In the first category, designated

as external control, person A creates conditions
that alter person B’s behavior without his
concurrence. The second and presumably more
humanitarian form, labeled influence, involves
processes in which A arranges conditions that

modify B’s behavior, to which he gives some
degree of consent. The distinction between
external control and influence, however, is more
apparent than real. In many instances certain
conditions are imposed upon individuals without

their agreement, knowledge, or understanding,
from which they can later free themselves by

willingly changing their behavior in a direction
subtly prescribed by controlling agents. Thus, for
example, persons who have been legally

committed to mental hospitals or penal
institutions may voluntarily enter into treatment
programs to acquire the types of behavior that will
improve their living circumstances in the

institution and ensure a speedy discharge. A more
fundamental ethical distinction can be made in

terms of whether the power to influence others is
utilized for the advantage of the controller or for
the benefit of the controllee, rather than in terms
of the illusory criterion of willing consent.
Internal control, Rogers’ third category,
involves a process in which a person arranges
conditions so as to manage his own
responsiveness. Although self-monitoring systems

play an influential role in the regulation of human
behavior, they are not entirely independent of
external influences. Self-monitoring systems are
transmitted through modeling and reinforcement
processes. After a person has adopted a set of
behavioral standards for self-evaluation he tends
to select associates who share similar value
systems and behavioral norms (Bandura &
Walters, 1959; Elkin & Westley, 1955). The

members of his reference group, in turn, serve to
reinforce and to uphold his self-prescribed
standards of conduct. A person who chooses a
small select reference group that does not share
the values of the general public may appear highly

individualistic and “inner- directed,” whereas in
fact he is very much dependent on the actual and
fantasied approval and disapproval of a few
individuals whose judgments he values highly.
During the course of psychotherapy, clients

likewise adopt, through modeling, their therapists’
values, attitudes, and standards of conduct for self-

evaluation (Pentony, 1966; Rosenthal, 1955).
Responsiveness to modeling influences is apt to be
particularly enhanced in a relationship in which a
person has developed a strong positive tie to a
prestigious model (Bandura & Huston, 1961;
Henker, 1964; Mussen & Parker, 1965), a
condition which is emphasized considerably in

most forms of psychotherapy. Studies of modeling
effects further disclose that persons tend to
perform the model’s behavior in his absence
(Bandura & Kupers, 1964; Bandura, Ross, & Ross,
1963), and they respond to new situations in a

manner consistent with the model’s dispositions
even though they have never observed the model’s
behavior in response to the same stimuli (Bandura
& Harris, 1966; Bandura & McDonald, 1963;
Bandura & Mischel, 1965). These findings indicate

that after the model’s attitudes and behavioral
attributes have been adopted, he continues to

influence and indirectly to control the subject’s
actions, though he is no longer physically present.
In fact, in Rogers’ (1951) conceptualization of

maladjustment, introjected parental values are
construed as continuing pathological influences
that maintain disturbing incongruities in the
clients’ self-structure. However, after internalized

parental values are supplanted by adoption of the
therapist’s attitudes and standards, the client is

flatteringly portrayed—by the psychotherapist—
as self-actualized, flexibly creative, and self-
directed!
Much of the controversy between Rogers and

Skinner centers around their own value
preferences for others. Skinner advocates that
people be made “truly happy, secure, productive,
creative, and forward-looking”; Rogers argues in

favor of self-direction and self-actualization of
potentialities as the prescribed objective of social
influence. It might be noted parenthetically that in
the context of proclaiming the self-actualization
objective, Rogers argues vigorously against self-
actualization in Skinnerian directions. The
leitmotif in this discourse appears to be one of
belief conformity rather than self-realization. As
usually happens in disputes over therapeutic

outcomes, “happiness” and “conformity to societal
norms” are selected as examples of unwholesome
outcomes equated with slothfulness; self-
actualization, on the other hand, is proffered as an
ennobling aim. To balance the evaluative scales, it

should be noted that the self-centered ethic of self-
actualization might be equally questioned on
moral grounds, particularly by innocent victims of
self-actualized despots or less notorious but
selfish, self-directed persons. Universally accepted

goals are difficult to come by because all the
various patterns of behavior enthusiastically

promoted by therapists of different persuasions
can be used to produce inimical human effects.
The most remarkable feature of the foregoing,
seemingly humanistic, rhetoric is that neither
participant acknowledges that the choice of
behavioral objectives is rightfully the client’s. A
person may seek from therapy neither Skinner’s

security nor a Rogerian conversion in the guise of
self-realization. We shall return shortly to this
issue of value standardization and the inclination
of therapists to impose their own cherished
objectives upon their clients.
Contrary to the beliefs of Rogers, Shoben, and

other critics, behaviorally oriented approaches
usually involve considerably less unnecessary
control and manipulation of attitudes and values

than do the procedures based upon the
psychodynamic model. In the latter treatments,
any behavior, no matter how trivial or apparently

irrelevant, tends to be viewed as a derivative of
concealed psychodynamic forces and is therefore
subject to analysis and reinterpretation in terms of
the therapist’s theoretical predilections. Thus
virtually no aspect of the client’s life—his social,
marital, and sexual behavior, his political and
religious beliefs, his vocational choice, his child-

training practices—escapes the therapist’s
repeated scrutiny and influence over a period of
several years. Since this approach tends to regard
behavioral difficulties as superficial
manifestations of more fundamental and often

unconscious internal events, influence attempts
are primarily directed toward subject matters of
questionable relevance. It is not uncommon,
therefore, to find clients whose belief systems
have been thoroughly modified despite little

amelioration of the behavioral difficulties for
which they originally sought help.
In contrast, behaviorally oriented therapists
generally confine their therapeutic efforts to the
behavioral problems presented by the client.
These are labeled as learned styles of behavior
rather than as expressions of esoteric unconscious
processes or as manifestations of mental disease.
Moreover, the procedures and objectives are

undisguised, the treatment is typically of shorter
duration, and clearly goal-directed. To be sure,
within this highly structured interaction, the
therapist must exercise responsible control over
conditions affecting relevant segments of the

client’s behavior if he is to fulfill his therapeutic
obligations. In this type of approach, however, the
psychotherapist is less inclined to condition and to
shape his client’s belief systems in accordance
with his own views. Paradoxical as it may seem,

the psychotherapists who pride themselves on
being nonmanipulative and noncontrolling are,

albeit unwillingly, often engaged in a more
disguised and manipulative enterprise than is true
of most behaviorally oriented practitioners. It

should be made clear, however, that behavioral
principles do not dictate the manner in which they
are applied. Undoubtedly some behavioral
therapists encroach on people’s rights to decide

the direction in which their behavior will be
modified, and act as therapeutic agents devoid of

consideration and regard for values.
ESTABLISHMENT OF FREEDOM OF CHOICE

THROUGH BEHAVIORAL APPROACHES
Discussions of the moral implications of
behavioral control almost always emphasize the
Machiavellian role of change agents and the self-

protective maneuvers of controllees. The fact that
most persons enter treatment only as a last resort,
hoping to modify patterns of behavior that are
seriously distressful to themselves or to others, is
frequently overlooked. To the extent that

therapists engage in moral agonizing, they should
fret more about their own limited effectiveness in
helping persons who are willing to undergo
financial hardships to achieve desired changes,
than in fantasizing about their potential powers.
The tendency to exaggerate the powers of

behavioral control by psychological methods
alone, irrespective of willing cooperation by the
client, and the failure to recognize the reciprocal
nature of interpersonal control obscure both the

ethical issues and the nature of social influence
processes.
In discussing moral and practical issues of
behavioral control it is essential to recognize that
social influence is not a question of imposing
controls where none existed before. All behavior is
inevitably controlled, and the operation of

psychological laws cannot be suspended by
romantic conceptions of human behavior, any

more than indignant rejection of the law of gravity
as antihumanistic can stop people from falling. As

Homme and Tosti (1965) point out, “either one
manages the contingencies or they get managed by
accident. Either way there will be contingencies,
and they will have their effect [p. 16].” The process
of behavior change, therefore, involves
substituting new controlling conditions for those
that have regulated a person’s behavior. The basic
moral question is not whether man’s behavior will
be controlled, but rather by whom, by what

means, and for what ends.
The primary criterion that one might apply in
judging the ethical implications of social influence
approaches (Kelman, 1965) is the degree to which

they promote freedom of choice. It should be
added, however, that if individualism is to be
guaranteed, it must be tempered by a sense of

social obligation. Custodial institutions created by
societies are highly populated with socially
injurious individualists. A person’s freedom of
self-expression can be restricted in several ways,
each of which presents somewhat different ethical
problems in the reestablishment of self-
determination.
Self-restraints in the form of conditioned
inhibitions and self-censuring responses often
severely curtail a person’s effective range of
behaviors and the types of options that they are
likely to consider for themselves. In many

instances, for example, persons are unable to
participate freely in potentially rewarding social
interactions because of severe phobias; they are
unable to engage in achievement, aggressive and

heterosexual activities; and they deny themselves
socially permissible gratification because of
austere, self-imposed standards of conduct.

Treatment programs designed to reduce rigid self-
restraints are rarely viewed as ethically
objectionable, since they tend to restore
spontaneity and freedom of choice among various
options of action. Ethical issues arise only if a
change agent uses his influence selfishly or to
make his clients socially irresponsible.
Behavioral deficits also greatly restrict freedom
of choice and otherwise curtail opportunities for
self-direction. Persons’ positions in various status
and power hierarchies are to a large extent
determined by their social, educational, and

vocational competencies. The degree of control
that one can exercise over one’s own activities, the
power to form and to modify one’s environment,
and the accessibility to, and control over, desired

resources increase with higher status positions.
Persons who have developed superior intellectual
and vocational capabilities enjoy a wide latitude of

occupational choices; they are granted
considerable freedom to regulate both their own
activities and the behavior of others; and they
have the financial means of obtaining additional
privileges that further increase their autonomy. By
contrast, high school dropouts who lack
sociovocational proficiencies are relegated to a

subordinate status, in which not only is their
welfare subject to arbitrary external controls, but
they are irreversibly channeled into an economic
and social life that further restricts their
opportunities to use their potentialities and to

affect their own life circumstances. Eliminating
such behavioral deficits can substantially increase
the level of self-determination in diverse areas of
social functioning.
Societally imposed restrictions on freedom of

self-expression occur as responses to deviant
behavior that violates legal codes. Chronic

alcoholics, drug addicts, sexual deviates,
delinquents, psychotics, and social nonconformists
and activists may have their liberties revoked for
fixed or indefinite periods when their public
actions are judged to be socially detrimental and
therefore to be subject to social control. Special
ethical problems are most likely to arise wherever

restoration of his freedom is made contingent
upon the individual’s relinquishing socially
prohibited patterns of behavior. If an agent of
change acts in opposition to the society which
supports him institutionally, then he evades his

broader social responsibilities with which he has
been entrusted. If, on the other hand, he imposes
conditions upon his captive clients designed to
force conformity to social norms, he is subverting
the client’s right to choose how he shall live his

life. These moral dilemmas are less difficult to
resolve in cases where the person’s behavior

injures or infringes the freedom of others. Such
persons have the choice of regaining their
autonomy by undergoing changes within a broad

range of socially tolerated alternatives, or setting
no limits on their own behavior and having society
restrict them to institutions. The ethical dilemma
is more serious when conventional norms are

questioned by many members of society and new
standards of behavior are advocated. Today there

are open controversies over the morality of
homosexuality, premarital sexual intercourse, use
of nonaddictive drugs, civil disobedience to unjust
rules, and many forms of social behavior that are

publicly defined as illegal. In such cases as these,
therapeutic agents may support changes in
socially prescribed directions or give legitimacy to

deviant patterns, depending upon the social and
personal consequences of the behavior, the client’s

preferences, and the therapist’s own value
orientation.
Most people whose freedom is curtailed by
societally imposed restrictions and who
voluntarily seek psychotherapeutic help are not
that strongly wedded to deviant behavior; but
because it is powerfully reinforcing, or because
they lack more satisfying alternatives, they have

difficulty relinquishing it. The establishment of
self-control and the reduction of positive valences
associated with deviant activities may sometimes
require the use of aversive procedures as part of
the treatment program. The use of aversive

methods is apt to be criticized as being, if not anti-
therapeutic, then certainly antihumanistic. But is it
not far more humanitarian to offer the client a
choice of undergoing a brief, painful experience to
eliminate self-injurious behavior, or of enduring

over many years the noxious, and often
irreversible, consequences that will inevitably

result if his behavior remains unaltered?
Restrictions of behavioral freedom arise also
from socially sanctioned discrimination. In such
cases a person’s freedom is curtailed because of
his skin color, his religion, his ethnic background,
his social class, or other secondary characteristics.
When a person’s warranted self-determination is

externally restricted by prejudicial social
practices, the required changes must be made at
the social systems level.
It is often mistakenly assumed (London, 1964)
that traditional psychotherapies fervently

embrace humanism whereas behavioral
approaches, for reasons never explicated, are
supposedly uninterested in the moral implications
of their practices or are antagonistic toward

humanistic values. In fact, behavioral therapy is a
system of principles and procedures and not a
system of ethics. Its methods, and any other

effective procedures for that matter, can be
employed to threaten human freedom and dignity
or to enhance them.
When freedom is discussed in the abstract it is
generally equated with nondeterminism;
conversely, automatonism is associated with a

deterministic position. Whether freedom and
determinism are compatible or irreconcilable

depends upon the manner in which causal
processes are conceptualized. According to
prevailing theories of personality, human actions
are either impelled from within by concealed

forces or externally predetermined. If individuals
were merely passive reactors to external
influences, then their behavior would be

inevitable; it would be absurd to commend them
for their achievements or to penalize them for

their transgressions. It would be more sensible,
from this point of view, to praise and to chastise
the external determinants. But since these events

are also unavoidably determined by prior
conditions, the analysis results in an infinite
regression of causes. Some degree of freedom is
possible within a deterministic view if it is

recognized that a person’s behavior is a
contributing factor to subsequent causal events. It
will be recalled from the previous discussion of

reciprocal influence processes that individuals
play an active role in creating their own
controlling environment.
From a social-learning point of view freedom is

not incompatible with determinism. Rather a
person is considered free insofar as he can partly
influence future events by managing his own
behavior. One could readily demonstrate that a

person can, within the limits of his behavioral
capabilities and environmental options, exercise
substantial control over his social life by having
him plan and systematically carry out radically
different courses of action on alternate days.
Granted that the selection of a particular course of
behavior from available alternatives is itself the
result of determining factors, a person can
nevertheless exert some control over the variables

that govern his own choices. Indeed, increasing
use is being made of self-control systems (Ferster,
Nurnberger, & Levitt, 1962; Harris, 1969; Stuart,
1967) in which individuals regulate their activities
to fulfill their own wishes by deliberate self-

management of reinforcement contingencies. The
self-control process begins by informing
individuals of the types of behaviors they will have
to practice to produce desired outcomes, of ways
in which they can institute stimuli to increase the

occurrence of requisite performances, and of how
they can arrange self-reinforcing consequences to

sustain them. Behavioral change procedures that
involve role enactment also depend upon the self-
determination of outcomes through clients’

regulation of their own behavior and the
environmental contingencies that reciprocally
influence it. Contrary to common belief, behavioral
approaches not only can support a humanistic

morality, but because of their relative
effectiveness in establishing self-determination

these methods hold much greater promise than
traditional procedures for enhancement of
behavioral freedom and fulfillment of human
capabilities.
BEHAVIORAL OBJECTIVES IN THE MODIFICATION

OF INTERNAL STATES AND COMPLEX
DYSFUNCTIONS
Thus far the failure to orient treatment to

desired behavioral outcomes has been attributed
to the prevalence of all-purpose single-method
therapies, to reliance on benign relationship
factors to produce diverse changes, and to
reluctance to acknowledge the issues of values and
behavioral control involved in the modification of
social behavior. The failure to specify objectives in
behavioral terms also stems in part from the view
that, in many cases, internal psychic states may

constitute the major problems requiring
modifications. These conditions are usually
defined in such broad terms as unhappiness,
absence of meaning and purpose in life, and

feelings of worthlessness. Before speculating on
how phenomenological events can be most
effectively altered, it should be noted that it is
highly fashionable to construe one’s concrete
behavioral problems in abstract, cosmic terms. It

is understandably less distressing to present one’s
plight as a manifestation of social maladies of

alienation, exploitation, or dehumanization than it
is to acknowledge despairing personal

shortcomings, evident heterosexual inadequacies,
intellectual failures, lack of vocational ingenuity

and productivity, and inability to form satisfying
interpersonal relationships.
Abstract problems such as “unhappiness,” and

“purposelessness” cannot be successfully modified
by any form of treatment as long as they remain
disconnected from their concrete experiential

determinants. A person does not feel abstractly
unhappy; he is most likely distressed about
specific problems arising from his mode of

functioning in social, vocational, sexual, or familial
areas. After the contributing conditions have been
identified, an appropriate treatment program can
be devised. The principal difficulty in modifying
complex conditions is not that behavioral

approaches are inapplicable, but that the
psychological phenomenon is generally described
in global abstract terms and the constituent
determinants are never clearly specified.
Greatest progress will be made in the
successful treatment of so-called complex
disorders when they are conceptualized, not as
nebulous general states, but as psychological

conditions involving multiple problems with
varying degrees of interdependence. From this
perspective, altering complex behavioral
dysfunctions does not require radically different

methods from those applied to the modification of
single disorders. This issue can perhaps be
illustrated by considering learning deficits. A child
may have developed satisfactory academic skills in
all areas except mathematics. Another child is

grossly deficient in mathematics and in other
academic skills, lacks social behavior skills that

would enable him to maintain satisfying
interpersonal relationships, and has not developed

motor competencies required for play activities.
There exists no single nonspecific treatment that

can simultaneously create competencies in
intellectual, linguistic, social, and motoric areas of
functioning. Separate programs would have to be
devised for each type of problem. But the

procedures used to develop arithmetic
competencies would be essentially the same in the

single-problem and the multiple-problem case.
This is precisely the approach employed by Lovaas
(1967) in establishing language functions,
interpersonal capabilities, and intellectual skills,

and in eliminating grossly bizarre behavior in
autistic children who present, in extreme forms,
one of the most generalized and complex

psychological disorders that therapists are called
upon to treat. Additional examples of successful

modification of multiform problems through
specific diverse treatments is provided by
Patterson & Brodsky (1967), and by Risley & Wolf

(1966). The developments in behavioral therapy
in some respects parallel those in medicine, where
global all-purpose treatments of limited efficacy
were eventually replaced by powerful specific
procedures for treating particular physical

disorders.
The behavioral change process is not as

piecemeal as the preceding remarks might imply.
Most psychological functions are at least partially
interdependent. Therefore, desirable changes in

one area of behavior may produce beneficial
modifications in other areas not directly involved
in the treatment program. Often, as will be shown
later, a relatively circumscribed problem has
widespread social consequences; and a change in a

specific deviant behavior can have pervasive
psychological effects.
If the major aim of therapy is the modification
of phenomenological events, the empirical
question remains how such changes can be made
most successfully. Some theorists hold that
behavior is essentially a byproduct of
phenomenological experiences; therefore they

select the latter events as the major subject matter
of therapeutic conversations. According to social-
learning theory, self-descriptions and
phenomenological experiences are partly by-

products of behaviorally produced outcomes.
People, for example, who lack the social and
vocational competencies required for meeting
environmental demands, and who resort to
defective coping strategies will undoubtedly

engender numerous adverse consequences, which
will give rise to despondency, negative self-

evaluations, and other subjective distresses.
Similarly, those who derive inadequate positive

reinforcement from their vocational and
interpersonal activities will experience feelings of

purposelessness and alienation. From a social-
learning perspective, phenomenological and other
internal events can be more effectively modified
through behavioral changes and the feedback from

resulting consequences than through conventional
interview procedures.
A laboratory study conducted by Keister
(1938) illustrates bow phenomenological events
can be altered by feedback from a series of

carefully guided mastery experiences. The author
selected a group of children who exhibited
extreme maladaptive tendencies, including
withdrawal, destructiveness, sulking and crying,
and expressions of feelings of helplessness when

faced with problem-solving tasks. Keister did not
obtain ratings of the children’s self-concepts, but it
is highly probable that, as a result of repeated
failure experiences, these children would
eventually evaluate themselves in negative terms.
In the treatment program the children solved a
series of graded problems that grew progressively
more difficult, thus ensuring a gradual build-up of
skill in coping with increasingly difficult tasks. In

addition, the experimenter consistently rewarded
the children’s successful solutions and persistent
task-oriented behavior. A pre- and post-test
comparison of the children’s responses to
exceedingly difficult tasks showed that the success

experiences were highly effective in replacing the
formerly maladaptive tendencies with
constructive, confidence-producing behavior.
Because cognitive and attitudinal changes have

rarely been systematically assessed in
behaviorally oriented programs, it is generally
assumed that these types of treatment approaches

alter only specific behavioral functioning. Several
experiments have recently been designed
especially to provide empirical evidence of the
affective and cognitive consequences of behavioral
changes. Bandura, Blanchard, and Ritter (1969)
found that elimination of phobic behavior was
accompanied by marked attitudinal changes

toward previously feared situations. In addition,
disturbing emotional responsiveness not only
toward the phobic stimulus but toward situations
beyond the specifically treated condition was
substantially reduced. In a preliminary study,

Wahler and Pollio (1968) similarly demonstrated
that behavioral changes produced in a boy
through selective social reinforcement altered
favorably his evaluations of himself and others. As
might be expected, his evaluation of events closely

related to the treatment objectives changed most
markedly.
Not only are self-attitudes and feeling states
fundamentally affected by behaviorally produced
experiences, but a favorable change also gains the
person acceptance and increased social status
(Hastorf, 1965). The positive social feedback
engendered by behavioral competence can thus
have important phenomenological consequences.

In subsequent chapters research evidence will be
presented showing that cognitive and affective
modifications can be achieved more successfully
through planned behavioral change than through
attempts to alter internal events directly. The

relative superiority of a behaviorally oriented
approach probably stems from the fact that a basic
change in behavior provides an objective and
genuine basis by which one feels self-respect, self-
confidence, and dignity.
INSIGHT AS A THERAPEUTIC OBJECTIVE
Most traditional approaches to psychotherapy
consider the achievement of insight or self-
awareness to be a prerequisite for the production
of widely generalized and enduring behavioral
changes. Therefore, development of insight
constitutes one of the primary objectives of
interview strategies. For this reason, among the

numerous technical issues discussed in
expositions of psychotherapeutic procedures,
those pertaining to timing and depth of
interpretations, methods for channeling

verbalizations into areas assumed to be conflict-
laden, strategies for handling clients’ resistances,
and explanations of the possible symbolic
significance of verbal and nonvocal responses
have all received considerable attention.
In therapeutic practice, the development of
insight is largely accomplished by therapists

repeatedly interpreting the verbal, affective, and
social responses that their clients report or exhibit

within the treatment setting. A number of
authorities have proposed rules for the optimal

level of interpretive responses for promoting
insights. According to Rogers (1951), for example,
clients will engage in progressively deeper self-
exploration provided that therapists label only the

feelings that are expressed more or less explicitly.
On the other hand, Fenichel (1941) and other
advocates of psychoanalytical procedures
recommend that therapists proceed slightly
beyond what the client is able to accept and

experience emotionally at any given time. By
contrast, Klein (1960), Berg (1947), and Rosen
(1953), among others, contend that rapid and
fundamental personality changes can be achieved
only by deep interpretations of internal processes

of which the client is completely unaware.
Research bearing on this issue (Collier, 1953;

Dittmann, 1952; Harway, Dittmann, Raush, Bordin,
& Rigler, 1955) has been mainly concerned with
attempts to scale the depth of therapists’

interpretive responses, which are typically rated
on a continuum ranging from superficial
restatements of clients’ remarks to suggestions of
causal relationships and psychological events that

are entirely foreign to clients’ views of themselves.
In addition, client- therapist verbal interchanges

have been occasionally analyzed in an effort to
establish relationships between variations in
therapists’ interpretive responses and different
verbal indices of therapeutic progress (Colby,

1961; Dittmann, 1952; Frank & Sweetland, 1962;
Speisman, 1959).
Despite the lack of consensus regarding
optimal interpretive procedures, it is generally

assumed that through skillful labeling of repressed
strivings, which manifest themselves in various
derivative forms, the unconscious determinants of
the client’s behavior are gradually made
conscious. After these unconscious events are
brought into awareness they presumably cease to
function as powerful instigators of behavior, or
they become more susceptible to cognitively
mediated control. Hence it is believed that with

the achievement of insight, flexible, voluntarily
guided behavior replaces automatic,
indiscriminate responding.
Although the acquisition of insight is
considered an essential goal of treatment and

supposedly results in a wide variety of beneficial
effects, insight has never been adequately defined
(Zilboorg, 1952), nor has the manner in which it
supposedly mediates behavioral change ever been

clearly specified or demonstrated. Apart from the
difficulties of defining insight, the history of a
client’s behavior is rarely known, and the

reconstructed content of both historical and
contemporary events is highly influenced by the
therapist’s suggestive probing and selective
reinforcement of the client’s verbalizations. Thus,
as Marmor (1962) has pointed out, schools of
psychotherapy have emerged with their own
favored set of hypothetical internal agents, and

their own preferred brand of insight; these can be
readily confirmed by self-validating interview
procedures. For these reasons, psychotherapists of
differing theoretical orientations repeatedly
discover their preferred psychodynamic agents,

but are unlikely to find evidence for the
underlying causes emphasized by their theoretical
rivals:
But what is insight? To a Freudian, it means

one thing, to a Jungian another, to a Rankian,
a Horneyite, an Adlerian or a Sullivanian, still
another. Each school gives its own particular
brand of insight. Whose are the correct
insights? The fact is that patients treated by
analysts of all these schools may not only
respond favorably, but also believe strongly
in the insights which they have been given.
Even admittedly ‘inexact’ interpretations
have been noted to be of therapeutic value!
Moreover, the problem is even more
complicated than this; for, depending upon
the point of view of the analyst, the patients
of each school seem to bring up precisely the
kind of phenomenological data which
confirm the theories and interpretations of
their analysts! Thus each theory tends to be
self-validating. Freudians elicit material
about the Oedipus Complex and castration
anxiety, Adlerians about masculine strivings
and feelings of inferiority, Horneyites about
idealized images, Sullivanians about
disturbed interpersonal relationships, etc.
The fact is that in so complex a transaction as
the psychoanalytic therapeutic process, the
impact of the patient and the therapist upon
each other, and particularly that of the latter
upon the former, is an unusually profound
one. What the analyst shows interest in, the
kinds of questions he asks, the kind of data
he chooses to react to or to ignore, and the
interpretations he makes, all exert a subtle
but significant suggestive impact upon the
patient to bring forth certain kinds of data in
preference to others [Marmor, 1962, p. 289].
The above assessment of the arbitrariness of
psychotherapeutically derived insights finds some
support in the findings of an experiment
conducted by Heine (1953), in which clients who

had been treated by psychoanalytic, Rogerian, and
Adlerian therapists were asked to specify the
factors responsible for their personality changes.
Although clients treated by therapists of these
different theoretical affiliations reported a similar

degree of improvement, they tended to account for
their behavior in terms of the explanation favored
by their respective therapists. These results, and
other findings that will be cited later, strongly
indicate that the content of a particular client’s

insights and emergent “unconscious” could be
predicted more accurately from knowledge of his

therapist’s theoretical belief system than from the
client’s actual social-learning history.
INSIGHT: A SOCIAL-CONVERSION OR A SELF-

DISCOVERY PROCESS?
In the preceding section it was suggested that
interpretive activities might be more accurately
represented as a direct social influence rather than

as a process involving delicate levitation of
repressed forces from the region of the client’s

unconscious mind. Psychotherapists’ reports that
their clients have achieved self-awareness
generally mean, in behavioral terms, that clients
have learned to label social stimulus events, past

and present causal sequences, and their own
responses in terms of the theoretical predilections
and language of their psychotherapists. In

traditional practice insight primarily represents a
form of self-evaluative behavior that is

conditionable and extinguishable, as are
nonverbal performances. By subsuming the
development of insight under the broad

framework of social persuasion, much of the
knowledge discovered by experimental social
psychology can be applied to the understanding of
how therapists induce, alter, and control their
clients’ self-insights—even though, in some cases,
therapists subscribe to such idiosyncratic beliefs
about the conditions governing human behavior as
to strain the broad limits of rationality.
Several factors of the treatment situation
contribute to the process of persuasion,

particularly as it applies to changes in the manner
in which clients construe their own actions and
what determines them. As noted in the preceding
chapter, because of initial selectivity and later
attrition of cases during the course of treatment,

the types of people who seek out and remain in
psychotherapy display personal attributes similar
to those of persons who, in laboratory studies of
conformity, attitude change, and conditionability,
are highly amenable to social influence. In addition
to the selection of persuasible clients, therapists,
by virtue of their advanced training and expertise,
are usually accorded high prestige and credibility.
Views expressed by sources of high credibility

generally exert a stronger influence on recipients’
opinions than those of low credibility sources
(Berg & Bass, 1961; Bergin, 1962; Hovland, Janis,
& Kelley, 1953). Interpretations made by
prestigious psychotherapists are, therefore, more

likely to alter clients’ opinions of themselves than
to produce disbelief or to destroy their confidence
in the therapist.
A closely related factor that seems both to

augment attitudinal conformity and to reduce
discrediting of the psychotherapist is the
ambiguity of the psychotherapeutic situation.

Usually the goals of treatment, if discussed in any
detail, are stated only in general terms; clients are
given only general instructions about the nature of
the therapeutic task and the manner in which the
objectives are to be realized. The therapist often
deliberately strives to remain ambiguous in order
to facilitate inappropriate generalization of

maladaptive patterns of behavior toward the
therapist. Most important, the subject matter of
interpretations is primarily concerned with
inferences about unobservable internal processes
rather than with more objective behavioral events.

Clients would, of course, have no difficulty in
ascertaining the validity of therapists’ judgments
of factual matters; however, clients have little
objective basis for evaluating whether they
possess Oedipus complexes, repressed hostilities,

latent homosexual urges, oral-sadistic drives, and
other esoteric motivational forces whose

identification is further complicated by the fact
that they are often inferred from both the high
incidence and the absence of the same behavior.

Studies of social compliance (Asch, 1952; Berg &
Bass, 1961) have abundantly documented that
persons can be more easily induced to accept the
opinions of others on subjective and unfamiliar

matters than on interpretations of events for
which objective cues are available. Having altered

their judgments, subjects typically underestimate
the extent of their compliance and the role of
social influence in modifying their opinions
(Rosenthal, 1963).
The fact that psychological treatment promises
relief from the distress occasioned by the client’s
behavioral difficulties also works against his quick
dismissal of insights proffered by the

psychotherapist, who is often sought out as a last
resort. Distress generally facilitates persuasion,
especially if solutions allegedly effective at stress
reduction are also made available (Chu, 1966;
Dabbs & Leventhal, 1966).
In attitude change research the opinions
selected for modification have generally involved
social rather than highly personal matters. A study

by Bergin (1962) of interpretations as persuasive
communications demonstrates that the variables

shown to control social attitudes play a similarly
influential role in altering the self-attitudes that
often concern psychotherapists.
In making interpretations, a therapist

communicates information about the client which
is somewhat discrepant with the client’s view of
himself. The controversy regarding the optimal
depth of interpretation might therefore be recast

in the following form: Can a person’s self-attitudes
be altered more rapidly by presenting him with a
progressive series of mildly discrepant
communications slightly beyond what the client is
willing to accept, or by confronting him with
extremely divergent communications as
recommended by Rosen (1953) and Klein (1960)?
The search for an optimal level of
interpretation may be a fruitless pursuit since,

according to persuasion theory, the effectiveness
of varying degrees of discrepant communications
is highly dependent upon the attributes,
credibility, social prestige, and power of the
communicator. Therapists to whom are attributed

low credibility and prestige, for example, may be
relatively ineffectual in producing attitudinal
changes even though they faithfully adhere to
interpretations that are only moderately at

variance with their clients’ beliefs about
themselves. On the other hand, when
psychotherapists are considered to be a source of

high credibility, and possess power to reward and
punish the client’s behavior, then “deep”
interpretations may be highly influential in
shaping clients’ self-insights. Perhaps this is the
reason why Rosen, who exercises considerable
rewarding and coercive power over his psychotic
patients, finds that deep interpretations produce

rapid attitudinal changes, whereas similar
interpretive strategies by therapists who lack
control over their patients’ environment generally
prove ineffectual. The interactive effects of these
different social variables upon conforming self-

evaluations are most clearly illustrated in Bergin’s
study (1962), which manipulated independently
both credibility of the communicator and degree
of incongruity of interpretations.
In the high credibility condition, college

students were seen individually in the Psychiatry
Department of a medical center by the

experimenter, who was ostensibly director of a
depth personality assessment project. To enhance
further the verisimilitude of the situation, students
were escorted by the clinic receptionist to the
experiment room, which was furnished with,
among other things, psychophysiological
recording equipment, an impressive array of

medical and psychiatric tomes, and a large portrait
of Sigmund Freud.
After the students had rated their
interpersonal characteristics on several rating
scales, they were administered an extensive

battery of psychological tests which were
presented as valid measures of underlying
personality dynamics. In a session conducted
several days later, the experimenter informed the

students that, according to results of the depth
assessment, their level of self-understanding was
quite accurate on all the traits rated except for the

area of masculinity-femininity. They then received,
according to random assignments, interpretations
that depicted them as either moderately, highly, or
extremely more feminine (masculine for girls)
than they judged themselves to be. Later the
students rated themselves again so that changes in
their self-evaluations could be assessed. Students

in the low credibility condition likewise completed
the initial self-ratings, received one of the three
levels of discrepant interpretations concerning
their masculine status, and then repeated the self-
evaluation. In these cases, however, the judgments

were made in a decrepit basement office by a
scrawny youngster on the basis of casual
observation.
The results, presented graphically in Figure 2-

1, show that under high credibility conditions the
more divergent the interpretation the greater the
change in self-attitudes; on the other hand, when

interpretations issued from a source of low
credibility, the amount of attitude change
decreased with increasing discrepancy between
the judgments of the participants.
Figure 2-1. Mean change in self-appraisal considered most
acceptable by subjects as a function of credibility of the
communicator and degree of discrepancy of the
interpretation from subjects’ view of themselves.
Bergin, 1962.
Although the generality of the self-evaluative
conforming behavior cannot be determined from
the findings of the foregoing study, it nevertheless

suggests strongly that people are willing to adopt
erroneous underlying attributes suggested to

them by prestigious specialists. It might be
supposed that the persuasive efforts of
psychotherapists would be especially effective
because the same interpretations are made

repeatedly during prolonged treatment and are
directed not only toward assumed unconscious
determinants but also toward clients’ resistances

against the prompted insights.
Suggestive communications offered by

prestigious agents under conditions of ambiguity
and high personal distress may be well suited for
imparting insights to clients, but after the self-
beliefs have been socially induced their
maintenance is strongly governed by existing
conditions of reinforcement. Results of
innumerable verbal conditioning experiments and
analyses of client-therapist interactions, which

have been cited earlier, furnish ample evidence
that psychotherapists selectively reinforce
conformity to their own opinions about the causes
of behavior, and that clients can readily secure
their therapists’ appreciation and approval by

reiterating the appropriate insights.
It would seem from the findings presented
above that interpretive psychotherapies may
primarily represent a conversion of the client to

the therapist’s point of view rather than a process
of self-discovery. It is not surprising, therefore,
that insight can be achieved without helping the

client with the difficulties for which the client
originally sought help. There is no reason to
expect, for example, that a stutterer converted to
Freudianism, Jungianism, Existentialism,
Behaviorism—or to any other theoretical system
—will begin to speak fluently. His stuttering is
more likely to be eliminated by necessary

relearning experiences than by the gradual
discovery of predetermined insights. To account
for the lack of relationship between insight and
social behavior, different varieties of insights have
been distinguished. On the one hand, there is

“intellectual insight,” which is believed to exist
when cognitive responses are present but the
accompanying social or emotional behavior is
absent. Then there is “emotional insight” which is
typically defined in terms of the effects which it

presumably causes: If the client exhibits
behavioral changes, he has achieved emotional

insight; if he fails to modify his social behavior
then he has simply acquired intellectual insight.
While the view that insight is a prerequisite of
behavioral change is widely accepted, some
theorists (Alexander, 1963) have considered
insight to be a consequence of change, rather than
its determinant. That is, as anxieties are

progressively reduced through the permissive
conditions of the treatment situation, formerly
inhibited thoughts are gradually restored to
awareness. In recent years, however, many
therapists have become increasingly skeptical

about the value of insights regarding hypothetical
psychodynamic events. The ethical and empirical
questions that have been raised with respect to
interpretive modes of therapy would apply equally
to behavioral approaches if these used interview

procedures similarly to teach clients to construe
their psychological functioning in behavioral

terms and did not effect any significant changes in
the personality problems for which the clients
sought aid.
Although insight into presumed psychic
determinants of interpersonal responses is of
questionable validity and has little effect on
behavior, considerable experimental evidence,

which will be reviewed in the concluding chapter,
suggests that awareness of response-
reinforcement contingencies can markedly
influence overt performances. Unlike the arbitrary
and enigmatic nature of psychodynamic events,

the controlling function of environmental
contingencies is readily demonstrable and
amenable to testing and verification.
BEHAVIORAL OBJECTIVES AND “POSITIVE MENTAL

HEALTH”
Discussions of psychotherapeutic and
socialization practices customarily decry the lack
of consensus among social scientists as to what

constitutes “positive mental health.” Underlying
this concern for agreement is the belief that

behavioral change principles cannot be judiciously
applied until an adequate conception of mental
health and the nature of the “good life” is
developed. The fact that a universal conception of

mental health would require value
standardization is usually obscured by the
abstract nature of the discourse. On the other
hand, when the issues are cast in a more specific

form, it becomes apparent that the search for
uniform criteria of “good” functioning is not only a
fruitless pursuit; it is also one that would raise
serious ethical concerns if standards were ever
officially adopted and imposed on the populace.

Who is to prescribe what is the “healthiest”
occupational activity, the “healthiest” political or

religious belief, the “healthiest” style of living, the
“healthiest” form of marital or social relationships,

or the “healthiest” artistic preferences?
People differ widely across social groups and

over time in their views of the ideal pattern of life.
Indeed, as noted in the introductory chapter,
modes of behavior that are judged abnormal and a
source of distress in one social group may be

regarded as commendable and emulative in
another subculture. In a society that values
individualism the “good life” may assume a wide
variety of acceptable patterns. Although some
common elements might be abstracted from the

heterogeneity, the distillation would most likely
yield a set of general bland attributes. Social
scientists can make their greatest contributions in
the ethical domain by assessing the consequences
of different styles of life. Such information would

provide others with useful bases for making value
choices.
Decision Processes in the Selection of

Objectives
A frequent objection to behavioral approaches
is that the people are often unaware that their
behavior is being modified, and verbal
conditioning studies are typically cited as
evidence. This portrayal of controlling power may

be flattering, but in fact it is exceedingly difficult to
influence the behavior of another person without
his awareness and concurrence. Indeed, as has
been pointed out elsewhere (Bandura, 1962),
verbal conditioning experiments actually

demonstrate the relative weakness of subtle
influence attempts. In the typical verbal
conditioning study, the response class to be
modified is not identified for the subject and the
experimenter purposely employs subtle verbal

and nonverbal reinforcers (e.g., “good,” “right,”
nods, smiles, and other gestures) so that the

subject will have difficulty in recognizing the
response-reinforcement contingency. Under these
circumstances subjects who discern the basis

upon which reinforcement is administered
produce incremental changes in the critical
responses, whereas those who remain unaware
generally show no conditioning effects at all. If, on

the other hand, the experimenter were to select
attractive incentives and specified what behavior

would be rewarded, it is safe to predict that
subjects would produce the desired responses at
asymptotic level almost instantaneously.
The psychological fascination with subtle and

disguised social influence processes, and the
comparative ineffectiveness of these procedures,
are also demonstrated by the short-lived interest
in experimentation on subliminal perception. The

initial studies generated considerable public alarm
that behavioral scientists had paved a freeway to
the “unconscious mind,” thus providing hidden
persuaders of Madison Avenue a means of
trafficking in subliminal messages that could
shape and control the interests, attitudes, and
social actions of persons without their awareness.
This picture is further reinforced by popular
descriptions of the potentialities of psychological

control conjuring up macabre associations of 1984
and Brave New World, in which people are
dominated by occult technocrats who possess
awesome methods of behavioral control. Some
state legislatures even enacted laws designed to

control the potential controllers. Research
evidence, as usual, introduced a sobering note into
extravagant fantasies. Investigations of subliminal
stimulation clearly showed that stimuli at
supraliminal levels have more pronounced effects

.upon subjects’ behavior than stimuli that are
below the threshold of awareness (McConnell,

Cutler, & McNeil, 1958). Subliminal stimulation
either produces no behavioral changes or, at most,
weak and fragmentary ones.
Nevertheless, the conduct of change programs
in shadowy ambiguity is sometimes recommended
on the assumption that persons’ awareness of
influence attempts will not only arouse interfering

counter control behavior, but will also reduce the
potency of reinforcing stimuli. Although these
assumptions may have some validity in situations
where the influence attempts are designed
primarily to induce persons to perform actions

contrary to their interests and value systems (e.g.,
advertising, political persuasion), they are less
appropriate for situations in which the learner
selects his own objectives. In fact, awareness of
and commitment to specified outcomes that are

shared by agents of change tend to enhance
positive evaluation of change agents’ efforts and to

facilitate the acceptance of their influence.
DECISIONAL RESPONSIBILITIES OF CHANGE

AGENTS AND CLIENTS
The ethical implications of behavioral control
cannot be discussed meaningfully without
specifying the scope of decision-making behavior
of both the client and the change agent. In any type

of social influence enterprise there exist two basic
decision systems. One set of decisions pertains to
the selection of goals; these decisions require
value judgments. The second set of decisions,

which involve empirical issues, relates to the
selection of specific procedures for achieving
selected goals. In the latter domain the agent of
change must be the decision-maker, since the
client is in no position to prescribe the learning

contingencies necessary for the modification of his
behavior. But though the change agent determines

the means by which specified outcomes can be
achieved, the client should play a major role in

determining the directions in which his behavior
is to be modified. To the extent that the client

serves as the primary decision-maker in the value
domain, the ethical questions that are frequently
raised concerning behavioral control become
pseudo issues.
When the client wishes to change a limited
range of deviant behavior, the objectives are self-

evident and the change agent can proceed with
treatment as soon as the learning experiences
appropriate to the desired outcomes have been

specified. More often, however, because clients are
uncertain about the benefits they hope to derive
from treatment, or because their goals are stated
too broadly, the identification of relevant
outcomes must constitute the initial objective of

the program. In such instances it is necessary to
conduct a thorough behavioral analysis in order to
identify the social conditions governing the client’s
response patterns and the range of behavioral and
situational modifications likely to promote the
desired psychological changes. After possible
alternative courses of action and their probable
consequences are specified, the client can
participate in the selection of his treatment

outcomes. This decisional process is not unlike
medical diagnosis in which a patient desires relief
from pain but cannot specify the cause of pain or a
remedy for it. Rather, the therapist must detect the
factors producing pain and indicate the chances

for immediate and long-term benefits from
alternative remedial interventions. Once the
patient has selected one of the alternatives, he not
only expects but demands that the therapist
manipulate and control events to accomplish the

desired relief. A physician who fails to assume full
control over the progress of treatment may be

charged with malpractice. On the other hand,
serious ethical problems would arise if a patient
consulting a medical specialist were promptly

subjected to radical surgical or medical
procedures without his concurrence based on a
clear understanding of the manner in which his
physical status was to be modified. Although the

preceding example has focused on the ethical
implications of therapeutic work, analogous

decision processes and value issues are involved
when a person consults lawyers, architects,
bankers, and other societal agents who possess
the power to influence by reason of their

expertise. Until recently the major obstacle to
serious use of a decisionmaking approach such as
this in behavioral change endeavors is that the

treatment alternatives were limited and the
outcomes uncertain.
It would be naive to assume that agents of
change play no role whatsoever in the
determination of goals. In psychotherapy, for
example, in order not to influence the client’s
choice of behavior, a therapist would be forced to
conduct with aloof objectivity an exhaustive
survey of all possible alternative outcomes from
which the client could make his choice. In practice,

however, only a few feasible objectives are likely
to be examined and compared. The
psychotherapist’s value orientation may partly
determine not only the range and types of
outcomes selected for consideration but also the

relative emphasis given to the probable
consequences associated with the various
alternatives. Thus some encroachment on the
client’s decision-making primacy in the value
domain is inevitable. If the change agent’s value

preferences are explicitly identified as his
personal biases and not represented to the client

as scientific truths, this problem is much less
serious. If values were stated more explicitly,
clients would be more inclined to select therapists

on the basis of similar moral commitments and
might well be more receptive to the therapist’s
influence.
Occasionally a person may select goals that the
change agent has no desire to promote because
the intended outcomes conflict with his basic

values or he lacks skill in the methods necessary
for attaining the chosen objectives. In such cases
he may refuse to participate in the treatment or, if

the desired changes seem appropriate, he may
refer the client elsewhere.
Special problems in goal selection also arise

when persons are confused over their own values
and purposes, or when they exhibit severe deficits
in reality-oriented behavior and low capacity for
communication. It might be questioned whether
such persons are capable of selecting meaningful
objectives for themselves. Fairweather, Sanders,
Maynard, and Cressler (1969) have shown in their
work with chronic schizophrenics that such
individuals can successfully participate in the
selection of personal goals provided the

alternatives are defined in comprehensible terms
of performance and the clients are given
responsibility for decision-making that affects
their daily lives. Some grossly deviant persons, of

course, may refuse to seek modifications of any
sort. Often they constitute threats to themselves or
to the welfare of others. If such persons are
unwilling to assist in selecting treatment goals, it
does not mean that one should abandon treatment

attempts. Sometimes it is necessary to assume that
the person cannot exercise sufficient control over

his behavior and to hope that, with appropriate
interventions, the person will reach a state of

aware self-interest in which he will desire further
modifications within a broad range of societally

tolerated alternatives.
REDEFINITION OF CLIENT’S OBJECTIVES
The preceding discussion has been mainly

concerned with problems created by uncertainties
about what people wish to gain from treatment. A
far more prevalent, but largely ignored ethical
issue, is raised by therapists’ unilateral

redefinition of the goals presented by the clients.
This revision of the therapeutic contract occurs

most frequently in approaches that focus major
attention not on the behavior of the client but on
inferential inner states. The therapist usually takes

the position that the client does not know what his
real problems are and that they can be revealed
only through a protracted series of interpretive
interviews; the client’s behavioral problems are
normally underrated as superficial derivatives of

underlying conditions that are believed to be most
effectively modified through the achievement of

insight. After restructuring the central problem
the therapist pursues objectives that are often
quite different from those originally sought by the
client. If the client has been sufficiently convinced

that he is resolving more generic problems his
behavioral difficulties assume secondary
importance in the course of therapy, so that even if
they are not modified, he supposes the contract to
be fulfilled. Insight has been attained.
A therapeutic contract involves an obligation

on the part of the therapist to modify the problems
presented by his clients. A therapist may market a
particular brand of insight without raising ethical

objections provided he adds two important
qualifiers: First, he informs his clients that the
insights they are likely to attain reflect his own

belief system and second, that attaining them is
apt to have little impact on the behavioral
difficulties that brought the client to treatment. It
is evident from the results of interpretive
approaches that a therapist who leads his clients
to believe that insight will alleviate their
behavioral malfunctioning is unlikely to

accomplish the changes he implies.
SEQUENTIAL DECISION-MAKING
Decisions about objectives are not irrevocable.

Consequences resulting from behavioral changes
representing the initially selected outcomes may

lead to revision of subsequent aims. The initial
objectives should be assigned a provisional status
in order to provide the client opportunities to

experiment with new behaviors and to experience
their consequences; then he can decide whether
he wishes to pursue further the chosen course of
action. Moreover, during the course of treatment,
previously ignored areas of behavioral functioning
may become more important than original goals.
Whenever this situation arises the treatment
program can be easily reoriented toward new
objectives and appropriate learning experiences.
By retaining flexibility in the selection, sequencing,

and timing of objectives, the treatment program
remains highly sensitive to feedback from
resultant changes and the therapist is less inclined
to invoke an extended moratorium on behavioral

modification while he searches for the
fundamental objective. Preoccupation with the
accurate identification of the core problem reflects
a remnant of the revivalist view of
psychopathology, according to which diverse

interpersonal problems are presumed to stem
from a central pathogenic experience. It is further

believed that interpretive revival and abreaction
of the core trauma will result in rapid and widely

generalized personality changes.
Contrary to the latter view, investigations of

the social-learning process (Bandura & Walters,
1963) provide considerable evidence that deviant
behavior is typically controlled by diverse
variables and is not generated by a single

pathogenic agent. Successful treatment therefore
requires the selection and attainment of a variety
of specific objectives rather than a single omnibus
outcome. The extent to which changes in one
system of behavior affect other areas of

functioning will be partly determined by the
similarity of the two systems and by the degree to
which the altered behavior brings the client into
contact with new role models and with new
patterns of reinforcement.
SELECTION OF CHANGE AGENTS AND THE LOCUS

OF TREATMENT
After the goals and requisite learning

experiences have been established, another set of
decisions arises in the selection of change agents

who, by virtue of their specialized training or close
relationship with the client, are best suited to
implement treatment procedures. In traditional
clinical practice, changes in behavior are

characteristically effected by professional
psychotherapists in office settings, mainly through
the modification of verbal-symbolic contents.
Although the decided preference for artificial

environments and symbolic substitutes for
naturally occurring events has been theoretically
justified, these treatment conditions were
probably adopted more for the therapists’
convenience than for any proven therapeutic

superiority. In fact, results of controlled studies
demonstrate that deviant behavior can be

modified more thoroughly and more expeditiously
by treating actual events rather than their

symbolic equivalents (Bandura, Blanchard, &
Ritter, 1969), and that change programs

conducted in natural settings are far superior to
similar ones administered in psychiatric
institutions (Fairweather, et al., 1969).
It follows from principles of generalization that
the optimal conditions for effecting behavioral
changes, from the standpoint of maximizing

transfer effects, would require people to perform
the desired patterns of behavior successfully in
the diverse social situations in which the behavior

is most appropriate. On the other hand, when
treatment is primarily centered around verbal
responses expressed in an invariant, atypical
context one cannot assume that induced changes
will necessarily generalize to real-life

performances to any great extent.
Issues regarding the locus and content of
treatment are closely linked with the choice of
change agents. From a social-learning perspective
those who have the most intensive contact with
the client, if given appropriate training, can serve
as the most powerful agents of change. Their
potential efficacy derives from the fact that in such

positions they exercise considerable control over
the very conditions that regulate the behavior.
Successful applications of this general principle
are provided in new approaches to child therapy

in which parents are utilized in the treatment of
their own children’s behavior (Hawkins, Peterson,
Schweid, & Bijou, 1966; O’Leary, O’Leary, &
Becker, 1967; Patterson, Ray, & Shaw, 1968; Risley
& Wolf, 1966; Russo, 1964; Wahler, Winkel,

Peterson, & Morrison, 1965; Williams, 1959).
In a well-designed program a thorough

behavioral analysis is first conducted to identify
the social conditions that maintain the various

behavior disorders. Next the deviant response
patterns to be eliminated and the desirable

behaviors to be strengthened are clearly specified.
The parents are then given a detailed description
of how they must alter their characteristic ways of
reacting to their child’s behavior to achieve

therapeutic changes. This typically involves a
reversal of parents’ differential reinforcement
practices. Whereas the child’s deviant behavior
previously commanded attention and his desirable
behavior received little special notice, the parents

are advised now to ignore or to reinforce
negatively his aberrant behavior and to respond
positively to the forms of behavior they wish to
promote. In the case of deficit problems (Lovaas,
1966), a program of graduated modeling is also

devised, while in fear-motivated disorders
(Bentler, 1962) a graduated reexposure to

threatening situations is implemented by the
parents.
It should be noted in this context that attempts
to modify behavior through giving advice, have an
extended history, mainly negative. Its paltry
outcomes probably result from the nature of the
advice given and from the fact that instructions

alone are of limited effectiveness unless they are
combined with other procedures that help to alter
and to support parental behavior. Parents may
understand principles of change but may have
difficulty translating them into appropriate

actions. To alleviate this problem, not only
treatment strategies are sketched out in
considerable detail, but initially the recommended
practices are modeled by the person planning the
program while parents observe the interactions.

After the procedures have been adequately
demonstrated and some control of the child’s

deviant behavior has been achieved, the parents
gradually take over the therapeutic function. The
parents are directly supervised until they attain

proficiency in handling their child’s behavior
without external direction.
Detailed instructions, combined with

demonstrations and supervised practices, are
effective means of instituting changes in parental
behavior, but favorable outcomes are necessary to

ensure adherence to the recommended practices.
The problem of parental reinforcement is
particularly critical in initial stages of treatment

when withdrawal of the positive consequences
that had been periodically evoked by the child’s
deviant behavior often produces a temporary
increase in such behavior. During this period it
may be necessary to provide extensive social

support to maintain the desired parental behavior.
In later phases beneficial changes in the child
serve as a natural and powerful source of reward
for the parents’ efforts so that the new familial
patterns become reciprocally reinforcing and
thereby self-sustaining. At times it may be difficult
for parents to carry out the necessary programs
because of social conditions, independent of the
child, that affect their behavior. Such hindrances

can be most successfully overcome by modifying
the conflicting influences impinging upon the
parents.
When a child’s deviant behavior is sufficiently
prevalent to occur frequently within a clinical

setting, parents may gain facility through
supervised consultation sessions on treatment
strategies that they can apply at home. On the
other hand, in instances where the major

behavioral problems are not readily reenacted at a
clinic, the change process can be most effectively
initiated in the home with the parents functioning

as therapists. The feasibility of the home
treatment approach has been demonstrated by
Hawkins et al. (1966).
Their illustrative case involved a four-year-old
boy who aggressively demanded constant
attention, often behaved in a physically abusive

and belligerent manner, and generally was
extremely difficult to manage. After a baseline

measurement of the incidence of hyperaggressive
behavior was made, the treatment program was
initiated. The mother was instructed to go about
her usual household activities and whenever her

son displayed behavior that required handling, the
observer would signal one of three modes of
response. Each time the boy behaved

reprehensibly the mother was advised either to
tell him to stop or to place him in his room for a

brief time. In contrast, when he behaved
commendably the mother was encouraged to
express interest and approval. As shown in Figure

2-2, the new reinforcement practices produced a
marked decrease in undesirable behavior. In the
Figure 2-2. Number of 10-second intervals in which the boy
displayed objectionable behavior during each one-hour
session. Hawkins et al., 1966.
next phase the mother was asked to resume her
original practices of chastising undesirable

behaviors while ignoring desirable ones, but she
found it difficult to recapture her former style. The
therapeutic contingencies were again reinstated,

and a follow-up study was conducted
approximately a month later in which the mother-

child interaction was observed for several sessions
without airy further guidance. The overall results
show not only that the mother maintained the
favorable changes in her son’s behavior long after
the supervising therapist had dropped out of the

picture, but that the boy generally behaved in a
more considerate and affectionate manner, which

contrasted markedly with his former
indiscriminate belligerence. As Hawkins points
out, a major benefit of enlisting parents as change
agents is that, having gained facility in effective

treatment methods, they can successfully apply
them to future developmental problems in a
variety of circumstances.
Although the discussion thus far has
concentrated on the implementation of change

programs by parents, the same general principles
apply when other change agents perform similar
functions. The direction of change must be defined
in terms of observable behavior; the methods for

achieving these outcomes must be clearly
specified and preferably modeled; enough

guidance must be provided to ensure success; and,
if necessary, special favorable consequences of
carrying out the recommended practices must be

arranged. Behavioral approaches, as will be shown
later, use teachers, nurses, peers, and students
extensively as agents of change under the
guidance of persons who possess professional

knowledge and competencies in principles of
behavioral change. To some extent, also,
individuals are called upon to function as their
own change agents by learning how to manage
contingencies and self-reinforcing consequences

in order to modify their own behavior in desired
directions. Nonprofessionals are frequently
selected to implement change programs, not just
as an economical way of alleviating serious
manpower shortages, but because they are in a

more advantageous position to effect better
outcomes than professionals, who may have only
brief contact with the client in an artificial setting
in which the deviant behavior is infrequently
displayed. When behavior is modified in the

natural social environment by persons who
normally exercise some control over the behavior,

the problems of induced changes failing to
generalize or to be sustained over time are much
less likely to arise.
In many behavioral change programs, the
supervisory staff instructs change agents on how
to implement selected procedures but fails to
demonstrate the desired practices or to arrange

for favorable consequences for their endeavors.
Since new behavioral practices often require
change agents to devote increased attention to the
persons whose behavior is being modified and to
discard old routines that had some functional

value, some resistance is to be expected. In the
initial phase of a project by Ayllon and Azrin
(1964) designed to restore self-care in chronic
schizophrenics it was noted, for example, that
hospital attendants often failed to put the

designated procedures into effect even though
they had repeatedly been instructed to do so. Only

after the attendants were provided feedback about
their own performances and social consequences
for their own behavior did they faithfully carry out

the prescribed program.
The efforts of change agents are reinforced and
maintained to some extent by positive experiences
resulting from favorable changes in the behavior

of their clients. In fact, some investigations
(Hawkins et al., 1966; Wahler & Pollio, 1968) have

encountered difficulties in employing the
intrasubject replication design to dramatize the
functional relationships between behavior and its
consequences because, after experiencing the

benefits from the behavioral changes they
produced in their children, the parents were
exceedingly reluctant to revert to their former

reinforcement practices. However, when the
required treatment conditions are difficult to

create and to sustain, when the rate of
improvement is relatively slow or evidence of
progress has weak reinforcing value, it is desirable

to provide adequate rewards for change agents as
well. For example, to enhance the performance of
remedial instructors, Wolf, Giles, and Hall (1968)
created a bonus monetary contingency that was
linked to their students’ productivity. The
provision of appropriate supports for the agent’s

behavior, which is a critical aspect of behavioral
change programs, is usually given little attention,
with the consequence that essential procedures

are halfheartedly or only sporadically applied. Any
temporary suspension of contingencies,
particularly in initial phases of a program, usually
results in intermittent reinforcement of the

undesired behavior. Therefore, treatment
programs should not be attempted unless the
appropriate contingencies will be systematically

applied.
ETHICAL ISSUES IN SOCIAL AND CULTURAL

CHANGE
Most of the preceding discussion of goal
selection was primarily concerned with the
achievement of behavioral changes on an
individual basis. It is generally acknowledged that
many of the problems confronting a society cannot

be solved at an individual level but necessitate
changes in entire social systems.
A variety of situations in which new

contingencies are introduced on a society-wide
basis raise important questions about the morality
and decision processes guiding instituted changes.

In cases involving widespread deviant behavior,
such as delinquency or prevalent deficit conditions
resulting from impoverished environments, major
social changes are required for rehabilitation. For
example, attempts to reduce the incidence of

antisocial behavior by treating individual
members who happen to be apprehended is a

futile endeavor. Group problems demand group
solutions. New social environments involving

appropriate contingencies, role models, and
incentives, must be created if constructive modes
of behavior are to be established and normatively
sanctioned.
As knowledge accumulates about the causes
and consequences of different social patterns and
efficacious principles of behavioral change are
further developed, a society gains the means not
only of preventing the development of major

social problems but also of realizing its avowed
aims. Preventive programs and improved systems
of social life entail new social practices, some of
which may clash with the ideologies and traditions

of various interest groups. Ethical controversies,
therefore, inevitably arise over the types of social
changes advocated as well as the methods by

which they are to be achieved.
The value conflicts resulting from intrasocietal
pressures for change occur on a much broader
scale in cross-cultural ventures in which one
society strives to introduce new patterns of
behavior into other societies occupying

subordinate positions. In many cases advocated
changes involving preventive medical practices,
reorganization of economic and agricultural
systems, creation of educational programs, and

introduction of technologies that release people
from demeaning labor have the potentiality for
enriching social life and enhancing human
freedom. Although the changes may have
beneficial outcomes, they often require radical

modifications of established beliefs and ways of
living and are therefore understandably opposed.

Moreover, intersocietal attempts at influence
typically involve the export not only of better

means of achieving cultural aims, but also of new
ideologies and ultimate ends themselves. It is

primarily the imposition of new moral standards,
some of which may be dysfunctional in the foreign
setting, and the external prescription of how
people within another culture should live their

lives, that give rise to ethical concerns.
The decision processes and value issues

involved in the selection of group goals are, in
many respects, similar to those that operate at the
individual level. First, it is necessary to decide

what social objectives from among a variety of
alternatives shall be pursued. The major question
here is whether the authority for goal selection
resides in a political or technological elite or is
determined through informed collaborative

participation of those whose lives will be affected
by whatever policies are adopted. If one seriously
subscribes to the value of group determination of
social objectives, then more attention must be
given to developing optimal methods for clarifying
the consequences associated with different value
choices, for ascertaining collective preferences,
and for resolving conflicts among different
interest groups. In addition, adequate safeguards

and social supports must be provided for
warranted attempts at personal influence of social
policies. Looking into the not-too-distant future,
Hofstadter (1967), for example, envisions the use
of computer technology, in which individual voting

devices are connected to computers which
assemble data almost instantaneously, to permit
greater individual participation in society’s
decision-making whenever feasible.
Under extensive bureaucratization, which

effectively obscures decision-making
responsibilities, most people come to feel that they

can exert little positive control over their
environment. Consequently they are inclined to
respond with grudging acquiescence to major
social changes that are often guided by economic
considerations, slide rule decrees, and political
expedience. Persons who are more actively
inclined are often thwarted by the lack of readily

accessible means of affecting decisions about the
cultural priorities that should be promoted.
However, the recent years have witnessed
vigorous demands, particularly among the
younger members of society, for a greater role in

making decisions that affect the course and quality
of their life.
Value conflicts arise not only in formulating
common goals, but also in selecting methods for

inducing preferred changes. In one way or another
decisions are made about how much social
objectives are advanced through coercive

methods, through positive reinforcement of
appropriate behaviors, or through provision of
models for emulation who exemplify the desired
behavioral patterns.
The notion of planned social change is likely to
arouse in people’s minds negative associations of
regimentation, invasion of privacy, and

curtailment of self-determination. In fact, as Benne
(1949) and Mannheim (1941) have cogently

argued, collectively planned social change, rather
than being anti-individualistic, generally
safeguards and extends human freedom. The need
for social planning stems from the fact that, in

many areas of behavioral functioning, people’s
outcome experiences are jointly determined by
each other’s actions. Thus if motorists did not have

the benefit of traffic codes they would repeatedly
obstruct and injure one another, whereas agreeing

to a few sensible regulations greatly enhances
their personal welfare and freedom of movement.
Without some social controls over human

behavior, personal freedoms would be
continuously in jeopardy. Paradoxically, zealous
individualists often attack the very social
institutions that are established to protect
freedom of self-expression.
Problems of dysfunctional restraints often
occur when social control is unwisely extended to

areas of functioning that do not involve
interdependent consequences to any significant
degree. Unconventional beliefs, styles of living,

and personal habits may be negatively sanctioned
even though these activities, apart from their
minor irritant value, rarely affect the welfare of
others. Such pressures toward the standardization
of life do constitute threats to personal freedom.
Summary
One of the major obstacles to the development
of effective change programs arises from the
failure to specify precisely what is to be
accomplished, or the more common practice of
defining the intended goals in terms of
hypothetical internal states. When the aims
remain ambiguous, learning experiences are

haphazard, and whatever procedures are
consistently applied tend to be determined more
by personal preferences of change agents than by
clients’ needs.
The appropriate methods and learning
conditions for any given program of behavioral

change cannot be meaningfully selected until the
desired goals have been clearly defined in terms of
observable behavior. Rapid progress is further
assured by designating intermediate objectives,
which delineate optimal learning sequences for

establishing the component behaviors of more
complicated social performances. The necessity

for behavioral specification of objectives is most
clearly illustrated in the case of complex patterns

of behavior which cannot be achieved with any
degree of success until they are analyzed into
essential constituent functions.
The selection of goals involves value choices.

To the extent that people assume major
responsibility for deciding the direction in which
their behavior ought to be modified, the frequently
voiced concerns about human manipulation
become essentially pseudo issues. The change

agent’s role in the decision process should be
primarily to explore alternative courses of action
available, and their probable consequences, on the
basis of which clients can make informed choices.

However, a change agent’s value commitments
will inevitably intrude to some degree on the goal
selection process. These biases are not necessarily

detrimental, provided clients and change agents
subscribe to similar values and the change agent
identifies his judgments as personal preferences
rather than purported scientific prescriptions.
Much more serious from an ethical standpoint is
the unilateral redefinition of goals by which
psychotherapists often impose insight objectives

(which mainly involve subtle belief conversions)
upon persons desiring changes in their behavioral
functioning.
Behavioral problems of vast proportions can
never be adequately eliminated on an individual

basis but require treatment and prevention at the
social systems level. As behavioral science makes
further progress toward the development of
efficacious principles of change, man’s capacity to

create the type of social environments he wants
will be substantially increased. The decision
processes by which cultural priorities are

established must, therefore, be made more explicit
to ensure that “social engineering” is utilized to
produce living conditions that enrich life and
behavioral freedom rather than aversive human
effects. Control over value choices at the societal
level can be increased by devising new systems of
collective decision-making which enable members

to participate more directly in the formulation of
group objectives.
In discussions of the ethical implications of
different modes of achieving personality changes,
commentators often mistakenly ascribe a negative

morality to behavioral approaches, as though this
were inherent in the procedures. Social-learning
theory is not a system of ethics; it is a system of
scientific principles that can be successfully

applied to the attainment of any moral outcome. In
actuality, because of their relative efficacy,
behavioral approaches hold much greater promise

than traditional methods for the advancement of
self-determination and the fulfillment of human
capabilities. If applied toward the proper ends,
social learning methods can quite effectively
support a humanistic morality.
REFERENCES
Alexander, F. Psychoanalysis and psychotherapy;

developments in theory, techniques, and training.
New York: Norton, 1956.
Alexander, F. The dynamics of psychotherapy in the

light of learning theory. American Journal of
Psychiatry, 1963, 120, 440-448.
Asch, S. E. Social psychology. New York: Prentice-Hall,

1952.
Ayllon, T., & Azrin, N. H. Reinforcement and

instructions with mental patients. Journal of
Experimental Analysis of Behavior, 1964, 7, 327-
331.
Bandura, A. Social learning through imitation. In M. R.

Jones (Ed.), Nebraska symposium on motivation:
1962. Lincoln: University of Nebraska Press,
1962. pp. 211-269.

experimental social psychology. Vol. II. New York:
Bandura, A. Social-learning theory of identificatory
processes. In D. A. Goslin (Ed.), Handbook of
socialization theory and research. Chicago: Rand
McNally, 1968.

13, 173-199.
Bandura, A., Grusec, J. E., & Menlove, F. L. Vicarious

extinction of avoidance behavior. Journal of
Bandura, A., & Harris, M. B. Modification of syntactic

style. Journal of Experimental Child Psychology,
1966, 4, 341-352.
Bandura, A., & Huston, A. C. Identification as a process

of incidental learning. Journal of Abnormal and
Social Psychology, 1961, 63, 311-318.

Bandura, A., Lipsher, D. H., & Miller, P. E.

Psychotherapists’ approach- avoidance reactions
to patients’ expressions of hostility. Journal of
Bandura, A., & McDonald, F. J. The influence of social

reinforcement and the behavior of models in
shaping children’s moral judgments. Journal of
281.
Bandura, A., & Mischel, W. The influence of models in

modifying delay of gratification patterns. Journal
705.
Bandura, A., Ross, D., & Ross, S. A. Imitation of film-

mediated aggressive models. Journal of Abnormal

New York: Ronald, 1959.
Bandura, A., & Walters, R. H. Social learning and

personality development. New York: Holt,
Rinehart & Winston, 1963.
Benne, K. D. Democratic ethics in social engineering.

Progressive Education, 1949, 26,201-207.
Bentler, P. M. An infant’s phobia treated with

reciprocal inhibition therapy. Journal of Child
Psychology and Psychiatry, 1962, 3, 185-189.
Berg, C. Deep analysis. New York: Norton, 1947.
Berg, J. A., & Bass, B. M. (Eds.) Conformity and

deviation. New York: Harper, 1961.
Bergin, A. E. The effect of dissonant persuasive

communications upon changes in a self-referring
attitude. Journal of Personality, 1962, 30, 423-38.
Brown, P., & Elliott, R. Control of aggression in a

nursery school class. Journal of Experimental
Child Psychology, 1965, 2, 103-107.
Chittenden, G. E. An experimental study in measuring

and modifying assertive behavior in young
children. Monographs of the Society for Research
in Child Development, 1942, 7 (1, Serial No. 31).
Chu, C. G. Fear arousal, efficacy and imminency.

Journal of Personality and Social Psychology, 1966,
4, 517-524.
Colby, K. M. On the greater amplifying power of

causal-correlative over interrogative inputs on
free association in an experimental
psychoanalytic situation. Journal of Nervous and
Mental Diseases, 1961, 133, 233-239.
Collier, R. M. A scale for rating the responses of the

psychotherapist. Journal of Consulting Psychology,
1953, 17, 321-326.
Dabbs, J. M., Jr., & Leventhal, H. Effects of varying the
recommendations in a fear-arousing
communication. Journal of Personality and Social
Psychology, 1966, 4, 525-531.
Dittes, J. E. Galvanic skin responses as a measure of

patient’s reaction to therapist’s permissiveness.
Journal of Abnormal and Social Psychology, 1957,
55, 295-303.
Dittmann, A. T. The interpersonal process in

psychotherapy: development of a research
method. Journal of Abnormal and Social
Psychology, 1952, 47,236-244.
Elkin, F., & Westley, W. A. The myth of adolescent

culture. American Sociological Review, 1955, 20,
680-684.
Fairweather, G. W., Sanders, D. H., Maynard, H., &

Cressler, D. L. Community life for the mentally ill:
an alternative to institutional care. Chicago:
Aldine, 1969.
Fenichel, O. Problems of psychoanalytic technique. New

York: Psychoanalytic Quarterly, 1941.
Ferster, C. B., Nurnberger, J. I., & Levitt, E. B. The

control of eating. Journal of Mathetics, 1962, 1,
87-109.
Frank, G. H., & Sweetland, A. A study of the process of
psychotherapy: The verbal interaction. Journal of
Gelfand, D. M., Gelfand, S., & Dobson, W. R.

Unprogrammed reinforcement of patients’
behavior in a mental hospital. Behaviour Research
and Therapy, 1967, 5, 201-207.
Goldman, J. R. The relation of certain therapist

variables to the handling of psychotherapeutic
events. Unpublished doctoral dissertation,
Stanford University, 1961.
Grossberg, J. M. Successful behavior therapy in a case

of speech phobia (“stage fright”). Journal of
Speech and Hearing Disorders, 1965, 30, 285-288.
Harris, F. R., Wolf, M. M., & Baer, D. M. Effects of adult

Children, 1964, 20, 8-17.
Harris, M. B. A self-directed program for weight

control: A pilot study. Journal of Abnormal
Psychology, 1969, 74, 263-270.
Hart, B. M., Reynolds, N. J., Baer, D. M., Brawley, E. R., &

Harris, F. R. Effect of contingent and non-
contingent social reinforcement on the
cooperative play of a preschool child. Journal of
Applied Behavior Analysis, 1968, 1, 73-76.
Harway, N. I., Dittmann, A. T., Raush, H. L., Bordin, E.

S., & Rigler, D. The measurement of depth of
interpretation. Journal of Consulting Psychology,
1955,19, 247-253.
Hastorf, A. H. The “reinforcement” of individual

actions in a group situation. In L. Krasner & L. P.
268-284.
Hawkins, R. P., Peterson, R. F., Schweid, E., & Bijou, S.

W. Behavior therapy in the home: Amelioration
of a problem in parent-child relations with the
parent in a therapeutic role. Journal of
Experimental Child Psychology, 1966, 4, 99-107.
Heine, R. W. A comparison of patients’ reports on

psychotherapeutic experience with
psychoanalytic, nondirective, and Adlerian
therapists. American Journal of Psychotherapy,
1953, 7, 16-23.
Henker, B. A. The effect of adult model relationships

on children’s play and task imitation. Dissertation
Abstracts, 1964, 24, 4797.
Herzberg, A. Active psychotherapy. New York: Grune &
Stratton, 1945.
Hofstadter, R. The instant electorate. Commencement

address presented at Carlton University, Ottawa,
Canada, May 1967.
Homme, L. E., & Tosti, D. T. Contingency management

and motivation. Journal of the National Society for
Programmed Instruction, 1965, 4, 14-16.
Hovland, C. I., Janis, I. L., & Kelley, H. H.

Communication and persuasion: Psychological
studies of opinion change. New Haven: Yale
University Press, 1953.
Jack, L. M. An experimental study of ascendant

behavior in preschool children. University of Iowa
Studies in Child Welfare, 1934, 9, 3-65.
Jourard, S. M. On the problem of reinforcement by the

psychotherapist of healthy behavior in the
patient. In F. J. Shaw (Ed.), Behavioristic
approaches to counseling and psychotherapy: A
Southeastern Psychological Association
Symposium. University, Ala.: University of
Alabama Press, 1961.
Keister, M. E. The behavior of young children in

failure: An experimental attempt to discover and
to modify undesirable responses of preschool
children to failure. University of Iowa Studies in
Child Welfare, 1938, 14, 27-82.
Kelman, H. C. Manipulation of human behavior: An

ethical dilemma for the social scientist. Journal of
Social Issues, 1965, 21, 31-46.
Klein, M. The psychoanalysis of children. Translated by

Alix Strachey. New York: Grove, 1960.
London, P. The modes and morals of psychotherapy.

New York: Holt, Rinehart and Winston, 1964.
Lovaas, O. I. A program for the establishment of

speech in psychotic children. In J. K. Wing (Ed.),
Early childhood autism. New York: Pergamon,
1966.
Lovaas, O. I. A behavior therapy approach to the

treatment of childhood schizophrenia. In J. P. Hill
(Ed.), Minnesota symposia on child psychology,
Vol. 1. Minneapolis: University of Minnesota
Press, 1967. pp. 108-159.
Lovaas, O. I., Freitag, G., Gold, V. J., & Kassorla, I. C.

Experimental studies in childhood schizophrenia:
Analysis of self-destructive behavior. Journal of
McConnell, J. V., Cutler, R. L., & McNeil, E. B. Subliminal

stimulation: An overview. American Psychologist,
1958, 13, 229-242.
Mager, R. F. Preparing objectives for programmed

instruction. San Francisco: Fearon, 1961.
Mahrer, A. R. The goals of psychotherapy. New York:

Appleton-Century- Crofts, 1967.
Mannheim, K. Man and society in an age of

reconstruction. New York: Harcourt, Brace &
World, 1941.
Marmor, J. Psychoanalytic therapy as an educational

process: Common denominators in the
therapeutic approaches of different
psychoanalytic schools. In J. H. Masserman (Ed.),
Science and psychoanalysis. Vol. 5. Psychoanalytic
education. New York: Grune & Stratton, 1962. pp.
286- 299.
Murray, E. J. A content-analysis method for studying

psychotherapy. Psychological Monographs, 1956,
70 (13, Whole No. 420).
Mussen, P. H., & Parker, A. L. Mother-nurturance and

girls’ incidental imitative learning. Journal of
O’Leary, K. D., O’Leary, S., & Becker, W. C. Modification

of a deviant sibling interaction pattern in the
home. Behaviour Research and Therapy, 1967, 5,
113-120.
Page, M. L. The modification of ascendant behavior in

preschool children. University of Iowa Studies in
Child Welfare, 1936, 12, 7-69.
Patterson, C. H. Control, conditioning, and counseling.

Personnel and Guidance, 1963, 41, 680-686.
Patterson, G. R., & Brodsky, G. A behavior modification

programme for a child with multiple problem
behaviors. Journal of Child Psychology and
Psychiatry, 1967, 7, 277-295.
Patterson, G. R., Ray, R., & Shaw, D. Direct intervention

in families of deviant children. Unpublished
manuscript, University of Oregon, 1968.
Pentony, P. Value change in psychotherapy. Human

Relations, 1966, 19, 39-46.
Pratt, C. C. The logic of modern psychology. New York:

Macmillan, 1939.
Risley, T., & Wolf, M. M. Experimental manipulation of

autistic behavior and generalization into the
home. In R. Ulrich, R. Stachnik, & J. Mabry (Eds.),
Control of human behavior. Glenview, Ill.: Scott,
Foresman, 1966. pp. 193-198.
Rogers, C. R. Client-centered therapy. Boston:
Houghton Mifflin, 1951.
Rogers, C. R., & Skinner, B. F. Some issues concerning

the control of human behavior: A symposium.
Science, 1956, 124, 1057-1066.
Rosen, J. N. Direct analysis: Selected papers. New York:

Grune and Stratton, 1953.
Rosenthal, D. Changes in some moral values following

1955, 19, 431-436.
Rosenthal, T. L. Anxiety-proneness and susceptibility

to social influence. Unpublished doctoral
dissertation, Stanford University, 1963.
Russo, S. Adaptations in behavioural therapy with

children. Behaviour Research and Therapy, 1964,
2, 43-47.
Shoben, E. J., Jr. The therapeutic object: Men or

machines? Journal of Counseling Psychology,
196.3, 10, 264-268.
Speisman, J. C. Depth of interpretation and verbal

resistance in psychotherapy. Journal of Consulting
Psychology, 1959, 23, 93-99.
Stuart, R. B. Behavioral control of overeating.

Behaviour Research and Therapy, 1967, 5, 357-
365.
Truax, C. B. Reinforcement and nonreinforcement in

Rogerian psychotherapy. Journal of Abnormal
Psychology, 1966, 71, 1-9.
Wahler, R. G., & Pollio, H. R. Behavior and insight: A

case study in behavior therapy. Journal of
Experimental Research in Personality, 1968, 3, 45-
56.
Wahler, R. G., Winkel, G. H., Peterson, R. F., &

Morrison, D. C. Mothers as behavior therapists for
their own children. Behaviour Research and
Therapy, 1965, 3, 113-124.
Wallerstein, R. S. The problem of the assessment of

change in psychotherapy. The International
Journal of Psycho-analysis, 1963, 44, 31-41.
Walters, R. H., & Brown, M. Studies of reinforcement

of aggression: III. Transfer of responses to an
interpersonal situation. Child Development, 1963,
34, 563-571.
Williams, C. D. The elimination of tantrum behavior by

extinction procedures. Journal of Abnormal and
Social Psychology, 1959, 59, 269.
Winder, C. L., Ahmad, F. Z., Bandura, A., & Rau, L. C.

Dependency of patients, psychotherapists’
responses, and aspects of psychotherapy. Journal
of Consulting Psychology, 1962, 26, 129-134.
Wolf, M. M., Giles, D. K., & Hall, R. V. Experiments with

token reinforcement in a remedial classroom.
Behaviour Research and Therapy, 1968, 6, 51-64.
Wolpe, J. Psychotherapy by reciprocal inhibition.

Stanford: Stanford University Press, 1958.
Zilboorg, G. The emotional problem and the

therapeutic role of insight. Psychoanalytic
Quarterly, 1952, 21, 1-24.
3
Modeling and Vicarious

Processes
One of the fundamental means by which new
modes of behavior are acquired and existing
patterns are modified entails modeling and
vicarious processes. Indeed, research conducted
within the framework of social-learning theory
(Bandura, 1965a; Bandura & Walters, 1963)

demonstrates that virtually all learning
phenomena resulting from direct experiences can
occur on a vicarious basis through observation of

other persons’ behavior and its consequences for
them. Thus, for example, one can acquire intricate
response patterns merely by observing the

performances of appropriate models; emotional
responses can be conditioned observationally by
witnessing the affective reactions of others
undergoing painful or pleasurable experiences;
fearful and avoidant behavior can be extinguished

vicariously through observation of modeled
approach behavior toward feared objects without

any adverse consequences accruing to the
performer; inhibitions can be induced by
witnessing the behavior of others punished; and,
finally, the expression of well-learned responses

can be enhanced and socially regulated through
the actions of influential models. Modeling
procedures are, therefore, ideally suited for
effecting diverse outcomes including elimination
of behavioral deficits, reduction of excessive fears

and inhibitions, transmission of self-regulating
systems, and social facilitation of behavioral
patterns on a group-wide scale.
Vicarious phenomena are generally subsumed
under a variety of terms. Among those in common
usage are “modeling,” “imitation,” “observational
learning,” identification,” “copying,” “vicarious

learning,” “social facilitation,” “contagion,” and
“role-playing.” In personality theory identification
has been most frequently differentiated from
imitation on the assumed basis that imitation
involves the reproduction of discrete responses,

whereas identification involves the adoption of
either diverse patterns of behavior (Kohlberg,
1963; Parsons, 1955; Stoke, 1950), symbolic
representations of the model (Emmerich, 1959),
or similar meaning systems (Lazowick, 1955).

Sometimes the distinction is made in terms of
differential antecedent or maintaining conditions

as illustrated by Parsons’ (1951) view that “a
generalized cathectic attachment” is a prerequisite
for identification but is unessential or absent in

the case of imitation. Kohlberg (1963), on the
other hand, reserves the term “identification” for
matching behavior that is presumed to be
maintained by the intrinsic reinforcement of

perceived similarity, and employs the construct
“imitation” for instrumental responses supported

by extrinsic rewards. Others define imitation as
matching behavior occurring in the presence of
the model, reserving identification for
performance of the model’s behavior in the latter’s

absence (Kohlberg, 1963; Mowrer, 1950). Not only
is there little consensus with respect to
differentiating criteria, but some theorists assume

that imitation produces identification, whereas
others contend, with equally strong conviction,

that identification results in imitation.
Unless it can be shown that vicarious learning
of different classes of matching behavior is
governed by separate variables, distinctions
proposed in terms of the types of emulated
responses not only are gratuitous but also cause
unnecessary confusion. Limited progress would be
made in elucidating behavioral change processes

if, for example, fundamentally different learning
mechanisms were invoked, without adequate
empirical basis, to account for the acquisition of
one social response versus ten interrelated social
responses that are arbitrarily designated as

various aspects of a given role. Results of
numerous studies to be reviewed later
demonstrate that the acquisition of isolated
matching responses and of entire behavioral
repertoires is, in fact, determined by the same

types of antecedent conditions. Further, retention
and delayed reproduction of even discrete

matching responses require representational
mediation of modeling stimuli. There is also little
reason to suppose, either on empirical or

theoretical grounds, that the principles and
processes involved in the acquisition of matching
responses that are performed in the presence of
models are different from those later performed in

their absence. Indeed, if the diverse criteria
enumerated above were seriously applied, either

singly or in various combinations, in categorizing
modeling outcomes, most instances of matching
behavior that have been traditionally labeled
imitation would qualify as identification, and much

of the naturalistic data cited as evidence of
identificatory learning would be reclassified as
imitation.
It is possible, of course, to draw distinctions

among numerous descriptive terms based on
antecedent, mediating, or behavioral variables.
One might question, however, whether it is
advantageous to do so, since there is every
indication that essentially the same learning
process is involved regardless of the generality of
what is learned, of the models from whom the
response patterns are acquired, and of the
stimulus conditions under which matching

behavior is subsequently performed.
THREE EFFECTS OF MODELING INFLUENCES
To elucidate vicarious influences it is essential

to distinguish among different types of behavioral
modifications resulting from exposure to modeling

stimuli, but the differentiation must be made in
terms of more fundamental criteria than those
discussed above. There is abundant evidence

(Bandura, 1965a; Bandura & Walters, 1963) that
exposure to modeling influences has three clearly
different effects, each of which is determined by a
separate set of variables. First, an observer may
acquire new response patterns that did not
previously exist in his behavioral repertoire. In
demonstrating this observational learning or
modeling effect experimentally, it is necessary for a
model to exhibit novel responses which the
observer has not yet learned to make and which

he must later reproduce in a substantially
identical form. Any behavior that has a very low or
zero probability of occurrence in the presence of
appropriate stimuli qualifies as a novel response.
Second, observation of modeled actions and
their consequences to the performer may

strengthen or weaken inhibitory responses in
observers. These inhibitory and disinhibitory
effects are evident when the incidence of imitative
and nonmatching behavior is increased, generally
as a function of having witnessed a model

experience positive outcomes, and decreased by
having observed a model undergo punishing

consequences.
Third, the behavior of others often serves
merely as discriminative stimuli for the observer
in facilitating the occurrence of previously learned
responses in the same general class. This response
facilitation effect can be distinguished from
disinhibition and modeling by the fact that no new
responses are acquired; disinhibitory processes
are not involved because the behavior in question
is socially sanctioned and, therefore, has rarely, if
ever, incurred punishment. A simple example of
social facilitation is provided in situations where a
person gazes intently into a display window and
passersby respond in a similar manner. In the
following sections the variables and mediating
processes governing these diverse modeling
phenomena will be discussed at length. The ways
in which modeling influences can be successfully
used to effect individual and broader social
changes will also be reviewed.
Theoretical Conceptions of
Observational Learning
The earliest formulations, dating back to

Morgan (1896), Tarde (1903), and McDougall
(1908), regarded modeling as an innate
propensity. These instinctual interpretations
dissuaded empirical investigations of the
conditions under which modeling occurs; and
because of the vehement reactions against the
instinct doctrine, until recently even the
phenomena subsumed under the concept tended
to be either repudiated or widely ignored in
theoretical explanations of learning processes.
ASSOCIATIVE AND CLASSICAL CONDITIONING
THEORIES
As the instinct doctrine fell into disrepute, a
number of psychologists, notably Humphrey
(1921), Allport (1924), and Holt (1931),
accounted for modeling behavior in terms of
associative principles. Temporal contiguity
between modeling stimuli and the imitator’s
matching response was considered to be a
sufficient condition for the occurrence of
imitation. According to Holt’s conceptualization,

for example, when an adult copies the response of
a child, the latter tends to repeat the reiterated

behavior, and as this circular associative sequence
continues, the adult’s behavior becomes an
increasingly effective stimulus for the child’s
responses. If, during this spontaneous mutual

imitation, the adult performs a response that is
novel for the child, the latter will copy it. Piaget
(1952) likewise depicted the modeling process as
one in which the imitator’s spontaneous behaviors
serve initially as stimuli for matching responses by

the model in alternating imitative sequences.
Allport also viewed modeling phenomena as
instances of classical conditioning of
verbalizations, motor responses, or emotions to
matching social stimuli with which they have been
contiguously associated.
The various associative theories isolated one of
the conditions under which modeling cues may
acquire eliciting functions for matching responses

that already exist in the imitator’s behavioral
repertoire. These theories failed to explain,
however, the psychological mechanisms governing

the acquisition of novel responses during the
model-observer interaction sequence. Moreover,
demonstrations of observational learning in
humans and animals do not ordinarily commence
by having a model reproduce semi-irrelevant

responses of the learner. In using modeling
procedures to teach a mynah bird to talk, for
example, the trainer does not engage initially in
circular crowing behavior; rather, he begins by
saying what he wishes to teach, which expressions
clearly do not exist in integrated form in the bird’s
vocal repertoire.
REINFORCEMENT THEORIES
With the advent of reinforcement principles,
theoretical explanations of learning shifted the
emphasis from classical conditioning to
instrumental response acquisition based on

reinforcing outcomes. Theories of modeling
phenomena similarly assumed that the occurrence

of observational learning is contingent upon
reinforcement of imitative behavior. This point of
view was most clearly expounded by Miller &

Dollard (1941) in the classic publication, Social
Learning and Imitation. According to this
formulation, the necessary conditions for learning
through modeling include a motivated subject who
is positively reinforced for matching the correct
responses of a model during a series of initially
random, trial-and-error responses.
The experiments conducted by Miller and
Dollard involved a series of two-choice
discrimination problems, in which a trained leader

responded to environmental stimuli that were
concealed from the subject so that he was

dependent solely upon the cues provided by the
model’s behavior. The leader’s choices were
consistently rewarded and the observing subject
was similarly reinforced whenever he matched

these choice responses. This form of imitation was
labeled by the authors “matched-dependent”
behavior, because the subjects relied on the leader

for relevant cues, and matched his responses.
Based on this paradigm, it was shown that

subjects readily learn to follow their respective
models, and generalize copying responses to new
situations, to new models, and to different

motivational states.
While these experiments have been widely
accepted as demonstrations of imitative learning,
they in fact represent only the special case of
discrimination place-learning, in which the

behavior of others provides discriminative stimuli
for responses that already exist in the subject’s
behavioral repertoire. Indeed, had the relevant
environmental cues been made more distinctive,
the behavior of the models would have been quite

irrelevant, and perhaps even a hindrance, to the
acquisition process. By contrast, most forms of
imitation involve responses rather than place-
learning, in which subjects combine behavioral
elements into new compound responses solely by

observing the performance of social models,
without any opportunity to perform the model’s

behavior at the time of exposure and without any
reinforcers administered either to the models or
to the observers (Bandura, 1965a). In the latter

instance, modeling cues constitute an
indispensable aspect of the learning process.
Moreover, since the reinforcement paradigm for
observational learning requires the subject to

perform the imitative response before he can
learn it, the theory advanced by Miller and Dollard

evidently accounts more adequately for the
performance of previously learned matching
responses than for their acquisition. Continuing
with the example of language learning, in order for

a mynah bird to learn the word “reinforcement”
imitatively, it would first have to utter the word
“reinforcement” in the course of random

vocalization, match it accidentally with the
trainer’s verbal responses, and secure a positive

reinforcement. The conditions that Miller and
Dollard assumed to be necessary for imitative
learning severely limit the types of behavioral

changes that can be attributed to the influence of
social models.
The Skinnerian analysis of modeling
phenomena (Baer & Sherman, 1964; Skinner,
1953), which is similar in many respects to the

one originally advanced by Miller and Dollard, also
specifies reinforcement as a necessary condition

for observational learning. In this approach
modeling is treated as a form of stimulus matching
in which a person matches the stimulus pattern
generated by his own responses to the

appropriate modeling cues. The stimulus
duplication is presumably achieved through a
process of differential reinforcement. When

matching behavior has been positively reinforced
and divergent responses either nonrewarded or

punished, the behavior of others comes to function
as discriminative stimuli for reinforcement in
controlling social responsiveness.
More recently, Gewirtz & Stingle (1968) have
conceptualized modeling as analogous to the
matching-to-sample paradigm used to study
discrimination learning. In this procedure a
subject chooses from among a number of

comparison stimuli one that shares a common
property with the sample stimulus. Although
modeling and matching-to-sample performances
have some likeness in that both involve a matching
process, they can hardly be equated. A person can

achieve errorless choices on matching comparison
operatic arias with a sample Wagnerian recital,
but remain totally unable to perform the vocal
behavior exhibited in the sample. Accurate
stimulus discrimination is a precondition for, but

not equivalent with, observational response
learning. The major controversy among theories of

modeling centers around the question of what are
the necessary and sufficient conditions for the
acquisition of new responses on an observational

basis.
Under naturalistic conditions the behavior
exhibited by models is typically reproduced in the
absence of direct reinforcement. Consequently,

theories that assume that some form of
reinforcement is necessary for learning tend to

invoke an intrinsic source of reinforcement. It is
assumed that if accurate reproduction of modeling
stimuli is consistently rewarded, behavioral
similarity per se acquires secondary reinforcing

properties. Thereafter a person will tend to
display a high incidence of precisely imitative
actions, which, because of their acquired reward

value, will be strengthened and sustained even
though they may never be externally reinforced.
Baer and his colleagues have conducted
several experiments designed to demonstrate
intrinsic reinforcement control of generalized
imitation. In one study (Baer & Sherman, 1964)
three imitative responses (head-nodding,
mouthing, and novel verbalizations) were
established in young children by social
reinforcement from a puppet who had explicitly

instructed the subjects to match his modeled
behavior. For a subgroup of children who showed
an increase in imitative responding the puppet
displayed nonreinforced bar-pressing
interspersed among the other three rewarded

matching responses. Under these conditions some
of the children imitated bar-pressing in varying
amounts even though this particular response was
never positively reinforced. In order to further
demonstrate the dependence of generalized

imitation on direct reinforcement of other
matching responses, social approval for imitative

head-nodding, mouthing and novel verbalizations
was discontinued with two subjects. This
extinction procedure resulted in decreased

imitative bar-pressing in one of the two children;
when reinforcement of the other three modeling
responses was reinstated, imitative bar-pressing
also reappeared.
The frequent references to the above study as
providing evidence for the self-reinforcing

function of response similarity overlook the fact
that, even under explicit demands, the imitative
behavior of one-third of the children was

completely unaffected by the reinforcement
operations, and that approximately half of the
remaining children whose data are reported
showed increments in reinforced imitative
behavior but failed to perform the nonreinforced

modeled response to any significant degree. Since
reinforcement exerted no clearly predictable
effects on the occurrence of generalized imitation
it must have been largely determined by other
unmeasured and uncontrolled variables.
Using similar reinforcement procedures with
social models and more powerful incentives, Baer,
Peterson, & Sherman (1967) were able to

establish generalized imitativeness in three
severely retarded children who initially displayed

a very low level of matching behavior (see Figure
3-1). After an extensive period of imitation-
contingent reinforcement had markedly increased
imitative behavior in these children (sessions 1-

14), some matching responses could be effectively
maintained without reinforcement by randomly
interspersing them among positively reinforced

imitations (sessions 15-26). However, both types
of imitative responses rapidly declined when

social approval and food were given to the
children on a temporal basis rather than
contingent upon imitative behavior (sessions 27-

31). It was further shown that both types of
matching responses could be quickly restored to
Figure 3-1. Percentage of reinforced and nonreinforced
modeled responses reproduced by a child during
periods when rewards were made contingent upon the
occurrence of matching responses or upon the passage
of a given period of time (DRO). Baer, Peterson, &
Sherman, 1967.
their previously high level by reintroduction of
response-contingent reinforcement (sessions 32-
38).
It has been similarly demonstrated that
schizophrenic children could acquire and maintain

Norwegian words imitatively without any
reinforcement (Lovaas, Berberich, Perloff, &
Schaeffer, 1966), and preschool children imitated

nonreinforced Russian words (Brigham &
Sherman, 1968) as long as the children were
rewarded for English words when correctly
reproduced.
Although a generalized disposition to imitate
the behavior of others can be developed by having
different persons reinforce diverse types of
responses in a variety of situations, this fact does
not necessarily demonstrate that reinforcing
properties inhere in behavioral similarity. If this
were in fact the governing mechanism, matching
responses would not undergo abrupt and marked
extinction (see Figure 3-1) the moment that

reinforcement for the larger subclass of imitative
responses is withdrawn, since one would not
expect similarity cues to lose their rewarding
value that suddenly. Rather, the intrinsic rewards

arising from precise response duplication should
sustain imitative behavior for some time even in
the absence of externally administered
reinforcers. Studies including more extensive
variations in incentive conditions, indeed, show

that generalized imitation is largely under
incentive control rather than its inherently

rewarding value. Berkowitz (1968) found that
retarded children who were rewarded for

imitative responses only at the end of the
experimental session displayed a high rate of

matching behavior as long as the food rewards
were present in the room. During sessions when
food was not displayed, imitation dropped
significantly; it was promptly re-established by

introducing the sight of food.
It should be noted that the laboratory

phenomenon that has been labeled “generalized
imitation” involves only imitation across
responses under conditions where subjects are

instructed to repeat the experimenter’s behavior.
A more stringent test of generalized imitation
would include different models performing
different responses in different social situations.
An alternative explanation for this limited form of

generalized modeling can be offered in terms of
discrimination rather than secondary
reinforcement processes. When a few
nonrewarded, modeled responses are randomly
distributed in a large number that are consistently
reinforced, the two sets of responses cannot easily
be distinguished and are therefore likely to be
performed with similar frequency. If, on the other
hand, the discriminative complexity of the

modeling task were reduced by having the model
portray a series of reinforced responses, followed
by the set of readily discriminable responses that
are never reinforced, the observer would
eventually recognize that the latter responses

never produce positive outcomes and he would, in
all likelihood, stop reproducing them. A
discrimination hypothesis thus leads to a
prediction which is opposite to that derived from
the principle of secondary reinforcement.

According to the acquired-reward interpretation,
the longer imitative responses are positively

reinforced, the more strongly behavioral similarity
is endowed with reinforcing properties and,
consequently, the greater should be the resistance

to extinction of unreinforced matching responses.
In contrast, a discrimination hypothesis would
predict that the longer the differential
reinforcement practices are continued, the more

likely the observer is to distinguish between
rewarded and unrewarded imitative behaviors,

with resulting rapid decline of unrewarded
imitative responses.
The occurrence of generalized modeling is also

probably determined in part by the invariant
conditions under which laboratory tests are
conducted. Reinforced and unreinforced
responses are typically exhibited by the same
model, in the same social setting, during the same

period of time, and after subjects have been
explicitly instructed to behave imitatively. On the
other hand, under natural conditions, which are
highly variable and more easily distinguishable,
there appears to be considerable specificity to
modeling behavior. If matching responses do, in
fact, automatically produce self-reinforcing effects,
then people should display widespread
reproduction of all types of behavior modeled by

children, barbers, policemen, delinquents,
professors and others. Actually, people tend to be
selective in what they reproduce, suggesting that
imitative performance is primarily governed by its
utilitarian value rather than by inherent

reinforcement derived from response similarity
per se. In other words, the theory of generalized
imitation explains more than has ever been
observed. The issue would appear to be one of
regulated performance rather than learning, since

people do know how to match the behavior of
others. Performance is primarily a function of

anticipated outcome which, in turn, are partly
determined by the degree of similarity between
new situations and past situations in which

particular responses have been reinforced.
Important treatment implications follow from
the interpretations of generalized modeling, since
in both cases the goal is to establish modeling

tendencies that will not be restricted to the
treatment setting but will generalize to other,

more natural settings. On the basis of a secondary
reinforcement hypothesis, the treatment program
should include considerable imitation training
under a generous schedule of reinforcement. The

assumption made is that the more reinforcement a
person experiences for behavioral matching, the
more reinforcing it will become for him to imitate

in any situation. On the basis of a discrimination
hypothesis, on the other hand, the program would

involve only as much reinforcement as is
necessary to establish matching behavior, which
would then be rewarded by different people in a

variety of situations. Generalization is not
assumed to occur automatically; it must be built
into the treatment program.
The Skinnerian analysis of modeling

phenomena relies entirely upon the standard
three-component paradigm Sd → R → Sr, where Sd
denotes the discriminative modeled stimulus, R

represents an overt matching response, and Sr
designates the reinforcing stimulus. It is difficult to
see how this scheme is applicable to observational

learning in which an observer does not overtly
perform the model’s responses during the
acquisition phase, reinforcers are not
administered either to the model or to the
observer, and the first appearance of the acquired

response may be delayed for days, weeks, or even
months. In the latter case, which represents one of
the most prevalent forms of social learning, two of
the events (R → Sr) in the three-term paradigm are
absent during acquisition, and the third element
(Sd or modeling stimulus) is typically absent from
the situation in which the observationally learned
response is performed. Like the Miller and Dollard
theory, the Skinnerian interpretation of modeling

phenomena accounts satisfactorily for the control
of previously learned matching responses by their
stimulus antecedents and their immediate
consequences. However, it fails to explain how a
new matching response is acquired

observationally in the first place. This occurs
through covert, symbolic processes during the
period of exposure to modeling stimuli, prior to
overt responding or to appearance of any
reinforcing events. Indeed, had the children in the

experiment by Baer & Sherman been tested for
vicarious learning immediately after the model

had demonstrated the four critical responses, they
could probably have reproduced the modeled
repertoire without undergoing any imitation-

contingent reinforcement. As will be shown later,
observational learning entails symbolic coding and
central organization of modeling stimuli, their
representation in memory, in verbal and imaginal

codes, and their subsequent transformation from
symbolic forms to motor equivalents. Because of

the inferential nature of these basic processes,
functional behaviorists are inclined to consider
them of limited scientific interest. However,
modeling phenomena must be analyzed in terms

not only of response-selection variables but also of
their mediational determinants before the
necessary and sufficient conditions for modeling

can be specified accurately.
In evaluating the role of reinforcement in

modeling processes, it is essential to distinguish
between response acquisition and performance
because these events are determined by different
variables. Numerous investigations, differing
considerably in the choice of incentives, types of
matching responses, and age of the subjects, have
shown that performance of matching responses is
substantially increased by rewarding such

behavior in either the model (Bandura, 1965a;
Kanfer, 1965; Parke & Walters, 1967) or the
subjects (Kanareff & Lanzetta, 1960; Lanzetta &
Kanareff, 1959; Metz, 1965; Schein, 1954; Wilson
& Walters, 1966); whereas imitative

responsiveness is reduced by direct or vicarious
punishment. However, results of an experiment
bearing on the learning-performance distinction
lend support to the theory that the acquisition of
matching responses results primarily from

stimulus contiguity and associated symbolic
processes, whereas the performance of

observationally learned responses will depend to
a great extent upon the nature of reinforcing
consequences to the model or to the observer.
In this study (Bandura, 1965b), children
observed a filmed model who exhibited a
sequence of novel physical and verbal aggressive
responses. In one treatment condition the model

was severely punished following the display of
aggressive behavior; in the second, the model was
generously rewarded with delectable treats and
lavish praise; the third condition presented no
response consequences to the model. A post-

exposure performance test of imitation revealed
that the reinforcement contingencies applied to
the model’s responses resulted in differential
degrees of matching behavior. Compared to
subjects in the model-punished condition, children

in the model-rewarded and the no-consequence
groups spontaneously performed a significantly

greater variety of imitative responses. Moreover,
boys reproduced substantially more of the model’s
behavioral repertoire than girls, the differences

being particularly marked in the model-punished
treatment (Figure 3-2).
Following the performance test, children in all

three groups were offered highly attractive
incentives contingent upon their reproducing the
model’s responses in order to promote

performance of what they had acquired through
observation. As shown in Figure 3-2, the
introduction of positive incentives completely

eliminated the previously observed performance
differences, revealing an equivalent amount of
learning among children in the model-rewarded,
model-punished, and no-consequence; conditions.
Similarly, the initially large sex differential, which

in similar studies has been typically interpreted as
reflecting a deficit in masculine-role identification
by girls, was virtually eliminated.
Findings of the foregoing experiment, and
others reviewed later, suggest that the behavior
Figure 3-2. Mean number of different matching responses
reproduced by children as a function of response
consequences to the model and positive incentives.
Bandura, 1965b.
analysis advocated by proponents of the
Skinnerian approach might further advance
understanding of modeling processes if it were

separated into a learning analysis and a
performance analysis. The learning analysis is
concerned with the manner in which variables
operating at the time of exposure to modeling

stimuli determine the degree to which the
modeled behavior is learned. The performance
analysis, on the other hand, is concerned with
factors governing persons’ willingness to perform
what they have learned.
Although there is ample evidence that
reinforcing consequences can significantly alter

the probability of future occurrence of preceding
matching responses, consequent events can hardly

serve as a precondition for the acquisition of
responses that have already been performed. The

major issue of whether reinforcement is a
prerequisite for observational learning can be
most definitively resolved by the use of
infrahuman subjects whose reinforcement history

can be controlled. In a preliminary study, Foss
(1964) found that birds will imitate unusual sound
patterns played on a tape recorder in the absence
of any prior reinforcement of matching responses.
In human learning, under conditions where

incentives are repeatedly given to a model as he
displays an ongoing series of responses,
observation of reinforcing outcomes occurring
early in the sequence might be expected to
increase the observer’s vigilance in respect to

subsequently modeled behavior. The anticipation
of positive reinforcement for matching responses

by the observer may, therefore, indirectly
influence the course of observational learning by
enhancing and focusing observing responses.
AFFECTIVE FEEDBACK THEORY
Mowrer’s (1960) sensory feedback theory of
imitation similarly highlights the role of
reinforcement but, unlike the preceding

approaches which reduce imitation to a special
case of instrumental learning, Mowrer emphasizes

the classical conditioning of positive and negative
emotions accompanying reinforcement to stimuli
arising from matching responses. Mowrer
distinguishes two forms of imitative learning in

terms of whether the observer is reinforced
directly or vicariously. In the former case, the
model performs a response and at the same time

rewards the observer. Through repeated
contiguous association of the model’s behavior

with rewarding experiences, these responses
gradually take on positive value for the observer.
On the basis of stimulus generalization, the

observer can later produce self-rewarding
feedback experiences simply by reproducing as
closely as possible the model’s positively valenced
behavior.
In the second or “empathetic” form of imitative
learning, the model not only exhibits the response
but also experiences the reinforcing consequences.

It is assumed that the observer, in turn,
experiences empathetically the sensory
concomitants of the model’s behavior, and also

intuits his satisfactions or discomforts. As a result
of this higher-order vicarious conditioning, the
observer will be predisposed to reproduce the
matching responses for the attendant positive
sensory feedback.
There is substantial evidence (Bandura &
Huston, 1961; Grusec, 1966; Henker, 1964;
Mischel & Grusec, 1966; Mussen & Parker, 1965)
that modeling can be augmented by increasing the
rewarding qualities of a model or by having the
observer witness the model experience rewarding
outcomes. These same studies, however, contain
some contradictory findings with regard to the

affective feedback theory. Even though a model’s
rewarding qualities are equally associated with
the different types of behaviors he performs,
modeling effects nevertheless tend to be specific

rather than general. That is, model nurturance
enhances the reproduction of some responses, has
no effect upon others, and may actually diminish
the adoption of still others (Bandura, Grusec &
Menlove, 1967a). A limited study by Foss (1964),

in which mynahs were taught unusual whistles
played on a tape recorder, also failed to confirm

the proposition that modeling is enhanced
through positive conditioning. His mynahs

imitated different sounds to the same extent
whether they were played in the absence of any

reinforcement or only when the birds were being
fed. It should be noted, however, that neither the
Foss study nor the experiments cited earlier
employed the type of temporal relationship

between modeling cues and the administration of
rewards that would be considered optimal for

endowing the modeled events with affective
valence.
In an elaboration of the affective feedback

theory of imitation, Aronfreed (1968) has
advanced the view that pleasurable and aversive
affective states become conditioned to cognitive
templates of a model’s behavior. Imitative
performances are presumed to be controlled by

affective feedback from intentions and from
proprioceptive cues generated during the
performance of an overt act. This
conceptualization of imitation is difficult to verify
empirically because it does not specify in sufficient
detail the characteristics of templates, the process
through which cognitive templates are acquired,
the manner in which affective valences become
conditioned to templates, or how the emotion-

arousing properties of templates are transferred
to intentions and to proprioceptive cues intrinsic
to overt responses. There is some experimental
evidence, however, that has important
implications for the basic assumptions of feedback

notions.
Feedback theories, particularly those that
partly rely on controlling functions of
proprioceptive cues, are seriously challenged by

findings of curare-conditioning experiments in
which animals are skeletally immobilized during
aversive conditioning or extinction. These studies

(Black, 1958; Black, Carlson, & Solomon, 1962;
Solomon & Turner, 1962) demonstrate the
occurrence of learning phenomena in the absence
of skeletal responding and its correlated
proprioceptive feedback. Results of
deafferentation studies (Taub, Bacon, & Berman,
1965; Taub, Teodoru, Ellman, Bloom, & Berman,

1966) also show that responses can be acquired,
performed discriminatively, and extinguished
even though sensory somatic feedback is
surgically abolished by limb deafferentation. It
would seem from these findings that the

acquisition, integration, facilitation and inhibition
of responses can be achieved through central
mechanisms independently of peripheral sensory
feedback.
It is also evident that rapid selection of

responses from among a varied array of
alternatives cannot be governed by proprioceptive

feedback since relatively few responses could be
activated, even incipiently, during
characteristically brief pre-decision periods
(Miller, 1964). In recognizing this problem,
Mowrer (1960) has conjectured that the initial
scanning and selection of responses may occur
primarily at the symbolic rather than at the action

level.
Human functioning would be inflexible and
unadaptive if responsiveness were controlled by
affectivity inherent in the behavior itself. Because
social responsiveness is highly discriminative, it is

extremely doubtful that behavioral patterns are
regulated by affective qualities implanted in
behavior. To take aggression as an example,
hitting responses directed toward parents, peers,

and inanimate objects differ little, if at all;
nevertheless, physically aggressive responses
toward parents are generally strongly inhibited,

whereas physical aggression toward peers is
freely expressed (Bandura, 1960; Bandura &
Walters, 1959). Moreover, in certain well-defined
contexts, particularly in competitive, physical
contact sports such as boxing, people will easily
initiate and maintain unattenuated, physically
aggressive behavior. One would, therefore, predict

more accurately the expression or inhibition of
identical aggressive responses from knowledge of
the stimulus context (e.g., church, athletic
gymnasium), the object (e.g., parent, priest,
policeman, or peer), and other cues that signify

predictable consequences, than from assessment
of the affective value of aggressive behavior. It has
been amply demonstrated (Bandura, 1968) that
the selection and performance of matching
responses is mainly governed by anticipated

outcomes based on previous consequences that
were directly encountered, vicariously

experienced, or self-administered.
Although feedback conceptions of modeling do
not require a response to be performed before it
can be learned, they nevertheless fail to explain
the acquisition of matching behavior when
reinforcers are not dispensed either to the model
or to observers. Moreover, a vast majority of the

responses that are acquired observationally are
not affectively valenced. This is exemplified by
studies of observational learning of perceptual-
motor tasks from filmed demonstrations (Sheffield
& Maccoby, 1961) that do not contain positive or

aversive stimuli essential for the classical
conditioning of emotional responses. Mowrer has,
of course, pointed out that sensory experiences
not only classically condition positive or negative
emotions, hut also produce conditioned sensations

or images. In most cases of observational learning
images or other forms of symbolic representations

of modeling stimuli may be the only important
mediators. Sensory-feedback theories of imitation
may therefore be primarily relevant to instances

in which the modeled responses incur relatively
potent reinforcing consequences capable of
endowing response-correlated stimuli with
motivational properties. Affective conditioning

should therefore be regarded as a facilitative
rather than a necessary condition for modeling.
CONTIGUITY-MEDIATIONAL THEORIES
When a person observes a model’s behavior,

but otherwise performs no overt responses, he can
acquire the modeled responses while they are
occurring only in cognitive, representational
forms. Any learning under these conditions occurs
purely on an observational or covert basis. This

mode of response acquisition has accordingly
been designated as no-trial learning (Bandura,
1965a), because the observer does not engage in
any overt responding trials, although he may
require multiple observational trials in order to

reproduce modeled stimuli accurately. Several
theoretical analyses of observational learning
(Bandura, 1962, 1965a; Sheffield, 1961) assign a
prominent role to representational mediators that

are assumed to be acquired on the basis of a
contiguity learning process. According to the
author’s formulation, observational learning
involves two representational systems—an

imaginal and a verbal one. After modeling stimuli
have been coded into images or words for memory
representation they function as mediators for
subsequent response retrieval and reproduction.
Imagery formation is assumed to occur
through a process of sensory conditioning. That is,
during the period of exposure, modeling stimuli

elicit in observers perceptual responses that
become sequentially associated and centrally

integrated on the basis of temporal contiguity of
stimulation. If perceptual sequences are

repeatedly elicited, a constituent stimulus
acquires the capacity to evoke images (i.e.,
centrally aroused perceptions) of the associated
stimulus events even though they are no longer

physically present (Conant, 1964; Ellson, 1941;
Leuba, 1940). Thus, for example, if a bell is
sounded in association with a picture of an
automobile the bell alone tends to elicit imagery of
the car. Under conditions where stimulus events

are highly correlated, as when a name is
consistently associated with a given person, it is
virtually impossible to hear the name without
experiencing imagery of the person’s physical
characteristics. The findings of studies cited above

indicate that, in the course of observation,
transitory perceptual phenomena produce

relatively enduring, retrievable images of modeled
sequences of behavior. Later reinstatement of
imaginal mediators serves as a guide for

reproduction of matching responses.
The second representational system, which
probably accounts for the notable speed of
observational learning and long-term retention of

modeled contents by humans, involves verbal
coding of observed events. Most of the cognitive

processes that regulate behavior are primarily
verbal rather than visual. To take a simple
example, the route traversed by a model can be
acquired, retained, and later reproduced more

accurately by verbal coding of the visual
information into a sequence of right-left turns
(e.g., RRLRR) than by reliance upon visual imagery

of the itinerary. Observational learning and
retention are facilitated by such codes because

they can carry a great deal of information in an
easily stored form. After modeled sequences of
responses have been transformed into readily

utilizable verbal symbols, later performances of
matching behavior can be effectively controlled by
covert verbal self- directions.
The influential role of symbolic representation

in observational learning is disclosed by a study
(Bandura, Grusec, & Menlove, 1966) in which
children were exposed to several complex

sequences of modeling behavior on film, during
which they either watched attentively, verbalized
the novel responses as they were performed by

the model, or counted rapidly while watching the
film to prevent implicit verbal coding of modeling
cues. A subsequent test of observational learning
disclosed that children who verbally labeled the
modeled patterns reproduced significantly more

matching responses than those in the viewing-
alone condition who, in turn, showed a higher
level of acquisition than children who engaged in
competing symbolization.
Further supporting evidence for the influence
of symbolic coding operations in the acquisition
and retention of modeled responses is furnished
by Gerst (1969). Subjects observed a filmed model

perform complex motor responses varying in the
ease with which they could be verbally coded.
They were instructed to code the items into either
vivid images, concrete verbal descriptions of the

response elements, or convenient summary labels
that incorporated the essential ingredients of the
responses. Compared to the performance of
control subjects who had no opportunity to
generate symbolic mediators, all three coding

operations enhanced observational learning
(Figure 3-3). Concise labeling and imaginal codes

were equally effective in aiding immediate
reproduction of modeled responses and both

systems proved superior in this respect to the
concrete verbal form. However, a subsequent test

for retention of matching responses showed
Figure 3-3. Percentage of modeled responses reproduced by
control subjects and those who coded the modeled
behavior as either images, concrete verbal descriptions,
or summary labels for memory representation. Gerst,
1969.
concise labeling to be the best coding system for
memory representation. Subjects in the latter
conditions retained a significant amount of what

they learned, whereas those who relied upon
imagery and concrete verbalizations displayed a

substantial loss of matching responses.
Results of a program of research using a
nonresponse acquisition procedure (Bandura,

1965a) indicate that the organization of behaviors
elements into novel patterns resembling modeled
responses can occur at a central level without
overt responding. The present theory assumes,
however, that stimulus contiguity is a necessary,

but not a sufficient, condition for observational
learning. Modeling phenomena, in fact, involve
several complexly interrelated subprocesses, each
with its own set of controlling variables. A
comprehensive theory of observational learning
must therefore encompass the diverse subsystems
governing the broader phenomena. The main
component functions that markedly influence the
nature and degree of observational learning are

discussed next.
ATTENTIONAL PROCESSES
Since repeated contiguous stimulation alone

does not always result in response acquisition, it is
evident that additional conditions are required for

the occurrence of observational learning. Simply
exposing persons to distinctive sequences of
modeled stimuli does not in itself guarantee that

they will attend closely to the cues, that they will
necessarily select from the total stimulus complex
the most relevant events, or that they will even
perceive accurately the cues to which their
attention has been directed. An observer will fail
to acquire matching behavior, at the sensory
registration level, if he does not attend to,
recognize, or differentiate the distinctive features
of the model’s responses. To produce learning,
therefore, stimulus contiguity must be

accompanied by discriminative observation.
A number of attention-controlling variables,
some related to incentive conditions, others to

observer characteristics, and still others to the
properties of the modeling cues themselves, will
be influential in determining which modeling

stimuli will be observed and which will be ignored.
Selectivity of modeling stimuli may be partly a
function of their inherent physical properties
based on intensity, size, vividness and novelty. Of
much greater importance for social learning,

however, is the acquired distinctiveness of model
attributes (Miller & Dollard, 1941). By being

repeatedly rewarded for imitating certain types of
models and not rewarded for matching the

behavior of models possessing different
characteristics, persons eventually learn to
discriminate between modeling cues that signify
differential probabilities of reinforcement. Thus,

models who have demonstrated high competence
(Gelfand, 1962; Mausner, 1954a, b; Mausner &
Bloch, 1957; Rosenbaum & Tucker, 1962), who are
purported experts (Mausner, 1953) or celebrities
(Hovland, Janis, & Kelley, 1953), and who possess

status-conferring symbols (Lefkowitz, Blake, &
Mouton, 1955) are likely to command more
attention and to serve as more influential sources
of social behavior than models who lack these
qualities. Other distinctive characteristics, such as

age (Bandura & Kupers, 1964; Hicks, 1965;
Jakubczak & Walters, 1959), sex (Bandura, Ross, &

Ross, 1963a; Maccoby & Wilson, 1957; Ofstad,
1967; Rosenblith, 1959, 1961), social power
(Bandura, Ross, & Ross, 1963b; Mischel & Grusec,

1966), and ethnic status (Epstein, 1966), which
are correlated with differential probabilities of
reinforcement, likewise influence the degree to
which models who possess these attributes will be

selected for emulation.
The affective valence of models, as mediated

through their attractiveness and other rewarding
qualities (Bandura & Huston, 1961; Grusec &
Mischel, 1966), may augment observational

learning by eliciting and maintaining strong
attending behavior. At the social level, one’s
organizational affiliations and living
circumstances, which affect associational
networks and preferences, will also determine to a

large degree the types of models to whom one is
repeatedly exposed, and consequently, the modes
of behavior that will be most thoroughly learned.
An adequate theory of vicarious learning must
also explain why, under essentially identical
conditions of modeling stimulation, some persons
display higher levels of response acquisition than
others. There is suggestive evidence that

characteristics of observers, deriving from their
previous social-learning experiences, may be
associated with different observational patterns.
The extent to which modeled patterns are

reproduced is significantly influenced by observer
characteristics such as dependency (Jakubczak &
Walters, 1959; Kagan & Mussen, 1956; Ross,
1966), self-esteem (de Charms & Rosenbaum,
1960; Gelfand, 1962; Lesser & Abelson, 1959),

level of competence (Kanareff & Lanzetta, 1960),
and socioeconomic and racial status (Beyer & May,

1968); and countless studies have shown that the
effects of modeling stimuli are partly determined

by the sex of observers. Persons who have been
frequently rewarded for displaying matching

behavior (Miller & Dollard, 1941; Schein, 1954)
are also apt to be most attentive to modeling cues.
Finally, motivational variables and transitory
emotional arousal significantly alter perceptual

thresholds and in other ways facilitate, impede,
and channel observing responses (Bandura &

Rosenthal, 1966; Easterbrook, 1959; Kausler &
Trapp, 1960).
It is difficult to evaluate from performance

measures alone whether the effects of observer
characteristics reflect differences in degree of
observational learning or in willingness to
perform what has been learned. Results of several
studies employing a learning analysis of modeling

(Bandura, Grusec, & Menlove, 1966; Grusec &
Brinker, 1969; Maccoby & Wilson, 1957) disclose
that observer characteristics can serve as
determinants of observational learning.
Viewers’ observing behavior can be effectively
enhanced and focused through arrangement of
appropriate incentive conditions. Persons who are
informed in advance that they will later be asked

to reproduce a given model’s responses and
rewarded in terms of the number of elements
performed correctly would be expected to pay
much closer attention to relevant modeling stimuli

than persons who are exposed to the same
modeled events without any predisposition to
observe and to learn them. The facilitative
influence of incentive set on observational
learning will be most operative under exposure to

multiple models requiring selective attentiveness
to conflicting cues. Indeed, incentive control of

observing behavior can, in most instances,
override the effects of variations in observer

characteristics and model attributes. It should be
noted, however, that in the present theory

reinforcement variables, to the extent that they
influence the acquisition process, do so principally
by augmenting and sustaining attentiveness to
modeling cues.
In addition to attention-directing variables,
stimulus input conditions (i e., rate, number,

distribution, and complexity of modeling stimuli
presented to observers) will regulate the
acquisition of modeled responses to some extent.

The observer’s capacity to process information
sets definite limits on the number of modeling
cues that can be acquired during a single exposure.
Therefore, if modeling stimuli are presented at a
rate or level of complexity that exceeds the

observer’s receptive capabilities, observational
learning will necessarily be limited and
fragmentary. Under such conditions repeated
presentations of the modeling stimuli would be
required in order to produce complete and precise
response matching.
Finally, the rate and level of observational
learning will be affected by the discriminability of
modeling stimuli. Modeled characteristics that are

highly discernible can be more readily acquired
than subtle attributes which must be abstracted

from heterogeneous responses differing on
numerous stimulus dimensions. In highly intricate
response systems, such as language behavior, for
example, children typically encounter

considerable difficulty in acquiring linguistic
structures because the identifying characteristics
of different grammatical constructions cannot be

readily distinguished within extremely diverse
and complex utterances. However, when verbal

modeling cues are combined with procedures
designed to increase syntactic discriminability
(Bandura & Harris, 1966; Lovaas, 1966a; Odom,

Liebert, & Hill, 1968) relatively complicated
linguistic patterns of behavior can be acquired and
modified observationally.
In therapeutic applications of modeling

procedures observational learning is often
retarded by discrimination failures arising from
deficiencies in cognitive skills, sensory-motor

handicaps, or faulty prior learning. In such cases a
program of discrimination pretraining may greatly
accelerate modeling processes. Winitz and Preisler

(1965) have shown, for example, that children
who learned to discriminate erroneous sounds
from correct sounds that they had misarticulated
subsequently displayed better imitative word
learning than children who did not receive

relevant discrimination pretraining.
RETENTION PROCESSES
with sensory registration and symbolic coding of

modeling stimuli. Another basic component
function involved in observational learning, but
one that has been virtually ignored in theories of
imitation, concerns the retention of modeled
events. In order to reproduce social behavior

without the continued presence of external
modeling cues a person must retain the original

observational inputs in some symbolic form. This
is a particularly interesting problem in instances
where persons acquire social patterns of behavior
observationally and retain them over extended

periods of time, even though the response
tendencies are rarely, if ever, activated into overt
performance until attainment of the age or social
status at which the activity is appropriate and
permissible.
There are a number of theoretical

controversies regarding memory processes which
will not be reviewed here since they fall beyond
the scope of this book. The major questions are
whether memory traces are established in an
incremental or an all-or-none fashion; whether
there exists a dual or a single memory mechanism;

and speculations about the biochemical and
neurophysiological processes whereby transient
neural after-effects of stimulation result in
relatively permanent structural alterations in the
central nervous system. Although memory

mechanisms have not as yet been adequately
explained, laboratory investigations have
identified a number of conditions that facilitate
retention, some of which have been shown to
augment modeling performances.
Among the many variables governing retention
processes, rehearsal operations effectively

stabilize and strengthen acquired responses. The
level of observational learning can, therefore, be
considerably enhanced through practice or overt
rehearsal of modeled response sequences,
particularly if the rehearsal is interposed after
natural segments of a larger modeled pattern
(Margolius & Sheffield, 1961). Of greater import is

evidence that covert rehearsal, which can be
readily engaged in when overt participation is
either impeded or impracticable, may likewise
enhance retention of acquired matching responses
(Michael & Maccoby, 1961). Data are meager,

however, on the types of responses that are most
susceptible to strengthening through covert
rehearsal. Several experiments involving a variety
of tasks (Morrisett, 1956; Perry, 1939; Twining,
1949; Vandell, Davis, & Clugston, 1943), have

shown that symbolic rehearsal of activities
significantly improves their later performance.

Such practice appears to be most effective in tasks
that rely heavily upon symbolic functions.
The influential role of covert practice of
modeled behaviors has received greatest
emphasis in Maccoby’s (1959) account of the
identification process. According to this view,
controlling, nurturing, and caretaking activities

require explicit reciprocal behaviors on the part of
parents and children. Consequently, in the course
of frequent mutually dependent interactions both
participants learn, anticipate, and covertly
rehearse each other’s customary responses. In

addition to the frequency md intimacy of social
interactions, the degree of power exercised by the
model over desired resources is considered to be
an important determinant of the frequency of
fantasy role-playing. In this theory, vicarious role-

rehearsal primarily serves a defensive function;
that is, in an effort to guide his behavior toward

models who possess controlling power, a person
will imagine different courses of action for
receiving help or avoiding censure, and he will try

to anticipate as accurately as possible the model’s
probable responses to these approaches. On the
other hand, there would be little incentive to
prepare oneself for, or to practice covertly, the

behavior of models who command no rewarding
or punishing power.
Anticipatory implicit rehearsal of modeled
responses may be supported to some extent by
role reciprocity and threat from resource

controllers, but it should be noted that persons
will also be inclined to practice modeled responses
that are effective in producing rewarding
outcomes. Moreover, according to social-learning
theory, the behavior of powerful models will be

attended to, rehearsed, and reproduced even
though observers have had no direct interaction
with them, because their behavior is likely to have
high utilitarian value. This is particularly true in
the case of models who possess expert power in
particular specialties. It would be unnecessary, for
example, for a novice to establish a
complementary role relationship with a qualified
automobile mechanic in order to master his skills

through observation during apprenticeship
training. Rehearsal behavior is undoubtedly
governed by different types of incentive
conditions, some of which may be entirely
independent of the model whose behavior is being

emulated.
It is generally assumed that the facilitative
effects of rehearsal result not from sheer
repetition, but rather from more active processes.

The interpolation of rehearsal in intricate modeled
sequences distributes the learning; this reduces
loss through intraserial interference from other

displayed elements (Margolius & Sheffield, 1961).
Reproduction of matching responses, either on an
overt or covert level, also provides the observer
with opportunities to identify the response
elements that he has failed to learn and thus to
direct his attention to the overlooked modeling
cues during subsequent exposure (McGuire,

1961). Finally, periodic reproduction of modeled
segments is likely to elicit and to sustain greater
attentiveness to modeling stimuli than passive
observation of lengthy, uninterrupted sequences
of behavior (Hovland, Lumsdaine & Sheffield,

1949; Maccoby, Michael, & Levine, 1961).
Symbolic coding operations, to which reference
was made earlier, an even more efficacious than
rehearsal processes in facilitating long-tern

retention of modeled events. During exposure to
stimulus sequence observers are inclined to code,
classify, and reorganize elements into familiar and

more easily remembered schemes (Bower, 1969;
Mandler 1968; Paivio, 1969; Tulving, 1968). These
coding devices may take various forms, such as
representing stimulus elements in vivid imagery
translating action sequences into abbreviated
verbal systems, and grouping constituent patterns
of behavior into larger integrated units. The

benefits accruing from rehearsal may, in fact, be
largely attributable not to associative
strengthening effects of repetition, but rather to
coding and organizational processes operating
during repeated enactments.
Decrements in retention often result primarily

from interference or unlearning arising from
either previously acquired contents or succeeding
observational inputs. These interference processes

are most influenced by the rate, temporal
distribution, and serial organization of stimulus
inputs. Under massed exposure conditions where

modeling stimuli are presented in lengthy,
uninterrupted sequences, substantial interference
effects are created which not only impair
retention, but may result in the development of
highly erroneous modeling responses. In one
study (Bandura, Grusec, & Menlove, 1966), for
example, children who had observed five

relatively complex sequences of modeled
responses during a single exposure sometimes
erroneously combined elements from the different
segments in subsequent behavioral reproductions.
The amount of forgetting and interpattern

intrusion will vary with the degree of similarity of
behavioral elements in the various modeled
sequences. On the other hand, modeling cues that
are presented in smaller units and at spaced
intervals are much less susceptible to loss through

associative interference.
MOTOR REPRODUCTION PROCESSES
The third major component of modeling
phenomena involves the utilization of symbolic
representations of modeled patterns in the form of
imaginal and verbal contents to guide overt
performances. It is assumed that reinstatement of
representational schemes provides a basis for self-

instruction regarding the manner in which
component responses must be combined and
sequenced to produce new patterns of behavior.
The process of representational guidance is

essentially the same as response learning under
conditions where a person behaviorally follows an
externally depicted pattern, or is directed through
a series of instructions to enact novel response
sequences. The only difference is that, in the latter

cases; performance is directed by external cues,
whereas, in delayed modeling, behavioral

reproduction is monitored by symbolic
counterparts of absent stimuli.
The rate and level of observational learning
will be partly governed, at the motor level, by the

availability of necessary component responses.
Behavior patterns of high-order complexity are
produced by combinations of previously learned
components which may, in themselves, represent

relatively intricate compounds. Modeled response
patterns are most readily achieved when they
require primarily the synthesis of previously
acquired components into new patterns exhibited
by models. On the other hand, observers who lack

some of the necessary components will, in all
probability, display only partial reproduction of a
model’s behavior. In such cases, the constituent
elements first must be established through
modeling and then, in a stepwise fashion,

increasingly complex compounds can be acquired
imitatively. Thus, for example, when a mute

autistic child failed to imitate the word baby, the
therapist modeled the component sounds, and
after these elements were established through

imitation, the child readily reproduced the word
baby (Lovaas, 1966b). As will be illustrated later,
graduated modeling procedures have proved
highly effective for modifying gross behavioral

deficits.
In many instances modeled response patterns

have been acquired and retained in
representational forms but they cannot be
reproduced behaviorally because of physical

limitations. Few basketball enthusiasts could ever
successfully match the remarkable performances
of a towering professional player regardless of
their vigilance and dutiful rehearsal.
Accurate behavioral enactment of modeling
cues is also difficult to achieve under conditions
where the model’s performance is governed by
subtle adjustment of internal responses that are
unobservable and not easily communicable. An
aspiring operatic singer may benefit considerably
from observing an accomplished voice instructor;
nevertheless, skilled vocal reproduction is
hampered by the fact that the model’s laryngeal

and respiratory muscle responses are neither
readily observable nor easily described verbally.
The problem of behavioral reproduction is further
complicated in the case of highly coordinated

motor skills, such as golf, in which a person cannot
observe many of the responses he is making and
must therefore primarily rely upon proprioceptive
feedback cues. For these reasons, performances
that contain many motor factors usually require,

in addition to the guidance of a proficient model,
some overt practice.
INCENTIVE AND MOTIVATIONAL PROCESSES
A person may acquire and retain modeled
events and possess the capabilities for skillful

execution of modeled behavior, but the learning
may rarely be activated into overt performance if
negative sanctions or unfavorable incentive
conditions obtain. Under such circumstances,

when positive incentives are introduced
observational learning is promptly translated into
action (Bandura, 1965b). Incentive variables not
only regulate the overt expression of matching

behavior, but they also affect observational
learning by exerting selective control over the
modeling cue to which a person is most likely to
be attentive. Further, they facilitate selective
retention by activating deliberate coding and

rehearsal of modeled responses that have high
utilitarian value.
It is evident from the foregoing discussion that
observers do not function as passive video-tape

recorders which indiscriminately register and
store all modeling stimuli encountered in

everyday life. From a social-learning perspective,
observational learning constitutes a complex
multiprocess phenomenon in which absence of
appropriate matching responses following

exposure to modeling stimuli may result from
failures in sensory registration, inadequate
transformation of modeled events to symbolic
modes of representation, retention decrements,
motor deficiencies, or unfavorable conditions of

reinforcement.
Establishment of New Response Patterns

through Modeling
Research and theoretical interpretations of
learning processes have focused almost

exclusively on a single mode of response
acquisition which is exemplified by the operant or
instrumental conditioning paradigm. In this

procedure an organism is instigated, in one way or
another, to perform responses, and
approximations progressively closer to the
desired final behavior are selectively reinforced. It
is generally assumed that complex human
behavior is likewise developed under naturalistic

conditions through this type of gradual shaping
process.
Fortunately, for reasons of survival and
efficiency, most social learning does not proceed
in the manner described above. In laboratory

investigations of learning processes
experimenters usually arrange comparatively
benign environments in which errors will not
produce fatal consequences for the organism. In

contrast, natural settings are loaded with
potentially lethal consequences that unmercifully
befall anyone who makes hazardous errors. For

this reason, it would be exceedingly injudicious to
rely primarily upon trial-and-error and successive
approximation methods in teaching children to
swim, adolescents to drive automobiles, or adults
to master complex occupational and social tasks. If
rodents, pigeons, or primates toiling in contrived
situations could likewise get electrocuted,

dismembered, or bruised for errors that inevitably
occur during early phases of learning, few of these
venturesome subjects would ever survive the
shaping process.
Apart from the question of survival, it is

doubtful if many classes of responses would ever
be acquired if social training proceeded solely by
the method of successive approximations through
differential reinforcement of emitted responses.

The technique of reinforced shaping requires a
subject to perform some approximation of the
terminal response before he can learn it. In

instances where a behavioral pattern contains a
highly unusual combination of elements selected
from an almost infinite number of alternatives the
probability of occurrence of the desired response,
or even one that has some remote resemblance to
it, will be zero. Nor is the shaping procedure likely
to be of much aid in evoking the necessary

constituent responses from spontaneously
emitted behavior. It is highly doubtful, for
example, that an experimenter could teach a
mynah bird the phrase “successive
approximations” by selective reinforcement of the

bird’s random squeaks and squawks. On the other
hand, housewives establish extensive verbal
repertoires in their feathered friends by verbally
modeling desired phrases either in person or by
means of recordings. Similarly, if children had no

exposure to verbalizing models it would probably
be impossible to teach them the kinds of verbal

responses that constitute a language. In cases
involving intricate patterns of behavior, modeling
is an indispensable aspect of learning.
Differential reinforcement alone can be
employed to evoke new patterns of behavior
under conditions where responses are composed
of readily available elements, stimuli exist that are

capable of arousing actions that resemble the
desired pattern, erroneous responsiveness does
not produce injurious consequences, and the
learning agent possesses sufficient endurance.
Even in these cases the response acquisition

process can be considerably shortened and
accelerated by the provision of appropriate social
models. This is particularly true if a pattern of
behavior contains some elements that are rarely
performed. For example, Luchins and Luchins

(1966) found that college students made over a
thousand errors and never did fully acquire a

complicated sequence of behavior when the only
response guidance they received was in the form
of differential feedback of correctly performed

elements. By contrast, subjects provided with
reinforced models learned the entire role behavior
rapidly and were spared the exasperation and
frustration evidenced by the trial- and-error

group.
A similar problem arises if the presence of

dominant pre-established behaviors precludes
emission of the desired subordinate responses
which seldom occur and, therefore, cannot be

influenced by reinforcement (Bandura & Harris,
1966; Bandura & McDonald, 1963). An experiment
designed to test whether moral judgments reflect
a fixed developmental sequence, as suggested by
Piaget’s theory (1948), or are modifiable by social-

learning variables illustrates the latter point. In
one condition of the study (Bandura & McDonald,
1963), children who exhibited a predominant,
subjective moral orientation either observed adult
models who expressed objective moral judgments,
or had no exposure to the models but were
positively reinforced whenever they expressed
objective moral judgments that ran counter to
their dominant evaluative tendencies. The

provision of models was found to be highly
effective in altering the children’s judgmental
responses (Figure 3-4). On the other hand, the
reinforcement procedure alone effected little
change in the children’s judgmental orientation

because of the relative absence of the desired
behavior.
It is evident from informal observation that
vicarious learning experiences and response

guidance procedures involving both symbolic and
live models are utilized extensively in social
learning to short-circuit the acquisition process.

Indeed, it would be difficult to imagine a culture in
which the language, mores, vocational and
avocational patterns, familial customs, and
educational, social, and political practices were
Figure 3-4. Mean percentage of objective moral judgment
responses produced by subjective children who were
either reinforced for objective judgments or exposed to
reinforced models who exemplified an objective
evaluative orientation. Plotted from the data of
Bandura & McDonald, 1967.
shaped in each new member through a gradual
process of differential reinforcement without the
response guidance of models who exemplify the

accumulated cultural repertoires in their own
behavior. In social learning under naturalistic
conditions responses are typically acquired
through modeling in large segments or in toto
rather than in a piecemeal, trial-and-error fashion.
Much social learning is fostered through

exposure to behavioral modeling cues in actual or
pictorial forms. However, after adequate language
development is achieved, people rely extensively

upon verbal modeling cues for guiding their
behavior. Thus, for example, one can usually
assemble relatively complicated mechanical

equipment, acquire rudimentary social and
vocational skills, and learn appropriate ways of
behaving in almost any situation simply by
matching the responses described in instructional
manuals. If the relevant responses are specified
clearly and in sufficient detail, verbally symbolized
models may have effects similar to those induced

by analogous behavioral displays (Bandura &
Mischel, 1965). The use of verbal forms of
modeling makes it possible to transmit an almost
infinite variety of values and response patterns
that would be exceedingly difficult and time

consuming to portray behaviorally.
The foregoing discussion is relevant to the
issue of instructional control of behavior. In
investigating the process of verbal control it is

essential to distinguish between the instigational
and the modeling functions of instructions.
Instructions are most likely to result in correct

performance when they both activate a person to
respond and describe the appropriate responses
and the order in which they should be performed.
Little would be gained, for example, by simply
instructing a person who has had no prior contact
with cars to drive an automobile. In studies
comparing the relative efficacy of instructions and

verbal modeling (Masters & Branch, 1969), both
types of influences produce their effects through
verbal modeling and they differ only in the
explicitness with which the required responses
are defined. As might be expected, greater

performance gains are achieved when the desired
behavior is clearly specified than when it must be
inferred from a few examples.
The basic components in the development of

complex integrated units of behavior are usually
present in subjects’ behavioral repertoires as
products either of maturation or of prior

observational learning and instrumental
conditioning. For example, persons can produce a
variety of elementary sounds as part of their
natural endowment. By combining existing sounds
one can create a novel and exceedingly complex
verbal response such as
supercalifragilisticexpialidocious. Similarly, people

are endowed with the capacity to move their
fingers, but intricate sequential arrangements of
movements are required to perform a piano
concerto. While most of the elements in activities
that are typically modeled in studies of

observational learning are undoubtedly present,
the particular combination of components in each
response may be unique.
There have been numerous experiments of

observational learning ii infrahuman species
dating back to the early studies of Thorndike
(1898) and Watson (1908). These initial

investigations, which were conducted at a time
when interpretations of imitation as instinct were
in vogue summarily dismissed the existence of
observational learning on the basis of
disappointing results from a few animals tested
under weak incentives and conditions that failed
to ensure adequate observation of the

demonstrator’s performance. Subsequent studies
conducted under more favorable experimental
conditions have generally shown that primates
can learn to solve manipulative problems (Hayes
& Hayes, 1952) and animals of lower order can

acquire discriminations (Bayroff & Lard, 1944;
Church, 1957; Miller & Dollard, 1941; Solomon &
Coles, 1954), lever-pressing responses (Corson,
1967), and escape behavior (Angermeier, Schaul,
& James, 1959) and can master relatively complex

tasks (Herbert & Harsh, 1944) more rapidly
through observation than the original models

achieved by trial-and-error or response-shaping
techniques. For example, Warden and his
associates (Warden, Fjeld, & Koch, 1940; Warden

& Jackson, 1935) spent a considerable amount of
time in training rhesus monkeys by trial-and-error
methods to master four problem-solving tasks in
which the animals opened doors to obtain raisins

by pulling chains, turning knobs, or manipulating
latches in certain prescribed ways. Following

training, the primate models manipulated the
puzzle devices while naive monkeys, presented
with a duplicate set of problems, observed the
skilled demonstrators. The naive observers

achieved instantaneous imitative solutions in 76
percent of the test trials! Adler and Adler (1968)
found that puppies solve problems through

observational learning soon after their eyes
become functional. Results of several experiments

(Darby & Riopelle, 1959; Herbert & Harsh, 1944)
show that the increments in performance
resulting from observation are not attributable to

the fact that the model’s demonstration may have
simply drawn attention to relevant nonsocial
stimuli in the situation.
The animal studies, with few exceptions, have

involved relatively simple responses that were
reproduced either simultaneously or immediately
after demonstration. Although relevant

comparative data are lacking, it is highly probable
that, unlike humans who are capable of acquiring
observationally and retaining large integrated

units of behavior, lower species would display a
limited capacity for delayed reproduction of
modeling stimuli due to sensory-motor
deficiencies. Delayed imitation also requires some
capacity for symbolization since the absent

modeling stimuli must be retained in symbolic
memory codes. As might be expected, the most
striking evidence of observational response
learning in animals comes from naturalistic
studies of both immediate and delayed imitation
of human responses by primates reared in human
families (Hayes & Hayes, 1952; Kellogg & Kellogg,
1933). Field studies of primate social behavior
(Imanishi, 1957; Kawamura, 1963) likewise

provide dramatic illustrations of the manner in
which idiosyncratic patterns of behavior are
acquired and transmitted to other members of the
subculture through modeling. The propagation
process is greatly influenced by pre-existing

associational networks and the social status of the
innovator.
The available cross-species data thus suggest
that the rate and level of observational learning

will be governed by the extent to which subjects
possess the requisite sensory capacities for
accurate receptivity of modeling stimuli, the motor

capacities necessary for precise behavioral
reproduction, and the capacity for representational
mediation and covert rehearsal, which is crucial for
successful acquisition and long-term retention of
extended complex sequences of behavior.
In the case of humans a wide variety of
response patterns differing considerably in

content, novelty, and complexity have been
transmitted through modeling procedures under
laboratory conditions. Among the diverse classes
of behavior that have been developed are included
stylistic response patterns (Bandura, Grusec, &

Menlove, 1966; Bandura, Ross, & Ross 1963b),
distinctive modes of aggressive behavior
(Bandura, Ross, & Ross, 1963a; Hicks, 1965; Kuhn,
Madsen, & Becker, 1967), dramatic play patterns

(Marshall & Hahn, 1967), prosocial frustration
reactions (Chittenden, 1942), and teaching styles
(Feshbach, 1967; McDonald & Allen, 1967). At an

even higher level of complexity, it has been shown
that through exposure to the behavior of models
persons can acquire standards for self-
reinforcement and self-evaluative responses
(Bandura & Kupers, 1964; Bandura & Whalen,
1966; Bandura, Grusec, & Menlove, 1967b),
conceptual behavior (Flanders & Thistlethwaite,

1969; Reed, 1966), moral judgmental orientations
(Bandura & McDonald, 1963), self-imposed delay-
of-gratification patterns (Bandura & Mischel,
1965), linguistic structures (Lovaas, 1966a), and
distinctive phonetic variations in verbal behavior

(Alyokrinskii, 1963; Hanlon, 1964).
GENERALITY OF MODELING INFLUENCES
It is widely assumed, on the basis of evidence

that people often produce new responses which
they have never formed or seen before, that
learning principles cannot account for innovative
behavior. Theories employing modeling principles
have often been similarly questioned on the
mistaken assumption that exposure to the
behavior of others can produce at most mimicry of
specific modeled responses.
In most experimental investigations of
modeling processes a single model exhibits a

limited set of responses, and observers are
subsequently tested for precise response
duplication under similar stimulus conditions

These restricted experimental paradigms cannot
yield outcomes that extend beyond the particular
responses demonstrated. On the other hand

studies employing more complex procedures
indicate that innovative behavior, generalized
behavioral orientations, and principles for
generating novel combinations of responses can
be transmitted to observers through exposure to

modeling cues. Under conditions in which
opportunities are provided to observe the

behavior of heterogeneous models (Bandura, Ross,
& Ross, 1963b), observers typically display novel

patterns of behavior representing diverse
combinations of elements from the different
models. Illustrations of the efficacy of modeling
procedures for developing generalized conceptual

and behavioral propensities are provided in
studies designed to modify moral judgmental
orientations (Bandura & McDonald, 1963) and
delay-of-gratification patterns of behavior
(Bandura & Mischel, 1965). In these experiments

the models and observers respond to entirely
different sets of stimuli in the social-influence
setting. Tests for generalized modeling effects are
conducted by different experimenters, in different
settings, with the models absent, and with

different stimulus items. The results disclose that
observers respond to new stimulus situations in a

manner consistent with the models’ dispositions
even though the subjects have never witnessed the
models’ behavior in response to the same stimuli.
In the higher-order form of modeling
described above the modeling stimuli convey
information to observers about the characteristics
of appropriate responses. Observers must abstract

common attributes exemplified in diverse
modeled responses and formulate a principle for
generating similar patterns of behavior. Responses
performed by subjects that embody the
observationally derived rule are likely to resemble

the behavior that the model would be inclined to
exhibit under similar circumstances, even though
subjects had never witnessed the model’s
behavior in these particular situations. The
abstraction of rules from modeling cues is

achieved through vicarious discrimination
learning (Bandura & Harris, 1966), in which the

model’s responses containing the relevant
attributes are reinforced, whereas those that lack
the critical features are consistently nonrewarded.
Although modeling variables play an important
role in the development of most social behaviors,
their position with respect to language learning is
unique. Since individuals cannot acquire words

and syntactic structures without exposure to
verbalizing models, it is obvious that some amount
of modeling is indispensable for language
acquisition. However, because of the highly
generative character of linguistic behavior, it is

usually assumed that imitation cannot play much
part in language development and production. The
main argument, which is based on the mimicry
view of modeling, is as follows: Children can
construct an almost infinite variety of sentences

that they have never heard. Consequently, instead
of imitating and memorizing specific utterances

that they have heard, children learn sets of rules,
on the basis of which they can generate an
unlimited number of grammatical sentences.
It is obvious that rules about grammatical
relations between words cannot be learned unless
they are exemplified in the verbal behavior of
models. An important question therefore concerns

the conditions that facilitate abstraction of rules
from verbal modeling cues. The principle
underlying a model’s varied responses can be
most readily discerned if its identifying
characteristics are repeated in responses

involving a variety of different stimuli. If, for
example, one were to place a series of objects on
tables, on chairs, on boxes and on other objects,
and simultaneously verbalized the common
prepositional relationship between these objects,

a child would eventually discern the grammatical
principle. He could then easily generate a novel

grammatical sentence if a toy hippopotamus were
placed on a xylophone and the child were asked to
describe the stimulus event enacted.
Unlike social responses which are often readily
acquired, language learning is considerably more
difficult, because sentences represent complex
stimulus patterns in which the identifying features

of syntactic structures cannot be easily
discriminated. The influential role of both
modeling and discrimination processes in
language development is shown by findings of an
experiment (Bandura & Harris, 1966) designed to

alter the syntactic style of young children who had
no formal grammatical knowledge of the linguistic
features that were manipulated. The grammatical
constructions chosen to be modified were the
prepositional phrase, which has a high base rate of

occurrence, and the passive voice, which is
grammatically more complex and rarely displayed

by young children.
As might be expected, social reinforcement,
even when combined with a strong attentional set
to identify the characteristics of “correct”
sentences, was ineffective in increasing the use of
passives in sentences generated by the children in
response to a set of simple nouns. The majority of

subjects did not produce a single passive sentence,
and consequently, no responses occurred that
could be reinforced. Nor were the children able,
within the relatively brief exposure period, to
discern the critical syntactic category simply from

observing a model construct a series of passive
sentences. In contrast, children generated
significantly more passives when verbal modeling
cues were combined with procedures designed to
increase syntactic discriminability. The most

powerful treatment condition was one in which
the attentional set was induced, modeled passive

constructions were interspersed with some
sentences in the active voice so as to enhance
differentiation of relevant grammatical properties,

and both the model and the children were
rewarded for passive constructions. In the case of
a syntactic category as common as prepositional
phrases, reinforcement together with an active

attentional set were effective in altering children’s
usage of prepositions, but modeling cues were not

a significant contributory factor.
Further evidence for the influential role of
modeling processes in language acquisition is

provided by naturalistic studies involving
sequential analyses of children’s verbalizations
and the immediately following parental responses.
Such studies disclose that young children’s speech
is at best semi-grammatical; in approximately 30

percent of instances adults repeat children’s
verbalization in a grammatically more complex
form accenting the elements that may have been
omitted or inaccurately employed (Brown &
Bellugi, 1964); and children often reproduce the
more complicated grammatical reconstructions
modeled by adults (Slobin, 1968).
The promising findings based on laboratory
studies of modeling processes indicate that an

efficacious program of behavioral modification is
one in which change agents model the behaviors
they wish their clients to acquire. During recent
years, a number of modeling procedures have
been devised and systematically applied to effect

psychotherapeutic changes. These treatment
approaches are reviewed next.
ELIMINATION OF DEFICIT CONDITIONS THROUGH

MODELING
Many of the generalized behavior disorders
that are most intractable are characterized by
gross deficits not only in behavior but also in the
basic psychological functions essential for
learning. The more severe cases, such as autistic

children and adult schizophrenics, generally
manifest little or no functional speech; they lack
social skills that are conducive to reciprocally
rewarding relationships; and interpersonal
stimuli, which ordinarily serve as the principal

medium of social influence, often have relatively
little impact on them. Since human behavior is

largely acquired through modeling and regulated
by verbal cues and symbolic reinforcers, profound
deficiencies in functions of this nature create
major obstacles to treatment. These issues are

best exemplified by the treatment of autism.
The elimination of autistic behavior is further
complicated by the fact that such children are
characteristically engrossed in repetitive motor
activities and other forms of self-stimulatory

behavior. Consequently, they remain oblivious
much of the time to relevant environmental
influences. The marked self-isolation is also
generally coupled with strong resistance to
situational demands, as evidenced by their
unwillingness to perform appropriate responses
that they are obviously capable of making (Cowan,

Hoddinoth, & Wright, 1965). When behavioral
demands within their capabilities are firmly
applied, the children are inclined to avoid
responding by evading the therapist or by
resorting to tantrums and bizarre motor activities

(Lovaas, 1966a; Colby, 1967). After such aversive
behaviors lose their functional value for avoiding
social demands through consistent
nonreinforcement, autistic children typically
respond with appropriate behavior (Risley & Wolf,

1967). However, the aversive countercontrol and
lack of positive responsiveness eventually

extinguish the concerted efforts of less durable
therapists. Disappointing treatment outcomes,
therefore, are frequently attributed to

neurophysiological malfunction.
Although physiological variables are probably
contributing factors in autism, it should be noted
that even biologically deficient organisms are

capable of learning provided that appropriate
conditions are arranged. It is evident, however,

from the adverse behavior characteristics of
autism that extraordinary interventions must be
employed, particularly in initial phases, if any
fundamental changes are to be effected in the

psychological functioning of autistic children.
One of the most provocative behavioral
approaches to the treatment of autism, in which
modeling procedures figure prominently, has been

developed by Lovaas and his colleagues (Lovaas,
1967). The therapeutic program is based on the
view that the total rehabilitation of autistic and
schizophrenic children can be best achieved
through the establishment of stimulus functions
which make one amenable to social influence. This
process primarily involves developing children’s
responsiveness to modeling cues, increasing the
discriminative value of stimulus events so that

children attend and respond appropriately to
aspects of their environment that they have
previously ignored, and endowing social approval
and other symbolic stimuli with reinforcing
properties. After a strong modeling set has been

created, and children have become adequately
responsive to environmental influences, the major
task of broadening children’s social and
intellectual competencies can be effectively
carried out by parents, teachers, and other agents.

Since interpersonal communication and social
learning are extensively mediated through

language, the development of linguistic skills is
also selected as a central objective of treatment.
As noted previously, modeling outcomes
depend upon accurate perceptual input. Autistic
children generally show defective reception of
external stimuli, a deficit which has been
attributed by some researchers to

neurophysiological impairment (Hutt, Hutt, &
Ounsted, 1965; Rimland, 1962). It cannot be
determined from the available data whether the
weak registration of external stimuli results from
the interfering effects of high central arousal, from

insufficient activation, from children’s intense
preoccupation with their own self-produced
stimulation, or from some other factors. Whatever
the reasons may be, it is evident that little
headway can be made toward effecting behavioral

change unless adequate control is gained over
children’s attending behavior. Lovaas’ method for

developing language functions in profoundly
autistic children, who display marked withdrawal
and bizarre self-stimulatory behaviors most of the

time, achieves attentional control through several
means. First, the therapist establishes close
physical contact by sitting directly in front of the
child so he cannot easily ignore the responses that

are being modeled. Second, during the session the
child is not permitted to avoid the therapeutic task

by withdrawal or by resorting to bizarre activities.
If necessary, the therapist physically restrains the
child from turning away, he establishes eye
contact by asking the child to look at him, and he

may withhold positive attention, address the child
sharply, or even slap him on the thigh to terminate
stereotyped bizarre behavior. Firm intervention of

this type, if thoughtfully employed, may serve a
therapeutic function when failure to respond

appropriately to situational demands reflects
unwillingness rather than inability. This is
dramatically illustrated in a telling sequence from

a film depicting the language learning program
(Lovaas, 1966b). A therapist repeatedly asks a girl
to name the color of a yellow crayon, to which she
responds with increasingly bizarre arm-flapping
and peculiar grimacing. Finally, the girl is slapped

on the thigh, and instructed to name the color,
whereupon she abruptly ceases the bizarre

behavior and calmly answers, “Yellow.” As a
further means of augmenting and sustaining the
child’s attentiveness to modeling cues, food
rewards, expressions of affection and social

approval are made contingent upon imitation.
If children’s behavioral repertoires are
impoverished, their behavioral reproductions may
be deficient even though they pay close attention

to modeling cues, because the requisite
components for the modeled responses are
lacking. In such cases complex patterns of
behavior must be reduced to small subunits of
behavior, each of which is established through
modeling. Poorly designed learning sequences,
which result in stressful failure experiences,
jeopardize attentional control by reducing the
child’s motivation to observe the modeled

responses and by arousing disruptive escape
behaviors. To obviate this problem modeled
responses are carefully graduated in complexity to
assure the child a high degree of success in
behavioral reproduction.
In teaching autistic children communicative

speech a modeling-reinforcement procedure is
employed in which the therapist displays
progressively more complex forms of verbal

behavior and rewards increasingly closer
reproductions of the modeled responses. In
teaching a mute child to talk, for example, the

therapist first rewards any visual attentiveness
and random sounds made by the child. When
vocalization has been increased, the therapist
utters a sound and the child is rewarded only if he
produces a vocal response within a certain time
limit. After the therapist’s speech is established as
an effective stimulus for the child’s vocalizations,

he is reinforced only for precise verbal
reproduction of specific sounds, words, and
phrases modeled by the therapist. By this method
children are first taught elementary sounds that
have pronounced visual components and can be

manually prompted, and then, in a stepwise
fashion, more complicated utterances and
combinations of words are added. Essentially
similar methods for establishing verbal
imitativeness are described in considerable detail

by Risley & Wolf (1967) in the treatment of
autistic children, and by Sloane, Johnston, & Harris

(1968) in remedial programs for speech-deficient
young children.
As exemplified by a case illustrated in Figure 3-
5, it may require several days for an autistic child
Figure 3-5. Rate of verbal imitation by a previously mute
autistic child during the first 26 days of training. The
words and sounds are printed in lower case letters on
the days they were introduced and trained, and in
capital letters on the days they were mastered. Lovaas,
Berberich, Perloff, & Schaeffer, 1966.
to master the first word, but subsequent imitative
word learning generally proceeds at a
comparatively rapid rate. The fact that the

establishment of two sounds and one verbal
response is accompanied by immediate
production of many new words composed of
elements that were never directly trained
indicates that autistic children possess greater

linguistic competencies and comprehension of
grammatical features than is commonly believed.
One would expect some language acquisition to
occur through observational learning as a function
of extensive exposure to grammatical speech. The

absence of verbal behavior in autistic children
may, therefore, partly represent a motivational

rather than a behavioral deficit. The question
remains as to whether the abrupt rise in
productivity results from children’s acquisition of

a modeling set, from realization that oppositional
tactics have become nonfunctional, or some other
factors.
Lovaas also provides some evidence to indicate
that, during the initial phase of imitation training,
extrinsic incentives may be essential for accurate

observation and reproduction of the therapist’s
performances. Children displayed a high level of
accurate imitative responsiveness when rewards

were made contingent upon matching the adult’s
speech perfectly; by contrast, when children were
equally generously rewarded after a certain time
had elapsed without regard to the quality of their
verbalizations, their imitative behavior

progressively deteriorated until it bore little
resemblance to the model’s responses (Figure 3-
6). However, in later stages of treatment, similar
shifts from response- to time-contingent
reinforcement did not adversely affect modeling
outcomes.
507
www.freepsychotherapybooks.org
Figure 3-6. Percentage of modeled responses correctly and incorrectly reproduced by an autistic child during periods
when rewards were made contingent upon matching perfectly the adult’s speech (response
contingent) or the elapsing of a certain amount of time (time contingent). Lovaas, 1967.
When children are able to imitate new words
they are taught a labeling vocabulary so that they
understand what the words mean. This is achieved
through a form of paired-associate learning in
which the therapist presents an object (e.g., glass
of milk) or models an activity (e.g., claps hands)

and simultaneously provides the correct verbal
label. On succeeding trials the adult’s verbal
prompt is gradually withdrawn until eventually
the child gives correct verbal responses to the
nonverbal events alone. In this way a wide variety

of object-word associations are learned and
discriminated. Reading skills are established in a

similar manner except that letter-picture and
letter-word associations are presented to the
children until they learn to make the appropriate
verbal responses to printed words in the absence
of pictorial or vocal prompts.
After children have been taught to speak and
to correctly label common objects and activities,
training in abstract linguistic functions begins.
This program consists essentially of rewarding the
child’s discriminative responsiveness to verbally

or behaviorally modeled events. Whenever the
child fails to respond or responds incorrectly he is
aided by verbal and manual prompts which are
gradually faded out on succeeding trials.

Prepositional training will illustrate the basic
discriminations that are developed. Behavioral
matching of a verbal stimulus can be more easily

achieved by autistic children than verbally
labeling nonverbal events. Therefore, initially the
adult gives a verbal instruction involving a
preposition (e.g., “Put the ball inside the box”) and
the child is rewarded for performing the motor
response appropriate to the verbal stimulus. If the
child fails to execute the response correctly, the

therapist moves the child’s hand with the ball to
the box while verbalizing the action. In the second
discrimination, objects are arranged in a particular
way and the child is asked to describe verbally the
relationships between the objects, using the

proper preposition. In the third step, which calls
for grammatical conversation, the child responds
verbally to a verbal stimulus (e.g., “Where did I put
the bicycle?”) without concomitant behavioral
enactment of the events to which reference is

made. As in other forms of rule learning, children
are taught to generalize the linguistic rule by

modeling a variety of objects in a variety of
prepositional relationships. Essentially the same
procedures have been successfully employed to

establish increasingly complex forms of linguistic
and conceptual behaviors (Lovaas, Berberich,
Kassorla, Klynn, & Meisel, 1966; Lovaas, Dumont,
Klynn, & Meisel, 1966). In the case of echolalic

children, inappropriate matching responses are
extinguished through reinforcement withdrawal,

but otherwise the training program is similar to
that employed with mute cases. However, since
echolalic children have already developed
imitative speech, they start at a more advanced

level and proceed at a much faster rate.
Formal language training is well suited for
establishing verbal skills but it may result in
speech that is lacking in spontaneity and overly

dependent upon specific external cuing. To
remove this problem, after the requisite skills for
generative grammatical speech have been
established, children are taught to use their
language to initiate and maintain social
interactions, to express their feelings and desires,
and to seek and exchange information about their
environment. Self-generated spontaneous speech
is initially fostered in several ways. First, by

withholding desired objects and activities until
children verbalize their wants, they are taught to
influence and control their environment verbally;
second, they are encouraged to develop comments
and stories about activities depicted pictorially in

magazines and books and are rewarded for
increasingly elaborate and novel verbalizations;
third, they are asked to recount, in detail, past
experiences; and finally, the concepts that they
have learned in the formal tasks are extended into

informal daily interactions. Indeed, as treatment
progresses the formal training procedures are

incorporated into more natural interpersonal
interactions, where verbal approval, affectional
expressions, play activities and a sense of

accomplishment replace primary rewards as
major reinforcing events.
Self-care skills, play patterns, appropriate sex-

role behaviors, intellectual skills, and
interpersonal modes of behavior can be
established in autistic children more rapidly than

linguistic patterns by modeling the appropriate
activities and rewarding the children’s emulations
(Lovaas, Freitag, Nelson, & Whalen, 1967). The

training program in nonverbal behavior relies
upon the same basic methods employed in
language learning. The therapist first establishes
control over children’s attending behavior;
complex response patterns are gradually

elaborated by modeling activities in small steps of
increasing difficulty; manual prompts are utilized
if children fail to respond. The prompts are
gradually withdrawn and reinforcement for
prompted behavior is later withheld to counteract
passive responsiveness. After imitative behavior is
strongly developed, stimulus control of children’s
behavior is shifted from modeling cues to verbal
prompts and appropriate environmental stimuli.

Children may, for example, initially engage in
painting activities only when they are modeled by
an adult, but by reinforcing painting in response to
verbal suggestions and art materials they
eventually learn to pursue such activities without

requiring a performing model.
The encouraging results of the project
described above would suggest that a modeling-
reinforcement approach merits serious

consideration in the treatment of schizophrenic
disorders. Since the beneficial outcomes are
achieved with nurses, parents, and college

students serving in the role of therapists, this
treatment approach gains further social
significance. However, evidence that children vary
tremendously in their rates of learning,
particularly in early stages of training, indicates
the need for comparative studies to evolve
procedures that would permit even greater

control over the change process. For example,
discrimination of modeling stimulus inputs is an
important prerequisite to their acquisition. In the
case of language learning, a brief program of
discrimination pretraining may greatly accelerate

modeling outcomes and reduce variability
resulting from deficiencies in speech perception.
For children who do not know the meanings of
modeled utterances, word reproduction is apt to

be a dull and tiresome exercise. A preliminary
program aimed at producing word comprehension
would make the situation more meaningful and

perhaps facilitate productive word learning. A
sequence similar to this type has been employed
by Humphery (1966) in developing language
functions in autistic children. As a way of ensuring
necessary attentiveness, children are seated in a
semi-darkened room and equipped with
earphones. In the initial language comprehension

phase of the program children see pictures of
objects projected on a screen and hear the
corresponding verbal labels without having to
reproduce them. After the word-object association
has been repeated sufficiently to establish the

meanings of the utterances, children are
reinforced for correct production of modeled
verbalizations. Generalization and discrimination
are not left to chance: Thus, children may first see
a dog as the focal object of a slide, but later it is

presented as part of increasingly complex arrays
of animals that will have to be accurately

discriminated. By including pictures or
demonstrations representing actions, qualifying
attributes, and object interrelationships the same

procedure can be extended to develop
increasingly complex linguistic skills. Humphery
has also found it advantageous to include samples
of the children themselves and their peers

pursuing activities in their natural surroundings,
because the immediacy of these stimuli make

them especially vivid and compelling inputs. This
approach is similar in many respects to language
learning under naturalistic conditions where
children observe a considerable amount of verbal

behavior before they are taught to produce words
and grammatical sentences. However, the optimal
sequences for word and meaning training remain

to be demonstrated.
Except for a few minor applications (Sherman,

1965; Wilson & Walters, 1966), there has been no
systematic use of modeling procedures in the
treatment of adult psychotics. This is all the more
surprising considering that a majority of the
chronic cases suffer from debilitating behavioral
deficits which must be overcome if they are to
function effectively in community life. The relative
neglect of this powerful approach probably results

in large part from therapists’ strong allegiances
solely to operant conditioning methods or to
interview procedures in which a great deal of time
is devoted to analyzing patients’ ineffectual
behaviors.
MODIFICATION OF PREPOTENT RESPONSE

PATTERNS THROUGH SYMBOLIC MODELING

with the use of modeling procedures to overcome
behavioral deficits. In many instances, a change
agent is faced with the opposite problem—that of
eliminating strongly established patterns of
deviant or maladaptive behavior. One might

attempt to accomplish this objective by a program
of differential reinforcement, in which socially

desirable behavior is positively reinforced and
deviant response patterns are either nonrewarded
or punished. Selective reinforcement is often a
slow and inefficient process when a person

displays a strong dominant response tendency and
desired alternative modes of behavior are only
weakly established or nonexistent in his
behavioral repertoire. Under these circumstances,

one may have to wait an unnecessarily long or
indefinite time for the appearance of alternative
responses. In such cases, the change process may
be greatly facilitated by the use of modeling
procedures designed to transmit, elicit, and

support modes of response that are incompatible
with the deviant behavior that a therapist is

attempting to eliminate. This, in effect, was the
strategy employed by Chittenden (1942) in

modifying children’s hyperaggressive and
domineering responses to frustration.
It has been widely assumed on the basis of
psychodynamic theories and energy models of
personality that either vicarious participation in,
or the direct expression of, aggressive behavior

serves to discharge “pent-up energies and affects”
and thereby to reduce, at least temporarily, the
incidence of aggressive behavior. Guided by this
catharsis theory, many parents, educators,
rehabilitation workers, and child psychotherapists

subtly or openly encourage hyperaggressive
children to express aggression in one form or
another. The overall evidence from laboratory
studies (Bandura, 1965a; Berkowitz, 1969)
strongly indicates that psychotherapies employing

these conventional cathartic or abreactive
procedures may be unwittingly maintaining

deviant behavior at its original strength or, more
likely still, increasing it rather than producing the
expected reductions in aggressive tendencies. In

contrast, therapy based upon social-learning
principles would concentrate, from the outset,
upon developing and strengthening constructive
alternative patterns of behavior. Proceeding on

this basis, Chittenden employed symbolic
modeling procedures for altering children’s

aggressive reactions to frustration.
Children who were excessively domineering
and hyperaggressive observed and discussed a

series of eleven 15-minute plays in each of which
dolls, representing preschool children, exhibited
an aggressive solution and a cooperative,
alternative solution to interpersonal conflicts
under circumstances that the children were likely

to encounter in everyday interactions. In addition
to modeling these two competing response
patterns, the consequences of aggression were
shown to be unpleasant and those of
cooperativeness to be rewarding. In one of the
modeled situations, for example, two boys engage
in a fight over the possession of a wagon; during
the struggle the wagon is broken, and both boys
end up unhappy. By contrast, the cooperative

alternative presents the boys enjoying themselves
as they take turns playing with the wagon.
Children for whom the different reactions and
consequences were modeled showed a decrease in
dominative aggressiveness (as measured by

situational tests in which two children were
placed in a room with a single attractive toy)
compared with a group of similarly
hyperaggressive children who received no

treatment. Of even greater interest is the finding
that children who had observed the discriminative
modeling displayed a significant decrease in

domination and an increase in cooperativeness as
assessed from behavior observations in the
nursery school made prior to treatment,
immediately after treatment, and a month later
(Figure 3-7). One cannot determine from these
data the relative contribution of vicarious
reinforcement and modeling to the obtained

outcomes. The children’s spontaneous comments
and enactments during test trials, in which they
were required to provide their own solutions to
social conflicts involving the dolls, indicated that
they had learned the cooperative strategies. Some,

however, gave evidence of also being strongly
affected by the consequences depicted: “Well, let’s
don’t have them fight; I don’t like to have them
bump their faces together, that hurts.…Let’s have
them take turns; then they won’t fight. Let them

ask Darrell (subject’s name) what to do. ‘Ask me,
Sandy and Mandy (dolls’ names). I’ll tell you to

take turns; then you won’t have a fight’
(Chittenden, 1942, pp. 53-54).”
In a preliminary report Gittelman (1965)
illustrates how behavioral enactment methods can
Figure 3-7. Amount of cooperative and domineering
behavior exhibited by hyperaggressive children before
and after receiving symbolic modeling treatment.
Drawn from the data of Chittenden, 1942.
be adapted for modifying aggressive behavior in
older children. They are first asked to describe
situations that typically provoke them to

aggression and belligerence. A hierarchy of
irritating situations is then constructed, ranging
from those causing only mild annoyance to
extremely instigatory ones. The child and other
group members enact these progressively

aggravating situations and practice effective
nonviolent means of coping with them.
The treatment program devised by Chittenden
primarily relied upon modeling techniques. After

desired patterns of behavior have been
established through some form of modeling, their
maintenance will be largely controlled by the

reinforcement practices existing within the
naturalistic setting. Hence, it may be necessary to
arrange favorable consequences to support newly
acquired response patterns. This would apply
particularly to behavior that is ordinarily
associated with less optimal reinforcement
conditions, as in the case of cooperativeness which

was more difficult to establish and to maintain.
The combined use of modeling and reinforcement
procedures is probably the most efficacious
method of transmitting, eliciting, and maintaining
social response patterns.
There is additional evidence that symbolic

modeling approaches, in which desired response
patterns are demonstrated concretely through
play activities, may be especially well suited for

modifying the behavior of young children,
Marshall & Hahn (1967) showed that preschool
children who participated in several sessions of

doll play with an adult who enacted topics
commonly used in children’s play subsequently
increased their dramatic play with peers in daily
interactions. The absence of any significant
changes in the play behavior of control groups of
children who either received the same amount of
adult warmth and attention during the assembly

of blocks and puzzles or had no contact with the
adult indicates that modeling and support of social
play behavior was the major determinant.
The foregoing studies illustrate the way in
which the same method, doll play, is utilized in

radically different ways depending upon whether
one views behavior from a psychodynamic or a
social-learning perspective. In the former case,
children are typically prompted to enact in doll

play assaultive and other negative response
tendencies toward parents, teachers, siblings and
peers which, if transferred to real life situations,

would further exacerbate their problems. In
contrast, the latter approach provides more
satisfactory solutions to interpersonal conflicts
and models beneficial modes of behavior that are
likely to foster positive social experiences.
Results of a study by O’Connor (1969)
involving positive symbolic modeling lend further

empirical support to the above view. Preschool
children were selected who showed extreme
social withdrawal, a behavior problem that often
reflects both deficits in social skills and fear of
close interpersonal contact. Half of these children

were shown a control film, while a matched group
of isolates observed a sound film depicting a
variety of social interactions at a progressively
more spirited activity level. Each filmed sequence

portrayed a child initially watching the ongoing
activities at a distance but eventually joining and
interacting with the children, with evident positive

consequences. In a behavioral assessment
conducted immediately after the treatment
session the controls remained markedly
withdrawn, whereas children who received the
symbolic modeling showed a substantial increase
in social interaction to the baseline level displayed
by nonwithdrawn children (Figure 3-8). With the

provision of adequate practice and reinforcement
of newly established social skills, such behavior
would undoubtedly assume greater functional
value and endure.
OTHER THERAPEUTIC AND INSTRUCTIONAL

APPLICATIONS OF MODELING
Applications of modeling procedures are by no
means confined to children or to grossly deviant

conditions. Behavioral enactment methods are
frequently utilized for a wide variety of purposes
in which people who want to develop new
competencies are provided with actual or
symbolic models of desired behavior. They are

given opportunities to perform these patterns
initially under nonthreatening conditions before

they are encouraged to apply them in their
Figure 3-8. Amount of social interaction shown by
withdrawn children in the symbolic modeling and
control conditions before and after the experimental
sessions. The dotted line represents the amount of
social interaction displayed by a group of nonisolate
children whose behavior was observed at the pretest
phase of the study. O’Connor, 1969.
everyday lives. Since, in modeling approaches, a
person observes and practices alternative ways of
behaving under lifelike conditions, transfer of

learning to naturalistic situations is greatly
facilitated.
Some treatment approaches, such as Kelly’s

(1955) fixed-role therapy, rely almost exclusively
upon modeling procedures. In the initial phase the
therapist writes a personality sketch suitable for

enactment by the client. He is then asked to
perform the role behaviors continuously as if he
were, in fact, the person portrayed in the sketch.
For example, a passive nonassertive person may
be assigned an active assertive role. The new

behavioral patterns, which are usually in marked
contrast to the client’s customary modes of

responses, are consistently enacted for several
weeks or some other preselected period. This

phase of the program is structured to the client as
representing brief experimentation with, rather
than permanent adoption of, new characteristics.
Moreover, the client is never told that he should be

the new character, only that he should act like him
on a trial basis. The emphasis on brief
experimentation and simulation is considered
essential for minimizing the initial threat of
making sweeping changes in one’s mode of life.
Prescribing a role by itself will be of limited

value unless a person knows how to translate it
into concrete actions under a variety of
circumstances. In Kelly’s approach the treatment

sessions, usually scheduled on alternate days, are
mainly devoted to rehearsing the prescribed role
as it might apply to everyday events involving

vocational and social relationships, heterosexual
interactions, parental relations, and life
orientations. Therapist and client usually alternate
in the role enactment. Through such role-reversal
the client not only benefits from the therapist’s
demonstration of skillful ways of relating to
others, but he also experiences how people are

likely to be affected by the behaviors being
modeled.
After new forms of responsiveness to different
types of interpersonal situations have been
adequately rehearsed, and the client’s actual

experiences in implementing the role have been
thoroughly discussed, the client decides whether
or not he wishes to adopt the new role behaviors
on a more lasting basis. If he has found the new

role effective and wishes to go on with the
program the behavioral rehearsals are continued
as long as necessary. With further experience the

client becomes increasingly skillful and
comfortable in the new role behaviors until
eventually they are spontaneously performed.
Although there is every reason to expect from
evidence of modeling studies that the type of
approach advocated by Kelly should be highly

efficacious, there have been no systematic
attempts made to measure the degree of success

associated with this particular method. Research
is also needed to determine whether the
recommended practices—the selection of
markedly contrasting behavior that is

continuously enacted under a simulated set in all
areas of social functioning—are, in fact, the
optimal conditions for establishing new role

behaviors. Desired outcomes might be more
consistently attained by gradual role adoption in

progressively more difficult social situations than
by complete role enactment from the outset.
Under a graduated procedure the behavioral

requirements would be adjusted to the client’s
capabilities at any given time and would hence
reduce the possibility that his initial attempts at
new ways of behaving would be poorly received

by others. By careful selection of both the real-life
situations in which the client enacts new modes of
behavior and the manner in which they are

expressed, the likely consequences of modeled
behavior can be controlled to a considerable
extent rather than left to fortuitous circumstances.
There are many other treatment approaches in

which modeling techniques, variously labeled
psychodramatic enactment (Moreno, 1958; Sturm,
1965), behavior rehearsal (Lazarus, 1966; Wolpe
& Lazarus, 1966), and role playing (Corsini &

Putzey, 1957) are employed to overcome specific
response deficits or to transmit more extensive
repertoires of social behavior. Modeling in the
form of role practice has also been extensively
adapted for training of industrial and managerial
skills (Corsini, Shaw, & Blake, 1961). Strategies to
be followed in implementing modeling principles
are presented in strong prescriptive terms and the
methods are credited with much success, but as is

generally true of the psychotherapy literature,
rigorously controlled studies of outcomes are
virtually nonexistent.
The efficacy of modeling approaches will be
largely determined by what is being enacted. If

change agents mainly encourage clients to
perform their customary ineffectual forms of
behavior, to reconstruct past relationship
experiences, and to revivify the emotional

reactions engendered by their inadequacies, then
these methods are unlikely to fare any better than
interpretive interview approaches that similarly

accentuate the negatives. On the other hand,
treatment approaches that employ modeling
procedures to establish effective modes of
behavior often lack an adequate transfer training
program in which clients are provided with
opportunities to test their newly acquired skills
under conditions likely to produce rewarding

consequences. If change agents themselves
portray requisite interpersonal competencies, and
arrange optimal conditions for their clients to
learn and to practice more effective means of
coping with potential problems, then this type of

approach is almost certain to prove successful.
Before turning to other issues we should like to
comment briefly on the nature of the effects
produced through modeling processes. When

people are deliberately instructed to observe and
to reproduce either the behavior exemplified by
others or an imaginatively reconstructed role,

there may be a tendency to view the resultant
changes as feigned and superficial. In fact, as will
be shown in the concluding chapter, role
enactment techniques have proved to be one of
the most effective means of inducing stable
affective and attitudinal changes. These findings
provide support for the view that self-evaluative

and cognitive events may be partly epiphenomena
arising from one’s competencies and the
consequences of one’s behavior. Modeling, even
under simulated conditions, can have far-reaching
effects.
MODELING PROCESSES IN INTERVIEW

PSYCHOTHERAPIES
It is generally assumed that personality

modifications in conventional verbal treatments
are achieved in part by clients’ identification with
their psychotherapists. However, as Mowrer
(1966) has noted, therapists characteristically
model a very limited range of social behavior, and

what they do exemplify most prominently may
have little utilitarian value for clients. The paucity

of helpful modeling cues applies particularly to
treatment approaches that advocate a behavioral
incognito in which therapists’ feelings, personal
opinions, and social responses are exhibited as

little as possible in order to facilitate the
occurrence of infantile transference reactions. To
the extent that therapists’ taciturnity and
interpretive behaviors are mimicked by clients in

their social relationships, as is not infrequent, they
are apt to be considered as bores or pests. In
contrast to conventional practices that invoke
some degree of therapist ambiguity and
concealment, Mowrer advocates that therapeutic

agents actively model what their clients are
supposed to learn and arrange conditions that will

foster identificatory outcomes. Hence, in integrity
therapy (Drakeford, 1967; Mowrer, 1964), which

is designed to get clients to recognize that they are
partly accountable for their life situations because

of their objectionable and duplicitous behavior,
the therapist himself consistently models self-
disclosure and personal accountability.
During the course of conversational treatment
some of the therapists’ attitudes and personal
preferences are inevitably revealed through their

selective responsiveness and interpretive
comments (Parloff, Iflund, & Goldstein, 1960).
These inferred attitudes are likely to be emulated

by clients even though therapists may strive to
maintain neutrality in the value domain. Some
suggestive evidence of this effect is reported by
Rosenthal (1955) who found that clients who
were judged as showing greatest clinical

improvement changed their values in the areas of
sex, aggression and authority in the direction of
their therapists’ values whereas clients who were
rated unimproved became less like their
therapists. The occurrence of value congruences
during the course of therapy is also shown by
Pentony (1966). It cannot be determined from
these data, however, whether the value
similarities are attributable to modeling or to

differential reinforcement of clients’
verbalizations; undoubtedly both kinds of
influence processes are operative.
There have been several recent
demonstrations that the classes of responses that

traditional psychotherapists are interested in
modifying can be significantly influenced by
modeling procedures. Schwartz & Hawkins (1965)
found that adult schizophrenics whose emotional

statements were positively reinforced in group
therapy increased affective expressions when
their group was provided with two patient models

who frequently verbalized their feelings; under
the same reinforcement conditions affective
responsiveness was decreased when the added
models displayed predominantly nonaffective
verbalizations. Marlatt, Jacobsen, Johnson, &
Morrice (1966) found that interviewees were
more inclined to reveal personal problems after

witnessing a brief waiting-room conversation in
which a model’s self-disclosure was either
accepted or socially rewarded by the interviewer
than if the model’s behavior was discouraged or
subjects had no exposure to a problem-admitting

model.
One of the obstacles to efficient conduct of
interview therapy arises from the fact that clients
are usually confused about what they are

supposed to do in order to achieve beneficial
effects, and verbal explanations inadequately
convey the requisite role behaviors. This

ambiguity can be easily overcome by providing
clients with concrete examples of appropriate
therapeutic responsiveness (Marlatt, 1968a,
1968b). In several studies Truax and his
colleagues (Truax & Carkhuff, 1967) demonstrated
that clients who listened to tape-recorded
excerpts exemplifying self-exploration

(considered to be “good” therapy behavior) prior
to undergoing treatment subsequently achieved
greater positive changes on a variety of
personality tests than did clients who received the
same type of treatment without the initial

modeling experience.
The foregoing studies indicate that modeling
procedures can be successfully employed to
induce changes in verbal behavior. However,

considering the weak relationships that exist
between alterations at the verbal level—whether
in the form of value preferences, verbal

statements, or endorsements of personality test
items—and nonverbal modes of response, it
would seem that models could be used far more
advantageously to promote effective interpersonal
behaviors directly.
Vicarious Conditioning of Emotional

Responsiveness
It is generally assumed that persons develop
emotional responses on the basis of direct painful

or pleasurable stimulation experienced in
association with certain places, people, or events.

Although many emotional responses are
undoubtedly acquired by means of direct classical

conditioning, affective learning in humans
frequently occurs through vicariously aroused
emotions. Many phobic behaviors, for example,
arise not from actual injurious experiences with

the phobic objects, but rather from witnessing
others either respond fearfully toward, or be hurt
by, certain things (Bandura, Blanchard, & Ritter,

1969; Bandura & Menlove, 1968). Similarly,
persons often acquire, on the basis of exposure to
modeled stimulus correlations, intense emotional
attitudes toward members of unpopular minority
groups or nationalities with whom they have had
little or no personal contact.
As suggested above, vicarious emotional
conditioning results from observing others
experience positive or negative emotional effects
in conjunction with particular stimulus events.
Both direct and vicarious conditioning processes

are governed by the same basic principle of
associative learning, but they differ in the source
of the emotional arousal. In the direct prototype,
the learner himself is the recipient of pain- or

pleasure-producing stimulation, whereas in
vicarious forms somebody else experiences the
reinforcing stimulation and his affective

expressions, in turn, serve as the arousal stimuli
for the observer. This socially mediated
conditioning process thus requires both the
vicarious activation of emotional responses and
close temporal pairing of these affective states
with environmental stimuli.
VICARIOUS EMOTIONAL AROUSAL
Experimental investigations of this
phenomenon have been concerned with
determining the factors that govern the degree to

which people become emotionally aroused by the
experiences of others. Some of the studies have

attempted to identify the social cues that are most
influential in producing vicarious arousal, while
still others have been designed to elucidate the

social-learning conditions whereby social cues
become endowed with emotion-eliciting potency.
One of the earliest studies of vicarious affective
arousal was reported by Dysinger & Ruckmick

(1933), who measured the autonomic responses
of children and adults to movie scenes depicting
dangerous situations and romantic-erotic displays.
The findings showed that scenes of danger,
conflict, or tragedy elicited the greatest emotional
reactions among young children, but
responsiveness decreased progressively with

increasing age. The inverse relationship obtained
was attributed to the greater ability of older

persons both to discriminate between fantasied
and realistic situations and to attenuate the
aversiveness of danger cues by forecasting
eventual favorable outcomes. As would be

expected, emotional reactions to erotic scenes
were stronger among subjects in older age groups.
More recent demonstrations of vicarious
emotional instigation through filmed stimulation
is provided in a series of experiments by Lazarus

and his associates (Lazarus, Speisman, Mordkoff, &
Davison, 1962). Continuous recordings of subjects’
autonomic responses were obtained during
presentation of a film portraying a primitive
puberty ritual of an Australian tribe in which a
native boy underwent a crude genital operation.
College students displayed heightened autonomic

responsiveness while viewing the genital
subincision scenes, the reactions being
particularly marked when the operation was
accompanied by sobs and other pain cues on the
part of the young initiate. Both the deletion of the

vocal pain cues and the provision of sound-tract
commentaries that minimized the aversiveness of
the depicted operation significantly reduced the
subjects’ level of emotional arousal; conversely,
commentaries highlighting the suffering and

hazards of such operations enhanced observers’
physiological arousal (Speisman, Lazarus,

Mordkoff, & Davison, 1964).
In an erudite analysis of vicarious processes,
Berger (1962) restricts the phenomenon of
vicarious instigation to situations in which an
observer responds emotionally to a performer’s
presumed affective experiences. Since the
emotional state of another person is not directly

observable, its presence, quality, and intensity is
typically inferred both from stimuli impinging
upon the performer, and behavioral cues
indicative of emotional arousal. As Berger points
out, a person may be vicariously instigated on the

basis of erroneous inferences from stimulus
events, as in the case of a mother who responds
fearfully at seeing her child fall, even though the
child is, in fact, unhurt and undisturbed. Similarly,
a bystander may react apprehensively to hearing a

sudden loud scream although, unknown to him,
the distressing vocalizations are simulated as part

of a game.
Berger has reasoned that a loud scream that
elicits a fear response from the observer may
represent a case of pseudovicarious instigation,
because the vocal cue may serve merely as a
conditioned fear stimulus independent of the
performer’s unconditioned emotional response or

the stimulus situation. The basis for this
distinction is debatable, since expressive cues are
the observable indicants of a performer’s assumed
emotional state and, as will be shown later, it is
precisely because such social cues have acquired

emotion-provoking properties that an observer
can be at all vicariously aroused by the
experiences of another person. There are,
however, instances in which covariations in the
emotional responses of observers and performers

do not necessarily involve vicarious instigation
processes. After a given environmental stimulus

has acquired strong eliciting potency for an
observer, his emotional responses are likely to be
evoked directly by the conditioned stimulus,

regardless of the behavior of others. Thus, for
example, when individuals become fearful upon
hearing the sound of a fire alarm in the building in
which they are working, they may be responding

similarly, because of like conditioning histories,
but independently to the same nonsocial cue.

Under these circumstances it is exceedingly
difficult to establish precisely the stimulus sources
of the observer’s emotional state since the
behavior of others, depending on its character,

undoubtedly augments or reduces the effects of
environmental eliciting stimuli. The most
convincing demonstration of vicarious instigation

is therefore provided under conditions where the
observer’s emotional responses are elicited

entirely by the performer’s affective expressions.
Such conditions are established by ensuring that
the stimuli which elicit emotional responses in the

performer either are unobservable by, or of
neutral valence for, the observing subject.
Miller and his colleagues (Miller, Banks, &
Ogawa, 1962, 1963; Miller, Murphy, & Mirsky,
1959) have identified, through the use of an

ingenious cooperative avoidance-conditioning
procedure, some of the social cues that serve as

conditioned stimuli for affective arousal in
observers. Rhesus monkeys were first trained to
avoid an electric shock by pressing a bar
whenever a stimulus light appeared. After the

avoidance training, the animals were seated in
different rooms, and the bar was removed from
the chair of one monkey and the stimulus light

from the other. Thus, the animal having access to
the light stimulus had to communicate by means

of affective cues to his partner, equipped with the
response bar, who could then perform the
appropriate instrumental response that would

enable both animals to avoid painful stimulation.
Distress cues exhibited by the stimulus monkeys
in anticipation of shock were highly effective in
eliciting fear in their observing companions as
reflected in increased heart rate and rapid
performance of discriminated avoidance

responses (Miller, 1967). The finding that color
slides showing the stimulus animal in fear or pain
elicited more avoidance responses than pictures of

the same animal in nonfearful poses indicates that
simple facial and postural expressions alone are
sufficient cues for eliciting emotional responses.
The investigators further showed that emotional

responses in monkeys could be vicariously
aroused not only by the sight of their experimental
counterparts, but also, through stimulus
generalization, by another monkey who was never
involved in the original aversive contingencies.

Moreover, mere exposure to a monkey reacting in
an apprehensive or fearful manner could reinstate

avoidance responses in the observer after they
had been extinguished to a zero level.
The above studies demonstrate that affective

expressions by others can serve as conditioned
aversive stimuli, but they do not explain how such
cues acquire their potency. That sensitivity to

expressive cues results from social-learning
experiences receives support from Miller, Caul, &
Mirsky (1967), who found that monkeys reared in

total social isolation during their infancy were
unresponsive, either behaviorally or
autonomically, to facial expressions of emotions of
other monkeys. There is evidence that social cues
signifying affective arousal acquire emotion-

provoking properties through essentially the same
process of classical conditioning that is involved in
the establishment of positive or negative valence
for nonsocial environmental stimuli. That is, if
affective expressions of others have been
repeatedly followed by emotional consequences
for observers, affective social cues alone gradually
attain the power to instigate emotional reactions
in observers. In naturalistic situations such

emotional covariations occur frequently. Persons
who are experiencing positive emotions are likely
to treat others in amiable ways which arouse in
them pleasurable affects; conversely, when
persons are dejected, ailing, distressed, or angry,

others are also likely to suffer negative
consequences. The clearest demonstrations of
how vicarious responsiveness is established are
furnished by laboratory studies with infrahuman
subjects in which the requisite social and temporal

contingencies are instituted.
Church (1959) subjected groups of rats either

to paired aversive consequences or unpaired
consequences, or assigned them to a control
condition in which no aversive stimuli were
presented. In the paired-consequences condition
animals were administered brief shocks after
another rat had been shocked for 30 seconds, with
the aversive stimulation to both animals

terminating simultaneously. Animals in the
unpaired-consequences condition received the
same number of brief shocks, but these were not
temporally associated with painful stimulation to
another rat. Following the emotional conditioning

phase of the experiment vicarious emotional
arousal was measured in response to the pain
reactions of another rat that was continuously
shocked in an adjacent cage. Animals that had
previously experienced paired consequences were

markedly affected by the pain responses of
another rat; the control group showed little

empathetic responsiveness; and animals whose
past distressing experiences were unassociated
with the pain responses of another member of

their species showed an effect intermediate
between the two groups.
Conditioning in humans is frequently mediated

through self-generated symbolic stimulation,
which also plays an influential role in vicarious
responding (Bandura & Rosenthal, 1966; Stotland,

Shaver, & Crawford, 1966). In personality theory
vicarious emotional arousal is typically discussed
under the concept of empathy. Within the

personality framework it is generally assumed
that an observer becomes empathetically aroused
as a result of intuiting the experiences and
affective states of another person. The research
reported by Stotland indicates, however, that a

somewhat different process may be involved.
Observers reacted more emotionally to the sight of
a person undergoing painful stimulation when
they were previously asked to imagine how they
themselves would feel if they were being hurt than
when they were told to imagine how the other
person felt during the treatment. These findings
suggest that modeled affective cues produce
vicarious arousal largely through an intervening

self-stimulation process involving imaginal
representation of aversive or pleasurable
consequences occurring to oneself in similar
situations.
Of the various interpersonal determinants of

empathetic responsiveness the perceived
similarity between model and observer has
received greatest attention. It has been generally
found that perceived similarity enhances vicarious

arousal (Stotland, 1969), but why this should be so
has not been adequately established. A likely
explanation could be put in terms of outcome

similarities. One would expect people who possess
similar interests and characteristics to share many
experiences and outcomes in common. It is much
easier for a person to imagine that the
consequences to individuals similar to himself
would apply to him than to imagine the same thing
about the experiences of people with whom he has

little in common. Thus, for example, a person who
often travels the airways is apt to be more
empathetically aroused upon hearing of fatalities
resulting from a commercial airplane accident
than someone who never flies. This explanation

assumes that vicarious responsiveness is based
upon active self-arousal rather than automatic
identification through similarity. Indeed, if people
who possess similar characteristics rarely
experienced concordant outcomes, they would

most likely exhibit weak empathy. The relative
influence of personal similarity and outcome

similarity on vicarious arousal could be best
evaluated by an experiment in which similar
people experience opposite consequences prior to

the empathy test, whereas dissimilar people
encounter identical outcomes. It would be
predicted from social-learning theory that
discrepant outcomes would override the effects of

personal similarity. The strongest empathetic
responsiveness would, of course, be expected to

occur under conditions of high observer- model
similarity and analogous consequences.
VICARIOUS CLASSICAL CONDITIONING
In the preceding section we reviewed some of
the conditions under which emotional responses
of a model, as conveyed through auditory, facial,
and postural manifestations, acquire the capacity

to arouse emotional responses in observers. In the
case of vicarious classical conditioning, the
observers’ vicariously elicited emotions become
conditioned, through contiguous association, to
formerly neutral stimuli. One of the earliest

laboratory investigations of this process was
reported by Kriazhev (1934), who conditioned
one animal in each of seven pairs of dogs to stimuli
presented in conjunction with food or electric

shock, while the other member of the pair merely
witnessed the procedure. The observing dogs
rapidly developed anticipatory salivary responses
to the signal for food, and conditioned agitation

and respiratory changes to the signal for shock.
However, this brief report does not contain
sufficient information on the details of the
experimental procedure to determine whether the
observers’ reactions to the conditioned stimulus

were tested in the absence of the models.
Laboratory investigations of vicarious classical

conditioning in humans (Barnett & Benedetti,
1960; Berger, 1962) typically involve the

conditioning of autonomic responses to neutral
environmental stimuli through observational

experiences. In Berger’s (1962) studies, for
example, one group of observers was informed
that the performing model would receive a shock
whenever a light dimmed, the dimming of the light

being in each trial preceded by a buzzer. A second
group of observers was instructed that the
performer would make a voluntary arm
movement whenever the light dimmed but that he
was receiving no aversive stimulation. In two

other conditions the model was supposedly
shocked but refrained from making arm
movements, or the model was neither shocked nor
withdrew his arm. The measure of vicarious
conditioning was the frequency of observers’

galvanic skin responses to the buzzer, which
served as the conditioned stimulus. Observers

who were informed that the model was receiving
aversive stimulation and who witnessed the model
simulate pain responses by jerking his arm

displayed a greater degree of vicarious
conditioning than observers in the other three
groups. In a further extension of socially mediated
conditioning, Craig & Weinstein (1965) found that

observation of a performer experiencing repeated
failure produces vicarious emotional arousal that

becomes conditioned to previously neutral
environmental cues.
Although the phenomenon of vicarious

conditioning has been clearly demonstrated,
people differ widely in the rate with which they
develop conditioned emotional responses
observationally and in the stability of the acquired
responses. Since this process requires the

observer to experience painful consequences
vicariously, thereby producing affective arousal,
variables that influence an observer’s general level
of emotionality are likely to enhance or retard
vicarious learning. There is some evidence
(Bandura & Rosenthal, 1966) that emotional
arousal is, indeed, a significant determinant of
vicarious conditioning, but the latter variables are
not related in a simple linear fashion. In this

experiment groups of adults observed another
person undergoing aversive conditioning
experiences in which a buzzer sounded at periodic
intervals and shortly thereafter the model feigned
pain, supposedly in response to having received

painful electric shocks. Prior to the vicarious
conditioning phase of the study, the groups of
observers were subjected to differential degrees of
emotional arousal manipulated both
psychologically and physiologically through the

administration of varying doses of epinephrine, a
sympathetic stimulant. The frequency with which

observers manifested conditioned galvanic skin
responses to the buzzer alone was found to be a
positive function of the degree of psychological

stress (Figure 3-9). However, a monotonic
Figure 3-9. Mean percentage of GSRs exhibited by subjects
during the acquisition phase, in which the tone and
model’s pain cues occurred in close temporal
association, and during tests in which the formerly
neutral tone was presented alone to assess its
conditioned aversive properties. The five treatment
conditions represent increasing degrees of affective
arousal. Bandura & Rosenthal, 1966.
decreasing function is obtained when, in addition
to situational stress, subjects experience
increasing physiologically induced arousal. If it

can be assumed that the five treatment conditions
represent incremental levels of emotional arousal

on a single dimension, then the combined results
suggest an inverted-U relationship between
magnitude of arousal and vicarious conditioning.
While the above study establishes a

relationship between arousal level and vicarious
conditioning, the manner in which high arousal
produces disruptive effects remains to be
demonstrated. Subjects’ reports suggested that

disruptive effects may, in part, be mediated by
self-generated competing responses designed to
reduce the aversiveness of the vicarious
instigation situation. In some cases, this took the
form of an intensive focus on irrelevant external
stimuli, to the exclusion of the disturbing pain
cues: “When I noticed how painful the shock was
to him I concentrated my vision on a spot which
did not allow me to focus directly on either his

face or hands.” Most observers attempted to
decrease the aversive stimulation arising from the
model’s pain reaction by conjuring up competing
cognitive activities: “I tried to be cool. I thought
about Latin verbs and about Latin composition.” A

few subjects, however, marshaled considerably
more potent contravening cognitive responses: “I
finally just tried to think about the girl I slept with
last night. It kept my mind off those damn shocks.”
To the extent that an observer who is faced with

distressing events succeeds either in attenuating
unpleasant arousal by producing competing

thoughts or in diverting his attention from
disturbing stimuli, associated stimulus events are
likely to become endowed with relatively weak

aversive properties. In the above experiment
deliberate use of avoidant and stimulus
neutralization stratagems was reported most
frequently by persons in the highest arousal

conditions.
The research discussed thus far has been

entirely concerned with vicarious conditioning
based on autonomic indices. Conditioned
emotionality is also often measured in terms of

behavioral suppression. If unpleasant experiences
are repeatedly paired with a neutral stimulus, it
acquires the power to evoke emotional reactions
that tend to inhibit instrumental behavior in its
presence. Crooks (1967) has shown that strong

behavioral suppression can be established solely
on the basis of observational experiences. After
being tested for the extent to which they handled
play objects, monkeys participated in a vicarious
fear conditioning experiment in which they
observed distress vocalizations sounded (through
a tape recorder) whenever a model monkey
touched a particular object. Later the observers
also received a control conditioning procedure

wherein they witnessed the model’s contacts with
a different object paired with the distress
vocalizations played backwards, thus obliterating
the distressing value of the sounds. In a
subsequent test the observing animals played

freely with the control items, but actively avoided
objects that accompanied supposedly painful
experiences for another animal.
Although emotional behavior is probably often

developed in everyday situations through
vicarious means, there are few occasions when
aversive forms of classical conditioning might be

intentionally employed for therapeutic purposes.
There are clinical reports (Miller, Dvorak, &
Turner, 1960), however, in which aversive
counterconditioning has been applied in a group
setting for creating aversion to alcohol in chronic
alcoholics. Aversion reactions are rapidly
established under such conditions, and most of the

clients display strong vicarious conditioning
effects. Positive vicarious conditioning, on the
other hand, has rarely been employed
systematically to develop empathy, pleasurable
reactions, and favorable social attitudes.
Vicarious Extinction
Emotional response patterns can be

extinguished as well as acquired on a vicarious
basis. Vicarious extinction of fears and behavioral

inhibitions is achieved by having persons observe
models performing fear-provoking behavior
without experiencing adverse consequences. How
avoidance responses can be extinguished without
having been elicited can be best explained in terms
of a dual-process theory of avoidance behavior. As
noted in the previous discussion of causal
processes, conditioned aversive stimuli evoke
emotional arousal that exerts some degree of

control over instrumental responding. It would
follow from this theory that if the arousal capacity
of a threatening stimulus is extinguished, then
both the motivation and one set of controlling
stimuli for avoidance behavior are removed. Black

(1958) has shown that neutralization of an
aversive stimulus through classical extinction
procedures alone markedly facilitates subsequent
elimination of avoidance behavior.
Some early suggestive evidence for the

occurrence of vicarious extinction is provided by
Masserman (1943) and Jones (1924) in

exploratory studies of the relative therapeutic
efficacy of modeling procedures. Masserman
produced strong feeding inhibitions in cats by
pairing food approach responses to a conditioned
stimulus with aversive stimulation. In the
remedial phase of the experiment, the inhibited
animals observed a cagemate, who had never been

negatively conditioned, exhibit prompt approach
and feeding responses. The observers initially
cowered at the presentation of the conditioned
stimulus, but with continued exposure to their
fearless companion, they advanced, at first

hesitantly and then more boldly, to the goal box
and consumed the food. Some of the animals,
however, showed little reduction in avoidance
behavior despite prolonged hunger and repeated
modeling trials. Moreover, avoidance responses

reappeared in a few of the animals after the
fearless cat was removed, indicating that in the

latter cases the modeling stimuli served merely as
temporary external inhibitors of avoidance
responses. Jones (1924) similarly obtained

variable results in extinguishing children’s phobic
responses by having them observe their peers
behave in a nonanxious manner in the presence of
the avoided objects.
Since nonoccurrence of anticipated aversive
consequences is a requisite condition for fear

extinction, the modeling displays most likely to
have strong effects on fearful observers are ones
in which performances that they regard as

hazardous are repeatedly shown to be safe under
a variety of threatening circumstances. However, if
people are to be influenced by modeled behavior
and its accompanying consequences, then the
necessary observing responses must be elicited

and maintained. Presentation of modeled
approach responses toward the most threatening
situation at the outset, as in the studies cited
above, is likely to generate high levels of fear
arousal in observers. To the extent that such
conditions activate avoidance responses (such as
withdrawing or looking away) designed to reduce
vicariously instigated distress, they will impede
vicarious extinction. Therefore, the efficacy of

vicarious extinction procedures may partly
depend on the manner in which modeled
performances are presented.
Avoidance responses can be consistently
extinguished with minimal distress if persons are

exposed to a graduated sequence of aversive
stimuli that progressively approximates the most
feared event. In the application of this stimulus
generalization principle to vicarious extinction,

persons initially observe a model responding in a
positive manner to situations that have low
arousal value. After emotional responses to

attenuated threats have been extinguished,
progressively more aversive modeling cues, which
are weakened by generalization of anxiety
extinction from preceding displays, are gradually
introduced and neutralized. Stimulus graduation is
not a necessary condition for vicarious extinction,
but it permits greater control over the change

process and it entails less anxiety elicitation than
approaches involving repeated exposure to
modeled events having high threat value.
In addition to stimulus exposure variables,
qualitative aspects of the modeled behavior are

likely to influence vicarious extinction outcomes.
The studies of vicarious emotional arousal
reviewed earlier demonstrate that negative
affective impressions by others can serve as

powerful cues for arousing fear and avoidance in
observers. One would therefore expect modeled
approach responses accompanied by positive

affective expressions to produce greater extinction
effects than those accompanied by anxiety. For
example, parental modeling efforts intended to
overcome children’s fears are frequently nullified
because the parents themselves suffer
apprehensions and force themselves into tense
contact with feared objects.
As part of a program of research designed to
elucidate the phenomenon of vicarious extinction,
several efficacious modeling procedures have
been developed for modifying anxiety disorders.
The first study in the series (Bandura, Grusec, &

Menlove, 1967b) involved a stringent test of the
degree to which strong avoidance behavior of long
standing can be extinguished vicariously. It also
explored the possibility that induction of positive

affective responses in observers during exposure
to potentially threatening modeling cues may
expedite the vicarious extinction process.
Young children, who exhibited fear of dogs as
revealed by parental ratings and an actual test of
dog avoidance behavior, were assigned to one of
four treatment conditions. One group participated
in eight brief sessions during which they observed
a fearless peer model exhibit progressively more

fear-provoking interactions with a dog. For these
children, the modeled approach behavior was
presented within a highly positive party context
designed to counteract anxiety reactions. The fear-

arousing properties of the modeled performances
were gradually increased from session to session
by varying simultaneously the physical restraints
on the dog, the directness and intimacy of the
modeled approach responses, and the duration of

interaction between the model and his canine
companion. A second group of children observed

the same graduated modeled performances, but in
a neutral context. In the two treatment conditions

described the stimulus complex contained both
modeling cues and repeated observation of the

feared animal. Therefore, in order to measure the
effects of exposure to the threatening object itself,
a third group of children observed the dog in the
positive context but with the model absent. A

fourth group participated in the positive activities
but was never exposed to either the dog or the

modeled displays.
Following completion of the treatment series,
children were readministered the avoidance test

consisting of the graded sequence of dog
interaction tasks. They were asked, for example, to
approach and to pet the dog, to release her from a
playpen, to remove her leash, to feed her dog
biscuits, and to spend a fixed period of time alone

in the room with the animal. The final and most
difficult set of tasks required the children to climb
into the playpen with the dog and, after having
locked the gate, to pet her and to remain alone
with the animal under the confining, fear-arousing
conditions.
Evidence that deviant behavior can be
modified by a particular method is of limited
therapeutic significance unless it can be

demonstrated that established response patterns
generalize to stimuli beyond those encountered in

treatment, and that induced changes endure after
the therapeutic conditions have been
discontinued. Therefore, the children were
readministered tests for avoidance behavior

toward different dogs after completion of the
treatment program, and again a month later.
The modeling procedure produced highly
stable and generalized vicarious extinction of

avoidance responses (Figure 3-10). The two
groups of children who had observed the peer
model interact fearlessly with the dog displayed
significantly greater approach behavior toward
both the experimental and an unfamiliar animal
than did children in both the dog exposure and
Figure 3-10. Mean dog-approach scores achieved by children
in each of the treatment conditions on the three
different periods of assessment. Bandura, Grusec, &
Menlove, 1967.
control conditions, who did not differ from each
other. The positive context, however, did not
contribute much to the favorable outcomes

obtained. Further evidence for the effectiveness of
this method is that 67 percent of the children
receiving the modeling treatment were able to
remain alone with the dog in the playpen. In
contrast, this ultimate test was met by relatively

few children in the two control conditions.
One would expect, from knowledge of
generalization processes, that vicarious extinction
effects would be partly determined by the variety

of stimulus elements that are neutralized.
Exposure to diverse models who display fearless
behavior toward variant forms of the feared object

without adverse consequences should produce
thorough extinction of fear arousal, and
consequently extensive reduction in avoidance
behavior. On the other hand, observers whose
emotional responsiveness to a restricted set of
modeled aversive elements is extinguished are apt
to display weaker extinction effects. Moreover,

under conditions where a series of aversive
stimuli is presented only once, certain observer
characteristics might also influence the extent to
which emotional responses are extinguished.
Observers who are highly susceptible to emotional

arousal would be inclined to respond to
threatening modeling displays with pronounced
fear and might, therefore, show relatively strong
resistance to vicarious extinction. Thus,
emotionality might serve as an additional

determinant of the rate at which avoidance
behavior is reduced through modeling procedures.
The above propositions were tested in a
second experiment (Bandura & Menlove, 1968)
employing the same assessment methodology
with children who displayed severe dog-avoidance
behavior. In this project, however, the
performances of models were presented in a
series of brief movies in order to test the efficacy

of symbolic modeling techniques that might lend
themselves conveniently to psychotherapeutic
applications. One group of children, who
participated in a single-model treatment, observed
a fearless male model display the same

progressively fear-provoking interactions with a
dog as in the preceding experiment. The second
group of children, receiving a multiple-model
treatment, observed several different girls and
boys of varying ages interacting positively with

many dogs. The size and fearsomeness of the dogs
increased progressively from that of small,

nonthreatening dogs to more massive varieties.
Children assigned to a control group were shown
movies that had no canine characters.
The dog-approach scores obtained by children
in each of the three conditions in the pre-test,
post-test, and follow-up phases of the experiment
are shown graphically in Figure 3-11. Children

who observed approach behavior modeled
without adverse consequences to the performer
displayed enduring and generalized reductions in
avoidance behavior, whereas the controls showed
no changes in this regard. Comparison of the

incidence of terminal performances (remaining
alone with the dog in the playpen) by children
presented with the single-modeling display and
those who witnessed the multiple modeling
showed the latter form of treatment to be superior

for completely eliminating dog-avoidance
behavior. Although modeling was equally effective

regardless of the severity of children’s phobic
behavior, those who manifested a wide variety of
fears benefited somewhat less from the multiple-

modeling treatment than children who had fewer
Figure 3-11. Median dog-approach scores achieved by
children who received either single-model or multiple-
model treatments, or who participated in a control
condition. Bandura & Menlove, 1968.
fears.
As a further test of the therapeutic value of

symbolic modeling, control children were
administered the multiple-model treatment after
the main experiment was completed. The control

children, whose avoidance behavior remained
unchanged in several tests conducted during the
control period, displayed a sharp increase in dog-

approach behavior following treatment. The
increased boldness of one of the control children
who had been subsequently treated is portrayed
in Figure 3-12. The top frames show the model’s
dauntless behavior; the lower frames depict the

child’s fearless interaction with the animals, both
of which she boldly corralled in the playpen, after
the formal test.
Comparison of results of the two experiments
suggests that symbolic modeling is less powerful
Figure 3-12. A girl who was apprehensive about dogs
engaging in fearless interactions with dogs after
exposure to the series of films in which a peer model
displays progressively threatening interactions with
dogs. Bandura & Menlove 1968.
than live demonstrations of essentially the same
behavior. Although the single-model treatment
effected significant reductions in children’s

avoidance responses, it did not sufficiently weaken
their fears to enable them to carry out the
threatening terminal approach behavior. However,
the diminished efficacy of symbolic modeling can

be offset by a broader sampling of models and
aversive stimulus objects. Children who received
the diverse modeling treatment not only showed
continuous improvement in approach behavior
between the post-test and follow-up periods, but

also achieved terminal performances at rates
comparable to equally avoidant children who, in

the previous experiment observed fearless
behavior performed by a single real-life model.

Hill, Liebert, & Mott (1968) and Spiegler, Liebert,
McMains, & Fernandez (1968) have also

successfully eliminated persistent avoidance
behavior in children and adults through brief
symbolic modeling, but in the latter studies the
modeled performances are accompanied by a

persuasive narrative and other fear-mitigating
variables.
The potency of modeling influences in the
transmission of anxiety is widely acknowledged,
but their therapeutic value has sometimes been

questioned (Jersild & Holmes, 1935) on the
grounds that fears persist even though modeling
frequently occurs under ordinary conditions of
life. The effectiveness of any principle of learning
depends not only on its validity but also on the

manner in which it is implemented. Inconsistent,
haphazard, and inadequately sequenced learning
experiences will produce disappointing outcomes
regardless of the cogency of the principle
supposedly guiding the treatment.
In many instances weak fears are undoubtedly
extinguished, or substantially reduced, through
fortuitous naturalistic modeling. However,

carefully planned modeling experiences are
essential for the modification of more tenacious

avoidance tendencies. There is some evidence
(Bandura & Menlove, 1968) that parents of
children who exhibit severe fearfulness make no
attempts to overcome their children’s fears

because they suffer from similar apprehension.
Consequently, they seldom model fearlessness
and, on the infrequent occasions when they do, the

modeling endeavors do not involve carefully
graded presentation of threatening stimuli,

without which this method is not only likely to be
ineffective but may actually exacerbate anxiety
reactions. A not uncommon domestic modeling

scene, for example, is one in which a parent is
busily petting a dog that is jumping about and
simultaneously bidding the child, who is clinging
fearfully, to touch the bounding animal. By

contrast, the modeling treatments, in addition to
utilizing the principle of graduation to reduce fear
arousal, involved concentrated exposures to

modeling displays under protected observation
conditions, and extensive variation of model

characteristics, intimacy of approach behavior,
and aversive properties of the feared object. Had
the modeling sequences been presented in a
widely dispersed and haphazard fashion and
restricted to the more reserved petting responses

by adults (whom children are likely to
discriminate as better able to protect themselves),

the vicarious extinction outcomes might have been
relatively weak and unpredictable.
The third project (Bandura, Blanchard, &
Bitter, 1969) employed an elaborate experimental
design to assess the comparative efficacy of
modeling and desensitization treatment
approaches for producing behavioral, affective,
and attitudinal changes. The participants were

adolescents and adults who suffered from snake
phobias that, in most cases, unnecessarily
restricted their activities and adversely affected
their psychological functioning in various ways.
Some of the people were unable to perform their

jobs in situations in which there was any remote
possibility that they might come into contact with
snakes; others could not take part in recreational
activities such as hunting, gardening, camping, or
hiking, because of their dread of snakes; and still

others avoided purchasing homes in rural areas,
or experienced marked distress whenever they

would be unexpectedly confronted with pet
snakes in the course of their social or occupational
activities.
In the initial phase of the experiment the
participants were administered a behavioral test
that measured the strength of their avoidance of
snakes. In addition, they completed a

comprehensive fear inventory to determine
whether elimination of fear of snakes is associated
with concomitant changes in other areas of
anxiety. Attitudinal ratings on several scales
describing various encounters with snakes and on

the evaluative dimensions of the semantic
differential technique were also obtained. The
latter measures were included to furnish data
regarding the interesting but inadequately
explored attitudinal effects of behavioral changes

induced through social-learning methods.
The cases were individually matched on the

basis of their avoidance behavior and assigned to
one of four conditions. One group participated in a
self-administered symbolic modeling treatment in
which they observed a film depicting young
children, adolescents, and adults engaging in
progressively threatening interactions with a large
king snake (Figure 3-13). To increase even further

the power of this method two other features were
added: subjects were taught to induce and to
maintain anxiety- inhibiting relaxation throughout
the period of exposure, and they were permitted
to regulate the rate of presentation of stimuli by

means of remote control starting and reversing
devices. The rationale for the second feature was
that a self-regulated modeling treatment should
permit greater control over extinction than one in
which persons are exposed to a sequence of

aversive cues without regard to their anxiety
reactions. Subjects were instructed to stop the film

whenever a particular modeled performance
provoked anxiety, to reverse the film to the
beginning of the aversive sequence, and to

reinduce deep relaxation. They then reviewed the
---
,....-!, -i M ' " " ; . A
• . ;._ 'J ' ' • '
Figure 3.13 Children and adults modeling

progressively fear-arousing
interaction with a king snake.
Bandura, Blanchard, & Rtiter, 1969.
threatening scene repeatedly in this manner until
it was completely neutralized before proceeding to
the next item in the graduated sequence. After

subjects became skillful in handling the projector
controls and the self-induction of relaxation, the

experimenter absented himself from the situation,
and the subjects conducted their own treatment
until their anxieties to the depicted scenes were
thoroughly extinguished.
The second group of subjects received a form
of treatment combining graduated modeling with
guided participation. The principal elements of
this method were developed by Ritter (1968,

1969a) as contact desensitization. In the
procedure employed in the present study, the
model initially demonstrates the desired behavior
under secure observational conditions, after which
subjects are aided through further demonstration
and joint performance to execute progressively
more difficult responses. Whenever subjects are
unable to perform a given behavior after
demonstration alone, they enact the feared

activities concurrently with the model. The
physical guidance is then gradually reduced until
they are able to perform the behavior alone.
In the application of this method to the
elimination of snake phobia, at each step the

experimenter himself performed fearless behavior
and gradually led subjects into touching, stroking,
and then holding the snake’s body with first
gloved and then bare hands while he held the

snake securely by the head and tail. If a subject
was unable to touch the snake after ample
demonstration, she was asked to place her hand

on the experimenter’s and to move her hand down
gradually until it touched the snake’s body. After
subjects no longer felt any apprehension about
touching the snake under these secure conditions,
anxieties about contact with the snake’s head area
and entwining tail were extinguished. The
experimenter again performed the tasks

fearlessly, and then he and the subject performed
the responses jointly; as subjects became less
fearful the experimenter gradually reduced his
participation and control over the snake until
subjects were able to hold the snake in their laps

without assistance, to let the snake loose in the
room and retrieve it, and to let it crawl freely over
their bodies. Progress through the graded
approach tasks was paced according to the
subjects’ apprehensiveness. When they reported

being able to perform one activity with little or no
fear, they were eased into a more difficult

interaction.
Subjects assigned to the third group received
the standard form of desensitization treatment
devised by Wolpe (1958). In this procedure deep
relaxation was successively paired with imaginal
representations of snakes arranged in order of
increasing aversiveness. As in the other

conditions, the treatment was continued until
subjects’ anxiety reactions were totally
extinguished or the maximum time allotment was
completed.
Subjects assigned to the control conditions

participated in the behavioral and attitudinal
assessments without intervening treatment. This
group primarily furnished a control for changes
resulting from repeated measurements. A

relationship pseudotherapy was not employed
because several previous investigations have
shown that snake-avoidance behavior is

unaffected by such experiences. In addition, the
controls were later administered the symbolic
modeling treatment without relaxation to evaluate
its contribution to the changes produced by this
method.
Following completion of the treatment series
the assessment procedures were readministered

to all subjects. In order to determine the generality
of extinction effects, half the subjects in each of the
conditions were tested initially with the familiar
brown-striped king snake and then with an
unfamiliar crimson-splotched corn snake that was

strikingly different in coloration; the remaining
subjects were tested with the two snakes in the
reverse order. The behavioral test consisted of a
series of tasks requiring the subjects to approach,

look at, touch, and hold a snake with bare and
gloved hands; to remove the snake from its cage,
let it loose in the room, and then replace it in the

cage; to hold it within five inches of their faces;
and finally to tolerate the snake in their laps while
they held their hands passively at their sides.
Immediately before and during the performance of
each task subjects rated the intensity of their fear
arousal on a 10-interval scale to measure
extinction of affective arousal accompanying

specific approach responses.
As shown in Figure 3-14, control subjects
remained unchanged in avoidance behavior,
symbolic modeling and desensitization produced
substantial reductions, and live modeling

combined with guided participation eliminated
snake phobias in virtually all subjects (92
percent). The modeling procedures not only
extinguished avoidance responses of long

standing, but they also neutralized the anxiety-
arousing properties of the phobic stimuli. Both of
the modeling treatments achieved marked

decrements in anticipatory and performance
anxiety. Although subjects who had received
desensitization treatment also experienced less
emotional arousal while performing snake-
Figure 3-14. Mean number of snake-approach responses
performed by subjects before and after receiving
different treatments. Bandura, Blanchard, & Ritter,
1969.
approach responses, their magnitude of fear
reduction was less than that of their counterparts
in tire modeling conditions.
Findings of this experiment also reveal that
applications of social-learning procedures have
important attitudinal consequences. Both
symbolic modeling and desensitization, which
primarily involve extinction of negative affect

aroused by aversive stimuli, produced favorable
changes in attitudes toward snakes. Consistent
with theoretical expectation, the treatment
condition that reduced the anxiety-arousing

properties of snakes and enabled subjects to
engage in intimate interactions with snakes
effected the greatest attitudinal changes. These

findings will be given more detailed consideration
in a later chapter specifically concerned with
processes governing the modification of attitudes.
Analysis of the fear inventory scores disclosed
some degree of fear reductions beyond the
specifically treated phobia, the decrements being

roughly proportional to the potency of the
treatments employed. Nontreated controls

showed no changes in either number or intensity
of fears. Desensitization produced a decrease only
in severity of fears toward other animals, whereas
symbolic modeling was accompanied by a

reduction in the number of animal fears and a
general diminution in the intensity of anxiety in
several other areas of functioning. Participant

modeling, on the other hand, effected widespread
fear reductions in relation to a variety of threats

involving both interpersonal and nonsocial events.
The transfer obtained reflects the operation of at
least two somewhat different processes. The first

involves generalization of extinction effects from
treated stimuli to related anxiety sources. The
second entails positive reinforcement of a sense of
capability through success, which mitigates

emotional responses to potentially threatening
situations. Having successfully overcome a phobia
that had plagued them for most of their lives,

subjects reported increased confidence that they
could cope effectively with other fear-provoking
events
After the posttreatment assessment, subjects

in the control group received the symbolic
modeling treatment without the relaxation
component. Symbolic modeling alone achieved
substantial decrements in fear arousal and

avoidance behavior: 45 percent of the subjects
exhibited terminal performances toward both
snakes. No significant differences were found in
approach behavior between subjects who were
administered symbolic modeling alone and those
who received symbolic modeling with relaxation.
However, subjects who paired modeling with
relaxation required fewer reexposures to
neutralize the aversive scenes, subsequently

experienced less fear arousal while performing
snake- approach responses, and showed greater
positive changes in their attitudes toward snakes.
In order to determine, in cases involving only
partial improvement, whether the deficiencies

reside in the treatment method or in the subject,
all persons who failed to achieve terminal
performances were subsequently administered
the participant modeling treatment. Snake-phobic

behavior was thoroughly extinguished in all these
subjects within a few brief sessions regardless of
their age, sex, proneness to anxiety, or severity of

avoidance behavior (Figure 3-15). Moreover, this
supplementary treatment produced further
reductions in fearfulness toward other types of
threats, and also additional attitudinal changes.
performed by subjects before and after (post-test)
receiving different treatments. Control subjects were
subsequently administered symbolic modeling
treatment without relaxation. All subjects in the
desensitization, symbolic modeling, and treated control
conditions who failed to perform the terminal
approach behavior were then given the live modeling
and guided participation treatment (post- live). The
snake-approach behavior of subjects in all four groups
was measured again in a follow-up study conducted
one month later. Bandura, Blanchard, & Ritter, 1969.
A one-month follow-up assessment revealed
that the beneficial changes produced in behavior,
attitudes, and emotional responsiveness were

effectively maintained. The clients also displayed
evidence that the behavioral improvements had

generalized from therapeutic to real-life
situations. They were able to participate in
recreational activities, such as hunting, camping,
hiking, and gardening, that they formerly avoided

because of their dread of snakes; they no longer
experienced marked distress when unexpectedly
confronted with snakes in the course of their

social or occupational activities; they were able to
handle harmless snakes; and a few even served as

model therapists for their own children and
fainthearted friends.
Ritter (1968) has obtained similarly uniform
success with group modeling procedures
administered to children who displayed fear of
snakes. Groups of children participated in two 35-
minute sessions in which they either merely
observed several fearless children exhibit intimate

interactions with a snake, or they received the
participant modeling form of treatment during
which the therapist displayed positive responses
toward the snake and then gradually eased the
children into performing the feared behavior.

Snake phobias were completely extinguished in 53
percent of the children by modeling alone and in
80 percent of the children who received modeling
combined with guided participation. The potency
of this approach receives further confirmation by

Rimm & Mahoney (1969), who successfully
extinguished snake-avoidance behavior with

participant modeling in adults who were unable to
achieve behavioral improvement when offered
increasing monetary rewards for performing a

graduated series of approach responses.
Within the participant modeling treatment
three processes are operative that might
contribute in varying degrees to such striking

psychological changes. These include observation
of fearless behavior being repeatedly modeled

without any unfavorable consequences, incidental
information received about the feared objects, and
direct personal contact with threatening objects
that engender no adverse effects. In an experiment

aimed at isolating the relative influence of these
component variables, Blanchard (1969) matched
subjects in terms of their snake-avoidance

behavior and assigned them to one of four
conditions. One subject in each quartet received

the standard procedure, which includes the
benefits of modeling, information, and guided
interaction with a snake. A second subject

simultaneously observed the modeling sessions
and listened to the verbal interchanges, thus being
exposed to both modeling and informational
influences. The third subject received only the
modeling component, while the fourth, who
merely participated in the testing procedures,

experienced none of the constituent influences.
Figure 3-16 summarizes the behavioral, affective,
and attitudinal changes associated with these

various treatment conditions. Modeling accounted
for approximately 60 percent of the behavior
change and 80 percent of the changes in attitudes
and fear arousal; guided participation contributed

the remaining increment. Informational
influences, on the other hand, had no effect on any

of the three response classes.
The guided participation component of the
modeling approach under discussion contains two
important aspects. Participant observers enact
progressively more difficult responses and, if
necessary, the model physically assists them in
Figure 3-16. Percentage of change in approach behavior,
fearfulness, and attitudes displayed by subjects who
received different components of the modeling-guided
participation treatment. Blanchard, 1969.
performing the behavior required at each step in
the graded series of tasks. In order to evaluate the
influence of these elements, Ritter (1969b)

administered one of three treatments to
acrophobic subjects during a single 35-minute
session. For one group of subjects, the
experimenter exhibited increasingly threatening
climbing responses and physically assisted

subjects in performing the same activities; in the
second condition the experimenter demonstrated
the behavior but only verbally guided subjects in
enacting matching performances; a third group
simply observed the demonstrated activities. At

the completion of the session all subjects were
readministered a behavioral test requiring them to

climb to a series of heights atop a seven-story
building. Modeling accompanied by physically
guided performance produced greater changes

than modeling with verbally guided enactment,
which in turn was superior to brief demonstration
alone.
Further research is needed to clarify the
mechanisms through which modeling procedures
achieve extinction effects. Results of the

experiment by Bandura, Blanchard, & Ritter
(1969) provide tentative support for the
proposition that avoidance behavior is reduced

through vicarious extinction of fear arousal.
During the symbolic modeling treatment subjects
rated the intensity of their fear arousal to each
modeled scene and to subsequent reexposures to
the same stimuli. As shown in Figure 3-17,

subjects showed a progressive decline in fear
arousal with each successive exposure to modeled
approach behavior. Subjects who combined
symbolic modeling with relaxation experienced a
greater reduction in fear on the second exposure
to the aversive scenes than their counterparts who
Figure 3-17. Mean level of fear arousal evoked by the
modeling stimuli initially and by each subsequent
exposure to the same filmed scenes in subjects
receiving symbolic modeling with relaxation and
symbolic modeling alone. The data are averaged across
scenes at each exposure and plotted for the first six
exposures only since subjects rarely required more
than six presentations to neutralize any given scene.
Bandura, Blanchard, & Ritter, 1969.
received symbolic modeling alone, but on
subsequent reexposures the rate of fear extinction
was essentially the same. The major theoretical

assumption receives additional support from the
study by Blanchard (1969), who also recorded
decrements in fear arousal with successive
reexposures to modeling stimuli. He found that the
more thoroughly fear arousal was vicariously

extinguished the greater was the reduction in
avoidance behavior and the more generalized
were the behavioral changes.
The process of change associated with the

powerful procedure involving modeling combined
with guided participation may be conceptualized
as follows: Repeated modeling of approach

responses decreases the arousal potential of
aversive stimuli below the threshold for activating
avoidance responses, thus enabling persons to
engage, albeit somewhat anxiously, in approach
behavior. Direct contact with threats that are no
longer objectively justified provides a variety of
new experiences that, if favorable, further

extinguish residual anxiety and avoidance
tendencies. Without the benefit of prior vicarious
extinction, the reinstatement of severely inhibited
behavior generally requires a tedious and
protracted program. After approach behavior

toward formerly avoided objects has been fully
restored the resultant new experiences give rise to
substantial reorganization of attitudes.
The findings of studies reviewed above

indicate that a powerful form of treatment is one
in which therapeutic agents themselves model the
desired behavior and arrange optimal conditions

for clients to engage in similar activities until they
can perform the behavior skillfully and fearlessly.
The therapeutic outcomes associated with this
approach are sufficiently promising to warrant its
further extension to other types of anxiety
conditions. It is undoubtedly best suited for
behavioral dysfunctions in which the feared

consequences are inspectional.
Inhibitory and Disinhibitory Effects of

Vicarious Experiences
In addition to the acquisition of instrumental

and emotional behaviors through observational
experiences, exposure to modeled events may

strengthen or weaken observers’ inhibitions of
well-learned response patterns. The occurrence of
inhibitory effects is indicated when, as a function of
observing negative response consequences to a

model, observers show either decrements in the
same class of behavior, or a general reduction of
responsiveness. It should be noted that when the

subject witnesses behavior that is subsequently
punished, the response-facilitative effects of
modeling cues are counteracted by the
suppressive effects of adverse outcomes. When
these opposing influences are of comparable
strength, persons who have observed modeled

behavior punished and those who have had no
exposure to the model may display an equally low
incidence of response. Therefore, inhibitory effects
can be best evaluated either by measuring
response decrements from baseline levels or by

comparison with performances of subjects who
have observed the same modeled behavior
without any consequences. In the experiment cited
earlier (Bandura, 1965b), for example, children
who had observed a model’s aggressive behavior

result in severe punishment performed
significantly fewer matching responses than

subjects who observed the same actions result
either in reward or in no evident consequences.
Indeed, the vicarious punishment produced

virtually complete suppression of imitative
aggression in girls, whose inhibitions regarding
physical forms of aggression are initially relatively
strong. Further evidence for the suppressive

effects of vicarious punishment is furnished by
studies comparing consistent vicarious reward

with successive reward and punishment of the
model’s behavior (Rosekrans & Hartup, 1967).
Subsequent punishment tends to cancel the
behavioral enhancement effects of rewarding

consequences to the model.
The above studies demonstrate the inhibitory
influence of observed negative outcomes to a
model on the aggressive behavior of viewers.

Walters and his associates (Parke & Walters,
1967; Walters, Leat, & Mezei, 1963; Walters,
Parke, & Cane, 1965) have likewise shown that
witnessing peer models punished for engaging in
forbidden play activities increased observers’
resistance to deviation when they were similarly
tempted with the prohibited objects. In a
comparative study, Benton (1967) found that
observers who witnessed others reprimanded for

handling prohibited toys later showed the same
amount of response inhibition as did the punished
performers. The possible mechanisms through
which vicarious punishment produces inhibitory
effects are discussed in some detail in the

introductory chapter of this book.
In many instances persons respond with self-
punitive and self-devaluative reactions to behavior
of their own that may be considered permissible,

or even rewardable, by others. Results of
investigations concerned with the social
transmission of self-monitoring reinforcement

systems (Bandura & Kupers, 1964; Bandura,
Grusec, & Menlove, 1967b) provide evidence that
witnessing punishments self-administered by a
model inhibits observers from performing the
devalued behavior. Observation of self-
administered reinforcements by a model have
been shown by Porro (1968) to have similar

effects on transgressive behavior. For children
who viewed a filmed model exhibit self-approving
responses to her transgressions, 80 percent
subsequently handled toys they were forbidden to
touch, whereas the transgression rate was only 20

percent for children who had observed the same
model express self-critical reactions concerning
her transgressions.
Behavioral restraints, established through

previous modeling or direct aversive conditioning,
can be reduced on the basis of observational
experiences. Such disinhibitory effects are evident

when observers display increases in socially
disapproved behavior as a function of viewing
models either rewarded or experiencing no
adverse consequences for performing prohibited
responses. The reduction of inhibitions through
modeling has been demonstrated most clearly in
studies of intense physical forms of aggression

that tend to be inhibited in viewers as a result of
past social training (Bandura, Ross, & Ross, 1963a;
Epstein, 1966; Walters & Llewellyn Thomas, 1963;
Wheeler, 1966). It has also been shown (Grosser,
Polansky, & Lippitt, 1951; Ross, 1962) that the

incidence of other types of deviation by observers
is significantly increased as a result of witnessing
models’ unpunished transgressions. On the other
hand, conforming models tend to strengthen the
observer’s self-controlling responses and thereby

to reduce conflictive behavior in tempting
situations (Ross, 1962).
Blake and his associates (Blake, 1958)
conducted investigations of some of the conditions
determining the influence of noncompliant and
conforming models on observers’ inhibitions in
prohibition situations. In one study, Freed,
Chandler, Mouton, & Blake (1955) found that,
although exposure to a noncompliant model

lowered students’ resistance to deviation,
transgressions occurred most frequently if the
restriction was relatively weak and the model
violated the prohibitory signs, whereas the
combination of a strong restriction and a

conforming model produced the lowest incidence
of deviation. A second experiment (Kimbrell &
Blake, 1958) demonstrated that the efficacy of
modeling cues for modifying inhibitions varies
with the observer’s level of instigation to

transgression. Under extreme provocation,
subjects disregarded both the imposed restriction

and the conforming model. However, under
conditions where the instigation was not so strong
as to force deviation, subjects who observed a

conforming model displayed more compliant
behavior than others who witnessed a model
violate the prohibition.
In naturalistic situations observers often
actually see a model’s transgressions rewarded or
punished. At other times, however, they can only

infer probable consequences from discriminative
symbols and attributes of the model that tend to
be correlated with differential reinforcements. The

manner in which distinctive model characteristics
signifying probable outcomes may increase a
model’s effectiveness in reducing inhibitions is
illustrated in an experiment conducted by
Lefkowitz, Blake, & Mouton (1955). Traffic-signal

violations by a presumably high-status person
attired in a freshly pressed suit, shined shoes,
white shirt, and tie produced a higher pedestrian
violation rate than the same transgression
performed by the same model dressed in soiled,
patched trousers, scuffed shoes, and a blue denim
shirt. The differential reduction in restraints noted
in the latter experiment is probably attributable to
the fact that transgressions by persons who

occupy a high position in a prestige hierarchy are
likely to be punished less frequently and less
severely than those performed by low-status
transgressors. The differential leniency is apt to be
temporarily extended to the imitator as well, when

the transgressive behavior is performed at the
same time along with the deviating model.
Other discriminative properties of the model,
such as age, sex, socioeconomic status, social

power, ethnic background, and intellectual and
vocational status, which are associated with
predictable contingencies of reinforcement, may

likewise influence the extent to which prohibited
acts will be imitated. Vicarious reinforcement
effects are, of course, considerably weakened or
nullified under conditions where the model’s
transgressive behavior is so markedly
inappropriate to the sex (Dubanoski, 1967), status,
or social role occupied by the observer that any

imitative tendencies are personally suppressed.
Under most circumstances people readily
adopt modeled responses that seem appropriate
or have utilitarian value. However, in some
problem-solving and achievement-like situations

they display counter-matching tendencies for fear
that imitative behavior will be considered
cheating, copying, or subservience, and, therefore,
socially disapproved (Luchins & Luchins, 1961;

Patterson, Littman, & Brown, 1968; Schein, 1954).
The inhibiting effect of anticipated negative
sanctions for imitation can be overcome in

observers through positive reinforcement of the
model’s responses (Clark, 1965).
It is interesting to note that, when a model
displays punishable behavior, absence of
anticipated adverse consequences increases
transgressive behavior in observers to the same

degree as witnessing the model experience
rewarding outcomes (Bandura, 1965b; Walters,
Parke, & Cane, 1965). These findings suggest that
nonreaction to formerly prohibited activities may

take on, through contrast, positive significance.
Similar contrast-of-reinforcement effects have
been demonstrated in studies of direct
reinforcement (Buchwald, 1959a, 1959b) in which
nonreward following punishment had functioned

analogously to a positive incentive, whereas
nonreward following a series of rewards had

operated as a negative reinforcer. In fact, even a
weak positive incentive, when contrasted with

more rewarding prior events, may acquire
negative reinforcing value (Buchwald, 1960). The

effects of witnessed outcomes on matching
behavior may therefore be determined to a large
extent by the context in which the events occur
and the customary sanctions associated with

particular modeled response patterns.
Because previous studies have utilized deviant

modes of behavior, which may be readily
disinhibited through omission of negative
consequences, the results provide no clear

evidence for the occurrence of positive vicarious
reinforcement. However, findings of an
experiment (Bandura, Grusec, & Menlove, 1967b)
involving modeled behavior that is positively
sanctioned reveal that social rewards dispensed to

a model augment matching responses compared
to a condition in which the exemplified actions
produce no evident consequences.
It is generally easier to disinhibit than to
inhibit response patterns through vicarious
means. The main reason for this difference is that
behavior which is customarily subject to negative
sanctions is often positively reinforcing for the

user, but it is socially suppressed for the
convenience and well-being of others. Thus, for
example, by violating prohibitions and restrictions
people can gratify their immediate needs more

directly and effectively than by following irksome
institutionalized requirements; similarly, by
adopting transgressive behavior they can gain
access to attractive resources that might
otherwise be denied them. Therefore, it does not

require much successful deviant modeling to
reduce vicariously the suppressive effects over

personally rewarding behavior. On the other hand,
inhibitory effects are far more difficult to establish

and to sustain through either direct or vicarious
punishment, when it involves relinquishing

behaviors that lead to immediate and direct
reinforcement. Considering that exposure to
deviant filmed models tends to weaken behavioral
restraints, one would expect televised portrayals

of successfully executed transgressions to have
disinhibitory effects on viewers.
Response Facilitating Effects of Modeling

Influences
The behavior of models often serves merely as
discriminative cues for observers in facilitating the
expression of previously learned responses that

ordinarily are not subject to negative sanctions.
Laboratory and field studies have shown that the

probability of occurrence of a wide variety of
neutral and socially approved behavior can be

substantially increased as a function of witnessing
the action of real-life or symbolic models. Some of
the behaviors that have been thus facilitated
include volunteering one’s services (Rosenbaum,
1956; Rosenbaum & Blake, 1955; Schachter & Hall,
1952), performing altruistic acts (Blake,
Rosenbaum, & Duryea, 1955; Bryan & Test, 1967;
Harris, 1968; Rosenhan & White, 1967), pledging

oneself to a course of social action (Blake, Mouton,
& Hain, 1956; Helson, Blake, Mouton, & Olmstead,
1956), assisting persons in distress (Bryan & Test,
1967), seeking relevant information (Krumboltz &
Thoresen, 1964; Krumboltz, Varenhorst, &

Thoresen, 1967), and selecting certain types of
foods (Duncker, 1938; Barnwell, 1966), activities
(Madsen, 1968), or articles (Bandura, Ross, &
Ross, 1963b; Gelfand, 1962). Some of the most
influential theoretical formulations of imitative

processes (Miller & Dollard, 1941; Skinner, 1953)
have, in fact, been almost exclusively concerned

with the discriminative function of social cues. In
the prototypic experiment the model’s responses
serve as the occasion upon which another

organism is likely to be reinforced for performing
similar responses. After a period of exposure to
differential reinforcement, imitative tendencies
become strongly established; conversely, by

reversing the contingencies so that matching
responses are never reinforced but nonmatching

behavior is consistently rewarded, imitativeness is
reduced to a very low or zero level (Miller &
Dollard, 1941).
Ethologists provide extensive documentation

of the response-facilitating function of social cues
in birds, fish, and mammals (Hall, 1963; Thorpe,
1956). Typically, the sight of certain responses
performed by an animal elicits a similar or

identical pattern of behavior in other members of
the same species. This process is generally
referred to as “social facilitation” or “behavioral
contagion” when it is presumably determined by
prior discriminative reinforcement, and “mimesis”
when corresponding unconditioned response
patterns are supposedly instinctively aroused.
As Hinde (1953) points out, the occurrence of
matching behavior in animals is often erroneously

attributed to mimetic processes. In the first place,
what appears to be mimetic behavior may involve
response patterns that have, in fact, been
established through prior social learning. Even in
cases where matching behavior is clearly

instinctive, it is frequently difficult to determine
whether social cues constitute the critical eliciting
stimuli. Readily discriminable “sign stimuli”
(Tinbergen, 1951) or “releasers” (Thorpe, 1956)

in the form of color displays, preparatory
movement sequences, postural cues, and specific
vocalizations frequently serve as unconditioned

stimuli for complete patterns of instinctive
behavior in other members of the species.
Therefore, when appropriate releasing stimuli are
displayed by a model during the performance of a
given activity, the corresponding responses on the
part of observing animals may be primarily
controlled by releasing stimuli, rather than the

model’s behavioral cues. Thus, for example, the
white tail feathers of a bird flying upward can
function as flight-eliciting stimuli for other
members of a flock (Armstrong, 1942). A suitably
feathered but nonflying artificial model might

likewise succeed in getting a flock of birds
airborne.
Pseudo-mimesis is also evident in instances
where a model’s behavior directs the observer’s

attention to environmental stimuli which, in turn,
elicit similar innate response patterns. It has been
shown, for example, that animals will consume

considerably more food when they are fed in pairs
than when they are fed alone, and satiated
chickens will begin to eat at the sight of other
birds feeding. It is entirely possible that in such
cases modeling cues primarily serve an orienting
function, whereas the consummately responses of
the sociable chicks are reinstated and maintained

by the grain to which their attention has been
redirected. The fact that the stimulus complex to
which observing animals are responding
frequently contains, in addition to social cues,
releasing stimuli and other response-controlling

environmental events complicates the
identification and analysis of genuine mimetic
phenomena.
The behavior of models may not only function

as discriminative cues for similar responses, but
also serve to direct the observers’ attention to the
particular stimulus objects manipulated by the

performer (Crawford & Spence, 1939). As a
consequence, observers may subsequently utilize
the same objects to a greater extent, though not
necessarily in an imitative way. In one modeling
experiment (Bandura, 1962), for example, the
model pummeled a plastic doll with a mallet.
Children who had observed this aggressive act

later displayed significantly more behavior in
which they pounded a peg board with the mallet
than did both the control subjects and those who
had viewed a nonaggressive model. Stimulus
enhancement effects are distinguished from social

facilitation in that the observer’s behavior in the
former case may bear little or no resemblance to
the model’s activities.
It is evident that observers are not equally

affected by the actions of others with whom they
may come into contact, nor are performers equally
influential in evoking the types of behaviors in

which they themselves are engaged. Susceptibility
to social facilitation is largely governed by three
sets of variables that have been discussed at
length in preceding sections and elsewhere. These
include observer characteristics, the
reinforcement contingencies associated with
matching behavior in the particular setting, and

the attributes of the model (Bandura, 1968;
Campbell, 1961; Flanders, 1968).
In learning analyses of response facilitation as
a function of model attributes (Miller & Dollard,
1941), stimulus generalization and differential

reinforcement are utilized as the main explanatory
principles. According to this interpretation, social
models differ in the extent to which their behavior
is likely to be successful in producing favorable

outcomes. Hence, persons are most frequently
rewarded for matching the behavior of models
who are intelligent, who possess certain social and

technical competencies, who command social
power, and who, by virtue of their adroitness,
occupy high positions in various status
hierarchies. On the other hand, the behaviors of
models who are ineffectual, uninformed, and who
have attained low vocational, intellectual, and
social status, are apt to have considerably less

utilitarian value. As a result of differential
reinforcement for matching models who possess
diverse attributes, the identifying characteristics
gradually come to serve as discriminative stimuli
that signify the probable consequences associated

with behavior modeled by different social agents.
Moreover, through the process of stimulus
generalization, the effect of a model’s prestige
carries over from one area of behavior to another,
and imitative responses tend to generalize to

unfamiliar persons to the extent that they share
similar characteristics with past reward-

producing models.
The fact that social behavior is extensively
under modeling stimulus control suggests that
social phenomena can be partly regulated through
alteration of focal modeling influences. Lippitt and
his colleagues (Lippitt, Polansky, & Rosen, 1952;
Polansky, Lippitt, & Redl, 1950) have shown in

several field studies that persons to whom is
attributed high social power are the major sources
of imitative behavior for other group members.
These findings indicate that the attitudes and
actions of entire groups can be modified most

rapidly and pervasively by changing the conduct
norms modeled by key sources of behavioral
contagion, whereas attempts to alter the behavior
of each member individually would prove
exceedingly laborious and ineffectual.
Utilization of Modeling Principles in

Planned Sociocultural Change
Societies are continuously faced with the

problem of introducing and gaining widespread
acceptance of new practices designed to improve
the quality of social life. This often involves
effecting changes in relatively circumscribed
groups, as in the case of specific community
development projects. At other times, however,

far-reaching modifications are sought in economic,
political, educational, and social practices that
implicate the entire culture.
Most socially significant changes involve some
negative consequences that are likely to serve

initially as barriers to change. In the first place,
people are required to expend a certain amount of
their time, energy, and resources, that might
otherwise be used for personal gratification, to

learn new personal habits and modes of living.
Second, the beneficial outcomes that may accrue
from new response patterns usually cannot be

clearly demonstrated until they have been tried
over a period of time. Since innovations often
prove unsuccessful and promoters generally
overvalue their potentialities, people are
understandably apprehensive about forsaking
existing behaviors of established utility for new
ones involving possible superior but uncertain

consequences. Most people, therefore, are
reluctant to change their customary practices until
they have observed new behaviors to be
rewarding for more venturesome adopters. Third,
conventional patterns are usually fortified by

belief systems and moral codes that portend
hazardous consequences for departures from
socially sanctioned practices. Thus, for example,
ineffective psychotherapeutic methods and folk
medical systems are much more difficult to

supplant when people are frightened by beliefs
that innovative procedures will adversely affect

them in ill-defined times and ways than when such
foreboding beliefs have not been used to reinforce
adherence to existing customs.
A fourth obstacle to the successful introduction
and diffusion of new modes of behavior is created
by individuals in positions of authority who have a
vested interest in preserving traditional prestige

and power structures. They are apt to oppose
actively any changes that may threaten their social
and economic status, particularly if new programs
are associated with outside agencies. Elite
countercontrol is generally maintained through

coercive pressures on less advantaged members
who have the most to gain from changes and are
therefore more receptive to new ways.
It is evident from the preceding discussion that

if programs designed to alter sociocultural
patterns are to meet with success, they must
employ powerful change procedures to overcome

the unfavorable conditions of reinforcement
initially associated with unaccustomed practices.
Attitude-change approaches have been employed
extensively on the assumption that a modification
in the belief system is a vital prerequisite to
acceptance of new behaviors. This strategy has
proved only partially successful. A different

approach, which concentrates on new alternatives
rather than on hindrances, arranges optimal
conditions for producing the desired behavioral
changes. New practices that are beneficial to the
user eventually become strongly established, and

incongruent attitudes either will be modified to
coincide with adopted behavior or they will be
construed in a manner that is consistent with pre-
existing beliefs.
Among the variety of methods available for

accelerating social changes, modeling plays a
highly influential role. If new response patterns

are to be learned, potential adopters must be
provided with models competent to display the
desired behavior and who are especially likely to
be emulated. Since vicarious reinforcement can
facilitate modeling, those exemplifying advocated
patterns should be appropriately rewarded to
demonstrate to others the benefits of new

practices. In addition to modeling influences, new
reinforcement contingencies must be introduced
into the social system to favor adoption and
continued performance of new behavior patterns
(Holmberg, 1960). The beneficial effects of new

skills and practices usually do not become
apparent until they have been applied over an
extended period. A change agent may, for example,
be faced with the problem of getting skeptical
people to institute and continue an irksome water

purification project over a long period before they
can obtain any clear evidence that it reduces

infectious diseases. As Erasmus (1961) has noted,
new cultural practices are most readily accepted
when they produce immediate inspectional

benefits and the causal relationship between new
behavior and advantageous outcomes can be
easily verified. The issue of spectacularity and
immediacy of results probably accounts for the

preference of aggressive over less obstreperous
means of forcing social changes.
In cases where the advantages to be gained
from new behavior patterns are considerably
delayed, it is necessary to provide subsidiary

immediate incentives to sustain them until the
long-term benefits occur and assume the
reinforcing function. These temporary substitute
rewards may involve financial compensations,
social recognition, positions in new leadership

hierarchies, and appropriate forms of status-
conferring symbolic rewards. It is quite possible
that many of the failures of cultural change
programs that are attributed to resistance arising
from conflicting beliefs in fact result from failure
to provide emulative models and adequate
reinforcement supports for unaccustomed
practices.
Another important factor that militates against

social change is that persons who adopt the new
patterns of behavior are often subjected to
negative sanctions from envious peers and
powerful officials whose vested interests may be
jeopardized. This creates especially difficult

problems when those in positions of power
undermine and block social reforms that do not
promote their own self-interests but are beneficial
for, and desired by, less advantaged persons.

Under these circumstances, little change will
result unless persons who adopt new patterns are
protected from maltreatment, and conditions are

arranged so that the new practices provide some
benefits for all concerned. This can be partly
achieved through the use of socially
interdependent contingencies in which a given
person’s rewarding outcomes are determined by
both the degree to which he performs the desired
behavior and a composite of individual

performances of the entire group. It will be shown
in Chapter 4 how the addition of group
reinforcement can favorably affect the
performance of its members. However, if an
organized minority continues to force compliance

with traditional practices, then aversive controls
must be applied. Desired objectives must be
enforced through social legislation, and defiance
must produce costly consequences. This
presupposes that change agencies exercise some

degree of control over the rewarding resources
available to communities and their leadership,

that they have the power to impose negative
sanctions, and that they have sufficient social
support to withstand the political repercussions of

enforced changes. In an effort to avoid offending
the existing leadership, social agencies generally
rely for the implementation of desired changes
upon the traditional elite, who, unfortunately,

often utilize such opportunities to further promote
their self-interests.
Under conditions where advocates of
innovations have no rewarding nor controlling
power, they must first establish their value by

demonstrating, in areas that engender little or no
resistance, that the practices they advocate yield
highly favorable outcomes. After they have thus
enhanced their credibility and modeling potency
they are in a more favorable position to attempt

modifications that conflict with existing traditions
and vested interests.
Summary
This chapter is principally concerned with
modeling processes whereby new modes of
behavior are acquired and existing response
patterns are extensively modified through
observation of other people’s behavior and its

consequences for them.
A multiprocess theory of observational
learning was advanced, according to which

modeled stimulus events are transformed and
retained in imaginal and verbal memory codes.
Later, reinstatement of these representational

mediators, in conjunction with appropriate
environmental cues, guide behavioral
reproduction of matching responses. Performance
of observationally learned responses is largely
regulated by reinforcing outcomes that may be

externally applied, self-administered, or
vicariously experienced. Since modeling

phenomena are controlled by several interrelated
subprocesses, the absence of modeling effects in

any given case may result from either failures in
sensory registration due to inadequate attention
to relevant social cues, deficient symbolic coding
of modeled events into functional mediators of

overt behavior, retention decrements, motor
deficiencies, or unfavorable conditions of
reinforcement.
Modeling procedures have been extensively
employed, with considerable success, for many

purposes, especially for developing conceptual
and interpersonal modes of behavior. In this
approach agents of change model requisite
behaviors and arrange optimal conditions for

recipients to learn and to practice the activities
until they are performed skillfully and
spontaneously. In addition to the utilization of

modeling principles for establishing social and
cognitive competencies, emotional responsiveness
can be conditioned and extinguished on a
vicarious basis. In the case of vicarious affective
conditioning, exposure to a model’s emotional
reactions arouses in observers emotional
responses which become conditioned, through

contiguous association, to distinctive cues present
in the situation. However, the degree of vicarious
responsiveness is partly dependent upon an
intermediary self-stimulation process involving
symbolic representation of similar consequences

occurring to oneself in the same situation.
Affective expressions of a model are most likely to
elicit high self-arousal in observers under
conditions where the participants have
experienced similar pleasurable or painful

experiences.
Vicarious extinction of emotional behavior is

achieved by exposing an observer to modeled
events in which a performer’s approach responses
toward feared objects do not produce adverse
effects or may engender positive consequences.
Studies of vicarious extinction reveal that this
procedure, particularly when combined with
guided participation, not only produces enduring

and generalized reductions in tenacious avoidance
behavior, but it also induces long-lasting
attitudinal changes and decrements in fearfulness
toward objects that were never specifically
included in the treatment program. A major factor

in modeling procedures that expedites behavioral
changes is assumed to involve vicarious extinction
of arousal reactions below the level for activating
avoidance responses, thus enabling persons to
perform approach behaviors. The fact that

elimination of the arousal potential of threatening
stimuli through a nonresponse extinction

procedure subsequently reduces avoidance
behavior provides further support for a dual-
process learning theory in which classically

conditioned effects partly govern instrumentally
learned responses.
Exposure to modeled events may also

strengthen or weaken observers’ inhibitions of
existing patterns of behavior. The occurrence of
these inhibitory and disinhibitory effects is mainly

determined by actual or inferred response
consequences to the model. Positive
reinforcement of models’ actions generally

facilitates similar behavior in observers if it is
appropriate to their social role and status,
whereas observation of punishing consequences
to models tends to inhibit similar responsiveness
in others. These vicarious reinforcement effects

may result from the information conveyed by the
model’s outcomes as to what constitutes
permissible or punishable actions in particular
situations, from motivational increases created by
witnessing others receive desired incentives, from
changes in model status produced by disparaging
or laudatory social reactions, and from vicarious
acquisition or extinction of emotional responses
through exposure to the affective expressions of

models undergoing rewarding or punishing
experiences.
The behavior of models often functions merely
as discriminative stimuli in facilitating the
expression by others of similar behaviors that

ordinarily are not subject to negative sanctions
and therefore do not involve inhibitory
mechanisms. Social models differ considerably in
the extent to which their behavior is likely to be

successful in producing favorable outcomes. As a
result of repeated differential reinforcement for
matching models who differ in intelligence, age,

socioeconomic status, social and vocational
competencies, prestige and power, model
attributes that signify probable consequences for
exemplified behavior determine in large part
which models will have greatest response-
facilitating effects. Because social behavior is
extensively under modeling stimulus control, the

attitudes and actions of groups can be modified by
altering the conduct norms modeled by major
sources of behavioral contagion.
REFERENCES
Adler, L. L., & Adler, H. E. Age as a factor in

observational learning in puppies. American
Dachshund, 1968, March, 13-14.
Allport, F. H. Social psychology. Cambridge, Mass.:

Riverside Press, 1924.
Alyokrinskii, V. V. Imitatsiya det’mi (4-7 let) russkikh i

nekotoroykh angluikikh glacnykh [Imitation by 4-
7 year-old children of Russian and several
English vowels], Voprosy Psikhologii, 1963, 9, 80-
88.
Angermeier, W. F., Schaul, L. T., & James, W. T. Social

conditioning in rats. Journal of Comparative and
Armstrong, E. A. Bird display and behaviour.
Cambridge: Cambridge University Press, 1942.
Aronfreed, J. Conduct and conscience. New York:

Baer, D. M., Peterson, R. F., & Sherman, J. A. The

development of imitation by reinforcing
behavioral similarity to a model. Journal of the
416.
Baer, D. M., & Sherman, J. A. Reinforcement control of

generalized imitation in young children. Journal
of Experimental Child Psychology, 1964, 1, 37-49.
Bandura, A. Relationship of family patterns to child

behavior disorders. Progress Report, 1960,
Stanford University, Project No. M-1734, United
States Public Health Service.

1962. pp. 211-269.

Academic Press, 1965, pp. 1-55. (a)
Bandura, A. Influence of models’ reinforcement
contingencies on the acquisition of imitative
responses. Journal of Personality and Social
Psychology, 1965, 1, 589-595. (b)

McNally, 1968.

13, 173-199.
Bandura, A., Grusec, J. E., & Menlove, F. L.

Observational learning as a function of
symbolization and incentive set. Child
Development, 1966, 37, 499-506.
Bandura, A., Grusec, J. E., & Menlove, F. L. Some social

determinants of self-monitoring reinforcement
Psychology, 1967, 5, 449-455. (a)
Bandura, A., Grusec, J. E., & Menlove, F. L. Vicarious

extinction of avoidance behavior. Journal of
(b)
Bandura, A., & Harris, M. B. Modification of syntactic
style. Journal of Experimental Child Psychology,
1966, 4, 341-352.
Bandura, A., & Huston, A. C. Identification as a process

of incidental learning. Journal of Abnormal and

Bandura, A., & McDonald, F. J. The influence of social

reinforcement and the behavior of models in
shaping children’s moral judgments. Journal of
281.
Bandura, A., & Menlove, F. L. Factors determining

vicarious extinction of avoidance behavior
through symbolic modeling. Journal of Personality
Bandura, A., & Mischel, W. The influence of models in

modifying delay of gratification patterns. Journal
705.

62.
Bandura, A., Ross, D., & Ross, S. A. Imitation of film-

mediated aggressive models. Journal of Abnormal
and Social Psychology, 1963, 66, 3-11. (a)
Bandura, A., Ross, D., & Ross, S. A. A comparative test

of the status envy, social power, and secondary
reinforcement theories of identificatory learning.
67, 527-534. (b)


Bandura, A., & Whalen, C. K. The influence of

antecedent reinforcement and divergent
modeling cues on patterns of self-reward. Journal
382.
Barnett, P. E., & Benedetti, D. T. Vicarious conditioning

of the GSR to sound. Paper read at the meeting of
the Rocky Mountain Psychological Association,
Glenwood Springs, 1960.
Barnwell, A. K. Potency of modeling cues in imitation
and vicarious reinforcement situations.
Dissertation Abstracts, 1966, 26, 7444.
Bayroff, A. G., & Lard, K. E. Experimental social

behavior of animals: III. Imitational learning of
white rats. Journal of Comparative Psychology,
1944, 37, 165-171.
Benton, A. A. Effects of the timing of negative

response consequences on the observational
learning of resistance to temptation in children.
Dissertation Abstracts, 1967, 27, 2153-2154.

466.
Berkowitz, L. Control of aggression. In B. M. Caldwell

& H. Ricciuti (Eds.), Review of child development
research. Vol. 3. New York: Russell Sage
Foundation, 1969.
Berkowitz, S. Acquisition and maintenance of

generalized imitative repertoires of profoundly
retarded children with retarded peers
functioning as models and reinforcing agents.
Unpublished doctoral dissertation, University of
Maryland, 1968.
Beyer, N. L., & May, J. G., Jr. The effects of race and
socioeconomic status on imitative behavior in
children using white male and female models.
Unpublished manuscript, Florida State
University, 1968.
Black, A. H. The extinction of avoidance responses

under curare. Journal of Comparative and
Black, A. H., Carlson, N. J., & Solomon, R. L. Exploratory

studies of die conditioning of autonomic
responses in curarized dogs. Psychological
Monographs, 1962, 76 (29, Whole No. 548).
Blake, R. R. The other person in the situation. In R.

Tagiuri & L. Petrullo (Eds.), Person perception and
interpersonal behavior. Stanford: Stanford
University Press, 1958. pp. 229-242.
Blake, R. R., Mouton, J. S., & Hain, J. D. Social forces in

petition signing. Southwestern Social Science
Quarterly, 1956, 36, 385-390.
Blake, R. R., Rosenbaum, M., & Duryea, R. Gift-giving as

a function of group standards. Human Relations,
1955, 8, 61-73.
Blanchard, E. B. The relative contributions of

modeling, informational influences, and physical
contact in the extinction of phobic behavior.
Unpublished doctoral dissertation, Stanford
University, 1969.
Bower, G. M. Mental imagery and associative learning.

In L. Gregg (Ed.), Cognition in learning and
memory. New York: Wiley, 1969.
Brigham, T. A., & Sherman, J. A. An experimental

analysis of verbal imitation in preschool children.
Journal of Applied Behavior Analysis, 1968, 1, 151-
158.
Brown, R., & Bellugi, U. Three processes in the child’s

acquisition of syntax. Harvard Educational
Review, 1964, 34, 133-151.
Bryan, J. H., & Test, M. A. Models and helping:

Naturalistic studies in aiding behavior. Journal of
407.
Buchwald, A. M. Extinction after acquisition under

different verbal reinforcement combinations.
48. (a)
Buchwald, A. M. Experimental alterations in the

effectiveness of verbal reinforcement
combinations. Journal of Experimental Psychology,
1959, 57, 351-361. (b)
Buchwald, A. M. Supplementary report: Alteration of

the reinforcement value of a positive reinforcer.
Journal of Experimental Psychology, 1960, 60,
416-418.
Campbell, D. T. Conformity in psychology’s theories of

acquired behavioral dispositions. In I. A. Berg &
B. M. Bass (Eds.), Conformity and deviation. New
York: Harper, 1961. pp. 101-142.

in Child Development, 1942, 7 (1, Serial No. 31).
Church, R. M. Transmission of learned behavior

between rats. Journal of Abnormal and Social
Psychology, 1957, 54, 163-165.
Church, R. M. Emotional reactions of rats to the pain

of others. Journal of Comparative and
Clark, B. S. The acquisition and extinction of peer

imitation in children. Psychonomic Science, 1965,
2, 147-148.
Colby, K. M. Computer-aided language development in

nonspeaking mental! disturbed children.
Technical Report No. CS85. Stanford University,
1967
Conant, M. B. Conditioned visual hallucinations.

Unpublished manuscript Stanford University,
1964.
Corsini, R. J., & Putzey, L. J. Bibliography of group

psychotherapy. Beacon, N. Y.: Beacon House,
1957.
Corsini, R. J., Shaw, M. E., & Blake, R. R. Roleplaying in

business and industry. New York: Free Press,
1961.
Corson, J. A. Observational learning of a lever pressing

response. Psychonomic Science, 1967, 7, 197-198.
Cowan, P. A., Hoddinoth, B. A., & Wright, B. A.

Compliance and resistance in the conditioning of
autistic children: An exploratory study. Child
Development, 1965, 36, 91.3-923.
Craig, K. D., & Weinstein, M. S. Conditioning vicarious

affective arousal. Psychological Reports, 1965, 17,
955-963.
Crawford, M. P., & Spence, K. W. Observational

learning of discrimination problems by
chimpanzees. Journal of Comparative Psychology,
1939, 27, 133-147.
Crooks, J. L. Observational learning of fear in
monkeys. Unpublished manuscript, University of
Pennsylvania, 1967.
Darby, C. L., & Riopelle, A. J. Observational learning in

the Rhesus monkey. Journal of Comparative and
deCharms, R., & Rosenbaum, M. E. Status variables

and matching behavior. Journal of Personality,
1960, 28, 492-502.
Drakeford, J. W. Integrity therapy. Nashville, Tenn.:

Broadman, 1967.
Dubanoski, R. A. Imitation as a function of role

appropriateness of behavior and response
consequences to the model. Unpublished
manuscript, University of Iowa, 1967.
Duncker, K. Experimental modification of children’s

food preferences through social suggestion.
33, 489-507.
Dysinger, W. S., & Ruckmick, C. A. The emotional

responses of children to the motion-picture
situation. New York: Macmillan, 1933.
Easterbrook, J. A. The effect of emotion on cue

utilization and the organization of behavior.
Ellson, D. G. Hallucinations produced by sensory

conditioning. Journal of Experimental Psychology,
1941, 28, 1-20.
Emmerich, W. Parental identification in young

children. Genetic Psychology Monographs, 1959,
60, 257-308.
Epstein, R. Aggression toward outgroups as a function

of authoritarianism and imitation of aggressive
models. Journal of Personality and Social
Psychology, 1966, 3, 574-579.
Erasmus, C. J. Man takes control. Minneapolis:

University of Minnesota Press, 1961.
Feshbach, N. D. Variations in teachers’ reinforcement

style and imitative behavior of children differing
in personality characteristics and social
background. Unpublished manuscript, University
of California, Los Angeles, 1967.
Flanders, J. P. A review of research on imitative

behavior. Psychological Bulletin, 1968, 69, 316-
337.
Flanders, J. P., & Thistlethwaite, D. L. Effects of

informative and justificatory variables upon
imitation. Unpublished manuscript, Vanderbilt
University, 1969.
Foss, B. M. Mimicry in mynas (Gracula religiosa): A

test of Mowrer’s theory. British Journal of
Psychology, 1964, 55, 85-88.
Freed, A., Chandler, P. J., Mouton, J. S., & Blake, R. R.

Stimulus and background factors in sign
violation, journal of Personality, 1955, 23, 499
(abstract) .
Gelfand, D. M. The influence of self-esteem on rate of

verbal conditioning and social matching
behavior. Journal of Abnormal and Social
Psychology, 1962, 65, 259-265.
Gerst, M. S. Symbolic coding operations in

observational learning. Unpublished doctoral
Gewirtz, J. L., & Stingle, K. C. The learning of

generalized imitation as the basis for
identification. Psychological Review, 1968, 75,
374-397.
Gittelman, M. Behavior rehearsal as a technique in

child treatment. Journal of Child Psychology and
Psychiatry, 1965, 6, 251-255.
Grosser, D., Polansky, N., & Lippitt, R. A laboratory

study of behavioral contagion. Human Relations,
1951, 4,115-142.
Grusec, J. E. Some antecedents of self-criticism.

4, 244-252.
Grusec, J. E., & Brinker, D. B., Jr. Reinforcement for

imitation as a social learning determinant with
implications for sex role development.
Unpublished manuscript, University of Toronto,
1969.
Grusec, J. E., & Mischel, W. The model’s characteristics

as determinants of social learning. Journal of
215.
Hall, K. R. Observational learning in monkeys and

apes. British Journal of Psychology, 1963, 54, 201-
226.
Hanlon, C. The effects of social isolation and

characteristics of the model on accent imitation
in fourth-grade children. Unpublished doctoral
Harris, M. B. Some determinants of sharing in

children. Unpublished doctoral dissertation,
Hayes, K. J., & Hayes, C. Imitation in a home-raised
chimpanzee. Journal of Comparative and
Helson, H., Blake, R. R., Mouton, J. S., & Olmstead, J. A.

Attitudes as adjustments to stimulus,
background, and residual factors. Journal of
322.
Henker, B. A. The effect of adult model relationships

on children’s play and task imitation. Dissertation
Abstracts, 1964, 24, 4797.
Herbert, J. J., & Harsh, C. M. Observational learning by

cats. Journal of Comparative Psychology, 1944, 37,
81-95.
Hicks, D. J. Imitation and retention of film-mediated

aggressive peer and adult models. Journal of
100.
Hill, J. H., Liebert, R. M., & Mott, D. E. W. Vicarious

extinction of avoidance behavior through films:
An initial test. Psychological Reports, 1968, 12,
192.
Hinde, R. A. The term “mimesis.” British Journal of

Animal Behavior, 1953, 1,7-11.
Holmberg, A. R. Changing community attitudes and
values in Peru: a case study. In R. N. Adams (Ed.),
Social change in Latin America today. New York:
Harper, 1960. pp. 63-107.
Holt, E. B. Animal drive and the learning process. Vol. 1.

New York: Holt, 1931.

Communication and persuasion. New Haven: Yale
Hovland, C. I., Lumsdaine, A. A., & Sheffield, F. D.

Experiments in mass communication. Princeton:
Princeton University Press, 1949.
Humphery, J. Personal communication, 1966.
Humphrey, G. Imitation and the conditioned reflex.

Pedagogical Seminary, 1921, 28, 1-21.
Hutt, S. J., Hutt, C., Lee, D., & Ounsted, C. A behavioural

and electroencephalographic study of autistic
children. Journal of Psychiatric Research, 1965, 3,
181-197.
Imanishi, K. Social behavior in Japanese monkeys,

Macaca fuscata. Psychologia, 1957, 1, 47-54.
Jakubczak, L. F., & Walters, R. H. Suggestibility as

dependency behavior. Journal of Abnormal and
Jersild, A. T., & Holmes, F. B. Methods of overcoming
children’s fears. Journal of Psychology, 1935, 1,75-
104.
Jones, M. C. The elimination of children’s fears. Journal

of Experimental Psychology, 1924, 7, 383-390.
Kagan, J., & Mussen, P. H. Dependency themes on the

TAT and group conformity. Journal of Consulting
Psychology, 1956, 20, 29-32.
Kanareff, V. T., & Lanzetta, J. T. Effects of task

definition and probability of reinforcement upon
the acquisition and extinction of imitative
responses. Journal of Experimental Psychology,
1960, 60, 340-348.
Kanfer, F. H. Vicarious human reinforcement: A

glimpse into the black box. In L. Krasner & L. P.
244-267.
Kausler, D. H., & Trapp, E. P. Motivation and cue

utilization in intentional and incidental learning.
Kawamura, S. The process of sub-culture propagation

among Japanese Macaques. In C. II. Southwick
(Ed.), Primate social behavior. Princeton: Van
Nostrand, 1963. pp. 82-90.
Kellogg, W. N., & Kellogg, L. A. The ape and the child: A

study of environmental influence upon early
behavior. New York: McGraw-Hill, 1933.
Kelly, G. A. The psychology of personal constructs. Vol.

11. Clinical diagnosis and psychotherapy. New
York: Norton, 1955.
Kimbrell, D. L., & Blake, R. R. Motivational factors in

the violation of a prohibition. Journal of Abnormal
Kohlberg, L. Moral development and identification. In

H. W. Stevenson (Ed.), Child Psychology: The sixty-
second yearbook of the National Society for the
Study of Education. Part I. Chicago: The National
Society for the Study of Education, 1963. pp. 277-
332.
Kriazhev, V. I. The objective investigation of the

higher nervous activity in a collective
experiment. Psychological Abstracts, 1934, 8,
2532.
Krumboltz, J. D., & Thoresen, C. E. The effect of

behavioral counseling in group and individual
settings on information-seeking behavior. Journal
of Counseling Psychology, 1964, 11, 324-333.
Krumboltz, J. D., Varenhorst, B. B., & Thoresen, C. E.
Nonverbal factors in the effectiveness of models
in counseling. Journal of Counseling Psychology,
1967, 14,412-418.
Kuhn, D. Z., Madsen, C. H., Jr., & Becker, W. C. Effects of

exposure to an aggressive model and
“frustration” on children’s aggressive behavior.
Child Development, 1967, 38, 739-745.
Lanzetta, J. T., & Kanareff, V. T. The effects of a

monetary reward on the acquisition of an
imitative response. Journal of Abnormal and
Social Psychology, 1959, 59,120-127.
Lazarus, A. A. Behaviour rehearsal vs. nondirective

therapy vs. advice in effecting behaviour change.
212.
Lazarus, R. S., Speisman, J. C., Mordkoff, A. M., &

Davison, L. A. A laboratory study of psychological
stress produced by a motion picture film.
Psychological Monographs, 1962, 76 (Whole No.
553).
Lazowick, L. On the nature of identification. Journal of

183.
Lefkowitz, M. M., Blake, R. R., & Mouton, J. S. Status
factors in pedestrian violation of traffic signals,
51, 704-706.
Lesser, G. S., & Abelson, R. P. Personality correlates of

persuasibility in children. In C. I. Hovland & I. L.
Janis (Eds.), Personality and persuasibility. New
Haven: Yale University Press, 1959. pp. 187-206.
Leuba, C. Images as conditioned sensations. Journal of

Experimental Psychology, 1940, 26, 345-351.
Lippitt, R., Polansky, N., & Rosen, S. The dynamics of

power. Human Relations, 1952, 5, 37-64.

Early childhood autism. Oxford: Pergamon, 1966.
pp. 115-144. (a)
Lovaas, O. I. Reinforcement therapy. (16 mm. sound

film). Philadelphia: Smith, Kline and French
Laboratories, 1966. (b)
Lovaas, O. I. A behavior therapy approach to the

treatment of childhood schizophrenia. In J. P. Hill
(Ed.), Minnesota symposia on child psychology,
Vol. 1. Minneapolis: University of Minnesota
Press, 1967. pp. 108-159.
Lovaas, O. I., Berberich, J. P., Kassorla, I. C., Klynn, G. A.,
& Meisel, J. Establishment of a texting and
labelling vocabulary in schizophrenic children.
Unpublished manuscript, University of California,
Los Angeles, 1966.
Lovaas, O. I., Berberich, J. P., Perloff, B. F., & Schaeffer,

B. Acquisition of imitative speech by
schizophrenic children. Science, 1966, 151, 705-
707.
Lovaas, O. I., Dumont, D. A., Klynn, G. A., & Meisel, j.

Establishment of appropriate response to, and
use of, certain prepositions and pronouns in
schizophrenic children. Unpublished manuscript,
University of California, Los Angeles, 1966.
Lovaas, O. I., Freitag, L., Nelson, K., & Whalen, C. The

establishment of imitation and its use for the
development of complex behavior in
schizophrenic children. Behaviour Research and
Therapy, 1967, 5, 171-181.
Luchins, A. S., & Luchins, E. H. Imitation by rote and by

understanding. Journal of Social Psychology, 1961,
54, 175-197.
Luchins, A. S., & Luchins, E. H. Learning a complex

ritualized social role The Psychological Record,
1966, 16, 177-187.
Maccoby, E. E. Role-taking in childhood and its
consequences for social learning. Child
Development, 1959, 30, 239-252.
Maccoby, E. E., & Wilson, W. C. Identification and

observational learning from films. Journal of
Abnormal and Social Psychology, 1957, 55, 76—
87
Maccoby, N., Michael, D. N., & Levine, S. Further

studies of student participation procedures in
film instruction: Review and preview, covert
practice, and motivational interactions. In A. A.
Lumsdaine (Ed.), Student response in
programmed instruction. Washington, D. C.:
National Academy of Sciences—-National
Research Council, 1961. pp. 295-325.
McDonald, F., & Allen, D. W. Training effects of

feedback and modeling procedures on teaching
performance. Unpublished manuscript, Stanford
University, 1967.
McDougall, W. An introduction to social psychology.

London: Methuen, 1908.
McGuire, W. J. Interpolated motivational statements

within a programmed series of instructions as a
distribution-of-practice factor. In A. A. Lumsdaine
(Ed.), Student response in programmed
instruction. Washington, D. C.: National Academy
of Sciences—National Research Council, 1961.
pp. 411-415.
Madsen, C., Jr. Nurturance and modeling in

preschoolers. Child Development, 1968, 39, 221-
236.
Mandler, G. Association and organization: Facts,

fancies, and theories. In T. R. Dixon & D. L. Horton
theory. Englewood Cliffs, N. J.: Prentice-Hall,
1968. pp. 109- 119.
Margolius, G. J., & Sheffield, F. D. Optimum methods of

combining practice with filmed demonstration in
teaching complex response sequences: Serial
learning of a mechanical-assembly task. In A. A.
Lumsdaine (Ed.), Student response in
National Academy of Sciences—National
Research Council, 1961. pp. 33—53.
Marlatt, G. A. Vicarious and direct reinforcement

control of verbal behavior in an interview setting.
Unpublished doctoral dissertation, Indiana
University, 1968. (a)
Marlatt, G. A. Exposure to a model and task ambiguity

as determinants of verbal behavior in an
interview. Paper read at the Western
Psychological Association meeting, San Diego,
April 1968. (b)
Marlatt, G. A., Jacobsen, E. A., Johnson, D. L., & Morrice,

D. J. Effect of exposure to a model receiving
varied informational feedback upon consequent
behavior in an interview. Paper presented at the
Midwestern Psychological Association meeting,
Chicago, 1966.
Marshall, H. R., & Hahn, S. C. Experimental

modification of dramatic play. Journal of
122.
Masserman, J. H. Behavior and neurosis. Chicago:

University of Chicago Press, 1943.
Masters, J. C., & Branch, M. N. A comparison of the

relative effectiveness of instructions, modeling,
and reinforcement procedures for inducing
behavior change. Journal of Experimental
Psychology, 1969 (in press).
Mausner, R. Studies in social interaction: III. Effect of

variation in one partner’s prestige on the
interaction of observer pairs. Journal of Applied
Psychology, 1953, 37, 391-393.
Mausner, B. The effect of prior reinforcement on the
interaction of observer pairs. Journal of Abnormal
and Social Psychology, 1954, 49, 65-68. (a)
Mausner, B. The effect of one partner’s success in a

relevant task on the interaction of observer pairs.
49,557-560. (b)
Mausner, B., & Bloch, B. L. A study of the additivity of

variables affecting social interaction. Journal of
256.
Metz, J. R. Conditioning generalized imitation in

autistic children, journal of Experimental Child
Psychology, 1965, 2, 389-399.
Michael, D. N., & Maccoby, N. Factors influencing the

effects of student participation on verbal learning
from films: Motivating versus practice effects,
“feedback,” and overt versus covert responding.
In A. A. Lumsdaine (Ed.), Student response in
programmed instruction, Washington, D. C.:
Miller, E. C., Dvorak, A., & Turner, D. W. A method of

creating aversion to alcohol by reflex
conditioning in a group setting. Quarterly Journal
of Studies on Alcohol, 1960, 21, 424-431.
Miller, N. E. Some implications of modern behavior

theory for personality change and psychotherapy.
In P. Worchel & D. Byrne (Eds.), Personality
change. New York: Wiley, 1964. pp. 149-175.
Miller, N. E., & Dollard, J. Social learning and imitation.

New Haven: Yale University Press, 1941.
Miller, R. E. Experimental approaches to the

physiological and behavioral concomitants of
affective communication in rhesus monkeys. In S.
A. Altmann (Ed.), Social communication among
primates. Chicago: University of Chicago Press,
1967. pp. 125-134.
Miller, R. E., Banks, J. H., & Ogawa, N. Communication

of affect in “cooperative conditioning” of rhesus
monkeys. Journal of Abnormal and Social
Psychology, 1962, 64, 343-348.
Miller, R. E., Banks, J. H., Jr., & Ogawa, N. Role of facial

expression in “cooperative avoidance
conditioning” in monkeys. Journal of Abnormal
Miller, R. E., Caul, W. F., & Mirsky, I. A. Communication

of affect between feral and socially isolated
monkeys. Journal of Personality and Social
Psychology, 1967, 7, 2.31-239.
Miller, R. E., Murphy, J. V., & Mirsky, I. A. Nonverbal

communication of affect. Journal of Clinical
Psychology, 1959, 15, 155-158.
Mischel, W., & Grusec, J. Determinants of the rehearsal

and transmission of neutral and aversive
behaviors. Journal of Personality and Social
Psychology, 1966, 3, 197-205.
Moreno, J. L. Fundamental rules and techniques of

psychodrama. In J. H. Masserman & J. L. Moreno
(Eds.), Progress in psychotherapy. Vol. 3.
Techniques of psychotherapy. New York: Grune &
Stratton, 1958. pp. 86-131.
Morgan, C. L. Habit and instinct. London: E. Arnold,

1896.
Morrisett, L. N., Jr. The role of implicit practice in

learning. Unpublished doctoral dissertation, Yale
University, 1956.
Mowrer, O. H. Identification: A link between learning

theory and psychotherapy. In Learning theory
and personality dynamics. New York: Ronald
Press, 1950. pp. 573-615.
Mowrer, O. H. Learning theory and the symbolic

processes. New York: Wiley, 1960.
Mowrer, O. H. The new group therapy. Princeton, N. J.:
Van Nostrand, 1964.
Mowrer, O. H. The behavior therapies, with special

reference to modeling and imitation. American
Journal of Psychotherapy, 1966, 20, 439—461.
Mussen, P. H., & Parker, A. L. Mother-nurturance and

girls’ incidental imitative learning. Journal of
O’Connor, R. D. Modification of social withdrawal

through symbolic modeling. Journal of Applied
Behavior Analysis, 1969, 2, 15-22.
Odom, R. D., Liebert, R. M., & Hill, J. H. The effects of

modeling cues, reward, and attentional set on the
production of grammatical and ungrammatical
syntactic constructions. Journal of Experimental
Ofstad, N. S. The transmission of self-reinforcement

patterns through imitation of sex-role
appropriate behavior. Unpublished doctoral
dissertation, University of Utah, 1967.
Paivio, A. Mental imagery in associative learning and

memory. Psychological Review, 1969, 76, 241-
263.
Parke, R. D., & Walters, R. H. Some factors influencing
the efficacy of punishment training for inducing
response inhibition. Monographs of the Society for
Research in Child Development, 1967, 32 (1, Serial
No. 109).
Parloff, M. B., Iflund, B., & Goldstein, N.

Communication of “therapy values” between
therapist and schizophrenic patients. Journal of
Nervous and Mental Diseases, 1960, 130, 193-199.
Parsons, T. The social system. Glencoe, 111.: Free

Press, 1951.
Parsons, T. Family structure and the socialization of

the child. In T. Parsons & R. F. Bales, Family,
socialization, and interaction process. Glencoe,
111.: Free Press, 1955. pp. 35-131.
Patterson, G. R., Littman, I., & Brown, T. R. Negative set

and social learning. Journal of Personality and
Pentony, P. Value change in psychotherapy. Human

Relations, 1966, 19, 39-46.
Perry, H. M. The relative efficiency of actual and

“imaginary” practice in five selected tasks.
Archives of Psychology, 1939, 34, No. 243.
Piaget, J. The moral judgment of the child. Glencoe,

111.: Free Press, 1948.
Piaget, J. Play, dreams, and imitation in childhood. New
York: Norton, 1952.
Polansky, N., Lippitt, R., & Redl, F. An investigation of

behavioral contagion in groups. Human Relations,
1950, 3, 318-348.
Porro, C. R. Effects of the observation of a model’s

affective responses to her own transgression on
resistance to temptation in children. Dissertation
Abstracts, 1968, 28, 3064.
Reed, J. L. Social modeling and object interaction as a

means of inducing change in children’s scientific
concepts. Unpublished manuscript, Harvard
University, 1966.
Rimland, B. Infantile autism. New York: Appleton-

Rimm, D. C., & Mahoney, M. J. The application of

reinforcement and modeling- guidance
procedures in the treatment of snake phobic
behavior. Unpublished manuscript, Arizona State
University, 1969.
Risley, T., & Wolf, M. Establishing functional speech in

echolalic children. Behaviour Research and
Therapy, 1967, 5, 73-88.
Ritter, B. The group treatment of children’s snake
phobias using vicarious and contact
desensitization procedures. Behaviour Research
and Therapy, 6, 1-6.
Ritter, B. Eliminating excessive fears of the

environment through contact desensitization. In
J. B. Krumboltz & C. E. Thoresen (Eds.),
Behavioral counseling: Cases and techniques. New
York: Holt, Rinehart & Winston, 1969. (a)
Ritter, B. Treatment of acrophobia with contact

desensitization. Behaviour Research and Therapy,
1969, 7, 41-45.
Rosekrans, M. A., & Hartup, W. W. Imitative influences

of consistent and inconsistent response
consequences to a model on aggressive behavior
in children. Journal of Personality and Social
Psychology, 1967, 7, 429-434.
Rosenbaum, M. E. The effect of stimulus and

background factors on the volunteering response.
53,118-121.
Rosenbaum, M. E., & Blake, R. R. Volunteering as a

function of field structure. Journal of Abnormal
Rosenbaum, M. E., & Tucker, I. F. The competence of
the model and the learning of imitation and
nonimitation. Journal of Experimental Psychology,
1962, 63, 183-190.
Rosenblith, J. F. Learning by imitation in kindergarten

children. Child Development, 1959, 30, 69-80.
Rosenblith, J. F. Imitative color choices in

kindergarten children. Child Development, 1961,
32, 211-223.
Rosenhan, D., & White, G. M. Observation and

rehearsal as determinants of prosocial behavior.
5, 424-131.
Rosenthal, D. Changes in some moral values following

1955, 19, 431-436.
Ross, D. Relationship between dependency,

intentional learning, and incidental learning in
preschool children. Journal of Personality and
Ross, S. A. The effect of deviant and nondeviant

models on the behavior of preschool children in a
temptation situation. Unpublished doctoral
Schachter, S., & Hall, R. Group-derived restraints and
audience persuasion. Human Relations, 1952, 5,
397-406.
Schein, E. H. The effect of reward on adult imitative

behavior. Journal of Abnormal and Social
Psychology, 1954, 49, 389-395.
Schwartz, A. N., & Hawkins, H. L. Patient models and

affect statements in group therapy. Proceedings of
the 73rd annual convention of the A.P.A.
Washington, D.C.: American Psychological
Association, 1965. pp. 265-266.
Sheffield, F. D. Theoretical considerations in the

learning of complex sequential tasks from
demonstration and practice. In A. A. Lumsdaine
(Ed.), Student response in programmed
instruction. Washington, D. C.: National Academy
of Sciences—National Research Council, 1961.
pp. 13-32.
Sheffield, F. D., & Maccoby, N. Summary and

interpretation of research on organizational
principles in constructing filmed demonstrations.
In A. A. Lumsdaine (Ed.), Student response in
Sherman, J. A. Use of reinforcement and imitation to
reinstate verbal behavior in mute psychotics.
Journal of Abnormal Psychology, 1965, 70, 155-
164.

Macmillan, 1953.
Sloane, H. N., Jr., Johnston, M. K., & Harris, F. R.

Remedial procedures for teaching verbal
behavior to speech deficient or defective young
children. In H. N. Sloane, Jr. & B. A. MacAulay
(Eds.), Operant procedures in remedial speech and
language training. Boston: Houghton Mifflin,
1968. pp. 77-101.
Slobin, D. I. Imitation and grammatical development

in children. In N. S. Endler, L. R. Boulter, & H.
Osser (Eds.), Contemporary issues in
developmental psychology. New York: Holt,
Rinehart & Winston, 1968. pp. 437^43.
Solomon, R. L., & Coles, M. R. A case of failure of

generalization of imitation across drives and
across situations. Journal of Abnormal and Social
Psychology, 1954, 49, 7-13.
Solomon, R. L., & Turner, L. H. Discriminative classical

69, 202-219.
Speisman, J. C., Lazarus, R. S., Mordkoff, A., & Davison,

L. Experimental reduction of stress based on ego-
defense theory. Journal of Abnormal and Social
Psychology, 1964, 68, 367-380.
Spiegler, M. D., Liebert, R. M., McMains, M. J., &

Fernandez, L. E. Experimental development of a
modeling treatment to extinguish persistent
avoidance behavior. Unpublished manuscript,
Vanderbilt University, 1968.
Stoke, S. M. An inquiry into the concept of

identification. Journal of Genetic Psychology, 1950,
76, 163-189.
Stotland, E. Exploratory investigations of empathy. In

L. Berkowitz (Ed.), Advances in experimental
social psychology, Vol. 4. New York: Academic
Press, 1969.
Stotland, E., Shaver, K., & Crawford, R. Empathizing

with pain and with pleasure, perceived similarity,
and birth order. Unpublished manuscript,
University of Washington, 1966.
Sturm, I. E. The behavioristic aspect of psychodrama.

Group Psychotherapy, 1965, 18, 50-64.
Tarde, G. The laws of imitation. New York: Holt, 1903.
Taub, E., Bacon, R. C., & Berman, A. J. Acquisition of a
trace-conditioned avoidance response after
deafferentation of the responding limb. Journal of
Comparative and Physiological Psychology, 1965,
59, 275-279.
Taub, E., Teodoru, D., Ellman, S. J., Bloom, R. F., &

Berman, A. J. Deafferentation in monkeys:
Extinction of avoidance responses, discrimination
and discrimination reversal. Psychonomic Science,
1966, 4, 323-324.
Thorndike, E. L. Animal intelligence: An experimental

study of the associative processes in animals.
Psychological Review Monograph Supplement,
1898, 2 (4, Whole No. 8).
Thorpe, W. H. Learning and instinct in animals.

London: Methuen, 1956.
Tinbergen, N. The study of instinct. Oxford: Clarendon,

1951.
Truax, C. B., & Carkhuff, R. R. Toward effective

counseling and psychotherapy. Chicago: Aldine,
1967.
Tulving, E. Theoretical issues in free recall. In T. R.

Dixon, & D. L. Horton (Eds.), Verbal behavior and
general behavior theory. Englewood Cliffs, N. J.:
Prentice-Hall, 1968. pp. 2-36.
Twining, W. E. Mental practice and physical practice

in learning a motor skill. Research Quarterly,
1949, 20, 432-435.
Vandell, R. A., Davis, R. A., & Clugston, H. A. The

function of mental practice in the acquisition of
motor skills. Journal of General Psychology, 1943,
29, 243-250.
Walters, R. H., Leat, M. & Mezei, L. Inhibition and

disinhibition of responses through empathetic
learning. Canadian Journal of Psychology, 1963,
17, 235-243.
Walters, R. H., & Llewellyn Thomas, E. Enhancement

of punitiveness by visual and audiovisual
displays. Canadian Journal of Psychology, 1963,
17, 244-255.
Walters, R. H., Parke, R. D., & Cane, V. A. Timing of

punishment and the observation of consequences
to others as determinants of response inhibition.
Journal of Experimental Child Psychology, 1965, 2,
10-30.
Warden, C. J., Fjeld, H. A., & Koch, A. M. Imitative

behavior in Cebus and Rhesus monkeys.
Pedagogical Seminary and Journal of Genetic
Psychology, 1940, 56, 311-322.
Warden, C. J., & Jackson, T. A. Imitative behavior in the

Rhesus monkey. Pedagogical Seminary and
Journal of Genetic Psychology, 1935, 46, 103-125.
Watson, J. B. Imitation in monkeys. Psychological

Bulletin, 1908, 5, 169-178.
Wheeler, L. Toward a theory of behavioral contagion.

Wilson, F. S., & Walters, R. H. Modification of speech

output of near-mute schizophrenics through
social-learning procedures. Behaviour Research
and Therapy, 1966, 4, 59-67.
Winitz, H., & Preisler, L. Discrimination pretraining

and sound learning. Perceptual and Motor Skills,
1965, 20, 905-916.

Wolpe, J., & Lazarus, A. A. Behavior therapy techniques.

Oxford: Pergamon, 1966.
4
Positive Control
In the modification of psychological conditions
that reflect primarily behavioral deficits, the
development of complex repertoires of behavior
and the strengthening of existing responses
constitute important objectives. Moreover, after
behavior has been well established, appropriate
conditions must be created to maintain it at a

satisfactory level. Reinforcement procedures are
best suited for these purposes.
It has been amply demonstrated that behavior

is controlled to a large extent by its consequences.
Any attempts, therefore, to produce enduring
changes in responsiveness must alter the
incidence, and often the nature, of reinforcing
outcomes that are customarily produced by given
modes of response. There are two broad classes of
consequences—rewarding and punishing events

—that serve as important determinants of
behavior. The present chapter is mainly concerned
with the establishment of response patterns and

their maintenance through systematic application
of positive reinforcement.
Theoretical Interpretations of
Reinforcement Processes
When a given response is followed by a

positively reinforcing consequence, it increases
the likelihood that the response will be repeated
on subsequent occasions. Although there is little
dispute about the validity of the empirical

principle of reinforcement, numerous alternative
explanations have been proposed for the manner

in which reinforcement produces its effects
(Hilgard & Bower, 1966; Kimble, 1961). The
various theoretical approaches differ in the extent
to which they consider drive states, stimulus
events, or response properties to be the critical

factors governing reinforcement processes.
DRIVE-REDUCTION HYPOTHESIS
One influential theory of reinforcement (Hull,

1943) assumes that the effects of reinforcing
consequences are produced by need-reduction. In

interpreting the process of primary reinforcement,
it is assumed that deprivation or painful
stimulation produces a physiological need giving

rise to a drive that activates behavior. A
reinforcing event is one that reduces a drive by
satisfying or removing the need. A more
comprehensive form of this drive-reduction view
was advanced by Miller & Dollard (1941), who
emphasize the activating properties of strong
stimuli rather than needs. According to their
stimulus-reduction theory, any stimulation,
regardless of whether it is based upon a need, can
become a drive if it is made sufficiently intense;

reduction of aversive stimulation has reinforcing
effects.
There exists an extensive body of evidence that

drive-inducing operations greatly enhance the
potency of reinforcing stimuli, and that
attenuation or termination of aversive stimulation

can have strong reinforcing effects on behavior.
The homeostatic conception of reinforcement
receives some further support from studies
demonstrating that procedures designed to alter
physiological states directly, while excluding

secondary reinforcement deriving from sensory
stimulation and consummatory responses, can

function as effective reinforcers for overt
behavior. Food-deprived animals, for example,

learn to make responses that result in nutrients
being delivered directly into their stomachs
(Miller & Kessen, 1952) or glucose being injected
intravenously (Chambers, 1956; Coppock &

Chambers, 1954) contingent upon correct
performances. Similarly, intravenous insulin
injections, which produce rapid decreases in blood
sugar level, have a punishing effect upon behavior
(Coppock, Headlee, & Hood, 1953).
The assumption that reinforcement requires

drive-reduction was initially questioned by
experiments demonstrating that nonnutritive
saccharin reinforced behavior (Sheffield & Roby,

1950). Similarly, copulation without ejaculation,
which produces no reduction in tension (Sheffield,
Wulff, & Backer, 1951; Whalen, 1961), could serve

as an effective reward. However, the conclusions
drawn from these findings regarding the validity
of the drive-reduction theory of reinforcement
have been challenged by Miller (1963), on the
basis of evidence that prefeeding hungry animals
with saccharin does in fact reduce their
subsequent consumption of food, and on the

assumption that sex may not involve a unitary
drive that can be reduced only by ejaculation. In
replying to criticisms of drive-reduction theory on
the grounds that people often engage in behavior
that produces heightened stimulation, Brown

(1955) pointed out that drive cannot be defined
solely in terms of intensity of stimulation. The
reason for this is that strong stimuli can lose their
activating function if presented in gradually
increasing values, if they have been associated

with rewarding experiences, or if they become
discriminative for less active responding.

Therefore, Brown cogently argues that definition
of a drive stimulus must include, in addition to its
intensity, other criteria such as the amount and

type of previous experience with the stimulus, and
the manner and context in which it is presented.
It should be noted in passing that, although

results from experiments involving fistula feeding
and nutritive injections demonstrate that
reduction of a physiological need can be sufficient

to reinforce an instrumental response, such
studies do not necessarily establish the
physiological bases of reinforcement. Any such

physiological explanation may ultimately be
carried to the point where reinforcing effects are
interpreted in terms of intracellular changes.
While relationships established at the molecular
level have considerable theoretical significance

regarding the basic mechanisms of reinforcement,
knowledge of this type is of limited usefulness in
devising incentive programs, since it is extremely
unlikely that in social practice one would alter
neurophysiological events directly in order to
influence responsiveness.
SENSORY-STIMULATION HYPOTHESIS
Although some reinforcement effects may be

governed by visceral drive states, there are many
reinforcing conditions that do not appear to
involve reduction of physiological needs or
removal of aversive stimuli, unless one were to

invoke a host of sensory and activity drives.
Animals will learn to perform responses that

produce visual and auditory stimulation or
opportunities to engage in manipulative and
exploratory activities (Barnes & Baron, 1961;

Butler, 1958a; Kish, 1966; Miles, 1958). A number
of studies, conducted principally with infants
(Rheingold, Stanley, & Doyle, 1964) and children
(Odom, 1964; Stevenson & Odom, 1964), have
likewise shown that visual and auditory feedback

can be effective in modifying and sustaining
behavior over time. These findings would seem to

indicate that much human behavior—particularly
approach, attending, and manipulative responses
—is reinforced by the sensory feedback that is
automatically produced.
Investigations of factors that might contribute
to the reinforcing properties of auditory and visual
events have revealed that novel and complex
stimuli function as more effective reinforcers than
simple and familiar stimulus events. The data

furthermore indicate that, as is the case with
biologically related incentives, the potency of
sensory reinforcers is increased by deprivation of
sensory experiences and diminished by stimulus-

satiation operations.
The existence of sensory reinforcement has
been convincingly demonstrated, but the nature of
the mechanisms underlying the phenomenon is by
no means clear. Berlyne (1960) and Harlow
(1953) have posited curiosity, manipulative, and
exploratory drives that are presumed to be
elicited by external, novel stimuli and reduced by
continuous exposure to such stimulation. In the

prototypic experiment, animals placed in a
lightproof and sound-attenuated box learn to
perform discriminative responses that open a
door allowing them either to view the outside

environment for a few minutes, or to press levers
for auditory stimulation. The major difficulty in
accounting for the animals’ behavior in terms of an
externally aroused curiosity drive is that the
animals are not exposed to the novel stimuli until

after the responses, of which the stimuli are the
presumed cause, have been successfully executed.

As Brown (1953) points out, “If visual exploration
provided the only significant motive, then the

monkeys must have been unmotivated until the
window was opened following a correct response.

But the monkeys did appear to be motivated. One
might conclude, therefore, that the effective
motivation was aroused before the monkeys were
allowed to see, not as a consequence of seeing [p.

54].” The more distally occurring visual and
auditory experiences can serve as reinforcing

events, but it is evident that the instrumental
responses must be activated by antecedent
stimuli.
Both Miller (Myers & Miller, 1954) and

Mowrer (1960) have reinterpreted sensory
reinforcement effects in terms of reduction of an
aversive drive. They assume that monotony
produces boredom, which has aversive properties,

and that close confinement and drastic reduction
of sensory contact with one’s environment can
generate considerable apprehension. Miller and
Mowrer therefore contend that, if changes in
sensory stimulation are boredom-relieving or
anxiety-reducing, then behavior is reinforced by
its sensory feedback in a manner consistent with
drive-reduction theory. One would encounter no
problems in testing the efficacy of sensory

reinforcers under conditions that do not arouse
anxiety, but it is exceedingly difficult to maintain a
low or unvaried level of sensory input, which in
large part determines the incentive value of
specific auditory and visual stimuli, without

producing concomitant boredom. The problem of
determining whether sensory reinforcement
effects are attributable to elimination of boredom
or to inherent rewarding properties of novel
stimuli is further complicated by the fact that most

exploratory activities undergo rapid satiation.
Some investigators (Fox, 1962; Isaac, 1962;

Leuba, 1955) have offered a neurophysiological
explanation of sensory reinforcement that is
similar in some respects to the operation of
homeostatic drives. Based on evidence that
prolonged deprivation of sensory input results in
psychological dysfunctioning, it is assumed that
for normal physiological functioning the organism

must maintain an optimal level of sensory
stimulation. Therefore, subjects will perform
instrumental responses to increase sensory input
if there is a deficit, and conversely, they will work
to reduce sensory stimulation if it exceeds the

optimal level.
It is difficult to explain, on the basis of a drive
for optimal quanta of sensory stimulation, why
animals confined in a drab, light-proof, sound-

attenuated box will work unflaggingly for certain
sights and sounds but will refrain from performing
responses that are instrumental in producing

negatively valenced stimuli in the same sensory
modalities. Thus, monkeys perform responses for
the opportunity to view other monkeys, movies,
electric trains, and to hear sounds of a monkey
colony in the outside environment, whereas the
sight or sound of a dog, and distress vocalizations,
readily suppress their exploratory tendencies

(Butler, 1954; 1958b). The findings clearly
indicate that the content rather than the amount of
sensory input may be the critical factor
determining the incidence of exploratory
responses. Studies in which both the amount of

sensory stimulation and the conditioned valence
of auditory or visual stimuli are systematically
manipulated would provide the basis for
determining whether sensory reinforcement
effects are best interpreted in terms of principles

of optimal stimulation, secondary reinforcement,
or their interactive effect.
The existence of sensory, manipulative, and
exploratory drives is usually inferred from
response patterns rather than defined in terms of
antecedent conditions for producing the specific
drive. Unless drives and responses are
operationally differentiated, there are no limits to
the proliferation of drive states which can be more

economically explained in terms of response
dispositions. If independent criteria are not
employed, new drives or motives may be invoked
for each reinforcing event or prevalent behavior.
PREPOTENT-RESPONSE HYPOTHESIS
In describing the essential properties of
reinforcing events, emphasis has usually been
placed on the nature of the reinforcing stimuli

(e.g., food, money, novel sensations, social
attention and approval, intracranial stimulation,
etc.), and their efficacy under varying conditions of
deprivation. On the basis of results from an
ingenious series of experiments, Premack (1965)
has presented an explanation of reinforcement
that emphasizes the reinforcing response rather
than the reinforcing stimulus. In these
investigations the reinforcement values of
different activities are estimated from the

duration for which subjects spontaneously engage
in particular behaviors when no time or response
restrictions exist. If the opportunity to engage in
the more rewarding activity is made conditional

upon the prior performance of low probability
behaviors, then the latter responses increase in
frequency. Based on these findings, Premack has
proposed the following principle of reinforcement:
For any pair of activities, the more probable one

will reinforce the less probable one.
Data observed by Premack indicate that, under

appropriate conditions, almost any activity can
function as an effective reinforcer. Thus, animals

that prefer running to eating will perform
consummatory responses in order to release an

activity wheel that permits them to sprint,
whereas subjects that would rather eat than run
will engage in running behavior in order to gain
access to food. Moreover, by manipulating

deprivational conditions, the reinforcement
relation between activities can be easily reversed,
with the result that reinforcing events are
converted into reinforceable events. To continue
with the previous example, eating will reinforce

running behavior in food-deprived animals, but
after they have been fed and their mobility
restricted, eating can serve as the instrumental
activity that is reinforced by opportunities to run.
Reversibility is apparently a general phenomenon

extending even to intracranial self-stimulation,
which can serve as an extremely powerful

reinforcer of instrumental responses. When the
probability of drinking, for example, is greater
than brain stimulation, drinking reinforces

intracranial self-stimulation (ICS), and conversely,
in animals for which ICS is more rewarding than
drinking, making ICS conditional upon drinking
produces an increase in drinking behavior

(Holstein & Hundt, 1965). If further research
should demonstrate that electrical stimulation of

the limbic system, which is assumed to be
inherently reinforcing, is itself reinforceable by
contingent response events, then one might
question the existence of an integrated brain

center that governs all reinforcement. As Premack
points out, one would have to locate another
region of central reinforcement in cases where

intracranial stimulation is increased by its
instrumental value in producing more highly

preferred response events. It is also unclear how
explanations in terms of central reinforcement
focuses can readily account for the reversal of

instrumental and rewarding functions of any given
activity. Indeed, the experimental evidence
convincingly demonstrates that reinforcement is a
relational rather than an absolute property of the
activity. A particular response event will have no

reinforcing potency in relation to a more highly
preferred activity, but it will function as an

effective positive reinforcer when paired with
responses of lesser value.
When reinforcing events are defined in terms

of their effects—as stimuli that increase the
probability of preceding responses—the empirical
principle of reinforcement is open to the criticism
of circularity. In rebuttal, Meehl (1950) contended
that this criticism ignores the fact that reinforcers

are transsituational, that is, a stimulus that has
been found to be effective in reinforcing one
response can be used predictively to reinforce
other types of responses. The findings discussed
above, however, indicate that the assumption of
transsituationality is valid only under certain
limiting conditions because most stimuli do not
possess generalized reinforcing potency. Premack
solves the circularity problem by defining the

reinforcing power of different consequences
independently of response changes.
It appears highly doubtful that the reinforcing
effects of prepotent responses are interpretable as
drive-reduction processes or sensory drive

mechanisms. There has been virtually no
speculation or research concerning the specific
aspects of prepotency that contribute to its
reinforcing potential, and, therefore, the

associated processes remain obscure. Although
the prepotent-response principle can encompass a
wide range of conditions which function as

effective reinforcers, it cannot account for the
efficacy of consequences that do not involve
performance of responses. Thus in cases where
behavior is strengthened either by nutrients
delivered directly into the stomach or blood
stream, or by more conventional consequences
such as praise, positive attention, monetary

rewards, or various types of sensory feedback, it
appears difficult to apply the prepotent- response
principle. Even when pairs of responses are
arranged in a contingent relationship, correct
specification of the reinforcing event is

complicated by the fact that changes in sensory
stimulation accompanying the behavior rather
than the activity per se may be primarily
responsible for the reinforcement effects. The
relative contribution of sensory consequences of

behavior to the total reinforcing effect can be
assessed by vicarious paradigms in which the

responses of a yoked observer are maintained by
witnessing the changes in visual and auditory
stimulation produced by a performer’s actions.
It is evident from this brief review that
exceedingly diverse events, which have no
apparent common properties, can all serve a
reinforcing function. A theory of reinforcement

that adequately integrates these heterogeneous
consequences into a unified system has yet to be
formulated. Considering that the reinforcing
potency of a given event is relationally
determined, a comprehensive theory of

reinforcement cannot be based on properties
inherent in the event itself. The experimental
findings nevertheless indicate that better use can
be made of a wider range of reinforcers than is
generally employed in programs of behavioral

change.
INCENTIVE FUNCTION OF REINFORCERS
The discussion thus far has been primarily
concerned with the performance-enhancing
effects of various contingent events, whether they
be drive-reducing, sensory, or in the form of
prepotent activities. Two different explanations
have been proposed as to how reinforcing

consequences affect behavior. Some reinforcement
theories assume that positive response outcomes
have a direct strengthening effect on stimulus-
response associations and that therefore learning

occurs only as a consequence of reinforcement.
Contiguity theory, on the other hand, distinguishes
between acquisition and performance. Learning,
according to this view, can occur through
contiguous association of stimulus events and

accompanying cognitive processes in the absence
of immediate rewards and punishments. To test

for contiguity learning a variety of experimental
paradigms have been employed in each of which

either overt responding or reinforcement, both
necessary for associative strengthening, are

eliminated. The overall results of these
investigations provide substantial support for the
contiguity principle. In sensory preconditioning
studies, for example, if one of two neutral stimuli

that have been repeatedly paired is then
conditioned to a response, the second stimulus

also becomes capable of evoking the response
without any direct reinforced association (Seidel,
1959). Many experiments, utilizing surgical and
curare procedures to prevent motor responding

during acquisition or extinction, have consistently
obtained learning in the absence of overt
responding. Similarly, innumerable modeling

studies have shown that new response patterns
can be acquired observationally without observers

themselves either engaging in any overt activity or
receiving any reinforcing stimulation.
Although response acquisition is largely
dependent upon stimulus contiguity,
reinforcement variables are considered to be
highly influential in regulating performance.
However, in this more cognitive interpretation of
behavioral change processes, reinforcers are

assumed to affect performance primarily through
their informative and incentive functions.
Reinforcing consequences convey information
about the type of behavior required in a given
situation. Anticipation of desired rewards for

performing the requisite behaviors can increase
and maintain appropriate responsiveness even
though presentation of the earned reinforcers may
be delayed for a considerable time. Indeed, in most
instances persons are motivated by, and work for,

anticipated rewards rather than immediate
reinforcing outcomes.
Contiguous occurrence of stimulus events is no
assurance that they will necessarily be observed.
Anticipated rewards can influence to some degree
what people will pay attention to. Thus by
arousing, focusing, and sustaining attentiveness to
relevant stimulus events, which is necessary for
learning, reinforcers may serve as indirect

determinants of response acquisition. The major
controversy between learning theories is
therefore concerned with the manner in which
reinforcement affects learning rather than with
whether reinforcement plays a role in the

acquisition process.
The basic assumption that reinforcement is a
prerequisite for learning is difficult to refute
empirically. Demonstrations of learning through

contiguity alone are often discounted by invoking
obscure or undetected sources of reinforcement
that are presumed to be operative in the situation.

A purely cognitive interpretation of reinforcement
effects is, however, challenged by results of
experiments with infrahuman subjects in which
reinforcing nutrients are introduced either
directly into the stomach or intravenously. In
these instances the reinforcers are not observable
and consequently, their response-enhancing

effects cannot be attributed to informative or
incentive factors. The overall evidence would
seem to indicate that reinforcers can have both
associative strengthening and performance-
enhancing effects.
Essential Components of Reinforcement

Practices
There are three essential features in the
successful application of reinforcement
procedures. First, one must select reinforcers that

are sufficiently powerful and durable to maintain
responsiveness over long periods while complex
patterns of behavior are being established and

strengthened. Second, the reinforcing events must
be made contingent upon the desired behavior if
they are to be optimally effective. And third, a
reliable procedure for eliciting or inducing the
desired response patterns is essential; otherwise,
if they rarely or never occur there will be few

opportunities to influence them through
contingent reinforcement.
INCENTIVE SYSTEM
It is generally acknowledged that motivation is
crucial for behavioral change. In most personality

theories motivation is conceptualized as enduring
energy systems within the organism, variously
labeled as needs, drives, or motives, which impel

and sustain responsiveness. When motivation is
treated as though it were a persisting internal
entity, this type of orientation not only impedes
development of efficacious change programs, but
it creates pessimism about the possibility of
treating persons who presumably lack the
requisite motivation. It also provides a convenient
rationale for failures that primarily result from
reliance upon weak methods of behavioral control.
Incentive theories of motivation assume that

behavior is largely activated by anticipation of
reinforcing consequences. From this point of view,
motivation can be regulated through arrangement

of incentive conditions and by means of satiation,
deprivation, and conditioning operations that
affect the relative efficacy of various reinforcers at

any given time. Thus, for example, in producing
intellectual strivings in children who display little
interest in academic pursuits, one would arrange
favorable conditions of reinforcement with
respect to achievement behavior rather than

attempt to create in some ill-defined way an
achievement motive, the presence of which is

typically inferred from the behaviors it
presumably actuates.
Given that performance is extensively
determined by reinforcement conditions, the
development and selection of an effective

incentive system is of central importance. The
influential role of reinforcement variables in

behavioral change is illustrated by results of
experiments comparing responsiveness with and
without contingent reinforcement. As part of a
program of research on reading, for example,

Staats and his colleagues (Staats, Staats, Schutz, &
Wolf, 1962) presented to preschool children
programmed material designed to teach them to

read words individually and then combined into
short sentences. When the children were praised

for correct responses but were offered no extrinsic
rewards, they worked at the reading tasks for 15
to 20 minutes and then became bored and restless

and asked to leave. After they no longer wished to
remain in the situation, tangible rewards,
consisting of candy treats, trinkets, and tokens
that could be exchanged for attractive toys, were

introduced. Under the influence of the positive
reinforcers, made conditional upon reading
achievements, the children’s limited “attention

span” suddenly expanded, and they not only
worked enthusiastically at the reading task for 45

minutes, but participated actively in additional
sessions.
A second group of four-year-olds originally
performed the reading task under reinforcement
conditions for two sessions, then the rewards
were discontinued until the children ceased to

participate, following which extrinsic incentives
were again reinstated. During the initial reinforced
sessions the children attended closely to the
reading material and worked actively at acquiring
new reading responses. When reinforcers were
withdrawn, however, the children’s attention,
participation, and reading achievements rapidly
deteriorated. Staats (1965) has further
demonstrated that, given an appropriate incentive

system, even very young children will engage in
complex learning activities with sustained interest
over an extended series of sessions.
The marked changes in positive
responsiveness noted in the above studies

illustrate how low persistence on academic tasks
resulting from inadequate incentives is often
erroneously attributed to basic deficits in the child
in the form of short attention span or low

frustration threshold. Levin & Simmons (1962)
similarly found that low persistence in hyper-
aggressive boys, which is generally interpreted in

clinical theory (Redl & Wineman, 1951) as
reflecting high impulsivity, weak ego control, and
a generalized inability to tolerate frustration, may
in fact be due to inadequate positive
reinforcement. When boys were merely praised
for appropriate responses, they rapidly ceased
responding, sometimes in a highly disruptive

fashion, by tossing the material out of the window
or by climbing on filing cabinets. On the other
hand, when food was used as a reinforcer, the boys
continued to work at the task even though
reinforcement was progressively reduced and

eventually discontinued altogether. The
supposedly short attention span of brain-damaged
and retarded children has also been markedly
increased by creating favorable incentive
conditions (Martin & Powers, 1967). The

foregoing studies, and results obtained by other
investigators (Slack, 1960; Whitlock & Bushell,

1967; Wolf, Giles, & Hall, 1968) indicate that
extrinsic incentives are often essential,
particularly during early phases of behavioral

change programs.
The incentive question poses greatest
problems in the treatment of people who present
severe and profound retardation in social

development. As previously noted, such
individuals are generally unresponsive to verbal

stimuli, customary social reinforcers are
ineffective in modifying their behavior, and the
selected activities often lack acquired reward
value for them. In such cases, change agents are

forced to rely initially upon primary reinforcers,
usually in the form of food. In order to enhance
treatment effects, conditioning sessions are

typically conducted prior to, or during, mealtimes
when food rewards are most effective in

sustaining a high level of responding (Lovaas,
Berberich, Perloff, & Schaeffer, 1966).
Although food rewards may be effectively
employed for short periods, they cannot be relied
upon exclusively in change programs. Food
preferences often vary considerably among
individuals and even within the same person from
time to time. More important, however, the

incentive value of food is highly dependent upon
the level of food deprivation at any given moment;
consequently, food rapidly loses its reinforcing
power through satiation. Since change programs
require frequent, and sometimes lengthy, sessions,

it is necessary to make use of reinforcing events
possessing more enduring incentive value.
For individuals who present gross deficiencies
in conditioned reinforcers, and who are therefore

responsive only to primitive physical
consequences, an important initial objective of
treatment is to endow social and symbolic stimuli

with reinforcing properties. The development of
social reinforcers is particularly critical, since
human behavior is frequently strengthened,
sustained, and modified by praise, approval,
encouragement, positive attention, and affection.
Ordinarily a neutral stimulus acquires
reinforcing properties through repeated

association with primary reinforcement (Kelleher
& Gollub, 1962; Zimmerman, 1957). In their work
with autistic children, Lovaas and his associates
(Lovaas, Freitag, Kinder, Rubenstein, Schaeffer, &
Simmons, 1966) found that negatively reinforcing

properties could readily be conditioned to the
verbal stimulus “no” through association with
aversive stimulation. On the other hand, numerous
sessions in which the word “good” was

contiguously paired with food failed to endow the
social stimulus with any reward value. The
contrasting results were attributed to differential

attentiveness on the part of the children. They
were highly attentive to external cues during
negative stimulation, whereas in sessions
employing rewards they engaged in considerable
self-stimulatory behavior and appeared oblivious
to the relevant social stimuli. It was therefore
decided to employ an instrumental conditioning

paradigm in which the children received food
rewards only if they approached the therapist
whenever he said the word “good.” The children
were thus required to attend closely to the
appropriate verbal cue and to discriminate it from

other stimuli occurring at the same time. After the
social stimulus had been established as
discriminative for primary reinforcement, the
children’s approach responses were intermittently
rewarded on a gradually increasing ratio in order

to further enhance the rewarding capacity of the
verbal cue. This procedure proved highly effective.

In later phases of the experiment, new responses
could be established and maintained in autistic
children through contingent presentation of

verbal approval alone. Moreover, the social
stimulus retained its reinforcing potency over an
extended period on the basis of periodic
association with food rewards.
With less severely autistic children social
reinforcers were established more readily. In

these cases verbal approval and affection in the
form of demonstrative pats and hugs sustained the
children’s positive responsiveness during

numerous sessions devoted to language learning
and the acquisition of social skills. Occasionally,
however, food rewards accompanied the social
reinforcers as a means of preserving their efficacy.
Many of the change programs discussed later rely

heavily upon interpersonal reinforcers in which
desired behavior is responded to with attention,
interest, and approval while undesired activities
are either consistently ignored or socially
disapproved.
A stimulus that has been associated on
numerous occasions with many types of primary
as well as secondary reinforcements acquires the

capacity to function as a generalized reinforcer. In
the treatment of young children or adults for

whom positive social and verbal stimuli have weak
incentive value, tangible generalized reinforcers
are frequently employed. Appropriate
performances are rewarded with monetary

credits, tokens, or points that can later be used to
obtain a variety of rewarding objects and special
privileges. A token incentive system has several

advantages over other forms of material rewards:
The reinforcing value of tokens is relatively

independent of momentary deprivational states;
tokens are not subject to satiation effects and
therefore retain their incentive properties over

long periods; they can be easily presented, if
necessary, immediately upon appropriate
performance; and finally, since individuals can
exchange their token savings for a variety of

attractive items of their own choosing, motivation
and responsiveness are likely to remain at a
consistently high level.
In recent years extensive use has been made of
Premack’s (1965) differential probability principle
in selecting reinforcing events. Because certain

preferred activities can reinforce activities of
lesser value, countless events may be effectively
employed to initiate and maintain desired
behavior. In practical applications of this principle
(Homme, 1966), a person essentially agrees to

perform a certain amount of low probability
behavior to engage in a more preferred activity for
a specified time. Apart from its flexibility and
simplicity, this type of reinforcement system
permits one to utilize naturally occurring activities
as reinforcers simply by arranging them in
appropriate temporal contingencies. Rewarding
activities are frequently used in combination with
generalized reinforcers. In such applications

individuals earn tokens for performing desired
behaviors which can later be used to obtain,
among other things, access to preferred activities.
The preceding discussion has emphasized the
influential role of extrinsic reinforcement with

persons who, for one reason or another, display
intrinsic motivation insufficient to develop the
behavioral repertoires necessary for coping
effectively with customary environmental

demands. It should be emphasized at this point,
however, that in a thoughtfully planned treatment
program, as newly established patterns of

behavior acquire secondary reinforcing properties
extrinsic incentives should be gradually
withdrawn and replaced with more symbolic and
self-monitored reinforcement systems. This issue,
which bears on questions of the durability of
induced changes and the humanistic implications
of different forms of behavioral influence, will be

discussed at length in a subsequent section of this
chapter.
ARRANGEMENT OF CONTINGENCIES
After appropriate reinforcers that have
sufficient incentive value to maintain stable

responsiveness have been chosen, the
contingencies between specific performances and
reinforcing stimuli must be arranged. Parents,

teachers, and psychotherapists intuitively employ
rewards in their attempts to influence and modify
behavior, but their efforts often produce limited
results because the methods are used improperly,
inconsistently, or inefficiently. In many instances
considerable rewards are bestowed, but they are
not made conditional upon the behavior that
change agents wish to promote; long delays often
intervene between the occurrence of the desired
behavior and its intended consequences; special

privileges, activities, and rewards are generally
furnished according to fixed time schedules rather
than performance requirements; and, in many
cases, positive reinforcers are inadvertently made

contingent upon the wrong types of behavior.
Most residential treatment programs, for

example, are conducted on a contingent-
punishment, noncontingent-reward basis. That is,
the participants obtain the maximum rewarding
benefits with few strings attached, but these
rewards and privileges are promptly withdrawn

whenever the residents are uncooperative,
defiant, or disruptive. In one residential treatment

center for delinquent boys that the writer had
occasion to visit, children are given 20 points upon

their arrival, which initially ensures them access
to all of the rewarding resources that the
institution has to offer. However, the boys are
penalized by loss of points and accompanying

privileges for deviant behavior and infractions of
house rules. Within an institutional setting in
which noncontingent rewards are provided at a
high level, the staff members are cast in the
unenviable role of punitive agents, and the boys

can move only in a downward direction. Thus, the
threat of punishment is ever present, but the
positive incentives for behavioral change, though
abundantly available, are poorly managed. Under
these circumstances, the majority of the

participants comply halfheartedly with the
minimum demands of the institution in order to

avoid penalties for any breach of the rules.
Similarly, in most psychiatric facilities, patients
can best maximize their rewards by merely

adopting a passive patient role.
The necessity for arranging appropriate
reinforcement contingencies is dramatically
illustrated by studies in which rewards are shifted

from a response-contingent to a time-contingent
basis (Lovaas, Berberich, Perloff, & Schaeffer,

1966; Baer, Peterson, & Sherman, 1967). During
sessions in which rewards are made conditional
upon occurrence of the desired behavior, the
appropriate response patterns are exhibited at a

consistently high level; by contrast, under
conditions where the same rewards are given but
after a certain time has elapsed, independent of

the client’s behavior, there is a marked drop in the
desired behavior. Reinstatement of response-

contingent reinforcement promptly restores the
high level of responsiveness. These behavioral
changes are particularly striking considering that

interpersonal relationship factors and the amount
of reward remain the same in all phases of
treatment except for the arrangement of the
contingencies. Essentially similar reductions in
responsiveness are obtained when individuals are
provided rewards in advance without any

performance requirements (Ayllon & Azrin, 1965;
Bandura & Perloff, 1967).
In an effective program of change
reinforcement contingencies should be arranged

to provide positive guidance and support for new
modes of behavior, rather than to extract minimal
compliance with situational demands. Social
change programs would become considerably
more efficacious, especially in modifying

pervasively aberrant disorders, if initially the
environment were devised to provide
noncontingent rewards at an adequate but
relatively low level, and preferred reinforcers
were readily available contingent upon the
occurrence of desired response patterns. Under
these conditions, a rehabilitative program can be
managed primarily on a positive reinforcement
basis without resort to those punitive measures

that are commonly employed in residential
treatments.
Behavior is influenced not only by the
contingencies operative in the situation, but also
by temporal aspects of reinforcement. Findings

from experimental studies (Renner, 1964)
demonstrate that behavioral changes proceed
most effectively when reinforcement is made
immediately contingent upon the behavior one

wishes to foster; generally, the degree of control
exercised by reinforcement decreases with
increasing delay. Whenever a delay occurs

between a particular response and its intended
consequences, other behaviors appear during the
intervening period and that response occurring
most closely to the delayed outcome is
immediately reinforced. Since one typically has
little control over the responses that may arise
during a specific temporal interval, delayed

reinforcement may actually strengthen forms of
behavior that change agents have no intention
whatsoever of promoting.
It is widely assumed, on the basis of results
from laboratory studies of delayed reinforcement,

that the effects of rewarding consequences will be
diminished, or even obviated, unless they are
made instantly contingent upon desired
performances. This conclusion needs qualification

because it is based on evidence from
experimentation either with infrahuman subjects
or under conditions where the basis for

reinforcement is not explained. When the
contingencies imposed upon an organism are not
clearly specified in advance, interposing a delay
between the occurrence of a response and its
consequences increases the difficulty of
identifying the arbitrary relationship, particularly
if a series of responses is performed during the

intervening period. As a result, inappropriate
responses tend to be adventitiously reinforced.
Although relevant experimental evidence is
lacking, there is every reason to expect from
informal observation that, in the case of humans,

symbolic activities can effectively mediate a
delayed reinforcement contingency without any
appreciable loss of behavioral control. Therefore,
if contingencies are explicitly defined for an
individual he is able to link eventual consequences

with particular performances. Verbal mediation
will, in all probability, eliminate irrelevant

responses even though a considerable time may
elapse between performance of the requisite
behavior and its consequences. A person who is

paid on a piecework basis, for example, is likely to
maintain a high performance level, although he
receives his total payment at the end of the month
rather than in small amounts immediately after

each unit of work has been completed.
With young children, grossly deviant adults

whose behavior is under weak stimulus control,
and individuals whose efforts extinguish rapidly
under delayed reinforcement contingencies, it may

be necessary initially to employ immediate
concrete rewards; otherwise, such persons are
likely to display rapid decrements in
responsiveness if reinforcing consequences are
postponed. On the other hand, persons who are

responsive to instructional control are usually able
to function adequately under delayed
reinforcement provided the contingencies are
explicitly defined and the incentives are
sufficiently attractive. Moreover, immediate
satisfactions derived from the activity itself and
signs of progress often supplement, and may
eventually replace, ultimate extrinsic
reinforcements in maintaining behavior.
RESPONSE INDUCTION AND EVOCATION
Selection of powerful incentives and skillful
contingency management will, in itself, be of little

consequence unless methods are available for
producing the responses to be reinforced. If the

behavior that a change agent wishes to strengthen
is already present and occurs with some
frequency, then contingent application of

incentives can, from the outset, increase and
maintain the desired response patterns at a high
level. Most cases referred for treatment, however,
present behavioral deficits, and therefore complex
modes of behavior must be organized in
incremental steps, each of which can be easily
acquired. When the initial level of the desired
behavior is extremely low, if the criterion for
reinforcement is initially set too high, most, if not
all, of the person’s responses go unrewarded, so

that his efforts are gradually extinguished and his
motivation diminished. Consequently, in the
beginning stages a low criterion for reinforcement
is generally adopted so that responses that are

within the individual’s capabilities, but may have
only slight resemblance to the desired behavior,
are reinforced. After gross approximations to the
complex pattern of behavior become more
frequent, reinforcement is made contingent upon a

closer response variant. The criterion for
reinforcement is thus raised in small successive

steps in the direction of more complicated forms
of behavior until eventually only the desired

behavior is reinforced.
The effective utilization of successive

approximation procedures is illustrated in a study
by King, Armitage, & Tilton (1960) designed to
increase interpersonal responsiveness in severely
withdrawn schizophrenics. Working on the

assumption that motor responses could be more
easily elicited from these patients than verbal or
social behavior, the therapists first set them the
task of performing a simple motor response which
brought social and material rewards. In successive

phases, the complexity of the task was increased,
and verbal and interpersonal responses were
elicited and rewarded. Also, in later phases,
rewards were presented only when the patients
communicated verbally and cooperated with the

therapist and other patients in order to solve
problems of some complexity. Three other groups

of patients, matched with the reinforcement group
for severity of disorder and length of
hospitalization, concurrently participated in either

traditional interview therapy, recreational
therapy, or received no treatment. The
reinforcement approach proved more efficacious
than all other three techniques in producing

favorable changes in social behavior assessed in
terms of ward observation and standardized

interviews. After 15 weeks of therapy, patients
treated by the reinforcement method displayed
more verbal behavior, less resistance to therapy,
more interest in occupational activities, and were

better prepared for transfer to a more advanced
ward, than patients in the other three groups.
Bensberg and his colleagues (Bensberg, 1965;

Bensberg, Colwell, & Cassel, 1965) provide
additional illustrations of how substantial

behavioral changes can be achieved even with
profoundly retarded children by rewarding small
increments in performance until the more

complex skills are established.
It is widely assumed among proponents of
operant conditioning that the above procedure,
which is variously labeled successive
approximation, shaping, or response differentiation,

is ideally suited for developing new, organized
modes of response previously absent from the
behavioral repertoire of the organism.
Consequently, many therapists spend countless
hours patiently shaping behavior bit by bit when

much of this tedious process can be drastically
reduced. As demonstrated in the preceding
chapter, complex patterns of behavior can be
developed in humans most rapidly through
graduated modeling combined with positive

reinforcement for matching responses. Operant
conditioning through successive approximation

may, however, be exclusively employed with
considerable success to reinstate previously
acquired responses that have been extinguished

and to strengthen performances weakly
established as a result of inadequate incentive
conditions. Thus, by selecting powerful reinforcers
and arranging the requisite contingencies, a

therapist can induce a mute catatonic who
possesses a language repertoire to resume verbal

communication (Isaac, Thomas, & Goldiamond,
1960); schizophrenics who have developed
adequate work repertoires can be led to
participate again in vocational activities (Ayllon &

Azrin, 1965); delinquents who refuse to attend to
school assignments can be motivated to improve
their academic performances (Cohen, 1968); and,

in verbal conditioning experiments, college
students who command an abundant supply of

personal pronouns can be subtly prompted to emit
these verbal responses at a relatively high rate
(Krasner, 1958).
In addition to utilizing the method of
successive approximation and behavioral
modeling for producing complex responses, one
can rely upon verbal prompts that specifically
instruct individuals how and when to perform the

reinforceable behavior (Baer & Wolf, 1967).
However, in cases that are unresponsive to social
forms of response guidance it may be necessary to
employ nonsocial stimuli that exercise strong
control over the behavior in question, even though

the eventual aim is to have the behavior occur in
response to quite different stimulus conditions. In
using nonsocial cueing procedures one initially
introduces discriminative stimuli that exert strong
control over the desired behavior. After the

responses have been evoked and firmly
established, the arbitrary stimulus supports are

“faded” or gradually withdrawn as control is
transferred to stimuli likely to function as the
major elicitors under naturalistic conditions. Thus,

for example, in augmenting attending behavior in
severely retarded children who were totally
unresponsive, Bensberg (1965) initially flashed
lights on a wall while he simultaneously instructed

them to look, and rewarded them for gazing at the
light. In this way attending responses, which are

prerequisite for social learning, were increased
and eventually brought under verbal stimulus
control. The use of tasks graduated in difficulty
also includes instances in which stimulus

conditions are arranged so that rewardable
behavior can be readily elicited at each successive
step.
A final method for evoking desired behavior,

which is sometimes employed with persons who
prove unresponsive to extensive stimulus
prompts, involves physical response guidance,
wherein individuals are assisted physically in
making the correct responses. In teaching autistic
children grammatical relationships between
objects (Lovaas, 1966), for example, if a child fails
to execute the response corresponding to the
verbal instruction, “Put the block inside the box,”

the therapist moves the child’s hand with the
block to the box and rewards the passively
performed action. On subsequent trials the
amount of manual guidance is gradually reduced
until the behavior is performed without

assistance.
Ethical Implications of Reinforcement

Practices
The deliberate use of positive reinforcement,
particularly in the form of tangible rewards, often

gives rise to ethical objections and concerns about
harmful effects that may result from such
practices. The attitude most commonly expressed

is that desirable behavior should be intrinsically
satisfying. It is feared that, if persons are
frequently rewarded, they will be disinclined to
behave appropriately unless continually paid to do
so, and when the customary rewards are
discontinued they will cease responding

altogether. It is further assumed that rewarding
practices not only establish weak and unenduring
behavior, but that contingent reinforcement is
likely to interfere with the development of
spontaneity, creativity, intrinsic motivational

systems, and other highly valued self-determining
personality characteristics. Some of the more
intemperate criticism considers the deliberate use
of reinforcement to be deceptive, manipulative,
and an insult to the personal integrity of human

beings.
For reasons presented above most persons

whose own behavior is strongly influenced by
social recognition, praise, approval, special
privileges, and monetary incentives are quick to
disclaim the use of rewarding practices (Bandura
& Walters, 1959; Sears, Maccoby, & Levin, 1957)
and to deny that their behavior has been
externally regulated (Rogers, 1960). It should also

be noted that, paradoxically, one is apt to
encounter less concern over the use of aversive
methods of control by threat, coercion, and
deprivation of privileges, methods which often do
produce the negative behavioral outcomes

inappropriately attributed to procedures relying
upon positive incentives.
The fact that behavior is strongly influenced by
its consequences is not a phenomenon created by

behavioral scientists, any more than physicists are
responsible for the laws of gravity. The process of
natural selection has favored organisms with

adaptive feedback control systems in which
reinforcing consequences serve as a major
regulator of behavior. Indeed, if behavior did not
change as a function of its outcomes, one’s life
span would be drastically curtailed. Selection of
the types of incentives by which the behavior of
others is to be established, guided, and maintained

is, of course, an ethical issue. However, the
behavioral effects resulting from applications of
different psychological procedures are entirely an
empirical matter. Available evidence from
laboratory and psychotherapeutic studies

suggests that reinforcement procedures, if
thoughtfully and skillfully implemented, can
produce enduring changes in social behavior and
facilitate the acquisition of self-monitoring
reinforcement systems. If, on the other hand,

operant techniques are crudely applied, and the
incentives are inappropriate to individuals’

developmental levels, then the change program
may be insulting as well as ineffective.
In discussing the systematic utilization of
positive incentives it is important to recognize
that a change program represents a continuum of
psychological experiences in which the type,
amount, and source of reinforcement regulating

behavior are gradually modified. Hence, the
incentives employed initially to establish new
patterns of social behavior and to develop
symbolic reinforcers may differ considerably from
the stimulus events that ultimately assume

controlling and reinforcing functions.
Critics of reinforcement methods generally
create the impression that change agents work
with mature and intrinsically motivated persons,

but, rather than appealing to higher symbolic
motivations, insist on imposing crass materialistic
incentives upon them. There are undoubtedly

some practitioners who apply incentive
procedures thoughtlessly and ineffectively.
Ordinarily, however, primary rewards are
employed in initial stages with persons who are
not reinforceable with other types of events and
who would otherwise remain inaccessible to
treatment. In the latter cases it would be no more

appropriate to rely upon developmentally
advanced incentives than to teach young children
how to count by commencing with the principles
of advanced mathematics. After reinforcing
functions have been imparted to social and

symbolic stimulus events, then more subtle and
naturally occurring reinforcers are increasingly
employed. Without the initial concrete training,
psychologically incapacitated persons are
relegated to a subhuman existence in custodial

institutions.
REINFORCEMENT SYSTEMS AND DURABILITY OF

BEHAVIORAL CHANGES
Demonstrations that behavior can be
maintained at a satisfactory level through

reinforcements mediated by change agents are of
limited significance unless the response patterns
endure long after the specially created
contingencies have been discontinued. There are

several ways in which reinforcement systems can
be devised and altered during the course of
treatment to ensure that existing behavior does
not readily extinguish.
Change in Frequency or Magnitude of
Reinforcement. After response patterns have been

firmly established through continuous
reinforcement, the schedule is gradually thinned
out, by providing rewards on increasingly variable
ratios and intervals so that the rewarding
consequences occur only periodically. As shown in

the introductory chapter, intermittently reinforced
behavior is extremely resistant to extinction. The

durability of behavior under less favorable
reinforcement conditions can also be increased by

gradually reducing the amount of reward after the
behavior has been sufficiently strengthened, or by
increasing the amount of work per reinforcement
(Staats & Butterfield, 1965).
Change in the Locus of Reinforcement. In most
instances many different interpersonal and
material rewards are potentially available, but
they remain inaccessible to individuals who lack
the social and vocational skills to attain them. In a

similar manner, because of behavioral deficits or
inhibitory tendencies persons may refrain from
participating in activities that would provide rich
sources of enjoyment. If proficiency in the

necessary skills and social responses were
established, they could be adequately supported
by reinforcements regularly available within the

environment. The main purpose of specially
arranged reinforcement is to develop and to
sustain behavioral repertoires to the point where
the individual makes successful contact with
existing sources of positive reinforcement. Once
this is achieved, the arbitrary contingencies can be
completely withdrawn without weakening or

reducing social behavior.
An excellent example of the successful shift of
the locus of reinforcement from adult change
agents to peer group members is provided in the
treatment of an extremely withdrawn girl,

referred to in Chapter 6 (Allen, Hart, Buell, Harris,
& Wolf, 1964). After a brief period in which
teachers’ interest and attention were made
contingent upon interaction with peers, the girl

entered into a great deal of social play with other
children. However, when adult reinforcement for
peer interactions was temporarily removed during

the early phase of treatment, she reverted to her
isolate pattern of behavior; reinstatement of the
therapeutic contingency restored social play to its
previous high level. As the girl derived increasing
enjoyment from play activities with peers, adult
rewards for interaction with children were
progressively diminished to a normal amount of

attention, and the schedule for nonreinforcement
of adult contacts gradually relaxed. Eventually the
treatment program was discontinued altogether
and no special contingencies were arranged
thereafter. The increase in social interaction with

other children nevertheless endured, as revealed
in observations of behavior conducted at various
times following the termination of the program.
Other case studies, specifically designed to
investigate the durability of behavioral change

(Baer & Wolf, 1967), have shown that if adults
maintain their reinforcement support of social

behavior in children until they achieve
reciprocally rewarding interactions with peers,
the children’s behavior comes increasingly under

peer control and is little affected by withdrawal of
adult social reinforcement.
Results from the above studies, and others

conducted in the same manner, show that
established patterns of behavior maintain their
strength after specially arranged consequences

are discontinued provided the behavior is brought
under the influence of favorable contingencies
within the individual’s social milieu. In cases,

however, where the reinforcement practices in
naturalistic situations are either deficient or
grossly deviant it is doubtful that lasting
behavioral changes can be achieved, unless the
program is extended to encompass significant

members of the individual’s social environment.
Change in the Form of Reinforcement. In
previous discussions of the treatment of autistic
children it was shown how their behavior could
initially be modified only through the use of
immediate primary reinforcers which were
gradually reduced and eventually discontinued as
social stimuli acquired reinforcing functions.

Another illustration of the transformation of
reinforcing supports for behavior during the
course of treatment is provided by Wahler (1968)
who successfully modified extreme oppositional

behavior in children by altering their parents’
reinforcement practices. An initial program in
which parents ignored their children’s resistance
to requests and rewarded cooperative behavior
with approval proved relatively ineffective. A

subsequent reinforcement system combining
parental approval with tokens exchangeable for

prized toys produced dramatic and enduring
increases in cooperativeness. Thereafter, the

tokens were gradually eliminated and cooperative
behavior was stably maintained by social approval

alone.
In order to acquire proficiency in complex
behaviors people must engage in long hours of
arduous work, give up attractive competing

activities, and delay a host of immediate
gratifications that may be readily available.

Because the learning process involves some
degree of self-denial and other negative aspects,
many people fail to develop minimal competencies
even though threats and coercive pressures are

continuously brought to bear. This prevalent
problem can be rectified more satisfactorily and
humanely by applying an arbitrary reward

contingency until the behavior is developed to the
stage at which it produces natural reinforcing

consequences. Thus, for example, extrinsic
rewards may be employed temporarily to teach
children how to read, but after written subject

matter becomes sufficiently reinforcing in itself to
sustain further development of reading skills, the
artificial contingency may be withdrawn. Many
forms of behavior, such as communicative facility
and manipulatory skills, which permit an
individual to regulate his environment more

effectively, persist with little external support
because they are functional in producing
rewarding outcomes. New performances are also

partially sustained by the sensory feedback that
they naturally produce.
If rewards are recurrently and explicitly
associated with cues that signify competency or
correctness, then symbolic stimulus events that
have informative value, and qualitative differences

in performance, acquire secondary reinforcing
properties. At this higher developmental level,
cues designating the adequacy of one’s
performance may be as effective a reinforcer as
monetary incentives (Lewis, Wall, & Aronfreed,
1963; Miller & Estes, 1961). Once informative
response feedback becomes a source of personal
satisfaction, maintenance of behavior is less
dependent upon extrinsic social or material

incentives. It should be noted, however, that any
reinforcement ensuing from confirmation of the
correctness of one’s responses is probably
mediated through a self-reinforcement process
rather than automatically generated. It is

exceedingly unlikely, for example, that correctness
feedback on tasks that are personally devalued or
are regarded as elementary will have much, if any,
reinforcing value. On the other hand, confirmation
of attainments that exceed personal standards of

what constitutes a worthy performance will tend
to activate positive self-evaluations.
The highest level of autonomy is achieved
when behavior generates self-evaluative and other
self-reinforcing consequences. In such instances, a
person sets himself explicit standards of
achievement and creates either self-rewarding or
self-punishing consequences depending upon the
quality of his behavior relative to his self-imposed

standards. Self-evaluative reactions can not only
maintain behavior under conditions of minimal
external support, but they may override the
influence of social rewards for behavior that
conflicts with the person’s own norms of

acceptable conduct. The manner in which
modeling and reinforcement procedures can be
used to establish intrinsic performance standards
by which a person monitors his own behavior has
been previously discussed.
Established patterns of behavior are most
likely to endure across a bewildering variety of

reinforcement contingencies if the significant
reinforcing events are either intrinsically related
to the behavior, or self-administered. The
conditions necessary for developing complex
behavioral repertoires through reinforcement
methods have been explicitly defined and are
usually faithfully implemented. Procedures for

developing reinforcement functions to symbolic
events need to be further refined and applied
systematically in programs of behavioral change.
Judging by evidence from laboratory studies, the
types of self-regulatory mechanisms that

humanistically oriented commentators consider to
be antithetical to behavioral approaches are, in
fact, most readily developed by methods derived
from social-learning theory.
It is also often erroneously assumed that

change programs based upon principles of
reinforcement involve a simple mechanistic

technology that can be applied almost reflexively
by anyone who possesses sufficient perseverance.
On the contrary, successful implementation of
social-learning methods calls for considerable
ingenuity and sensitive responsiveness to
psychological changes in individuals throughout
the period of treatment. Among other

requirements, one must devise efficacious
incentive systems, select appropriate
reinforcement schedules, arrange essential
contingencies and gradually modify them as
treatment progresses. One must also create

methods for evoking desired responses with
sufficient frequency for them to be strongly
established. And finally, it is necessary to select
and to train appropriate persons to implement the
procedures in naturalistic contexts. Whereas in

conventional treatments people are frequently left
to their own devices in transferring whatever they

may have learned to their everyday life, social-
learning approaches devote considerable attention
to arranging conditions necessary to ensure

optimal generalization and maintenance of newly
established modes of behavior. In view of the
amount of inventiveness required, it is not
surprising that, although principles of

reinforcement have been in existence for many
decades, the derivation of efficacious procedures

has been disappointingly slow. For similar
reasons, the implementation of contingency
systems by amateur or less artful operant
practitioners is often appallingly crude.
SPECIFICATION OF REINFORCEMENT
CONTINGENCIES
In most experimental investigations of
reinforcement processes, instructions that specify
the desired behavior and its programmed
consequences are deliberately minimized or
ignored altogether. Indeed, Skinner (1963) has

strongly cautioned against the use of instructional
control procedures in learning experiments on the
grounds that they circumvent and obscure the
functional analysis of behavior. Laboratory
experimentation designed to explore the extent to
which behavioral phenomena can be brought

under the control of different types of
reinforcement conditions should, of course, avoid
combining variables in ways such that their
individual contributions cannot be distinguished.

However, rigid adherence to isolated procedures
is ill-advised in change programs which must
frequently combine a variety of methods in order
to achieve optimal results. Some devoted partisans
of the operant approach, nevertheless, often rely

exclusively upon reinforcement practices to
develop response patterns that can be readily

produced by the use of simple instructions,
behavioral demonstrations, or appropriate verbal

modeling cues.
Considerable experimental evidence, which

will be reviewed in a later chapter, demonstrates
that awareness of response-reinforcement
contingencies can markedly accelerate behavioral
change. It is also abundantly evident, as any parent

will attest, that advice, instructions, requests, and
other verbal forms of guidance often have no
enduring effects or go unheeded. Systematic
studies conducted with children (O’Leary, 1968)
and adolescents (Phillips, 1968) indeed show that

prescribing behavioral rules alone is relatively
ineffective in changing their behavior. The power
of verbal influence is largely determined by the
anticipated or accompanying response
consequences. This is revealed in a study by

Ayllon and Azrin (1964) designed to evaluate the
relative efficacy of instructions and reinforcement,

used singly and in combination, for reinstating
acceptable dining behavior in adult
schizophrenics.
The treatment program attempted to get
patients to pick up cutlery at the serving center,
which they rarely did; instead they ate their food
with their hands. Following the baseline period of

observation during which no special consequences
were arranged, a reinforcement procedure was
introduced in which patients who picked up the
necessary utensils were immediately given,
without any explanation, their choice of extra food

or cigarettes. In a subsequent phase, instructions
were added to the reinforcement procedures, by
having the attendants explain, “Please pick up
your knife, fork, and spoon, and you have a choice
of extra milk, coffee, cigarettes, or candy.”
As shown in Figure 4-1, reinforcement alone
produced no change whatsoever in the patients’

behavior. Here the reinforcement procedure
proved totally ineffective because the vast
majority of patients never exhibited any responses
that could be reinforced, and the few who
Figure 4-1. Percentage of patients who picked up cutlery
during the baseline period, during the reinforcement
phase in which appropriate responses were promptly
rewarded, and during a period when instructions were
combined with reinforcement. Ayllon & Azrin, 1964.
occasionally picked up cutlery never figured out
the basis upon which they were given extra treats.
By contrast, when instructions were combined

with reinforcing consequences the patients
showed a marked and sudden increase in
appropriate behavior, and a number of them
maintained this change after the contingent
rewards were discontinued.
In order to assess the efficacy of instructions

alone, a second group of patients was simply
asked to pick up the necessary eating utensils at
each mealtime. Instructions were found to be

initially effective for about half of the patients, but
in the absence of any consequences for following
or ignoring the requests, verbal directives

eventually lost most of their controlling power
(Figure 4-2). During the next phase, the
instructions were continued but, in addition,
patients gained immediate access to the food
Figure 4-2. Percentage of patients who displayed the
appropriate responses during baseline, instruction, and
combined reinforcement-instruction phases of the
study. Ayllon & Azrin, 1964.
counter whenever they picked up the cutlery and
were sent to the end of the serving line if they did
not. Under the combined influence of instructions

and response consequences virtually every patient
regularly exhibited the appropriate dining
behavior. Comparison of the two sets of data
reveals that delayed access to rewards produced
by nonresponsiveness was considerably more

powerful in modifying the patients’ behavior than
extra treats for performance of appropriate
responses.
Applications of Contingency Systems
The extensive growth of programs utilizing

positive incentives in one form or another
precludes a complete review of the countless
clinical, remedial, and developmental applications
of reinforcement principles. Rather, some
representative contributions that illustrate the
procedures and their efficacy in modifying a wide
range of behaviors will be discussed. Since
investigations of reinforcement variables alone

are most often conducted by researchers working
within the Skinnerian conceptual framework, the
degree of success of these methods is rarely
evaluated through the use of control groups to
measure the contribution of uncontrolled

variables, or comparisons between groups
involving different experimental operations.
Instead, the intrasubject replication design is
generally employed for isolating the variables
governing change. In this method of research a

given pattern of behavior is repeatedly induced
and eliminated in the same subject through

successive reversal of treatment conditions
(Sidman, 1960). Intrasubject replication is the
most convincing means of demonstrating the

functional relationship between behavioral
phenomena and their controlling conditions.
However, there are certain limitations and
evaluative problems in the use of this

methodology.
Intrasubject replication is well-suited for

investigating performance control processes but it
cannot be employed in studying learning
phenomena in which certain experiences produce

a more or less irreversible change in the behavior
of an organism. For example, after persons have
acquired communicative speech, reading skills,
and various social and psychomotor competencies,
one cannot erase these response capabilities and

thus restore the original behavioral deficits
through nonreinforcement operations or any
other psychological procedure.
A number of interpretive complications arise
in the use of this methodology even in the study of
performance changes. There are no difficulties in
evaluating findings when large successive changes
in behavior occur rapidly and consistently across

different subjects. In many instances, however, not
only are the accompanying behavioral changes
less dramatic, but some individuals remain
essentially unaffected by repeated exposure to the

same treatment conditions. Replicative failures
are usually attributed to inadequacies in the
reinforcers employed, whereas successes are
assumed to result from the manipulated
reinforcement variables. It is entirely possible,

however, that in successful cases the behavioral
changes are largely due to the influence of

unobserved variables that happen to co-vary with
the reversal of treatment conditions. The

evaluative problem is further complicated by the
fact that, in cases where successful behavioral

control is achieved, no statistical criteria have
been developed to evaluate whether the
magnitude of change produced by a given
treatment exceeds the variability resulting from

uncontrolled factors operating while the
treatment condition is not in effect. Many readers

have undoubtedly experienced frustration in
attempting to evaluate conclusions drawn by
investigators on the basis of visual inspection of
frequency curves that not only involve

considerable variability during baseline
conditions, but differ widely between subjects and
are somewhat irregular across successive

replications.
The intrasubject replication design also

precludes accurate assessment of the relative
efficacy of different treatment variables. Changes
that are sequentially produced in a given
individual by the application of diverse methods
cannot be directly compared for several reasons.
The degree of influence required to create an
initial change may differ from that needed to effect
additional improvements in the same behavior. In

developing linguistic functions in autistic children,
for example, Lovaas (1966) found that children
were slow to acquire the first few words but
subsequent word learning proceeded at a
relatively rapid rate. A similar increase in the rate

of response acquisition as treatment progressed
was noted by Staats, Minke, Goodwin, & Landeen
(1967).
In many instances the original baseline is not

recoverable; consequently, the effects that
different variables have on behavior must be
compared against dissimilar performance levels.

Even if the original response baseline can be
recovered, it may be much easier to reinstate a
given behavior than to create it initially.
Zeilberger, Sampen, and Sloane (1968) have, in
fact, shown that behavior can be modified more
quickly the second time, a finding that is evident in
most graphs based on experiments in which the

same controlling variables are successively
applied and withdrawn. Not only does prior
performance of responses increase the speed with
which they can be reinstated after being
extinguished, but during the initial experiences

learning sets are acquired which can result in
marked improvement in performance of quite
different responses (Kimble, 1961; Harlow, 1949).
Finally, the reinforcement value of a given

event can be markedly altered through contrast
with previous or contemporary conditions of
reinforcement (Buchwald, 1960; Dunham, 1968).

Therefore, sequential changes associated with
different types of reinforcement operations reflect
relational influences as well as the specific
properties of reinforcement procedures. Hence,
the implicit assumptions that repetitive control
does not alter the modifiability of the behavior in
question, that behavior at different levels is

equally modifiable, and that reinforcement
operations are unaffected by contrast in incentive
conditions are all probably untenable. The relative
potency of different controlling variables can,
therefore, be best assessed through experimental

designs involving matched groups.
Some of the published reports involving
reinforcement techniques are based upon
individual cases in which, for practical or ethical

reasons, successive reversals of contingencies
have not been attempted. Although these types of
studies provide less convincing evidence

regarding the variables responsible for observed
changes, the results nevertheless have important
suggestive value, particularly when changes are
effected in persons who have proved totally
unresponsive to other methods of change.
MODIFICATION OF DEVIANT SOCIAL BEHAVIOR
Differential reinforcement has been widely
employed for the modification of deviant behavior
in both adults and children. In these treatment
programs rewarding consequences for desired

behavior are typically combined with extinction,
modeling, and in some cases with punishment

procedures. However, the studies reported in the
present section are primarily concerned with the
control of behavior through its positive

consequences.
Ayllon and his associates have conducted an

extensive program of research in the development
of reinforcement procedures for the modification

of gross behavior disorders in adult psychotics. In
the early studies (Ayllon & Michael, 1959) nurses
and hospital attendants were trained to record the
frequency with which patients exhibited specific
patterns of behavior, and to arrange in naturalistic
situations the reinforcement contingencies—
usually in the form of social attention and food

rewards— necessary to bring about desired
changes. By withholding attention for bizarre

forms of behavior while selectively reinforcing
rational response patterns, nurses succeeded in
markedly reducing or completely eliminating
psychotic verbalizations (Ayllon & Haughton,

1964), chronic anorexia (Ayllon, Haughton, &
Osmond, 1964), and a host of other deviant
behaviors of long standing considered indicative
of psychotic pathology (Ayllon, 1963; Ayllon &
Michael, 1959). In later studies (Ayllon & Azrin,

1965, 1968) the scope of operant conditioning
methods was broadened by the use of a wider
range of positive incentives, which were applied
on a group basis for establishing social and
vocational competencies in chronic psychiatric
patients. A detailed account of the latter program
will be presented later.
Numerous projects have been reported in
which reinforcement principles are systematically

employed to alter deviant behavior in children.
These studies provide impressive testimony that
children’s behavior can be powerfully controlled

by the social consequences provided by adults.
Each case involves intrasubject replication in
which the incidence of particular response
patterns is objectively recorded under naturally
occurring contingencies and during subsequent

periods when therapeutic contingencies are
alternately applied and withdrawn. The findings
demonstrate that persistent problem behaviors
can be successfully eliminated, reinstated, and
extinguished a second time by altering the amount
of adult interest and attention produced by the
deviant behavior. Among the disorders
successfully treated through such selective
reinforcement are extreme withdrawal (Allen, et

ah, 1964; Brawley, Harris, Allen, Fleming, &
Peterson, 1969; Johnston, Kelley, Buell, Harris &
Wolf, 1963), regressive crawling (Harris, Johnston,
Kelley, & Wolf, 1964), extreme passivity (Johnston,
Kelley, Harris, & Wolf, 1966), hyperactivity and

aggressive behavior (Allen, Henke, Harris, Baer, &
Reynolds, 1967; Hall, Lund & Jackson 1968), and
depressive feelings and marked overdependency
(Wahler & Pollio, 1968).
A noteworthy feature common to the

procedures discussed above, apart from their
demonstrated efficacy, is the fact that the change

programs are conducted within natural settings by
teachers and parents utilizing reinforcing events
that form a natural part of spontaneous
interpersonal relationships. It is possible, of
course, to modify behavior under contrived
conditions with incentives that are highly
attractive but are rarely employed, for practical or

other reasons, in everyday situations. Although
results of such studies may have some value in
demonstrating that a particular form of behavior
can be controlled by artificially arranged
consequences, such treatment procedures

eventually have to be supplanted for several
reasons: First, behavioral changes established
under artificial conditions must be supplemented
with generalization training, with respect to both
the types of tasks and incentives employed, in

order to ensure adequate transfer effects. Even if
contrived procedures were equally effective, they

frequently necessitate specialized equipment and
personnel, which limit their applicability.
Furthermore, although relevant data are rarely

obtained, one must also take into account possible
self-evaluative consequences that arbitrary
interventions have upon their recipients, as well
as social effects upon the attitudes and behavior of

others who have opportunities to observe the
treatment.
Countless studies employing reinforcement
techniques have been published in recent years,
but they lack the replicative control feature.

Nevertheless, they report favorable outcomes with
such diverse clinical problems as autistic behavior
(Lovaas, 1968) severe anorexia (Bachrach, Erwin,
& Mohr, 1965; Leitenberg, Agras, & Thomson,
1968), school phobias (Patterson, 1965), socially

disruptive behavior (Zimmerman & Zimmerman,
1962), mutism (Sherman, 1965; Straughan, 1968),
psychogenic seizures (Gardner, 1967), self-
mutilative activities (Allen & Harris, 1966),
antisocial behavior (Colman & Baker, 1968), and
innumerable other types of deviant behavior,
some of which are discussed in sections that
follow. Although one finds a paucity of
experiments in which the relative efficacy of

diverse methods of therapy is systematically
evaluated, results of several controlled
investigations with schizophrenic patients (King,
Armitage, & Tilton, 1960; Peters & Jenkins, 1954;
Schaefer & Martin, 1966) and antisocial character

disorders (Colman & Baker, 1968), disclose that
treatment based upon reinforcement principles
produces greater change in interpersonal behavior
than do programs following conventional lines.
In a significant extension of reinforcement

procedures, Patterson and his colleagues
(Patterson, Ray, & Shaw, 1968) have achieved

some success in modifying deviant behavior by
altering the reinforcement patterns of familial
systems and peer groups. According to the
authors’ etiological formulation, deviant behavior
typically arises under conditions of low levels of
positive reinforcement and nonreciprocal social
interactions among family members. The children

are, therefore, forced to resort to extreme forms of
behavior to elicit reinforcing reactions from
others. As a further consequence of
nonreciprocity, children are likely to become
increasingly controlled by the peer group and less

responsive to adults. This change, in turn, leads
adults to resort to aversive forms of control, which
further reduces their influence as reinforcing
agents.
The treatment approach, which involves a

four-step program, takes place in the home. After
two weeks of baseline observation of familial

interactions, parents are provided with a specially
prepared booklet designed to familiarize them
with general principles of reinforcement,
extinction, aversive control, inadvertent
reinforcement of deviant behavior, and with
procedures for recording interpersonal behavior.
In the second phase, parents are asked to list the

child behaviors they wish to modify. They are then
assigned a special hour during each day to record
the incidence of these behaviors, the various
consequences they engender, and the family
members who provide the consequences. After

they have learned to observe interpersonal
contingencies accurately, parents are helped,
through ample demonstration and supervised
practice, to alter the reinforcement contingencies
that they provide for both deviant and desired

response patterns. The family problems are thus
modified one at a time.
When necessary, new reinforcement practices
are also introduced in the classroom setting and in
the peer group. Control over deviant behavior in
extra-familial situations is typically achieved by a
peer contingency procedure in which both the
child and his peers initially earn desired rewards
for his good behavior. The material reinforcers are

then gradually withdrawn until eventually the
child’s behavior is entirely maintained by social
reinforcement from teachers and peers. Following
termination of the formal treatment program,
telephone contact is maintained on a diminishing

schedule, and home observations are conducted
periodically over a six-month follow-up period.
Results based on six families that have
participated in the above program show that

parents reduced the frequency with which they
positively reinforced deviant behavior from an
average rate of 35 percent during the baseline

period to 10 percent at the end of the intervention
program. Modification of familial contingencies
not only decreased the family’s output of deviant
behavior, but it increased the amount of positive
social reinforcement in the entire social system,
and it produced a more reciprocal quality to the
interactions between the various family members.

Moreover, these favorable changes tend to be
effectively maintained over time. Although this
approach appears to hold promise, the marked
variability in baseline rates of deviant behavior
before the new reinforcement practices were

inaugurated, and the differential responsiveness of
families to the program, indicate that further
refinements and assessments are needed.
DEVELOPMENT OF SOCIAL AND SELF-

MANAGEMENT SKILLS IN SEVERELY RETARDED
CHILDREN
Reinforcement techniques have also proved to

be of value for establishing basic social and self-
care skills in profoundly retarded children who,
because of their primitive level of behavior, are
usually considered uneducable and are hence
relegated to institutional back wards. Such
children have been toilet-trained (Giles & Wolf,
1966; Hundziak, Mowrer, & Watson, 1965), taught

to dress and undress themselves, to feed
themselves with utensils, to manage their personal
grooming, and to respond to verbal directions,
which is of considerable aid in the social training

process (Bensberg, Colwell, & Cassel, 1965;
Girardeau & Spradlin, 1964; Minge & Ball, 1967;
Roos, 1965). In addition, the efficacy of these
methods for developing communication skills,
interpersonal response patterns, and other

complex forms of behavior in severe retardates is
being explored. Bensberg notes parenthetically

that such programs have not only benefited
retarded children, but the functions of attendants,

who implement the training procedures, have
changed from dismal custodial care to active

participation in assisting children to gain
competencies within the limits of their ability.
SYMBOLIC LEARNING
In recent years, reinforcement procedures
have been used effectively in conjunction with

programmed instructional materials to establish
complex symbolic forms of behavior. Staats’
(1965) program of research on the acquisition of

reading behavior furnishes one such example.
Reading involves complicated processes in
which children must learn both to discriminate
among intricate verbal symbols and to associate
appropriate verbal responses to them. Complexity

arises primarily because the same elements in a
compound word stimulus must elicit different

responses depending upon the context in which
they occur. Since words contain many common
stimulus properties (e.g., counsel, council) and in

most cases, word differentiation relies upon subtle
cues, the development of reading responses
constitutes a demanding associative form-
discrimination task. In addition to difficulties
created by high stimulus similarity, the

instructional material itself typically serves as a
weak source of positive reinforcement,

particularly for young children. An effective
reading program therefore requires extensive
training utilizing material that is carefully
sequenced, repeated pairing of words with their

pictorial or verbal associates, immediate and
continuous feedback of the correctness of the
responses, and an incentive system capable of
sustaining children’s attention and active
responsiveness over long periods. These essential

conditions are incorporated in the semi-
automated method devised by Staats to study
reading acquisition.
In the initial pre-training phase, children are
reinforced for imitating single vowels or words
spoken by the experimenter. After verbal
modeling has been well established, a word is

shown with several drawings of simple objects,
one of which corresponds to the printed stimulus,
and the experimenter pronounces the word. When
children have learned through discriminative
consequences to match words with their pictorial

representations without verbal prompts, the
actual reading training is begun. The reading task
is presented to children primarily in the form of a
matching-to-sample discrimination procedure. In
each learning sequence a sample word is shown in

the top aperture of a panel, and is matched by one
of three words presented simultaneously in a

bottom row of windows. The experimenter names
the stimulus word and asks the children to repeat
the word and to select the comparable item from

among the alternatives. If the child reads the word
correctly, he is immediately reinforced with
tokens which may be exchanged for preselected
toys or other desired items. Whenever the child’s

response is incorrect, the entire sequence is
repeated. After children have learned to read

words individually, they are presented in
sentences and in short paragraphs composed of
previously learned material.
In a discrimination learning task the error rate

can be effectively controlled by the use of cueing
procedures and stimulus alternatives that are
easily discriminable from the correct choice. By
gradually fading out the stimulus supports for the

appropriate behavior and employing
progressively finer contrasts between the stimulus
alternatives (Rocha e Silva & Ferster, 1966),
children can eventually learn to respond to subtle
features of words. Most other symbolic activities,
such as abstraction and concept formation,
similarly depend upon the establishment of subtle
discriminations. This can be best achieved by
working from relatively gross contrasts to

successively smaller differences between stimuli.
Although comparisons with other instructional
methods have not been made because of the
exploratory nature of this research, nevertheless
Staats has accumulated considerable data (Staats,

Finley, Minke, & Wolf, 1964; Staats, Minke, Finley,
Wolf, & Brooks, 1964) regarding the potential
value of this approach for establishing reading
behavior in preschool children. Further, the

influence of schedules of reinforcement on rate of
reading acquisition has been studied
systematically in several cases with intrasubject

replications. The results, though somewhat
variable, demonstrate that under conditions of
reinforcement children maintain strong interest in
the reading task, and continue to acquire new
reading responses over an extended series of
sessions even though, for experimental reasons,
the social interaction between tutor and child is

severely curtailed. In educational or remedial
applications one would, of course, question the
wisdom of rigid adherence to such highly
impersonal practices and exclusive reliance upon
material reinforcers. As might be expected from

previous research, intermittent reinforcement
generally produced higher reading response rates
than continuous reinforcement. Moreover, during
periods when reinforcement was temporarily
discontinued reading behavior rapidly

deteriorated.
These procedures were initially extended by

Staats to the study of reading acquisition in
retarded children and remedial reading in a
delinquent adolescent (Staats & Butterfield, 1965).
The latter case involved a 14-year-old boy who, in
addition to accumulating a long and varied history
of aggressive destructive behavior, had never
received a single passing grade in any school

subject; despite eight and a half years of classroom
instruction, his reading achievement was only at
the second grade level. Because of his uncharitable
attitudes and behavior toward school personnel,
and occasional dismantling of school property, he

was considered to be uneducable, incorrigible, and
mentally retarded.
Paralleling the methods employed in the
laboratory studies, the boy learned first to read

words presented singly, then combined in
sentences, and finally organized into short stories.
The vocabulary items, which were selected from

standard reading material arranged according to
difficulty level, were presented individually on
index cards and the boy was asked to pronounce
them. Following each correct reading response the
boy received token rewards which he saved for
various articles and money allotments. After the
boy had mastered the vocabulary items he earned

additional tokens for accurate oral reading of
paragraphs, silent reading of entire stories, and for
correct answers to questions assessing
comprehension of story content.
During the training program the boy covered a

sizable range of reading material with
undiminished interest. He not only acquired new
reading responses at a relatively high rate, but he
showed increasing ability to read new words on

first presentation, and retained much of what he
had learned. This notable progress was also
reflected in reading achievement test scores

obtained prior to, during, and after completion of
approximately 40 hours of reading training
distributed over a four and a half month period
(Figure 4-3). That the brief treatment program
Figure 4-3. Reading test scores achieved after 8/2 years of
regular classroom instructions and after 4K months
during which reading responses were positively
reinforced. Staats & Butterfield, 1965.
had produced generalized educational and
behavioral effects is indicated by the fact that the
boy received undistinguished but passing grades

in all subjects for the first time in his turbulent
academic career, and he markedly decreased and
eventually ceased his aggressively defiant
behavior. The entire program, which was
administered by a probation officer, involved a

total expenditure of $20.31 for token exchange
items.
Essentially the same procedures were applied
with some degree of success by adult volunteers

and high school seniors in teaching reading skills
to retarded, emotionally disturbed, and culturally
deprived children (Staats et al., 1967). The

children acquired reading responses at an
accelerating rate even though the instructional
material was increasing in difficulty and the
amount of extrinsic reinforcement was
progressively reduced. However, several
methodological problems associated with the
criterion tasks and the fact that control subjects

received special training with similar materials
precluded a stringent test of whether a program
involving tight contingencies and material
reinforcers yields better outcomes than the
standard educational practices.
Self-instructional Systems. The acquisition of

basic cognitive skills and knowledge, prerequisite
for higher symbolic activities, requires repeated
presentation of substantial amounts of abstract

content and principles, as well as intricate
discrimination training. Since many of these
functions can be performed more efficiently by

programmed self-instructional techniques than by
conventional training methods, increasing use is
made of semi-automated instructional systems for
symbolic learning. In evaluating the role of these
approaches in the educational process, it should
be emphasized that the critical issue is not
reliance upon mechanical versus social

presentation of stimulus material, but rather
which tutorial systems, applied either singly or in
combination, best approximate optimal conditions
for learning. Although social commentators often
attribute legions of virtues to conventional modes

of instruction and hosts of pernicious effects to
programmed methods, many instructors do not, in
fact, provide the type of content organization that
would ensure rapid learning and effective
retention; many present material in ways that

extinguish students’ intellectual interests; and
often many inadvertently establish strong

avoidance tendencies toward the subject matter
being taught. As a consequence, many students,
particularly those who are weakly motivated or

less well endowed intellectually, display marked
intellectual deficits despite numerous years
expended fruitlessly in school attendance.
Provided that they are skillfully designed and
adaptive to individual requirements, self-
instructional systems possess several features that

can facilitate the learning process. First, they
present material to the student in a well-organized
graduated order. The utilization of logically

ordered sequences prevents students from
becoming contused or lost through omission of
essential intermediate steps in exposition; this
removes one major aversive aspect of
conventional instruction. Second, they provide the

student with immediate feedback about the
accuracy of his responses, helping him to
continuously monitor his comprehension of the
subject matter. Third, since a student can proceed
to new information only by making correct
responses to preceding items, the required active
participation of the student forces careful
observation of stimulus material. Thus, if a student
should lapse into classroom reverie, the

instructional content, like Old Man River, keeps
rolling along, whereas in programmed instruction
the patient tutor remains idle as long as the
student is disengaged. Fourth, the self-pacing
feature of programmed teaching methods makes

individualized instruction possible for persons
who differ in ability and mastery of the material.
In computerized systems, in which new
instructional content is selected at each step on
the basis of the learner’s past performances,

students can generate their own optimal learning
sequences. Finally, because errors are drastically

reduced by gradual progression in content
difficulty, learning from self-instructional
programs is minimally threatening. The self-

pacing and nonthreatening characteristics are
particularly important in clinical and remedial
applications to persons who have undergone
extensive failure experiences and who differ

widely in the content areas in which they show
deficits.
In an effort to assuage fears that machines may
displace warm-blooded pedagogues, it has been
customary to relegate the teaching of hackneyed,

factual material to programmed instructional
devices, reserving creative, problem-solving, and
conceptual skills to teachers who would be
liberated from mundane functions. As Resnick
(1963) cogently points out, the difficulty in

teaching complex intellectual skills results
primarily not from inherent limitations in
programmed instruction procedures but from the
fact that cognitive activities are usually described
in very general terms or remain essentially
undefined. For this reason, even talented teachers
are often at a loss in deciding what types of
learning experiences might be most appropriate to
the development of abstract abilities. After

component behaviors of more complex skills have
been adequately specified there is no reason why
they cannot be taught by carefully prepared
learning sequences. Indeed, there is some
evidence to suggest that self-instructional

methods can be utilized effectively for teaching
relatively complex intellectual skills such as
language functions, mathematical reasoning,
decision making, abstract thinking, problem-
solving strategies, a wide range of vocational and

avocational skills, and the basic concepts and
principles of diverse fields of study. Moreover,

numerous comparative investigations (Silberman,
1962; Stolurow, 1963) have generally shown
programmed instruction to be at least as effective

as, and sometimes better than, conventional
teaching methods while also substantially
reducing student time, cost, and professional
personnel. Since the development of favorable

attitudes toward subject matters is as important a
part of the learning process as the acquisition of

specific intellectual skills (Mager, 1968), the
assessment of tutorial systems should measure
both attitudes and achievements. Unfortunately,
the attitudinal effects of programmed instruction,

or conventional teaching for that matter, are
rarely taken into account.
With further advances in auto-instruction, both
technologically (in the areas of picture projecting

systems, vocal programming, and the use of
computers to permit students more complete
control over learning sequences), and in
knowledge of acquisition processes, it should be
possible to arrange optimal learning conditions
more readily and to extend programmed
instruction to even more complex forms of
symbolic behavior. Indeed, in an elaborate study
of computer-assisted instruction conducted by

Atkinson and Suppes (Atkinson, 1968) first grade
students receive their entire reading and
mathematics instruction through television
learning units controlled by a central process
computer. The computer presents the material,

monitors the students’ performances, and
continuously adjusts the learning sequence to the
capabilities and individual rate of achievement of
each pupil. Instruction provided by this system
produces greater proficiency than does the regular

classroom method of teaching. If costs can be
reduced, computer-based systems of education

may eventually replace conventional techniques of
instruction in many areas of study. Such
instructional systems could, in fact, furnish varied

forms of educational material at different levels
under optimal learning conditions to homes,
schools, commercial facilities, and any other
settings equipped with television learning units.
Progress in this area has been hampered to
some extent by the identification of programmed

learning with audio-visual aids. As a consequence,
research activities have been unduly preoccupied
with comparisons of conventional instructions and

minor variations of program characteristics,
rather than undertaking systematic investigations
of acquisition processes associated with self-
instructional systems. The latter type of research
would not only increase the effectiveness of

programmed instruction, but would also elucidate
fundamental learning processes involved in
complex skills. For example, the teaching of
reading through an automated matching-to-
sample method, applied to young children over an
extended period of time, should provide basic
information about discriminative processes to
supplement knowledge derived from brief
laboratory studies employing identical procedures

with less meaningful content.
Although ample attention is currently being
devoted to hardware and program variables, the
incentive requirements for learning have been
essentially ignored. This oversight results in large

part from widespread assumptions that the
programmed format is effective in holding
attention and that informative feedback about the
accuracy of one’s responses functions as an

automatic positive reinforcer in sustaining
responsiveness. This may be true for persons who
have learned to value intellectual achievements, or

who expect to derive some immediate benefit
from increasing their competency in specific areas
of functioning. However, for highly motivated
bright students, small step, linear programming
involving strongly prompted responses, which
ensures successful learning in less talented
individuals, may provide such trivial increments in

performance that the learning task becomes
neither challenging nor personally rewarding.
While these negative effects can be avoided to
some extent by adjusting the size of informational
units to ability level, the incentive problem grows

far more serious in the case of persons for whom
signs of intellectual achievement must be
established as effective rewards. In these
instances an extrinsic incentive system must be
added to self- instructional procedures if students’

interest and responsiveness are to be durably
maintained.
SELF-MANAGED BEHAVIORAL CHANGE
Most of the programs that have been discussed
so far achieve behavioral changes primarily
through external management of reinforcement
contingencies. Recent years have witnessed a
growing interest in self- control processes in

which individuals regulate their own behavior by
arranging appropriate contingencies for
themselves. These self-directed endeavors
comprise a variety of strategies, some of which

were originally proposed by Ferster, Nurnberger
& Levitt (1962).
Efforts at self-influence usually prove
unsuccessful because they involve vague self-
instructions that have no immediate behavioral
implications. Moreover, unless self-directives are
supported by reinforcement operations, either

external or self-administered, they are unlikely to
exert much control over behavior. The selection of

well-defined objectives, both intermediate and
ultimate, is an essential aspect of any self-directed

program of change. The goals that individuals
choose for themselves must be specified in
sufficiently detailed behavioral terms to provide
adequate guidance for the actions that must be

taken daily to attain desired outcomes.
To further increase goal commitment
participants are asked to make contractual
agreements to practice self-controlling behaviors
in their daily activities. Thus, for example, in

modifying smoking behavior (Tooley & Pratt,
1967) and obesity (Ferster, Nurnberger, & Levitt,
1962), clients agree to restrict increasingly, in
graduated steps, the times and places in which

they will engage in the undesired behavior. Under
conditions where individuals voluntarily commit
themselves to given courses of action, subsequent

tendencies to deviate are likely to be counteracted
by negative self-evaluations. Through this
mechanism, and anticipated social reactions of
others, contractual commitments reinforce
adherence to corrective practices.
Satisfactions derived from evident changes
help to sustain successful endeavors. Individuals

can, therefore, utilize objective records of
behavioral changes as an additional source of
reinforcement for their self-controlling behavior.
In studies of self-directive processes by Kolb,
Winter, & Berlew (1968) students used miniature

counters to keep an accurate record of the
frequency with which they displayed desired and
undesired behavior throughout each day. These
data were plotted graphically to provide a clear

picture of the behavioral improvements students
were accomplishing by their own efforts. Daily
feedback of this type not only serves a reinforcing

function but it also safeguards against irregular
and halfhearted implementation of self-prescribed
procedures. In a study designed to improve self-
instruction behavior, Fox (1966) found that
students who recorded their daily productivity
continued to work on assignments until they
exceeded their preceding performances, thus

ensuring continued improvement.
Since behavior is extensively under external
stimulus control, persons can regulate the
frequency with which they engage in certain
activities by altering stimulus conditions under

which the behavior customarily occurs.
Overeating, for example, will arise more often
when appetizing foods are prominently displayed
in frequented places in the household than if they

are stored out of sight and made less accessible.
Indeed, it has been shown that, compared to
individuals of normal weight, obese persons are

less responsive to internal hunger states
(Stunkard & Koch, 1964), whereas their eating
behavior is excessively dependent upon external
food-related stimuli (Schachter, 1967). Some
degree of self-control can thus be achieved by
judicious environmental arrangements that
reduce the external determinants of the behavior.

Conversely, the incidence of desired activities can
be increased by introducing appropriate stimuli
for them.
Behavior that provides immediate positive
reinforcement, such as eating, smoking, and

drinking, tends to be performed in diverse
situations and at varied times. Therefore, another
important aspect of self-managed change involves
progressive narrowing of stimulus control over

behavior. Continuing with the obesity illustration,
individuals are encouraged gradually to delimit
the circumstances under which they eat until

eventually their eating behavior is brought under
the control of a specific set of stimulus conditions.
This outcome is achieved by having clients commit
themselves to a graduated program in which they
refrain from eating in non-dining settings,
between regular mealtimes, and while engaging in
other activities such as watching television,

reading, or listening to the radio. An essentially
similar procedure is employed to increase effortful
behavior that is under weak situational control.
Thus in getting students to study productively a
specific desk and time are designated for study

and all potentially distracting stimuli are removed.
To preserve the desk’s stimulus value for study
behavior, whenever students find their thoughts
wandering or their interests lagging they are
instructed to leave the situation and turn to other

activities. In this way progressively longer periods
of concentrated study are achieved (Fox, 1966;

Goldiamond, 1965).
The foregoing procedures are primarily aimed
at instituting self-controlling behavior, but unless
positive consequences are also arranged the well-
intentioned practices are likely to be short-lived.
Self-controlling behavior is difficult to sustain
because it is associated, at least initially, with

relatively unfavorable conditions of
reinforcement. Prepotent activities typically
provide immediate positive reinforcement for the
individual, whereas their aversive consequences
are not experienced for some time. Conversely,

self-control measures usually produce immediate
unpleasant effects while the personal benefits are
considerably delayed. Self-reinforcing operations
are, therefore, employed to provide immediate
support for self-controlling behavior until the

benefits that eventually accrue take over the
reinforcing function.
The contingencies that individuals arrange for
themselves may involve different types of
reinforcing events. They are asked to select a
variety of activities that they find rewarding and
to make them contingent on the performance of
desired behavior. Viewing television, drinking
coffee, reading magazines, recreational breaks,

monetary rewards, or food treats may, for
example, be made conditional upon a certain
amount of study behavior. As was noted in the
introductory chapter, self-managed reinforcement
can serve an important behavior maintenance

function (Bandura & Perloff, 1967). In the case of
powerful appetitive behaviors, positive competing
activities may be engaged in, to aid self-control, at
times when the instigation to perform undesired
behavior is high. The disposition to perform

prepotent behavior can also be reduced by
generating immediate aversive consequences

either symbolically (Cautela, 1966; Homme, 1965),
or through the use of portable stimulation devices
(McGuire & Vallance, 1964). The manner in which

self-generated aversive effects have been utilized
to gain control over sexual perversions, chronic
alcoholism, and other types of addictive behavior
receives detailed consideration in Chapter 8.
As a final feature of self-directed change
programs, increases in desired behavior and

reductions in undesired behavior are attempted
gradually. In this way the incidence of experienced
discomforts is kept low, and steady progress

toward the eventual goal can be achieved.
The efficacy of self-directed approaches to
behavioral change is best illustrated in the

modification of obesity, which has proved
refractory to a variety of medical and
psychological procedures. Stunkard (1958)
succinctly describes the usual outcomes
associated with traditional treatments as follows:
“Most obese persons will not stay in treatment for
obesity. Of those who stay in treatment, most will
not lose weight and of those who lose weight, most
will regain it [p. 79].” Unlike approaches that focus
on caloric intake or inferential inner causes of

overeating, self- control programs attempt to
accomplish an enduring change in the pattern of
eating behavior by regulating stimulus conditions
and self-generated consequences for the behavior.

Stuart (1967) reports uniformly marked and
lasting weight reductions in eight obese women
who followed a program combining the various
elements discussed above (Figure 4-4). In a
controlled study Harris (1969) found that men

and women who were trained to use similar self-
control procedures lost weight and maintained the

loss, whereas a matched control group given
calorie charts and urged to reduce remained

obese. Both studies further reveal that this
approach is accompanied by low dropout rates

and no unpleasant emotional effects. The
Figure 4-4. Weight losses achieved by eight women using
self-control procedures. Stuart, 1967.
encouraging preliminary findings indicate that
self- control methods merit further systematic
investigations both as treatments in their own

right and as adjuncts to other procedures.
VERBAL CONDITIONING
The method of positive reinforcement has also
been widely applied in studies of verbal

conditioning. These investigations typically utilize
either a free-interview or discrimination-learning
situation in which an experimenter selectively
reinforces certain classes of verbal responses but
ignores all other verbalizations. The reinforcement

usually consists of nodding, smiling, repeating or
paraphrasing the interviewee’s remarks, or simple
verbal utterances with positive connotations. The
simplicity of the procedure and its similarity to
clinical interviews led to ready adoption of verbal-

conditioning paradigms for testing hypotheses
concerning psychotherapeutic interaction
processes. The overall results of innumerable
studies (Kanfer, 1968; Krasner, 1962; Salzinger,

1959) reveal that interviewers can exercise
substantial control over the content of subjects’
verbal behavior through selective responding. It
has been demonstrated, for example, that affective

expressions, positive or negative self-reference
statements, confiding, hostile, and affiliative
verbalizations, expressions of opinion or beliefs,
“hallucinatory” and “neurotic” responses, maternal
references, early childhood memories, and

common responses to word association stimuli
can be increased by minimal social reinforcement,

and decreased by withholding reactions conveying
interest or approval. Experiments designed to

isolate variables governing the extent of verbal
conditioning have shown that subjects’

responsiveness to social reinforcers is affected by
such factors as the characteristics of the
experimenter, the types of reinforcing events
employed, the content of responses selected for

modification, subjects’ interpretations of the
reinforcing stimuli, the personality characteristics

and emotional state of interviewees, and the
quality of the relationship between the
experimenter and his subjects.
The theoretical and practical significance of

experiments demonstrating that verbal content is
modifiable through reinforcement largely depends
upon the psychological events that one wishes to
explain or to modify. If one is interested in

elucidating the communication processes
associated with conversational forms of treatment,
then verbal conditioning procedures furnish a
laboratory analogue to the clinical interview,
provided that the situation bears some
resemblance to psychotherapy, the interviewer’s
reinforcing interventions are analogous to those
regularly employed by therapists, and the classes
of responses chosen for study exemplify

psychotherapists’ treatment concerns. In this
connection, verbal-conditioning studies strongly
indicate that the specific content of clients’
verbalizations, which is often assumed to reflect
intrapsychic processes, may be largely determined

by interviewers’ selective interest and attention.
This is corroborated by analyses of response-
reinforcement contingencies as they occur
naturally in psychotherapeutic interactions
(Bandura, Lipsher, & Miller, 1960; Goldman, 1961;

Murray, 1956; Truax, 1966; Winder, Ahmad,
Bandura, & Rau, 1962). Therapists’ positive

reinforcement of certain types of verbal responses
increases their occurrence, whereas clients avoid
discussing matters that produce less favorable

reactions. Considering that people often remain in
treatment for extended periods, and that the
therapist’s potency as a reinforcing agent is
enhanced by his high status and his client’s

emotional dependence upon him, it is not
surprising to find that clients frequently undergo

marked changes in verbal behavior.
People rarely enter psychotherapy and
willingly expend considerable time, money, and

effort merely to learn to talk differently. A
question of the utility of verbal conditioning as a
method of treatment therefore arises. Little
therapeutic importance can be attached to verbal-
conditioning procedures unless it is demonstrated

that verbal behaviors established in treatment
settings generalize to other persons in the natural
environment and, even more important, that
verbal changes influence nonverbal behaviors to a
significant extent. Several investigators have
found that changes in verbal responses display
some transfer to different situations (Ullmann,
Krasner, & Collins, 1961) and can affect nonverbal
responding (Lovaas, 1961, 1964); however, most

studies have failed to obtain generalization as
measured by tasks varying in similarity (Rogers,
1960; Tobias, 1960; Ullmann, Krasner, & Edinger,
1964; Williams, 1959).
The contradictory and weak generalization

effects noted above are not at all surprising when
one considers that experimental manipulations in
most conditioning studies barely suffice to
produce a conditioning effect, let alone stimulus or

response generalization. It is possible that greater
changes could be achieved through verbal
conditioning if interviewers continued the

treatment for longer periods and instituted
systematic programs of generalization training.
Generality is usually ensured by varying stimulus
configurations. This requires changing reinforcing
agents and modifying treatment conditions so that
both the social settings and the responses being
reinforced are increasingly similar to those

encountered in the natural environment. As noted
previously, therapists often choose to modify
verbal behavior in hospital or office settings rather
than to alter social behavior directly under natural
conditions, more for reasons of convenience than

therapeutic efficacy. It would be far more
meaningful and advantageous to effect desired
behavioral changes from the outset and to provide
clients with graduated performance tasks to carry
out in their social milieux. Such an approach

avoids the unnecessary problems associated with
circuitous treatment strategies that commence

with verbal conditioning, which must later be
supplemented by a series of procedures designed
to establish and to transfer social response

patterns to extratherapeutic situations. There are
occasions, of course, when change agents are faced
with the problem of reinstating verbal
communication in mute persons (Isaac, Thomas, &

Goldiamond, 1960; Salzinger, Feldman, Cowan, &
Salzinger, 1965; Sherman, 1965) or of modifying

delusional and other types of deviant
verbalizations (Ayllon & Haughton, 1964; Ayllon &
Michael, 1959; Richard & Dinoff, 1962). Verbal
conditioning may be an appropriate, though not

necessarily the most efficient, procedure for these
purposes.
In addition to the use of verbal conditioning as
a method for gaining understanding of interview

processes and as a treatment technique in its own
right, this approach has been employed to study
the influence of certain variables on learning
processes. Initial findings from verbal
conditioning experiments were accepted as
striking demonstrations of automatic, unconscious
learning. However, more detailed analyses of
conditioning performances as a function of
subjects’ awareness of reinforcement

contingencies raised fundamental questions about
what, in fact, is being learned in such experiments
(i.e., verbal responses or hypotheses about
contingencies). With conceptualization of verbal
conditioning as a hypothesis-testing rather than

an automatic response-strengthening process, the
focus of research interest shifted from traditional
learning variables to the role of awareness in the
learning process. Results of this line of research
and their implications for theories of behavioral

change will be reviewed later.
Social Organizational Applications of

Reinforcement Contingencies
The discussion thus far has been primarily
concerned with the alteration of circumscribed
responses of single individuals through positive
reinforcement. In many cases change agents are
faced with the far more complex task of effecting
extensive modification in the attitudes and

behavior of large groups of individuals in
educational, rehabilitative, and other social
establishments. In order to accomplish
widespread behavioral changes it is necessary to
alter institutionally organized practices of the

larger social system in ways that will
simultaneously affect the behavior of each
member in beneficial ways. Some of the issues and
problems associated with group-oriented
contingency systems are best illustrated in

treatment applications of reinforcement
procedures to institutionalized populations.
Over the years there have appeared numerous
sociological studies of the psychiatric hospital as a
social system (Dunham & Weinberg, 1960;
Goffman, 1961; Stanton & Schwartz, 1954;
Wessen, 1964), each of which documents the
debilitating effects that prevailing institutional
practices have upon inmate populations. Because

of the large numbers of patients who must be
cared for with limited staff resources and facilities,
most institutions, regardless of their avowed
objectives, are primarily concerned with the
management and social control of patients, rather

than with, their rehabilitation. In order to
maintain efficiency and economy, the institution
must effect certain changes in patients’ behavior
that are incompatible with achievement of
effective social functioning and often prove more

deleterious than the behavioral problems that
originally led to the patients’ hospitalization.
The initial socialization of inmates usually
involves some degree of suppression of
individualized modes of behavior. On admission to
the institution, people are characteristically
deprived of most of their personal possessions,
their civil rights, their social status, their
accustomed satisfactions, their privacy, and their

individuality, so that they can be handled
expeditiously in large groups. Throughout the
period of institutionalization, the patients’
behavior is closely regulated and accommodated
to fixed hospital routines. Under these types of

organizational contingencies, initiative, self-
reliance, and self-determination, which are
necessary for attaining satisfactory independent
adjustment outside the hospital, are generally
extinguished, whereas the more docile patient-

role behaviors bring about the greatest rewards
and promotion in graded ward systems. Moreover,

whatever chores the patients may be assigned
contribute primarily to hospital maintenance
rather than to the further development of

occupational skills. With prolonged withdrawal of
the customary incentives for sustaining complex
behavioral repertoires, patients generally display
a progressive loss of social and vocational

competence, which renders them even less
capable of managing their lives on the outside.
In addition to reinforcement of institutional
dependence and behavioral losses through
deficient incentives, the gradual abandonment of

patients by their relatives, their stigmatization as
mentally diseased, and their loss of contact with
persons and contemporary events outside the
hospital further contribute to chronicity. Most
patients who are subjected to traditional

contingency patterns in psychiatric facilities over
a period of several years become permanently
resigned either to a simple regimented
institutional life or to a pattern of “intermittent
patienthood” (Friedman, von Mering, & Hinko,
1966). Although, in exchange for self-
determination, hospital residency provides more
physical comforts and fewer taxing demands than
patients with limited personal resources can attain

within their own social environments, it is evident
that they are rarely overjoyed or contented with
their sheltered existence. Major changes in
reinforcement contingency patterns at the social
system level are therefore required if institutional

establishments are to serve an important
rehabilitative function.
The recent years have witnessed increasing
use of contingent reinforcement on a group basis.

These procedures have been systematically
applied, for example, to hospital wards for
severely debilitated psychotics (Atthowe &

Krasner, 1968; Ayllon & Azrin, 1965) and
alcoholics (Narrol, 1967), in social-educational
programs for retarded children (Bijou, 1965;
Girardeau & Spradlin, 1964), in remedial academic
programs for school dropouts and low achievers
(Clark, Lackowicz, & Wolf, 1968; Wolf, Giles, &
Hall, 1968), in educational systems for managing

behavior disorders (O’Leary & Becker, 1967) and
fostering productive classroom behavior (Hall,
Panyan, Rabon & Broden, 1968), and in
rehabilitative institutions for delinquent
adolescents (Cohen, 1968). Unlike traditional

treatment systems, these programs contain work-
payment incentive systems and contingency
structures that are highly compatible with those in
the larger society.
There are three main characteristics of group-

oriented reinforcement practices as they are being
applied to populations that require, at least

initially, the use of extrinsic incentives. First,
behaviors essential for effective day-to-day
functioning (e.g., self-management, educational
accomplishments, appropriate social behavior,
satisfactory work performance, etc.) are
designated by the staff as reinforceable responses.
Second, a simulated currency, which is

exchangeable for a variety of desired objects,
activities, and special privileges, can be earned by
performance of the appropriate behaviors. Third,
an exchange system is devised in which a specified
number of points or tokens is required for the

purchase of various objects and privileges, similar
to monetary transactions in the outside
community.
The powerful control exercised by group

reinforcement procedures over the behavior of an
entire psychiatric ward population is best
exemplified by a series of experiments (Ayllon &

Azrin, 1965) in which organizational
reinforcement contingencies were systematically
varied. In each of the studies, tokens earned could
be used to secure, among other things, privacy
(e.g., choice of bedroom, choice of eating group,
selection of personal furniture and a room
divider), freedom to leave the ward and hospital

grounds, private audience with members of the
hospital staff, recreational opportunities (e.g.,
movies, band concerts, social functions, exclusive
rental of a radio or a television set), and a varied
array of commissary articles that could be

obtained by special request.
One experiment, conducted with a small group
of chronic female patients, studied the influence of
reinforcement contingencies on patients’

performance of off-ward jobs that were staffed by
paid hospital personnel. In the initial phase each
patient selected her preferred work assignment

and received 70 tokens for each full 6-hour day. In
order to ascertain whether job selection was
indeed determined by extrinsic incentives or by
social and intrinsic rewards deriving from the
vocational activity itself, in the second phase of the
study patients were no longer paid for
participating in their preferred jobs, but tokens

could be earned by work in non-preferred
assignments. During the third phase, the original
reinforcement for the preferred job was
reinstated.
In marked contrast with the usually erratic and

lethargic work performance of patients prior to
the incentive program, under the simulated
economy system all patients reported promptly
and regularly to work without complaint, even

though they were free to take time off from their
jobs whenever they wished. That the contingency
structure was highly influential in regulating the

behavior of the group is further shown by the fact
that all but one of the patients changed their work
assignments immediately when reinforcement
was shifted from preferred to non-preferred jobs
(Figure 4-5). When informed that the people with
whom they had been working were very pleased
with their performance and would like them to

continue, but in order to give other patients a turn
at the job, no tokens would be available, several
patients explained to ward attendants: “No, honey,
I cannot work at the laundry for nothing. I’ll work
at the lab. I just couldn’t make it to pay my rent, if I

didn’t get paid.”…“You mean if I work at the lab I
won’t get paid? I need tokens to buy cigarettes for
my boy friend and to buy new clothes so I’ll look
nice like the other girls [pp. 363-365].”
In a subsequent experiment, when patients

were paid the tokens on a noncontingent basis
each morning rather than at completion of the

day’s assignment, they all ceased working within a
week. On the other hand, when reinforcement was
again made conditional upon work performance,
the patients immediately resumed their job
Figure 4-5. Mean number of hours patients worked per day
when positive reinforcement was varied between
preferred and nonpreferred jobs. Ayllon & Azrin, 1965.
assignments. These marked changes in
performance are particularly interesting in view of
patients’ reports that they worked to keep active,

because of enjoyable social relationships, personal
satisfactions deriving from the work itself, and
other intrinsic gratifications. Obviously in this case
self-reports were unreliable indicants of the actual
conditions maintaining their behavior.
The manner in which patients spent their

currency is also highly informative. It was used
mainly to secure privacy (45%), commissary items
(34%), and ward leaves (21%), whereas virtually

no tokens were expended for private audiences
with hospital staff (.001%), for religious services
(.0002%), or recreational activities (.0008%).
The contingency system was found to be
equally effective in maintaining active patient
participation in varied on-ward activities
including complex duties as dietitians, secretarial
assistants, waitresses, and sales clerks in the
commissary, in janitorial, ward cleaning, and

laundry work, assisting in recreational programs
and personal care of other patients, and special
services. As shown in Figure 4-6, when rewards
were dependent upon successful completion of

performances the patients worked industriously;
when they were simply given the tokens
noncontingently they gradually stopped working;
and when contingent reinforcement procedures
were reinstated their participation was restored

immediately and maintained at the previously
high level.
It is of particular interest that when the
incentives were completely withdrawn and the

institutional rewards and privileges were made
freely available in a manner similar to usual

hospital practices, a marked loss in behavior
Figure 4-6. Total number of hours a group of 44
schizophrenic patients participated in rehabilitative
activities when rewards were conditional upon
successful completion of assignments and when the
same rewards were provided regardless of whether or
not the patients took part in the activities. Ayllon &
Azrin, 1965.
resulted (Figure 4-7).
The rapid behavioral changes produced by
alternating incentive conditions does not

necessarily demonstrate that reinforcement is the
sole controlling factor. In social situations
behavior always remains partly under modeling
stimulus control. Persons who occupy a
prestigeful position in a social group usually serve

as major sources of social behavior for other
members. Consequently, to the extent that the
incentive conditions modified the behavior of
prestigious models, other patients may have

emulated their actions independently of the token
rewards. As Schwartz & Hawkins (1965) have
shown, substantially different group behavior can

occur under the same reinforcement conditions
depending upon the behavior of influential
models. The ubiquitous influence of modeling cues
complicates interpretation of intragroup
Figure 4-7. Total number of hours spent each day by the
group of 44 patients performing “on-ward” activities
during periods when rewards were given upon
completion of work assignments, when positive
incentives were not used and the various activities and
privileges were freely available, and when the
reinforcement contingencies were reintroduced into
the social system. Ayllon & Azrin, 1965.
replicative control through manipulation of
reinforcement variables. An additional
complication was unfortunately introduced in

several of the studies by designating the
noncontingent period as a “vacation with pay,”
which carries a strong suggestion that work might
be temporarily discontinued. The changes
observed were, therefore, probably a compound

function of incentive conditions, modeling
influences, and instructional sets.
The successive reversals of work performances
consistently achieved through systematic

variation of contingency structures provide
convincing experimental demonstrations that
organizational reinforcement practices can

determine the degree to which persons will
exhibit initiative, involvement, and active
participation in available activities. These
principles and procedures have been applied on
an extended basis with deliberate therapeutic
intent in various institutional settings.
Atthowe & Krasner (1968) established an

incentive program in an entire psychiatric ward
population of 86 chronic schizophrenic patients,
whose length of hospitalization varied from 4 to
49 years with a median institutional residency of
24 years. As a group, these patients had

maintained an apathetic, isolated, and almost
vegetative existence on the ward. They
participated minimally in the hospital routines,
and they often neglected to care for themselves or

to change their clothes unless aided by attendants.
Most had lost complete contact with the outside
community and were essentially resigned to a

permanent institutional residency.
In an effort to reinstate interpersonal and self-
directing behavior so that the patients could
function more effectively, a token system involving
most phases of ward and hospital life was
adopted. The patients were informed that, in the

future, certain privileges, recreational activities,
passes, money allowances, and numerous other
rewarding events could be purchased with tokens.
The patients were provided with many

opportunities for earning tokens through
appropriate social behavior. In addition to self-
care, responsibleness, and housekeeping
behaviors, the patients could earn tokens for
participating in various social and vocational

activities. Each patient was rewarded immediately
after successful completion of some desired

activity. A system of negative sanctions and token
fines was established to control seriously

disruptive behaviors and theft of tokens. The latter
problem attested to the fact that the range of

behaviors created by the simulated economy was,
indeed, analogous to that occurring in the larger
outside community.
Those who functioned more independently
and could assume full responsibility for their work
assignments were compensated more generously.

However, they were paid the tokens only once a
week in order to teach them to budget their
expenditures. By utilizing banking and budget

management systems patients were further
prepared for successful community life. Special
token bonuses were also offered if patients
received satisfactory ratings in their industrial
training and made worthy contributions to their

social group.
The ward incentive program mirrored societal
practices in still another respect. Patients who
were able to function productively on their jobs,
who had a realistic discharge plan, and who had
sufficient token savings for a substantial
membership fee could, if vacancies existed, join a
privileged group on the ward which freed them

from most of the institutional constraints.
Members received a credit card that permitted
them considerable freedom to regulate their own
behavior. It entitled them, not only to all the

privileges within the token system, but additional
advantages as well. They were provided, for
example, choice dining and sleeping
arrangements, extra passes on weekdays and
unlimited pass privileges on weekends, and they

could bank their money in the community without
any restrictions on withdrawal.
The efficacy of this treatment program was
evaluated in terms of several indices including

both changes in the specific behaviors that were
involved in the reinforcement contingencies and

more generalized outcomes. Consistent with the
results of Ayllon & Azrin (1965), it was found that
the target behaviors were restored quickly and
dramatically after introduction of incentives. This

is shown most clearly by the marked decrease in
the frequency with which attendants had to
awaken patients, assist them with personal
grooming, make their beds, and prepare them for
daily activities. The number of morning infractions

dropped from a baseline rate of approximately 75
per week prior to reinforcement to about 9 after
several months of treatment. The patients likewise
displayed increased participation in group
activities, which were also specifically reinforced

performances. Some additional evidence that the
incentive system was, in fact, a significant

contributor to the observed changes is shown in
patients’ increased social participation when they
were rewarded more generously, whereas

participation in group activities declined
somewhat when the token rate was later reduced
to its original level.
Among generalized changes observed were the
patients’ display of substantially more social
communication as measured by behavioral

ratings, and greater interest in the outside
community. This was reflected in increased use of
passes, from an average of 9 per week before

treatment to 37 per week after the incentive
program had been instituted. Indeed, 26 percent
of the patients left the hospital on daytime or
overnight passes for the first time in many years,
and one patient, for whom the hospital had

become a permanent abode, ventured out for the
first time in 43 years!
Discharge rates also verified the beneficial
effects of the new reinforcement practices.
Twenty-one patients left the hospital via the
privileged group, almost double the discharge rate
for the same ward the previous year. The overall
findings of this project indicate that alteration of

contingency structures in a hospital social system
can not only counteract the stupefying effects of
prolonged institutionalization, but also produce
generalized increases in self-directive and

interpersonal modes of behavior.
A comprehensive treatment program does not

simply aim to produce effective functioning in
circumscribed areas within the institution, but
rather to establish the requisite social and
vocational competencies for successful self-
management in the larger society.

Institutionalized patients who possess marketable
skills and who have a reasonably adequate

environment to which to return can undoubtedly
benefit greatly from programs that provide

adequate guidance and reinforcing support for
adaptive patterns of behavior. Favorable outcomes
can be further ensured by the use of transitional
facilities embodying contingency systems that

foster gainful employment and progressively more
active participation in community life. However,
chronically hospitalized patients, who are grossly
lacking in personal and social resources, present
the most challenging rehabilitation problem.

Results based on follow-up studies disclose that
approximately 70 percent of chronic patients who
are discharged from mental hospitals return
within 18 months regardless of the type of
treatment received during the period of

hospitalization (Fairweather, Simon, Gebhard,
Weingarten, Holland, Sanders, Stone, & Reahl,

1960; Fairweather & Simon, 1963). As a
consequence of this high readmission rate,
hospitals are faced with a continuously growing

population of chronic patients. Moreover, for
reasons given earlier, most of these patients are
destined for a permanent institutional residency.
It is generally acknowledged that unless the
social roles and competencies acquired by patients
in institutional settings approximate the behaviors

essential for meeting the more rigorous demands
of community living, it is highly unlikely that they
will be able to achieve a satisfactory independent

adjustment. With this in mind, Fairweather and his
coworkers (Fairweather, 1964) conducted an
elaborate field experiment to determine whether
chronically hospitalized patients can be restored
to the community by providing them with

opportunities to develop, within the hospital,
some of the problem-solving and self-management
behaviors required in the outside community.
The participants, predominantly schizophrenic
patients, were matched on the basis of age,
diagnosis, and length of hospitalization and
assigned randomly either to a conventional
hospital program or to patient-led problem-

solving groups. The treatment programs were
similar in all respects, with the single exception
that patients in the two groups took part in
different activities for two hourly sessions each

day. Those in the traditional treatment program
participated in individual work assignments and a
recreational hour during the time that their
counterparts engaged in group work assignment
and decision-making sessions.
In the task group condition an incentive
system was applied in which the participants

received increasing monetary and pass privilege
rewards, contingent on the development of four

levels of progressively complex social and self-
directive behavior. The responsibility for

evaluating and modifying the behavior of each
member, and for implementing the incentive
system, was delegated to the group. It met daily to
discuss the progress of individual members, their

problems, and constructive ways in which these
might be managed or modified. Any staff member
could be invited to furnish factual information
needed by the group to reach a reasonable
decision, but the staff refrained from

recommending what courses of action should be
taken.
Each week the group also met with the staff to
present their recommendations concerning each

individual member’s step-level and concomitant
money and passes for the following week, the
actions taken with respect to problem behavior,

and their evaluations of the progress, morale, and
functioning of their group. The staff then either
approved all recommendations made by the
group, approved some and disapproved others, or
rejected their recommendations. If warranted, the
entire group could be rewarded or penalized by
being raised or lowered one step- level, depending

upon the appropriateness of the group’s decision-
making behavior.
Thus, the social-role behaviors required of
both the patients and the staff differed markedly
in the two programs. In the conventional

treatment the hospital staff was primarily
responsible for regulating the patients’ daily
activities, for making decisions concerning money
allotments, passes and other types of privileges,

and for implementing disciplinary and remedial
courses of action. The patients, in turn, occupied
the usual subordinate, patient role. By contrast,

although the staff in the group decision-making
program could overrule or amend action-oriented
recommendations proposed by patients, staff
members functioned primarily as consultants and
resource persons. This social structure provided
patients considerable freedom and responsibility
to manage their own activities and to make

decisions that significantly affected each other’s
behavior.
In an effort to control for possible effects of
different staff characteristics on patients’ social
behavior, the two sets of staff members changed

wards halfway through the experiment. The
relative efficacy of the treatment approaches was
objectively assessed in terms of diverse criteria
including a multitude of behavior ratings,

sociometrically derived preferences, self-
evaluations, administrative indices, and attitude
questionnaires. Most of the behavioral

assessments were conducted throughout the 27
weeks that the experiment was in progress, while
others were obtained at the completion of the
study, and six months later.
The voluminous data from this ambitious, well-
executed field study demonstrate that the
program specifically designed to reinstate

interpersonal responsiveness and self-directive
behavior in patients yielded consistently superior

outcomes. Patients in the latter treatment
condition rapidly formed cohesive groups, in
which the members exhibited increasing mutual
interest, help, and responsibility. They organized

their own employment group, interviewed and
counseled patients, and assumed full
responsibility in locating employment for eligible

members. They also established informal
educational programs taught by group members

who possessed specialized skills or knowledge.
The contrasting ward climates are even more
strikingly revealed in patients’ specific attitudes
and social behaviors. Relative to the patients
receiving the traditional care and treatment, those
in the graded reward program showed
significantly greater interpersonal responsiveness,
more verbal communication, and less bizarre

behavior. Some of these comparative data are
summarized graphically in Figure 4-8. In their
sociometric and questionnaire responses, patients
in the incentive-group program viewed their
fellow members more positively; they regarded

their treatment as more difficult and demanding
but, at the same time, more beneficial; and they
displayed greater optimism about their eventual
discharge, future employment, and development
of close interpersonal relationships upon their

return to the community. These positive
expectations were largely confirmed by the follow-

up study. Compared to patients who received the
traditional treatment, those who participated in
the incentive-group program spent a shorter time

in the hospital, were more frequently gainfully
Figure 4-8. Behavioral changes displayed by patients
receiving either conventional hospital treatment or a
program designed to establish problem-solving and
self-management competencies, (a), percentage of
observations during which patients manifested
pathological behavior; (b), percentage of observations
during which patients engaged in social interactions
involving three or more persons; (c), total time the two
groups of patients remained silent during weekly ward
meetings; (d), mean number of patients participating in
weekly ward meeting discussions. Fairweather, 1964.
employed, met with friends more regularly, and
engaged in a higher level of verbal interaction with
others.
It is interesting to note that in the present
treatment program favorable generalized changes

were effected in patients’ behavior even though
the staff explicitly provided only a general set of
contingencies and relatively few concrete

incentives. The investigators report that the
monetary and pass privilege rewards were
essential in the early phase of treatment, but after
the patients established mutually rewarding
relationships with each other, pride in their

accomplishments, competitiveness with other
groups, and mutual social approval and
disapproval became the major reinforcing events
regulating their day-to-day behavior. These
findings suggest that patients can successfully
modify and sustain each other’s behavior through
mutual social reinforcement provided that
behavioral guidelines and incentives are
furnished, conditions are arranged that lead to the

formation of a cohesive group, and the
responsibility for change is primarily delegated to
the group. Since the behavioral demands that a
group places upon its members through common
agreement are less likely to be resisted than staff-

imposed contingencies, a group-mediated
treatment program may achieve a more natural
and higher degree of change in social behavior. In
a project evaluating the efficacy of a token
reinforcement system for treating predelinquent

boys, Phillips (1968) found that large token
penalties administered by the staff to the entire

group failed to reduce their undesired behavior,
whereas it was promptly and enduringly
eliminated when lesser fines were individually

levied by a group member who assumed
managerial responsibility for the troublesome
activities (Figure 4-9).
The success of group-managed contingency
systems still relies upon reinforcement processes,
except that interpersonal reinforcers are favored

over material rewards, and group members rather
than the staff serve as the main reinforcing agents.
Although Fairweather emphasizes the

autonomous functioning of the groups, and the
gradual emergence of group norms, it should be
noted that the staff members continuously
imposed specific contingencies on the groups,
both in their written communications describing

problem behaviors of individual members that the
group was expected to control, and in their
evaluative responses to the groups’ decisions. It is
therefore not surprising that each group evolved
patterns of reinforcement that fostered desired
behavioral changes. In the absence of adequate
Figure 4-9. Number of infractions occurring under two types
of contingency systems: (1) In the manager condition
one of the boys purchased a managership which
assigned to him the responsibility for selecting
individuals to perform the tasks and for paying or
fining them according to the quality of their work. The
manager received or lost monetary points according to
the number of tasks completed. (2) In the fines
condition the staff imposed penalties on the entire
group for failure to perform the required tasks. The
numbers under the arrows indicate the number of
monetary points the group was fined. Phillips, 1968.
staff monitoring of a group’s activities, particularly
in early stages of treatment, the types of
reinforcement contingencies that a given patient

group might adopt would be left to fortuitous
factors.
In view of the generalized changes in
interpersonal responsiveness achieved by
Fairweather, it would be of considerable interest

to compare systematically the relative efficacy of
staff-administered reinforcement systems
involving an elaborate set of precise contingencies
of the type employed by Ayllon, and Atthowe &

Krasner, with one in which some incentives are
used to foster strong group cohesiveness but
controlling functions are in large part delegated to

group members. The merits of these two
approaches, which involve many common
principles, could be easily combined to form a
program in which a specified set of reinforcement
contingencies is developed and implemented by
the patients themselves under staff guidance.
Another important contingency variable that

requires systematic investigation is concerned
with whether rewards are tied to individual
performances or to entire sets of behavior. In the
latter system, successive phases are devised which
require increasingly higher levels of functioning in

several different areas. As individuals progress
through these sequential steps by adopting the
requisite patterns of behavior they receive
increased rewards and privileges. In treating a

group of delinquent adolescents, Martin,
Burkholder, Rosenthal, Tharp, & Thorne (1968)
found that a phase-contingent system of

reinforcement produced more rapid and
uniformly positive changes in behavior than a
previous system in which specific responses were
individually reinforced. Indeed, the latter
contingency structure produced much wrangling
and accusations of unfairness because, in an effort
to ensure adequate reinforcement of progress

made by youngsters functioning at different levels,
they were required to meet different behavioral
standards and achievements for similar rewards.
The authors attribute the greater efficacy of the
reinforcement system linked to role behaviors to

the fact that clear specification of sequential goals
and the behaviors required for promotion from
one phase to the next serve as prompts and
positive guides for changes in desirable directions.
In most applications of reinforcement

principles to severely incapacitated persons,
behavioral improvements are initially achieved by

immediate reinforcement of specific
performances. However, as their competencies are
increased, individuals are promoted to a phase
system analogous to hierarchical reinforcement
structures existing in community life.
Despite the favorable changes in behavior
produced by the incentive group program

developed by Fairweather, readmission rates for
patients who had received the different forms of
treatment were essentially the same. Eighty
percent of the short-term psychotics maintained
an adequate community adjustment, whereas only

45 percent of those who had been institutionalized
for two or more years remained outside the
hospital by the end of six months. These outcome
data provide further justification for questions

that are increasingly raised about the wisdom of
conducting behavioral change programs within
hospital settings. The behavioral requirements for

effective hospital and community adjustment
differ in so many fundamental respects that, even
though institutional reinforcement practices may
establish and strengthen some of the response
patterns consistent with those in the larger
society, most of the social and vocational
behaviors that can be developed in hospitals have

limited transfer value. More radical departures
from conventional institutional approaches are
clearly required if chronically hospitalized
psychotics are to be restored to society as socially
productive members.
A second experimental project conducted by

Fairweather and his associates (Fairweather,
Sanders, Maynard, & Cressler, 1969) represents
one example of an innovative program that holds

considerable promise for the rehabilitation of
chronic psychotics. Based on evidence that the
type of social environment to which patients

return is the major determinant of successful
community adjustment, Fairweather decided to
institute the social subsystem that proved highly
successful in the hospital in a community
residence where behavioral requirements are
essentially identical to those of any member
residing in the community. One group of chronic

patients was formed in the hospital and, after they
achieved a sufficient degree of organization and
cohesiveness within the hospital and adequate
training in decision-making behavior to function
effectively, the group was transferred to a lodge

located in the community. There the members
were responsible, as a unit, for regulating each
other’s behavior, managing the affairs of the lodge,
including the purchase and preparation of food,
keeping records of expenditures and personal

loans from their own savings bank, and, when
necessary, obtaining information about drug

dosages from a local physician as well as
administering the necessary medication. In
addition, the patients operated an income-

producing business (a janitorial service, both
commercial and residential, supplemented by
yardwork, general hauling, and painting) in the
community. They assumed major responsibility

for receiving and processing work orders,
arranging transportation to and from work,

assignment of patients to specific jobs appropriate
to their level of functioning, and management of
the monetary incentive system. Initially a research
staff member coordinated the daily operations of

the lodge, but later this function was performed
even more successfully by a lay person with
periodic consultation provided by a member of the

research staff. The primary function of the
consultation was to supply needed information, to

assess the group’s functioning, to review any
personal or organizational problems that arose,
and to evaluate the group’s recommended

solutions. The income derived from the business,
which amounted to a total of $52,000 over a
period of two and three-fourths years, was
distributed weekly among the participants with
each member’s share being determined by his

productivity and the amount of responsibility that
he assumed in the organization.
In order to evaluate the relative efficacy of this
community-based program, an individually
matched group of 75 patients received the group

decision-making treatment in the hospital along
with the traditional types of assistance and out-
patient therapy following discharge from the
institution. Repeated measurements were made of
changes in patients’ self-evaluations, attitudes,

interpersonal behavior, and ability to maintain a
satisfactory independent adjustment in the
community.
Results of this project show that, within a
supportive subcommunity, chronic marginal
individuals can manage their daily affairs and lead
a worthwhile and constructive life. A forty-month
follow-up study revealed that the experimentally

created subsystem sustained patients in the
community, whereas those treated in the hospital
were unable to adjust to life on the outside (Figure
4-10). The differences between these groups are

even more striking in their vocational functioning:
the lodge system enabled patients to maintain
gainful employment, while none of the patients
who received treatment within the hospital setting
were employed full time (Figure 4-11). These

beneficial results were obtained at an individual
cost of $6 per day, as compared to $14 at the

hospital from which the participants were drawn,
$12 at a local state hospital, $45 at a local private

hospital, and $56 at a local county hospital.
The main purpose of the type of residential

program discussed above is to create a semi-
Figure 4-10. Percentage of time that patients in the lodge
and hospital programs spent in the community for 40
months of follow-up. The lodge program was
discontinued after 33 months. Fairweather et al., 1969.
Figure 4-11. Percentage of time that patients in the lodge
and hospital programs were employed full time for 40
months of follow-up. Fairweather et al., 1969.
autonomous subcommunity in which marginal
individuals can function in a socially productive
manner, rather than to serve as a transitional

facility. Some of the participants may eventually
leave after they have developed the social and
vocational competencies required to live
independently. However, those who have no
immediate families to return to or little or no

financial resources, and who are unlikely to gain
employment individually in the open job market
because of their advanced age, limited vocational
skills, and stigmatization, may achieve the most
satisfactory adjustment in their own supportive

social system. Individually they may be unable to
meet the demands of community life, but as a

group they can function as self-supporting and
productive individuals who would otherwise be
institutionalized for the rest of their lives.
GROUP REINFORCEMENT SYSTEMS IN THE
MODIFICATION OF DELINQUENCY
In a program devised by Cohen (1968; Cohen,

Filipczak, Bis, Cohen, Goldiamond, & Larkin, 1968)
for overcoming educational deficits in adolescent
delinquents, the institutional reinforcement
practices are, in many respects, also analogous to

the contingency structures that obtain in the
larger community.
The adoption of, and adherence to, prosocial

patterns of behavior is highly dependent upon
adequate proficiency in educational and
vocational skills necessary for legitimate

acquisition of resources that are highly valued by
the culture. Most delinquents are handicapped by
gross deficits in the educational abilities requisite

for satisfactory vocational adjustment. As a
consequence, those who value costly possessions

and the social rewards attendant upon symbols of
high status are forced to resort to deviant means
for obtaining them. Regardless of whatever other
objectives may be selected in the rehabilitation of
delinquents, little progress can be made in altering

antisocial behavior unless habitual offenders are
provided with vocational skills that will permit
legitimate attainment of desired social and
material rewards. Therefore, the experimental

program devised by Cohen focused primarily upon
the creation of educational competencies and
favorable attitudes toward academic pursuits.
The adolescent boys, who were randomly
selected for the experimental program from the
institutional population, presented exceedingly
poor academic histories: All had dropped out of

school, they showed little interest in academic
matters, and none derived much intrinsic

satisfaction from intellectual activities. A system of
extrinsic rewards based on a point economy was

therefore employed to establish academic
behaviors. Points earned for high achievement
scores in programmed instructional courses and
other academic activities could purchase favorable

accommodations and a wide range of services and
commodities. Starting at a base pay of $10 to $15 a
week, payable on high scholastic achievement and
a minimum number of hours of academic work,
the boys could earn additional money points by

studying.
Most coercive sanctions imposed on
delinquents by correctional instructions were
absent from this program. Rather, heavy reliance

was placed on positive incentives, individual
initiative, and self-determination. Consistent with
the basic contingency system in daily life on the

outside, boys paid for their private rooms ($6 a
week, or 600 points) and selected meal plans that
varied in choice of foods. They used point currency
to rent recreational items and private offices, and
to pay tuition for requested courses. They could
also purchase admission to entertainment or
leisure-time activities provided at a project

lounge, and merchandise available in either the
project store or through mail order catalogues.
Whenever applicable, the prices of items
corresponded to the rates in the outside
community. Within this environment the boys

were provided with considerable freedom: They
determined their own study and bedtime routines,
they selected their own leisure-time activities and
planned their own outside programs, they aided in
planning menus, and they had open mailing and

visiting privileges.
A boy was free to choose not to participate in

any of the scholastic or rehabilitative activities,
but if his points fell below 1200 he was placed “on
relief.” Although this problem rarely arose, while
on relief status the boy lost his private room, was
served the regular institutional food, and enjoyed
no extra luxuries. This system of treatment
corresponds closely to principles suggested earlier

for altering behavior on a positive reinforcement
basis by creating an environment where
noncontingent gratifications are provided at an
adequate but low level, and in which the
performance of desired modes of behavior

produces further rewards. If such a system is well
managed, hostile-resistive actions by participants
should be markedly reduced because their own
behavior determines their welfare and not
management fiat.
Under environmental contingencies
specifically designed to support learning,

delinquent boys who had received few prior
accolades for scholarship and were school
dropouts worked productively at self-managed
educational activities. They studied
conscientiously in their spare time, and gained
more than two grade levels on standard
achievement tests within an eight-month period.

These favorable results suggest that reinforcing
environments for learning can be successfully
employed in the rehabilitation not only of
delinquents, but of other educationally
disadvantaged children as well.
The contingency system discussed above

aimed almost exclusively at educational behaviors
and, although beneficial changes in attitudes and
social behavior were noted, no concerted effort

was made to alter interpersonal response
patterns. The contingencies could, of course, be
easily extended to include cooperative,

responsible, and self-controlling behavior which,
supplemented with proficiency in a selected
occupation, would remove the major personal
barriers to a successful prosocial adjustment.
INTERDEPENDENT CONTINGENCY SYSTEMS
The preceding social systems primarily involve

group-wide contingencies, but reinforcements are
still administered on an individual basis. Work
payment arrangements, for example, are usually
the same for all members of the group, although

the actual compensation depends upon the type
and amount of work performed by each person.

Under such conditions, a member’s outcomes are
not affected by the behavior of others. One can, of
course, influence the degree of social interaction

between persons even on an individualistic
reinforcement basis simply by rewarding
cooperative responses (Cohen & Lindsley, 1964),
or other forms of social behavior.
On some occasions a change agent may be

called upon to increase the level of mutual
responsibility, cohesiveness, commitment, and

contribution to a common goal among all
members of a group. This objective can be
accomplished most effectively by instituting
reinforcement contingencies on a group basis.

Under these circumstances individual rewarding
outcomes depend upon the level of group
performance and, conversely, censurable behavior
by any given member may produce negative

consequences for the entire group. By having
people share in the consequences of their
decisions and actions, the degree of social
responsibility and involvement is thereby
increased.
There have been some laboratory
investigations of group productivity as a function

of different types of group contingency structures.
Glaser & Klaus (1966) found that group behavior

was influenced by reinforcement contingencies in
the same way as individual behavior when all

members were required to perform correctly to
produce rewarding outcomes for themselves.
Group output increased when correct team
performances were positively reinforced, whereas

the rate of correct responding declined when
group performance no longer resulted in
reinforcing consequences. On the other hand, a
contingency structure in which a correct response
by any member produced reinforcement for the

entire group often resulted in a deterioration of
performance for the group or individual members.
The latter condition is analogous to naturalistic
situations where individuals benefit from
someone else’s efforts. The decline of

responsiveness can be easily offset by the use of a
double contingency so that a given individual’s

outcomes are determined by both the extent of his
own contribution and the group’s overall
performance. It will be recalled that Fairweather’s

residential treatment program relies upon this
type of contingency system in the allocation of
monetary rewards.
Some suggestive evidence concerning the
relative efficacy of individual and group
contingency systems is provided by Wolf & Risley

(1967). They studied the amount of disruptive
classroom behavior displayed by a child in the
absence of any special reinforcement and during

subsequent periods when either she alone earned
five points, or she and her immediate peers each
earned one point for her commendable behavior.
It is interesting to note that the child’s activities
were more effectively controlled under the peer

contingency even though it produced only one-
fifth of the amount of reinforcement provided on
the individual basis. Apparently, through the
group reward, change agents were able to enlist
the peers’ aid in modifying the behavior of their
companion. The findings of the present case study,
and those cited earlier, are sufficiently interesting
to warrant further systematic exploration of the
effects of different types of group contingency

structures on social performance.
Group-oriented reinforcement practices have
been adopted on a society-wide basis in the Soviet
Union for the explicit purpose of developing
strong collectivistic morality in its citizenry

(Bronfenbrenner, 1962). This aim is implemented
by use of school collectives where children’s
behavior is regulated by rewards and
punishments administered on a group basis, so

that all members of a given social unit are affected
by the actions of each individual. Socialization at
the school is commenced in the primary grades by

assigning children to row units. Daily records are
kept of each group’s performance on a variety of
social and academic activities. The grades that a
person receives are based on the overall
performance of his unit rather than his individual
accomplishment, and, from time to time, the most
successful rows are publicly acclaimed, and given

special privileges. Collective achievement is
further reinforced by introducing competition not
only between rows, but also between classrooms,
between schools, between districts, and so on. In
this manner cooperative contingencies are

maintained within groups of expanding
membership, while simultaneously, strong
competitive contingencies are instituted between
progressively larger social units.
In addition to inculcating collective

orientations through group incentives, the control
and modification of behavior is delegated to the

peer group. Initially teachers set the standards of
behavior and evaluate the group’s performance.
As soon as possible, however, the responsibility
for evaluating the behavior of individuals and for
administering rewards and sanctions is turned
over to members of the social unit. This transfer is
accomplished by designating row monitors for

each activity to judge work assignments, to
criticize shortcomings of their comrades, and to
devise effective means for helping other members
of one’s collective. Whenever individuals are
praised for their mutual aid, the contribution to

group achievement is carefully noted.
Pervasive and unrelenting application of
group-oriented systems of reinforcement which
stifle autonomy and self-determination clearly are

antithetical to goals that are highly valued in most
societies. Therefore, where interdependent
contingencies are instituted to increase group

unity and responsibility, each member should also
be given opportunities for independent
accomplishment.
Summary
The present chapter reviewed theories of
reinforcement in terms of their relative emphasis
upon associative-strengthening or incentive

functions of reinforcers. Evidence bearing on
alternative explanations of reinforcement effects

in terms of drive reduction, sensory stimulation,
and opportunities to engage in prepotent
responses was also reviewed. The fact that
markedly diverse events can assume reinforcing

functions and their value is determined by
relational rather than fixed properties present
difficulties for the construction of an inclusive
theory of reinforcement.
Since behavior is largely governed by its

consequences, differential reinforcement has been
extensively applied, both singly and in
combination with other methods, to overcome

behavioral deficits, to maintain existing response
patterns, and to alter deviant behavior which is
supported by its rewarding effects.
Three sets of variables are importantly

involved in effective implementation of
reinforcement principles. First, it is essential to
devise an incentive system that is capable of

maintaining a high level of responsiveness over
long periods. Without adequate incentives
behavioral control is likely to remain weak and

unstable. Second, the reinforcing events must be
made conditional upon occurrence of the desired
behavior. Although reinforcement practices are
commonly used to modify behavior in natural
settings, these efforts are often minimally effective

because the intended consequences are poorly
timed, the wrong responses may be inadvertently

reinforced, and, even when appropriate
contingencies are arranged, they are often only

sporadically applied.
The third requirement concerns methods
powerful enough to induce or to elicit the desired
modes of response with sufficient frequency for

them to be strongly established through positive
reinforcement. Several different strategies can be

used for this purpose. The most popular approach,
though not necessarily the most efficient one,
relies upon a process of gradually shaping emitted
responses into desired patterns by selective

reinforcement of successively closer
approximations. However, in most cases complex
responses can be more rapidly created by the

provision of performance guides in the form of
appropriate verbal or behavioral modeling cues.

Another method, applicable when responses are
already available but rarely exhibited, depends
upon prompting and fading techniques in which

any stimuli known to exercise strong control over
the desired behavior are enlisted until its
incidence is sufficiently increased, after which the
ancillary stimulus supports are gradually
withdrawn.
The behavioral change process is further
complicated because additional learning

conditions must be arranged if established
response patterns are to generalize beyond the
specific treatment situation, and to persist long

after the specially created contingencies have been
discontinued. Enduring changes in behavior can
be achieved by gradually reducing the frequency
or magnitude of reinforcement; by shifting the
locus of reinforcement from transitory change

agents to favorable contingencies existing within
the person’s social milieu; and by altering the form
of the events that assume reinforcing functions.
Thus arbitrary extrinsic incentives, which may be
necessary during early stages of treatment, can be
gradually withdrawn and replaced by symbolic
cues signifying performance achievements
coupled with self-evaluative and other self-
reinforcing consequences. As such stimulus events

which are more intrinsically related to
performances acquire reinforcing capacity,
personally valued patterns of behavior can be self-
maintained with minimal external support.
Changes effected in social response patterns

can be widely generalized by conducting
treatment within natural social contexts, by
directly modifying the reinforcement practices of
persons who normally exercise some control over

the crucial behavior, and by varying systematically
the stimulus conditions under which the behavior
is established.
Results of myriad projects in which requisite
conditions are adequately arranged demonstrate
that contingent reinforcement can be a highly
effective means of establishing and modifying
diverse classes of response. This is most
impressively revealed by studies in which

tenacious deviant behavior is successively
eliminated and reinstated by varying its social
consequences. Such powerful replicative control
over behavior not only has considerable

therapeutic significance, but accents the influential
role played by environmental contingencies in the
regulation of behavior disorders: Conditions
which are usually attributed to complex internal
determinants presumed to be operating largely at

an unconscious level respond in orderly fashion to
external control.
In recent years there has been increased
recognition that most social problems must be

treated primarily at the social-systems rather than
at the individual level. Therefore, considerable

importance is attached to extensions of
reinforcement procedures demonstrating that
widespread changes in attitudes and behavior can
be achieved by applying organizational

contingencies to members of a given group.
Interdependent contingency systems, in which the
outcomes for individual members are dependent
upon the composite group performance rather
than their own attainments, have also been

successfully employed to increase the productivity
and level of mutual support, responsibility and
cohesiveness in social groups. Double
reinforcement contingencies in which individuals’
outcomes are jointly determined by the nature of

their own contributions and by the group’s overall
performance are likely to produce the most

socially productive functioning.
Although behavior can be effectively controlled
by varying its positive consequences, negative
sanctions are commonly employed, particularly in
efforts to reduce deleterious response patterns.
The processes involved in aversive control are
reviewed next.
REFERENCES
Allen, K. E., & Harris, F. R. Elimination of a child’s

excessive scratching by training the mother in
reinforcement procedures. Behaviour Research
and Therapy, 1966, 4, 79-84.
Allen, K. E, Hart, B, Buell, J. S., Harris, F. R., & Wolf, M.

M. Effects of social reinforcement on isolate
behavior of a nursery school child. Child
Development, 1964, 35, 511—518.
Allen, K. E., Henke, L. B., Harris, F. R., Baer, D. M., &

Reynolds, N. J. Control of hyperactivity by social
reinforcement of attending behavior. Journal of
Educational Psychology, 1967, 58, 231-237.
Atkinson, R. C. Computerized instruction and the

learning process. American Psychologist, 1968,
23, 225-239.
Atthowe, J. M., Jr., & Krasner, L. Preliminary report on

the application of contingent reinforcement
procedures (token economy) on a “chronic”
psychiatric ward. Journal of Abnormal Psychology,
1968, 73, 37-4.3.
Ayllon, T. Intensive treatment of psychotic behavior

by stimulus satiation and food reinforcement.

instructions with mental patients. Journal of the
331.
Ayllon, T., & Azrin, N. H. The measurement and

reinforcement of behavior of psychotics. Journal
357-383.
Ayllon, T., & Azrin, N. H. A motivating environment for

therapy and rehabilitation. New York: Appleton-
Ayllon, T., & Haughton, E. Modification of

symptomatic verbal behaviour of mental
patients. Behaviour Research and Therapy, 1964,
2, 87-97.
Ayllon, T., Haughton, E., & Osmond, H. O. Chronic

anorexia: A behaviour problem. Canadian
Psychiatric Association Journal, 1964, 9, 147-154.

Bachrach, A. J., Erwin, W. J. & Mohr, J. P. The control of

eating behavior in an anorexic by operant
conditioning techniques. In L. P. Ullmann & L.
Krasner (Eds.), Case studies in behavior
modification. New York: Holt, Rinehart &
Winston, 1965. pp. 153-163.
Baer, D. M., Peterson, R. F., & Sherman, J. A. The

development of imitation by reinforcing
behavioral similarity to a model. Journal of the
416.
Baer, D. M., & Wolf, M. M. The entry into natural

communities of reinforcement. Paper read at the
American Psychological Association meeting,
Washington, September, 1967.
Bandura, A., Lipsher, D. H., & Miller, P. E.

Psychotherapists’ approach- avoidance reactions
to patients’ expressions of hostility. Journal of
Bandura, A., & Perloff, B. The efficacy of self-

monitoring reinforcement systems. Journal of
Personality and Social Psychology, 1967, 7, 111—
116.
Barnes, G. W., & Baron, A. Stimulus complexity and

sensory reinforcement. Journal of Comparative
and Physiological Psychology, 1961, 54, 466-469.
Bensberg, G. J. (Ed.) Teaching the mentally retarded: A

handbook for ward personnel. Atlanta: Southern
Regional Education Board, 1965.
Bensberg, G. J., Colwell, C. N., & Cassel, R. H. Teaching

the profoundly retarded self-help activities by
behavior shaping techniques. American Journal of
Mental Deficiency, 1965, 69, 674-679.
Berlyne, D. E. Conflict, arousal, and curiosity. New

York: McGraw-Hill, 1960.
Bijou, S. W. Experimental studies of child behavior,

normal and deviant. In L. Krasner, & L. P. Ullmann
(Eds.), Research in behavior modification. New
York: Holt, Rinehart & Winston, 1965. pp. 56-81.
Brawley, E. R., Harris, F. R., Allen, K. E., Fleming, R. S.,

& Peterson, R. F. Behavior modification of an
autistic child. Behavioral Science, 1969, 14, 87-97.
Bronfenbrenner, U. Soviet methods of character

education: Some implications for research.
American Psychologist, 1962, 17, 550-564.
Brown, J. S. Comments on Professor Harlow’s paper.
In Current theory and research on motivation: A
symposium. Lincoln: University of Nebraska
Press, 1953. pp. 49-55.
Brown, J. S. Pleasure-seeking behavior and the drive-

reduction hypothesis. Psychological Review, 1955,
62, 169-179.
Buchwald, A. M. Supplementary report: Alteration in

the reinforcement value of a positive reinforcer.
416-418.
Butler, R. A. Incentive conditions which influence

visual exploration. Journal of Experimental
Psychology, 1954, 48, 19-2.3.
Butler, R. A. Exploratory and related behavior: A new

trend in animal research. Journal of Individual
Psychology, 1958, 14, 111-120. (a)
Butler, R. A. The differential effect of visual and

auditory incentives on the performance of
monkeys. American Journal of Psychology, 1958,
71, 591-593. (b)
Cautela, J. R. Treatment of compulsive behavior by

covert sensitization. Psychological Record, 1966,
16, 33-41.
Chambers, R. M. Effects of intravenous glucose
injections on learning, general activity, and
hunger drive. Journal of Comparative and
Clark, M., Lackowicz, I., & Wolf, M. A pilot basic

education program for school dropouts
incorporating a token reinforcement system.
188.
Cohen, D. J., & Lindsley, O. R. Catalysis of controlled

leadership in cooperation by human stimulation.
Journal of Child Psychology and Psychiatry, 1964,
5, 119-137.
Cohen, H. L. Educational therapy: The design of

learning environments. In J. M. Shlien (Ed.),
Research in psychotherapy. Washington, D. C.:
American Psychological Association, 1968. pp.
21-53.
Cohen, H. L., Filipczak, J., Bis, J., Cohen, J., Goldiamond,

I., & Larkin, P. Case II—Model: A contingency-
oriented 24-hour learning environment in a
juvenile correctional institution. Silver Spring,
Md.: Educational Facility Press, 1968.
Colman, A. D., & Baker, S. L., Jr. Utilization of an

operant conditioning model for the treatment of
character and behavior disorders in a military
setting. Unpublished manuscript, Walter Reed
Army Institute of Research, Washington, D. C.,
1968.
Coppock, H. W., & Chambers, R. M. Reinforcement of

position preference by automatic intravenous
injections of glucose. Journal of Comparative and
Coppock, H. W., Headlee, C. P., & Hood, W. R. Negative

reinforcing effects of insulin injections. American
Psychologist, 1953, 8, 337 (Abstract).
Dunham, H. W., & Weinberg, S. K. The culture of the

state mental hospital. Detroit: Wayne State
Dunham, P. J. Contrasted conditions of reinforcement:

A selective critique. Psychological Bulletin, 1968,
69,295-315.
Fairweather, G. W. Social psychology in treating

mental illness: An experimental approach. New
York: Wiley, 1964.
Fairweather, G. W., Sanders, D. H., Maynard, H., &

Cressler, D. L. Community life for the mentally ill:
An alternative to institutional care. Chicago:
Aldine, 1969.
Fairweather, G. W., & Simon, R. A further follow-up
comparison of psychotherapeutic programs.
Journal of Consulting Psychology, 1963, 27, 186.
Fairweather, G. W., Simon, R., Gebhard, M. E.,

Weingarten, E., Holland, J. I., Sanders, R., Stone, G.
B., & Reahl, J. E. Relative effectiveness of
psychotherapeutic programs: A multicriteria
comparison of four programs for three different
patient groups. Psychological Monographs, 1960,
74 (5, Whole No. 492).

control of eating. Journal of Mathetics, 1962, 1,
87-109.
Fox, L. Effecting the use of efficient study habits. In R.

Ulrich, T. Stachnik & J. Mabry (Eds.), Control of
human behavior. Glenview, Ill.: Scott, Foresman,
1966. pp. 85-93.
Fox, S. S. Self-maintained sensory input and sensory

deprivation in monkeys: A behavioral and
neuropharmacological study. Journal of
55, 438-444.
Friedman, I., von Mering, O., & Hinko, E. Intermittent

patienthood—the hospital career of today’s
mental patient. Archives of General Psychiatry,
1966, 14, 386-392.
Gardner, J. E. Behavior therapy treatment approach to

a psychogenic seizure case. Journal of Consulting
Psychology, 1967, 31, 209-212.
Giles, D. K., & Wolf, M. M. Toilet training

institutionalized, severe retardates: An applicant
of operant behavior modification techniques.
American Journal of Mental Deficiency, 1966, 70,
766-780.
Girardeau, F. L., & Spradlin, J. E. Token rewards in a

cottage program. Mental Retardation, 1964, 2,
345-351.
Glaser, R., & Klaus, D. J. A reinforcement analysis of

group performance. Psychological Monographs:
General & Applied, 1966, 80 (13).
Goffman, E. Asylums. Chicago: Aldine, 1961.
Goldiamond, I. Self-control procedures in personal

behavior problems. Psychological Reports, 1965,
17, 851-868.
Goldman, J. R. The relation of certain therapist

variables to the handling of psychotherapeutic
events. Unpublished doctoral dissertation,
Hall, R. V., Lund, D., & Jackson, D. Effects of teacher
attention on study behavior. Journal of Applied
Hall, R. V., Panyan, M., Rabon, D., & Broden, M. Teacher

applied contingencies and appropriate classroom
behavior. Paper presented at the meeting of the
American Psychological Association, San
Francisco, September, 1968.
Harlow, H. F. The formation of learning sets.

Harlow, H. F. Motivation as a factor in the acquisition

of new responses. In Current theory and research
in motivation: A symposium. Lincoln: University of
Nebraska Press, 1953. pp. 24-49.
Harris, F. R., Johnston, M. K., Kelley, C. S., & Wolf, M. M.

Effects of positive social reinforcement on
regressed crawling of a nursery school child.
Journal of Educational Psychology, 1964, 55, 35-
41.
Harris, F. R., Wolf, M. M., & Baer, D. M. Effects of adult

Children, 1964, 20, 8-17.

Psychology, 1969, 74, 263-270.
Hilgard, E. R., & Bower, G. H. Theories of learning. (3d

ed.) New York: Appleton-Century-Crofts, 1966.
Holstein, S., & Hundt, A. G. Reinforcement of

intracranial self-stimulation by licking.

Homme, L. E. Contiguity theory and contingency

management. Psychological Record, 1966, 16,233-
241.
Hull, C. L. Principles of behavior. New York: Appleton-

Hundziak, M., Mowrer, R. A., & Watson, L. S., Jr.

Operant conditioning in toilet training of severely
mentally retarded boys. American Journal of
Isaac, W. Evidence for a sensory drive in monkeys.

Psychological Reports, 1962, 11, 175-181.
Isaac, W., Thomas, J., & Goldiamond, I. Application of

operant conditioning to reinstate verbal behavior
in psychotics. Journal of Speech and Hearing
Disorders, 1960, 25, 8-12.
Johnston, M. S., Kelley, C. S., Buell, J. S., Harris, F. F., &

Wolf, M. M. Effects of positive social
reinforcement on isolate behavior of a nursery
school child. Unpublished manuscript, University
of Washington, 1963.
Johnston, M. S., Kelley, C. S., Harris, F. R., & Wolf, M. M.

An application of reinforcement principles to
development of motor skills of a young child.

1968. pp. 254-290.
Kelleher, R. T,, & Gollub, L. R., A review of positive

conditioned reinforcement. Journal of the
597.

1961.
King, G. F., Armitage, S. G., & Tilton, J. R. A therapeutic

approach to schizophrenics of extreme
pathology. Journal of Abnormal and Social
Psychology, 1960, 61, 276-286.
Kish, G. B. Studies of sensory reinforcement. In W. K.

Honig (Ed.), Operant behavior. New York:
Appleton-Century-Crofts, 1966. pp. 109-159.
Kolb, D. A., Winter, S. K., & Berlew, D. E. Self-directed

change: Two studies. Journal of Applied
Behavioral Science, 1968, 4, 453-471.
Krasner, L. Studies of the conditioning of verbal

170.
Krasner, L. The psychotherapist as a social

reinforcement machine. In H. H. Strupp & L.
Luborsky (Eds.), Research in psychotherapy. Vol.
II. Washington, D. C.: American Psychological
Association, 1962. pp. 61-94.
Leitenberg, H., Agras, W. S., & Thomson, L. E. A

sequential analysis of the effect of selective
positive reinforcement in modifying anorexia
nervosa. Behaviour Research and Therapy, 1968,
6, 211-218.
Leuba, C. Toward some integration of learning

theories: The concept of optimal stimulation.
Levin, G. R., & Simmons, J. J. Response to food and
praise by emotionally disturbed boys.
Lewis, M., Wall, A. M., & Aronfreed, J. Developmental

change in the relative values of social and
nonsocial reinforcement. Journal of Experimental
Psychology, 1963, 66, 133-137.
Lovaas, O. I. Interaction between verbal and

nonverbal behavior. Child Development, 1961, 32,
329-3.36.
Lovaas, O. I. Control of food intake in children by

reinforcement of relevant verbal behavior.
68, 672-678.

Early childhood autism. Oxford: Pergamon, 1966.
pp. 115-144.
Lovaas, O. I. Some studies on the treatment of

childhood schizophrenia. In J. M. Shlien (Ed.),
103-129.
Lovaas, O. I., Berberich, J. P., Perloff, B. F., & Schaeffer,
B. Acquisition of imitative speech by
schizophrenic children. Science, 1966, 151, 705—
707.
Lovaas, O. I., Freitag, G., Kinder, M. I., Rubenstein, B. D.,

Schaeffer, B., & Simmons, J. Q. Experimental
studies in childhood schizophrenia. III.
Developing social behaviors using electric shock.
Journal of Experimental Research in Personality,
1965, 1, 99-109.
Lovaas, O. I., Freitag, G., Kinder, M. I., Rubenstein, B. D.,

Schaeffer, B., & Simmons, J. Q. Establishment of
social reinforcers in two schizophrenic children
on the basis of food. Journal of Experimental Child
Psychology, 1966, 4, 109-125.
McGuire, R. J., & Vallance, M. Aversion therapy by

electric shock: A simple technique. British
Medical Journal, 1964, 1, 151-153.
Mager, R.F. Developing attitudes toward learning. Palo

Alto, Calif.: Fearon, 1968.
Martin, G. L., & Powers, R. B. Attention span: An

operant conditioning analysis. Exceptional
Children, 1967, 33, 565-570.
Martin, M., Burkholder, R., Rosenthal, T. L., Tharp, R.

G., & Thorne, G. L. Programming behavior change
and reintegration into school milieux of extreme
adolescent deviates. Behaviour Research and
Therapy, 1968, 6, 371-383.
Meehl, P. E. On the circularity of the law of effect.

Miles, R. C. Learning in kittens with manipulatory,

exploratory, and food incentives. Journal of
51, 39-42.
Miller, L. B., & Estes, B. W. Monetary reward and

motivation in discrimination learning. Journal of
Miller, N. E. Some reflections on the law of effect

produce a new alternative to drive reduction. In
M. R. Jones (Ed.), Nebraska symposium on
motivation: 1963. Lincoln: University of Nebraska
Press, 1963. pp. 65-112.
Miller, N. E., & Dollard, J. Social learning and imitation.

New Haven: Yale University Press, 1941.
Miller, N. E., & Kessen, M. L. Reward effects of food via

stomach fistula compared with those of food via
mouth. Journal of Comparative and Physiological
Psychology, 1952, 45, 555-564.
Minge, M. R., & Ball, T. S. Teaching of self-help skills to
profoundly retarded patients. American Journal of
Mowrer, O. H. Learning theory and behavior. New

York: Wiley, 1960.
Murray, E. J. A content-analysis method for studying

psychotherapy. Psychological Monographs, 1956,
70 (13, Whole No. 420).
Myers, A. K., & Miller, N. E. Failure to find a learned

drive based on hunger: Evidence for learning
motivated by exploration. Journal of Comparative
Narrol, H. G. Experimental application of

reinforcement principles to the analysis and
treatment of hospitalized alcoholics. Quarterly
Journal of Studies on Alcohol, 1967,28, 105-115.
Odom, R. D. Effects of auditory and visual stimulus

deprivation and satiation on children’s
performance in an operant task. Journal of
O’Leary, K. D. A token reinforcement system in the

public schools. Paper presented at the meeting of
the American Psychological Association, San
Francisco, September, 1968.
O’Leary, K. D., & Becker, W. Behavior modification of
an adjustment class: A token reinforcement
program. Exceptional Children, 1967, 33, 637-
642.
Patterson, G. R. A learning theory approach to the

treatment of the school phobic child. In L. P.
Ullmann & L. Krasner (Eds.), Case studies in
behavior modification. New York: Holt, Rinehart &
Winston, 1965. pp. 279- 285.
Patterson, G. R., Ray, R., & Shaw, D. Direct intervention

in families of deviant children. Unpublished
manuscript, University of Oregon, 1968.
Peters, H. N., & Jenkins, R. L. Improvement of chronic

schizophrenic patients with guided problem-
solving motivated by hunger. Psychiatric
Quarterly Supplement, 1954, 28, 84-101.
Phillips, E. L. Achievement place: Token

reinforcement procedures in a home- style
rehabilitation setting for “pre-delinquent” boys.
223.
Premack, D. Reinforcement theory. In D. Levine (Ed.),

Nebraska symposium on motivation: 1965.
Lincoln: University of Nebraska Press, 1965. pp.
123-180.
Redl, F., & Wineman, D. Children who hate: The
disorganization and breakdown of behavior
controls. Glencoe, Ill.: Free Press, 1951.
Renner, K. E. Delay of reinforcement: A historical

review. Psychological Bulletin, 1964, 61, 341-361.
Resnick, L. B. Programmed instruction and the

teaching of complex intellectual skills: Problems
and prospects. Harvard Educational Review,
1963, 33, 439-471.
Rheingold, H. L., Stanley, W. C., & Doyle, G. A. Visual

and auditory reinforcement of a manipulatory
response in the young child. Journal of
Experimental Child Psychology, 1964, 1,316-326.
Richard, H. C., & Dinoff, M. A follow-up note on “verbal

manipulation in a psychotherapeutic
relationship.” Psychological Reports, 1962, 11,
506.
Rocha e Silva, M. I., & Ferster, C. B. An experiment in

teaching a second language. Sonderdruck aus
IRAL, 1966, 4, 85-113.
Rogers, J. M. Operant conditioning in a quasi-therapy

setting. Journal of Abnormal and Social
Psychology, 1960, 60, 247-252.
Roos, P. Development of an intensive habit-training
unit at Austin State School. Mental Retardation,
1965, 3, 12-15.
Salzinger, K. Experimental manipulation of verbal

behavior: A review. Journal of General Psychology,
1959, 61, 65-94.
Salzinger, K., Feldman, R. S., Cowan, J. E., & Salzinger,

S. Operant conditioning of verbal behavior of two
young speech-deficient boys. In L. Krasner & L. P.
82-105.
Schachter, S. Cognitive effects on bodily functioning:

Studies of obesity and eating. In D. C. Glass (Ed.),
Neurophysiology and emotion. New York:
Rockefeller University Press and Russell Sage
Foundation, 1967, pp. 117-144.
Schaefer, H. H., & Martin, P. L. Behavioral therapy for

“apathy” of hospitalized schizophrenics.
Schwartz, A. N., & Hawkins, H. L. Patient models and

affect statements in group therapy. Proceedings of
the 73rd Annual Convention of the A.P.A.
Washington, D. C.: American Psychological
Association, 1965. pp. 265- 266.
Sears, R. R., Maccoby, E. E., & Levin, H. Patterns of child
rearing. Evanston, Ill.: Row, Peterson, 1957.
Seidel, R. J. A review of sensory preconditioning.

Sheffield, F. D., & Roby, T. B. Reward value of a non-

nutritive sweet taste. Journal of Comparative and
Sheffield, F. D., Wulff, J. J., & Backer, R. Reward value of

copulation without sex drive reduction. Journal of
Comparative and Physiological Psychology,
1951,44, 3-8.
Sherman, J. A. Use of reinforcement and imitation to

reinstate verbal behavior in mute psychotics.
164.
Sidman, M. Tactics of scientific research. New York:

Basic Books, 1960.
Silberman, H. F. Characteristics of some recent studies

of instructional methods. In J. E. Coulsen (Ed.),
Programmed learning and computer-based
instruction. New York: Wiley, 1962. pp. 13-24.
Skinner, B. F. Operant behavior. American

Psychologist, 1963, 18, 503-515.
Slack, C. W. Experimenter-subject psychotherapy: A
new method of introducing intensive office
treatment for unreachable cases. Mental Hygiene,
1960, 44, 238-256.
Staats, A. W. A case in and a strategy for the extension

of learning principles to problems of human
behavior. In L. Krasner & L. P. Ullmann (Eds.),
Research in behavior modification. New York:
Holt, Rinehart & Winston, 1965. pp. 27-55.
Staats, A. W., & Butterfield, W. H. Treatment of

nonreading in a culturally deprived juvenile
delinquent: An application of reinforcement
principles. Child Development, 1965, 36, 925-942.
Staats, A. W., Finley, J. R., Minke, K. A., & Wolf, M.

Reinforcement variables in the control of unit
reading responses. Journal of the Experimental
Analysis of Behavior, 1964, 7, 1.39-149.
Staats, A. W., Minke, K. A., Finley, J. R., Wolf, M., &

Brooks, L. O. A reinforcer system and
experimental procedure for the laboratory study
of reading acquisition. Child Development, 1964,
35, 209-231.
Staats, A. W., Minke, K. A., Goodwin, W., & Landeen, J.

Cognitive behavior modification: ‘motivated
learning’ reading treatment with subprofessional
therapy-technicians. Behaviour Research and
Therapy, 1967, 5, 283-299.
Staats, A. W., Staats, C. K., Schutz, R. E., & Wolf, M. M.

The conditioning of textual responses using
“extrinsic” reinforcers. Journal of the
Experimental Analysis of Behavior, 1962, 5, 33-40.
Stanton, A. H., & Schwartz, M. S. The mental hospital.

New York: Basic Books, 1954.
Stevenson, H. W., & Odom, R. D. Visual reinforcement

with children. Journal of Experimental Child
Psychology, 1964, 1, 248-255.
Stolurow, L. M. Programed instruction for the

mentally retarded. Review of Educational
Research, 1963, 33, 126-136.
Straughan, J. The application of operant conditioning

to the treatment of elective mutism. In H. N.
Sloane, Jr., & B. A. MacAulay (Eds.), Operant
procedures in remedial speech and language
training. Boston: Houghton Mifflin, 1968. pp. 242-
255.

365.
Stunkard, A. J. The management of obesity. New York
State Journal of Medicine, 1958, 58, 79-87.
Stunkard, A., & Koch, C. The interpretation of gastric

motility. I. Apparent bias in the reports of hunger
by obese persons. Archives of General Psychiatry,
1964, 11,74-82.
Tobias, S. Effects of verbal reinforcement on response

changes in a nonreinforced situation. American
Psychologist, 1960, 15, 390 (Abstract).
Tooley, J. T., & Pratt, S. An experimental procedure for

the extinction of smoking behavior. Psychological
Record, 1967, 17, 209-218.
Truax, C. B. Reinforcement and nonreinforcement in

Rogerian psychotherapy. Journal of Abnormal
Psychology, 1966, 71, 1-9.
Ullmann, L. P., Krasner, L., & Collins, B. J. Modification

of behavior through verbal conditioning. Journal
of Abnormal and Social Psychology, 1961, 62, 128-
132.
Ullmann, L. P., Krasner, L., & Edinger, R. L. Verbal

conditioning of common associations in long-
term schizophrenic patients. Behaviour Research
and Therapy, 1964, 2, 15-18.
Wahler, R. G. Behavior therapy for oppositional

children: Love is not enough. Paper read at
Eastern Psychological Association meeting,
Washington, April, 1968.
Wahler, R. G., & Pollio, H. R. Behavior and insight: A

case study in behavior therapy. Journal of
56.
Wessen, A. F. (Ed.). The psychiatric hospital as a social

system. Springfield, Ill.: Charles C Thomas, 1964.
Whalen, R. E. Effects of mounting without

intromission and intromission without
ejaculation on sexual behavior and maze
learning. Journal of Comparative and Physiological
Psychology, 1961, 54, 409-415.
Whitlock, C., & Bushell, D., Jr. Some effects of “back-

up” reinforcers on reading behavior. Journal of
Williams, R. I. Verbal conditioning in psychotherapy.

American Psychologist, 1959, 14, 388 (Abstract).
Winder, C. L., Ahmad, F. Z., Bandura, A., & Rau, L. C.

Dependency of patients, psychotherapists’
responses, and aspects of psychotherapy. Journal
of Consulting Psychology, 1962, 26, 129-134.
Wolf, M. M., Giles, D. K., & Hall, R. V. Experiments with

token reinforcement in a remedial classroom.
Wolf, M. M., & Risley, T. Analysis and modification of

deviant child behavior. Paper read at the
American Psychological Association meeting,
Washington, September, 1967.
Zeilberger, J., Sampen, S. E., & Sloane, H. N., Jr.

Modification of a child’s problem behaviors in the
home with the mother as therapist. Journal of
Applied Behavior Analysis, 1968, 1, 47-53.
Zimmerman, D. W. Durable secondary reinforcement.

Zimmerman, E. H., & Zimmerman, J. The alteration of

behavior in a special classroom situation. Journal
59-60.
5
Aversive Control
Negative consequences are widely used to
modify behavior, but such practices are generally
disavowed. There are several reasons, some
rationally grounded and others unwarranted, why
punishment is regarded with disfavor. One of the
principal objections to aversive control stems
from the widespread belief that internal, and often

unconscious, forces are the major determinants of
behavior. From this perspective, punishment may
temporarily suppress certain expressions but the

underlying impulses retain their strength and
press continuously for discharge through
alternative actions. Moreover, when punishment is

described in behavior theory as having inhibitory
or suppressive effects, it also carries the
implication that no expressed response tendencies
remain active at a covert level and, therefore,
require continual counteraction.
Both the immediate and long-term effects of
negative sanctions are viewed with a less

jaundiced eye when behavior is assumed to be
largely determined by its consequences. As long as
a given response pattern creates aversive
outcomes of sufficient strength to override the

effects of other maintaining conditions, it will not
be utilized. If more effective means of securing
desired outcomes are developed during this
period, the behavioral changes initiated through
punishment endure after the aversive

contingencies have been discontinued. Thus, for
example, if competitive behavior is rendered
nonfunctional by arranging adverse consequences
while cooperativeness proves more rewarding,
competitiveness will eventually be abandoned
without any internal energy strains. When the
person is behaving cooperatively it does not mean

that he is actively suppressing a simultaneously
aroused competitive response, any more than a
person who is sitting is continually subduing a
standing response that is simultaneously pressing
for expression.
The use of aversive control is also frequently

questioned on the grounds that it produces a
variety of undesirable by-products. This concern is
warranted, as we shall see later. Many of the

unfavorable effects, however, that are sometimes
associated with punishment are not necessarily
inherent in the methods themselves but result

from the faulty manner in which they are applied.
A great deal of human behavior is, in fact, modified
and closely regulated by natural aversive
contingencies without any ill effects. On the basis
of negative consequences people learn to avoid or
to protect themselves against hazardous falls,
flaming or scalding objects, deafening sounds, and

other hurtful stimuli; they change their clothing to
remain comfortable in sweltering or frigid
temperatures; and they engage in a considerable
amount of behavior that is supported almost
entirely by removal of irritants. In instances where

certain activities can have injurious effects
aversive contingencies must be socially arranged
to ensure survival. Punishment is rarely indicted
for ineffectiveness or deleterious side effects when
used, for example, to teach young children not to

insert metal objects into electrical outlets, not to
cross busy thoroughfares against red signal lights,

nor to perform behavior that would result in
mutilation. Certain types of negative sanctions, if
applied considerately, can likewise aid in

eliminating self-defeating and socially detrimental
behavior without creating any special problems.
Punishment is rarely employed as a sole

method for modifying behavior; but if it is used
judiciously in conjunction with other techniques
designed to promote more effective response

options, such combined procedures can hasten the
change process. In addition, aversive
consequences are frequently used to modify

deviant behavior that is automatically self-
reinforcing upon occurrence and in cases where
certain response patterns must be brought rapidly
under control because of their injurious effects
upon the performer or other persons.
The negative consequences may involve either
the removal of positive reinforcers or the
presentation of aversive stimuli. The events in the
latter category may take the form of physical
punishment, shock stimulation, unpleasant
auditory feedback, or verbal reprimands. In the
former case response costs may be instituted or
persons may be deprived of privileges,

possessions, social contacts, and other positively
reinforcing events to which they are accustomed.
Although both operations represent forms of
punishment, they not only have somewhat

different effects on behavior, but they may
produce different side effects and contrasting
reactions toward prohibiting agents.
Consequently, the research findings and
applications of these two forms of punishment will

be discussed separately.
Presentation of Negative Reinforcers
Theories and research regarding punishment
have been primarily concerned with the extent to

which response patterns can be removed through
direct administration of aversive stimuli. For
obvious ethical reasons studies of the behavioral
changes produced by punishment of high intensity
or of long duration have been confined to

infrahuman subjects. Laboratory investigations of
the effects of punishment on humans typically
employ either relatively weak physical stimuli or
negative symbolic consequences. In these studies
punishment is applied either to responses that are

concurrently being maintained on an intermittent
schedule of positive reinforcement or to response
patterns that are undergoing extinction.
THEORETICAL CONCEPTIONS OF PUNISHMENT

EFFECTS
When aversive consequences follow upon any
behavior, they generally produce a reduction or
cessation of responsiveness. After the aversive

contingencies have been discontinued the
behavior sometimes reappears, suggesting that

punishment suppresses response tendencies but
does not eliminate them. Under some conditions,
however, punishment may produce enduring
changes in behavior. As will be shown later, the

degree to which behavior is positively reinforced
is one determinant of both the reductive power of
punishment and the extent to which punished
responses are subsequently reinstated. In addition

to the reward contingencies maintaining the
punished behavior, the effects of punishment may
vary considerably as a function of many other
variables (Azrin & Holz, 1966; Church, 1963;
Solomon, 1964) including the intensity, duration,

frequency, and distribution of aversive
consequences; their temporal relation to the

behavior to be modified; the strength of punished
responses; the availability of alternative

behavioral patterns that are positively
reinforceable; the presence of discriminative

stimuli that signify the probability that a given
performance will result in adverse consequences;
the level of instigation to perform the negatively
sanctioned behavior; and the characteristics of

punishing social agents. The degree of control
exercised by aversive outcomes over behavior in

any particular case is, therefore, highly dependent
upon a host of other operative variables in
addition to the punishment contingencies.
Several theoretical formulations have been

advanced to account for the varied behavioral
effects produced by different types of punishment.
Conditioned emotionality theories (Estes, 1944),
for example, primarily attribute the effects of

punishment to emotional reactions that are
classically conditioned to environmental stimuli
during the course of aversive treatment. Later,
exposure to the threatening situation generates
emotional arousal that may disrupt or inhibit
responding. Major support for this view comes
from countless studies demonstrating that
behavior can be suppressed by the presentation of
environmental stimuli that had been previously

paired with aversive experiences. The theory of
punishment advanced by Mowrer (1960) assumes
that negative emotions are conditioned not only to
environmental stimuli but also to proprioceptive
cues generated by the punished behavior itself.
According to competing-response
interpretations (Guthrie, 1935), punishment
produces behavioral changes by eliciting
incompatible responses in the presence of cues

that formerly controlled the punished behavior.
Competing responses presumedly are directly
conditioned by contiguity to stimuli present at the

time of punishment. The generality of this
nonmediational explanation is most seriously
challenged by results of experiments in which
previously neutral stimuli are endowed with
response-inhibiting properties under conditions
where animals are skeletally immobilized by
curare to prevent motor responding (Rescorla &

Solomon, 1967). Nevertheless, in ordinary
circumstances punishment activates instrumental
responses as well as fear; consequently,
punishment can produce markedly different
behaviors under similar levels of emotional

arousal depending upon the types of responses
that it originally evoked (Bolles, 1967). The same
punishing stimulus may thus accelerate or retard
performance of the same behavior depending
upon whether it is applied in such a way as to

evoke responses that are compatible or in conflict
with the ongoing activities (Fowler & Miller,

1963).
Whereas the preceding theory construes
punishment effects in terms of competing
associative learning, avoidance conditioning
explanations (Dinsmoor, 1954) emphasize the
reinforcing consequences that are produced by,
and serve to maintain, incompatible behavior. To

the extent that withdrawal, avoidance, and
response inhibition provide relief from disturbing
stimulation the behavior is thereby reinforced.
The above-mentioned theories emphasize the
motivating, reinforcing, and response-eliciting

capacity of punishment. On the other hand,
discrimination hypotheses (Holz & Azrin, 1961,
1962) highlight the informative function of
punishing events. The inhibitory effects produced

through vicarious punishment, for example, result
in large part from the information conveyed to
observers that certain types of performances are

negatively sanctioned in particular situations
(Bandura, 1965). The discriminative properties of
punishment are also clearly revealed by
experiments in which the temporal order of
punishment, reward, and extinction are
systematically varied. Severe punishments that
regularly precede reward come to signal positive

reinforcement and accelerate responding;
conversely, otherwise weak and ineffective
punishments that precede the removal of rewards
decrease responding. Similarly, punishments that
initiate a period of relief from further painful

stimulation become safety signals that lead to
increased responsiveness (Hendry & Van Toller,
1964).
Further evidence of the way in which

punishing stimuli can acquire markedly different
properties through their relation with other
reinforcing events is strikingly demonstrated by

Sandler and Quagliano (1964). After monkeys
learned to press a lever to avoid being shocked, a
second contingency involving self-administered
painful stimulation was introduced. A lever press
prevented the occurrence of the original shock,
but it also produced an electric shock of lesser
magnitude. As the experiment progressed the self-

administered shock was gradually increased until
it equalled the aversive stimulus being avoided.
However, the animals showed no reduction in the
frequency of self-punishing lever-pressing
responses though this behavior no longer served

as a “lesser of two evils.” Even more surprising,
after the avoided shock was completely
discontinued but lever-pressing responses (which
had now become objectively functionless) still
produced painful consequences, the animals

continued to punish themselves unnecessarily
with shock intensities that they had previously

worked hard to avoid. Anyone observing the
needless self-injurious behavior of these animals
without knowledge of their prior learning history

would undoubtedly be baffled by their tenacious
“masochism.”
Findings of a study by Ayllon & Azrin (1966)

identify conditions under which a formerly
punishing stimulus can not only maintain
responding through its informative value, but also

serve as a positive reinforcer to strengthen new
performances. After white noise completely
suppressed a rewarded response in chronic

schizophrenics, the white noise was then
intermittently paired with tokens having reward
value. Later, contingent noise alone maintained
several thousand responses on a new task. The
preceding experiments illustrate how punishing

events can achieve enduring reinforcing functions
that become dissociated from the original
conditions of reinforcement. The resultant self-
punitive behavior, whether maintained through
fear of threats that no longer exist or by
anticipation of occasional rewards that are no
longer forthcoming, is clearly inappropriate if
viewed in terms of objective reality.
In the material presented above we reviewed

separately the various effects that punishing
stimuli can have on behavior. Most theorists
subscribe to a multi-process conceptualization of
the manner in which punishment achieves
reductive effects and the factors that govern any

subsequent recovery of punished responses.
According to this view, painful stimulation
produces both generalized emotional arousal and
escape- withdrawal responses, which are usually

incompatible with and, therefore, capable of
replacing the ongoing behavior. Any
environmental stimuli and responses that

regularly precede or accompany the aversive
experiences acquire, through their contiguous
association, the capacity to arouse emotional
reactions for some time after punishment has
been discontinued. In addition to emotional
conditioning, any responses that successfully
terminate or avoid aversive stimulation are

instrumentally reinforced. The punished
responses remain suppressed as long as the
threatening events maintain their capacity to
generate prepotent fear reactions and
incompatible avoidance behavior. This period

provides the opportunity for other modes of
behavior to be established which, if sufficiently
strong, will permanently supplant punished
response tendencies. However, in the absence of
adequately rewarded alternatives, after

punishment has been discontinued, the aversive
properties of the maintaining stimuli may be

extinguished through either repeated exposure to
fear-provoking situations or elicitations of
partially suppressed responses without adverse

consequences, by observation of nonpunished
performances, or by other informative means.
Under these types of conditions inhibited behavior
will eventually reappear.
LOCUS OF AVERSIVE CONTROL
As noted in the preceding discussion several
alternative interpretations have been proposed
concerning the source of the controlling stimuli for

inhibitory processes created through punishment.
According to one explanation, control primarily
resides in environmental stimuli. That is, if a
person regularly undergoes punishing experiences
in a particular situation, the environmental cues,

through their endowed emotion-provoking
properties, produce either generalized
suppression of behavior, avoidance of the
threatening situational events, or activities
designed to prevent their recurrence.
A second interpretation assumes that aversive
control resides in the punished behavior itself.
When punishing consequences are made
contingent upon the occurrence of certain types of
responses, it is assumed that proprioceptive cues

accompanying the responses acquire the capacity
to arouse conditioned emotional reactions.

Aversive stimulation generated by initiation of
previously punished behavior prevents
completion of the action sequence. Response
inhibition is believed to be reinforced by the

termination of response-produced emotional
arousal. Thus, in the latter case, individuals learn
to avoid self-generated distress by suppressing the
negatively valenced behavior.
A number of investigators have compared the

suppressive power of punishment when it is
administered either on a response-contingent
basis or in conjunction with specific
environmental stimuli independently of the
ongoing behavior. These studies reveal that both
types of aversive contingencies decrease
responding, but there is disagreement as to which

method produces greater behavioral suppression.
In the original experiment, Estes (1944) found
that animals which had been shocked only at times
when they were not engaging in lever-pressing
behavior displayed essentially the same degree of

suppression and subsequent recovery of lever-
pressing responses as subjects whose punishment
was strictly contingent upon the occurrence of the
response. In a further test of whether suppressive
effects are governed by environmental stimuli or

response- produced cues, animals whose lever-
pressing responses had been punished were left in

the situation for an adaptation period with the
lever removed. This arrangement prevented
elicitation of the punished response but permitted

the neutralization of threatening situational cues.
A subsequent test for extinction revealed that
mere re-exposure to the fear-provoking situation
with no further unpleasant experiences resulted in

almost complete extinction of conditioned
suppression. The findings of this study thus

suggested that internal cues accompanying the
punished response exercised relatively little
influence upon the inhibitory process.
Hunt & Brady (1955) extended the above

research in a comparative study of the influence of
response-contingent and stimulus-correlated
punishment upon the acquisition, generalization,
and extinction of conditioned suppression of

responses that were intermittently rewarded. For
subjects in the stimulus group, shocks were
contiguously associated with a tone, but lever-
pressing responses were never punished; on the
other hand, in the response condition, the tone
was presented and animals were shocked only
when they pressed the lever in the presence of the
auditory stimulus. Both procedures resulted in
almost complete response suppression whenever

the tone was presented. The method designed
specifically to endow the environmental cue with
aversive properties, however, produced greater
emotional disturbance, and greater generalized
inhibition that was more resistant to extinction.

Essentially identical results were obtained in an
earlier study (Hunt & Brady, 1951) even though
subjects in the response-contingent treatment
received more shocks. In a well-designed
experiment that equated for the number and

temporal distribution of shocks, Hoffman &
Fleshier (1965) found that animals that were

punished only if they responded in the presence of
certain cues displayed less behavioral suppression
and extinguished more rapidly than their

counterparts that were punished during
presentations of the same cues without regard to
their behavior. The foregoing results thus provide
evidence that, under certain conditions,

inhibitions are primarily situation- rather than
response-bound.
At variance with the above conclusion, Azrin
(1956) found that response-produced punishment
was considerably more effective than

noncontingent aversive stimulation in suppressing
rewarded behavior. Azrin attributes the
conflicting results to the fact that subjects in his
experiment received severe shocks over an
extended time, whereas previous studies had

employed relatively brief periods of punishment.
This interpretation does not fully explain the
diverse results, because Boe & Church (1967)
report that response-contingent punishment is
more suppressive than noncontingent shock even
when administered during a brief period.
Moreover, Camp (1965), employing a wide range
of shock intensities, found that the suppressive
effects of response-contingent punishment were

greater than the inhibitory effects of
noncontingent punishment, but the differences
between the two procedures were relatively small.
Response inhibition is most likely to be highly
situation-bound when aversive experiences occur

repeatedly in the same setting. If, on the other
hand, selected responses are punished in the
presence of specific stimuli but incur no negative
consequences within the same environment when

other factors are absent, then omnipresent stimuli
are less likely to become endowed with strong
suppressive power. Under natural conditions, of

course, punishment is seldom contingent upon
behavior alone; nor are people invariably
punished in certain situations. Rather, the same
behavior may be permissible or punishable in the
same settings depending upon the person toward
whom the behavior is directed, the role occupied
by the performer, the time and specific

circumstances under which the behavior is
exhibited, the instigating conditions, and many
other factors.
Peripheral theories of response inhibition that
emphasize skeletal responding and its attendant

proprioceptive feedback cannot adequately
account for the highly discriminative manner in
which the same responses are freely performed or
suppressed under slightly differing circumstances.

Regulation of human behavior on the basis of
punishing experiences is undoubtedly mediated to
a considerable degree through central

mechanisms. The major theoretical issues and
supporting evidence concerning symbolic control
of responsiveness, which constitutes a third locus
of aversive control, is presented in the final
chapter of this book. This view assumes that, on
the basis of information conveyed through prior
response consequences experienced in

conjunction with different situational, temporal,
and social cues, individuals infer the probability
that a given course of action will be ignored,
rewarded, or punished. Behavior is then partly
guided by anticipatory consequences that are

symbolically produced. That is, anticipatory
aversive consequences will have response-
inhibiting effects, whereas anticipation of
rewarding outcomes will facilitate performance of
the same behavior. In addition to the influence of

expectant outcomes, self-evaluative consequences
also figure prominently in the self-regulation of

behavior.
A number of studies employing punishing
operations have been conducted with children in
an effort to elucidate internalization processes. In
a comprehensive monograph, Aronfreed (1968)
conceptualizes internalization as essentially a
process in which aversive or pleasurable affective

states become attached to both response-
produced cues and their cognitive
representations. This outcome is presumably
achieved on the basis of classical conditioning of
anxiety to the proprioceptive stimuli inherent in

the action itself and, through the temporal
gradient of reinforcement, to its behavioral and
cognitive precursors. Thereafter, stimuli
accompanying preparatory movements, or even
intentions, arouse anxiety which in turn prevents

further responding. Behavioral suppression is self-
reinforced by the resultant reduction in anxiety.

Thus, according to this view, behavior becomes
internalized or self-maintained when changes in
affective states, which constitute intrinsic

reinforcing consequences, are mediated by
response-produced cues.
Empirical tests of the affective feedback theory

of internalization have primarily investigated
behavioral suppression as a function of timing of
punishment. In the typical experimental paradigm,

on each of a series of trials children are asked to
choose between a highly attractive toy and one
which is relatively uninviting. In one condition the

experimenter verbally reprimands children as
soon as they make a motion toward the attractive
toy; in other treatments children are verbally
rebuked only after they have picked up the toy and
handled it for varying periods of time. In the test

for internalized behavioral suppression children
are presented with a highly attractive and an
uninteresting toy and it is noted whether or not
they touch the attractive object during the
experimenter’s absence. The rationale for
selecting the temporal variable is that punishment
at the initiation of a transgression is expected to
attach maximal anxiety to cues that accompany
preparatory responses; as a result, behavior is

suppressed in its early stages. On the other hand, if
punishment is administered following
transgression, anxiety and its attendant inhibitory
responses are not evoked until after the
disapproved act has been completed.
Before evaluating the main findings of these

studies it should be noted that tests of internalized
behavioral control typically involve so many
external stimulus supports that response

suppression cannot be attributed solely to
intrinsically mediated consequences. Post-training
measures of self-control are characteristically

obtained by the same experimenter, during the
same experimental session, in the same
experimental room, in which children are
presented either identical or similar stimulus
objects possessing the same relational properties
as the stimuli utilized during training. Although
the punishing agent absents himself from the

room, nevertheless many controlling external cues
remain. It would, in fact, be of interest to study the
progressive decrease in self-controlling responses
as progressively more cues common to the
punishment situation are removed.
Results of several experiments based on this

procedure (Aronfreed, 1968; Aronfreed & Reber,
1965; Benton, 1967; Walters, Parke, & Cane, 1965)
show that children who have been punished early

in the response sequence deviate less frequently
than children punished only after completion of
the transgression. Moreover, the longer that

punishment is withheld while children are
engaging in the deviant behavior, the weaker the
subsequent behavioral suppression. These
findings have been interpreted as providing strong
evidence that behavior is regulated by affective
consequences conditioned to sequential intrinsic
stimulus correlates of a punished act. Results of

animal experiments investigating the effects of
delay of punishment on resistance to temptation
(Solomon, Turner, & Lessac, 1968) are also
frequently cited as corroborating the intrinsic
mediation of behavioral suppression.
Findings yielded by the timing-of-punishment

experiments cannot be meaningfully interpreted
with respect to the affective feedback hypotheses
because, among other methodological problems,

the timing manipulations involve a confounding of
variables. In the early punishment condition
transgressive behavior produces only punishment,

whereas in delayed conditions the behavior
receives both immediate reward and subsequent
punishment. The suppressive effect of instant
reward followed some time later by punishment is
much weaker than punishment alone. In the study
conducted by Solomon, for example, food-
deprived puppies that were swatted by the

experimenter as they touched forbidden food
exhibited stronger resistance to deviation than
puppies that were punished shortly after they had
begun to consume the palatable horsemeat. The
difference in response inhibition, which has been

often attributed by other authors to the
attachment of anxiety at different points in the
sequence of response-produced cues, simply
demonstrates that a response is more effectively
inhibited by punishment alone than if it is initially

rewarded and then punished. Confounding of
reward and punishment effects, as Solomon points

out, is advantageous for understanding natural
socialization practices, but the data are of limited
value for elucidating intrinsic conditioning

processes.
It is likewise difficult to ascertain whether
similar differences obtained in investigations cited
earlier derive from sequential aversive

conditioning of response-produced cues, or from
the fact that in late-punishment conditions

children are rewarded by being allowed to play
briefly with a highly attractive toy before the
reprimand is administered. As might be expected
from the response-enhancing effects of positive

reinforcement, the longer children are allowed to
engage in the rewarding transgressive behavior
the weaker is the subsequent punishment. On the

other hand, when children are punished almost
immediately upon committing the deviant act

(Parke & Walters, 1967; Walters & Demkow,
1963), differences in response suppression
between early-punished and late-punished

subjects become quite small and, for the most part,
statistically insignificant. The latter negative
results have been attributed to insufficient
separation of sequential stimulus components in
early- and late-punishment conditions, but an
alternative interpretation may be cast in terms of

curtailment of confounding rewarding effects
which occur under delayed punishment.
It also seems highly probable that a person
who reprimands a child even before a deviant act

actually occurs would be viewed as considerably
more forbidding and punitive than one who shows
no disapproval until after the child has engaged in
the transgressive behavior for some time. Indeed,
in the latter case, the disciplinarian must appear to

be a more lenient, or at least an inconsistent,
person. Early-punished children may therefore be
reluctant to transgress not because of attachment
of affective states to incipient response-produced
cues, but rather due to stronger anticipatory fears
evoked cognitively by the stricter disciplinarian.
In addition to the methodological problems
considered above, the type of discrimination task
employed in most timing-of-punishment

experiments is not the most suitable one for
determining the extent to which response

suppression is mediated by its proprioceptive
cues. This is because selection of both attractive
and unattractive objects involves virtually
identical reaching responses, and therefore

arousal of the same pattern of interoceptive
stimulation. Consequently, to the extent that
anxiety is conditioned to cues inherent in the

response itself, the act of reaching for the
unattractive object should be equally suppressed.

Demonstration of proprioceptive stimulus control
of behavior would require subjects to make
distinctive responses, each producing dissimilar

patterns of internal stimulation which would have
been associated with differential consequences.
Moreover, in order to assess what contribution, if
any, is made by proprioceptively aroused affects
to response suppression, it would be necessary to

measure the independent effects of cognitive
discriminations and symbolically generated

affective consequences. By analogy with the
timing-of-punishment experiments, a person
reaching for a repulsive snake and for a savory
dessert would in all likelihood experience aversive

and pleasurable emotional states, respectively.
Since, however, the reaching responses in both
instances produce essentially identical patterns of

interoceptive stimulation, the resultant affective
states must be centrally, rather than peripherally,

mediated.
For reasons given above it is doubtful that
timing-of-reinforcement paradigms can provide
decisive evidence concerning sensory feedback
theories of internalization. However, curare
conditioning experiments—in which animals are
skeletally immobilized during aversive
conditioning or extinction—conducted by
Solomon and his associates (Black, 1958; Black,
Carlson, & Solomon, 1962; Solomon & Turner,
1962), shed considerable light on the issue of
central or peripheral mediation of inhibitory
response tendencies. The findings demonstrate

that conditioned emotional responses can be
readily acquired and extinguished independently
of skeletal responding and its correlated
proprioceptive feedback. It would appear from the
overall findings of the series of curare

experiments that when approach responses
toward certain discriminable stimulus objects are

undergoing punishment, the external stimuli also
acquire the capacity to elicit conditioned
emotional responses which can, in turn, control

instrumental responses quite independently of
response-correlated feedback mechanisms.
The role of central processes in behavioral

inhibition could be established through traditional
procedures by including training conditions in
which subjects merely verbalize their object

choices and are punished whenever they elect
disapproved items without performing any
reaching responses. If children, in these conditions

displayed a degree of response suppression equal
to that of subjects whose motor choice responses
were punished, then the findings would indicate
that proprioceptive feedback does not enter into
the regulatory process.
This issue can also be easily investigated by
comparing the inhibitory effects of punishment
when administered to the responding as opposed
to the nonresponding part of the body. In an
experiment reported by Kaufman (1964), adults
participated in a series of sessions that consisted
of alternating periods of punished and unpunished
responding. In half the punishment periods shock

was delivered to the performing hand, while in the
other half the nonresponding hand was shocked.
At low and high intensities, variations in the locus
of punishment had no differential effects, but at

moderate intensities shock applied to the
responding hand produced slightly more
suppression. Some further evidence that response
inhibition is primarily a cognitively controlled
phenomenon is that under severe punishment

subjects completely suppressed responding but
the same responses were performed at a stable

high rate during nonshock periods in the same
session.
Another effective means of assessing whether
response-produced cues assume controlling

functions through affective conditioning is to
include observing partners in punishment
paradigms. This procedure was, in fact, employed
by Benton (1967). While groups of children were

verbally reprimanded for either approaching or
handling tabooed toys in a two-choice
discrimination task, their matched partners simply
observed the punished performances. In a
subsequent test for transgression, the observers

showed the same amount of response inhibition as
the performers whose motor responding was
repeatedly punished.
In Aronfreed’s (1968) theory of internalization,

affective states become affixed not only to
proprioceptive cues but also, through backward
generalization, to intentions and cognitive

representations of the punished act. No empirical
investigations have as yet been conducted in
which verbalized intentions alone are punished.
However, there is some evidence, which is
reviewed in Chapter 8, demonstrating that
punishment of imaginal representations of deviant
sexual activities is associated with a reduction in

the corresponding behavior. Aronfreed has shown
that if during punishment children are informed
that selection of the forbidden attractive toys is
disapproved because their functions are difficult
to describe and they are therefore only

appropriate for older children, subjects are later
less inclined to violate the prohibition. It is
difficult to evaluate without additional
assessments whether such instructions facilitate
compliant behavior because they “inject a

cognitive structure” or for other reasons. One
might expect, for example, punishments that are

arbitrarily administered to generate more
resentment and oppositional behavior than when
the basis for the negative sanctions is

considerately explained (Pastore, 1952). Under
conditions where the basis for punishment is
prominently conveyed by the physical attributes
of the forbidden objects, the role of cognitive

functions is considerably reduced. On the other
hand, when the factors determining whether given

performances will incur punishment are complex
and not easily distinguishable, statement of the
contingency rules governing punishment would be
expected to facilitate self-regulation of

responsiveness.
In evaluating the role of cognitive processes in
behavioral control it is essential to distinguish
between the effects of cognitive representation of

responses themselves and symbolically generated
anticipatory consequences. In theories assuming
that emotion-arousing properties are directly
conditioned to responses, the initiation of
negatively valenced behavior or its cognitive
equivalents is presumed automatically to evoke
negative affects that actuate response inhibition.
An alternative conceptualization of the control
mechanisms is that performances are internally

regulated not by emotions directly affixed to the
behavior but rather by anticipated aversive
consequences. Given different expected outcomes,
the same behavior may be inhibited or freely
expressed, which could not occur if emotional

arousal were directly response-cued.
Several studies that have been conducted
within the framework of dissonance theory also
raise an issue that is relevant to aversive control.

It is assumed in this formulation that if a person is
provided with insufficient justification for his
behavior, the resultant inconsistent cognitions

generate an aversive state that can be reduced by,
among other means, devaluating the activity.
When an individual refrains from transgressing
because of a severe threat, he presumably has
adequate excuse for his compliant behavior and,
therefore, continues to value the tabooed activities
highly. On the other hand, if he displays response

inhibition following a mild threat, he may, to
eliminate disturbing dissonance created by the
incongruous behavior, convince himself that the
desired objects are less worthwhile.
Aronson & Carlsmith (1963) tested the above

notion by having children rate their second-
ranked toy after each of three interventions. In the
phase involving mild threat the experimenter
informed children that he would be annoyed if

they played with the forbidden toy; in the strong
threat condition children were told that if they
played with the toy the experimenter would be

very angry and that he would take all his toys,
never to return again; while in the control phase,
the experimenter simply removed the crucial toy
to determine whether its value may be enhanced
by being singled out for attention. Both threats
produced complete behavioral compliance in all
children, suggesting that the presence of other

attractive toys, including the unprohibited most
preferred item, greatly weakened instigation to
transgression. With respect to toy preferences,
following mild threat 36 percent of the children
decreased their liking for the forbidden toy,

whereas none of the children decreased their
rating of the tabooed object following severe
threat or physical removal of the items. Although
threats varying in severity affected preference
ratings, it remains to be demonstrated whether

one could produce negative valuations of desired
activities by threats alone, regardless of their

severity.
Punishment is usually applied with the intent
of creating behavioral controls that will endure in
appropriate situations even when punishing
agents may no longer be present. Preference
changes alone are therefore of limited significance
unless it is also demonstrated that such changes

influence subsequent self-controlling behavior. In
a well-designed experiment by Freedman (1965)
groups of children were threatened with either
mild or severe punishment for touching an
attractive but forbidden toy. Half the subjects in

each treatment were provided with a brief period
free of surveillance during which they could
transgress without risk of punishment, whereas
the prohibiting experimenter remained in the
room with the other half of the subjects. Virtually

none of the children in any of the groups violated
the prohibition. Immediately after the session they

re-rated their liking for the different toys and,
approximately a month later, they were given an
opportunity under permissive conditions to play

with the forbidden object. Transgressive behavior
was lowest among children who received the mild
threat and refrained from deviating in the absence
of social surveillance. Children who were

threatened with severe punishment and complied
in the absence of the experimenter were more

inclined to handle the previously forbidden toy
and did not differ in this respect from subjects in
the condition combining low threat and social
surveillance. The results are somewhat difficult to

explain in terms of dissonance reduction because,
contrary to expectation and the findings of
Aronson & Carlsmith, the forbidden toy was not

devalued any more under mild sanctions than
under threats of severe punishment. The author

interprets the discrepant findings as indicating
that subjects did not select devaluation of the
forbidden object as the principal means of

dissonance reduction in this particular
experiment. This explanation may have some
validity, but it raises questions as to whether the
hypothesized dissonance processes could ever be
refuted empirically.
Results of the preceding study differ from a
large body of evidence based on experiments with

animals (Azrin & Holz, 1966; Church, 1963),
children (Parke & Walters, 1967), and adults
(Powell & Azrin, 1968; Rotenberg, 1959)

demonstrating that the degree of response
reduction is an increasing function of the intensity
of punishment. Mild punishment generally
produces little change in positively reinforced
performances; at intermediate levels aversive

consequences have partially suppressive effects;
while intense punishment typically results in large
and stable reductions in behavior.
The conflicting evidence may be interpreted in
several ways. In the above experiments, punishing
consequences were actually administered
contingent upon occurrence of transgressive
behavior, whereas the dissonance studies involved

a single verbal threat of punishment. A second and
more critical difference concerns the type of
behavior that is being controlled. In the
dissonance paradigm approach responses are

inhibited toward one of several positive
alternatives. Under these advantageous conditions
the instigation to transgression is apparently so
weak that a mild verbal threat is sufficient to
produce compliance in all subjects regardless of

whether the prohibiting agent is present or
absent. Given a response tendency of any strength,

transgressive behavior is ordinarily performed
more frequently in situations free of social

surveillance than when the disapproving agent is
physically present (Hicks, 1968). By contrast, in

studies of aversive control investigators either
select responses that are highly resistant to
change, or the behavior to be eliminated is
increased in strength and concurrently maintained

on a favorable schedule of positive reinforcement.
It is precisely under these types of circumstances

that the punishment is employed in everyday life.
In applications of aversive procedures one would
employ the minimum negative sanctions needed
to achieve adequate behavioral control.

Thereafter, control is usually maintained by
occasional negative consequences in weaker and
largely symbolic forms.
POTENTIAL PROBLEMS ARISING FROM AVERSIVE

CONTROL
Because of the varied and complex effects of
punishment, aversive control, particularly when

socially mediated, must be employed with care
and skill in programs of behavioral change. Many
of the undesirable consequences that may
accompany punishment are to some extent
preventable. Some of these common by-products,
and ways of minimizing them, are discussed next.
Generalization of Conditioned Inhibition.
Punishment is most often employed to reduce the
incidence of recurrent disturbing patterns of

behavior. The effects of aversive consequences,
however, are not specifically confined to the
responses that are negatively sanctioned. Severe

punishments, particularly if applied over a long
time, can result in broad generalization of
suppressive effects to socially desirable patterns
of behavior. Thus, for example, repeated harsh
punishment of aggression may not only eventually

eliminate the troublesome behavior, but stifle
assertiveness as well. The extent of generalized

inhibition arising from contingent punishment
varies inversely with the degree of similarity of

the new situations to those of the original aversive
training (Desiderato, 1964; Hoffman & Fleshier,
1961; Honig & Slivka, 1964). Large segments of
behavior are therefore most likely to be adversely

affected under conditions where punishment
contingencies are ambiguous, or where the
negative sanctions are applied to a wide range of
social responses in diverse settings.
Although at first conditioned inhibition

transfers broadly, if further punishment is
discontinued the generalization gradient gradually
narrows until eventually the subject suppresses
responding only to the stimulus context in which

his behavior was punished. Nevertheless, a study
by Hoffman, Fleshier & Jensen (1963) suggests
that some aftereffects may persist. When animals

were subjected to unrelated emotional stress
three years after behavioral suppression was
almost completely extinguished, the animals again
exhibited substantial inhibition in the presence of
the formerly aversive stimuli. These cues not only
partly regained, under general emotional arousal,
their previous capacity to suppress behavior, but

the original conditioned stimuli retained increased
potency for some time after termination of the
stress condition. Evidently, punishments
administered at an early period may sensitize the
organism to formerly aversive stimuli so that their

suppressive power is easily reinstated, at least
temporarily, by stressful experiences arising from
other sources.
Inappropriate or excessive generalization of

inhibitions and sensitivities can be easily
prevented by the use of discrimination
procedures, as demonstrated by Hoffman &

Fleshier (1964). During the development of
conditioned suppression animals were punished
for responding in the presence of a 1000 cycles
per second (cps) tone, but response to the adjacent
900 cps tone was never accompanied by shock.
Subsequent tests for generalization revealed that,
whereas generalization tones at or above 1000 cps

had strong suppressive effects, the discrimination
training produced very little behavioral inhibition
to stimuli on the lower side of the gradient (Figure
5-1). Moreover, when the animals were placed
under emotional stress after inhibitions were

largely extinguished, response suppression was
markedly enhanced in the presence of 1000 cps
and louder tones, but rate of responding to
auditory stimuli of 900 cps or lower remained
essentially unaffected. As depicted graphically in

Figure 5-2, discrimination procedures both
sharply curtailed the spread of inhibitory effects

and prevented certain generalized stimuli from
acquiring and retaining latent suppressive power
that could subsequently be reinstated by

unrelated stressful experiences.
Figure 5-1. Generalization of response suppression exhibited
by subjects after discrimination training during which a
1000 cps tone was paired with shock but a 900 cps tone
was never accompanied by shock. A ratio of 1.0
indicates complete suppression. Plotted from data of
Hoffman & Fleshier, 1964.
Figure 5-2. Effects of emotional stress on the generalization
of response suppression produced by prior
discrimination training. A ratio of 1.0 represents
complete suppression. Hoffman & Fleshier, 1964.
It would follow from the above findings that a
change agent who wishes to restrict the range and
direction of behavioral suppression should not

merely apply negative sanctions to undesirable
response patterns but should also arrange

different consequences for related forms of
behavior in different social contexts. For instance,
physically assaultive behavior may be punished
but desirable assertiveness rewarded. In addition

to selective reinforcement, discrimination is
greatly facilitated by the use of verbal aids. By
clearly labeling the modes of behavior that are

permissible and those that are punishable, and by
specifying the times and places at which certain

courses of action are appropriate or unsuitable,
greater specificity of punishment effects can be
ensured.
Emotional Conditioning. Another possible
accompaniment of aversive control, which has
been discussed at length in earlier sections, is the
capacity of punishment to endow formerly neutral
stimulus events with emotion-provoking

properties. Any behavior that serves to avoid,
remove, or postpone threatening stimuli reduces
emotional distress and is thereby automatically
reinforced even though the punishment
contingencies may no longer be in effect. These

inadvertently established fear elicitors often give
rise to avoidant patterns of behavior capable of
creating their own maintaining conditions. The
resulting avoidant responses may be more socially
undesirable than the behavior that punishment

was originally intended to reduce and, once
established, they may be considerably more

difficult to eliminate.
As we shall see later, not all forms of
punishment create conditioned emotional arousal.
Fear learning is most likely to accompany
procedures based on the social presentation of
painful stimuli. Negative sanctions involving
chiefly the removal of positive reinforcers

generally do not have fear-arousing effects. By
careful choice of punishment procedures one can
prevent or minimize the occurrence of undesirable
emotional conditioning.
Behavioral Inflexibility. In many instances

change agents are faced with the task of
discouraging patterns of behavior that are not
only permitted but expected at some later period
of life. Such problems are most likely to occur

when marked discontinuities exist in cultural
demands, as in the case of sexual behavior. Thus, a
child who has been severely punished for all

expressions of sexual curiosity may be rendered
anxious about sex and remain sexually inhibited in
later life when such behavior is socially approved
and expected of him. When marked temporal or
situational discontinuities exist, the use of social
training procedures that result in rigid and
inflexible behavior are contraindicated.
It is apparent from the results of an
experiment by Whiting & Mowrer (1943) that,
depending upon the manner in which it is
employed, punishment may be most efficient in
inhibiting behavior at the expense of later

adaptability to changing circumstances. Using a
socialization paradigm, Whiting & Mowrer
employed three methods to train animals to select
an effortful, circuitous route to food reward

instead of a considerably easier and direct path.
Whenever the easy route was chosen during
training one group of animals encountered a

physical barrier, the second group was denied
reward, while the third group was administered
an electric shock. The punished animals
abandoned the short passage most rapidly but
they also persisted longest on the effortful
tortuous route after the original negative
sanctions were removed. Lest the reader conclude

that behavioral rigidity is an unavoidable
concomitant of aversive control, it should be
emphasized that punishment combined with
discrimination training would likewise have
resulted in appropriately flexible responsiveness.

If, for example, short cuts were shocked in the
presence of a red light, but rewarded whenever
the light was absent, the animals would
undoubtedly have acquired discriminative
inhibitions and reverted rapidly to the easy route

after discontinuance of the socialization
contingency was signalled by permanent removal

of the environmental cue signifying punishment.
The transitory nature of behavioral
suppression induced through punishment,
frequently noted in laboratory studies, is not
surprising in view of the short periods during
which punishment is typically applied. However,
one finds limited but consistent evidence

(Solomon, 1964) that under certain conditions
even exceedingly brief punishment can result in
profound autonomic disturbances and lasting
inhibitions. A few shocks administered to an
animal during a consummatory response may

produce extremely powerful feeding inhibitions
(Lichtenstein, 1950), often leading to self-
starvation in the midst of plentiful food supplies
(Appel, 1961; Masserman, 1943). The factors
responsible for such extremely rapid and enduring

inhibitory learning have not as yet been identified,
but the timing of punishment appears to be an

influential variable. It is not possible, from the
limited data available, to determine whether the
extraordinary suppressive power of well-timed

punishment is primarily confined to
consummatory responses in subhuman species.
One would need to exercise considerable caution
in the use of aversive procedures if the latter

phenomena were replicated with instrumental
behavior performed by humans.
Avoidance of Punishing Agents and Situations.
With few exceptions, in laboratory investigations
of punishment effects, animals are confined to the

apparatus or their freedom of movement is
otherwise physically limited so that no escape
from the experimental situation is possible. Nor
are the subjects at liberty to choose whether they
will return to situations in which their behavior is

punished, or to cut short their stay whenever they
are displeased with inhospitable treatment
inflicted by their hosts. In naturalistic situations,
however, persons can generally avoid or restrict,
to some extent, contact with punishing agents and
social settings in which negative sanctions are
frequently applied (Azrin, 1958; Powell & Azrin,
1968). For this reason, even if punishment proved
to be highly effective in controlling behavior, it

might be of limited value under circumstances
where unpleasant events could be easily avoided.
The few studies permitting escape from
situations containing some aversive features
reveal that subjects will repeatedly withdraw to

nonpunitive situations even though the places of
refuge are less rewarding (Azrin, Halse, Holz, &
Hutchinson, 1965), require greater expenditure of
effort for a given reward (Dardano & Sauerbrunn,

1964), or offer no positive reinforcement at all
(Hearst & Sidman, 1961). Apparently withdrawal
from partly aversive situations is sufficiently

reinforcing to outweigh the effects of relatively
unfavorable rewarding conditions brought about
by escape behavior. Punishments at relatively low
intensities, although ineffective in suppressing
behavior, will nevertheless drive the subject out of
the situation (Azrin et al., 1965). This illustrates
the limitations of aversive controls when applied

in the absence of constraints or attractions to
check withdrawal.
To the extent that negative sanctions foster
fear and active avoidance of change agents, their
opportunities to influence the behavior of others is

reduced. This is a particularly serious hindrance
when widespread changes in attitudes and social
behavior are brought about most effectively
through unforced modeling. With restricted social

contact, there can be little identificatory learning.
It should not be assumed from these comments
that punishment invariably reduces spontaneous

modeling. Risley (1968) has shown that in the case
of children who are so preoccupied with bizarre,
self-injurious activities that they remain oblivious
to social stimuli, elimination of the hindering
behavior through punishment increases their
attentiveness and responsiveness to modeling
cues.
Escape tendencies can, of course, be
counteracted to some degree by increasing the
reward value of environments in which
punishments are periodically administered, and
by enhancing the positive valence of change

agents. A high level of positive reinforcement may
create psychological barriers against withdrawal
so strong that persons will remain in situations
that have punishing aspects as well. It should also

be noted that punishment based upon removal of
positive reinforcers, though temporarily
frustrating, nevertheless tends to maintain, and

may even increase, approach tendencies toward
change agents.
Negative Modeling. In social applications of
aversive control one must also consider the
modeling function of punishing behavior. In many
instances modeling cues furnished by disciplinary

actions are inconsistent with, and therefore
contravene, the effects of direct training. If, for
example, a parent punishes his child physically for
having struck a playmate, the intended outcome of

the punishment is that the child should refrain
from hitting others. Concomitantly with the
intentional training, however, the parent is
unwittingly providing vivid examples of the very
behavior that he is attempting to reduce in the

child. From fear of retaliation, the child may not
counter-aggress in his parent’s presence, but he

may nevertheless model his behavior after that of
the parent when the child must himself cope with

or control the behavior of others.
Consistent with modeling theory, Hoffman

(1960) found that mothers who forced compliance
with their demands through the use of power-
assertive techniques had children who exhibited
aggressive power-assertiveness in controlling the

behavior of their peers. Although the direction of
the causal relationship cannot be unequivocally
established from these data, results based on
laboratory studies (Bandura, 1965) clearly
demonstrate that aggressive patterns of behavior

can be readily transmitted through adult
modeling. In an experiment referred to earlier,
Mischel & Liebert (1966) found that negative
sanctions imposed on children were less effective
when the agent’s modeling behavior was

inconsistent than when the imposed and modeled
standards of behavior were congruent. These

findings indicate that anyone attempting to
control specific troublesome responses should
avoid modeling punitive forms of behavior that

may not only counteract the effects of direct
training but also increase the probability that on
future occasions the individual may respond to
interpersonal thwarting in an imitative manner.
EFFECTS OF PUNISHMENT ON CONCURRENTLY

REINFORCED BEHAVIOR
Punishment is frequently resorted to as a
method of social control when the positive

reinforcements maintaining troublesome behavior
cannot be identified, or, if known, cannot be
readily modified. It is considerably easier, though
less effective, for example, to punish the antisocial

behavior of delinquents than to remove the
subcultural contingencies that mold and control
their actions.
In most of the research reviewed earlier,
punishment was applied to responses after the

rewards maintaining them were removed, in order
to determine whether the addition of aversive
consequences accelerated the extinction process.
Of considerably greater significance are
investigations of the effects of punishment on
behavior that is concurrently maintained by

positive reinforcement, since the response
patterns that people frequently attempt to modify

result in some rewarding outcomes for the
performer. The available evidence generally
indicates that punishment does not have enduring
reductive effects on behavior that is

simultaneously being maintained by a favorable
schedule of positive reinforcement. Mild and
moderately punishing stimuli typically reduce the
occurrence of intermittently reinforced behavior,
but as punishment is continued subjects adapt to

the aversive consequences and exhibit some
recovery of responsiveness even while the
punishment contingency is still in effect (Azrin,
1959, 1960; Holz, Azrin, & Ulrich, 1963). Parents
who make frequent use of aversive controls would
undoubtedly provide corroborative testimonials
that, after a time, their punishments cease to have

much effect on their children’s behavior.
Severe punishment produces marked

reduction of intermittently reinforced behavior
with little or no recovery as long as responses
continue to incur aversive consequences. High

intensities of punishment are required, however,
to maintain behavioral suppression if punishment
is administered only intermittently, rewarded
alternatives do not exist, the behavior is strongly
established, and is concurrently supported by

highly favorable conditions of reinforcement
(Azrin & Holz, 1966; Boe, 1964).
PUNISHMENT AND AVAILABILITY OF ALTERNATIVE

MODES OF RESPONSE
Severe punishing consequences would have to
be applied over a long period to reduce behavior
effectively in persons who, because of their
restricted behavioral repertoires, possess few
alternative means for securing positive

reinforcement. The effects of response availability
on the suppressive power of punishment are

clearly illustrated in a laboratory experiment
conducted by Mowrer (1940), in which animals
were taught two modes of adjustment to an
aversive situation. One group first learned to sit on

the hind legs in order to reduce the intensity of
shock stimulation. After this response was
acquired, conditions were altered so that the
subjects could turn the shock off completely by
striking a pedal. Since the latter competing

response was considerably more effective it soon
became strongly established. A second group of
subjects learned only the pedal-pressing response.
When pedal-pressing was subsequently punished
in both groups, the animals that had learned a
single response continued to perform the
punished behavior for the entire duration of the

experiment, whereas all but one of the animals
that had available two different modes of

adjustment quickly reverted to the earlier pattern
of behavior.
Azrin and his associates (Herman & Azrin,

1964; Holz, Azrin, & Ayllon, 1963) similarly
demonstrated with adult psychotics that
punishment was generally ineffective when the
punished response constituted the sole means of
securing positive reinforcement. On the other

hand, when patients were provided an alternative
means of obtaining rewards, punishment
produced an immediate and complete reduction of
the undesired behavior.
The above findings suggest that persons who
have few response options will be slow to
abandon behavior that results in negative
outcomes. This is one reason that punishment is

ineffective in modifying the antisocial patterns of
delinquents and adult offenders who lack
alternative prosocial modes of response for
acquiring possessions that they value highly.

Under these conditions, punishment of antisocial
behavior is likely to lead offenders either to adopt
safer forms of illegitimate activities, or to alter
their techniques in order to avoid detection and
punishment on future occasions. An excellent

example of the way in which punishment results
in refinement rather than elimination of antisocial

behavior is provided in the autobiography of a
talented habitual offender.
My prison surroundings have been

completely a life apart, something so far
away that at times it was my real
circumstances that seemed so fantastic. In
between I went over jobs which I had pulled
off and mentally surveyed them to see how
they could be improved upon. Then I went
over my mistakes again, and learned how
they had occurred and let me down. So you
see there was always plenty for me to do
when I lay on that board with no occupation
but thinking. And plan future jobs. Oh yes, if
a survey could be taken it would be proved
that most of the big criminal jobs, and
thousands of small ones, are planned in gaol.
Planned to the last detail because there is not
sufficient alternative interest to occupy
prisoners’ minds [Hill, 1955, p. 39],
To the extent that refinements in deviant behavior
increase an individual’s confidence that he can
avoid detection and punishment on subsequent

occasions the behavior will most likely be
repeated.
BEHAVIORAL REDUCTION THROUGH POSITIVE

REINFORCEMENT OF COMPETING RESPONSES
The laboratory findings reviewed above
indicate that pre-existing modes of behavior are
likely to emerge when dominant response
patterns are reduced through punishment. Stimuli
that exercise some degree of control over different

types of responses can also be utilized in
conjunction with aversive procedures to elicit
weaker response tendencies. The temporary

suppression of disapproved behavior thus
provides an opportunity to strengthen either
emergent or actively elicited responsiveness.

When the desired alternatives do not exist in the
individual’s behavioral repertoire, positive
modeling and reinforcement procedures can be

effectively employed to establish and to
strengthen response patterns incompatible with

the maladaptive behavior.
The results of several experiments (Boe, 1964;

Holz, Azrin, & Ayllon, 1963; Whiting & Mowrer,
1943) have consistently shown that responses can

be rapidly and durably eliminated when the
behavior in question is punished and competing
responses are simultaneously rewarded. A
treatment program relying upon a combination of

aversive control of deviant responses and positive
reinforcement of desired responses may be most
effective for eliminating highly disturbing patterns
of behavior. When differential reinforcement is
applied to competing modes of behavior, the

punished responses may be counteracted by
either interfering responses elicited by the
aversive stimuli or competing behavior
maintained on the basis of positive reinforcement.
Under these circumstances, long-lasting

suppression is probably achieved less through
inhibitory responses established by punishment

than by the prepotency of rewarded alternatives.
This is suggested by findings from an experiment
by Timmons (1962), who compared the relative

efficacy of extinction, verbal punishment, and
counterconditioning of a competing response in
eliminating formerly appropriate verbal
responses. Counterconditioning proved most

powerful, but the addition of punishment to
counterconditioning contributed little to the

change process. However, this conclusion may not
hold true when undesired behavior is so strongly
established that few opportunities arise to reward
competing tendencies. It has also been shown that

even mild punishment, which has more
informative than inhibitory value, may facilitate
behavioral change to the same extent as severe

levels of punishment, provided that alternative
responses are concomitantly rewarded (Boe,

1964).
In a comprehensive analysis of punishment
effects Solomon (1964) has termed the
widespread belief that punishment is only a
temporarily effective controller of behavior a
legend springing from tenderheartedness and
sentimentalism. The findings reviewed in the
preceding sections, though not entirely consistent,

nevertheless provide considerable empirical
documentation that under a wide variety of
conditions the reductive effects of punishment
tend, in fact, to be impermanent. The conclusions
drawn from this large body of research cannot be

readily dismissed as legendary or spurious. It is
equally true that punishments administered in
conjunction with other procedures may produce
enduring changes in behavior (Beach, Conovitz,
Steinberg, & Goldstein, 1956; Boe & Church, 1967;

Storms, Boroczi, & Broen, 1963). Moreover,
innumerable studies have shown that tenacious

inhibitions and avoidance behaviors are created
when emotional arousal is conditioned to
environmental and self-generated stimuli through

aversive means.
It would appear from these diverse outcomes
that any sweeping judgment regarding
punishment effects results in self-contradiction.

One cannot contend that punishment is ineffective
because it has only temporary suppressing effects

and at the same time argue against its use on the
grounds that it produces behavior that is
unusually resistant to change. Similarly, self-
contradictions arise when studies in which

punishment results in self-starvation
(Lichtenstein, 1950) or needless inhibition of
effective means of securing reinforcement

(Whiting & Mowrer, 1943) are cited as evidence of
powerful aversive control of behavior, but

behavioral rigidity and other undesirable by-
products of punishment are treated as
unsubstantiated concerns.
Applications of Aversive Contingency

Systems
When aversive procedures are essential for
alleviating a detrimental condition, they are
generally viewed by adults as an unpleasant

though necessary part of the treatment,
comparable to the painful routines of physical

medicine, rather than as an interpersonal assault.
For this reason, patients rarely develop widely
generalized fears and hostile behavior toward
dentists and surgeons whose ministrations

initially produce highly distressing experiences.
Consequently, if presented in a treatment context,
aversive contingencies may have fewer adverse

side effects than when they are used dictatorially
to eliminate behavior that has high functional

value for the performer.
When aversive control is employed to modify
harmful social behavior the same punishing
consequences may be strongly resented or
willingly accepted depending upon the perceived
intent of the agent, and whether the sanctions are
applied mainly for his own convenience or for the
benefit of the recipient. Undesirable emotional

effects can be substantially reduced by arranging
in advance explicit contractual contingencies
which clearly define the broad limits of
permissible and punishable behaviors. Whenever
the undesirable behavior is performed the

aversive consequences should be applied
immediately, consistently, and in a matter-of- fact
way. If prearranged contingencies are
implemented in a nonpunitive fashion, the
negative sanctions will tend to be regarded by the

recipient as legitimate, predictable consequences
of his behavior, rather than as arbitrary and

vindictive reactions.
SPEECH DISORDERS
Aversive contingencies have been extensively
employed by Goldiamond (1965a) in both
experimental production of stuttering behavior
and its elimination. Before discussing the details of

this approach and its therapeutic efficacy, some of
the conditions that have been presumed to control
disfluencies will be reviewed. Learning
interpretations of functional speech disorders

generally conceptualize stuttering as avoidance
responses that are evoked by verbal stimuli in the
presence of potentially threatening situational
cues. The various explanatory schemes (Brutten &
Shoemaker, 1967; Shames & Sherrick, 1963;

Sheehan, 1958; Wischner, 1950) primarily differ
in the roles they assign to conditioned

emotionality and to positive and negative
reinforcement processes in the regulation of dis

fluencies.
Innumerable assessments have been made of

the personality characteristics of stutterers and
their parents, but observational studies of the
naturally occurring contingencies associated with
disfluencies are lacking. Based on retrospective

data from families of stutterers and nonstutterers,
Johnson (1942), who advocated a semantic-
learning approach, considered the following
conditions to be critical in the initial development
of speech disorders. All young children display

some repetitions of words, phrases, and syllables
without any accompanying emotional arousal or
self-awareness that their speech is defective
(Davis, 1939, 1940). However, parents of children
who later exhibit serious speech problems label

normal disfluencies as stuttering, to which they
then respond with increased vigilance, verbal

reprimands, and anxious remedial efforts. It is
assumed that, as a consequence of negative
evaluations and mislabeling of natural

disfluencies, anxiety reactions become
conditioned to the act of verbal communication.
Formerly effortless repetitions are now
accompanied by blockages, prolongations,

muscular tension, respiratory changes, and
compensatory facial and body movement. Once

disfluencies take on aversive properties they are
transformed into stuttering responses, and, like
other forms of avoidance behavior, become
capable of creating their own maintaining

conditions.
Results of laboratory studies that will be
reviewed later are not entirely consistent with the
above etiological formulation. Stimuli that have

been regularly associated with punishment can
have disruptive effects on speech, but negative
consequences made specifically contingent on
disfluencies generally reduce their occurrence
(Brookshire & Martin, 1967; Siegel & Martin,
1966; Quist & Martin, 1967). It would follow from
these findings that punishment might increase the
frequency of disfluencies only if the disruptive
effects of classically conditioned arousal outweigh

the reductive effects of punishing consequences.
Another important determinant of disfluencies,
which is usually either ignored or assigned a
secondary role in anxiety theories of stuttering, is
that solicitous parental concern made contingent

upon disfluencies can function as a powerful
positive reinforcer for such behavior. Parents may
thus inadvertently increase their children’s
disfluencies through their selective attention.
Familial modeling may also play an influential

role, either directly or indirectly, in the
development of disfluent speech patterns. Nelson

(1939) compared the incidence of stutterers in
filial, parental, and grandparental generations of
families of 204 stutterers and of a matched group
that manifested no speech disorders. Stuttering
appeared in more than one generation in only 2
percent of the nonstutterers, whereas the
incidence was 51 percent in the families of

stutterers. More direct evidence for the force of
example is provided by Van Riper (1937), who
found that stutterers not only displayed a higher
rate of disfluency following exposure to a
stuttering model as compared to a nonstuttering

model, but they even adopted some of the
idiosyncratic features of the stutterer’s verbal
behavior. Nelson argues for a genetic transmission
on the grounds that, in some cases, the parents no
longer stuttered or contact with disfluent

grandparents was limited. The genetic
interpretation may very well be valid, but an

explanation in terms of social learning is equally
tenable. Adults who themselves suffered from
speech disorders at an earlier period of their life,

or whose parents stuttered, would be prone to
respond with excessive concern to their children’s
natural disfluencies, thus increasing such behavior
even in the absence of modeled stuttering.
Sheehan (1958) and Wischner (1950) provide
the most detailed accounts of the negative

reinforcement mechanism presumed to maintain
stuttering responses. Their interpretations differ
mainly on the point in the speech sequence at

which reinforcement supposedly occurs.
According to Wischner, anxiety elicited by specific
words and situational cues results in momentary
blocking of a later portion of a verbal response in
an attempt to postpone or avoid anticipated social

disapproval, embarrassment or other negative
experiences. Stuttering behavior is believed to be
reinforced by virtue of its close temporal
juxtaposition with anxiety- tension reduction
accompanying successful completion of the word
on which difficulty was experienced. Although
stuttering generally produces negative
consequences as well, Wischner assumes that the
rewarding effects of immediate tension reduction

outweigh the inhibitory effects of temporarily
more remote punishment.
In his approach-avoidance conflict theory of
stuttering, Sheehan (1958) similarly regards
stuttering as an anxiety reducer, but he posits a

two-stage reinforcement process. Stuttering,
according to Sheehan, is a resultant of competing
urges to communicate and to avoid speaking.
Whenever the conflicting approach and avoidance

tendencies reach a point of equilibrium, the flow of
speech is interrupted or blocked. Momentary
inhibition of speech reduces the fear generated by

verbal communication which both reinforces
blocking and, by lowering the fear-motivated
avoidance gradient, releases the blocked word. In
addition, tension reduction following completion
of the word reinforces the preceding stuttering
responses, as well as any accompanying facial and
body movements designed to help restore fluency.
Brutten & Shoemaker (1967) consider
stuttering to be a phenomenon involving both
emotional and instrumental conditioning
processes, with the former assuming the more
influential role. According to this view,

disfluencies reflect the disruptive effects of
emotional arousal that have become classically
conditioned to certain situational and word cues
through unpleasant experiences. Several studies

have been published that provide some
supporting evidence for the influence of classically
conditioned arousal on disfluencies. Hill (1954)

found that students displayed disorganization of
speech in the presence of a light that had
previously been paired with shock stimulation.
Similar increases in disfluencies have also been
obtained during periods of emotional stress
produced by arbitrary punishment (Stassi, 1961),
or by exposure to stimuli that signaled the

occurrence of punishing events (Savoye, 1959).
While stuttering is conceptualized primarily as a
form of behavioral disorganization rather than an
avoidance response, it is assumed to include a
secondary instrumental component. That is,

stutterers adopt idiosyncratic phonatory,
articulatory, and resonatory modes of expression
designed to escape or forestall emotional
disturbances occasioned by disfluencies. These
adjustive responses, most of which take nonverbal

forms, are instrumentally reinforced by
subsequent word completion and attendant

reduction of distress. It would follow from this
explanatory scheme that stuttering can be
effectively eliminated only by extinguishing the

emotion-provoking properties of threatening
stimuli.
In view of the widespread belief that anxiety-

reduction is the primary maintaining mechanism
in functional speech disorders, it is somewhat
surprising that relatively few laboratory studies

have been conducted for the specific purpose of
evaluating this central hypothesis. The
reinforcement mechanisms adopted in the

theories discussed earlier are well suited to
account for unusual persistence of speech
disorders. If stuttering responses produce almost
instantaneous reinforcement through anxiety
reduction upon their occurrence, as assumed, one

would expect stuttering never to undergo
extinction unless special consequences were
somehow promptly interposed between the onset
of blocking and completion of the word. However,
a retrospective study by Sheehan & Martyn (1966)
showing that approximately 80 percent of college
students who had been stutterers eventually
achieved fluent speech suggests that recovery
from stuttering without special intervention

requires explanation just as much as does the
persistence of stuttering.
In addition to the meager experimental data on
the acquisition and maintenance of stuttering
responses, the anxiety-reduction theories

furnished no distinctive modes of treatment. An
early experiment by Sheehan (1951), however,
had therapeutic implications that were never
pursued. In this study it was assumed that the

point of reinforcement of stuttering responses is
the anxiety reduction following completion of the
word. Therefore, in an effort to prevent

reinforcement of stuttering responses, conditions
were so arranged that only fluent speech could be
instrumental in terminating each word in spoken
sentences. Adult stutterers read passages aloud on
different days under two conditions in
counterbalanced order. Under the control
conditions subjects merely read the material six

times in their usual way without arrangement of
any special consequences for disfluencies. In the
nonreinforcement treatment, however, subjects
read the passages aloud five times, except that
they were required to repeat each stuttered word

until they had said it fluently before proceeding to
the next word. Thus, stuttering prolonged rather
than terminated the attendant stress and tension.
On the sixth experimental trial and on the seventh
trial of both conditions subjects read the passages

as they normally would. Comparison of the
frequency of disfluencies between conditions and

successive readings (Figure 5-3) shows that
stuttering was substantially reduced and
remained significantly lower when it produced

negative consequences; whereas, under ordinary
Figure 5-3. Frequency of stuttering through successive
readings in experimental and control conditions.
Sheehan, 1951.
conditions, stuttering rate decreased slightly,
probably as a function of adaptation, but was
subsequently restored to its original level. Despite

the encouraging findings, the experimental
procedure was never extended to test its

therapeutic efficacy. Instead the author (Sheehan,
1954) advocated the same time-consuming
interview approaches that have proved of limited
value in altering other forms of deviant behavior.
In the preceding experiment any possible
decremental reinforcement of stuttering
responses occurring at the terminal point of the
verbal sequence was removed. The effectiveness

of reinforcement can be reduced if an interval of
time and other activities intervenes between a
response and its intended consequences. In
Sheehan’s study the temporal arrangement of
events was not the most favorable one for
eliminating disfluency; i.e., stuttering responses
eventually ended in word completions which, in
turn, were followed by negative consequences of
having to repeat the word on which the subject

encountered difficulty. Hence, successful word
completion was punished more strongly than
temporally remote repetitions and blockages.
Results of a study by Daly & Cooper (1967),
however, raise doubt that stutter-contingent

punishment would have achieved a greater
reductive effect. These investigators compared the
frequency of disfluencies under conditions where
shock was administered either during the act of
stuttering or immediately following completion of

each stuttered word. The stutter-contingent
system reduced disfluencies most, but the

difference was not of statistically significant
magnitude. In the method devised by Goldiamond
(1965a) for modifying chronic stuttering, which

will be reviewed next, each moment of disfluency
produces an immediate unpleasant consequence.
It is not entirely clear from Goldiamond’s

otherwise informative article what he considers to
be the critical conditions for the acquisition and
maintenance of stuttering responses. While

acknowledging that stuttering may involve an
emotional component, the anxiety-reduction
theory is summarily dismissed. Instead, numerous

anecdotal reports are presented to illustrate that
stutterers are generally subjected to lower
achievement demands, they are called on less
often to perform onerous tasks, their errors are
more likely to elicit sympathetic reactions from

listeners, they are provided more time to
formulate answers to questions, and they can
effectively command polite attention through
disfluency. These reports highlight both the
avoidance function of stuttering responses and
their value for drawing positive responses from
others.
Because of the nature of the contingencies
employed, findings from an experiment designed

to produce stuttering responses in a normally
fluent subject (Flanagan, Goldiamond, & Azrin,
1959) are of limited value in elucidating the
conditions under which stuttering is most likely
acquired. In this experiment, a subject received

continual shock; a verbal blockage, however,
turned off the shock for 10 seconds, and each
additional disfluency occurring during the interval
further delayed the aversive stimulation for a fixed

time. As might be expected, blockage rate
increased markedly to the point where the subject
almost completely averted the unpleasant

stimulation by continuous disfluent speech. While
this study demonstrates that it is possible to
induce deviant verbal behavior, it is exceedingly
improbable that parents of stutterers continuously
punish their children’s fluent verbal patterns, but
respond nonpunitively whenever their children
block and stutter. Indeed, the findings discussed

earlier indicate that the pattern of naturalistic
contingencies is probably the exact opposite of the
one imposed in the above study. We may grant
that experiments are not designed to reproduce in
every detail the stimulus events that occur in real

life, but we should require experimental
contingencies to bear some resemblance to social
reality if their findings are to have explanatory
value.
The studies previously reported suggest that

disfluent speech can be influenced to varying
degrees by at least three sets of controlling

conditions. First, threatening stimulus events can
produce speech disorganization through arousal
of anticipatory emotional responses. Second,
anxiety and tension reduction associated with the
completion of stuttering responses can serve to
reinforce them. And third, disfluency may have
some functional value in commanding attention

and lessening performance demands as suggested
by Goldiamond. A situational analysis of
disfluencies in children by Davis (1940), in fact,
disclosed that disfluent speech occurred most
frequently when children either were emotionally

aroused or wished to gain the attention of others.
Practically every form of psychotherapeutic
approach has been applied at one time or another
to speech disorders, with limited degrees of

success. Goldiamond (1965a) reports uniformly
favorable outcomes with stutterers by disfluency-
contingent punishment which, if confirmed by

more extensive assessments conducted over a
longer period, would represent a notable
therapeutic achievement.
In the first step of the procedure, stutterers are
instructed to read aloud from printed pages for
approximately 50-minute periods to provide a

baseline measure of stuttering. In this initial
assessment reading speed and frequency of

disfluencies are recorded. During the treatment
phase of the program reading rate is deliberately
slowed down and negative consequences are
made contingent upon the occurrence of stuttering

responses. Delayed auditory feedback of the
person’s own voice is utilized as the stutter-
produced aversive stimulus.
Pauses frequently occur as natural parts of

fluent speech and, consequently, an independent
observer would have to delay judgment as to
whether a particular hesitancy represented a
natural pause or a speech block. In order to ensure
that aversive consequences are immediately
contingent upon disfluencies, the client self-
administers the negative feedback on the basis of
his own response definitions. During oral reading
of material presented at a low rate the client

presses a microswitch for each word blocked,
which immediately shunts speech to a delayed
feedback device for a fixed time. This procedure
eventually establishes a slow stutterless pattern of
verbal behavior.
After the substitute fluent pattern is stabilized,

the client’s verbal behavior is progressively
modified in the direction of normal speech. This is
achieved by accelerating the reading rate in

graduated steps through mechanical control of the
material to normal or beyond baseline levels. At
the same time, delayed feedback is also gradually

faded out. For example, the delay period may be
reduced from an initial 250 millisecond duration
to 200, 150, 100, 50 milliseconds, and finally
eliminated completely.
A summary of the procedures and concomitant
changes in verbal fluency achieved with the first
subject to receive this form of treatment is

presented in Figure 5-4. During the baseline
period, which extended over the first 21 days, the

subject read approximately 110 words per minute
and stuttered on about 15 words per minute. In
sessions 22 through 33, he engaged in self-
definition of stuttering without any response-

contingent consequences. Beginning with the 34th
session, reading rate was lowered and stutter-
produced delayed feedback was introduced. At

session 47, time control of reading rate was
instituted, and several sessions later delayed

feedback was gradually faded out on successive
days. As shown in the figure, during the terminal
phase of the experiment the subject was reading

approximately 140 words per minute, well above
his previous baseline, while stuttering responses,
1037
Figure 5-4. Reduction in stuttering rate and establishment of fluent reading
in a chronic stutterer. Goldiamond, 1965a.
which ranged between 0.2 and 0.6 words per
minute, were almost completely eliminated.
Goldiamond reports even more dramatic
changes in verbal fluency (Figure 5-5) in a second
stutterer who, because of time limitations,

participated in a highly condensed version of the
standard procedure. Stutter-produced delayed
feedback combined with slow presentation of

reading material was introduced after three
sessions; shortly thereafter reading rate was
raised in successive stages, and the aversive
contingency was gradually removed. By the eighth
and final day the client was reading 256 words per
minute without manifesting a single disfluency.
Quantitative data are presented for eight
chronic stutterers. In each case fluent speech was
achieved and maintained in the laboratory
situation even when rate of verbalization was
1039
Figure 5-5. Elimination of disfluencies through stutter-produced delayed
feedback and development of rapid fluent reading. Goldiamond,
1965a.
increased and negative consequences withdrawn.
Goldiamond also reports concomitant
improvements in clients’ speech in naturalistic

settings, but apparently no objective assessments
were conducted. According to a later research
report (Goldiamond, 1965b), fluent patterns of
rapid reading, which endure under laboratory

conditions, have been established in 30 stutterers
within a remarkably short time. These preliminary
findings indicate that the procedures devised by
Goldiamond may have considerable promise.
However, evaluation of their therapeutic efficacy

must be deferred until more stringent tests are
carried out, and long-term follow-up results are

furnished.
Reports of changes in stuttering behavior must

be accepted with caution in view of evidence that
disfluency varies considerably as a function of the

social characteristics and the communicative
demands of different situations. A review of the
relevant literature by Bloodstein (1949) reveals
that stuttering responses are greatly diminished

under conditions of reduced requirements for
interpersonal communication. For example,
stutterers can generally sing and count fluently,
they can speak smoothly when acting, when
imitating another person’s verbal style, or when

alone. For this reason attainment of rapid
stutterless reading by a person alone in a booth
does not constitute a powerful test. Disfluency is
likewise decreased in situations where negative
social reactions from listeners are minimized.

Thus stutterers may experience little verbal
difficulty when speaking to persons much younger

than themselves, to audiences regarded as socially
or intellectually inferior, or to persons over whom
they exercise authority. These findings suggest

that stringent tests of fluency would require social
communication rather than oral reading in
evaluative situations involving intimidating
audiences. Tape recordings of speech under these

types of conditions can be obtained without
difficulty.
When behavior is modified in clinic offices or
laboratories the problem of insufficient transfer of
changes to everyday situations frequently arises.

Goldiamond has attempted to overcome this
difficulty in the elimination of stuttering by
utilizing additional procedures specifically
designed to aid generalization of fluent speech to
the natural social environment. Speech exercises

are prescribed for the client to perform at home.
Essentially these involve reading for brief periods
in the slow stutterless manner developed in the
laboratory, interspersed with rapid oral reading.
The method devised by Goldiamond follows an
exclusively operant approach. If all stutterers who
receive this form of treatment are able to converse
fluently, even under stressful social conditions,

then obviously no additional procedures are
required. On the other hand, if their speech
performance is substantially improved but they
continue to exhibit some disfluencies in certain

emotion-arousing situations they could benefit
from a program of desensitization. Some case data
have been published to suggest that disfluencies
controlled by conditioned aversive stimuli can be
successfully eliminated by extinguishing

emotional responsiveness to such threats.
An illustrative case is provided by Walton &

Mather (1963) in the treatment of a 40-year-old
architect who suffered from an articulatory

disorder characterized by speech blocks
associated with teeth grinding. The controlling

stimuli for the client’s disfluencies included
situations in which he was required to convey
specific information, particularly in the presence
of strangers, supervisors, and persons whom he

wished to impress. Initially a “speech shadowing”
technique was employed in which the therapist
read passages from a book and the client matched
his verbalizations without seeing the text.
Although this method proved highly effective in

producing fluent speech in the treatment situation,
the stammering nevertheless persisted in natural
contexts. The second treatment strategy was
aimed directly at neutralizing the stimulus
determinants of the speech blocks. At the

completion of a desensitization treatment, in
which stammer-generating situations were

repeatedly paired with relaxation, the client was
able to converse fluently, even when
communicating specific information to persons in

authority. Because of the multiple determination
of disfluencies, a treatment strategy combining
reinforcement procedures that restore fluent
speech with methods that can extinguish the

arousal potential of stutter-eliciting stimuli would
prove most effective in eliminating disfluencies

under stressful social circumstances.
CONTROL OF SELF-INJURIOUS BEHAVIOR
One of the most perplexing and dangerous

disorders, which is especially prevalent among
schizophrenic children, involves self-injurious
behavior. In its more extreme forms, children
pummel their faces repeatedly, they bang their

heads forcibly against hard or sharp objects, they
tear and bite off pieces of flesh from their bodies,
or they exhibit some other type of self-mutilating
behavior. Because of the serious risk of permanent
physical injury, such children are usually kept

continuously in physical restraints.
The maintaining conditions of self-injurious
behavior are not fully understood, but several
experiments have shown that it is amenable to

control through variation of reinforcing
consequences. Lovaas, Freitag, Gold, & Kassorla

(1965) found that the self-injurious responses are
readily cued off by stimuli signifying withdrawal
of social reinforcement for other behaviors and
that they tend to increase in frequency and

intensity when social reactions are made
contingent upon their occurrence. Demonstrations
that self-injurious behavior can be reduced

through reinforcement of physically incompatible
responses and increased by extinction of

competing activities cast little light on the
variables that control this behavior. Of much
greater interest are the changes produced by

variation of the contingencies applied directly to
self-injurious behavior.
It has been demonstrated (Bucher & Lovaas,
1968; Lovaas, et al., 1965) that physical affection,
and sympathetic and reassuring comments made

to children whenever they engage in self-injurious
behavior, leads them to respond even more self-

destructively, whereas such behavior is promptly
reduced to its original baseline level when positive
reactions are discontinued. These findings suggest
that self-injurious behavior may be partly

maintained by its social consequences, though the
evidence is conflicting on this point. Subtle
changes in social reinforcement, such as an

experimenter remaining with a child without
attending to him, does not seem to affect the rate

of self-injurious responses (Lovaas et al., 1965;
Risley, 1968). More complete withdrawal of social
interaction for a brief period whenever children

behave self-injuriously reduces such activities, but
occasionally they are unaffected even by these
more powerful consequences (Risley, 1968).
Whatever the controlling variables might be in any
given case, preliminary findings indicate that
chronic and intractable self-injurious behavior can

be successfully eliminated with beneficial effects
by brief application of an aversive contingency.
Punishment is generally used as an adjunct to
either extinction procedures or differential

reinforcement of competing response patterns.
Brief social withdrawal made contingent on self-
destructive behavior is usually sufficient for its
permanent removal (Hamilton, Stephens, & Allen,
1967; Lovaas et al., 1965; Wolf, Risley, & Mees,

1964). In some cases, however, when self-
mutilating behavior is first placed on extinction it
may temporarily increase in intensity, which could
be potentially dangerous. These responses can be
promptly and completely eliminated by contingent
application of a few painful shocks. In one of
several cases reported by Bucher & Lovaas (1968)
a seven-year-old schizophrenic boy who had been
self-injurious since the age of two performed

approximately 3,000 self-pummelling responses
during a period of 90 minutes when his physical
restraints were removed. This behavior was
almost totally eliminated in four sessions by the
use of 12 contingent shocks. Self-destructive

behavior that had persisted over a period of six
years was also rapidly and durably removed in a
schizophrenic girl after she received a total of 15
shocks for beating her head (Figure 5-6). In each
case reported, contingent aversive consequences

not only removed self-mutilating behavior, but the
children whined less and they were much more

inclined to attend to the therapists.
Tate & Baroff (1966) similarly achieved quick
control over chronic self-injurious behavior in a
partially blind psychotic boy through punishment
Figure 5-6. Frequency of self-injurious behavior and amount
of avoidance and crying displayed by an autistic child
during pre-treatment sessions (1 through 15), and
when such behavior was punished by shock (P) or a
verbal reprimand “no” (N) during sessions 16, 17, 19,
and 21. The numerals below the session numbers
identify the therapist present during the session.
Bucher & Lovaas, 1968.
procedures. The behavior, which he had exhibited
continuously for five years, included vigorous
head-banging, face-slapping, and self- kicking.

Because physical contact with people was highly
reinforcing to the boy it was employed as the
consequent event in the first phase of the
treatment program. A series of daily walks was
arranged in each of which the therapist removed

his hand from the child’s grasp and ceased talking
whenever a self-injurious response occurred;
physical contact was reinstated if the boy did not
hit himself for a period of three seconds. As can be
seen in Figure 5-7, the time-out contingency

produced a dramatic reduction in self-injurious
behavior.
In the second phase of the program, response-
contingent shock was used to eliminate the
remaining head-banging that threatened further
damage to his eyes. It was explained to him that if
Figure 5-7. Daily average frequency of self-injurious
responses performed per minute on control days when
such responses were followed by no special
consequences, and on experimental days when self-
striking behavior produced brief withdrawal of
physical contact. Tate & Baroff, 1966.
he continued to hit himself he would receive
painful shocks. The aversive contingency,
combined with verbal praise and affectionate

reactions for desirable behavior, permanently
eliminated the self-injurious responding. Whereas
previously the boy had been physically restrained
in bed, after the treatment program was
completed he participated freely in daily activities

with increased enjoyment and spontaneity.
Risley (1968) provides a detailed report of a
case in which social consequences were totally
ineffective in decreasing self-injurious behavior. It

involved a markedly deviant 6-year-old girl who
suffered many serious body and facial injuries
from hazardous climbing activities which she

engaged in continuously. A program of differential
reinforcement, wherein climbing behavior was
ignored and incompatible responses were
rewarded, failed to produce any significant
changes. Unlike the cases cited above, brief
physical isolation contingent upon dangerous
climbing behavior also proved ineffectual. After

these other methods failed, shock punishment was
applied in conjunction with a verbal reprimand.
Administration of several shocks, and later verbal
reprimands alone, completely eliminated
inappropriate climbing when the experimenter

was present, but there was no noticeable decrease
in this behavior at home. When the mother
subsequently applied contingent shock at home,
perilous clambering promptly declined from an
average rate of 29 to 2 per day within a few days.

Thereafter, the behavior was successfully
controlled by having the child sit in a chair for a

brief period after each instance of climbing.
It is noteworthy that surprisingly brief
programs of contingent shock and reinforcement
withdrawal are not only effective in removing self-
injurious behavior of long standing, but they
generally improve social functioning as well.
These related changes are most clearly revealed in

the findings reported by Risley (1968). After
disruptive self-injurious behavior is removed,
children become more attentive, socially
responsive, and display increased imitativeness,
thus enabling them to acquire new patterns of

behavior. If further studies support this
conclusion, there is no justification for having
children mutilating themselves or spending their
early years uselessly in physical restraints.
MOTOR DYSFUNCTIONS
In one of the early applications of aversive
contingencies, Liversedge & Sylvester (1955;
Sylvester & Liversedge, 1960) treated 39 cases of
writer’s cramp with a procedure employing
response-contingent shock. In the majority of
cases, tremors and spasms of hand muscles were
elicited only by highly specific writing stimuli, but
the same muscle groups were unaffected when

involved in nonwriting situations. As has been
found in other types of deviant behavior, the
individuals exhibiting this occupational
impairment shared no common psychological

characteristics, suggesting that specific
reinforcement contingencies rather than
psychodynamic factors were the critical
determinants. It is therefore not surprising that a
number of these clients who had undergone

various conventional forms of psychotherapy
experienced little or no amelioration of their “craft

neurosis.” Consequently, Liversedge & Sylvester
explored the efficacy of aversive consequences for

altering each component of the physical disorder.
In order to remove tremors, one element of the

motor disability, the clients were required to
insert a stylus into a series of progressively
smaller holes; each time the stylus made contact
with the side of the hole it resulted in a shock. The

removal of the spasm component of the disorder
was obtained in two ways: First, the clients traced
various line patterns (similar to the movements
required in writing) on a metal plate with a stylus,
and any deviation from the path produced a shock.

Following training on the apparatus, they then
wrote with an electrified pen which delivered a
shock whenever excessive thumb pressure was
applied. In treating typists’ cramp a small
electrified pad was attached to the palm of the

hand so that whenever the fingers contracted into
the palm a shock was delivered.
The authors report that after 3 to 6 weeks’
training, writing of satisfactory quality was
restored in 24 of the cases; the clients were able to
resume work, which often involved writing for
extended periods, and follow-up studies
conducted up to four and a half years later
disclosed that the improvement was being

maintained. Five clients responded favorably to
the treatment but subsequently experienced a
recurrence of muscular dysfunction, while 10
cases showed no improvement. The failures had
exhibited the motor disability over a longer period

(6 to 21 years), which may partly explain why they
were less responsive to the treatment.
The procedures devised by Liversedge &
Sylvester are sufficiently effective to merit

controlled studies to isolate factors responsible for
the success of this treatment approach. Since the
technique involves both response guidance and

punishment of spasmodic and tremorous
responses, it is conceivable that guided retraining
with nonaversive feedback may in itself effect
changes. In addition, information is needed
regarding the types of individuals who achieve
lasting benefit from application of aversive
contingencies alone, those who require

supplementary or alternative forms of treatment,
and the cases for whom aversive procedures are
contraindicated.
It would appear, from the findings of
Liversedge & Sylvester, that aversive

contingencies are most successful in modifying
motor dysfunctions controlled directly by stimuli
in writing situations that do not have strong
emotion-provoking value. On the other hand, in

instances where muscular tremors and tension
reflect the disruptive effects of high emotional
arousal, punishment may further augment the

negative valence of writing situations and thereby
exacerbate the condition. In this connection, Beech
(1960) has furnished some preliminary
experimental evidence that extinction and
desensitization methods may prove successful
with highly anxious individuals whose writing
disturbances are unimproved by contingent

negative consequences.
In one case, for example, the person had
received a full year of psychoanalysis, a year of
hypnotherapy, and 10 sessions of aversive
conditioning without any amelioration of a

writer’s cramp of five years’ duration. Whenever
he grasped a pen the index finger would contract
and the wrist would bend sharply, causing severe
pain, fatigue, and immobilization of the hand. The

client participated in seven extinction sessions,
each of which consisted of massed evocation of
effortful finger contraction, to the point at which

he could no longer retract the finger. Writing tests
administered following each extinction period
disclosed not only a rapid and progressive decline
in the incidence of finger contraction, but also a
similar decremental change in the untreated arm
spasm. Moreover, on the day after the final
experimental session the client was able to write

24 lines with only one instance of hand spasm.
Similar, though less stable, results were
obtained by extinction involving repeated
response evocation with a second client, both of
whose hands were severely immobilized.

Attempts at writing produced intense painful
muscular contractions and, at best, illegible
scribbling; his left hand, which he had previously
used in typing, was clenched tightly into a fist

which he could open for only a few seconds at a
time. In an attempt to extinguish fist-clenching the
client repeatedly performed effortful hand-closure

responses on a dynamometer until he could no
longer squeeze the handle of the apparatus. After
33 extinction sessions he was able to open his
hand for an indefinite period; however, thoughts
about his work, and stressful interpersonal
experiences, typically reinstated the motor
disorder. Since writing situations elicited

considerable anxiety in the client, the tremors and
spasms associated with the act of writing were
treated by the standard desensitization procedure,
in which imagined writing activities were
progressively performed in the context of

relaxation. Tremorless writing of satisfactory
quality was temporarily restored in this manner,
only to be followed by repeated relapses.
In the above case the motor dysfunction was

apparently controlled, in large part, by distressing
vocational and interpersonal events. Had
emotional arousal to the latter situations also been

neutralized, it is possible that more enduring
changes would have been achieved. This type of
approach was successfully utilized with a foreign
student who was unable to take lecture notes due
to marked hand tremors and attendant fatigue.
After participating in a brief desensitization
program in which imagined and actual writing

situations were presented in conjunction with
relaxation, the student was able to write in a rapid
and relaxed manner without manifesting any
muscular disturbances. Announcement of a final
examination, however, re-established the

tremorous responses, though in less marked form.
A second series of desensitization sessions,
directed toward examination situations,
effectively eliminated the student’s writing
disability.
The case studies reported by Beech were
primarily designed to demonstrate that writing

disorders which apparently are under aversive
stimulus control and may therefore be
exacerbated by punishment can be modified under
laboratory conditions by other methods. While
these findings have suggestive value, more
extended applications and assessments of these
procedures are necessary to establish their utility

for eliminating occupational motor disorders.
A study by Barrett (1962) provides a
laboratory demonstration of reduction in tics by
the use of automatically programmed response
consequences. The client, a 38-year-old

accountant, suffered from extensive multiple tics
that proved refractory to psychotherapeutic and
pharmacological treatments. According to the
client’s report, the tics developed after a

frightening experience in the army when he awoke
one night with a choking sensation accompanied
by a momentary inability to breathe or swallow. At

the time of the study, his motor pattern included
contractions of neck, shoulder, chest, and
abdominal muscles, head-nodding, eye-blinking,
mouth-opening, other facial movements, and
swallowing difficulties.
During the experimental sessions, the tiqueur
was seated comfortably in an armchair designed

so that spasmodic movements produced by the tic
were automatically recorded, and activated the
contingency controller. The contingency
arrangement was programmed so that each tic
produced either a brief interruption of music, or

white noise. Music was chosen as the positive
stimulus since the client was a part-time musician
and the interruption of recorded jazz concerts that
he had personally selected could serve as an

adequate negative event. The effects of the
withdrawal of music and the presentation of noise
were compared with the client’s deliberate efforts

to control his tics. Results based on eight sessions
showed that, while the rate of tics could be
reduced somewhat by self- control, tic-produced
white noise, and contingent music, the most
dramatic and stable decrements resulted from tic-
produced interruption of music. The latter
procedure reduced the rate of tics from 55 to 85

percent below the baseline level during the
various sessions, but the question remains
whether continuance of the mild punishment
contingency for a longer period could eventually
eliminate or markedly reduce the occurrence of

tics in naturalistic settings when negative
feedback is absent. The author suggests that in
therapeutic applications of this method, the client
might be able to modify motor and other
disturbances under his own direction if he were

furnished with a portable contingency controller
to plug into a home radio, television, or record

player. The intricate apparatus could be dispensed
with, however, if comparative studies disclosed
that a simple extinction procedure involving

repeated nonreinforced evocation of tics was
equally or more effective than methods utilizing
response-contingent punishment.
SEXUAL DEVIATIONS
Chapter 8 describes classical conditioning

procedures designed to eliminate sexual
aberrations by endowing stimuli that elicit the
behavior with aversive properties. Some attempts

have been made to bring sexually deviant behavior
under control through response-contingent
aversive stimulation. Feldman & MacCulloch
(1964, 1965) provide a detailed account of a
treatment method, primarily based on an

avoidance conditioning paradigm, that they have
developed for the modification of homosexuality.
Clients are asked initially to rate the
attractiveness of an extensive series of slides
depicting both clothed and completely nude males.

A similar hierarchy of slides of females varying in
attractiveness is prepared. In order to facilitate
transfer effects, whenever possible photographs of
males and females with whom the client is closely
acquainted are used.
In the avoidance conditioning procedure, a
picture of a male is presented on a screen in a

darkened room and the client is instructed to
leave the picture on the screen as long as he finds
it sexually attractive. The client is informed that he
might be administered shocks during the viewing

periods, but that he can depress a switch that
simultaneously terminates the picture and the
aversive stimulation. If he turns off the slide
within 8 seconds, shock is avoided; if, on the other
hand, he continues to view the picture beyond the

designated period he is administered an
unpleasant shock through electrodes attached to
his leg.
On the assumption that variable training
procedures will produce avoidance responses that
are highly resistant to extinction, the occurrence

of negative consequences is varied randomly with
respect to their frequency and timing.

Unpredictable aversive contingencies can produce
stable rates of avoidance responses in the
treatment situation, but it is questionable whether
they will have much effect on extinction in

everyday situations. The reason for this, which
will be explained more fully later, is that clients
can easily recognize that the arbitrary punishment

contingency employed in treatment is completely
absent in extratherapeutic situations. Under

conditions where the schedules of reinforcement
in different situations are highly distinguishable,
the partial reinforcement effect would not be

expected to carry over to the new context. In any
event, during the treatment series, one-third of the
client’s avoidance responses to pictures of males
are reinforced by shock termination; on one-third

of the trials aversive stimuli are administered even
though the client performs appropriate avoidance
responses within the allotted time; and on the

remaining trials termination of the picture is
delayed for varying periods of time after the
avoidance response has been performed.
In addition to establishing avoidance

responses toward males, an effort is made to
condition anxiety relief properties to females by
introducing slides of females contiguously with
the removal of pictures of males. To further

increase approach tendencies toward females, the
client can request return of a slide of a female after
it has been removed and thus postpone aversive
experiences periodically associated with male
pictorial stimuli. The presentation and removal of
the slides of females is controlled by the therapist
in order to prevent any inadvertent reinforcement
of avoidant behavior toward females.
Initially, the slide of the least attractive male is

paired with that of the most fetching female. A
particular male slide is displayed repeatedly until
the client reports indifference or a dislike for it,
and, in addition, turns off the slide within a second
or two; the next feminine picture is introduced

when the client consistently requests the return of
the preceding feminine item. This same process is
repeated with succeeding pairs of stimuli in the
pictorial hierarchy. A typical session involves

completion of about 30 trials and requires some
20 minutes to conduct. The treatment series,
which averages about 15 sessions, is continued

until a client exhibits a clear change of sexual
orientation. In addition, the client returns for
about 8 or 10 “booster” sessions during the year
following completion of the program.
Feldman & MacCulloch (1965) present
preliminary outcome data for 19 unselected
chronic homosexuals treated by this method.

Three cases discontinued therapy; 10 out of 12
clients under 40 years of age, and 1 out of 4 cases

over 40, achieved marked changes in their sexual
orientations. According to the authors, the clients’
interests in men have greatly diminished and
homosexual practices have been virtually

eliminated, whereas heterosexual interests,
fantasies, and behavior have been substantially
increased. Moreover, in all but one case the

increased heterosexual behavior noted
immediately after treatment has been either

maintained or augmented during follow-up
periods ranging from 2 to 14 months.
MacCulloch, Feldman, & Pinshoff (1965) also
recorded response latencies and physiological
concomitants of avoidance conditioning for a small
subsample of cases to determine if response
measures obtained during treatment have value in
predicting post-therapy sexual behavior. Clients

who achieved and maintained a heterosexual
orientation displayed progressively shorter
avoidance response latencies to homoerotic
stimuli; they showed strong approach responses
to feminine photographs as therapy progressed,

and they exhibited conditioned autonomic
responses to pictures of males. In contrast, those
who reverted to homosexual practices exhibited
considerable irregularity in response latencies,
weak approach tendencies toward feminine

stimuli, and little or no autonomic conditioning.
These results, while most interesting, and

consistent with the correlates that one would
expect for successful avoidance learning, must be
confirmed on a larger sample before their

predictive value can be adequately evaluated.
Feldman & MacCulloch attribute the
therapeutic outcomes primarily to the avoidant
component of the treatment procedure. The

behavioral changes also reflect the effects of
classically conditioned aversion to male stimuli

and stress-reducing value to feminine cues. The
highly favorable outcomes reported by Feldman
and MacCulloch are particularly striking when one
considers that there was no biased selection of

cases, and that supplemental programs designed
to develop adequate heterosexual repertoires of
behavior were rarely employed.
Bond & Evans (1967) successfully eliminated

underwear fetishism by contingent application of
aversive consequences in two boys who
repeatedly raided clotheslines for women’s
undergarments. The boys were presented in
random order 20 fetishistic and 20 neutral objects
and intermittently shocked while removing the
fetishistic items. The authors report that after
several sessions the boys lost their interest in
female underwear and permanently discontinued

their clothesline forays. The methods employed in
the above studies appear to hold some promise for
modifying deviant sexual behavior, but full
evaluation must await controlled studies.
Removal of Positive Reinforcers
Aversive consequences in the form of physical
punishment are seldom employed as methods for

controlling behavior in naturalistic situations.
Removal of positive reinforcers, on the other hand,

is a very common mode of aversive control
(Bandura & Walters, 1959; Sears, Maccoby, &
Levin, 1957). This method is exemplified by
negative sanctions in which persons are deprived
for a time of rewards and privileges that are
ordinarily available, such as use of television,
automobiles, or certain facilities; they are
temporarily removed from rewarding situations;
they are restricted from going out or participating

in enjoyable activities; or they are temporarily
forced to relinquish other objects and pastimes
they value. Monetary penalties are also frequently
levied as punishments, particularly with adults.
The removal of positive reinforcers as a

punishment technique should be distinguished
operationally from extinction procedures,
although both methods may reduce responding
through some common processes. In extinction,

consequences that ordinarily follow the behavior
are simply discontinued; in punishment, behavior
results in the application of aversive consequences

through forfeiture of positive reinforcers. Thus, in
extinguishing aggression sustained by peer
attention, the behavior is consistently ignored;
under the punishment contingency, however, the
rewards of peer attention are pitted against the
negative effects of confinement to one’s room, loss
of television privileges, or some other type of

negative outcome. As is true of other forms of
aversive control, the amount of behavioral
reduction produced by punishment through
removal of positive reinforcers will depend,
among other factors, upon the relative magnitude

of the opposing consequences. The most effective
way of permanently eliminating undesirable
responses, of course, would be to remove the
positive conditions maintaining the behavior. In
many situations, however, the rewards dispensed

by others cannot be easily controlled. Under these
circumstances negative sanctions may be

effectively employed to reduce deviant response
patterns and to hold them in check while
alternative modes of behavior are being

established and strengthened.
EXPERIMENTAL STUDIES OF PUNISHMENT BY
REINFORCEMENT WITHDRAWAL
Laboratory studies of the efficacy of

punishment by reinforcement withdrawal have
been largely confined to procedures in which the
occurrence of selected responses produces a loss
or temporary time-out from positive
reinforcement that was previously available. A

number of these experiments were primarily
designed to determine whether time-out
punishment functions as an aversive stimulus
analogous to shock stimulation or other physically
hurtful events (Azrin & Holz, 1966; Leitenberg,

1965). The findings generally show that
reinforcement time-out consequences can produce

durable reductions in responses if an alternative
mode of behavior is available to the subject (Holz,

Azrin, & Ayllon, 1963), or the contingencies
maintaining the behavior have been removed
(Baer, 1961; Nigro, 1966).
Results of an illustrative experiment by Baer
(1962), however, disclose that reductive effects

are transitory in the case of behavior that
continues to produce powerful self-reinforcing
consequences. Young boys who displayed
persistent thumbsucking were shown cartoons.
Both boys watched the films together, but

interruption of the cartoon was made contingent
upon thumbsucking for one of the subjects, whose
response brought noncontingent film stoppages
for the other boy. Halfway through the experiment

their roles were reversed. Contingent interruption
of enjoyable cartoons produced a marked
decrement in thumbsucking, but no decrease

occurred when punishment was randomly related
to the behavior. However, during periods when
the punishment contingency was suspended the
boys promptly reverted to sucking their thumbs.
The preceding studies demonstrate that brief
reinforcement withdrawal can function
analogously to an aversive stimulus in reducing

behavior. Similar reductive effects are achieved by
punishment through response-cost contingencies

in which monetary points are lost each time
negatively sanctioned behavior is performed
(Elliott & Tighe, 1968; Weiner, 1962). Laboratory
investigations of other behavioral effects of

reinforcement withdrawal provide further
evidence that it possesses some of the functional
properties of an aversive event. It has been shown

(Ferster, 1958; Morse & Herrnstein, 1956;
Zimmerman, 1963) that behavior which prevents

or terminates reinforcement withdrawal is
effectively maintained in much the same way as
avoidance responses are sustained by their

success in preventing the occurrence of physically
painful consequences. Moreover, neutral stimuli
that are regularly associated with reinforcement
withdrawal tend to acquire negative properties
(Ferster, 1960; Ferster, Appel, & Hiss, 1962).

Unlike the effects of physical punishment,
however, stimulus events that signal the advent of

reinforcement withdrawal do not seem to
generate disruptive emotional arousal
(Leitenberg, 1965).
Several attempts have been made to compare

the relative power of reinforcement withdrawal
and presentation of physically aversive stimuli in
reducing selected response patterns. Tolman &
Mueller (1964) employed different types of

punishers with a young rhesus monkey who
developed a marked affinity for one toe,
commonly sleeping with it in his mouth and
sucking it while climbing about on two hands and
one foot. Since the monkey had mainly interacted
with humans (monkey-rearing practices would
probably not have produced a chronic toe-sucker),
visual contact with people (viewed through a
small window) served as the positive reinforcer.

Punishment periods, in which the window was
closed whenever the monkey placed its toe in its
mouth and was opened upon removal of the toe,
alternated with nonpunished periods during
which the window remained continually open. The

second phase of the experiment followed a similar
procedure except that window-closing was
replaced by unpleasant sounds at the onset of toe-
sucking; the sounds terminated as soon as the toe
was removed. During noncontingent punishment

the aversive sounds were simply presented at
periodic intervals without regard to the animal’s

behavior. Punishment through aversive sound
stimulation produced a marked reduction in toe-
sucking; recovery was delayed, but once the

response recurred it was emitted at a relatively
high rate. On the other hand, punishment by the
removal of social stimuli resulted in less abrupt
decrement; recovery was more gradual and less

complete.
Comparative data are difficult to evaluate

when based on an experiment with a single
subject, in which potency of the punishing stimuli
was not equated, and possible order effects were

not controlled. However, the findings of Tolman &
Mueller are essentially corroborated by McMillan
(1967), who assessed the relative efficacy of
contingent shock and temporary withdrawal of
rewards in eliminating a concurrently reinforced

response. Both types of punishers reduced
responding to about the same degree, but the
time-out procedure was associated with less
behavioral recovery.
The adjunctive use of punishment by
reinforcement withdrawal has certain advantages
over physically aversive procedures. As previously
shown, aversive interventions may arouse fear

and avoidance of punishing agents, and thus
weaken their potential influence. In contrast,
methods that chiefly involve the removal of
positive reinforcers not only generate much

weaker emotional effects, but they tend to foster
and maintain orientation toward the agents who
control the desired positive resources. If
restoration of the positive reinforcers is made
contingent upon performance of alternative

modes of behavior, rapid behavioral changes may
in fact result.
BEHAVIORAL CONTROL BY WITHDRAWAL OF

POSITIVE REINFORCERS
Reinforcement withdrawal has proved to be an
effective means of managing deleterious behavior

that often impedes the person’s own development
and seriously infringes on the well-being of others.
If combined with methods that foster constructive
alternatives this form of behavioral control can aid

achievement of enduring changes in social
behavior.
In applications of time-out procedures,

behaviors that are considered unacceptable and
the consequences that they will produce are
clearly explained in advance. When social

exclusion is employed as the negative outcome, as
is usually the case, each transgression results in
brief social withdrawal that is carried out
immediately, naturally, and in a firm but
nonhostile manner. If, during the time-out

interval, the person continues to display
obstreperous behavior, the period of exclusion is

extended until cessation of the behavior. Under
this type of contingency self-control is quickly

established. Since social attention accompanying a
disciplinary intervention may reinforce the
preceding deviant behavior, the change agent
minimizes social and verbal interaction as much as

possible while the negative sanction is being
applied.
The way in which time-out contingencies may
be employed as part of a broad program is
illustrated in the treatment of an autistic three-

year- old boy by Wolf, Risley, & Mees (1964). In
addition to grossly retarded social and verbal
development, the boy exhibited violent tantrums
that included head-banging, face-slapping, hair-

pulling, and face-scratching. After a tantrum he
was badly bruised and bleeding, and refused to
sleep at night, forcing one or both parents to

remain by his bed. Sedatives, tranquilizers, and
physical restraints were applied without success.
When it became clear that refusal to wear
eyeglasses (necessitated by the removal of
cataractal lenses) might result in ultimate
blindness, psychologists were invited by the
hospital staff to devise a treatment program for

him. Ward attendants, and later his parents,
carried out the prescribed program under the
guidance and direction of the consultants.
Most of the boy’s recurrent problem behaviors,
which obstructed any treatment efforts, were

eliminated by a procedure combining extinction
and punishment through reinforcement
withdrawal. In modifying the tantrum behavior,
for example, whenever the boy slapped himself

and whined he was placed in his room where he
remained until the tantrum ceased. During the
initial phase of treatment the attendants offered

elaborate apologetic explanations while escorting
him to his room, and showered him with attention
when he returned. These accompanying reactions
converted the exclusion into a rewarding
experience, with the result that the boy displayed
frequent tantrums followed by brief perfunctory
trips to the room. A minimum time of 10 minutes

in the room was therefore instituted and the
attendants were instructed to minimize the
inadvertent social reinforcement. Under this
contingency the frequency of violent tantrums
gradually declined and eventually disappeared.

Eating problems, in which the boy grabbed food
from other children’s plates, threw it about the
room, or ate with his fingers, were rapidly
eliminated in a similar manner. The attendants
simply removed him from the dining room for the

remainder of his meal for snatching or tossing
food after a warning, and withdrew his plate for a

few minutes whenever he ate with his fingers.
Use of social exclusion sometimes leads to new
problems which must be dealt with. During the
period of isolation the child may, for example,
proceed to wreck the furnishings in the room, he
may inflict injury upon himself, or exhibit other
harmful behaviors. This problem arose in mild

form during the treatment of the boy’s sleeping
patterns, which were extremely irregular and
required the extended presence of the parents or
attendants at bedtime. After completion of
pleasant bedtime routines, the boy was put to bed,

and left with the door open. If he refused to
remain in bed, the door was closed, which initially
gave rise to violent temper tantrums. These
tantrums were controlled by extending the time
that the door remained closed until after the

tantrum subsided. Under the cumulative
punishment contingency the tantrums rapidly

disappeared and normal sleeping patterns were
established by the sixth night. Similarly,
destructive discarding of eyeglasses ceased within

five days when the boy was placed in his room for
10 minutes whenever he threw his glasses, or if
tantrums developed, until they abated.
The foregoing account has emphasized
aversive controls. It should be noted that, in the
total program, positive reinforcement was also

extensively employed to get the boy to wear
eyeglasses, and he was positively guided and
rewarded for more appropriate behavior patterns.

Prior to treatment, the boy was totally lacking in
communicative skills, which were gradually
established through reinforced modeling. The
development of more rewarding competencies
undoubtedly contributed to the effectiveness of

mild punishment.
As the boy’s condition improved, contacts with
his family and home were progressively increased.
At first the parents visited the hospital for one
hour and observed the way in which tantrums and
bedtime problems were handled by the
attendants. Subsequently the parents made
several visits a week, during which an attendant

observed and instructed them in their handling of
their son. Then he began short home visits
accompanied by an attendant, followed by
progressively longer visits. After discharge, he no

longer manifested severe behavior problems, he
had become increasingly verbal, and the family
interactions were considerably more enjoyable.
The reinforcement procedures were extensively
applied for several years in a nursery school

setting where the boy made sufficient progress to
enroll in public school (Risley & Wolf, 1966; Wolf,

Risley, Johnston, Harris, & Allen, 1967). Time-out
procedures were used occasionally in early phases

to eliminate tantrum behavior, exhibited
whenever he was asked to perform an

instructional task, and to control hurtful behavior
toward other children.
Reference has already been made to several

studies in which self-injurious behavior in autistic
children was completely eliminated or markedly
reduced by reinforcement withdrawal. Similar

methods have been shown by Hamilton, Stephens,
& Allen (1967) to be uniformly successful in
eliminating injurious aggressive and self-

destructive behavior in severely retarded
adolescents. In each instance, the individual was
physically confined to a chair in a time-out area for
a fixed period following the occurrence of
injurious behavior. In one case, for example, a girl

beat her head and back against the wall a total of
35,906 times during four 6-hour observations—
about once every three seconds! When the time-
out contingency was later put in effect, head-
banging precipitously dropped to a negligible level
of 7, 2, 0, 1, 0 for five successive weeks and never
reappeared during nine months of follow-up
study. Of considerable import, both from a clinical
and ethical standpoint, is that after the continual

self-mutilating behavior was eliminated the
retardates participated with evident enjoyment in
daily social and recreational activities. The authors
make the interesting observation that, following
removal of pervasive deviant behavior, individuals

begin to make contact with potentially rewarding
aspects of the environment that were always
available, which automatically reinforce beneficial
modes of behavior. Once a self-regulating
interaction between behavior and environmental

contingencies is initiated, widespread changes
may result even though alternative behaviors

were never deliberately established.
Several additional case reports have been
published which furnish quantitative data
indicating the efficacy of time-out procedures in
modifying diverse behavioral disorders. Sloane,
Johnston, & Bijou (1968) rapidly eliminated
extreme aggressiveness in a preschool boy, and

Burchard & Tyler (1965) reduced the antisocial
behavior of a delinquent adolescent by contingent
social exclusion. This procedure has also been
employed by Tyler & Brown (1967) on a group
basis with institutionalized delinquents. The staff

of a rehabilitation center was unable to check
aggressive, disruptive behavior of delinquent boys
that was highly troublesome during recreational
periods but not serious enough to warrant severe
sanctions. A program involving short periods of

mild punishment for misbehavior was instituted
to control the group. Every time a boy displayed

offensive behavior he was immediately placed in a
room in the cottage for 15 minutes without any
threats, invective, sermonizing, or negotiations for

second chances. Consistent with previous findings,
brief social withdrawal produced a marked
decline in the incidence of disruptive behavior. In
order to determine whether punishment had

lasting effects, the aversive consequences were
discontinued; instead, the staff reprimanded the

boys verbally and occasionally closed the
recreational facility temporarily when their
behavior got completely out of hand. During the
period when negative sanctions were removed

there was a rapid increase in the rate of offenses.
However, when the punishment contingency was
subsequently reinstated disruptive behavior

subsided with equal rapidity and remained at a
low level. Apparently the boys were quick to

discriminate the changes in conditions of
reinforcement and regulated their behavior
accordingly. The aversive control thus proved to

be an exceedingly effective management technique
but, unlike previous findings, it produced no
enduring changes in behavior.
The conflicting results are most likely due to

the types of reinforcement systems created by
peers in institutions for delinquents. An
observational study by Buehler, Patterson, &

Furniss (1966) revealed that delinquent peers
provide extensive positive reinforcement for
antisocial behavior, whereas they typically punish

attitudes and behavior that conform to
institutional norms. If negative sanctions applied
by staff members are strong enough to outweigh
peer influences, socially conforming behavior may
be achieved and sustained as long as the

institutional sanctions remain in effect. However,
when aversive controls are removed, peer
reinforcement practices will quickly reinstate
deviant patterns of behavior. To achieve stable
changes in behavior would require modifications
of the contingency systems practiced by peers like
those, for example, that Cohen (1968) successfully
accomplished by placing delinquents’ livelihood
within the institution on a self-determining basis.

When contingencies are arranged so that
constructive behavior is adequately reinforced
and antisocial behavior has limited pay-off,
antisocial personalities are less inclined to seek
their rewards from outwitting staff members or

from perturbed reactions of others to crisis-
producing transgressions (Colman & Baker, 1968).
“Systematic exclusion” is increasingly
employed in school settings (Chapman, 1962;

Kiersey, 1958) as a means of controlling seriously
disturbing behavior in children after other
available methods have failed. In this program the

child, his parents, his teacher, the school
psychologist, and the principal meet as a group to
arrange explicit contingencies between the child’s
disruptive conduct and its social consequences. At
this meeting, each participant’s role is specifically
outlined. It is explained that the school cannot
permit a child continually to disrupt the

educational activities of an entire class. The child’s
help is enlisted to control, as best he can, behavior
that has disturbing effects on all concerned.
Whenever he displays behavior that exceeds
certain clearly defined limits, the teacher must ask

him to leave school for the remainder of the day.
In order to remove any inadvertent positive
reinforcement of the deviant behavior, the teacher
is instructed not to threaten, coax, urge, or scold
the child, nor is she to engage in persuasive

attempts to alter his behavior. Rather, the
previously agreed-upon sanctions are applied

immediately, straightforwardly, and in a matter-
of-fact way. By having the teacher apply
prearranged contingencies in a consistent and

objective fashion, the child is more likely to regard
the disciplinary interventions as natural,
inevitable consequences of his behavior, than as
arbitrary or malevolent treatment by his teacher.

If the child should refuse to leave when he is
requested, the principal removes him and the

period of exclusion is automatically extended.
When the child arrives home, the parents are
instructed to keep him on the premises during the

remaining school hours, but to refrain from
punishing, scolding, or applying other disciplinary
measures. The psychologist’s function is to
structure and to supervise the program, to offer
the participants positive assistance when needed,

and to decide when the contingencies are to be
discontinued.
The authors report that systematic application
of exclusion consequences produces rapid and
lasting reduction in chronically disruptive
behavior, but no quantitative data are presented,
nor is there any specification of the conditions
under which this method is most efficacious. It is

apparent that an exclusion procedure will produce
no behavioral changes, or may even augment
deviant responsiveness, if the situation from
which a person has been removed is unrewarding

or highly unpleasant. Under these circumstances,
withdrawal has positively reinforcing rather than
punishing effects. It has been shown in laboratory
studies (Herrnstein, 1955), for example, that
behavior which produces a temporary time-out

from reinforcement increases in frequency if the
conditions of reinforcement that the behavior

avoids are relatively unfavorable.
The effectiveness of exclusion procedures can

probably be greatly enhanced if, in addition to the
punishment contingency, the child is provided

with certain privileges and rewards for each class
period during which he does not engage in
disruptive behavior. It would also be of interest to
explore the rate at which deviant behavior

decreases as a function of varying durations of
exclusion, the types of situations to which the
child is removed, and the attractiveness of the
setting from which he is withdrawn. Findings of
studies employing time-out contingencies to

control grossly deviant behavior indicate that
much briefer periods of exclusion might work as
well, or even better, than full-day suspensions
from classroom activities.
Summary
The present chapter has discussed the
processes whereby response patterns are
eliminated through the use of punishing stimuli.
Punishing consequences may involve either
removal of positive reinforcers or presentation of
aversive events. Punishment is believed to achieve
its reductive effects by producing conditioned fear
that elicits inhibitory behavior’s, or by facilitating

the appearance of responses that are incompatible
with and, therefore, capable of supplanting the
punished behavior. The degree of control exerted
by punishment is largely a function of the
intensity, duration and distribution of aversive

consequences, their temporal relation to the
behavior to be modified, the strength with which
punished responses are concurrently reinforced,
the availability of alternative modes of behavior
for securing rewards, the level of instigation to

perform the negatively sanctioned behavior, and
the psychological characteristics of punishing

agents.
Several different theories have been proposed
concerning the locus of aversive control.
Environmental stimuli that are regularly
associated with punishing experiences may
become conditioned elicitors of fear and have
suppressive effects on behavior. A second

interpretation is that proprioceptive stimuli
arising from the punished behavior itself acquire
negatively reinforcing properties through
association with punishing experiences on
previous occasions. Although response-contingent

punishment produces reductive effects, evidence
that behavioral inhibitions can be readily acquired
and extinguished under curare or through
vicarious experiences without any motoric
responding seriously challenge peripheral views

that aversive control resides mainly in response-
produced cues. Rather, these findings lend weight

to the theory that the effects of punishment are
mediated through central controlling mechanisms.
On the basis of previous response consequences,

experienced either directly or vicariously under
differing circumstances, individuals infer the
likelihood that a given course of action will
produce punishing outcomes. Symbolic

representation of these anticipated consequences
can exercise some degree of control over overt

responsiveness. The same behavior may,
therefore, be freely expressed or inhibited in
similar environmental situations as a result of
intricate discriminations of differences in

reinforcement contingencies.
Aversive forms of control have been primarily
employed to eliminate persistent responses that
are automatically self-reinforcing upon

occurrence, to reduce the incidence of seriously
disturbing patterns of behavior for which the
maintaining positive reinforcements cannot be
identified or readily eliminated, and to bring
rapidly under control responses that have
injurious consequences for the performer or
others. While findings of controlled studies are
still somewhat limited, preliminary results of
applications of aversive procedures to the

modification of persistent deviant behavior are
considerably more encouraging than laboratory
findings would lead one to expect. Chronic self-
mutilating behavior, debilitating occupational
cramps of long standing, stuttering, antisocial

aggression, and deviant patterns of sexual
behavior have been substantially reduced or
eliminated by methods relying upon contingent
application of negative consequences.
The relative ineffectiveness of punishment in

producing durable reductive effects in laboratory
situations has probably resulted from the fact that,

with few exceptions, the punished response
constitutes the sole means of securing rewards.
Hence, it comes as no surprise that in single-
response situations punished behavior is
performed for some time even though it incurs
aversive consequences, and it often reappears
when punishment is discontinued. In contrast,

people generally have numerous options available
in everyday life. Even though punishment may
only temporarily inhibit dominant responses,
during the period of suppression alternative
modes of behavior may be strengthened

sufficiently to supplant the original response
tendencies. Moreover, brief cessation of behavior
that is highly disturbing to others eventually
draws positive reactions from appreciative
associates. The new conditions of reinforcement

created by cessation of deviant responses may
foster and maintain their relinquishment. By the

same token, rodents or pigeons that were
suddenly showered with food pellets and
increased positive attention from relieved

comrades, after inhibiting a socially distressing
bar-press response, would undoubtedly abandon
the cherished bar more rapidly than they would if
no alternative activities were available and

response inhibition produced no outcomes other
than the removal of aversive stimulation and loss

of food rewards. For these and other reasons, the
social implications of laboratory findings
regarding aversive control must be accepted with
reservations.
Some of the cited negative by-products of
aversive control can be avoided or reduced to a
large extent by the use of discrimination
procedures in conjunction with punishing

consequences. Moreover, punishment based upon
the removal of positive reinforcers ordinarily
reduces undesired behavior without producing
fear learning or avoidant behavior. This procedure
also tends to maintain strong approach tendencies
toward change agents, and, when reinstatement of
approval, possessions, or privileges is made
conditional upon more appropriate behavior, it
provides more positive support and guidance than

mere administration of negative outcomes.
Lasting elimination of detrimental behavior
can be most effectively facilitated by punishment if
competing response patterns are simultaneously
rewarded. Negative sanctions may therefore be

successfully employed to hold undesired
responses in check while alternative modes of
behavior are being established and strengthened.
Another means of weakening undesired behavior,

relying upon extinction operations, is discussed in
the next chapter.
REFERENCES
Appel, J. B. Punishment in the squirrel monkey Saimiri

sciurea. Science, 1961, 133, 36.
Aronfreed, J. Conduct and conscience. New York:

Aronfreed, J., & Reber, A. Internalized behavioral
suppression and the timing of social punishment.
1, 3-16.
Aronson, E., & Carlsmith, J. M. Effect of the severity of

threat on the devaluation of forbidden behavior.
66, 584-588.
Ayllon, T., & Azrin, N. H. Punishment as a

discriminative stimulus and conditioned
reinforcer with humans. Journal of the
419.
Azrin, N. H. Some effects of two intermittent

schedules of immediate and non- immediate
punishment, Journal of Psychology, 1956, 42, 3-
21.
Azrin, N. H. Some effects of noise on human behavior.

Journal of the Experimental Analysis of Behavior,
1958, 1, 183-200.
Azrin, N. II. Punishment and recovery during fixed-

ratio performance. Journal of the Experimental
Azrin, N. H. Effects of punishment intensity during
variable-interval reinforcement. Journal of the
142.
Azrin, N. H., Halse, D. F., Holz, W. C., & Hutchinson, R.

R. Motivational aspects of escape and
punishment. Journal of the Experimental Analysis
of Behavior, 1965, 8, 31-44.
Azrin, N. H., & Holz, W. C. Punishment. In W. K. Honig

(Ed.), Operant behavior. New York: Appleton-
Century-Crofts, 1966. pp. 380-447.
Baer, D. M. Effect of withdrawal of positive

reinforcement on an extinguishing response in
young children. Child Development, 1961, 32, 67-
74.
Baer, D. M. Laboratory control of thumbsucking by

withdrawal and reinstatement of reinforcement.
1962, 5, 525-528.


New York: Ronald, 1959.
Barrett, B. Reduction in rate of multiple tics by free
operant conditioning methods. Journal of Nervous
and Mental Diseases, 1962, 135, 187-195.
Beach, F. A., Conovitz, M. W., Steinberg, F., & Goldstein,

A. C. Experimental inhibition and restoration of
mating behavior in male rats. Journal of Genetic
Psychology, 1956, 89, 165-181.
Beech, H. R. The symptomatic treatment of writer’s

cramp. In H. J. Eysenck (Ed.), Behavior therapy
and the neuroses. New York: Pergamon, 1960. pp.
349-372.
Benton, A. A. Effects of the timing of negative

response consequences on the observational
learning of resistance to temptation in children.
Dissertation Abstracts, 1967, 27, 2153-2154.

Black, A. H., Carlson, N. J., & Solomon, R. L. Exploratory

studies of the conditioning of autonomic
responses in curarized dogs. Psychological
Bloodstein, O. Conditions under which stuttering is

reduced or absent: a review of literature. Journal
of Speech and Hearing Disorders, 1949, 14, 295-
302.
Boe, E. E. Extinction as a function of intensity of

punishment, amount of training, and
reinforcement of a competing response. Canadian
Journal of Psychology, 1964, 18, 328-342.
Boe, E. E., & Church, R. M. Permanent effects of

punishment during extinction. Journal of
63, 486-492.
Bolles, R. C. Theory of motivation. New York: Harper &

Row, 1967.
Bond, I. K., & Evans, D. R. Avoidance therapy: Its use in

two cases of underwear fetishism. Canadian
Medical Association Journal, 1967, 96, 1160-
1162.
Brookshire, R. H., & Martin, R. R. The differential

effects of three verbal punishers on the
disfluencies of normal speakers. Journal of Speech
and Hearing Research, 1967, 10, 496-505.
Brutten, E. J., & Shoemaker, D. J. The modification of

stuttering. Englewood Cliffs, N. J.: Prentice-Hall,
1967.
Bucher, B., & Lovaas, O. I. Use of aversive stimulation
in behavior modification. In M. R. Jones (Ed.),
Miami symposium on the prediction of behavior,
1967: Aversive stimulation. Coral Gables, Fla.:
University of Miami Press, 1968. pp. 77-145.

4, 157- 167.
Burchard, J., & Tyler, V., Jr. The modification of

delinquent behaviour through operant
conditioning. Behaviour Research and Therapy,
1965, 2, 245- 250.
Camp, D. S. Response suppression as a function of

punishment intensity and of response-
punishment contiguity. Dissertation Abstracts,
1965, 26, 3494.
Chapman, R. W. School suspension as therapy.

Personnel and Guidance Journal, 1962, 40, 731-
732.
Church, R. M. The varied effects of punishment on

behavior. Psychological Review, 1963, 70, 369-
402.
Cohen, H. L. Educational therapy: The design of

learning environments. In J. M. Shlien (Ed.),
21-53.
Colman, A. D., & Baker, S. L., Jr. Utilization of an

operant conditioning model for the treatment of
character and behavior disorders in a military
setting. Unpublished manuscript, Walter Reed
Army Institute of Research, Washington, D. C.,
1968.
Daly, D. A., & Cooper, E. B. Rate of stuttering

adaptation under two electroshock conditions.
Dardano, J. F., & Sauerbrunn, D. Selective

reinforcement of concurrent progressive ratio
behavior. Journal of the Experimental Analysis of
Behavior, 1964, 7, 51-65.
Davis, D. M. The relation of repetitions in the speech

of young children to certain measures of
language maturity and situational factors: Part I.
Journal of Speech Disorders, 1939, 4, 303-318.
Davis, D. M. The relation of repetitions in the speech

of young children to certain measures of
language maturity and situational factors: Part II.
Journal of Speech Disorders, 1940, 5, 235-246.
Desiderato, O. Generalization of conditioned
suppression. Journal of Comparative and
Dinsmoor, J. A. Punishment: I. The avoidance

hypothesis. Psychological Review, 1954, 61, 34-
46.
Elliott, R., & Tighe, T. J. Breaking the cigarette habit: A

technique involving threatened loss of money.
Estes, W. K. An experimental study of punishment.

Psychological Monographs, 1944, 57 (3, Whole
No. 26.3).
Feldman, M. P., & MacCulloch, M. J. A systematic

approach to the treatment of homosexuality by
conditioned aversion: Preliminary report.
American Journal of Psychiatry, 1964, 121, 167-
171.
Feldman, M. P., & MacCulloch, M. J. The application of

anticipatory avoidance learning to the treatment
of homosexuality. 1. Theory, technique and
preliminary results. Behaviour Research and
Therapy, 1965, 2, 165- 183.
Ferster, C. B. Control of behavior in chimpanzees and

pigeons by time out from positive reinforcement.
No. 461).
Ferster, C. B. Suppression of a performance under

differential reinforcement of low rates by a pre-
time-out stimulus. Journal of the Experimental
Ferster, C. B., Appel, J. B., & Hiss, R. A. The effect of

drugs on a fixed-ratio performance suppressed
by a pre-time-out stimulus. Journal of the
Flanagan, B., Goldiamond, I., & Azrin, N. H.

Instatement of stuttering in normally fluent
individuals through operant procedures. Science,
1959, 130, 979-981.
Fowler, H., & Miller, N. E. Facilitation and inhibition of

runway performance by hind- and forepaw shock
of various intensities. Journal of Comparative and
Physiological Psychology, 196.3, 56, 801-805.
Freedman, J. L. Long-term behavioral effects of

cognitive dissonance. Journal of Experimental
Goldiamond, I. Fluent and nonfluent speech

(stuttering): Analysis and operant techniques for
control. In L. Krasner & L. P. Ullmann (Eds.),
Holt, Rinehart & Winston, 1965. pp. 106-156. (a)
Goldiamond, I. Operant properties of verbal fluency

and nonfluency. Progress Report, U.S.P.H.
Research Grant M-8876. Institute for Behavioral
Research, 1965. (b)
Guthrie, E. R. The psychology of learning. New York:

Harper, 1935.
Hamilton, J., Stephens, L., & Allen, P. Controlling

aggressive and destructive behavior in severely
retarded institutionalized residents. American
Journal of Mental Deficiency, 1967, 71, 852-856.
Hearst, E., & Sidman, M. Some behavioral effects of a

concurrently positive and negative stimulus.
1961, 4, 251-256.
Hendry, D. P., & Van Toller, C. Fixed-ratio punishment

with continuous reinforcement. Journal of the
300.
Herman, R. L., & Azrin, N. H. Punishment by noise in

an alternative response situation. Journal of the
188.
Herrnstein, R. J. Behavioral consequences of removal
of a discriminative stimulus associated with
variable-interval reinforcement. Unpublished
doctoral dissertation, Harvard University, 1955.
Hicks, D. J. Effects of co-observer’s sanctions and adult

presence on imitative aggression. Child
Development, 1968, 39, 303-309.
Hill, B. Boss of Britain’s underworld. London: Naldreth,

1955.
Hill, H. E. An experimental study of disorganization of

speech and manual responses in normal subjects.
Journal of Speech and Hearing Disorders, 1954, 19,
295-305.
Hoffman, M. L. Power assertion by the parent and its

impact on the child. Child Development, 1960, 31,
129-143.
Hoffman, H. S., & Fleshier, M. Stimulus factors in

aversive controls: The generalization of
conditioned suppression. Journal of the
378.
Hoffman, H. S., & Fleshier, M. Stimulus aspects of

aversive controls: Stimulus generalization of
conditioned suppression following
discrimination training. Journal of the
239.
Hoffman, H. S., & Fleshier, M. Stimulus aspects of

aversive controls: the effects of response
contingent shock. Journal of the Experimental
Hoffman, H. S., Fleshier, M., & Jensen, P. K. Stimulus

aspects of aversive controls: The retention of
conditioned suppression. Journal of the
583.
Hoffman, M. L. Power assertion by the parent and its

impact on the child. Child Development, 1960, 31,
129-143.
Holz, W. C., & Azrin, N. H. Interactions between the

discriminative and aversive properties of
punishment. Journal of the Experimental Analysis
of Behavior, 1962, 5, 229-234.
Holz, W. C., Azrin, N. H., & Ayllon, T. Elimination of

behavior of mental patients by response-
produced extinction. Journal of the Experimental
Holz, W. C., Azrin, N. H., & Ulrich, R. E. Punishment of
temporally spaced responding. Journal of the
122.
Honig, W. K., & Slivka, R. M. Stimulus generalization of

the effects of punishment. Journal of the
Hunt, H. F., & Brady, J. V. Some quantitative and

qualitative differences between “anxiety” and
“punishment” conditioning. American
Psychologist, 1951,6,276-277 (abstract).
Hunt, H. F., & Brady, J. V. Some effects of punishment

and intercurrent anxiety on a simple operant.
Journal of Comparative and Physiological
Psychology, 1955, 48, 305-310.
Johnson, W., et al. A study of the onset and

development of stuttering. Journal of Speech
Disorders, 1942, 7, 251-257.
Kaufman, A. Increased suppression during

punishment applied to the responding member.
Kiersey, D. W. Systematic exclusion, an experimental

regimen for chronic misbehavior. Paper read at
the Annual Conference of California Association
of School Psychologists and Psychometrists,
Asilomar, Calif., 1957.
Leitenberg, H. Is time-out from positive reinforcement

an aversive event? A review of the experimental
evidence. Psychological Bulletin, 1965, 64, 428-
441.
Lichtenstein, F. E. Studies of anxiety: I. The production

of a feeding inhibition in dogs. Journal of
43, 16-29.
Liversedge, L. A., & Sylvester, J. D. Conditioning

techniques in the treatment of writer’s cramp.
Lancet, 1955, 2,1147-1149.
Lovaas, O. I., Freitag, G., Gold, V. J., & Kassorla, I. C.

Experimental studies in childhood schizophrenia:
Analysis of self-destructive behavior. Journal of
MacCulloch, M. J., Feldman, M. P., & Pinshoff, J. M. The

application of anticipatory avoidance learning to
the treatment of homosexuality. II. Avoidance
response latencies and pulse rate changes.
McMillan, D. E. A comparison of the punishing effects

of response-produced shock and response-
produced time out. Journal of the Experimental
Masserman, J. M. Behavior and neurosis. Chicago:

Mischel, W., & Liebert, R. M. Effects of discrepancies

between observed and imposed reward criteria
on their acquisition and transmission. Journal of
Morse, W. H., & Herrnstein, R. J. The maintenance of

avoidance behavior using the removal of a
conditioned positive reinforcer as the aversive
stimulus. American Psychologist, 1956, 11, 430
(Abstract).
Mowrer, O. H. An experimental analogue of

“regression,” with incidental observations on
“reaction formation.” Journal of Abnormal and
Mowrer, O. H. Learning theory and behavior. New

York: Wiley, 1960.
Nelson, S. E. Personal contact as a factor in the

transfer of stuttering. Human Biology, 1939, 11,
393-401.
Nigro, M. R. Punishment of an extinguishing shock-

avoidance response by time-out from positive
reinforcement. Journal of the Experimental
Parke, R. D., & Walters, R. H. Some factors influencing

the efficacy of punishment training for inducing
response inhibition. Monographs of the Society for
No. 109).
Pastore, N. The role of arbitrariness in the frustration-

aggression hypothesis. Journal of Abnormal and
Powell, J., & Azrin, N. The effects of shock as a

punisher for cigarette smoking. Journal of Applied
Quist, R. W., & Martin, R. R. The effect of response

contingent verbal punishment on stuttering.
Journal of Speech and Hearing Research, 1967, 10,
795-800.
Rescorla, R. H., & Solomon, R. L. Two-process learning

theory: Relationships between Pavlovian
conditioning and instrumental learning.
Risley, T. R. The effects and side effects of punishing

the autistic behaviors of a deviant child. Journal
of Applied Behavior Analysis, 1968, 1, 21-34.
Risley, T., & Wolf, M. M. Experimental manipulation of
autistic behaviors and generalization into the
home. In R. Ulrich, T. Stachnik, & J. Mabry (Eds.),
Control of human behavior. Glenview, 111.: Scott,
Foresman, 1966. pp. 193-198.
Rotenberg, I. C. Effect of schedule and severity of

punishment on verbal behavior. Journal of
Sandler, J., & Quagliano, J. Punishment in a signal

avoidance situation. Paper read at the
Southeastern Psychological Association Meeting,
Gatlinburg, Tenn., 1964.
Savoye, A. L. The effect of the Skinner-Estes operant

conditioning punishment paradigm upon the
production of non-fluencies in normal speakers.
Unpublished master’s thesis, University of
Pittsburgh, 1959.

rearing. Evanston, 111.: Row, Peterson, 1957.
Shames, G. H., & Sherrick, C. C., Jr. A discussion of

nonfluency and stuttering as operant behavior.
Journal of Speech and Hearing Disorders, 1963, 28,
3-18.
Sheehan, J. G. The modification of stuttering through

non-reinforcement. Journal of Abnormal and
Sheehan, J. G. An integration of psychotherapy and

speech therapy through a conflict theory of
stuttering. Journal of Speech and Hearing
Disorders, 1954, 19, 474-482.
Sheehan, J. G. Conflict theory of stuttering. In J.

Eisenson (Ed.), Stuttering: A symposium. New
York: Harper, 1958. pp. 121-166.
Sheehan, J. G., & Martyn, M. M. Spontaneous recovery

from stuttering. Journal of Speech and Hearing
Research, 1966, 9, 121-135.
Siegel, G. M., & Martin, R. R. Punishment of

disfluencies in normal speakers. Journal of Speech
and Hearing Research, 1966, 9, 208-218.
Sloane, H. N., Jr., Johnston, M. K., & Bijou, S. W.

Successive modification of aggressive behavior
and aggressive fantasy play by management of
contingencies. Unpublished manuscript,
University of Illinois, 1968.
Solomon, R. L. Punishment. American Psychologist,

1964, 19, 239-253.

69, 202-219.
Solomon, R. L., Turner, L. H., & Lessac, M. S. Some

effects of delay of punishment on resistance to
temptation in dogs. Journal of Personality and
Stassi, E. J. Disfluency of normal speakers and

reinforcement. Journal of Speech and Hearing
Research, 1961, 4, 358-361.
Storms, L. M., Boroczi, C., & Broen, W. E., Jr. Effects of

punishment as a function of strain of rat and
duration of shock. Journal of Comparative and
Sylvester, J. D., & Liversedge, L. A. Conditioning and

the occupational cramps. In H. J. Eysenck (Ed.),
Behaviour therapy and the neuroses. New York:
Pergamon, 1960. pp. 334-348.
Tate, B. G., & Baroff, G. S. Aversive control of self-

injurious behavior in a psychotic boy. Behaviour
Research and Therapy, 1966, 4, 281-287.
Timmons, E. O. Weakening verbal behavior: A

comparison of four methods. Journal of General
Psychology, 1962, 67, 155-158.
Tolman, C. W., & Mueller, M. R. Laboratory control of
toe-sucking in a young rhesus monkey by two
kinds of punishment. Journal of the Experimental
Tyler, V. O., Jr., & Brown, G. D. The use of swift, brief

isolation as a group control device for
institutionalized delinquents. Behaviour Research
and Therapy, 1967, 5, 1-9.
Van Riper, C. The influence of empathic response on

frequency of stuttering. Psychological
Walters, R. H., & Demkow, L. Timing of punishment as

a determinant of resistance to temptation. Child
Development, 1963, 34, 207-214.
Walters, R. H., Parke, R. D., & Cane, V. A. Timing of

punishment and the observation of consequences
to others as determinants of response inhibition.
Journal of Experimental Child Psychology, 1965, 2,
10-30.
Walton, D., & Mather, M. D. The relevance of

generalization techniques to the treatment of
stammering and phobic symptoms. Behaviour
Research and Therapy, 196.3, 1, 121-125.
Weiner, H. Some effects of response cost upon human

operant behavior. Journal of the Experimental
Whiting, J. W. M., & Mowrer, O. H. Habit progression

and regression—a laboratory study of some
factors relevant to human socialization. Journal of
Comparative Psychology, 1943, 36, 229-253.
Wischner, G. J. Stuttering behavior and learning: A

preliminary theoretical formulation. Journal of
Wolf, M., Risley, T., Johnston, M., Harris, F., & Allen, E.
Application of operant conditioning procedures
to the behavior problems of an autistic child: A
follow-up and extension. Behaviour Research and
Therapy, 1967, 5, 103-111.
Wolf, M., Risley, T., & Mees, H. Application of operant

conditioning procedures to the behavior
problems of an autistic child. Behaviour Research
and Therapy, 1964, 1, 305-312.
Zimmerman, J. A. A multiple schedule of time out

avoidance with monkeys. American Psychologist,
1963, 18,422 (Abstract).
6
Extinction
When reinforcement for a learned response is
withheld, individuals will continue to exhibit that
behavior for a limited time. Under repeated
nonreinforcement, however, the behavior
decreases and eventually disappears. This decline
in responsiveness as a function of nonrewarded
repetition of a response is called extinction. In

naturally occurring situations response patterns
sustained by positive reinforcement are frequently
eliminated simply by discontinuing the rewards

that ordinarily produce the behavior. Avoidance
behavior, which is strongly maintained by its
success in forestalling anticipated aversive

experiences, can similarly be extinguished if, when
it is prevented from occurring in the presence of
threatening stimuli no adverse consequences are
engendered. With repeated nonreinforced
exposure to subjective threats, protective behavior

that is inappropriate to the altered circumstances
is eventually abandoned. In both instances the

extinction process is essentially the same,
although the procedures vary according to the
nature of the maintaining consequences.
Interpretations of the Extinction Process
The rate of extinction is governed by a number
of factors, among them the irregularity with which
the behavior was reinforced in the past, the

amount of effort required to perform it, the level
of deprivation present during extinction, the ease

with which changes in conditions of reinforcement
can be discerned, and the availability of alternative
modes of response. Because of the diversity of
controlling variables, a number of different
theoretical conceptualizations of extinction have
been proposed (Kimble, 1961), each emphasizing

a separate aspect of the phenomenon. These
formulations and their supporting evidence are
reviewed next.
CONDITIONED INHIBITION THEORY
According to Hull’s (1943) theory of extinction,

repeated nonreinforced elicitation of an effortful
response generates an inhibitory state analogous
to fatigue, which tends to counteract the

recurrence of the response. Since this reactive
inhibition dissipates with time, it presumably
exerts only a temporary suppressing influence on
behavior. Some support for the fatigue hypothesis
is provided by the well-known phenomenon of
spontaneous recovery. When a response has been
initially extinguished, it tends to reappear, though
typically at reduced strength, with the passage of
time, suggesting the dissipation of a transitory
form of inhibitory control. The fact, however, that

the amount of recovery following successive
extinctions becomes progressively smaller until it
eventually reaches zero clearly indicates that
additional processes are involved. Therefore, Hull

postulated that in addition to response-produced
inhibition, extinction also involves the production
of conditioned inhibition. The enduring decrement
in behavior was explained as follows: When
reactive inhibition reaches a high level, cessation

of activity alleviates the aversive motivational
state, and, consequently, any stimuli associated

with stoppage of the response become
conditioned inhibitors. In this formulation, fatigue

reduction resulting from the termination of
behavior that produces aversive stimulation

constitutes the primary reinforcing event.
It would follow from inhibition theory that any
conditions which increase the amount of relative
inhibition should facilitate extinction. The two

factors that have been investigated most
intensively in this regard are the rate of response

evocation and the amount of effort required to
perform the behavior. In general, research
findings (Kimble, 1961) indicate that there is little
diminution in response when extinction trials are

widely distributed over time, whereas extinction
occurs rapidly with massed response evocation.
The evidence concerning the effect of effortfulness

on extinction is somewhat equivocal, but the
majority of investigations has shown that

performances involving a large expenditure of
effort extinguish more rapidly than those
requiring little exertion.
Although a fatigue theory accounts for certain
extinction phenomena, there are many facts that
cannot be adequately explained in terms of
response-generated inhibition (Gleitman,
Nachmias, & Neisser, 1954; Kimble, 1961; Mowrer,

1960). First, both reactive and conditioned
inhibition depend upon repeated response
evocation. There is ample evidence, however, that
nonreinforced performance of behavior is a
facilitative, but not a necessary, condition for

extinction. These findings are based on different
paradigms, in each of which responses are either
partially or totally extinguished without ever
being performed. In “latent extinction” studies, for
example, animals that are merely placed directly

in the empty goal box a number of times
subsequently extinguish instrumental responses

more rapidly than control groups placed in other
situations lacking rewards (Deese, 1951; Seward
& Levy, 1949; Moltz, 1955). Extinction of

avoidance responses likewise can be greatly
facilitated independently of skeletal responding by
repeatedly presenting a conditioned aversive
stimulus to subjects who are muscularly

immobilized by curare (Black, 1958). Moreover,
autonomic responses, which are not known to

generate fatigue states, also undergo extinction
with repeated nonreinforced elicitation.
Nonresponse extinction is perhaps most clearly
revealed by experiments of vicarious extinction

(Bandura, 1968) in which intractable avoidance
responses of long standing are completely
eliminated on the basis of observational

experiences alone.
The limitation of explanations of extinction in

terms of response-produced inhibition is also
apparent in investigations of resistance to
extinction as a function of variations in the
percentage, magnitude, and serial patterning of
reinforcement, and of changeable acquisition
conditions. Though the terminal levels of learning
achieved under constant and variable
circumstances are essentially identical, increasing

the irregularities of reinforcement and training
conditions increases subsequent resistance to
extinction. Finally, in some instances
nonreinforced elicitation may result in extremely
rapid or even single-trial extinction before much

reactive inhibition could possibly develop. The
overall research evidence thus indicates that,
while response-produced inhibition may be one
determinant of extinction, additional processes
are undoubtedly involved.
COMPETING-RESPONSE THEORY
In the interference interpretation of extinction
(Guthrie, 1935; Estes, 1959), the decrement of a
response during nonreinforced evocation results
from the appearance of incompatible responses
strong enough to supersede the ongoing behavior.
These competing responses may be linked either
to the same stimuli or to different stimulus events.

In the latter case, response diminution primarily
reflects external inhibition brought about by
simultaneous occurrence of new prepotent stimuli
that evoke antagonistic tendencies, or by

attentional shifts to other distinctive features of
the environment. In the former instance, which
essentially involves a counterconditioning
process, extinction results from the development
of new incompatible responses to the same

stimuli, or the reappearance of interfering
responses that have been previously learned.
Any conditions, apart from the omission of
reinforcement, that reduce the probability of

occurrence of the original behavior will facilitate
the appearance of competing response patterns.

Some of these conditions, originally proposed by
Guthrie (1935), include introducing the
controlling stimuli at such reduced intensity that
they fail to evoke the undesired behavior;

repeating the controlling stimuli after the original
responses have been inhibited through fatigue so
that new behavioral tendencies can be learned;
and presenting the stimuli in the context of more
powerful stimulus events that evoke incompatible

responses of sufficient strength to override the
undesired behavior. Response-prevention
techniques relying upon physical barriers (Carlson
& Black, 1959; Solomon, Kamin, & Wynne, 1953)
also provide a means of ensuring the occurrence of

alternative responses in the presence of
conditioned aversive stimuli. In a theory of

extinction emphasizing conditioned relaxation
processes, Denny and his associates (Denny &
Weisman, 1964; Weisman, Denny, Platt, &

Zerbolio, 1966) contend that termination or
omission of aversive stimulation automatically
produces relaxation responses that generalize
backward to the fear-provoking situation and

compete with avoidance behavior.
Given the emergence of competing responses,

whatever their source, the nature of the
reinforcement that maintains new behavioral
tendencies within the extinction situation remains

to be explained. It will be recalled that reduction in
fatigue associated with the cessation of effortful
behavior was assumed in Hullian theory to
reinforce inhibitory responses. Some research
evidence suggests that the termination of aversive

emotional effects generated either by fear-
producing stimuli or by repeated nonreward may
also provide supporting reinforcement.
Page reports a series of experiments that
illustrate the extinction of avoidance behavior
through the development of incompatible
responses. In the initial study (Page & Hall, 1953),
animals learned to avoid shocks administered in

one compartment of a shuttle box by escaping into
a neutral chamber. The avoidance responses were
then extinguished in two different ways: Control
animals were given regular extinction trials in

which they performed avoidance responses until
they stopped running from the threatening
compartment; the experimental subjects were
detained in the fear-provoking box for the first five
trials and then given traditional extinction trials.

The barrier group extinguished approximately
three times as fast as the controls.
In order to determine if elimination of
avoidance responses in the first study was due to

the acquisition of competing protective responses
or to neutralization of the fear-evoking stimuli,

Page (1955) conducted a second experiment that
proceeded in the following manner: The initial
phase of the study, which duplicated the
procedure of the earlier experiment, similarly

demonstrated that animals first detained in the
threatening compartment subsequently
extinguished much more rapidly than controls
given regular extinction trials. In the second phase
of the study, designed to measure the aversive

properties of the conditioned stimuli, the animals
were placed in the neutral chamber after being
deprived of food and the speed with which they
entered the shock compartment for food was
measured. In addition, a control group of animals

never exposed to shock stimulation was tested.
The approach response latencies averaged

approximately 25, 60, and 110 seconds for the
control, regular extinction, and response-
prevention groups, respectively. It seems clear

from the two sets of data that under forced
exposure to fear-arousing stimuli a dominant
mode of avoidance behavior was eliminated, but
the animals nevertheless retained some fear of the

negative compartment. These findings indicate
that the threatening stimuli continued to generate

aversive stimulation and whatever protective
responses the animals adopted in the situation
were reinforced by the omission of painful shocks.
In the extinction of avoidance behavior,

absence of expected adverse consequences
provides a powerful source of reinforcement for
competing responses. In the elimination of
behavior previously maintained by positive

reinforcement, reduction of aversive emotional
arousal produced by omission of anticipated
rewards may likewise constitute the main
reinforcement for antagonistic responses.
According to the frustration interpretation of
extinction (Amsel, 1962; Wagner, 1966),
nonrewarded repetition of responses generates
aversive arousal capable of evoking conflicting
response tendencies that interfere with the

ongoing behavior. By supplanting the
nonrewarded behavior the competing responses
reduce disturbing emotional arousal and are
thereby negatively reinforced. Consistent with
these theoretical speculations, it has been shown

that nonreward produces aversive effects
analogous to punishment operations. Stimuli
previously associated with nonreward acquire
arousal properties (Wagner, 1963), their presence
attenuates responding (Amsel & Surridge, 1964),

and escape from cues signifying nonreward can
reinforce new performances (Wagner, 1963).
The appearance of new behavior that is
antagonistic to nonreinforced responses will
undoubtedly accelerate the extinction process. In
many instances, however, rapid elimination of
nonrewarded behavior results from the
development of expectations about the future
probability of reinforcement rather than from the

gradual conditioning of incompatible responses to
the same controlling stimuli. The discrimination
theory of extinction, which is reviewed next, treats
extinction as a centrally, rather than a
peripherally, mediated phenomenon.
DISCRIMINATION THEORY
Interpretations of extinction in terms of
discrimination emphasize the role of

observational and cognitive processes. According
to this formulation, behavior is performed for
some time after reinforcement has been
discontinued because the subject has failed to
recognize that previous reinforcement
contingencies are no longer in effect. It would
follow from this hypothesis that variables which
reduce the discriminability between prior
conditions of reinforcement and those of
extinction should prolong nonrewarded

responding.
Numerous experiments, in which the
difference between acquisition and extinction has

either been systematically varied or is easily
inferable, provide supporting evidence for the
discrimination hypothesis. Behavior established

under intermittent reinforcement, for example, is
more resistant to extinction than responses
following continuous reinforcement. When
rewards are suddenly discontinued, it is
reasonable to suppose that persons who have

been reinforced each time they respond will
recognize the change more readily than those who

have always been reinforced irregularly. The
lower the frequency of reinforcement, the less

discernible are the changes. It is perhaps for
similar reasons that, even under the same degree
of partial reinforcement, behavior that has been
rewarded irregularly extinguishes less rapidly

than responses established by a regular,
predictable schedule of intermittent
reinforcement (Ferster & Skinner, 1957; Kimble,
Mann, & Dufort, 1955; Longenecker, Krauskopf, &
Bitterman, 1952). Moreover, behavior is highly

resistant to extinction following training that
includes long series of consecutively
nonreinforced trials (Slamecka, 1960). The latter
conditions would make it particularly difficult to
determine when extinction has begun.
Discriminations can be formed not only on the
basis of the frequency and patterning of

reinforcing stimuli, but also in terms of other
distinguishable features of the environment that
signify a change in reinforcement practices. The
presence during extinction of stimuli that have
previously signified that appropriate performance
will be reinforced results in faster extinction than
if the positive discriminative cues are absent

(Elam, Tyler, & Bitterman, 1954; McNamara &
Paige, 1962; Slamecka, 1960). These findings,
which are contrary to secondary reinforcement
theory, are in accord with the discrimination
hypothesis. If responses in the presence of stimuli

that formerly signaled a high likelihood of reward
are no longer reinforced, it is made apparent that
the original reinforcement contingencies have
been discontinued. However, Longstreth (1966)
interpreted similar findings with children in terms

of frustration theory, which holds that when a
stimulus previously paired with reward is

subsequently presented alone it generates
aversive emotional responses that interfere with
ongoing behavior.
Irregular conditions of learning would also be
expected to increase the complexity of
discrimination and hence to prolong the
persistence of nonrewarded performances. In

several experiments, response patterns that are
acquired under variable stimulus conditions (e.g.,
gross changes in drive states, environmental
stimuli, required performances, and in the
frequency, magnitude, and delay of reinforcement)

are extinguished under unchanging circumstances.
Results of these studies disclose that resistance to
extinction increases with increased variability in
conditions of learning (McClelland & McGown,
1953; McNamara & Wike, 1958; Mackintosh,

1955).
While the above findings are consistent with

the discrimination hypothesis, they can be
adequately explained without the necessity for
invoking symbolic processes. Under variable
circumstances different types of responses are
learned to a variety of stimuli. It would therefore
require a longer time to extinguish a varied
assortment of responses than a single response

made to a limited number of stimuli presented
under invariant conditions. The results of a study
by Brown & Bass (1958), however, bring into
question both the discrimination and the multiple-
learning interpretations. In this experiment

subjects were both trained and extinguished
under constant or variable stimulus conditions.
Persistence of non-rewarded behavior was
primarily affected by variation in stimulation
during extinction rather than by the degree of

contrast between acquisition and extinction that
should facilitate discrimination, or by the amount

of change in stimulus conditions during
acquisition that should promote more generalized
learning.
The potentially influential role of observational
experiences in extinction is disclosed by studies
alluded to earlier, employing nonresponse
extinction procedures. Mere observation that

formerly rewarding situations no longer contain
rewards facilitates the elimination of instrumental
approach responses. Moreover, the larger the
number of positively discriminative stimuli
retained in the situation in which rewards were

previously dispensed, the faster the response
extinction (Denny & Ratner, 1959; Moltz, 1955).
While the efficacy of observational extinction
procedures has been well established under a

variety of conditions (Deese, 1951; Dyal, 1963;
Koppman & Grice, 1963; Seward & Levy, 1949;
Wilson & Dyal, 1963), the response decrements

may be interpreted in several ways. One possible
explanation is in terms of cognitive processes.
Repeated observation that formerly available
rewards are now absent undoubtedly conveys
information to the observer about the altered
conditions of reinforcement. However, when
environmental stimuli that signify the

reinforcement contingencies associated with given
behavior are also removed, the observed situation
bears little resemblance to, and hence provides
little information about, the original one. There is
no reason to suppose that under such

circumstances expectations regarding the
customary reinforcement contingencies should be
modified to any significant extent. Repeated
exposure to formerly rewarding stimulus contexts
may also extinguish the secondary reinforcing

properties of environmental stimuli that had been
regularly associated with primary reinforcement

(Moltz & Maddi, 1956). The latter outcome would
likewise serve to hasten the extinction process.
The research cited above has been primarily
confined to infrahuman subjects; perhaps for this
reason fractional anticipatory response
mechanisms and associated proprioceptive cues
have frequently been invoked as explanatory

factors. In the case of humans, who possess
superior discriminative and symbolic capacities,
the informative value of observational experiences
regarding reinforcement contingencies would
assume considerably greater importance in

eliminating nonreinforced behavior.
According to this more cognitive view,
extinction primarily reflects the operation of
inhibitory sets rather than the loss of behavior or

its disconnection from previous controlling
stimuli. For this reason behavior can be discarded
even without having been performed on the basis

of observation that such behavior is no longer
reinforced, it can be promptly replaced by more
utilitarian modes of response, and readily
reinstated whenever the original reinforcement
contingencies are restored.
The theory of cognitive control of extinction is
supported by several lines of evidence, some of

which are discussed fully in the concluding
chapter. Extinction is greatly facilitated by
awareness that the usual consequences have been
discontinued; and, conversely, it is retarded under
diverting instructions that reduce discriminability

of the change in reinforcement (Spence, 1966). In
fact, when presentation of reinforcement is
embedded in a diverting task that ensures
exposure to stimulus events but prevents

recognition of their contingent relationship, the
rate of extinction is the same for responses
originally acquired under either partial or

continuous reinforcement.
The common finding that irregular
reinforcement produces behavior that is more
resistant to extinction than continuous
reinforcement is also disconfirmed when
discriminability between acquisition and

extinction is equalized by informing subjects at the
onset of extinction that painful stimuli will not be
presented on subsequent trials (Bridger & Mandel,
1965). As can be seen from Figure 6-1, the partial

reinforcement effect was obtained for
noninformed subjects but induced awareness
essentially abolished conditioned autonomic
responses, regardless of whether they were
acquired on a 100 percent or 25 percent schedule

of reinforcement. Numerous other investigations
of symbolic control of extinction reveal that

persons who are simply informed that
reinforcement has been discontinued display a

precipitous decrement in both conditioned
autonomic responses (Cook & Harris, 1937; Grings

& Lockhart, 1963; Notterman, Schoenfeld, & Bersh,
Figure 6-1. Rate of extinction of GSRs as a function of
awareness and the schedule of reinforcement
employed during the acquisition phase. Bridger &
Mandel, 1965.
1952; Wickens, Allen, & Hill, 1963) and
instrumental avoidance behavior (Lindley &
Moyer, 1961; Moyer & Lindley, 1962), whereas

uninformed subjects show a more gradual decline
in responsiveness.
The influence of verbalized contingencies in
facilitating extinction presupposes a history of
differential reinforcement on the basis of which

verbal cues become reliable indicants of probable
response consequences. In cases where social
agents or verbal communications are considered
untrustworthy, where the real or imagined
consequences of certain actions are highly

injurious, and where environmental events are not
entirely predictable, verbal control of extinction is
apt to be relatively weak. It is extremely unlikely,
for example, that informing snake-phobic persons
that a particular reptile is harmless will result in
any appreciable decrease in snake avoidance
behavior.
The powerful symbolic control over emotional
responses developed under laboratory conditions

contrasts sharply with the refractory quality of
fears acquired through natural experiences. The
difference may arise partly from the degree of
control exercised by change agents over the feared
events. By turning off the shock apparatus or

removing shock electrodes, experimenters can
completely remove any potential threats from the
situation. In contrast, naturally feared objects that
are ordinarily innocuous can nevertheless

occasionally produce hurtful effects despite
assurances to the contrary. Even harmless snakes
or dogs do bite. However, this explanation does

not fully reconcile the divergent findings because
snake phobics experience considerable emotional
disturbance to pictures of reptiles (Bandura,
Blanchard, & Ritter, 1969) while acknowledging
that the agitation is groundless because pictorial
snakes cannot possibly inflict any injury. The
overall evidence would seem to indicate that

emotional behavior may be controlled by two
different stimulus sources. One is the emotional
arousal self-generated by symbolic activities in the
form of emotion-provoking thoughts about
frightening or pleasurable events. The second is

the response evoked directly by conditioned
aversive stimuli. The former component would be
readily susceptible to extinction through cognitive
restructuring of probable response consequences,
whereas elimination of the latter component may

require repeated nonreinforced exposure to
threatening events either directly or vicariously

(Bridger & Mandel, 1964). The differential rate of
extinction of emotional responses arising from
symbolic self-stimulation and from direct external

evocation are discussed more fully in later
considerations of symbolic regulation of behavior.
Laboratory investigations in which verbal reports
of experimenters are accorded high credulity, the

aversive stimuli are of comparatively weak
intensities, and experimenters have full control

over the occurrence of aversive events, may
therefore provide an insufficient explanation of
the process of extinction, particularly as it applies
to refractory avoidance behavior.
COGNITIVE DISSONANCE THEORY
Lawrence & Festinger (1962) have proposed
an explanation, in terms of cognitive dissonance

processes, of why behavior that has been
originally acquired under relatively unfavorable
conditions of reinforcement may be especially
resistant to extinction. According to this
interpretation, when a subject is induced to

engage in behavior that is insufficiently rewarded,
an aversive state of cognitive dissonance is created
by the conflicting information of having expended
effort for inadequate recompense. Under

circumstances where the subject cannot easily
cease responding, the resultant dissonance is
reduced by enhancing the attraction or value of
the activity to justify continuation of the behavior.

After a subject has persuaded himself that he
really likes engaging in the behavior it become
more resistant to extinction when extrinsic
rewards are later removed.
Several different types of reinforcement
conditions are identified as especially prone to
induce dissonance during the period of

acquisition. Behavior that frequently goes
unrewarded, that requires high expenditure of

effort, and for which rewards are delayed, would
be most resistant to extinction. Lawrence &

Festinger have demonstrated in a series of well-
designed experiments with infrahuman subjects
that responses established under these less
advantageous conditions are indeed more

persistent than those that are rewarded
continuously, immediately, and at little
expenditure of effort.
Other investigators have, of course, attributed
the influence on extinction of these reinforcement

variables to the operation of other mechanisms
involving discrimination processes, frustration
effects, and counterconditioning of competing
responses. These alternative theories, therefore,

need to be tested under conditions where they
make opposing predictions. For example,
resistance to extinction following both a highly

variable and an entirely regular schedule of the
same total partial reinforcement has been studied.
The absolute number of unrewarded trials is
identical in both conditions; consequently,
subjects are provided with the same number of
occasions on which dissonance could be aroused
and presumably reduced. Dissonance theory

would predict the same rate of extinction under
both conditions, whereas discrimination theory
would lead one to expect the unpredictable
schedule to produce the more durable behavior.
Bitterman and his associates have conducted

several such experiments in which subjects are
reinforced on 50 percent of the training trials; for
one group the rewards are administered
haphazardly, while subjects in the other group are
regularly reinforced on odd-numbered trials.

Behavior is much more resistant to extinction
(Tyler, Wortz, & Bitterman, 1953) after random 50

percent reinforcement than after regularly
alternated 50 percent reinforcement. Analogous
results are obtained in extinction of autonomic

reactions which do not involve performance of any
effortful responses (Longenecker, Krauskopf, &
Bitterman, 1952). Moreover, increased training,
which provides more dissonance-reducing

opportunities, facilities extinction following
alternating reinforcement, but it has no effect on

behavior rewarded according to an unpredictable
pattern (Capaldi, 1958).
Whether or not findings of the type reported

above contradict dissonance theory cannot be
resolved as long as there exists some ambiguity as
to the conditions most likely to produce high
dissonance. Do subjects who repeatedly perform
effortful behavior knowing that it is unlikely to be

rewarded experience more or less dissonance
than if they had expected rewarding outcomes?
The authors assume that the latter condition is
more dissonance producing. It would seem,
however, that subjects in the former condition are
exhibiting the more irrational behavior and would,
therefore, experience greater pressure to justify
their actions by persuading themselves that they
really enjoy the activity. On the other hand,

subjects who performed because they expected to
be rewarded have adequate justification without
needing to endow the activity with additional
attractions. If reluctant performance of an action is
accepted as behavioral evidence for the existence

of dissonance (Lawrence & Festinger, 1962) then
subjects trained under alternating reinforcement
in the above experiments experienced more
severe dissonance. As training progressed they
continued to perform the behavior on unrewarded

trials albeit with evident hesitancy.
The literature contains other findings that

cannot be adequately accounted for by either
dissonance theory or the alternative formulations.
Young (1966) measured resistance to extinction
as a function of variations in the effortfulness of
the response, the frequency of reward, and the
number of unrewarded trials. The results showed
that, under conditions where animals experienced

many unrewarded trials during acquisition, the
more effort required to perform the response, the
more rapidly it was extinguished, whereas the
opposite relationship was obtained when the
training period included fewer nonreinforced

trials. To further complicate the picture, no
relationship was found between effortfulness of
response and resistance to extinction for subjects
trained under continuous reinforcement.
It is apparent from the experimental findings

reviewed in the preceding sections that no single
theoretical conceptualization can encompass

adequately all of the diverse variables governing
decrements in behavior when reinforcement is
withdrawn. A person may cease responding for
many different reasons and, therefore, a
comprehensive explanation of extinction
phenomena requires a multi-process theory.
Extinction of Positively Reinforced Behavior
It was previously shown how persistent
deviant behavior is often maintained by
intermittent positive reinforcement. Such

contingencies are most likely to arise under
conditions where desirable responses and even
mild forms of deviant behavior are typically
disregarded. On the other hand, the more

persistent or intense responsiveness produces
aversive consequences for others, who unwittingly

reinforce the troublesome behavior in their efforts
to terminate it. Because of its aversive quality,
deviant behavior readily creates conditions that
are likely to perpetuate it. It is true, of course, that
troublesome activities are also frequently
punished, but a reinforcement schedule combining
intermittent reward with occasional punishment
generally results in behavior that is resistant to
change. Moreover, many interventions intended as

punishments actually serve as positive reinforcers
that maintain undesirable behavior. Self- defeating
contingencies usually go unnoticed because
people tend to see only the immediate results,
whereas they rarely systematically evaluate the

changes produced by their practices, or the long-
term effects that their behavior may have on
others. Therefore, detrimental social systems are
often unknowingly created and mutually
sustained, because deviant behavior is rewarded

by the attention it commands and ineffective
control techniques are reinforced by their success

in temporarily checking disturbing performances.
The positively reinforcing effects of verbal
reprimands are well illustrated in a field study by
Madsen et al. (1968) of disruptive behavior in
classroom settings. After the frequency with which
children left their desks was recorded, teachers
consistently reprimanded them for standing and

told them to sit down. The admonishments
promptly increased the number of children
standing (Figure 6-2). In the subsequent baseline
phase teachers reduced reprimands to their
original moderate level, which produced a

corresponding reduction in the number of upright
students. However, they were springing up again
at high rates when procedures were altered so
that transgressive behavior evoked frequent “sit-
down” commands. During the final phase of the

study, teachers ignored standing and praised the
children for working at their desks, a practice that

reduced the incidence of disruptive behavior to its
lowest level.
In eliminating behavior maintained by positive
reinforcement, extinction can be accomplished
Figure 6-2. Number of children standing in class during
baseline periods and when such behavior produced
verbal admonishments or incompatible responses were
positively reinforced. Madsen et al., 1968.
simply by discontinuing the reinforcing
consequences. In social change programs
extinction procedures are often combined with

positive reinforcement of incompatible behavior.
When extinction is a component in a multiform
procedure its separate contribution to behavioral
changes is difficult to evaluate, and categorization
of the treatment method is somewhat arbitrary.
Several issues relating to extinction processes

are revealed in Williams’ (1959) successful
elimination of aggressively demanding behavior in
a young boy. This child had been ill for the first 18
months of his life and had required considerable
attention and special care. During this period,
strong dependency behavior was undoubtedly

established. When the child recovered, his parents
attempted to withdraw some of the attention that
they had previously given him. The child
responded with intense protests, forcing the
parents to attend to him and thus unwittingly to
reinforce crying spells, tantrums, and demands for
their extended presence and undivided attention,

especially at bedtime. The parents were instructed
to put the child to bed in a leisurely and
nonpunitive fashion and, after the completion of
bedtime routines, to ignore the child’s screaming
and raging. Under the extinction contingencies an

immediate marked drop in the duration of
tantrums occurred, followed by almost complete
extinction of tantrums within a few days (Figure 6-
3). The child no longer cried when left in his room,
but instead played happily until he dropped off to

sleep.
In everyday situations it is sometimes difficult

to achieve generalized extinction of deviant
behavior because different social agents are
inconsistent in their reinforcement practices.
Thus, if parents no longer reward temper
tantrums, but other significant adults continue to
do so, a child will, in all likelihood, exhibit a
discriminative pattern of negative behavior

toward others in accord with their customary
reinforcement practices. The rate of extinction can
be further impeded if the same agent, through
inconsistency in his own actions, places the
behavior on a schedule of partial reinforcement. In

the case discussed above, for example, the parents
and an aunt alternated in the painful bedtime
routines. The tantrums were briefly reinstated and
reinforced, after having been extinguished, by
attention from the aunt on an occasion when the

child fussed after having been put to bed. A second
extinction series was therefore carried out and

resulted in complete and stable elimination of the
tantrums (Figure 6-3). Food-throwing and other
disruptive behaviors were similarly extinguished

by promptly removing the child from the table
Figure 6-3. Duration of crying in two extinction series in
which tantrum behavior was no longer socially
reinforced. Williams, 1959.
whenever he deliberately spilled or threw food
from his plate. When this practice was first
instituted the boy was removed from his high

chair 12 times, but thereafter food-throwing
declined rapidly and ceased by the seventh meal

(Williams, 1962).
A child’s aversive control of his parents, if
intermittently reinforced, is likely to generalize to

other areas of behavior and to other people. As
shown by Williams, after inappropriate coercive
behaviors are extinguished the familial
atmosphere changes from one of recurrent drawn-
out battles to reciprocally rewarding interactions.
The importance of establishing uniform
contingencies in implementing a change program
based on extinction is further shown in a case
reported by Ayllon & Michael (1959). A female
patient, who exhibited extremely persistent
psychotic talk, had been subjected to considerable
verbal abuse and beaten on several occasions by
other patients in an effort to keep her quiet.

Patients responded negatively to this woman’s
delusional verbalizations, but the nurses from
time to time attended to her bizarre talk in order
to “get at the roots of her problems” or responded

with perfunctory statements of sympathy and
understanding. The patient was thus provided
intermittent social reinforcement for behavior that
was at other times punished or ignored. The
nurses were instructed not to attend to psychotic

talk and to reinforce sensible verbalizations.
Although the patient’s psychotic responses had

persisted over the previous three years, during a
relatively brief period of treatment the percentage

of these responses dropped from 91 to less than
25. However, an increase in psychotic talk

occurred during the ninth week of treatment,
when, unknown to the ward personnel, a social
worker had been conducting interviews with the
patient and inadvertently reinforcing her

psychotic verbalizations; the effects of these
interviews generalized to the patient’s

interactions with the nurses and other patients as
well. Reinforcements provided by hospital
employees and other visitors to the ward
produced other temporary increases. However,

the psychotic talk still remained less frequent than
it had been at the commencement of the extinction
program and therefore no longer provoked

punitive behavior from the other patients.
A report by Groot (1966) provides some

information on the collateral changes that may
result from extinction of a related deviant
behavior. It also illustrates how, under conditions
where treatment practices are supported
independently of the consequences they have on
recipients, it is frequently more difficult to modify
the practices of the treatment staff than to change
the behavior of clients. A chronic schizophrenic,

who had received insulin and shock treatment,
individual and group therapy, and a lobotomy,
engaged continually in psychotic talk and wrote
numerous bizarre letters. An extinction program
was applied to psychotic talk, and its concomitant

effects on bizarre letter writing and incidence of
disturbed behavior were measured. When
extinction was first instituted, the number of
bizarre letters increased from a baseline rate of
approximately 13 letters a week to 43 letters in

two days, after which it declined and stabilized at
about 5 letters a week. The relative frequency of

disturbed behavior also decreased from 71
percent during the baseline period to only 16
percent when the extinction program was in

effect. Both disturbed behaviors and the number
of psychotic letters increased after the staff
reverted to their own preferred practices, whereas
deviant performances diminished when the nurses

were again persuaded to withhold attention from
psychotic verbalizations.
As part of a program of research in the
development of procedures for the modification of
psychotic behavior, Ayllon and his associates

(Ayllon & Haughton, 1962; Ayllon & Michael,
1959) provide numerous examples in which
deviant behavior of hospitalized psychotics is
extinguished by withdrawal of its positively
reinforcing consequences. In one study (Ayllon &

Haughton, 1962), a group of schizophrenics, who
exhibited severe eating problems of long standing,
remained totally unresponsive to announcements
that meals were being served and to other
persuasive appeals. Consequently, the patients
had been individually escorted to the dining room
by ward personnel, spoon-fed, tube-fed, and
subjected to electroshock “therapy” and other
forms of infantilizing and punitive treatments.
It was assumed by the research staff that the
nurses’ inadvertent social reinforcement in the
form of coaxing, persuading, and feeding the
patients maintained their eating problems, a
contingency that also served to reduce the

controlling function of verbal stimuli. All social
reinforcement for ignoring the announcement of
mealtime and refusals to eat was therefore
withdrawn; following meal call, the dining room

remained open for 30 minutes and any patient
who failed to appear unassisted during that time
simply missed his meal. Under this new

reinforcement contingency, patients responded
promptly to meal call and the chronic feeding
problems were completely eliminated.
It is interesting to note that delusional
statements to the effect that the food was
poisoned, or that God had instructed the patients

to refuse to eat, dropped out soon after they began
to feed themselves (Ayllon & Michael, 1959).

These findings suggest that in some cases
delusional responses may be a product, rather
than a source, of deviant behavior. By adopting a
sick role, supported by delusional justifications,

patients can be more successful in forcing
attending and caretaking responses from busy
ward personnel, who would otherwise reject

inappropriate demands for personalized attention.
Indeed, the nurses frequently encouraged and

positively reinforced infantile response patterns
on the assumption that the patients were
incapable of more mature reality-oriented

behavior because they were “mentally ill.” The
changes noted in delusional beliefs following the
reinstatement of self-feeding are in accord with
considerable research evidence from social

psychological studies which demonstrates that
after a response pattern has been durably
modified, cognition appears to accommodate over
time to support or to justify the new behavior.
Similar extinction procedures were applied by
Ayllon & Haughton (1964) in modifying aberrant

verbal repertoires in three females diagnosed as
chronic schizophrenics. In each case, following a
baseline period of observation, the deviant verbal
behavior was extinguished by withholding social
attention and tangible rewards whenever the

patients engaged in either psychotic talk or
psychosomatic complaints. At such times the
nurses and ward attendants appeared distracted,
bored, or would simply shift their attention to
some other event taking place in the ward. In
order to demonstrate that the changes observed
during the extinction period were not due to some
other variables, the deviant verbalizations were
positively reinforced in the second stage of the

experiment, following which extinction
contingencies were again reinstated.
Figure 6-4 shows the modification of
delusional talk in a patient whose conversational
content over the preceding 14 years had been

dominated by regal self-references (e.g., “I’m the
Queen. The Queen wants a smoke. . . . How’s King
George, have you seen him?”). The findings
provide striking evidence that the staff’s social

attention exercised considerable control over the
client’s aberrant verbal behavior. Psychotic verbal
responses were progressively increased as a

function of positive reinforcement, but they
decreased rapidly when social attention was
withdrawn. The frequency of appropriate verbal
responses was similarly decreased or increased by
Figure 6-4. Reversals in the incidence of psychotic and
neutral verbal behavior as a result of variations in
social reinforcement of these two classes of
verbalizations. Ayllon & Haughton, 1964.
altering the reinforcement contingencies.
Figures 6-5 and 6-6 illustrate the modification
of somatic complaint behavior in two women who

continually alleged innumerable physical
symptoms in the absence of any organic
dysfunction. Their somatic preoccupation and
accompanying emotional responses such as crying
and sobbing were drastically reduced when they

were no longer consoled, or given sympathy or
attention for complaints of various aches and
pains. The extinction process was undoubtedly
hastened in the latter case by the concurrent

positive reinforcement of more appropriate verbal
responsiveness.
As shown in the studies of Ayllon and his
colleagues, institutional environments that are
lacking in positive social reinforcement are likely
to produce a high incidence of attention-
1185
Figure 6-5. Frequency of somatic complaints during the baseline period and
while somatic verbalizations were successively rewarded with attention
and ignored. The temporary increase in somatic complaints shown by
the arrow in the fourth phase of the treatment coincides with a
visit by a relative. Ayllon & Haughton, 1964.
1186
Figure 6-6. Frequency of somatic complaints during the baseline period and while somatic
verbalizations were successively rewarded with attention and ignored. Ayllon &
Haughton, 1964.
compelling deviant behavior. In those familial
environments that likewise provide minimal social
reinforcement, interest and attention may be

supplied primarily for somatic disturbances or
psychological disorders that cannot be easily
ignored. Walton (1960a), for example, reports a
case of a 20-year-old woman who suffered from

long-standing neurodermatitis on the back of her
neck, which was continuously aggravated by
persistent scratching. The client had undergone
numerous medical treatments, including
ointments, pills, lotions, and x-ray therapy, but her

skin condition remained essentially unchanged.
Assessment of the family interrelationships

revealed that the son, who had always enjoyed a
preferred status, was currently receiving most of

the parental attention and their limited financial
resources, whereas the daughter had been

relegated to an inferior, ignored position. With the
advent of the dermatitis, however, the daughter
received more solicitous attention than ever
before, and her fiancé, who similarly expressed

concern over the dermatitis, frequently assisted
with the application of the prescribed ointments.
On the assumption that the scratching
behavior, which perpetuated the skin condition,
was being inadvertently reinforced by the high

level of attention, the family members were
instructed to ignore the dermatitis and the fiancé
to discontinue the ointment routine. Following
discontinuance of the solicitous ministrations, the
scratching decreased and at the end of three

months the dermatitis had completely
disappeared. A four- year follow-up study
disclosed no recurrence of the neurodermatitis;
the client was happily married and successfully
employed. Walton attributes the elimination of
compulsive scratching to the rapid accumulation
of inhibitory potential with nonreinforced
performance and the eventual selection of less
effortful means of securing attention. A more likely

explanation would be in terms of clients’
discriminating the altered reinforcement
contingencies rather than the build-up of
response-produced inhibitory potentials.
In the above case a somatic disorder was

indirectly perpetuated through reinforced motor
behavior that exacerbated the condition. An
ingenious program of research by Miller (1969)
provides impressive evidence that physiological

responses involved in psychosomatic conditions
can be directly modified by contingent
reinforcement. In these studies animals are

curarized to eliminate skeletally mediated
influences, and spontaneous physiological
responding is altered by administering reinforcing
consequences whenever visceral responses of a
selected rate or magnitude occur. The animals are
typically reinforced by rewarding brain
stimulation or by the cessation of electric shock.

Results of numerous experiments consistently
show that when increases or decreases in
physiological responding are reinforced, visceral
changes are produced in the rewarded direction. A
wide variety of visceral responses, including

changes in heart rate, blood pressure, vasomotor
activity, intestinal contractions, and rate of urine
formation, have been substantially modified by
this procedure. The preciseness of reinforcement
control of visceral responses is most strikingly

illustrated by an experiment (Di Cara & Miller,
1968) in which animals were rewarded for

relatively greater vasomotor responses in one ear
than in the other. The animals learned to respond
with differential vasomotor activity in the two

ears, indicating much greater specificity to
autonomic functioning than has been previously
assumed.
Based on the evidence that visceral responses
are subject to reinforcement control, Miller (1969)
speculates that psychosomatic conditions may be

partly developed through contingent attention and
other reinforcing consequences. If this is the case,
it should be possible to modify the visceral

responses involved in psychosomatic disorders by
the use of extinction and differential
reinforcement procedures.
TEMPORAL COURSE OF EXTINCTION
When reinforcement for previously rewarded
behavior is discontinued, the subject is likely to

exhibit, during initial phases of extinction, a
temporary acceleration or intensification of the
behavior in an effort to produce the customary

reinforcement. This is particularly true of
performances that have been maintained on a
continuous schedule of reinforcement. A temper
tantrum that continues to go unheeded will
typically soar to deafening intensities; initially

mild dependency demands, if disregarded, may
culminate in a sharp kick in the shins; and

negative attention-getting behavior that is
consistently ignored typically assumes
increasingly ludicrous forms. But if the more
vigorous responding also proves unsuccessful, it

gradually declines and alternative patterns of
behavior emerge.
The particular sequences and patterns of
responses that appear during later phases of
extinction are primarily determined by the

response options available to the individual. As
dominant modes of behavior are extinguished, a
person will try alternative courses of action that
have been successful on previous occasions in
similar situations. No special problems are created
by the use of extinction alone, provided the
available alternatives are desirable. If, however,

the responses in the individual’s repertoire are
largely inadequate, a change agent may be faced
with the arduous task of extinguishing a long
succession of inappropriate patterns of behavior.
This problem can be easily avoided by combining

extinction procedures with methods that foster
more effective modes of response.
The simultaneous utilization of extinction and
reinforcement procedures is illustrated by one of

several studies (Allen, Hart, Buell, Harris, & Wolf,
1964) designed to modify behavior disorders in
young children on the basis of reinforcement

principles. The case involved a preschool girl who
exhibited marked passivity and withdrawal from
peer interactions, and high dependence on adults,
which generally took the form of hypochondriacal
complaints, drawing attention to collections of
incidental objects, and to simple stereotyped
activities that might gain and hold adult interest.

Most of the time, however, she spent either
standing alone, or sitting in packing boxes, despite
her teachers’ supportive efforts to structure play
activities for her with a group of accepting
children. These attempts were partly unsuccessful

because the teacher would leave soon after the girl
became involved in the group, thus making the
loss of the teacher contingent upon social
participation. Before long she would be standing
on the periphery, seeking out a teacher, or

reverting to some form of solitary activity. In
addition to the autistic behavior, the girl displayed

speech defects and a variety of tic-like behaviors.
In order to increase her social responsiveness,
isolate behavior was consistently nonrewarded
and she was given a minimum of attention for
adult contacts that competed with peer-group
play. Attention was, however, given freely and
warmly whenever she approached or interacted

with peers. Initially, because of her strong
avoidance behavior, approximations to social
interactions, such as standing near other children
or parallel play, were positively reinforced. A
teacher would go to her whenever she was in the

proximity of peers, converse with her, comment
on the nearby play activity, and suggest ways of
joining in the game. Later, adult attention,
approval, and closeness were made contingent
upon direct play with other children. The

introduction of these new contingencies produced
a rapid and marked change in the girl’s social

behavior: peer interactions increased from
approximately 10 to about 60 percent, while adult
contacts dropped from 40 to less than 20 percent

(Figure 6-7).
Figure 6-7. Amount of social behavior displayed by a child as
a result of variations in social reinforcement of adult
and peer interactions. Allen et al., 1964.
In order to determine whether the differential
reinforcement was the determinant of the
behavioral changes, the original contingencies

were reinstated. During this period the teachers
gave full attention to the girl when she contacted

them, played with her when she was alone,
watched her when she engaged in solitary
activities, and conversed with her as long as she
remained in close proximity. No attempts were

made, however, to initiate or to reward social
interaction with peers. Under these conditions, the
girl’s contacts with adults increased, the

hypochondriacal complaints and articulation
problems reappeared, and she promptly reverted

to the isolate pattern of behavior (Figure 6-7).
Several days after the therapeutic contingencies
were again instituted, the girl’s contacts with

adults stabilized at about 25 percent, and social
interactions with peers increased to the previous
level of about 60 percent.
As play activities and peers became

increasingly enjoyable, adult rewards for
interaction with children were progressively
diminished, and the schedule of nonreinforcement

of adult contacts gradually relaxed. At the end of
the 10-day period, the program was discontinued
and no special contingencies were arranged

thereafter. Observational studies conducted
during several periods following treatment
revealed that the girl continued to display a high
level of social interaction with other children
(Figure 6-7). Not only was the isolated pattern of

behavior successfully modified, but her speech,
which had been slow, drawn-out, and frequently
inaudible, improved considerably. She no longer
demanded unnecessary medication, and she
displayed increased capabilities to assert and
defend herself when necessary.
Similar results were obtained with a combined
extinction-reinforcement procedure in the
treatment of two preschool boys who exhibited

frequent crying episodes whenever mildly
frustrated or thwarted by other children (Hart,

Allen, Buell, Harris, & Wolf, 1964). Observation
revealed that their vociferous crying generally
elicited comforting ministrations and solicitous
concern from teachers. Consequently, teachers

were instructed to pay no attention to the crying
episodes, unless the child was actually hurt. If he
was in close proximity to the teacher when he

began to cry, she walked away or occupied herself
with other duties. On the other hand, whenever

that child handled stressful situations more
constructively he received prompt approving
attention. Within five days after introduction of

the new contingencies the rate of crying declined
from 5-10 times per morning to practically a zero
level and remained negligible thereafter. Etzel &
Gewirtz (1967) achieved equally favorable results
with persistent infant criers by combining

extinction of crying with reinforcement of more
cheerful behavior.
As revealed by the above studies, modification
of behavior through extinction can be
accomplished quickly and predictably by eliciting

and reinforcing desirable modes of response in
addition to eliminating the rewarding
consequences for dysfunctional behavior. Some of
the emotional effects that accompany extinction
can also be avoided in this manner. If, however,

constructive alternatives are only weakly
established or nonexistent in an individual’s
behavioral repertoire, a change agent may have to
wait unnecessarily long or indefinitely for their
appearance. Under these conditions, rather than
relying on the fortuitous occurrence of favorable
events, a change agent can both facilitate and
more effectively control the course of treatment
by providing rewarded models who display

alternative means for securing desired outcomes.
Individuals undergoing extinction alone may learn
what is no longer effective but remain uncertain
about possible constructive courses of action,
whereas the combined use of extinction and

modeling provides considerable positive guidance.
MODIFICATION OF AGGRESSIVE RESPONSE

PATTERNS
Theorizing about the conditions governing the
occurrence and modification of aggressive
behavior has, until recent years, been considerably
influenced by the frustration-aggression
hypothesis (Berkowitz, 1962; Dollard, Miller,

Doob, Mowrer, & Sears, 1939). According to this
view, aggression is the naturally dominant

response to frustration, and nonaggressive
reactions are likely to occur only if aggression has
been nonrewarded or punished. Since aggression
was considered an unlearned response to

frustration, the research stimulated by this theory
was mainly concerned with the effects of
frustration on aggression, with its inhibition and
displacement, and with the occurrence of

catharsis. On the other hand, the crucial issues of
how relatively complex patterns of aggressive
behavior are originally learned, and the influence
of a variety of controlling variables other than
“frustration” were largely ignored.
Man is endowed with neurophysiological
mechanisms that enable him to behave

aggressively, but the activation of these
mechanisms depends on stimulation and is subject

to cortical control. Therefore, the frequency with
which aggressive behavior is displayed, the

specific forms that it takes, the situations in which
it is expressed, and the targets that are selected for
attack are strongly influenced by social
experience. A social-learning theory of aggression

distinguishes the acquisition of instrumental
responses that have destructive or pain-producing
potential from the conditions governing their
subsequent performance. Aggressive response
patterns are characteristically acquired under

nonfrustrating conditions in the absence of
injurious intent and often toward inanimate
objects. Thus, for example, military recruits
acquire and perfect combat skills through many
hours of target practice and simulated skirmishes;

boxers develop hurtful pummeling abilities by
using punching bags and sparring partners whom

they do not necessarily intend to hurt; and
huntsmen acquire the basic rudiments of hunting
by shooting at inanimate targets before they go

out in search of game. Indeed, if aggressive
repertoires were taught only while individuals
were hostilely aroused and entertained injurious
designs, many of the tutors and learners would

probably be maimed during the acquisition phase.
Most theories of aggression single out a limited set

of variables that may influence performance of
aggressive responses, but for the most part they
overlook the essential skill component.
The process of acquisition and subsequent

utilization of aggressive behavior is best
exemplified in laboratory studies employing
markedly different training and test situations.
Walters & Brown (1963) showed that boys who

had been intermittently reinforced with marbles
for punching an automated Bobo doll later
exhibited more physically aggressive behavior
toward other children in a competitive situation
than did boys who had received no prior training
in punching responses. Conversely, in the
previously reviewed study by Chittenden (1942),
children whose aggressive responses were
reduced in a doll-play situation through

differential modeling and vicarious reinforcement
were much less prone to respond aggressively to
interpersonal thwarting, both in situational tests
and in their everyday interactions.
There is a substantial body of evidence

(Bandura & Walters, 1963) that novel modes of
aggressive behavior are readily acquired through
observation of aggressive models. Findings of
these controlled investigations lend support to

field studies demonstrating the crucial role of
modeling in the genesis of antisocial aggressive
behavior (McCord & McCord, 1958), and in the

cultural transmission of aggressive response
patterns (Bateson, 1936; Whiting, 1941). Modeling
influences continue to regulate aggressive
responsiveness to some extent even after the
behavior has been acquired. The behavior of
models continually exerts selective control over
the types of responses exhibited by others in any

given situation. Moreover, seeing individuals
behaving aggressively without adverse
consequences reduces restraints in observers,
thereby increasing both the frequency with which
they engage in aggressive activities (Wheeler,

1966) and the harshness with which they treat
others (Epstein, 1966; Hartmann, 1969).
After aggressive patterns of behavior have
been learned, they can be maintained by a variety

of reinforcing events. Theories that invoke
aggressive drives (Dollard et al., 1939; Feshbach,
1964) assume that pain cues and other injurious

consequences experienced by the victim constitute
the major reinforcers of aggressive behavior. The
process whereby signs of injury and distress
acquire positively reinforcing properties has never
been established. Sears, Maccoby, & Levin (1957)
suggest that expressions of pain and discomfort
produced in others by aggressive behavior are

frequently followed by removal of frustrations or
rewarding outcomes for the aggressor. Through
such paired association pain cues acquire
conditioned reward value. One might also expect
expressions of suffering to gain satisfying

properties under conditions of interdependent
competing consequences in which success for one
member produces punishing outcomes for the
other. Feshbach (1964) offers a somewhat
different interpretation of the phenomenon.

Through example and precept children learn a
retaliation norm: infliction of injury requires that

the initial aggressor must be hurt. It is further
assumed that perception of pain in one’s
tormentors is experienced as satisfying because

successful retaliation restores self-esteem.
To assess the reinforcing function of infliction
of pain, Feshbach, Stiles, & Bitter (1967)
conducted an experiment in which female

students participated in a verbal-conditioning task
after being treated in either a hostile or a friendly

manner by a confederate of the experimenter. For
half the subjects in each condition the correct
verbal response was reinforced by witnessing the
confederate being shocked, while for the

remaining students the contingent event was a
light flash. Angered subjects who observed the
provocateur experiencing pain showed an

increase in conditioned responses, but the
noninsulted subjects did not. These findings would

seem to indicate that under anger arousal pain
cues can serve as positive reinforcers, although
additional information is needed before the

conditioning differences can be unequivocally
attributed to the influence of anger arousal.
Witnessing another person experience pain
evokes emotional responses (Berger, 1962), and
observers who are moderately aroused are more
easily conditioned by such displays than

nonaroused subjects (Bandura & Rosenthal,
1966). Since emotional arousal facilitates
conditioning, an experimental design comparing

the effects of pain cues on fear-aroused and anger-
aroused subjects would be best suited for
evaluating whether conditioning outcomes are
due specifically to anger or to the general effects of

emotional arousal.
A comprehensive theory of aggression must

account not only for aggressive actions that are
primarily reinforced by injurious consequences,
but also for much broader classes of aggressive
behavior in which the infliction of injury is
essentially irrelevant or, at best, a secondary
controlling condition. If there is any validity to
naturalistic studies demonstrating that social
recognition is often made contingent upon
performance of assaultive behavior, and if

aggression is defined as behavior intended to
produce injurious consequences, then some of the
most violent interpersonal activities would be
excluded from consideration. It is evident that
people frequently resort to aggression not for pain

cues but because it has high utilitarian value. By
aggressive behavior, or dominance through
physical and verbal force, individuals can obtain
material resources, change rules to fit their own
wishes, gain control over and extract subservience

from others, terminate provocation, and remove
physical barriers which block or delay attainment

of desired outcomes. It is, therefore, not surprising
that aggressive-domineering patterns of behavior
are so prevalent.
Systematic analysis of reinforcement
contingencies in naturally occurring interactions,
indeed, reveals that aggressive behavior is
strengthened and maintained to some extent by its

positive social consequences. In deviant
subcultures, where physical aggression is
regarded as emulative behavior (Buehler,
Patterson, & Furniss, 1966; Wolfgang & Ferracuti,
1967; Yablonsky, 1962) aggressiveness is often

deliberately rewarded and collectively sanctioned.
Studies of aggression as it occurs in familial and
other social situations disclose that individuals are
often inadvertently trained to behave aggressively
by persons who normally eschew such behavior.

Because of its aversive properties aggressive
behavior not only commands attention, but is

often successful in removing unpleasant
performance demands and in controlling the
behavior of others. Both parents (Hawkins,

Peterson, Schweid, & Bijou, 1966) and peers
(Patterson, Littman, & Bricker, 1967) thus
intermittently reinforce aggressive responding.
A similar reinforcement process is typically
operative at broader social levels. When legitimate
demands and constructive efforts to produce

needed changes are repeatedly thwarted by
persons who benefit from the prevailing system
this evokes more intense and disruptive actions

that cannot be ignored. In many instances, existing
practices lack sufficient justification to withstand
any concerted and aggressive efforts to force
changes. As a consequence, aggressive behavior
eventually succeeds in securing desired goals and,

like any other efficacious, modeled behavior, it is
widely emulated as a method of achieving social
change.
Drive theories of aggression assume that
frustration arouses an aggressive impulse or drive
that can be reduced only through some form of
aggressive behavior. From a social-learning
perspective, frustration is regarded as a facilitative

rather than a necessary condition for aggression.
That is, frustration produces a general state of
emotional arousal that may lead to a variety of
responses depending upon the types of frustration

reactions that have been previously learned, and
the reinforcing consequences typically associated
with different courses of action. This
conceptualization is supported by several lines of
evidence.
Psychophysiological studies, some of which are
discussed in Chapter 8, demonstrate that fear-

arousing and anger-provoking situations produce
highly similar physiological arousal (Ax, 1953;

Schachter, 1957). Moreover, the same state of
physiological arousal induced directly by a

sympathetic stimulant may be experienced as
anger or euphoria depending upon the type of
emotional behavior exhibited by others in the
same situation (Schachter & Singer, 1962). These

data question the assumption that frustration
creates a distinct form of emotional arousal that
can be reduced only through aggressive behavior.
Indeed, contrary to aggressive drive theory,
findings of numerous controlled experiments

reveal that far from producing a cathartic
reduction of aggression, performance of
aggressive behavior within a permissive setting
maintains it at its original level (Feshbach, 1956;
Kenny, 1952; Mallick & McCandless, 1966). It has

also been shown repeatedly that vicarious
participation in aggressive activities increases the

probability that observers will behave in an
aggressive fashion (Bandura, 1965; Berkowitz,
1969).
The persistence of emotional frustration
effects may be more adequately interpreted in
terms of a self-arousal mechanism rather than as a
lingering aggressive drive requiring discharge

through assaultive or destructive behavior. After a
person has been insulted, unjustly treated, or
otherwise thwarted, the resultant emotional
arousal is repeatedly revivified and often
enhanced on later occasions through symbolic

reinstatement of the anger-provoking incidents.
Thus thinking about the ill treatment and the
possible adverse consequences of disturbing
episodes can reinstate intense feelings long after
the initial reactions to the situation have subsided.

The persistence of elevated arousal, according to
the social-learning view, is attributed to self-

generated stimulation rather than to the existence
of an undischarged reservoir of aggressive drive. If
the person should become immersed in new

activities that supersede the arousing ruminations,
or if he should reinterpret the original provoking
experience in a more favorable light, then
“tension” is likely to be reduced noticeably. For

example, a person who is angered over a
presumed snub will probably experience a rapid

and thorough reduction in anger arousal, without
having to assault someone, upon learning that he
has been invited to the social function after all.
Similarly, on the supposition that diminution of

emotional arousal is a consequence of changes in
symbolic activities rather than a cathartic effect of
having experienced aggression vicariously, one

would expect aroused individuals to experience
equally salutary effects from getting involved in an

absorbing book, a movie, a stage play, or a
television program lacking aggressive displays.
Results of several experiments, differing
considerably in subject characteristics, in the form
in which aggression is expressed, and in
dependent measures, are in general agreement
with the foregoing theoretical considerations.
Kahn (1966) subjected college students to anger-

arousing experiences, after which they either
expressed their feelings of anger to an
encouraging and sympathetic “physician,” or they
merely sat for an equivalent period. Students who
participated in the cathartic interview disliked the

provocateur significantly more than the controls,
and during the recovery period, the catharted
subjects were generally more aroused
physiologically than students in the control
condition. In an experiment conducted with

children, Mallick & McCandless (1966) found that
reinterpretation of the anger-eliciting events

substantially reduced aggressive behavior toward
the antagonist, whereas free expression of
physical aggression did not lessen their punitive

behavior. Kaufmann & Feshbach (1963a, b)
provide further suggestive evidence for the anger-
mitigating effects of cognitive processes.
The term frustration has been applied to
exceedingly diverse conditions including the
obstruction, omission, or delay of reinforcement,

the withdrawal of rewards, and the administration
of punishing stimuli. It has been demonstrated in
research involving both human and animal

subjects that these operations produce quite
different outcomes, and even the same condition
does not have an invariant behavioral effect. The
manner in which individuals respond to
conditions regarded as frustrative is primarily

determined by the patterns of behavior that they
have previously learned for coping with such
situations.
The importance of direct training in the
development of frustration responses is
demonstrated in an experiment by Davitz (1952).
After first being observed in free interaction,
several groups of children participated in training

sessions in which competitive and aggressive
behaviors were praised and encouraged. In
contrast, other groups of children were rewarded
for constructive and cooperative behavior. All

children were then severely frustrated and
immediately following this experience they were
observed for a second time in free interaction.
Children who had been trained to behave
aggressively showed an increase in aggression,

whereas children who had received training in
cooperativeness behaved more constructively in

response to frustration. This study illustrates how
frustration serves as an arousal stimulus that

enhances whatever responses are dominant in
subjects’ behavioral repertoires.
The influential role played by social-learning
factors in aggression is further shown by studies
in which aggressive behavior is induced in
primates through electrical stimulation of the

hypothalamus (Delgado, 1967). Thalamic
stimulation of a monkey who assumes a dominant
role in the colony instigates him to attack
subordinate male members. By contrast, thalamic
stimulation elicits cowering and submissive

behavior in a monkey of low social rank. Even
more impressive is evidence that electrical
stimulation of the same cerebral mechanism can
evoke differential amounts of aggression in the
same animal as his social rank is modified by

changing the membership of the colony. Thus,
thalamic stimulation elicits submissiveness in the

animal when he occupies a low hierarchical
position and marked aggressiveness when he is
the dominant member in the group.
In human learning, responses to frustration
frequently originate from observation of parental
and other models who provide repeated examples
of how to deal with thwarting events.

Consequently, when observers encounter stressful
situations they are more likely to respond
imitatively than to engage in initial trial-and-error
behavior. Only when a person has learned
aggression as a dominant response to emotional

arousal will there be a high probability of his
reacting aggressively to frustration. For example,
Bandura (1962) found that children who had
observed a model behaving in an aggressive
manner responded to frustration with kicking,

striking with mallets, and other imitative
aggressive responses, while equally frustrated

children who had watched a nonaggressive model
displayed less aggression than a control group that
matched the inhibited behavior of their model.

The influential role of modeling in shaping
frustration or stress reactions is well documented
in countless studies showing that deviant behavior
patterns are often transmitted through familial

modeling.
It is evident that, as a result of differential

modeling and reinforcement patterns, frustration
may elicit a wide variety of responses. When
thwarted, some people become dependent and

seek help and support, some display withdrawal
and resignation, some experience psychosomatic
dysfunctions, some seek refuge in drug-induced
experiences and anaesthetic doses of alcohol,
some respond aggressively, and most simply

intensify constructive efforts to overcome the
obstacles they face. It therefore comes as no
surprise that in laboratory studies in which
frustration is systematically varied it sometimes
increases aggression (Berkowitz, 1964; Geen,
1968; Hartmann, 1969), has no effect on
aggressive behavior (Buss, 1966; Jegard &
Walters, 1960; Walters & Brown, 1963), or
reduces aggressive responding (Kuhn, Madsen, &

Becker, 1967). In experiments reporting positive
results frustration usually exerts an influence only
in conjunction with prior training in aggression or
exposure to aggressive modeling cues. The fact
that the negative findings occurred in studies in

which other variables were highly influential lends
support to the view that frustration is only one,
but not necessarily the most important, variable
determining aggressive behavior. Indeed,
according to social-learning theory, one could

readily produce highly aggressive individuals by
providing them with successful aggressive models

and intermittently rewarding aggressive behavior,
while keeping frustration at a low level. It would
follow from the findings reviewed in the preceding

sections that lasting changes in aggressive
behavior can be most successfully achieved by
reducing the utilitarian value of aggression
through the development of more effective

alternative modes of response.
A variety of social-learning procedures has

been employed with success in modifying extreme
aggressive behavior. Chittenden (1942) achieved
reductions in aggression by modeling more

constructive means of coping with interpersonal
conflicts. Several investigators (Hawkins et al.,
1966; Sloane, Johnston, & Bijou, 1968; Zeilberger,
Sampen, & Sloane, 1968) have eliminated violent
temper tantrums and physically assaultive

behavior by reducing the amount of social
reinforcement that parents and teachers provide
for such behavior. In the foregoing programs
aversive consequences, usually consisting of social
withdrawal, are administered for physical assault
and destructive behavior, and desirable social
behavior is actively promoted.
Aggression has also been effectively modified
by differential social reinforcement in which

aggressive behavior is placed on an extinction
schedule. Brown & Elliott (1965) instructed
teachers to ignore aggression and reward
cooperative behavior in an effort to reduce the
amount of aggression exhibited by 27 boys in a

nursery school class. Under these reinforcement
contingencies the incidence of both physical and
verbal aggression declined. After the program was
discontinued physically aggressive behavior

showed some recovery during a follow-up period,
which was attributed to the fact that teachers
found it difficult not to attend to and interact with

the boys when they engaged in such activities. The
social reinforcement procedures were again
consistently applied and produced additional
reductions in physical and verbal assaults. Scott,
Burton, & Yarrow (1967) report similar results in
a controlled study of a nursery school boy who
displayed frequent unprovoked aggression. When

the usual conditions of adult reinforcement were
in effect the boy exhibited a high rate of negative
behavior toward peers. In contrast, during periods
when significant adults consistently ignored
aggressive actions and concurrently attended to

desired behavior the boy showed a substantial
increase in positive forms of interaction with other
children.
Extinction of Defensive Behavior
As Mowrer (1950) has previously noted,

human behavior is frequently activated not by
immediate physical discomforts but by anticipated
aversive effects. That is, housewives do not
depend upon hunger pangs to prompt them to
purchase groceries; homeowners do not wait until
they experience the discomfort of a burning house
to buy fire insurance; students ordinarily do not
rely upon distress created by examination failures

to begin to study; and usually motorists do not
wait until inconvenienced by a stalled automobile
to replenish gasoline. Through representational
mechanisms future events can be converted into
current stimuli that are functionally similar to

physical stimuli in their capacity to evoke adaptive
courses of action.
Similarly, avoidance behavior can be strongly
reinforced by its success in preventing the

occurrence of anticipated painful experiences. This
process is illustrated by the apocryphal case
report of a severe compulsive who, when asked by

a psychiatrist why he incessantly snapped his
fingers, replied that it kept ferocious lions away.
When informed that there were no lions in the
vicinity, the compulsive client responded, “See, it
works.” The reality of the phenomenon is most
strikingly demonstrated by laboratory studies of
unsignaled avoidance in which animals can

forestall shock for a fixed period each time they
perform a defensive response (Sidman, 1966).
Under these conditions animals display a stable
rate of avoidance behavior and, as a result, they
rarely encounter the actual punishing events.

Moreover, the avoidance behavior persists for a
long time after aversive stimuli have been
withdrawn, and it is easily reinstated on later
occasions by a few negative experiences.
After successful avoidance behavior has been

developed it may be controlled cognitively and by
discriminative stimuli without requiring

emotional arousal. In a comparison of different
extinction procedures, Notterman, Schoenfeld, &
Bersh (1952) conditioned heart-rate responses to
a tone through shock stimulation. When the
extinction trials began one group of subjects was
presented the tone without comment, the second
group was informed that the shocks had been

permanently discontinued, while the third group
was told that they could successfully avoid being
shocked by tapping a telegraph key whenever the
tone was presented. Awareness of the altered
contingencies facilitated extinction, but autonomic

responding was almost completely eliminated
when subjects had a suitable means of avoiding
aversive stimulation (Figure 6-8).
Under naturally occurring conditions

individuals periodically encounter punishing
experiences and frequently find themselves in
fear-provoking situations. Defensive activities are,

therefore, reinforced not only by forestallment of
potential threats but also by fear reduction
accompanying escape from aversive situations
that produce disturbing arousal. In experiments
Figure 6-8. Extinction of heart-rate responses to a
conditioned stimulus in subjects who either were
uninformed that shock was discontinued, were told
there would be no further shocks, or were provided
with a motor response for avoiding shock. B-l
represents the baseline heart-rate response to the tone
before aversive conditioning. Redrawn from
Notterman, Schoenfeld, & Bersh, 1952.
designed to evaluate separately the various factors
that might reinforce avoidance behavior, Kamin
(1956, 1957) found that either termination of the

fear-provoking signal or avoidance of physically
painful stimulation increased the frequency of
avoidance responses and the speed with which
they were performed. Avoidance behavior was
most pronounced when it both terminated fear-

arousing cues and prevented shock stimulation.
Extinction of avoidance behavior is achieved
by repeated exposure to threatening events
without the occurrence of any adverse

consequences. The major obstacle to obtaining
rapid extinction is the self-reinforcing character of
avoidant behavior deriving from its capacity to

remove or postpone anticipated threat. Moreover,
inhibition of responses that have been punished in
the past and the successful avoidance of fear-
provoking situations effectively prevent the
individual from reappraising the currently
prevailing conditions of reinforcement.
Anticipatory arousal and defensive behaviors that

are no longer objectively justified are thus
protected from extinction. Continued reexposure
to fear-producing stimuli without unfavorable
consequences eventually eliminates both
emotional and instrumental behavior through

mechanisms previously discussed. Several variant
extinction procedures have been devised in an
effort to accelerate and to gain better control over
extinction processes.
FORCED EXPOSURE THROUGH RESPONSE

PREVENTION
Avoidance behavior can be rapidly eliminated
by blocking its occurrence in the presence of fear-

arousing stimuli. However, there is some evidence
to indicate that forced exposure may produce only

temporary changes without altering the arousal
potential of subjectively threatening situations. In
some instances, for example, cessation of
avoidance responding simply reflects the

formation of erroneous discriminations that
protect the fear-eliciting capacity of threatening
stimuli from extinction. Solomon, Kamin, & Wynne
(1953) trained dogs to jump over a barrier in a

shuttle box whenever a buzzer was sounded
preceding an intense electric shock. After the
avoidance behavior was well learned, extinction
was instituted. Under the regular extinction
procedure, the animals continued to perform the

effortful avoidance response to the buzzer with
undiminished speed for several hundred trials

without receiving a further shock. At this stage of
the investigation, various modification procedures

were introduced. For some animals, a glass
barrier, which prevented the jumping response,

was placed between compartments of the shuttle
box, but this method also proved ineffective in
facilitating extinction for most animals. In this
experiment, the physical obstruction was utilized

only on the fourth and seventh of the ten trials
which the dogs received during each daily session.

Under these predictable stimulus changes the
animals apparently discriminated between trials
on which the barrier was present as being safe and
those on which it was absent as dangerous.

Consequently, they continued to jump rapidly on
the latter trials, but remained unperturbed by the
aversive tone whenever the glass barricade was

introduced.
Evidence for the discrimination interpretation

of the marked resistance to extinction is provided
by Carlson & Black (1959), who replicated the
above experiment, with the exception that the
glass barrier was employed throughout the initial
series of extinction trials, after which it was
permanently removed. Under this procedure,
avoidance behavior was rapidly eliminated. Using
a similar method, Page & Hall (1953) likewise

demonstrated that the response-prevention
technique can accelerate extinction, provided it is
employed on every trial in a lengthy series during
the initial phase of extinction. Weinberger (1965)
has further shown that the rate of extinction of

avoidance behavior is increased with longer
durations of forced exposure to fear-provoking
events.
Response blocking in the presence of aversive

stimuli can produce behavioral changes through
several different means. It may extinguish the
aversive properties of threatening stimuli so that

they lose their capacity to evoke fear and
avoidance. Alternatively, it may eliminate the
obstructed avoidance responses without altering
the arousal potential of feared stimuli by
producing new forms of defensive behavior that
are inevitably successful in forestalling
nonexistent threats. This process of response

substitution is best exemplified by Miller’s (1948)
experiment, in which animals confined in a
threatening situation acquired a long succession of
avoidance responses as each preceding one was
obstructed. The importance of distinguishing

between changes reflecting stimulus
neutralization and response substitution is further
illustrated by evidence that subjects that have
extinguished avoidance responding to a given CS
may nevertheless be somewhat fearful of that

stimulus, as measured by suppression of rewarded
behavior whenever the stimulus is presented

(Kamin, Brimer, & Black, 1963).
Assessments of the varied effects
accompanying response prevention suggest that
this method may produce rapid behavioral
changes without achieving fear extinction. This is
shown by evidence that, compared to regular
extinction, behavior that has been eliminated by

response blocking reappears with greater
frequency to later presentations of the
conditioned stimuli (Benline & Simmel, 1967);
subjects are more reluctant to approach the fear-
provoking cues, indicating that they have retained

some of their aversive properties (Page, 1955);
and not only are subjects whose defensive
behavior has been removed in this manner more
susceptible to subsequent aversive conditioning,
but the reestablished avoidance behavior is more

resistant to extinction (Polin, 1959).
Response-prevention methods have rarely

been employed clinically. However, Meyer (1966)
presents interesting results with severe
obsessional disorders, which suggest that this
approach may have value in modifying certain
behavioral conditions. One case involved a 33-
year-old woman who was almost totally
incapacitated by washing rituals. She phobically

avoided touching common objects and
discontinued sexual relations because of fear of
contamination. Most of her time was spent
compulsively washing and scrubbing the house.
The second woman suffered for 36 years from

intrusive “blasphemous thoughts” that centered
around having sexual intercourse with the Holy
Spirit. The resultant guilt and anxiety were
alleviated by performing various ritualistic
behaviors a certain number of times. These

obsessive thoughts and rituals remained unaltered
by electric shock therapy and a leucotomy but,

according to the client, the condition was
aggravated by 11 years of psychoanalysis where
most stimuli were interpreted as sexual symbols.

Consequently she stopped eating oblong objects,
and any activity that could conceivably have
sexual connotations (e.g., shutting drawers,
putting in plugs, wiping tall receptacles, entering

underground trains, etc.) evoked ritualistic
behavior.
The clients were requested to perform
threatening activities (e.g., touching door knobs,
handling dust bins, imagining sexual relations with

the Holy Spirit, eating sausages, etc.) and the
nursing staff prevented them from engaging in the
ritualistic behavior designed to forestall
foreboding consequences. The women displayed
intense distress when performance of the

ritualistic behavior was first blocked. However,
their emotional reactions gradually diminished,
and both avoidance behavior and the compulsive
rituals were substantially reduced after the
restrictions had been removed. According to
follow-up studies, the first client continued her
washing routines, but she was much less disturbed
by dirt, her family relationships improved greatly,
she resumed sexual relations, and she was able to

participate in a number of social activities which
she previously had avoided for fear of
contamination. The second client decreased the
ritualistic behaviors from approximately 80 to 4
per day and the occurrence of intrusive thoughts

was similarly reduced. These encouraging
preliminary results suggest that avoidance
behavior that is powerfully maintained by staving
off fantasied direful consequences may initially
require a forced extinction procedure for their

elimination.
EXPOSURE TO THREATS GRADUATED IN

AVERSIVENESS
Inappropriate defensive behavior is frequently
extinguished by introducing aversive stimuli at

weak intensities that do not evoke avoidance
responses, and then gradually increasing their
threat value until the most fearsome situations
have been neutralized. By skillful application of

stimulus change procedures the arousal capacity
of aversive stimulus events can be eliminated
without eliciting fear or alternative forms of
defensive behavior.
Several studies have compared the relative
efficacy of regular extinction and a stimulus

change procedure used either alone or as a
component of a multiple method for eliminating
emotional responding. In an experiment by Kimble
& Kendall (1953) animals performed avoidance
responses to a light (CS) that was previously

associated with shock. The avoidance responses of
half the animals were extinguished by the

conventional method of repeatedly presenting the
fear-provoking CS at training intensity without the

shock. For subjects in the second condition the
intensity of the CS was gradually raised in small
steps from zero to the level used during training,
and thereafter the light was presented at the

training intensity as in the first group. Subjects
initially exposed to the conditioned aversive
stimulus at graduated intensities abandoned the
avoidance behavior twice as fast as the group
confronted from the outset by the CS at full

intensity. In fact, 80 percent of the animals in the
stimulus change condition rapidly extinguished
emotional responses to weaker versions of the CS
and consequently, they displayed no avoidance
behavior at all when confronted with the formerly

aversive stimulus at high intensity.
A graduated procedure can produce

comparatively rapid extinction because
superseding competing responses are more likely
to occur to weak aversive stimuli than to more
intense forms which activate strong avoidance
behavior. Once nonavoidant responses occur to
situations containing few fear-provoking
elements, the competing responses generalize also

to similar stimuli containing more fearsome
elements. In this way it is possible to extinguish
emotional behavior without eliciting any
avoidance behavior, provided the conditioned
aversive stimuli are increased in sufficiently small

increments.
The process of extinction can be further
hastened by combining aversive stimulus change
with positive stimulus conditions designed to

evoke behavior capable of supplanting avoidance
tendencies. By employing this type of multiple
procedure, which is treated at length in the next

chapter, the occurrence of nonfearful behavior to
subjective threats can be better controlled. This is
corroborated by Poppen (1968) in a laboratory
study comparing five different methods for
eliminating behavioral inhibitions. After animals
learned to press a lever for food, they were
shocked immediately following a tone until it

acquired the capacity to suppress responding
thoroughly. The shock stimulation was then
discontinued and the animals were assigned to
one of five treatment conditions. For one group of
subjects receiving regular extinction, the fear-

arousing tone was presented periodically at the
intensity used in training; for a
counterconditioning group the training tone was
accompanied by food rewards; a third group was
administered graduated extinction in which the

aversive tone was introduced at low intensity and
progressively increased as weaker variants were

neutralized; for subjects receiving the graduated-
counterconditioning treatment aversive stimulus
change was combined with food rewards; and

finally, a fifth group of subjects participated in a
flooding procedure wherein the training tone was
presented continuously for 10-minute periods.
The degree of response suppression was

measured by differences in rate of responding
prior to, and in the presence of, the threatening

tone.
The reductions in response suppression
achieved by the various treatment procedures are

summarized in Figure 6-9. Regular extinction was
found to be least effective for eliminating
conditioned fear, but the potency of this method
was considerably enhanced by presenting the
threatening stimuli in a gradual fashion.

Behavioral inhibitions were most rapidly and
thoroughly removed by reducing the aversiveness
of fear- provoking stimuli through graduated
presentation, and simultaneously eliciting
incompatible responses. An essentially similar
pattern of results was obtained for the number of
1246
Figure 6-9. Reductions in response suppression achieved by subjects in each of the
five treatment conditions during ten extinction sessions. A value of
zero indicates complete suppression of the intermittently rewarded
lever-pressing response, whereas a ratio of 0.50 represents no
response inhibition. Poppen, 1968.
trials required to eliminate completely the
behavioral inhibitions. Presenting fearsome
stimuli in a graded series and eliciting competing

responses both accelerated extinction, but the
procedure combining these two factors reduced
extinction time by half.
A number of case studies have been published
in which a graduated extinction procedure was

employed to modify severe emotional behavior.
An illustration of the use of this principle is
provided by Grossberg (1965), in the treatment of
a woman who suffered from a public speaking

phobia so incapacitating that she was unable to
complete a speech course required for college
graduation, even with the aid of tranquilizers,

group therapy with other speech phobics, and 30
hours of individual psychotherapy.
The extinction program consisted of 17
sessions in which the student delivered
increasingly longer speeches to progressively
larger audiences in situations that gradually

approximated speech-class conditions. Initially the
student read familiar and unfamiliar passages
from a book and then made brief speeches to the
therapist alone in his office and in a small, empty

classroom. Auditors were later introduced one by
one until she delivered a variety of speeches to an
audience of nine listeners. To further ensure
adequate generalization of extinction effects, the
stimulus conditions were continuously varied by

utilizing several different classrooms, by having
the student visualize her new classmates during

demonstration speeches, and by having the
therapist himself absent during some of the

sessions. At the completion of the extinction
series, the student delivered six speeches and

attained a grade of “B” in the speech course from
which she had previously fled to a physician after
a frightful struggle to complete a one-minute
oration.
According to psychoanalytic theory,
interpretations designed to reduce the strength of

defenses should precede the labeling of impulse
expressions. On the basis of deductions consistent
with the latter theory made from Miller’s (1948)

conflict paradigm, Dollard & Miller (1950)
advanced the view that anxiety that motivates
avoidance responses in an approach-avoidance
conflict should be reduced before attempting to
actuate approach behavior.
The person with a severe neurosis who does

reach the psychotherapist is a specially
selected case with extremely strong
avoidance tendencies. Therefore, trying to
increase his motivation to approach goals
will only increase his fear and conflict. This
increase in misery will tend to drive him out
of therapy. This is indeed what seems to
happen. Therapists have found that the first
thing to do is to concentrate on reducing the
fears motivating avoidance (i.e., to analyze
resistances) rather than to try to increase the
motivation to approach the feared goal [p.
359].
This theory implies that avoidance behavior

can be modified most effectively by interpretive
interview procedures and that no attempt should
be made, during initial stages of treatment, to have
clients perform the feared behavior. It is highly
probable that if therapists were to force their

clients to approach the most fear-provoking
situations at the outset, they would indeed
experience intense anxiety, and might even
terminate psychotherapy. However, the
experimental evidence previously reviewed
demonstrates that avoidance responses can be

readily extinguished if subjects are exposed to
initially weak, but gradually increasing, aversive
stimuli. An extinction procedure of this type was,
in fact, also successfully employed by Herzberg
(1941) in the treatment of an agoraphobic
housewife. This woman displayed severe anxiety

and psychosomatic reactions whenever she went
out alone; she consequently refused to leave the
household unless accompanied by another person,
or transported in a taxi. She was first assigned the
task of walking by herself in a park, which

constituted a considerably weaker threat than
walking in the street. The anxiety reactions to the
park situation were readily extinguished, and she
was then instructed to walk alone on a quiet street
in her neighborhood. In this way, the client was

gradually reexposed to progressively more fear-
provoking cues, until eventually she would walk

almost anywhere alone without experiencing
anxiety or psychosomatic reactions. Herzberg
(1945) has employed similar graded performance

tasks in conjunction with interview methods for
eliminating varied forms of avoidance behavior
and for promoting new response patterns. In some
of these cases, however, clients are presented

relatively demanding tasks with insufficient
preparatory experiences that would assure more

effective progress.
It should be noted in passing that treatment
approaches utilizing graded performance tasks are

just as consistent with Miller’s conflict paradigm
as theories that advise focusing on avoidance
tendencies. That is, unpunished evocation of
weakly inhibited responses produces extinction
effects that will generalize to the more strongly

inhibited forms of behavior, thus reducing the
entire avoidance gradient. In this manner anxiety
associated with successively closer variants of the
desired behavior can be progressively
extinguished until clients are able to execute the
goal responses without experiencing undue
emotional arousal. This strategy has, in fact, been
successfully applied to the modification of
agoraphobias (Jones, 1956; Meyer, 1957; White,

1962), claustrophobias (Meyer, 1957; Walton &
Mather, 1963a), compulsive response patterns
(Walton, 1960b), school phobias (Garvey &
Hegrenes, 1966; Kennedy, 1965), severe sexual
inhibitions (Haslam, 1965; Walton, 1960c), and

more circumscribed avoidance responses
(Freeman & Kendrick, 1960). In a series of
interesting individual studies Foster (1967; Foster
& Campos, 1964) was able to ameliorate clinical
seizures and EEG dysrhythmia evoked by

stroboscopic stimulation or certain musical
selections through repeated presentation of the

eliciting sensory stimuli initially at innocuous
levels and gradually approaching the evocative
forms.
Walton & Mather (1963b) report that similar
extinction procedures yielded variable results
with obsessive-compulsive behavior that
presumably was originally conditioned to

response-produced stimuli rather than to
environmental cues. In attempting to account for
these diverse outcomes, Walton & Mather
distinguish between treatment strategies aimed at
extinguishing “the more basic conditioned

autonomic drive (CAD)” from those directed
toward the elimination of avoidance behavior to
generalized environmental stimuli. According to
their reasoning, in acute anxiety disorders
instrumental avoidance responses are elicited by

the underlying conditioned autonomic drive;
consequently, in the latter condition, treatment

should concentrate on the extinction of autonomic
responsiveness since its removal will eliminate
associated avoidance responses without any direct

intervention.
To support this supposition, Walton & Mather
cite two successfully treated males who displayed
severe obsessive-compulsive disorders of recent

origin. In the one case, a handwashing ritual,
which was believed to be evoked by anxiety and

guilt over violently aggressive fantasies,
disappeared after the passive client received
training in self-assertive behavior. Similar
development of self-assertiveness in the second

client resulted in reduction of obsessional
thoughts about homosexuality and
destructiveness, which were assumed to arise

from anticipatory concern over negative social
reactions to his obsequious behavior.
In behavioral disorders of long standing,
Walton & Mather maintain that cues other than
those originally involved in the aversive
conditioning may, through the process of stimulus
generalization, acquire eliciting potency so that
avoidance responses become “functionally
autonomous” or partially independent of the
chronologically earlier CAD. Hence, modification of

chronic disorders would require the extinction of
both the initial conditioned autonomic responses
and the avoidance behavior. Results from several
chronic cases, which displayed only partial
improvement when their treatment was restricted

either to the original conditioned stimuli or to the
avoidance responses themselves, are presented as
tentative support for the latter hypothesis.
The assumptions made by Walton & Mather

about the conditions regulating avoidance
behavior are disputed by a substantial body of
evidence that avoidance behavior is not under

autonomic control. These findings, which are
discussed at length in Chapter 7, support the view
that autonomic and instrumental avoidance
responses are coeffects of aversive conditioning
rather than causally linked events. When
emotional responses are conditioned to a
particular stimulus, other cues falling on the same

physical or semantic stimulus dimension also
acquire eliciting potency through the process of
generalization. A systematic behavioral
assessment would most likely reveal that both
primary and secondary stimuli evoke autonomic

and avoidant responsiveness. Nor is the range of
stimulus generalization necessarily determined by
temporal factors. The autonomic-motor focus in
treatment approaches proposed by Walton &
Mather, in fact, reduces to the question of whether

emotional reactions should be extinguished to
primary or to generalized stimuli rather than to a

drive-behavior distinction. The outcomes reported
by the authors are completely in accord with
predictions from the principle of generalization,

that reduction in emotional behavior will be
greatest toward the stimuli that have been
neutralized, regardless of where they happen to
fall on the generalization gradient. The

decrements in conditioned emotionality would
become progressively smaller the farther the

nontreated evocative stimuli are removed from
those selected for extinction treatment.
In one of the reports (Walton & Mather,

1963b) that lends support to the above
formulation, a 24-year-old single female who had
had an extremely moralistic upbringing suffered
from severe sexual anxieties. Any form of physical
or social contacts with men, and even sexual

intercourse in wedlock, was considered sinful.
Following adolescence, when exposure to sexual
information and a masturbatory episode
generated intense guilt feelings, the child’s marked
sexual anxieties generalized to urogenital
functions. During this time she developed an
obsessive concern about urination and defecation,
and instituted elaborate toilet rituals designed to
ensure complete cleanliness. The anxiety

responses further transferred to animal feces and
urine so that she also carefully avoided park seats,
lamp posts, and chairs in private homes containing
pets. Because of her inability to use public
lavatories and benches, and her marked

curtailment of social interactions with men, the
woman was eventually forced to give up her job.
The treatment was designed specifically to
decrease sexual anxieties by pairing drug-induced

relaxation with scenes of progressively more
intimate interactions with men. As a result, her
anxiety responses to heterosexual stimuli were

markedly reduced:
She could pass men in the street, sit next to

them on public vehicles, wait in shop or bus
queues with them and speak to them. She
related two such incidents with satisfaction.
She had waited with a young man, a stranger,
for half an hour at a bus stop and had
become engaged in a lengthy conversation.
This almost resulted in a date. On a second
occasion she renewed a childhood
acquaintanceship with a young man of her
age [p. 167].
In accord with generalization principles, the
client’s generalized fear of urination, defecation,

and excreta, stimuli far removed from the primary
sexual stimuli, was only partially reduced.
In a second set of outcome data presented by
Walton & Mather a single woman developed

obsessive concerns about contamination by dirt,
and compulsive handwashing rituals, subsequent
to a guilt-producing love affair with a married
man. In this particular case, however, sexual
anxieties were untreated but compulsive

responses to generalized eliciting stimuli were
extinguished. The woman was required to perform
a series of tasks graded according to their
contamination value and potency in evoking
handwashing, such as use of public washbowls,

toilets, and seats, touching door knobs, picking up
objects from floors, and walking along dusty
thoroughfares. The client’s compulsive behavior
was substantially reduced by this program but her

sexual anxieties were undiminished.
It would appear from the data discussed above,

together with laboratory findings on the
generalization of extinction effects (Bass & Hull,
1934; Hoffeld, 1962; Hovland, 1937), that
decisions about whether to orient an extinction
program toward primary or generalized stimuli,

or both types of events, should be determined by
the nature and range of changes that one wishes to

produce.
An interesting group application of graduated
extinction is described by Saul and his associates
(Saul, Rome, & Leuser, 1946), in the treatment of
severe and widely generalized anxiety arising

from traumatic military experiences. The soldiers
were shown, in a secure and relaxing context, a
graded series of movies of battle scenes beginning
with exposures that they could tolerate. Initially

the films depicted preparatory battle scenes,
followed by motion pictures of surface and aerial
bombardment from which displays of injury and
destruction were deleted. In later sessions the
soldiers were gradually presented more

frightening combat scenes. In addition to
regulating the aversiveness of pictorial stimuli, the

presentation of anxiety-arousing combat noises
was likewise controlled. At first the battle scenes

were shown silently, and only gradually was
sound introduced. Day by day the sounds of

gunfire, explosions, and aerial bombardment were
increased until full intensity was reached. As a
further safeguard against excessive emotional
arousal, each soldier was provided his own

volume control with which he could regulate the
amount of aversive stimulation.
An average of 12 showings of approximately
15 minutes each effectively extinguished the
soldiers’ intense emotional responses, as shown

by reactions of calm, and even boredom, to scenes
that had previously terrified them. Additional
evidence that the soldiers had been successfully
desensitized is provided by their relatively
undisturbed responses to a test film of a Marine

invasion depicting intense combat and severe
casualties. Moreover, they were able to attend
commercial movies, which most of them had
previously avoided because of the newsreels, and
they displayed a generalized diminution of
emotional responsiveness to a variety of sounds,
noises, and even music to which they had been
formerly hypersensitive.
Results of the above study cannot be fully

evaluated in the absence of an untreated control
group and more systematic assessment of changes
in emotional responsiveness. However, the
favorable outcomes yielded by modeling studies
utilizing films graduated in aversiveness (Bandura

& Menlove, 1968) suggest that group extinction
procedures involving pictorially presented threats
could be employed effectively to extinguish
common fears that are no longer appropriate.
ROLE OF POSITIVE INCENTIVES IN EXTINCTION
The selection of appropriate performance
tasks and their sequential arrangement is usually

given detailed consideration, but the important
role played by incentive factors in extinction
programs is often overlooked. Even though a
change agent has planned an optimal sequence of
activities, his efforts will be of little avail unless
individuals carry out the necessary procedures
that have been prescribed for them. Considering

that the program, if it is to be successful, requires
people to enter into threatening situations that

were previously avoided and to perform fear-
provoking responses, some degree of reluctance
or even opposition is not unexpected. Fortunately,
there are several factors that support approach

efforts despite anxiety. The distress and impaired
functioning created by inappropriate fears and
inhibitions, coupled with expectations of eventual
benefits, undoubtedly serve as strong
inducements to engage in formerly inhibited

activities. In addition, social rewards in the form of
interest and approval from change agents and
other significant individuals function as positive
incentives for performing essential behaviors.
Finally, the skill with which extinction experiences
are organized is an influential factor facilitating or
hampering behavioral change. If people are

initially encouraged to carry out inhibited
behaviors under highly favorable conditions, the

possibility of adverse outcomes, which might
jeopardize positive motivation, is minimized. By
making the progression in each successive
assignment so gradual that failures seldom occur

the rewards associated with continual tangible
progress will help to strengthen willingness to
attempt more difficult tasks. In some cases,

however, the change agent may have to introduce
more powerful positive incentives to keep

individuals in subjectively threatening situations.
Moreover, when the extinction program is self-
managed in everyday situations, as is frequently

the case, performance tasks must be specified in
considerable detail if they are to be implemented
by clients with high probability of success.
The influential role of feedback in sustaining

and accelerating extinction of phobic behavior is
shown by studies in which precise feedback of
performance is sequentially added and removed.

In one experiment (Leitenberg et ah, 1968)
claustrophobic and knife phobic women were
instructed to engage in feared activities for

progressively longer durations, under conditions
where for each trial, they recorded the exact time
spent in a small room or looking at knives or did
not receive any time scores. Explicit feedback
facilitated behavioral change, omission of time

scores produced a decline in performance, and
reinstatement of feedback led to renewed
improvement. However, adding praise to
informative feedback did not further enhance the
rate of progress. Using a similar sequential design,
Agras, Leitenberg, & Barlow (1968) demonstrated
that performance feedback was also a powerful
factor in eliminating severely agoraphobic
behavior (Figure 6-10). These findings indicate

that failure to recognize progressive improvement
in performance can seriously hinder progress and
create unnecessary feelings of discouragement.
It is important to bear in mind that not all
avoidance behavior necessarily represents an

anxiety problem. In some instances, the original
aversive conditions have ceased to operate and
the avoidance behavior is, in fact, primarily
maintained by its positive consequences. A school-

phobic child, for example, may continue to avoid
scholastic situations after they have lost their
threatening value because of increased attention

and other rewards associated with remaining at
home. Under these conditions, a fear-extinction
program would be inappropriate and fruitless. If
any significant behavioral change is to be achieved
1269
Figure 6-10. Effect of social reinforcement and
nonreinforcement of performance improvements upon
the rate of progress of two agoraphobic clients. Agras,
Leitenberg, & Barlow, 1968.
the rewards associated with avoidance behavior
must be withdrawn and made contingent upon
more beneficial modes of response.
In many instances avoidance behavior is
supported by both positive and negative
reinforcers. Hence, partial attainment of the
treatment objectives may produce some
disappointment because of the loss o benefits

formerly derived from the behavior disorder. In
such case; adequate substitute rewards must be
provided. It may also be advisable to delay
temporarily the removal of deviant behavior that

has high functional value until alternative sources
of reward are established. One must, therefore,
identify the factors maintaining deviant behavior

before embarking on change programs, and utilize
this information in preparing individuals for
changes in accustomed reinforcement that their
recovery will most likely produce.
EXTINCTION THROUGH PROLONGED OR MASSIVE
EXPOSURE TO AVERSIVE STIMULI
In the preceding extinction approach, aversive

stimuli are initially presented at low intensities
that are easily tolerable, and more stressful
situations are gradually introduced as emotional
responses to weaker threats are progressively
eliminated. Considering that in laboratory

investigations extinction is typically carried out in
relation to aversive stimuli at training intensity, it
is evident that fear extinction can be achieved
without stimulus graduation. Indeed, even
prolonged or massive exposure to aversive stimuli

at high intensities may produce rapid and stable
extinction of avoidance responses.
Polin (1959) trained animals to jump a hurdle
at the sound of a buzzer in order to avoid electric
shock. The animals were then given four days of
differential extinction training: One group
received 20 trials daily of five-second exposures to
the buzzer with a physical barrier erected to
prevent the avoidance response; the “flooding”

group each day received 100 seconds of
continuous auditory stimulation in a free-
responding situation; the control group was
merely given a four-day rest. In subsequent phases

of the experiment, all animals received an identical
series of regular extinction trials, followed by two
days of avoidance reconditioning in which the
buzzer was again associated with electric shock,
and a final series of regular re-extinction trials.
As summarized graphically in Figure 6-11,
animals that experienced long durations of

continuous exposure to the fear-arousing stimulus
extinguished avoidant responding considerably

faster than either the control or barrier group in
both extinction phases. The results further

revealed that extinction based on forced exposure
Figure 6-11. Rate of extinction and reconditioning of
avoidance behavior eliminated by different extinction
procedures. Polin, 1959.
through physical restraints had no lasting value
and may, in fact, have prevented the elimination of
avoidance behavior. If responses are physically

restrained, they cannot occur and therefore they
cannot be eliminated through nonreward. Thus,
initially, the physically restrained subjects showed
more rapid extinction than the controls, but in
both the later phases and in the re-extinction, the

barrier group displayed a much higher incidence
of avoidant responding even though the groups
started at the same reconditioned level.
Stampfl (Stampfl & Levis, 1967) has developed

a method of treatment, entitled implosive therapy,
which is based on massive exposure of clients to
highly aversive stimuli in imaginal form.

Evaluation of this particular approach is
somewhat complicated by the fact that the
conceptualization of psychological disorders
appears to have limited relationship to the
extinction procedure actually employed. It is
assumed that stimuli most closely associated with
traumatic experiences are invested with intense

anxiety and are, therefore, repressed and
inaccessible. Other stimuli more remotely
connected with the trauma also acquire anxiety-
arousing properties, but to a lesser degree. These
weaker aversive stimuli, that are experienced as

frightening, elicit avoidance behavior even though
the focal threat is absent. The avoidance responses
activated at the early stage of the stimulus
sequence successfully protect the remaining, more
threatening elements from extinction.
In laboratory studies cited in support of the
above formulation, Levis (1966, 1967) employed a

paradigm in which animals underwent regular
aversive conditioning except that several distinct
stimuli preceded the onset of shock stimulation. In
one experiment, first a door of the shock
compartment was raised, six seconds later lights
flashed, then a buzzer was sounded which was, in
turn, followed by shock. During extinction trials

animals quickly exited from the threatening
compartment at the appearance of the first signal
of danger, thus preventing reexposure to the
remaining aversive stimuli. Eventually avoidance
responses to the initial cue were extinguished but

the resultant contact with the second feared cue,
which retained aversiveness transmitted by the
primary experiences, temporarily reinstated the
arousal potential of the first stimulus so that it
regained its capacity to maintain avoidance

behavior for some time before it was permanently
neutralized. A similar, though progressively

shorter, reacquisition process occurred with each
stimulus in the sequence, resulting in a
phenomenal amount of avoidance responding. One

animal, for example, performed 921 avoidance
responses to the most remote stimulus, 75 to the
second fearsome cue, but only 4 to the stimulus
directly associated with the painful experiences.
The above paradigm is well suited for
demonstrating sequential conditioning and

extinction of aversive stimuli, but it is unclear
what relevance, if any, the concept of repression
has to the phenomenon. The term repression is

usually employed to denote thought inhibition.
The arousal potential of an aversive stimulus can
be preserved from extinction by avoidance of
preceding cues regardless of whether or not the
protected events are symbolically represented.

Furthermore, the fact that one stops thinking
about fear-provoking situations does not prevent
him from being repeatedly exposed to them.
Implosive therapy is based on the premise that
extinction of anxiety can be most effectively
achieved by repeated elicitation of intense
emotional responses without the occurrence of
physically injurious consequences. Mainly for

reasons of ease, the emotional responses are
activated symbolically. The therapist vividly
describes the most revolting and terrifying
experiences conceivable, and clients are urged to

imagine themselves actively engaged in these
shocking activities. A compulsive hand washer
who is obsessed about dirt, for example, is asked
to visualize himself reaching into a wastebasket
and then withdrawing his hand, which is depicted

as dripping with a sickening mixture of mucus,
saliva, vomit, and feces. If the dirt phobia is

believed to arise from anxiety over anal functions,
the client is further instructed to imagine himself

residing in a septic tank where he eats his meals,
entertains his friends, and mushes around in this

soggy abode. Stampfl reasons “that he who has
lived in a septic tank need not fear the dirt found
in a wastebasket.” This is obviously not a method
for fastidious therapists. The distressing scenes

are presented again and again with appropriate
embellishments until they cease to evoke

emotional reactions. This procedure is repeated
with other variations on major sources of
disturbances. In order to accelerate the process of
extinction, clients are also instructed to recreate

disturbing scenes imaginally on their own
between treatment sessions.
Relatively little time is spent in ferreting out
the crucial sources of anxiety in any given case.

This is due, in part, to the assumption that
extinction of emotional responsiveness to
extremely fearsome situation will generalize
broadly to less frightening ones. A second reason
is that the anxiety elicitors assumed to be
repressed are routinely selected from a limited set
of categories relating to aggression, sex, rejection,
oral an anal functions, bodily injury, punishment,
loss of impulse control, and guilt.
Extinction is initially carried out with
environmental stimuli that are evident elicitors of
avoidance behavior. After these “symptom-
contingent” cues, which are believed to be the
least anxiety provoking, have been neutralized,

clients are repeatedly presented with the
hypothesized repressed events in harrowing
forms. It seems exceedingly improbable from the
case material cited that the heterogeneous cues

selected for extinction could have occurred
sequentially in traumatic conditioning. There is
also some ambiguity in the implementation of

implosive procedures because no explicit criteria
are presented for determining when treatment
should be confined to the evident determinants of
avoidance behavior or extended to hypothetical
sources of anxiety. Clients may therefore be
needlessly subjected to aversive stimulation while
therapists are neutralizing hypothesized

determinants of questionable relevance. The loose
relationship between conceptual rationale and
practice is further shown in experimental
evaluations of implosive therapy where
supposedly dynamically significant contents are

never pursued.
Results of animal experimentation (Polin,
1959; Poppen, 1968) and a few clinical
applications (Malleson, 1959) indicate that

avoidance behavior can be extinguished by
prolonged or massive exposure to subjectively
threatening stimuli. Preliminary studies (Hogan,

1966; Levis & Carrera, 1967) demonstrating that
implosive therapy produced greater reduction in
deviant responses on the MMPI test than
conventional treatment were somewhat
unconvincing because of the weak criterion
measure employed. Subsequent laboratory
investigations present evidence, based on

objective measures of behavioral change, that this
method can achieve extinction of avoidance
behavior. In one experiment (Kirchner & Hogan,
1966) coeds who feared rats were either assigned
to a control condition in which they were

instructed to imagine pleasant scenes while
listening to music, or they received group
implosive therapy. To minimize possible social
influences, subjects in the latter condition listened
through earphones in a language laboratory to a

one-hour tape-recording that described, among
other frightening scenes, rats biting, ripping flesh

and attacking a person en masse. A test for
avoidance behavior disclosed that 62 percent of
the subjects in the implosive condition were able

to pick up a white rat, while 26 percent of the
controls performed the same behavior. Essentially
similar outcomes were obtained in a second
experiment (Hogan & Kirchner, 1967) on the basis

of a single session of individual implosive
treatment. Sixty-seven percent of the treated

subjects, and 9 percent of the controls, could pick
up a rat in a subsequent behavioral test.
The efficacy of this method was further

evaluated in a comparative study (Hogan &
Kirchner, 1968) with coeds who feared snakes.
One group participated in a 45-minute implosive
session in which they were asked to imagine slimy
snakes crawling over them, biting them

relentlessly, and finally wrapping tightly around
their necks, slowly strangling them. A second
group of subjects, assigned to a verbal therapy
condition, discussed their interpersonal
relationships, their prior experiences with snakes,
and were reassured that snakes are harmless. The
third group of subjects read material concerning
the myths and habits of snakes. Whereas the
percentage of subjects able to pick up a snake after

treatment did not differ significantly in the
implosive (70 percent) and the verbal therapy (40
percent) groups, these subjects were considerably
braver than those receiving bibliotherapy (10
percent). Moreover, the implosive procedure was

successful with 67 percent of the coeds who had
previously failed to pick up the snake after
completing verbal therapy or bibliotherapy. The
fact that a brief verbal discussion produced
criterion performances in 40 percent of the cases

suggests that either the avoidance behavior of
many of the subjects was relatively weak to begin

with or the test was not sufficiently difficult. In
future assessments of implosive therapy, it would
therefore be of value to require more fear-

provoking performances toward the phobic
objects, and to test for the generality and stability
of behavioral changes brought about by this
method.
Contrary to results of the foregoing studies,
Mealiea (1967) reports findings based on a well-

designed experiment that cast doubt on the
efficacy of the implosive method. Snake-phobic
subjects were administered either taped

desensitization, implosive therapy that evoked
extremely anxiety-arousing imagery, a modified
desensitization procedure in which relaxation was
paired with scenes taken from the implosive
treatment, a pseudotherapy combining relaxation

with pleasant imagery, or no treatment at all. The
snake-approach behavior performed by the
different groups of subjects prior to treatment,
immediately after treatment, and a month later is
summarized graphically in Figure 6-12. Graduated
desensitization proved superior to the other
performed by subjects in each of five conditions before
treatment, immediately after treatment, and a month
later. Plotted from data of Mealiea, 1967.
conditions in reducing avoidance behavior toward
a snake and a second feared animal that served as
a measure of generalization. However, subjects

who received implosive therapy did not differ
from any of the control groups. In view of these
negative results, the implosive method should be
utilized with caution until it has been subjected to
further laboratory tests.
A distinction should be drawn between

flooding procedures in which conditioned aversive
stimuli are simply presented in intense forms from
implosive procedures that provide vivid accounts

of hazardous consequences that the feared objects
can produce. There is considerable difference
between exposing people repeatedly to a fearsome

collection of rodents without any adverse effects
and depicting them eating human flesh. Some of
the portrayed consequences may never have
occurred to phobic subjects and could establish, at
least temporarily, a new basis of fearful self-
arousal. It is interesting to note in this regard that
the types of shocking and nauseating

consequences employed in implosive therapy to
extinguish avoidance behavior are also being used
to create strong aversions toward attractive or
addictive objects. It is entirely conceivable that
aversive stimuli may have markedly different

effects depending upon subjects’ fear level and the
valence of the objects with which they are paired.
There is also reason to expect that conditioned
aversive stimuli might initially increase negative
responsiveness but with repeated presentations

would eventually lose their emotion-arousing
capacity. To gain a better understanding of both

fear-extinction and aversion-conditioning
processes would require detailed analysis of
changes in the magnitude and quality of emotional

arousal over a series of trials in which aversive
experiences are elicited in conjunction with
attractive, neutral, and fear-provoking objects.
EXTINCTION BASED ON MASSED EVOCATION OF

RESPONSES
Repeated nonreinforced evocation of effortful

behavior creates aversive consequences in the
form of pain and fatigue, which inhibit responses

that will produce the discomfort. Successive
extinction operations of this type typically result
in progressive decline, and eventually complete
elimination, of the behavior. The method of

massed performance has been applied on a limited
basis to incapacitating tics and other spasmodic
movements that have proved refractory to a host
of medicative and psychological ministrations.
These patterns of muscular contraction are

usually conceptualized as conditioned avoidance
responses that were originally evoked in highly
traumatic situations (Yates, 1958). It is assumed
that tics probably occurred by chance in close
temporal proximity to the termination or
reduction of intense aversive stimulation and,

through the accidental correlation, they acquired
emotion-reducing qualities. Muscular contraction

has some inherent pain- and tension-reducing
value which would further enhance the self-
reinforcing character of such responses. The fact
that the rate of tics increases under conditions of

stress and excitement is considered as suggestive
evidence for the “tension-reduction”
interpretation.
It is impossible to unravel from retrospective
accounts the actual contingencies under which tics

are established; there is nevertheless ample
observational evidence that animals in aversive
conditioning experiments frequently acquire tic-
like responses that are highly resistant to
extinction long after noxious stimulation has been
discontinued. Considering, however, that
persistent stereotyped movements have also been

established through adventitious reinforcement in
experiments employing reward contingencies
(Skinner, 1948), it is probable that tics, in fact,
originate under a variety of learning conditions.
Avoidance responses that are automatically

self-reinforcing upon occurrence can be
eliminated in several different ways. In a stimulus-
oriented approach one would neutralize the
conditioned aversive stimuli controlling the

occurrence of avoidance behavior. On the other
hand, in a response-oriented approach, efforts are
made to nullify the rewarding value of avoidance

responses either by externally administered
negative consequences (Barrett, 1962;
Goldiamond, 1965), or by massed evocation that
results in response-produced aversive effects.
Desensitization procedures have not been utilized
to eliminate tics, but reactive extinction methods
have.
One factor that influences the rate of extinction
based on reactive procedures is the frequency
with which the responses are performed. In
general, there tends to be little diminution in the
strength of avoidance behavior when extinction

trials are widely distributed, whereas under
conditions of massed evocation aversive effects
presumably build up rapidly, and, consequently,
extinction is accelerated (Calvin et al., 1956;

Edmonson & Amsel, 1954). In accord with
laboratory findings, Yates (1958) reports that
massed evocation of responses in which a woman

voluntarily performed multiple tics for as long as
one hour, followed by prolonged rest, was the
most effective procedure for extinguishing the
motor responses. After the experimental program,
the client carried out the exercises at home (Jones,
1960) and recorded the number of tics that she
was able to produce intentionally during each one-

minute period. The results show a progressive
decline in the rate of voluntarily emitted
responses with successive extinction periods, as
well as a significant reduction in her involuntary
tics in everyday situations.
Clark (1966) treated three adults, all of whom

manifested explosive repetition of obscenities or
other expletives along with various motor tics.
Because of this peculiar behavior one of the males

was unable to appear in public or retain any
friends, while the second was in danger of losing
his job because of his incessant loud barking. A

massed-practice regimen was employed in which
the clients repeated the verbal tics as often as
possible until they could no longer emit them. One
of the three cases, a female in whom the motor tic
was more prominent, discontinued treatment
after a reluctant start; in the other two, the
muscular spasms disappeared spontaneously as

the verbal tics were successfully extinguished. The
clients remained free of tics as corroborated by
recorded follow-up interviews.
Similar positive results are reported by
Costello (1963) in the treatment of a 12-year-old

boy who displayed a persistent eye-blink tic. The
boy was initially instructed to produce the tics
deliberately in front of a mirror for five-minute
periods several times a day. The duration of the

massed practice was then gradually increased to
one-hour sessions. Although no quantitative data
are presented, it is reported that the frequency of

tics declined markedly and remained at a low level
when evaluated again one year later.
On the assumption that extinction occurs faster
under conditions of low rather than high drive
states, Walton (1961; 1964) utilized massed
performance combined with medication to reduce

emotional arousal in eliminating severe tics in two
boys. One of the tiqueurs exhibited violent arm
and leg movements that made it exceedingly
difficult for him to eat his meals, and unsettled

anyone in his immediate vicinity; the second
suffered for eleven years from vigorous head-
shaking and explosive nasal expiration. In both
cases the tics were durably eliminated by reactive
extinction. The contribution of the medication to

these changes cannot be assessed, however, in the
absence of cases treated without the

pharmacological supplement. Even though
responses may be extinguished more rapidly in a

tranquilized than in a nondrug state, the clinical
use of drugs may be contraindicated. The reason

for this, as will be shown later, is that extinction
effects achieved in the drug condition often fail to
transfer to the nondrug state.
Contrary to the preceding favorable outcomes,
Feldman & Werry (1966) were unable to achieve
any decline in head-jerking and eye-blinking tics

in an adolescent boy through massed practice. The
authors ascribe the failure to the presence of high
anxiety. It is entirely possible that the conflicting

findings are partly due to considerable differences
in the way in which extinction was conducted.
Previous investigators have utilized prolonged
periods of massed performances lasting several
hours, whereas in the study by Feldman & Werry

the boy performed the head tic for only five-
minute sessions because of dizziness, while the
eye-blink served as an unpracticed control.
However, this interpretation may not fully account
for the discrepancies. Data published by Abi Rafi
(1962) show that even the same procedure of
extended response evocation may produce
differential results. In one case, a man who lost
considerable sleep because of interference from

pronounced facial grimacings benefited greatly
from this method. The second case was an elderly
woman who was forced to relinquish many
activities she enjoyed because of a foot-tapping tic
that was highly disturbing to others. Prolonged

massed evocation failed to produce any
appreciable response decrement. Her obstinate tic
was subsequently modified successfully by self-
control training in which incipient foot
movements activated a buzzer to prevent further

responding. One might wonder, because of the
client’s favorable response to a simple alternative

treatment, whether greater progress would be
achieved in behavioral modification if failures
were less frequently attributed to inferred anxiety

states.
The studies reported above indicate that a
program of prolonged massed practice can
extinguish extremely persistent tics, but the

specific factors responsible for the observed
changes and the most efficacious procedures

cannot be determined from these data. The
interpretation of results is especially complicated
when medicinals or other methods are utilized in
conjunction with repeated performance. Even the

recommended optimum conditions of massed
practice must be accepted with reservation, since
the supporting experimental data (Yates, 1958)

are based on a single case in which both duration
and intensity of responding were continuously

varied; consequently, it is by no means clear
whether changes in extinction rate represent the
cumulative effects of prior nonreinforced

performance or variations in duration of evocation
sessions. If response-produced aversive effects
play an influential role in the efficacy of massed-
practice methods, it would be of considerable
interest to investigate extinction rates as a
function of both length of repeated performance

and effortfulness of response. According to the
inhibition hypothesis, increased effortfulness,
which could be varied in terms of the vigor with

which responses are performed, should result in
faster extinction.
The elimination of persistent behavior under
conditions of massed nonreinforced evocation is
generally attributed to the development of
conditioned inhibitory responses arising from

reactive fatigue states. As noted earlier, however,
interfering responses can originate from a variety
of sources; therefore, extinction outcomes may
reflect several different processes. Moreover,
some reduction in tics is probably attributable to
increased efforts at self-control (Barrett, 1962).
EXTINCTION IN INTERVIEW APPROACHES
Interview treatment approaches generally

consider permissiveness to be an important
condition for therapeutic change. It is expected
that when a client repeatedly expresses thoughts
and feelings which, as a result of a previous

history of punishment, elicit anxiety or guilt, but
which the therapist does not disapprove or

criticize, the client’s inappropriate emotional
responses will be gradually extinguished through
nonreinforcement. It is further assumed that

extinction effects will generalize to thoughts
concerning related topics which also may be
inhibited, and to the corresponding verbal and
physical behavior as well (Dollard & Miller, 1950).
Some suggestive evidence for the relationship

between permissiveness and extinction of
conditioned emotionality associated with verbal

behavior is provided in two studies reported by
Dittes (1957a, b). In one investigation (1957b),
involving analyses of specific client-therapist
interaction sequences, Dittes found that

permissive responses on the part of the therapist
toward sexual statements were followed by
decreases in the client’s autonomic responses,
resistive or avoidant remarks, and interruptions in

speech. A sequential analysis of 30
psychotherapeutic interviews with the same client
revealed that initially sexual statements were
accompanied by strong emotional arousal, but
with repeated evocation anxiety responses to

verbal sexual expressions were gradually
extinguished.
There is every reason to expect that if
psychotherapists respond favorably toward their

clients’ verbal expressions of thoughts and feelings
that were previously punished, the attendant

emotional responses will eventually extinguish.
The critical issues, therefore, that remain to be
answered are concerned with the degree of
generalization of extinction effects to nonverbal

behavior and to other persons. These questions
are particularly critical since it is not uncommon
for clients to express thoughts and feelings freely
within the safety of the interview setting, but to
remain inhibited and fearful in their everyday

interactions. If a satisfactory degree of transfer can
be demonstrated, which is doubtful in view of the
generally discouraging outcome data, controlled
studies would be needed to assess the relative
efficacy of verbal extinction procedures and

approaches employing graded performance tasks
in eliminating inappropriate affective and

avoidant behavior.
“Abreaction” and Extinction. Changes effected
by abreactive procedures, in which clients are
induced by hypnosis, intravenous barbiturates, or
inhalational anesthetics to revivify past traumatic
events, may likewise be explained in terms of an
extinction process. During symbolic reinstatement

of traumatic episodes, individuals typically
express the intense emotional responses
experienced at the time of the fear-provoking
incidents. The emotional expression is also
frequently accompanied by a reduction in

avoidance behavior that was originally elicited in
the traumatic situation and subsequently
generalized to other similar stressful situations.
The process of traumatic aversive

conditioning, generalization, and extinction is
illustrated in a successful ether abreaction of an
anxiety disorder apparently originating in a

gruesome battle experience 18 years earlier (Little
& James, 1964). The client had shot two young
soldiers in the back with a concealed weapon
while being taken captive near enemy lines. After
disarming two other soldiers, he burst through a
doorway in a small farmhouse to find 12 enemy
troops in the process of wakening. He stood guard

over them for 10 strained hours, finally shot their
sergeant, who kept urging the soldiers to rush
their captor, and brought in the prisoners when
night fell. The next day he developed a temporary
paraplegia when a rifle grenade exploded nearby.

Following his army discharge, the client continued
to experience chronic anxiety and guilt, avoided all
military functions and, for 18 years, was unable to
open and walk through a door if he could hear
voices on the other side.
The client received five abreaction sessions in
which he recreated, in action and with violent

emotion, the traumatic military episode. The
authors report progressive reduction in anxiety
and guilt with each session. Moreover, the door
phobia was eliminated, and according to the 12-
month follow-up report, the client extended his
range of social interactions and continued to
experience no difficulties in walking through

doors anywhere.
The hypnotic and pharmacological abreactive
techniques currently in use are derived
historically from the early work of Freud & Breuer
(1940), who employed hypnotic abreaction in the

modification of functional sensory-motor
disorders such as anesthesias, neuralgias,
paralyses, visual disorders, epileptoid convulsions,
and other forms of defensive reactions. This

method, however was abandoned by Freud in
favor of free association and interpretive
procedures because affective expression appeared

to produce only a temporary elimination of the
associated behavioral disorders.
As shown earlier, in order to achieve
permanent or complete extinction of emotional
responses it is necessary to present the fear-
provoking stimuli repeatedly without

reinforcement. During the course of extinction,
emotional responses are likely to reappear at
some strength, although the amount of recovery
diminishes with successive extinctions. It is,

therefore, not surprising that a few extinction
sessions, in which a client was symbolically
reexposed to highly traumatic stimulus events,
failed to reduce emotional responding to a stable
zero level. Had Freud extended the extinction

series it is probable that his original “cathartic”
procedure might have proved more efficacious

than the protracted interpretive form of treatment
that he subsequently adopted.
An interesting laboratory demonstration of the
progressive decline of emotional behavior with

repeated hypnotic abreactions of a traumatic
episode is furnished by Lifshitz & Blair (1960).
The subject, a 23-year-old female, revivified under
hypnotic age regression a near drowning that she

had experienced at 10 years of age. “She was at the
beach with her father and waded too far out into
the water, was knocked down by a succession of
waves, inhaled and swallowed water, and was in
fear of drowning when rescued by her father [p.

248].”
Under hypnosis the subject spontaneously
recalled this specific episode seven times, during
each of which the following autonomic reactions

were continuously recorded: duration of
abreaction as revealed principally by facial
expressions; basal heart rate immediately prior to

the description of the unpleasant experience;
maximum respiratory rate during abreaction; GSR
reactivity; frontalis muscle activity; cheek
temperature; and total body movement. As shown
in Table 6-1, repeated nonreinforced evocation of
emotional responses to the past traumatic event
produced a diminution of emotional

responsiveness. The fact that the subject
continued to exhibit marked physiological
responses to other unrelated traumatic incidents
indicates that the former changes reflect a genuine
extinction effect, rather than a general adaptation

process.
Table 6-1 Analysis of Poly graphic Recordings (Lifshitz &

Blair, 1960)
Duration of 2' 2' 1'10" 1' 1'
abreaction
Maximum 36/min 39/min 18/min 24/min 24/min
respiratory
rate during
abreaction
No. of GSR 16 16 6 9 1
waves
during
abreaction
Basal heart 81/min 72/min 90/min 87/min 87/min
rate
(20 sec.
average)
Maximum 100/min 120/min 95/min 100/min 100/min
heart rate
during
abreaction
Frontalis Moderate Moderate Slight Very Very
muscle slight slight
activity
during
abreaction
Cheek Fall Fall (toe No No No
temperature temp) change change change
during
abreaction
Bed Moderate Marked None None None
movement
during
abreaction
Abreactive procedures are probably best

suited for producing rapid and stable extinction of
emotional responses developed in traumatic

conditioning situations, provided that the threats
are no longer present. Clinical case data (Shrovon

& Sargant, 1947) seem to bear this out, although in
instances where abreaction is used in conjunction

with drastic environmental changes as well as
other treatment procedures, it is impossible to

isolate the factors responsible for the reported
behavioral modifications.
Traditional accounts of the abreactive process
generally ascribe beneficial outcomes to the

“discharge of pent-up emotions” and the “working
through” of spontaneously recalled material. From
a learning point of view, the critical therapeutic

factor is the repeated elicitation of emotional
responses without reinforcement rather than the
energy discharges or the historical insights. For

this reason, it is not unexpected that persons who
express strong hostile, dependent, or depressive
feelings while under the effects of barbiturates or
anesthetics fail to attain enduring benefits
(Hordern, 1952) when the reinforcement

contingencies generating and maintaining these
unpleasant emotional states remain unaltered.

Discussions of the efficacy of abreactive
procedures are usually confined to the influence of

clients’ personality characteristics and emotion
induction procedures, hypnotic, barbiturate, or
anesthetic. However, both in theorizing and in
clinical practice, virtually no attention has been

devoted to the variables that determine the rate of
extinction.
Summary
In the process of extinction, when the
reinforcing consequences for a particular response
pattern are consistently discontinued, the
recurrence of the behavior is diminished and

eventually ceases. Since the decremental effects of
nonreinforcement are controlled by many

variables, several different theoretical
interpretations of extinction have been proposed.
Contrary to the connotation of the term,
extinguished behavior is displaced rather than
permanently lost. In fact, nonreinforced behavior
is often abandoned without being performed
solely as a result of observing changes in
conditions of reinforcement, and it is easily
recovered by reinstating the original

reinforcement contingencies. Such rapid changes
in behavior suggest that extinction phenomena
primarily reflect the operation of cognitively
mediated inhibitory sets rather than the alteration

of specific stimulus-response associations. That is,
when an organism discerns that the usual
response consequences have been discontinued,
the behavior is discarded and supplanted by
alternative response patterns. However, in the

case of severe avoidance behavior, cognitive
control may be relatively weak and the absence of

aversive consequences must be repeatedly
experienced, either directly or vicariously, before

the behavior is abandoned.
Under conditions where no reinforcement is

externally administered during the extinction
phase, it is assumed that continued performance
of nonrewarded behavior generates aversive
effects and its cessation provides negative

reinforcement for competing response patterns.
This may take the form of fatigue reduction,
elimination of aversive emotional effects resulting
from the omission of expected rewards, or fear
reduction resulting from the absence of aversive

consequences.
Behavior that is maintained by positive
reinforcement is extinguished by discontinuing its
rewarding consequences. Omission of expected

rewards for given performances can generate
aversive emotional effects that function
analogously to punishment, as shown by evidence

that stimuli previously associated with nonreward
acquire arousal capacity, their presence reduces
responsiveness, and escape from cues signifying
nonreward can reinforce new performances. As
behavior is reduced through nonreward,
alternative modes of response eventually emerge.
The degree of behavioral variability and the

characteristics of the new actions occurring during
the course of extinction depend upon the options
that individuals have previously learned for
securing reinforcement. Reliance on extinction
alone, therefore, does not guarantee that desired

response patterns will appear unless they happen
to be strongly developed. Behavioral changes can
be hastened and effectively controlled, however,
by combining extinction of undesired responses
with stimulus control procedures and with

modeling and positive reinforcement of competing
response tendencies.
Extinction of avoidance behavior is achieved
by repeated exposure to subjectively threatening
stimuli under conditions designed to ensure that
neither the avoidance responses nor the
anticipated adverse consequences occur. The
major obstacle to eliminating defensive behavior
arises because successful avoidance of events that

are no longer dangerous preserves their
aversiveness and forestallment of anticipated
consequences reinforces the defensive activities.
Attempts have been made to facilitate extinction of
avoidance behavior by blocking its occurrence in

the presence of fear-arousing stimuli. Such forced
exposure may simply produce other types of
avoidance responses without altering the arousal
potential of the feared situations.
Inappropriate defensive behavior is most

frequently eliminated by an extinction procedure
involving gradual stimulus change. This is

achieved by reexposing individuals initially to
aversive stimuli at weak intensities that do not
evoke avoidance responses, and then to gradually
increasing threats until the most fearsome
situations have been completely neutralized. If the
aversive stimuli are increased in sufficiently small
increments emotional behavior can be successfully

extinguished with minimal fear elicitation and
avoidance responding.
The third and most recently developed
extinction procedure involves prolonged and
massive exposure to intensely disturbing events

that are symbolically created. Preliminary findings
show that avoidance behavior can be eliminated in
this manner, but the full effects of this method
have not as yet been adequately evaluated.
REFERENCES
Abi Rafi, A. Learning theory and the treatment of tics.

Journal of Psychosomatic Research, 1962, 6, 71-
76.
Agras, S., Leitenberg, H., & Barlow, D. H. Social

reinforcement in the modification of
agoraphobia. Archives of General Psychiatry,
1968, 19, 423-427.
Alexander, F. Psychoanalysis and psychotherapy. New
York: Norton, 1956.
Allen, K. E., Hart, B., Buell, J. S., Harris, F. R., & Wolf, M.
M. Effects of social reinforcement on isolate
behavior of a nursery school child. Child
Development, 1964, 35, 511-518.
Amsel, A. Frustrative nonreward in partial

reinforcement and discrimination learning: Some
recent history and a theoretical extension.
Amsel, A., & Surridge, C. T. The influence of magnitude

of reward on the aversive properties of
anticipatory frustration. Canadian Journal of
Psychology, 1964, 18, 321-327.
Ayllon, T., & Haughton, E. Control of the behavior of

schizophrenic patients by food. Journal of the
352.
Ayllon, T., & Haughton, E. Modification of

symptomatic verbal behaviour of mental
patients. Behaviour Research and Therapy, 1964,
2, 87-97.

Ax, A. F. The physiological differentiation between

fear and anger in humans. Psychosomatic
Medicine, 1953, 15, 433-442.

1962. Lincoln: Univer. of Nebraska Press, 1962.
pp. 211-269.

Bandura, A. Modelling approaches to the modification

of phobic disorders. In Ciba Foundation
Symposium. The role of learning in psychotherapy.
London: Churchill, 1968. pp. 201-217.

attitudinal changes, Stanford Univer., 1969, 13,
173-199.


62.

Barrett, B. H. Reduction in the rate of multiple tics by

free operant conditioning methods. Journal of
Nervous and Mental Disease, 1962, 135, 187-195.
Bass, M. J., & Hull, C. L. The irradiation of a tactile

conditioned reflex in man. Journal of Comparative
Psychology, 1934, 17, 47-65.
Bateson, G. The Naven. Stanford: Stanford Univer.

Press, 1936.
Benline, T. A., & Simmel, E. C. Effects of blocking of the

avoidance response on the elimination of the
conditioned fear response. Psychonomic Science,
1967, 8, 357-358.

466.
Berkowitz, L. Aggression: A social psychological
analysis. New York: McGraw Hill, 1962.
Berkowitz, L. Aggressive cues in aggressive behavior

and hostility catharsis. Psychological Bulletin,
1964, 71, 104-122.
Berkowitz, L. Control of aggression. In B. M. Caldwell

& H. Ricciuti (Eds.), Review of child development
research, Vol. .3. 1969.

Bridger, W. H., & Mandel, I. J. Abolition of the PRE by

instructions in GSR conditioning. Journal of
Bridger, W. H., & Mandel, I. J. A comparison of GSR

fear responses produced by threat and electric
shock. Journal of Psychiatric Research, 1964, 2,
31-40.
Brown, J. S., & Bass, B. The acquisition and extinction

of an instrumental response under constant and
variable stimulus conditions. Journal of
51, 499-504.
Brown, P., & Elliott, R. Control of aggression in a
nursery school class. Journal of Experimental

4, 157-167.
Buss, A. H. Instrumentality of aggression, feedback,

and frustration as determinants of physical
aggression. Journal of Personality and Social
Psychology, 1966, 3, 153-162.
Calvin, A. D., Clifford, L. T., Clifford, B., Bolden, L., &

Harvery, J. Experimental validation of
conditioned inhibition. Psychological Reports,
1956, 2, 21-56.
Capaldi, E. J. The effect of different amounts of

training on the resistance to extinction of
different patterns of partially reinforced
responses. Journal of Comparative and
Carlson, N. J., & Black, A. H. Traumatic avoidance

learning: Note on the effect of response
prevention during extinction. Psychological
Reports, 1959, 5, 409-412.
in Child Development, 1942 (1, Serial No. 31).
Clark, D. F. Behaviour therapy of Gilles de la

Tourette’s syndrome. British Journal of
Psychiatry, 1966, 112, 771-778.
Cook, S. W., & Harris, R. E. The verbal conditioning of

the galvanic skin reflex. Journal of Experimental
Psychology, 1937, 21, 202-210.
Costello, C. G. The essentials of behaviour therapy.

Canadian Psychiatric Association Journal, 1963, 8,
162-166.
Davitz, J. R. The effects of previous training on

postfrustration behavior. Journal of Abnormal and
Deese, J. The extinction of a discrimination without

performance of the choice response. Journal of
44, 362-366.
Delgado, J. M. Social rank and radio-stimulated

aggression in monkeys. Journal of Nervous and
Mental Disease, 1967, 144, 383-390.
Denny, M. R., & Ratner, S. C. Distal cues and latent

extinction. Psychological Record, 1959, 9, 33-35.
Denny, M. R., & Weisman, R. G. Avoidance behavior as
a function of length of nonshock confinement.
Psychology, 1964, 58, 252-257.
Di Cara, L. V., & Miller, N. E. Instrumental learning of

vasomotor responses by rats: Learning to
respond differentially in the two ears. Science,
1968, 159, 1485-1486.
Dittes, J. E. Extinction during psychotherapy of GSR

accompanying “embarrassing” statements.
54, 187-191. (a)
Dittes, J. E. Galvanic skin response as a measure of

patient’s reaction to therapist’s permissiveness.
55,295-303. (b)
Dollard, J., & Miller, N. E. Personality and

psychotherapy. New York: McGraw-Hill, 1950.
Dollard, J., Miller, N. E., Doob, L. W., Mowrer, O. H., &

Sears, R. R. Frustration and aggression. New
Haven: Yale University Press, 1939.
Dyal, J. A. Latent extinction as a function of number of

training trials. Psychological Reports, 1963, 13,
407^14.
Edmonson, R. W., & Amsel, A. The effects of massing
and distribution of extinction trials on the
persistence of a fear-motivated instrumental
response. Journal of Comparative and
Elam, C. R., Tyler, D. W., & Bitterman, M. E. A further

study of secondary reinforcement and the
discrimination hypothesis. Journal of
47, 381-384.
Epstein, R. Aggression toward outgroups as a function

of authoritarianism and imitation of aggressive
models. Journal of Personality and Social
Psychology, 1966, 3, 574-579.
Estes, W. K. The statistical approach to learning

theory. In S. Koch (Ed.), Psychology: A study of a
science. Vol. 2. New York: McGraw-Hill, 1959. pp.
380-491.
Etzel, B. C., & Gewirtz, J. L. Experimental modification

of caretaker-maintained high-rate operant crying
in a 6- and a 20-week-old infant (infans
tyrannotearus): Extinction of crying with
reinforcement of eye contact and smiling. Journal
of Experimental Child Psychology, 1967, 5, 303-
317.
Feldman, R. B., & Werry, J. S. An unsuccessful attempt
to treat a tiqueur by massed practice. Behaviour

Crofts, 1957.
Feshbach, S. The catharsis hypothesis and some

consequences of interaction with aggressive and
neutral play objects. Journal of Personality, 1956,
24, 449-462.
Feshbach, S. The function of aggression and the

regulation of aggressive drive. Psychological
Review, 1964, 71, 257-272.
Feshbach, S., Stiles, W. B., & Bitter, E. The reinforcing

effect of witnessing aggression. Journal of
Experimental Research in Personality, 1967, 2,
133-139.
Foster, F. M. Conditioning of cerebral dysrhythmia

induced by pattern presentation and eye closure.
Conditional Reflex, 1967, 2, 236-244.
Foster, F. M., & Campos, G. B. Conditioning factors in

stroboscopic-induced seizures, Epilepsia, 1964, 5,
156-165.
Freeman, H. L., & Kendrick, D. C. A case of cat phobia:
Treatment by a method derived from
experimental psychology. British Medical Journal,
1960, 2, 497-502.
Freud, S., & Breuer, J. On the psychical mechanisms of

hysterical phenomena. In E. Jones (Ed.), Collected
papers. Vol. 1. London: Hogarth, 1940. pp. 24-41.
Garvey, W. P., & Hegrenes, J. R. Desensitization

techniques in the treatment of school phobia.
American Journal of Orthopsychiatry, 1966, 36,
147-152.
Geen, R. G. Effects of frustration, attack, and prior

training in aggressiveness upon aggressive
behavior. Journal of Personality and Social
Psychology, 1968, 9, 316-321.
Gleitman, H., Nachmias, J., & Neisser, U. The S-R

reinforcement theory of extinction. Psychological
Review, 1954, 61, 23-33.
Goldiamond, I. Fluent and nonfluent speech

(stuttering): Analysis and operant techniques for
control. In L. Krasner & L. P. Ullmann (Eds.),
Holt, Rinehart & Winston, pp. 106-156.
Grings, W. W., & Lockhart, R. A. Effects of “anxiety-

lessening” instructions and differential set
development on the extinction of GSR. Journal of
Groot, H. Extinction of psychotic letter writing. Paper

read at the California State Psychological
Association, San Francisco, 1966.
Grossberg, J. M. Successful behavior therapy in a case

of speech phobia (“stage fright”). Journal of

Harper, 1935.
Hart, B. M„ Allen, K. E., Buell, J. S., Harris, F. R., & Wolf,

M. M. Effects of social reinforcement on operant
crying. Journal of Experimental Child Psychology,
1964, 1, 145-153.
Hartmann, D. Influence of symbolically modeled

instrumental aggression and pain cues on
aggressive behavior. Journal of Personality and
Social Psychology, 1969,11, 280-288.
Haslam, M. T. The treatment of psychogenic

dyspareunia by reciprocal inhibition. British
Journal of Psychiatry, 1965, 111, 280-282.
Hawkins, R. P., Peterson, R. F., Schweid, E., & Bijou, S.

W. Behavior therapy in the home: Amelioration
of problem parent-child relations with the parent
in a therapeutic role. Journal of Experimental
Child Psychology, 4, 99-107.
Herzberg, A. Short treatment of neuroses by

graduated tasks. British Journal of Medical
Psychology, 1941, 19, 36-51.
Herzberg, A. Active psychotherapy. New York: Grune

and Stratton, 1945. Hoffeld, D. R. Primary
stimulus generalization and secondary extinction
as a function of strength of conditioning. Journal
1962, 55, 27-31.
Hogan, R. A. Implosive therapy in the short-term

treatment of psychotics. Psychotherapy: Theory,
Research and Practice, 1966, 3, 25-32.
Hogan, R. A., & Kirchner, J. H. Preliminary report of

the extinction of learned fears via shortterm
implosive therapy. Journal of Abnormal
Psychology, 72, 106-109.
Hogan, R. A., & Kirchner, J. H. Implosive, eclectic

verbal and bibliotherapy in the treatment of fears
of snakes. Behaviour Research and Therapy,
6,167-171.
Hordern, A. The response of the neurotic personality

to abreaction. Journal of Mental Science, 1952, 98,
630-639.
Hovland, C. I. The generalization of conditioned

responses. I. The sensory generalization of
conditioned responses with varying frequencies
of tone. Journal of General Psychology, 1937, 17,
125-148.
Hull, C. L. Principles of behavior. New York: Appleton-

Century-Crofts, 1943. Jegard, S., & Waiters, R. H. A
study of some determinants of aggression in
young children. Child Development, 1960, 31,
739-747.
Johnston, M. S., Kelley, C. S., Buell, J. S., Harris, F. R., &

Wolf, M. M. Effects of positive social
reinforcement on isolate behavior of a nursery
school child. Unpublished manuscript, University
of Washington, 1963.
Jones, H. G. The application of conditioning and

learning techniques to the treatment of a
psychiatric patient. Journal of Abnormal and
Social Psychology, 1956, 52,414-419.
Jones, H. G. Continuation of Yates’ treatment of a

tiqueur. In H. J. Eysenck, (Ed.), Behaviour therapy
and the neuroses. New York: Pergamon, 1960. pp.
250-258.
Kahn, M. The physiology of catharsis. Journal of
286.
Kamin, L. J. The effects of termination of the CS and

avoidance of the US on avoidance learning.
Psychology, 1956, 49, 420-424.
Kamin, L. J. The effects of termination of the CS and

avoidance of the US on avoidance learning: An
extension. Canadian Journal of Psychology, 1957,
11,48-56.
Kamin, L. J., Brimer, C. J., & Black, A. H. Conditioned

suppression as a monitor of fear of the CS in the
course of avoidance training. Journal of
56, 497-501.
Kaufmann, H., & Feshbach, S. Displaced aggression

and its modification through exposure to
antiaggressive communications. Journal of
Abnormal and Social Psychology, 1963, 67, 79-83.
(a)
Kaufmann, H., & Feshbach, S. The influence of

antiaggressive communications upon the
response to provocation. Journal of Personality,
1963, 31, 428-444. (b)
Kennedy, W. A. School phobia: Rapid treatment of fifty
cases. Journal of Abnormal Psychology, 1965, 70,
285-289.
Kenny, D. T. An experimental test of the catharsis

theory of aggression. Unpublished doctoral
dissertation, University of Washington, 1952.

1961.
Kimble, G. A., & Kendall, J. W., Jr. A comparison of two

methods of producing experimental extinction.
90.
Kimble, G. A., Mann, L. I., & Dufort, R. H. Classical and

instrumental eyelid conditioning. Journal of
Kirchner, J. H., & Hogan, R. A. The therapist variable in

the implosion of phobias. Psychotherapy: Theory,
Research and Practice, 1966, 3, 102-104.
Koppman, J. W., & Grice, R. G. Goal-box and alley

similarity in latent extinction. Journal of
Kuhn, D. Z., Madsen, C. H., Jr., & Becker, W. C. Effects of

exposure to an aggressive model and
“frustration” on children’s aggressive behavior.
Lawrence, D. H., & Festinger, L. Deterrents and

reinforcement: The psychology of insufficient
reward. Stanford: Stanford University Press,
1962.
Leitenberg, H., Agras, W. S., Thomson, L. E., & Wright,

D. E. Feedback in behavior modification: An
experimental analysis in two phobic cases.
137.
Levis, D. J. Effects of serial CS presentation and other

characteristics of the CS on the conditioned
avoidance response. Psychological Reports, 1966,
18, 755-766.
Levis, D. J. Implosive therapy: Part II. The subhuman

analogue, the strategy, and the technique. In S. G.
Armitage (Ed.), Behavior modification techniques
in the treatment of emotional disorders. Battle
Creek, Mich.: Veterans Administration, 1967. pp.
22-37.
Levis, D. J., & Carrera, R. Effects of ten hours of

implosive therapy in the treatment of
outpatients: A preliminary report. Journal of
Lifshitz, K., & Blair, J. H., The polygraphic recording of
a repeated hypnotic abreaction with comments
on abreaction psychotherapy. Journal of Nervous
and Mental Disease, 1960, 130, 246-252.
Lindley, R. H., & Moyer, K. E. Effects of instructions on

the extinction of a conditioned finger-withdrawal
response. Journal of Experimental Psychology,
1961, 61, 82-88.
Little, J. C., & James, B. Abreaction of conditioned fear

reaction after eighteen years. Behaviour Research
and Therapy, 1964, 2, 59-63.
Longenecker, E. G., Krauskopf, J., & Bitterman, M. E.

Extinction following alternating and random
reinforcement. American Journal of Psychology,
1952, 65, 580-587.
Longstreth, L. E. Frustration and secondary

reinforcement concepts as applied to human
instrumental conditioning and extinction.
No. 619).
Mackintosh, I. The resistance to extinction of

responses acquired under irregular conditions of
learning. Journal of Comparative and Physiological
Psychology, 1955, 48, 363-370.
Madsen, C. H., Jr., Becker, W. C., Thomas, D. R., Koser,
L., & Plager, E. An analysis of the reinforcing
function of “sit down” commands. In R. K. Parker
(Ed.), Readings in educational psychology. Boston:
Allyn and Bacon, 1968. pp. 265-278.
Malleson, N. Panic and phobia. Lancet, 1959, 1, 225-

227.
Mallick, S. K., & McCandless, B. R. A study of catharsis

of aggression. Journal of Personality and Social
Psychology, 1966, 4, 591-596.
McClelland, D. C., & McGown, D. R. The effect of

variable food reinforcement on the strength of a
secondary reward. Journal of Comparative and
McCord, J., & McCord, W. The effects of parental role

model on criminality. Journal of Social Issues,
1958, 14, 66-75.
McNamara, H. J., & Paige, A. B. Resistance to extinction

as a function of presenting a discriminative
stimulus before and after the runway response.
McNamara, H. J., & Wike, E. L. The effects of irregular

learning conditions upon the rate and
permanence of learning. Journal of Comparative
Mealiea, W. L., Jr. The comparative effectiveness of

systematic desensitization and implosive therapy
in the elimination of snake phobia. Unpublished
doctoral dissertation, University of Missouri,
1967.
Meyer, V. The treatment of two phobic patients on the

basis of learning principles. Journal of Abnormal
Meyer, V. Modification of expectations in cases with

obsessional rituals. Behaviour Research and
Therapy, 1966, 4, 273-280.
Miller, N. E. Theory and experiment relating

psychoanalytic displacement to stimulus-
response generalization. Journal of Abnormal and
Miller, N. E. Learning of visceral and glandular

responses. Science, 1969, 163, 434-445.
Moltz, H. Latent extinction and the reduction of

secondary reward value. Journal of Experimental
Psychology, 1955, 49, 395-400.
Moltz, H., & Maddi, S. R. Reduction of secondary

reward value as a function of drive strength and
latent extinction. Journal of Experimental
Psychology, 1956, 52, 71-76.
Mowrer, O. H. Learning theory and personality

dynamics. New York: Ronald Press, 1950.
Mowrer, O. H. Learning theory and behavior.. New

York: Wiley, 1960.
Moyer, K. E., & Lindley, R. H. Supplementary report:

Effects of instructions on extinction and recovery
of a conditioned avoidance response. Journal of
Notterman, J. M., Schoenfeld, W. N., & Bersh, P. J. A

comparison of three extinction procedures
following heart rate conditioning. Journal of
677.
Page, H. A. The facilitation of experimental extinction

by response prevention as a function of the
acquisition of a new response. Journal of
48, 14-16.
Page, H. A., & Hall, J. F. Experimental extinction as a

function of the prevention of a response. Journal
1953, 46, 33-34.
Patterson, G. R., Littman, R. A., & Bricker, W. Assertive
behavior in children: A step toward a theory of
aggression. Monographs of the Society for
No. 113).
Polin, A. T. The effects of flooding and physical

suppression as extinction techniques on an
anxiety motivated avoidance locomotor
response. Journal of Psychology, 1959, 47, 235-
245.
Poppen, R. L. Counterconditioning of conditioned

suppression. Unpublished doctoral dissertation,
Saul, L. J., Rome, H., & Leuser, E. Desensitization of

combat fatigue patients. American Journal of
Psychiatry, 1946, 102,476-478.
Schachter, J. Pain, fear, and anger in hypertensives

and normotensives: A psychophysiological study.
Psychosomatic Medicine, 1957, 19, 17-29.
Schachter, S., & Singer, J. E. Cognitive, social, and

physiological determinants of emotional state.
Scott, P. M., Burton, R. V., & Yarrow, M. R. Social

reinforcement under natural conditions. Child
Development, 1967, 38, 53-63.
rearing. Evanston, Ill.: Row, Peterson, 1957.
Seward, J. P., & Levy, N. Sign learning as a factor in

extinction. Journal of Experimental Psychology,
1949, 39, 660-668.
Shrovon, H. J., & Sargant, W. Excitatory abreaction:

With special reference to its mechanism and the
use of ether. Journal of Mental Science, 1947, 93,
709-732.

Crofts, 1966. pp. 448-498.
Skinner, B. F. “Superstition” in the pigeon. Journal of

Slamecka, N. J. Tests of the discrimination hypothesis.

Journal of General Psychology, 1960, 63, 63-68.
Sloane, H. N., Jr., Johnston, M. K., & Bijou, S. W.

Successive modification of aggressive behavior
and aggressive fantasy play by management of
contingencies. Unpublished manuscript,
University of Illinois, 1968.
Solomon, R. L., Kamin, L. J., & Wynne, L. C. Traumatic

avoidance learning: The outcomes of several
extinction procedures with dogs. Journal of
302.
Spence, K. W. Cognitive and drive factors in the

extinction of the conditioned eye blink in human
subjects. Psychological Review, 1966, 73, 445-458.
Stampfl, T. G., & Levis, D. J. Essentials of implosive

therapy: A learning- theory-based
psychodynamic behavioral therapy. Journal of
Tyler, D. W., Wortz, E. C., & Bitterman, M. E. The effect

of random and alternating partial reinforcement
on resistance to extinction in the rat. American
Journal of Psychology, 1953, 66, 57-65.
Wagner, A. R. Conditioned frustration as a learned

drive. Journal of Experimental Psychology, 1963,
66, 142-148.
Wagner, A. R. Frustration and punishment. In R. N.

Haber (Ed.), Current research in motivation. New
York: Holt, Rinehart and Winston, 1966. pp. 229-
239.
Walters, R. H., & Brown, M. Studies of reinforcement

of aggression: III. Transfer of responses to an
interpersonal situation. Child Development,
1963,34, 563-571.
Walton, D. The application of learning theory to the
treatment of a case of neurodermatitis. In H. J.
Eysenck (Ed.), Behaviour therapy and the
274. (a)
Walton, D. The relevance of learning theory to the

treatment of an obsessive- compulsive state. In H.
J. Eysenck (Ed.), Behaviour therapy and the
164. (b)
Walton, D. Strengthening of incompatible reactions

and the treatment of a phobic state in a
schizophrenic patient. In H. J. Eysenck (Ed.),
Behaviour therapy and the neuroses. New York:
Pergamon, 1960. pp. 170-180. (c)
Walton, D. Experimental psychology and the

treatment of a tiqueur. Journal of Child Psychology
and Psychiatry, 1961, 2, 148-155.
Walton, D. Massed practice and simultaneous

reduction in drive level: Further evidence of the
efficacy of this approach to the treatment of tics.
In H. J. Eysenck (Ed.), Experiments in behaviour
therapy. New York: Pergamon, 1964. pp. 398-400.

Research and Therapy, 1963, 1, 121-125. (a)
Walton, D., & Mather, M. D. The application of learning

principles to the treatment of obsessive-
compulsive states in the acute and chronic
phases of illness. Behaviour Research and
Therapy, 1963, 1, 163-174. (b)
Weinberger, N. M. Effect of detainment on extinction

of avoidance responses. Journal of Comparative
Weisman, R. G., Denny, M. R., Platt, S. A., & Zerbolio, D.

J., Jr. Facilitation of extinction by a stimulus
associated with long nonshock confinement
periods. Journal of Comparative and Physiological
Psychology, 1966, 62, 26-30.
Wheeler, L. Toward a theory of behavioral contagion.

Psychological Review, 1966, 73,179-192.
White, J. G. Neurotic habit formations and the

experimental analysis of learning (conditioning).
Paper read to the Ulster Neuropsychiatric
Society, Belfast, April, 1962.
Whiting, J. W. M. Becoming a Kwoma. New Haven: Yale

Wickens, D. D., Allen, C. K., & Hill, F. A. Effects of
instruction on extinction of the conditioned GSR.
235-240.
Williams, C. The elimination of tantrum behavior by

extinction procedures. Journal of Abnormal and
Social Psychology, 1959, 59, 269.
Williams, C. D. Extinction and other principles of

learning in the treatment and prevention of
children’s disorders. Paper read at the meeting of
the American Psychological Association, St. Louis,
September, 1962.
Wilson, W. J., & Dyal, J. A. Effects of nonresponse

acquisition on latent extinction in a differential
reinforcement setting. Psychological Reports,
1963, 13, 879-884.
Wolfgang, M. E., & Ferracuti, F. The subculture of

violence. London: Tavistock Publications, 1967.
Yablonsky, L. The violent gang. New York: Macmillan,

1962.
Yates, A. J. The application of learning theory to the

treatment of tics. Journal of Abnormal and Social
Psychology, 1958, 56, 175-182.
Young, A. G. Resistance to extinction as a function of

number of nonreinforced trials and effortfulness
of response. Journal of Experimental Psychology,
1966, 72,610-613.
Zeilberger, J., Sampen, S. E., & Sloane, H. N., Jr.

Modification of a child’s problem behaviors in the
home with the mother as therapist. Journal of
Applied Behaviour Analysis, 1968, 1, 47-53.
7
Desensitization through
Counterconditioning
Of the various methods of behavioral
modification derived from learning theory, those
based on the principle of counterconditioning are
most widely applied to behavior in which
conditioned emotionality plays a prominent role.
These psychological conditions, which are most

frequently seen in conventional interview
treatments, include anxiety states, chronic
tensions, and other forms of autonomic

overactivity reflected in a variety of somatic
disturbances of a functional nature. Conditioned
emotionality is also involved in most behavioral

inhibitions and avoidant response patterns.
The introductory chapter discussed the
process whereby formerly neutral events, through
their conjunction with aversive experiences,

acquire emotion-arousing properties. If negatively
valenced events are repeatedly associated with

positive experiences, the stimuli gradually lose
their aversive quality. This outcome is achieved by
eliciting activities that are incompatible with
emotional responses in the presence of fear- or

anxiety-arousing stimuli.
RELATIONSHIP BETWEEN EMOTIONAL

CONDITIONING AND INSTRUMENTAL RESPONDING
Most of the desensitization procedures that
will be discussed in this chapter are predicated on
the assumption that elimination of conditioned
emotional arousal will decrease or eliminate

instrumental avoidance behavior. In the present
discussion emotional arousal encompasses both
autonomic and central arousal processes. This
outcome presupposes that classically conditioned
effects exert mediating control over instrumental!)
learned behavior. It has been demonstrated in

numerous experiments with infrahuman subjects
that avoidance responses to a given stimulus can
be established through prior classical pairings of
that stimulus with aversive experiences (Rescorla

& Solomon, 1967). These studies typically employ
a three-stage paradigm in which animals initially
learn to make instrumental avoidance responses
to aversive stimulation in the absence of the
critical cues. In the second phase, the animals

undergo a classical conditioning procedure in
which one tone (S–) is repeatedly paired with

shock to endow it with arousal potential, whereas
a different tone (S°) is never associated with shock

stimulation so as to preserve its neutral
properties; during the phase of classical

conditioning the animals are skeletally
immobilized by curare to prevent instrumental
responding. After the differential emotional
conditioning has been completed, S– and S° are

presented at random intervals under conditions
where subjects are free to make motor responses,

and the frequency with which these two stimuli
elicit avoidance responses is measured. The
stimulus that has been endowed with arousal
capacity characteristically evokes high rates of

avoidance behavior, which rarely occurs in the
presence of the neutral stimulus. Moreover, other
variables in classical conditioning that affect the

activating properties of stimuli generally produce
corresponding differences in instrumental

avoidance behavior. It has also been shown that
pairing of a stimulus with rewarding experiences
in advance later facilitates learning and retards

extinction of instrumental responding to the same
or similar cues (Bower & Grusec, 1964; Trapold &
Winokur, 1967).
Although the influence of classically

conditioned processes on instrumental
responding has been well established, the nature
of the mediators and the mechanisms through

which behavioral control is achieved have not
been determined. Several alternative explanations
have been proposed and tested primarily with

aversive conditioning paradigms (Rescorla &
Solomon, 1967). Painful stimulation elicits not
only internal emotional reactions but also
previously acquired and unlearned escape
responses. It is possible that, under conditions

where subjects are free to respond motorically
while undergoing classical conditioning,
instrumental responses are also being learned and
operantly reinforced. Some evidence suggesting
that skeletal mediators alone cannot account for
transfer is provided by Solomon & Turner (1962).
They observed such transfer effects even when
animals were classically conditioned under curare,
which prevents skeletal responding. However,

curare procedures do not conclusively rule out
motor mediators because, under lower levels of
curarization, electromyographic responses can be
increased through contingent reinforcement and
later can facilitate the occurrence of avoidance

responses in the normal state (Black, 1967). In
addition, as Rescorla & Solomon (1967) have
noted, even in completely curarized subjects
efferent neural events, which regulate responding
centrally, may be elicited and modified during

classical conditioning.
Most currently popular theories of

psychopathology assume that the effects of
aversive conditioning control avoidance behavior
through autonomic mediators. According to this
interpretation, negatively valenced cues elicit
autonomic arousal (usually designated as anxiety)
that produces afferent feedback having both
stimulus and drive properties. Avoidance behavior

eventually becomes conditioned to this
autonomically produced stimulation, so that it
both instigates and directs the performance of
defensive response patterns.
Such an anxiety theory receives little empirical

support if anxiety is equated with peripheral
autonomic reactivity, as is generally the case.
Studies in which autonomic and avoidance
responses are measured concurrently reveal that

these two sets of response events may be partially
correlated but not causally related. Black (1959)
found that avoidance responses during extinction

persisted long after autonomic responses had
been extinguished. Notterman, Schoenfeld, &
Bersh (1952) likewise demonstrated that after
subjects were provided with an effective means of
coping with a potentially threatening situation,
they continued to perform appropriate avoidance
behavior although their autonomic responsiveness

was completely extinguished. This finding is
further corroborated by Grings & Lockhart (1966),
who report that subjects exhibit a sudden drop in
autonomic arousal after learning that they can
successfully avert painful stimulation by

performing an appropriate avoidance response.
The generality of the preceding results is limited,
however, by the fact that only a single autonomic
response was measured. In view of evidence
(Lacey, 1950) that individuals display

considerable variation in their characteristic
modes of physiological reactivity to stress, and

that different responses are not highly
intercorrelated, no single measure of autonomic
reactivity can be considered an adequate index of

autonomic arousal.
Laboratory investigations of the acquisition
and maintenance of avoidance responses in
sympathectomized animals (Wynne & Solomon,

1955) provide a more critical test of the
hypothesis that autonomic responses serve a

mediating function in avoidance behavior. In the
latter experiment, autonomic function was
eliminated in a group of dogs by surgical section of
the sympathetic segment of the autonomic

nervous system, by vagus-drug parasympathetic
blocking procedures, or by combined surgical and
pharmacological treatments. The animals were

then trained to avoid an intense shock by jumping
over a barrier at a light signal. Following

avoidance learning, the shock was discontinued to
test for extinction of jumping responses to the
light alone. Unoperated animals, which served as

the comparison control group, participated in the
same experimental situation. In addition, two dogs
underwent the surgical-drug procedures and the
test for extinction after the avoidance response

had been well established.
The results disclose that removal of peripheral
autonomic responses has only a partial effect on

the acquisition of avoidance behavior, with the
differences occurring mainly in the initial phase of
learning. Sympathectomized animals were more

delayed than the controls in escaping shock,
required significantly more trials to learn their
first avoidance response, and tended to extinguish
more rapidly, although differences were slight in
this respect. However, speed of extinction in

animals deprived of normal autonomic functioning
after avoidance responses had been firmly
established did not differ from that of the normal
controls. Moreover, no consistent relationship was
obtained between the avoidance learning pattern
and the portion of the autonomic nervous system
that was blocked or resected.
All the sympathectomized animals eventually
acquired stable avoidance responses. This

suggests that autonomic arousal may play a
facilitative role, but is not required for the
establishment of avoidance behavior; maintenance
of previously learned avoidance responses is
apparently even less dependent upon autonomic

feedback stimulation. The overall evidence thus
indicates that mechanisms other than autonomic
arousal govern avoidance responding. Indeed, the
latencies of autonomic reactions and their

associated feedback are much longer than those of
skeletal responses; consequently, avoidance
behavior is typically executed before autonomic

reactions could possibly be elicited. This factor
alone precludes autonomic control of avoidance
behavior.
In a comprehensive review of the pertinent
literature, Rescorla & Solomon (1967) propose the
tenable view, principally on the basis of exclusion

rather than direct corroborative evidence, that
instrumental responsiveness is mainly regulated

by central mediators which can be established and
eliminated through classical conditioning
operations. Since central processes exert control
over both autonomic and instrumental

responding, these two response systems are, in
general, partially correlated. Major obstacles to
clarification of the role of central mediators in

avoidance behavior are created by the failure to
specify the locus and nature of the mediating

systems and the most valid indices of their
activities. The problem is further complicated by
suggestive evidence (Lacey, 1967) that the

different arousal systems—
electroencephalographic, autonomic, and
behavioral—are functionally separable. Although
they generally appear concomitantly, physiological

and behavioral arousal can be markedly
dissociated pharmacologically. Thus organisms
may be centrally aroused but behaviorally

unresponsive, or conversely, they may be
behaviorally excited in the absence of central

activation as measured by standard electrocortical
signs. These findings indicate that, under certain
conditions, external stimuli may control avoidance
responses independently of physiological arousal.
Nevertheless, it is clear from studies in which

stimuli are endowed with physiological arousal
properties under curare that the stimuli are not

directly conditioned to avoidance responses since
these never occur. Rather, in the early stages the
responses appear to be controlled by mediating
events that are common to other stimuli to which

avoidance responses have been previously
learned. After avoidance responses habitually
occur with reinforcing consequences in the
presence of cues, such conditioned aversive

stimuli eventually acquire discriminative value
and can exert control over avoidance behavior
without emotional arousal. This type of shift in the
locus of stimulus control accords with common

observation that mediating functions diminish as
response patterns become routine.
Whatever the specific regulatory mechanisms

might be, the fact that overt behavior is modifiable
by classical conditioning procedures has
important treatment implications. Of particular

relevance are studies demonstrating that
neutralization of an aversive stimulus alone
markedly facilitates subsequent extinction of
avoidance behavior. In an experiment conducted

by Black (1958), after animals learned to make
shock avoidance responses to a tone, they were
skeletally immobilized by curare to prevent
confounding results from any performance
extinction during the treatment period. Animals in

the control group received 50 performance
extinction trials and were then given curare
without special treatment. The experimental
group simply experienced 50 presentations of the
fear-provoking tone while in a curarized state. In a

subsequent test, subjects who had received the
classical extinction treatment required
approximately 40 regular extinction trials to
completely eliminate their avoidance behavior,
whereas the group that was given performance

extinction needed an additional 450 trials before
they discontinued making avoidance responses.
In the foregoing study, classical extinction was
achieved by repeated exposure to anxiety-
arousing stimuli without any adverse experiences.
The extinction process can be hastened by
presenting threatening stimulus events along with
positive stimuli that elicit incompatible responses
strong enough to supersede anxiety reactions. For

example, Gale, Sturmfels, & Gale (1966) found that
conditioned emotionality was eliminated more
rapidly by repeated paired presentations of
aversive stimuli with food than when the same
aversive stimuli were presented alone.
The facilitative effects of eliciting antagonistic

responses on extinction of emotional behavior are
even more clearly illustrated by Poppen (1968) As
part of a larger experiment, he compared the

speed with which behavioral inhibitions were
eliminated in animals when graduated aversive
stimuli were presented either alone or in

conjunction with food rewards. Figure 7-1 shows
the mean number of exposures required to
extinguish fear at each of the stimulus values of
the aversive hierarchy for subjects receiving
Figure 7-1. Mean number of nonreinforced trials required to
eliminate response suppression at each of the stimulus
values of the aversive hierarchy through extinction and
counterconditioning procedures. Poppen, 1968.
graduated extinction and graduated
counterconditioning. Emotional responsiveness in
both groups was eliminated at an equally rapid

rate toward stimuli of low and of moderate threat
value. When confronted with severe threats,
however, subjects administered the
counterconditioning treatment required
substantially fewer exposures to overcome their

fearfulness.
In the behavior therapy literature operations
in which aversive stimuli are presented alone are
typically designated “extinction,” whereas multiple

procedures combining fear-arousing and positive
stimuli are labeled “counterconditioning.” These
methods are often discussed as though they

involved fundamentally different processes.
Actually, counterconditioning is a major factor in
extinction. That is, during nonreinforced
repetitions of a stimulus, temporary inhibitory
states and inevitable changes in the stimulus
complex eventually elicit competing responses of
sufficient strength to replace original reactions.

According to this interpretation of the process,
conventional extinction procedures often involve a
form of unguided counterconditioning. A major
advantage of methods that include counter-
response elicitation is that the occurrence and

strength of competing activities are managed
rather than left to fortuitous factors; this permits
greater control over desired outcomes.
CONCEPTUALIZATION OF THE
COUNTERCONDITIONING PROCESS
Although applications of the principle of
counterconditioning were reported by Jones as
early as 1924, the approach received little

attention until Wolpe (1958) devised an ingenious
procedure that greatly increased the range of

disorders subject to treatment by this method.
Based on careful analysis of the stimulus
determinants of emotional responsiveness, the
therapist constructs a ranked list of situations to

which the client reacts with increasing degrees of
anxiety or avoidance. When counterconditioning is
based on relaxation procedures, the therapist
induces in the client a state of deep relaxation,

which presumably counteracts anxiety, and asks
the client to visualize the weakest item in the
hierarchy of emotion-arousing stimuli. If the client
experiences any emotional disturbance the
aversive scene is promptly withdrawn, relaxation

is reinstated and then the item is repeatedly
presented until it ceases to evoke anxiety; if the

relaxation remains unimpaired in the imagined
presence of the threat, the client’s emotional

responses to the next item in the hierarchy are
extinguished and so on throughout the graduated

series. In this manner the intensity of aversive
stimuli is increased from session to session until
the most threatening events have been completely
neutralized. Further details of this particular

method and its variant forms have been published
by Wolpe (1961), Wolpe & Lazarus (1966), and

Lazarus (1964).
Wolpe considers three sets of variables
essential for achieving consistent

counterconditioning outcomes. First, it is
necessary to select an anxiety-neutralizing
stimulus capable of inducing a competing
condition of sufficient strength to overcome the
reactions ordinarily evoked by the emotion-

arousing cues. Second, the aversive events are
presented initially in attenuated forms so that
emotional responses to be counteracted are
relatively weak and hence can be readily
extinguished. The arousal potential of more
aversive situations supposedly is progressively
reduced by generalization of anxiety extinction
from preceding weaker items. Through successive
advances of extinction and generalization, stimuli

of increasing aversiveness can be gradually
neutralized without evoking anxiety of an
intensity difficult to countercondition. The third
variable pertains to necessary temporal
prerequisites, that is, both the anxiety reducing

and the aversive stimuli must be contiguously
associated.
As will be shown later, Wolpe’s desensitization
method has generally proved successful in

modifying emotional behavior, but the theoretical
speculations about the manner in which anxiety is
acquired and the mechanisms governing the

counterconditioning process (Wolpe, 1958) are
largely disputed by empirical findings. In accord
with Hullian formulations, Wolpe favors a drive-
reduction theory of classical conditioning, and a
fatigue theory of extinction. In contrast to this
view, experimental results (Mowrer, 1960;
Solomon & Brush, 1956) clearly support a

contiguity theory of conditioning, in that
emotional responsiveness is best acquired and
strengthened through association of a stimulus
with shock onset rather than shock reduction.
Although fatigue resulting from nonreinforced

evocation of effortful behavior may foster the
appearance of inhibitory responses in
performance extinction, it is doubtful that
sufficient fatigue, if any, can be generated by
symbolic and autonomic responses to account for

anxiety decrements achieved through distributed
trials in symbolic desensitization treatments. A

more plausible interpretation of extinction under
the latter conditions is that emotional responses
are gradually eliminated by deliberate elicitation

of incompatible responses and by superimposing
aversive stimuli onto positive events that mitigate
self-generated arousal.
In accord with most traditional theories of
psychopathology, Wolpe (1958) adopts the
position that anxiety is a major causal determinant

of inappropriate avoidance behavior. Anxiety is
defined principally in terms of over-reactivity of
the sympathetic division of the automatic nervous

system. Counterconditioning effects are explained
by Wolpe in terms of reciprocally inhibitory
processes occurring at the level of the autonomic
nervous system. This conceptual scheme is mainly
based on the assumption that sympathetic and

parasympathetic responsiveness are generally
physiologically antagonistic. It is further assumed
that muscular relaxation, sexual behavior,
assertive responses, and other pleasurable
stimulation elicit parasympathetic responsiveness
which, if sufficiently strong, inhibits the
predominantly sympathetic responses of anxiety.
It is important to bear in mind that the
psychological principle of counterconditioning and

the efficacy of procedures based on this principle
are independent of the validity of Wolpe’s
neurophysiological speculations. Indeed, contrary
to Wolpe’s peripheral theory of anxiety, research
findings cited earlier clearly indicate that

autonomic and avoidance responses are coeffects
rather than causally linked events. To the extent
that extinction is governed by mutually inhibitory
mechanisms, they are more likely to operate

subcortically rather than in the autonomic system.
It is interesting to note in this connection that
some evidence exists (John, 1961) for two

reciprocally inhibitory arousal systems in the
reticular formation which mediate defensive and
approach behavior.
Controlling Variables in Desensitization
A number of laboratory experiments have been
conducted to determine whether the component
variables in desensitization procedures are

necessary, facilitative, or irrelevant to extinction
outcomes associated with this method.

Unfortunately, results of many of these studies are
uninterpretable because their sample sizes are
much too small to prove anything, outcome
measures are inadequate, treatment conditions

are applied for exceedingly brief periods, and
other methodological deficiencies exist that are
typically excused on the grounds that the
experiments are merely exploratory. Although
such studies are usually acknowledged by the

authors to be technically insufficient, the resultant
findings are rarely dismissed as having little
evidential value. The findings of some experiments

that are otherwise well designed may be
misleading because the behavioral test used
requires, at the most, brief contact with the feared
object (e.g., touching or holding a snake).

Treatment conditions that appear equally
efficacious on the basis of a weak criterion test
might yield differential outcomes if more
threatening and demanding performance tasks
were utilized.
The research discussed in subsequent sections

is mainly confined to experiments that are
sufficiently well designed to permit meaningful
interpretation of the data. Many of these studies

employ the snake phobia paradigm originally
devised by Lazovik & Lang (1960). This type of
phobic disorder is especially well suited for

clarifying the role of variables considered to be
influential contributors to the counterconditioning
process. The reason for this is that the incidence of
snake phobias is relatively high, the strength of
avoidance behavior can be objectively measured,
and extra-experimental encounters with snakes
that might confound treatment effects rarely occur

or can be easily controlled.
ROLE OF ANXIETY-COMPETING RESPONSES AND

EXTRANEOUS PROCESSES
If, in fact, desensitization methods involve a
counterconditioning process, then contiguous

association of threatening stimuli with anxiety-
inhibiting responses would constitute a necessary
condition for rapid elimination of avoidance

behavior. In a relatively complex treatment
containing numerous elements it is possible that
any number of variables, operating either singly or
in combination, may be responsible for observed
outcomes. Thus, for example, avoidance behavior

may be reduced to some degree by relaxation
training alone, by gradual exposure to

progressively more threatening situations, or by
expectations that participation in a treatment
program will result in favorable changes. Another
possible source of influence is the social

relationship that develops between change agents
and their clients. In order to test whether
desensitization methods achieve their effects
through counterconditioning or extraneous

processes, Davison (1968) conducted an
experiment that proceeded in the following
manner.
Snake-phobic students were individually
matched on the basis of the strength of their
avoidance behavior toward a snake and assigned
to one of four conditions. For students who

received the treatment that fulfilled
counterconditioning requirements, imaginal

representations of progressively more threatening
interactions with snakes were contiguously paired

with muscular relaxation, as in the standard
practice. A second group participated in a pseudo-
counterconditioning treatment that was identical
to the procedure employed in the first condition

except that the symbolic contents paired with
relaxation were childhood experiences entirely
unrelated to snakes. Because of the widespread
belief that anxiety disorders represent derivative
manifestations of reactivated infantile conflicts, it

was possible to use snake-irrelevant items without
jeopardizing the verisimilitude of the treatment
approach. This group provided a control for the
effects of extraneous variables associated with
relationship experiences, expectations of

beneficial changes, relaxation training, or other
possibly unrecognized factors. The third group

was administered the same set of graded fear-
provoking scenes involving snakes but in the
absence of relaxation. This exposure condition

primarily served as a control for the influence of
repeated exposure to threatening stimuli. Finally,
a small control group of students participated in
the assessment procedures without receiving any

form of intervening treatment.
In order to equate the groups for durations and

specific patterns of experiences, students in the
pseudo-counterconditioning and the exposure
groups were yoked to their matched partners in

the counterconditioning treatment, whose
progress determined the total number of
treatment sessions, the length of each session, and
the number and duration of each stimulus
exposure. After the treatments were completed

students were readministered the behavioral test
involving 13 progressively more intimate
interactions with a snake. Only the subjects for
whom fear-arousing events were paired with
relaxation displayed substantial increases in
snake-approach behavior, whereas students in the
pseudo-counterconditioning, exposure, and
control groups failed to achieve any significant
reductions in avoidance responses.
The question of whether extinction of
avoidance behavior through desensitization is
attributable to relationship influences or other
unsuspected variables has also been investigated
by Lang and his associates. In the initial project

(Lang & Lazovik, 1963; Lang, Lazovik, & Reynolds,
1965), which involved snake-phobic adults, one
group received the standard form of
desensitization treatment; a second group

participated in a form of relationship therapy in
which, after receiving a plausible explanation for
their placebo treatment, they discussed

experiences unrelated to their phobia in the
context of deep relaxation. A no-treatment control
group was also included.
Control subjects and those who received the
pseudotherapy showed no significant changes
either in their snake-avoidance behavior or on any

of the self-report indices of anxiety. By contrast,
students who received the counterconditioning

treatment exhibited greater snake approach
behavior relative to the combined results of the
latter two groups, and they experienced less
anxiety about snakes. However, results of this

study must be accepted with reservations for
several reasons. During the measurement of
phobic behavior the tester modeled each approach

response before requesting the subject to perform
the same task. Although the amount of modeling

may not be sufficient to reduce inhibitory
responses in control subjects, it may facilitate
approach behavior in subjects whose avoidance

tendencies had been weakened to some extent
through prior counterconditioning. Results of
other studies (Wolpin & Raines, 1966) are
similarly confounded by extensive modeling of

intimate interactions with feared objects during
the administration of avoidance tests. It also
appears from the magnitude of change scores that

some of the group differences at a borderline level
of significance would most likely prove

nonsignificant had desensitization been compared
to pseudotherapy alone rather than combined
with the nontreated control group. There is no
question, however, that students who had been
successfully desensitized to most of the items in

the anxiety hierarchy achieved substantially
greater reductions in avoidance behavior than

subjects in either the relationship or control
conditions.
Lang (1968) has devised a self-directed
desensitization procedure that makes it possible
to manage counterconditioning variables more
reliably and permits greater experimental control
over extraneous processes. Graduated sequences
of threatening situations and relaxation

instructions are prerecorded on magnetic tapes
that are controlled by the person undergoing the
treatment. After relaxation instructions have been
played an anxiety-arousing item is automatically
presented. Whenever subjects signal distress they

are instructed to stop visualizing the scene,
relaxation is reinduced, and then the item is
repeated. If subjects signify an increase in anxiety
during reexposure to the same aversive scene they
are returned to the preceding item in the

hierarchy. As long as subjects signal decreasing
arousal to successive exposures to a given scene, it

is repeatedly presented until it ceases to elicit
emotional responses. In this way, subjects manage
their own desensitization treatment throughout

the graduated series.
The above procedure was primarily designed
for research purposes, but a case report by Migler
& Wolpe (1967) suggests that it may have clinical

applications as well. A male client who was unable
to participate in staff meetings because of severe

public speaking anxieties successfully desensitized
himself at home through the use of a modified tape
recorder that contained prerecorded relaxation
instructions and scenes of increasingly

threatening public speaking situations. These
preliminary clinical data are corroborated by
results from comparative studies by Melamed &

Lang (1967), Donner (1967), and Krapfl (1967),
who found that self-administered desensitization

produced the same amount of reduction in
avoidance behavior as the standard, socially
administered form. Lang has also successfully

employed the semiautomated procedure to
investigate changes in autonomic indicants of
emotional arousal through the course of the
desensitization process.
Results of a study designed by Moore (1965) to
assess some of the factors operative in the
desensitization method are of interest because

they essentially replicate the findings cited above
with a radically different type of emotional
dysfunction. Asthmatics who had proved

unresponsive to medical treatment were given
either relaxation alone, relaxation combined with
suggestions that they would show both
progressive improvement in respiratory function
and reduced sensitivity to situations that evoke

asthmatic attacks, or they received the
counterconditioning treatment. In the latter
procedure deep relaxation was paired with graded
situations based on respiratory difficulties,
infective and allergic factors, and stress-provoking
events. Each patient was administered two of the
treatments over a period of two months according
to an experimental design that presented the
methods in every combination and in every order.

The effects of these various procedures were
assessed in terms of subjective reports of
asthmatic attacks and two objective measures of
respiratory function. These included the Maximum
Peak Flow and the percentage this performance

represented of optimal flow after an inhaled dose
of isoprenalin.
The changes accompanying the different
treatment conditions are summarized graphically

in Figure 7-2. All three treatments produced
reductions in attacks of wheezing according to
clients’ self-reports, but only the

counterconditioning method significantly
improved respiratory function based on physical
measurements.
Figure 7-2. Changes in reported asthmatic attacks, and two
physiological measures of respiratory function
associated with each of three treatment conditions.
Moore, 1965.
Although the findings reported by Davison
(1968) and Rachman (1965) would seem to
indicate that relaxation plays a highly influential

role in symbolic desensitization, this conclusion
requires qualification in the light of results of a

study conducted by Schubot (1966). It will be
recalled that in Davison’s experiment, subjects in
the exposure condition were often required to
continue visualizing disturbing scenes after they

had signaled anxiety, in order to equalize exposure
durations in the different treatments. Miller
(1967) has shown that the desensitization

procedure achieves equally beneficial results
regardless of whether the subject or the

experimenter controls the termination of aversive
stimuli. It is conceivable, however, that if subjects
who were administered only the aversive stimuli

had been allowed to terminate threatening scenes
before they generated excessive anxiety, which
was the procedure followed for the
counterconditioning group, repeated exposure

alone eventually might have produced some
extinction of avoidance behavior. To test this
notion, Schubot (1966) compared the elimination

of phobic behavior in groups of adults who were
administered either exposure to anxiety-

provoking scenes paired with relaxation or
exposure alone under conditions where aversive
stimuli were promptly terminated for all subjects
whenever they signified distress.
Interpretation of results of this otherwise well-
designed experiment is somewhat complicated by

lack of a yoked exposure condition in which
visualization durations are externally controlled
independently of subjects’ emotional responses.
Nevertheless, the available data (Figure 7-3)
demonstrate that relaxation was essential for
modifying extreme phobic behavior, but it did not
Figure 7-3. Differential responsiveness of moderate and
severe phobic subjects to counterconditioning and
extinction procedures. Schubot, 1966.
facilitate extinction of avoidance responses of
moderate strength.
It is of considerable interest to find that covert

extinction operations, in which conditioned
emotionality is eliminated by repeated symbolic
reinstatement of threatening events below the
stress threshold level, significantly reduce
avoidance behavior. If people could extinguish

inappropriate emotional reactions by thought
alone, one would expect such disorders to be
much less prevalent. To be sure, under naturalistic
conditions people rarely construct hierarchies of

emotionally disturbing situations and
systematically engage in covert extinction trials.
Moreover, on the occasions when they do think

about threatening events they are apt to revivify
the aversive consequences that accompanied their
behavior, thereby reinforcing rather than
weakening their fears. Because of the relevance of
covert extinction for the general issue of symbolic
control of overt behavior, it would be of
importance to determine whether this

phenomenon is easily reproducible and its limiting
conditions.
Several investigators have found that the
components of desensitization separately achieve
reductions in avoidance behavior, but they do not

gain additional efficacy in combination with each
other. In some of these experiments, however, the
behavioral test is exceedingly brief, experimenters
lack experience in the use of the method (Cooke,

1968), or subjects receive limited training in
relaxation (Proctor, 1968). Other studies, such as
the one reported by Folkins, Lawson, Opton, &

Lazarus (1968), suffer from methodological
deficiencies and a tendency to read more into the
data than they actually yield. Folkins and his
associates measured self-reports and
physiological responses of students to a film
depicting industrial accidents after they had
received one of four types of treatments. For

subjects in the simulated desensitization
condition, brief scenes from the film were
presented verbally in the context of muscular
relaxation and pleasant imagery during three
sessions. Unlike the standard procedure, however,

the stressful stimuli were presented by tape
recording without regard to the students’
emotional reactions. The second condition
included both the positive imagery and exposure
to the aversive scenes. A third group received only

relaxation training, while students in the fourth
condition served as nontreated controls.
A treatment effect was obtained on one of
three self-report indicators of stress reactions, and
on the skin conductance measure. No significant
differences were found, however, in heart rate
responses. The authors conclude, on the basis of
visual inspection of the data rather than statistical
evaluation of inter-group differences, that

complete desensitization is a less effective stress
reducer than either of its components, relaxation
or cognitive rehearsal, and that of the two
elements, cognitive rehearsal is the more
powerful. They further suggest that insight-

oriented approaches may be superior to
desensitization techniques. These conclusions are
supported by neither the data nor the treatment
operations. All treatments probably reduced
emotional responding compared to the control

group, but it appears from the summary data that
the treatments do not differ significantly among

themselves. With regard to the procedures, since
the “cognitive rehearsal” involved both stimulus
exposure and positive imagery that has

counterconditioning potential, this method
actually represents a variant form of
desensitization. It might also be noted in passing
that if visualization of aversive stimuli is

conceptualized as an insight operation, then the
term has little meaning.
The various findings, taken as a whole, indicate
that relaxation is a facilitative rather than a
necessary condition for elimination of avoidance

behavior. Evidence that relaxation often hastens
the extinction process does not verify that the
benefits derive from the explicit manipulation of
muscular activities. Indeed, Rachman (1968) has
argued that feelings of calmness induced by the

procedure rather than muscular relaxation per se
is the decisive factor at work. In this alternative
explanation, relaxation instructions and
presentation of pleasant scenes to the imagination
reduce affective arousal which attenuates
responsiveness to aversive stimuli. This
interpretation is consistent with the view
advanced in this book that behavioral changes are
largely governed by central mechanisms rather

than by peripheral processes.
If the competing activities that function to
reduce the arousal capacity of threatening stimuli
are, in fact, symbolically mediated, then certain
changes in the standard desensitization practice

may be advantageous. As Rachman suggests,
greater emphasis would be placed on the
development of tranquil and pleasant imagery
than on motor relaxation exercises. This issue can

be best resolved by laboratory studies of the
anxiety- mitigating effects of positive imagery and
muscular relaxation when used alone and in

combination with each other.
GRADUATED STIMULUS PRESENTATION
Research discussed thus far discloses that
deconditioning effects occur even when social and
expectancy influences are controlled, and that
relaxation functions as a facilitative rather than as
a necessary condition for change. The question of

whether graduated presentation of aversive
stimuli is a fundamental requirement of

desensitization has not been systematically
investigated. If the process of anxiety elimination
through performance extinction and symbolic
desensitization involves analogous processes, then

reductions in avoidance behavior can be achieved,
both by reexposure to progressively more
threatening events and by repeated confrontation
with the most feared situation at the outset.
However, these two treatment strategies would be

associated with markedly different amounts of
anxiety elicitation. The more stressful
confrontation method is apt to generate high
levels of emotional arousal that are gradually
reduced with successive nonreinforced occurrence
of fear- provoking events. On the other hand, when
aversive stimuli are introduced in attenuated form

and gradually increased to their full threat value,
extinction effects can be attained with minimal

anxiety arousal. Indeed, by initially presenting an
aversive stimulus in a weak form so that it will not
elicit any anxiety, and by increasing its duration
and intensity in small, progressive steps, it should

be possible to extinguish emotional
responsiveness without the occurrence of
emotional responses.
Anxiety-free extinction in aversive situations

has received little study. Experiments with
infrahuman subjects (Kimble & Kendall, 1953;
Poppen, 1968) have shown that exposure to
stimuli graduated in aversiveness produces more
rapid extinction of emotional behavior than when
they are repeatedly presented at their full value.
Terrace (1966) has provided considerable
evidence that discriminative behavior can be
established with virtually no responses to negative

stimuli, through the use of progressive stimulus
change procedures. Thus to alter the responses

made to a negative stimulus (S–) it is gradually
introduced into a positive stimulus complex (S+)
that evokes a desirable form of behavior. The

elements in the latter stimulus are progressively
reduced until eventually S– alone produces the
responses that were originally controlled by S+.
Evidence that stimulus control can be transferred
by this method without negative responses brings

into question the widely shared belief that the
occurrence of anxiety responses is a necessary

condition for their elimination. It is essential,
however, that the stimuli that are being
neutralized should have anxiety-arousing potency

in their original form. One would expect little
therapeutic gain from programs that fostered
nonemotional reactions to stimuli that ordinarily
do not exercise control over emotional responses.

In other words, the requirements pertain to the
motivational properties of stimuli rather than to

the elicitation of emotional responses.
In the counterconditioning paradigm the
introduction of anxiety-competing responses

presumably enables people to tolerate higher
levels of threat without responding anxiously.
Some suggestive evidence that this does, in fact,
occur is furnished by Davison’s study (1968), in
which subjects in different treatment conditions

were individually yoked, and hence received the
same number, order, content, and duration of
stimulus exposures. Students who were
administered threatening scenes in the context of
deep relaxation signaled distress on 27 percent of
the stimulus presentations, whereas those who
received the same items without relaxation
registered anxiety on 61 percent of the trials.
Schubot (1966) reports similar differences in

anxiety elicitation, under conditions where
subjects themselves controlled the termination of
aversive scenes, between subgroups that
displayed extreme avoidance behavior. The rate of
anxiety- signaling by markedly avoidant subjects

who received exposure only was three times as
high as that shown by their equally fearful
counterparts for whom exposure was paired with
relaxation. Moreover, the emotional reactions of
the latter group were neutralized more rapidly to

individual scenes and they completed significantly
more items in the hierarchy during the same

period of time. Interestingly, the anxiety-signaling
rates did not differ between the moderately fearful
subgroups who achieved comparable gains in

approach behavior. Consistent with the view
expressed earlier regarding anxiety responding
during extinction, Schubot found that frequency of
anxiety elicitation was inversely related to degree

of extinction of avoidance behavior.
Assuming that subjects’ imaginal self-

stimulation corresponds closely to scenes
presented by the experimenter, the above findings
indicate that an approach that combines

graduated stimulus presentation with deliberate
elicitation of anxiety-neutralizing responses is
best suited for reducing avoidance behavior with
minimal stress. In evaluating differential
treatment approaches one must consider not only

the rate at which they modify avoidant behavior,
but also the emotional costs to the client. The
latter criterion is particularly important if a given
method achieves relatively quick results but
drives away many of the participants because it
engenders excessive distress.
Findings of an experiment by Krapfl (1967) are
relevant to several of the issues discussed above.
Snake-phobic subjects received socially

administered desensitization or several forms of
desensitization conducted via tape recordings. In

the semiautomated treatments the aversive
stimuli were presented either in order of
increasing aversiveness as in the standard
procedure, in a descending arrangement from

most to least anxiety- arousing, or in a random
order. Two control groups, one that received no
treatment and another that was presented

pleasant but snake-irrelevant stimuli, were also
included. Behavioral avoidance tests were

administered after five sessions of treatment and
again six weeks later.
Subjects in all treatment conditions achieved
enduring increases in approach behavior and
differed in this respect from the two control
groups (Figure 7-4). No significant differences
were found among the experimental conditions,
except that the randomized procedure

consistently produced weaker effects than did
procedures employing the ascending aversive
order. Although desensitization that proceeded
from most to least aversive items proved
efficacious on behavioral indices, it elicited

initially a high level of emotional responding and
negative reactions to the procedure. In clinical
applications this method, therefore, runs a higher
risk that clients might terminate their
participation.
TEMPORAL FACTORS IN DESENSITIZATION
If stimulus events are to lose their arousal
capacity through a process of counterconditioning,
then emotion-provoking and emotion-countering
stimuli must be contiguously associated. Melvin &
Figure 7-4. Mean number of approach responses performed
by subjects in each of four treatment conditions and in
two control conditions. Plotted from data of Krapfl,
1967.
Brown (1964) found that repeated paired
presentations of a positive reinforcer with a
physically noxious stimulus reduced its aversive

power, but temporally dissociated presentations
of these same events did not alter its negative
valence. Moreover, reduction in aversiveness
varied directly with the number of paired

associations. It is evident, however, that the
temporal relationship between stimulus events in
the standard desensitization procedure, while
meeting the association requirement, differs
markedly from that generally considered optimal

for classical conditioning. In the latter case,
conditioned responses are most readily produced

when the CS precedes the UCS by a very brief
interval. By contrast, in the desensitization

procedure, relaxation, which is supposed to serve
the same function as UCS-induced states, is

maintained continuously, whereas the conditioned
aversive stimuli are presented briefly at irregular
intervals.
If one equates classical conditioning with a
limited set of operations in which stimulus events
are presented episodically and in very close

temporal conjunction, then obviously Wolpe’s
procedure does not fulfill these specific
requirements, nor, incidentally, does a large body

of literature demonstrating classical aversive
conditioning. For example, experiments in which
animals are shocked in compartments painted a
certain color, the CS (i.e., the color cue) is
continuously present and the painful stimulation

is experienced intermittently. Similarly,
conditioned emotionality can be extinguished
when aversive cues and positive stimuli that elicit
antagonistic responses are both continuously
present (Farber, 1948).
In order to establish conditioned responses it
is essential that the effects induced by a UCS occur
in conjunction with the CS. Under circumstances

where the events to be conditioned to the CS are
controlled almost exclusively by the UCS then a

close temporal relationship between these two
sets of stimuli is required. However, in human
learning, emotional responses are generally
elicited not only by an external UCS but also by

symbolic representations of aversive or
pleasurable experiences. The influential role of
self-generated arousal in classical conditioning is

indicated by studies showing that conditioned
responses can be developed by having subjects

merely associate a CS with imagined stimulation in
the absence of the appropriate UCS; conversely,
conditioned responses generally fail to develop,

even though the CS and UCS are presented
repeatedly under optimal temporal contiguity, if
subjects do not recognize that the two stimuli are
related. Research bearing on these issues and

alternative interpretations of empirical findings
are discussed more fully in the concluding chapter.
To the extent that self-generated stimulation

substitutes for, or supplements, external inputs,
conditioning can occur under a variety of external

temporal arrangements. The question of whether
desensitization outcomes are achieved through
conditioning in the traditional sense should
perhaps be held in abeyance until the
psychological processes underlying classical

conditioning have been more adequately
explicated.
In addition to the temporal issue, questions
also arise about the nature of the conditioned
response. It is apparent from laboratory studies of
classical conditioning that the conditioned
response is rarely, if ever, identical to the behavior
originally evoked by the unconditioned stimulus.
For example, a painful shock delivered to a
person’s forearm following a tone will typically

elicit autonomic reactions, arm retraction, and
vocalization of pain. After a series of trials,
presentation of the tone alone will most likely
elicit autonomic and central arousal without the
motor and vocal elements. Indeed, if the organism

were so constructed that a conditioned stimulus
had the capacity to create the tissue damage
accompanying physically injurious events, then
learning would have self-destructive rather than
survival value. A conditioned stimulus not only

evokes merely a component part of the original
reaction, but it often activates anticipatory

responses that bear little resemblance to the
unconditioned response. It would, therefore, seem
more plausible to view conditioning outcomes as

reflecting the operation of mediating mechanisms
rather than the direct coupling of stimuli with
responses evoked by other events. Since
responses to an aversive stimulus contain a self-

generated component one would not expect them
to be identical to those produced by the external

stimulus.
Similarly, after an aversive stimulus has been
repeatedly paired with relaxation, it is exceedingly

unlikely that formerly threatening cues will
promptly evoke muscular flaccidity. Rather, such
cues no longer generate emotional arousal. The
foregoing interpretation of counterconditioning
differs from explanations based on Guthrie’s

theory (1935) or Wolpe’s model (1958), both of
which assume a recoupling of conditioned
responses to stimuli.
INFLUENCE OF SOCIAL REINFORCEMENT AND

COGNITIVE VARIABLES ON DESENSITIZATION
The studies reviewed thus far investigated the
influence of traditional learning variables on
desensitization processes. This class of

determinants is primarily concerned with
stimulus events, that is, their content, intensity,

valence, frequency, mode of presentation, and
temporal relationship. Learning variables often
exert differential effects on behavior depending
upon concomitant social and cognitive variables.

The relationships obtained between responses
and their programmed stimulus consequences, for
example, may vary considerably when the same

reinforcing stimuli are administered by persons
who vary in prestige (Prince, 1962), attractiveness

(Marder, 1961), sex (Epstein & Liverant, 1963;
Stevenson, 1965), ethnic status (Smith & Dixon,
1968), and friendship (Hartup, 1964; Patterson &

Anderson, 1964). Similarly, informational
variables may be influential in determining
responsiveness to stimulus presentations. For
these reasons, the behavioral changes effected by

desensitization cannot be solely attributed to the
effects of stimulus pairings.
Not only is learning multiply controlled by

interacting variables, but the willingness to
perform responses that have been acquired can be
affected by a host of motivation-related influences.

Subtle situational demands, self-imposed
achievement pressures, expectations that a given
method will result in beneficial changes, and the
desire to please conscientious therapists are often
invoked as unsuspected determinants of

psychotherapeutic changes. Such factors are most
likely to exercise some control over existing
behaviors that persons can readily perform should
they wish to. On the other hand, motivational
influences alone will not produce response
patterns that are lacking, nor will they meet with
much success in restoring severely inhibited
behaviors that have proved intractable to
repeated social pressure and persuasion. In the

latter case, requisite learning experiences must be
provided.
The prevalent tendency to view multiple
determination of behavior in terms of rival
preferences for certain component variables has

given rise to much unproductive argument and
research. This is particularly true of investigations
that are explicitly designed to negate learning
influences. In an effort to demonstrate that

expectations alone regarding treatment might
account for reductions in avoidance behavior in
desensitization therapy, Efran & Marcia (1967)

administered to snake-fearful students a
pseudotherapy in which they were occasionally
shocked while observing blank slides in a
tachistoscope that they were led to believe
contained pictures of the phobic stimuli presented
at subliminal levels. Half of these subjects were
assigned to a “low expectancy” condition in which

they were informed that the treatment lacked a
crucial element, whereas subjects in the “high
expectancy” condition were told that the
treatment yielded promising results. To further
enhance the expectancy manipulation, subjects

were shown fictitious polygraph records
indicating that their physiological reactivity to the
“subliminal stimuli” had decreased over the course
of treatment. A control group participated in the
assessment procedures without exposure to any of

the experimental procedures. Unlike previous
researchers, who have tested changes in

avoidance behavior toward live reptiles that have
some realistic threat value, for some unexplained
reason Efran & Marcia used lifeless specimens of

the phobic objects. Lifeless objects are likely to
arouse relatively weak inhibitions that would be
especially susceptible to motivational
manipulations.
On the basis of a partial analysis of the data,
the authors conclude that positive expectations

can reduce fear responses. This finding is
considered especially significant since punishing
subjects for reacting fearfully should, if anything,

increase their avoidance behavior.
In point of fact, the three groups of subjects did
not differ significantly in degree of self-reported

fear experienced during the test of approach
behavior. The authors attribute this negative
finding to a lack of independence between
fearfulness and total amount of approach behavior
performed and to the susceptibility of this
measure to situational demands. The
methodological confounding of the dependent
variable could have been easily avoided by
measuring the degree of fear associated with each
specific approach response rather than obtaining a

single aggregate rating after the avoidance test
was completed. Considering that expectancy cues,
through which situational demands are most
frequently conveyed (Rosenthal, 1966), were

explicitly manipulated in this study, it seems
illogical to discount the absence of differences on a
measure because of its susceptibility to situational
influences. Statistical comparisons were also made
on measures of approach behavior and interview

ratings of improvement between various
combinations of groups except high and low

expectancy, the conditions of major relevance to
the expectancy hypothesis. Complete analysis of

their published data reveals that experimental
subjects achieved a higher rate of improvement

than the controls, but the high and low expectancy
groups did not differ significantly from each other
on either of these measures. In direct contrast to
the authors’ conclusion, the results show in fact

that the increases in approach behavior produced
toward the attenuated threats, though in the

predicted direction, were essentially comparable
regardless of whether subjects expected the
treatment to be effective or ineffective. It should
also be noted that “principles of conditioning”

would not lead one to predict that several random
pairings of shock with a blank card interspersed in
a hundred trials should necessarily increase

avoidance of snakes or spiders. Indeed, the
opposite outcome is entirely possible considering

that subjects were led to believe that unconscious
responses to subliminal phobic stimuli would be
followed by painful shocks, whereas the

occurrence of the assumed stimuli did not result in
aversive experiences on 84 of the trials! It remains
an open question whether the authors’ procedure
is more likely to condition anxiety relief, rather
than aversive reactions, to whatever the subjects

were imagining to the blank slides.
Research in which social and cognitive

variables are studied as components of
counterconditioning procedures can provide
valuable information about the degree to which

these different factors, both singly and in
combination with each other, facilitate extinction
outcomes. If it were found that response to
desensitization was partly determined by induced
expectations and other informational inputs, their

mechanisms of action would still remain to be
explained. Since induced negative expectations
tend to decrease behavioral participation (Kelley,
1950), supposedly cognitive variables might affect
conditioning outcomes mainly through peripheral
processes, by reducing attentiveness and by
eliciting only halfhearted adherence to required
procedures.
A study by Valins & Ray (1967), designed to

demonstrate that cognitive labeling of one’s
internal physiological reactions can affect
avoidance behavior, is likewise marred by
methodological deficiencies that yield data easily
subject to misinterpretations. According to the

authors’ reasoning, to the extent that persons can
be led to believe that they are no longer affected
internally by a feared object, they will consider
their fear to be unwarranted and accordingly

reduce their avoidance behavior. To test this
notion, paid volunteers who rated themselves as
fearful of snakes were shown pictures of

increasingly fearsome snakes as well as slides with
the word “shock” followed by shock stimulation.
One group received false heartbeat feedback
suggesting that their heart rates were unaffected
by snake stimuli but adversely affected by shocks.
A control group was exposed to the identical tape
recording except that subjects were informed that

they were hearing meaningless sounds rather than
their internal reactions amplified. All subjects
were then tested for approach responses toward a
snake.
In order to provide a meaningful test of the

cognitive labeling hypothesis it is essential to
preselect subjects who, in fact, display emotional
arousal and avoidance behavior toward snakes. It
is of little value, for example, to demonstrate that

subjects who do not fear snakes will perform
approach responses after being informed that they
are internally unaffected by pictures of snakes. On

the other hand, if misinforming fearfully avoidant
subjects that they are no longer internally affected
by conditioned aversive stimuli produces
significant reductions in avoidance behavior, then
cognitive labeling processes might serve as
contributory influences in desensitization
treatments. Because of the absence of objective

evidence concerning subjects’ initial fears of
snakes, this study provides no basis for judging
whether cognitive labeling is an irrelevant, weak,
or strong variable. Since 44 percent of the control
subjects successfully performed the terminal

approach task, it is evident that a sizable
proportion of the sample was completely fearless
to begin with. Other investigators have similarly
found that approximately 40 percent of subjects
who label themselves as fearful of snakes turn out

to be relatively fearless, much to their surprise,
when administered a behavioral test.
Contrary to the authors’ conclusion that
“cognitions about internal reactions are important
modifiers of behavior,” the heart-rate feedback
and the control group did not differ significantly in
approach behavior. When subjects who reported
that they had previously touched a snake were
excluded from the sample, a difference emerged;

but the effect cannot necessarily be attributed to
cognitive labeling, because the independent
variable in the reconstituted groups includes both
labeling and self-selection influences. A number of
unknown factors associated with accuracy of past

recall may enter in as plausible determinants.
Results of a second experiment also provide
limited evidence, because subjects’ initial
fearfulness was never objectively assessed, and
treatment effects were not evaluated by the

amount of snake approach behavior but by the
amount of money required to induce subjects

merely to touch a snake, a relatively weak
behavioral requirement.
Unlike desensitization, which eliminates fear
arousal through nonreinforced reexposures to
subjective threats, the cognitive approach
discussed above attempts to produce behavioral
changes by mislabeling existing emotional

responsiveness. Any fear reduction resulting from
deceptive feedback is apt to be short-lived unless
the mislabeling occasions genuine changes in
persons’ anticipatory arousal reactions. If our
assumption is correct that conditioned stimuli

generate emotional effects partly through an
intervening self-arousal mechanism, then persons
who are led to believe that they are no longer
frightened by threatening events may
subsequently reduce fear-arousing cognitions in

response to these situations and thus diminish
emotional responsiveness. A test of the self-

arousal theory would require measurement of
physiological and self-evaluative responses to
conditioned aversive stimuli prior to, and after,

cognitive mislabeling of internal states.
It should be remembered that cognitive claims
have been mined many times with disappointing
therapeutic yields. In the case of persons who

display relatively weak inhibitions, erroneous
explanations for physiological arousal to fear-

provoking situations may lower their fear to the
point where they can perform desired behavior. It
is doubtful, however, that strong fears and
inhibitions can be eliminated through either

mislabeling internal reactions or attributing them
to erroneous sources. A severe acrophobic, for
example, may be temporarily misled into believing

that his fear is no longer physiologically justified,
but he is likely to encounter unnerving internal

feedback when confronted with actual dreaded
heights. There is little reason to expect that
auspicious cognitions induced through deceptive

labeling can substitute for corrective learning
experiences in the stable modification of human
behavior.
Leitenberg, Agras, Barlow, & Oliveau (1969)

provide evidence that therapeutically oriented
instructions and social reinforcement may
enhance the favorable response changes

accompanying desensitization treatment. Snake-
phobic students who were administered this
procedure in the guise of an experiment on

visualization achieved some reduction in
avoidance behavior. The behavioral changes,
however, were much greater for subjects who had
been told that they were receiving a form of
therapy that is successful in reducing fears and

were praised for completion of items in the
hierarchy. These social variables would not
account for the success of self-administered
desensitization (Melamed & Lang, 1967; Krapfl,
1967) unless, as the authors suggest, self-observed
signs of progress assume a similar reinforcing
function.
TRANSFER OF EXTINCTION EFFECTS
In the standard desensitization procedure
emotional responses are extinguished to symbolic
representations of fear-provoking situations. The
treatment is not only directed toward attenuated

forms of actual threats, but a relatively limited set
of aversive stimuli is usually neutralized. Thus, for

example, a person with a widespread social phobia
may be desensitized to a dozen or so imagined
situations which cannot possibly encompass the

wide variety of interpersonal circumstances that
provoke anxiety. Under these conditions
extinction effects must generalize from thinking to
acting and to situations that may contain some
aversive elements which were never neutralized.
Results of laboratory studies reviewed in
preceding sections amply demonstrate that the
effects of symbolic desensitization do exert a
significant influence on behavioral functioning.
This is shown not only in improved performance

but also in the fact that the number of hierarchical
items successfully neutralized is positively
correlated with degree of reduction in phobic
behavior (Davison, 1968; Lang, Lazovik, &

Reynolds, 1965). However, the extent of
behavioral transfer is somewhat less than is
frequently claimed on the basis of clinical
observations.
Agras (1967) compared progress in
desensitization and reduction in GSR responses to
imagined test scenes with reports of performance

in the actual feared situation by a small group of
severely agoraphobic clients. Parallel changes

were obtained on all measures, but improved
performance in the real-life situations lagged

behind extinction of anxiety to symbolic stimuli.
However, no consistent relationship was found
between extinction of GSR responses to test items
and reported behavioral change. Hoenig & Reed

(1966) likewise found only partial
correspondence between degree of reduction in
phobic behavior and extinction of GSR responding
when the phobic cues were represented by word
labels, by imagination, and by the actual stimulus

objects. These results are somewhat at variance
with those reported by Rachman (1966), who
tested subjects’ self-reported anxiety to actual
phobic stimuli immediately after they had been
desensitized to imaginal representations of

identical situations. Immediate generalization of
anxiety reduction, as measured by fear reports,

was found in 82 percent of the tests.
Unfortunately, neither of the preceding studies
involves a systematic test of avoidance behavior.

Considering that avoidance behavior appears to be
influenced more by central than by autonomic
mediators, reliance upon GSR responsiveness as
the sole indicant of emotional arousal leaves much

to be desired.
The apparent discrepancies in findings

probably depend, in part, on differences in
severity of phobic disorders, on whether one
measures transfer in terms of improved

performance or decreases in subjective distress,
and on variations in the test procedures
themselves. In an experimental design in which
the same subjects are repeatedly tested with real
stimuli following neutralization of imagined

counterparts, any observed changes reflect the
combined effects of performance extinction and
symbolic desensitization. A precise appraisal of
generalization unconfounded by test-produced
changes would require desensitizing different
groups of subjects to different levels in the
hierarchy of anxiety-provoking stimuli and then
measuring their avoidance behavior in the actual
feared situation that corresponds to their highest

neutralized item. It would also be of interest to
investigate generalization systematically as a
function of degree of similarity between the
imagined stimuli that are desensitized and those
encountered in the real-life situation.
From knowledge of stimulus generalization

one would not expect symbolic desensitization
alone to exert vast transfer effects on instrumental
responding. The degree of generalization will

depend upon, among other factors such as
similarity of mediators, the number of stimulus
elements that different situations have in common.

The traditional desensitization procedure involves
too limited a sample of aversive stimulus
elements, and the threats are neutralized in too
attenuated a form to produce complete extinction
of emotional responses to threatening events
encountered in actual life circumstances. The
more that imagined treatment stimuli differ from

their actual counterparts, the greater the transfer
decrements. In accord with theoretical
expectation, the overall evidence of laboratory
studies (Agras, 1967; Bandura, Blanchard, &
Ritter, 1969; Davison, 1968) shows that symbolic

desensitization significantly reduces both
subjective distress and avoidance behavior, but
the number of approach responses that subjects
can perform behaviorally is generally less than the
number that have been successfully desensitized

in imagination. Moreover, new approach
responses are usually accompanied by relatively

high anxiety when first performed.
It has sometimes been erroneously concluded,
because of evidence that persons experience
anxiety while performing responses which have
ceased to be threatening in symbolic form, that
desensitization procedures do not achieve
behavioral changes through extinction of the

arousal potential of aversive stimuli. Since some
transfer loss is operative in symbolic
desensitization, this procedure is more likely to
produce anxiety decrements rather than complete
anxiety extinction. It has been repeatedly shown in

laboratory studies that control subjects manifest
undiminished anxiety when they perform their
pretest approach responses a second time,
whereas the same responses have lost much of
their anxiety-provoking capacity for matched

subjects after they have undergone desensitization
treatment. To the extent that emotional arousal is

reduced below the threshold which would activate
avoidance responses, people will be able to engage
in approach behavior, although with some residual

anxiety.
In clinical practice symbolic desensitization is
typically supplemented, either deliberately or
unwittingly, with other procedures that tend to

facilitate transfer effects. Symbolic desensitization
is most often combined with performance

extinction in which clients are urged to perform
formerly inhibited behavior in carefully selected
naturalistic situations as their fears extinguish to
equivalent imagined threats. Even though change

agents may not prescribe appropriate
performance tasks, most people nevertheless
eventually engage in approach behavior as their

avoidant tendencies gradually weaken through
treatment.
The desire to please the change agent and
others may induce individuals to venture fear-
provoking behavior. Positive social reinforcement
and other rewarding outcomes accruing from
successful performance of previously inhibited
activities may further extinguish any residual
anxieties. In some cases desensitization is also
supplemented by modeling procedures which, in

themselves, can produce substantial vicarious
extinction of emotional arousal. Thus, for example,
in the frequently quoted case by Jones (1924),
extinction of the boy’s animal phobia was achieved
not only by feeding him his favorite food in the

presence of gradually increasing anxiety-arousing
stimuli, but also by having him observe the
positive response of other children as they played
with the feared animal. Abrupt increases in
approach behavior were associated with each of

several modeling experiences.
In laboratory investigations, of course, these

various “extraneous” influences are intentionally
excluded. Because clinical outcomes are usually
obtained by diverse combinations of methods,
results are difficult to evaluate and to compare
with changes produced by single procedures
under laboratory conditions. Nevertheless,
findings bearing on transfer effects question the

wisdom of relying solely upon symbolic
desensitization for eliminating behavioral
inhibitions and conditioned anxiety. When such
methods are employed for clinical purposes they
should be supplemented with graduated

performance tasks, positive reinforcement of
approach behavior to overcome initial reluctance
of phobic persons to re-expose themselves to
feared situations, and modeling procedures to
further augment change in behavior. The use of

supplementary procedures to obtain more
consistent extinction outcomes will be discussed

more fully later.
COMPARATIVE EFFICACY OF
COUNTERCONDITIONING PROCEDURES
A number of experiments have been
specifically designed to compare outcomes of
desensitization with those obtained by other
methods of change. Paul (1966) reports a

methodologically sophisticated study in which he
compared, with appropriate controls, the relative
success of desensitization and interview
approaches for modifying debilitating

performance anxiety in college students who
experienced high distress in public speaking
situations. The students were initially
administered a series of personality
questionnaires measuring both generalized

emotional responsiveness and apprehension
about speaking before an audience. Those who

received high scores on the pretest measures
participated in a relatively stressful situational

test in which they were asked to deliver an
impromptu speech before an unfamiliar audience

including several clinical psychologists who, the
subjects were informed, would be evaluating their
performance. Immediately prior to the test
speeches the students’ pulse rates and palmar

sweating were measured; in addition, their actual
speech performance was rated in terms of

customary behavioral indicants of anxiety. On the
basis of these scores, students were randomly
assigned from stratified blocks to different
treatment conditions and control groups.
One group of students received insight-
oriented psychotherapy in which self-
understanding and insight into the psychological
determinants of their speech problems were

pursued through conventional interpretive
techniques. In order to assess the degree of change
resulting from effects of social interaction and
expectation of beneficial outcomes, a second group
of students was assigned to an “attention-placebo”
condition. During each session the latter subjects
were administered a placebo with strong
suggestions that the drug which they had received
effectively reduces the occurrence of anxiety in

stressful situations. Following the administration
of placebos subjects performed a supposedly
stressful task that, in fact, produced feelings of
drowsiness. For students in the
counterconditioning group, relaxation was

progressively associated with public speaking
items on a temporal anxiety hierarchy, graded
from reading about a speech two weeks before
presentation to delivering a speech before a large
audience. Students in a no-treatment control

group merely participated in all of the assessment
procedures.
The therapy was limited to five sessions
distributed over a period of six weeks. Five
practicing clinicians, who had considerable
experience in the use of insight-oriented
treatment approaches, administered each of the
three therapeutic procedures; this controlled for
possible variability stemming from differences in

therapists’ characteristics. Following completion
of the treatment series, the subjects’ degree of
physiological arousal, behavior dysfunctioning,
and self-reported distress were measured in the
threatening speech test situation; approximately

six weeks later the original set of personality
questionnaires was also readministered.
The percentage of students in each group who
exhibited decreases in emotional behavior of a

specified magnitude, as objectively assessed in the
standardized test situation, is summarized
graphically in Figure 7-5. Statistical analyses of a

variety of measures, including magnitude of
change as well as the percentage of cases
displaying decrements in emotionality, reveal that
subjects in all three treatment conditions showed
1436
Figure 7-5. Percent of subjects in each of the four conditions who displayed decreases in anxiety as measured by
behavior ratings, self-reports of emotional disturbance, and measures of physiological arousal. Drawn
from data of Paul, 1966.
significantly less overt behavior indicative of
anxiety, and reported less distress in the
situational test as compared to the no-treatment

control group. However, only the subjects
receiving the counterconditioning treatment
achieved a significant reduction in physiological
arousal relative to the controls. Additionally, the
counterconditioning group proved consistently

superior on all measures to subjects in the insight
and the attention-placebo conditions, which did
not differ significantly from each other. Follow-up
data similarly disclosed that the students treated
by means of counterconditioning reported

experiencing less anxiety related to giving
speeches than did students in either the other

treatment or the control groups.
It is also of interest that the therapists, who in
their regular clinical practice favored insight-
oriented methods, not only rated subjects treated
by the desensitization procedure as having
improved to a greater degree, but indicated a
significantly better prognosis for them as well.

Their prognostications were borne out by an
additional follow-up assessment (Paul, 1967), in
which all subjects were readministered the
personality tests approximately two years after
the formal experiment was completed.

Desensitization yielded the highest percentage of
subjects (85 percent) who showed decrements
(from pre-therapy ratings) in speech anxiety two
years later, followed by insight (50 percent),
placebo (50 percent), and non-treated controls

(22 percent).The corresponding percentages of
improvement on generalization measures of

interpersonal anxiety were 36, 25, 25, and 18
percent for the four groups, respectively. Not only
did desensitized cases maintain their gains over

time, but none showed increases in performance
anxiety or any evidence of substitute forms of
deviant behavior. These follow-up results,
although most impressive, should be accepted

with caution considering that they are based
entirely upon self-report measures.
The foregoing project was later extended (Paul
& Shannon, 1966) by administering a group
desensitization treatment to students, selected

from a waiting list, who, as controls of the earlier
study, had shown no reduction in anxiety during
the waiting interval. Students in this condition
participated in nine sessions during which their
emotional responses were neutralized to a

common hierarchy of public speaking situations.
An additional control group, matched for sex, age,
class, and equated on personality test scores, was
also included. In order to determine whether
elimination of speech fright improves academic
functioning, students’ grade point averages for the
semester prior to, and following, treatment were
obtained.
Group desensitization produced significant

reductions in self-reported anxiety with respect to
speaking and other interpersonal situations, and
increases in extroversion. Additional comparisons
involving these same measures show group
desensitization to be equally effective as

individual desensitization, but superior to the
insight and placebo treatments. Moreover,
students treated with group desensitization
showed a modest gain, whereas nontreated

controls suffered a substantial loss, in grade point
average. The latter finding is somewhat surprising
because one would not expect most academic

grades to be determined to any appreciable degree
by amount of public participation. The extensive
generalization of favorable changes revealed also
in the personality tests— assuming that they are
manifested in actual social behavior as well—may
occur because verbal communication figures
prominently in virtually all social and intellectual

activities. The noted improvements were
successfully maintained as revealed by a two-year
follow-up study (Paul, 1968a). Group differences
in academic performance provide even more
impressive testimony for the lasting benefits

accompanying desensitization treatment. Two
years after the project was completed, 90 percent
of the students who received group
desensitization had either graduated or were
completing their studies in good standing,

whereas 60 percent of the nontreated controls had
dropped out of school. The grade point averages

for students in the group desensitization and
control conditions in the follow-up semester were
3.5 and 2.4, respectively.
The beneficial effects of desensitization on
academic performance are further corroborated
by Mann & Rosenthal (1969) with elementary
school children. Compared to nontreated controls,

children who suffered from examination anxiety
showed significant changes in test anxiety and
reading achievement scores after receiving either
individual or group desensitization. Interestingly,
participant observers benefited to the same

degree as did direct recipients of the treatment
procedures.
Another comparative test of the efficacy of
counterconditioning methods for modifying

diverse forms of phobic disorders is provided by
Lazarus (1961), who employed an experimental
design in which the behavioral outcomes of group

desensitization were compared with those of
conventional group psychotherapy. The
experiment included acrophobics,
claustrophobics, cases of impotence, and clients
who each exhibited a different type of phobic
reaction. The participants were matched in pairs
on the basis of age, sex, and severity of the phobic

behavior, and were randomly assigned to
desensitization and interpretive treatment
conditions. In addition, a third group of phobic
clients, who received interpretive treatment plus
relaxation at the conclusion of each session, was

subsequently added in order to assess the possible
effects of relaxation per se on avoidance behavior.
The same person served as the therapist for all
three treatment conditions.
Only clients who exhibited severe phobic

behavior, as measured by actual behavioral tests,
were selected for the experiment. In order to

minimize the possible influence of preliminary
relationship experiences, the relevant anxiety
hierarchies were constructed from clients’ written
responses to questionnaire items, rather than
from personal interviews. The clients who
participated in the group desensitization were
treated in small, separate, homogeneous groups.

For the acrophobic clients, a common stimulus
hierarchy was constructed utilizing primarily a
physical proximity dimension beginning with a
scene in which a subject is looking down from a
height of about 10 feet, and terminating with

highly anxiety-provoking items. The
claustrophobic anxiety hierarchy represented a
stimulus continuum in which the degree of spatial
constriction and ventilation were varied
simultaneously from scenes depicting the client

“sitting in a large and airy room with all the
windows open” to “sitting in front of an open fire

in a small room with the doors and windows shut.”
Finally, the stimulus hierarchy items constructed
for the impotent men described progressively

intimate sexual situations ranging from sitting
close to a woman to pre-coital love-making in the
nude.
During desensitization sessions, a deep level of
relaxation was induced, following which the
participants were all asked to visualize the scene

which had been presented to them, and to signal
the psychotherapist whenever any stimulus item
proved disturbing. The rate and duration of

presenting the hierarchical items were paced
according to the most anxious group-member.
Clients who participated in the interpretive
treatment condition received a traditional form of
group psychotherapy which highlighted personal

exploration of feelings and interpersonal
relationships, permissiveness toward and
acceptance of emotional expressions, and the
development of insight into the origins and the
factors underlying their phobic disorders. These
clients received a mean of 22 sessions, the same
number as their matched counterparts in the
desensitization groups.
The therapeutic outcomes for acrophobics and

claustrophobics were objectively assessed by
situational tests administered one month
following the termination of the treatment
sessions. The acrophobics were required to climb
a fire escape to a height of approximately 50 feet,

then to accompany the experimenter in an
elevator to the roof of the building eight stories
above the street, from where for two minutes they
were to count the automobiles passing below. A

similarly rigorous objective criterion was adopted
as evidence that claustrophobic reactions were
successfully extinguished: The subjects were

required to remain in the cubicle with the movable
screen a few inches away without experiencing
any disturbance for a period of five minutes. A
second observer was present during the
situational tests to provide a reliability check on
whether the client had successfully met the
tolerance criteria. For obvious reasons, the

impotent men were not subjected to a situational
test, nor was the behavior of the four clients in the
mixed group objectively measured. The
improvement rates in these cases were based on
reports of significant behavioral change. Although

capacity performance tests are less subject to
extraneous influences than self-report measures,
results of this study would have been more
definitive had the assessments been conducted by
a tester who had no knowledge of the conditions

to which the subjects were assigned rather than by
the therapist himself.
Phobic behavior was completely extinguished
in 13 of the 18 clients who received
desensitization treatment, whereas the
interpretive and interpretive-plus-relaxation
treatment successfully modified the phobic
responses in only 2 out of 17 cases. Additional
evidence for the effectiveness of

counterconditioning is provided by the finding
that 10 of the 15 clients whose phobias were
essentially unmodified by the interpretive
procedures were successfully treated by group
desensitization within ten sessions. A follow-up

study conducted at varying intervals after the
termination of the treatment program revealed
that 80 percent of cases who were successfully
treated by means of counterconditioning
procedures maintained their behavioral changes,

according to a stringent criterion in which even
the recurrence of weak phobic responses was

rated as a relapse. Gelder & Marks (1968)
similarly found that a group of phobic clients who
had not responded to 18 months of group

psychotherapy showed significant reductions in
phobic behavior after a few months of
desensitization treatment.
Although most of the laboratory studies
discussed in earlier sections were primarily
designed to isolate the contribution of component

variables in the desensitization procedure, their
results nevertheless have some bearing on the
efficacy of this approach. In evaluating outcomes a

distinction should be drawn between behavioral
improvement and complete elimination of
avoidance behavior. The findings generally
disclose that a relatively brief program of
desensitization involving five to ten sessions

produces improved performances in virtually all
participants. In the study by Bandura, Blanchard,
& Ritter (1969), for example, 90 percent of the
subjects who received desensitization treatment
displayed increases in approach behavior that
exceeded the performances of their matched
nontreated controls. However, depending on the
stringency of the criterion test, only between 30
and 50 percent of the subjects achieve complete

extinction as evidenced by their ability to perform
the terminal approach tasks in behavioral tests
(Bandura, Blanchard, & Ritter, 1968; Davison,
1968; Lang, Lazovik, & Reynolds, 1965; Schubot,
1966). It should be noted, of course, that these

improvement rates are based on very brief, time-
limited therapies in which all subjects are tested
after several treatment sessions regardless of the
number of hierarchical items to which they have
been desensitized. The actual therapeutic limits of

this particular form of counterconditioning can be
best established by studies in which behavioral

improvement and complete extinction rates are
objectively measured after subjects have been
thoroughly desensitized to the entire set of

anxiety-arousing stimuli. In evaluating the efficacy
of treatment procedures the incidence of terminal
performances should be of major concern to
discourage the development of complacency about

methods that consistently achieve performance
gains but leave many of the participants

behaviorally incapacitated to some degree.
There are a number of clinical reports that
present outcome data in the form of therapists’

judgments of their success rates. Wolpe (1958)
and Lazarus (1960, 1963b) state that between 75
and 90 percent of the clients whom they have
treated were “markedly improved or completely
recovered.” Hain, Butcher, & Stevenson (1966)

report that desensitization was effective in 78
percent of the cases and that improvements often
occurred in areas of occupational, sexual, and
social functioning beyond the specifically treated
phobias. Follow-up studies, conducted at intervals
ranging from six months to several years after the
termination of treatment, with few exceptions
disclose that clients not only maintain their gains,
but many display additional positive changes in

behavior. In another paper Wolpe (1964)
reaffirms the efficacy of desensitization therapy
for modifying complex behavior disorders.
Somewhat less favorable outcomes than those
given above have been published by Cooper

(1963) and Marks & Gelder (1965) in
retrospective comparisons of clients treated by
“behavior therapy” and “psychotherapy” made in
terms of general judgments of improvement rates

from case notes. In a spirited rejoinder Wolpe &
Lazarus (1966) discount the discrepant results as
due to “the fledgling efforts of novices who have

learned the rudiments of systematic
desensitization [p. 159].” Conflicting data of this
sort are not at all surprising as long as they are not
erroneously considered as measures of behavior
outcome but are understood instead as differences
between therapists’ judgmental responses (which
rarely correlate perfectly with clients’ actual

behavior functioning). Indeed, one would expect
diminishing correspondence between actual
behavior and subjective ratings as one moves from
objective measures of clients’ behavior to their
own self-assessments, from clients’ verbal reports

of performance changes to therapists’ judgments
of improvement, from therapists’ inferences based
on clients’ self-reports to information that
happens to get recorded in case notes, and from
case notes of undetermined reliability to

retrospective global ratings made by still another
set of judges who never had any contact with the

client. Major differences in the types of case notes
kept by therapists of behavioral and
psychodynamic persuasions further preclude any

meaningful comparison between success rates.
Much progress can be made in reducing fruitless
controversies about the relative superiority of
rival methods by abandoning the outcome

numbers game in which therapists’ judgments of
their clients’ verbal reports of their behavioral

changes are evaluated against a legendary
baseline figure of two-thirds improvement.
These types of quasi-outcome data have, at

best, only suggestive value. Apart from the
subjective and general nature of clinical ratings,
the therapeutic interventions are exceedingly
varied, including assertion training, graded
reexposure to feared situations, anxiety-relief

procedures, aversive counterconditioning, role
playing, symbolic desensitization, verbal
prompting of desired response patterns, social
reinforcement of behavioral changes, and a host of
unrecognized treatment factors a: well as a variety
of unmeasured environmental influences. It is
consequently impossible to identify which
treatment variables are responsible for observed
changes, even if the outcome figures were valid

and the behavioral modifications were
attributable to the therapeutic interventions. Of
much greater concern, however, is the fact that
subjective evaluations of treatment outcomes lend
themselves readily to unwarranted claims of

efficacy by their proponents and to premature
rejection of potentially promising approaches by
their theoretical rivals.
Proof of the efficacy of procedures of treatment

must rest on objective assessment of behavioral
changes and evidence of lawful covariation
between specific learning variables and

designated outcomes, rather than on general
judgments of improvement in “neurotic illness,”
achieved through the use of varied combinations
of treatment procedures. Laboratory experiments
and controlled individual studies involving well-
defined manipulations of treatment variables and
objective measurements of behavioral changes

will eventually yield reliable information
regarding the critical parameters in
counterconditioning methods. Research of this
kind should not only greatly increase
understanding of conditioning processes, but it

also provides the basis for refinements in
treatment procedures. The laboratory studies
reviewed earlier represent an encouraging
advance in this direction.
Innumerable single case reports are of interest

not because they validate anything but because
they illustrate how counterconditioning principles

can be applied to exceedingly diverse anxiety
disorders. Most of these studies include detailed
reports, often independently verified, of the
modifications achieved in clients’ behavioral
functioning. However, in treatments involving the
combined use of different procedures,
desensitization is obviously not the sole

determinant of observed changes. It is also
important to bear in mind that case reports may
convey an overly favorable impression of the
efficacy of a given treatment method because of
selective publication of successful cases.
APPLICABILITY OF DESENSITIZATION TREATMENTS
Desensitization procedures have been utilized
to extinguish countless circumscribed, but

partially incapacitating, phobias including
avoidance of places and activities that might result
in contact with feared animals, birds, reptiles, and
insects (Clark, 1963; Cooke, 1966; Friedman,
1966; Ramsay, Barends, Brenker, & Kruseman,
1966); fearful avoidance of automobiles, airplanes,
and other types of transports (Kraft & Al-Issa,
1965a; Lazarus, 1960; Rosenthal, 1967; Wolpe,
1962); fear of water (Bentler, 1962); storms
(Costello, 1963); aversion to heat that prevented

the person from washing in warm water and from
drinking or eating hot foods (Kraft & Al-Issa,
1965b); and dread of atomic disasters that
resulted in avoidance of radios, television, movies,

newspapers, conversations, and other forms of
communication that might bring news of
international disharmony (Ashem, 1963); school
phobias (Chapel, 1967); anxiety reactions to
hypodermic injections and the use of sanitary pads

(Rachman, 1959), and to hospitals and
ambulances (Lazarus & Rachman, 1957);

hyperesthesia of taste and touch (Beyme, 1964);
autonomic disorders (Cohen & Reed, 1968); and

persistent apprehensions about illness, physical
injury, and death (Rifkin, 1968; Wolpe, 1961).
In addition to isolated phobias, desensitization
methods have been applied to the modification of
pervasive behavioral dysfunctions resulting from
incapacitating obsessions and compulsions

(Haslam, 1965; Walton & Mather, 1963b), from
claustrophobias and agoraphobias, from
articulatory disorders (Gray, England, & Mohoney,
1965; Lazarus & Rachman, 1957; Rosenthal, 1968;
Walton & Mather, 1963a), from recurrent

nightmares (Geer & Silverman, 1967) and
insomnia (Geer & Katkin, 1966), and from chronic
alcoholism (Kraft & Al-Issa, 1967a). Finally,
complex interpersonal problems have been
eliminated by extinguishing clients’ anxieties

associated with sexual intimacy, aggressive and
hostile behavior, close social relationships, social

disapproval and rejection, failure to meet external
or self-imposed achievement demands, and fear of
persons in positions of authority (Hain, Butcher, &

Stevenson, 1966; Kraft & Al-Issa, 1967b; Madsen &
Ullmann, 1967; Wolpe, 1958; Wolpe & Lazarus,
1966).
It is commonly assumed, especially in
psychiatric theorizing, that grossly deviant
behavior is primarily a function of biochemical

determinants, whereas deviations of lesser
magnitude are governed by experiential factors.
Granted that physiological variables can

contribute significantly to behavioral variability,
such evidence does not justify a dichotomous
theory of psychopathology, particularly in view of
the absence of any objective criteria as to where
the line of demarcation should be drawn between

so-called “neurotic” and “psychotic” response
patterns. In keeping with the dichotomous thesis,
Wolpe (1958) questions whether schizophrenics
can profit from desensitization therapy. This
position implies, among other things, that persons
who are considered to be psychotic are incapable
of classical conditioning. Contrary to this view,
laboratory studies provide evidence that
schizophrenics not only exhibit emotional

conditionability (O’Connor & Rawnsley, 1959;
Vinogradov, 1962), but apparently they differ
little, if at all, in this respect from groups judged to
be normal (Howe, 1958; Spence & Taylor, 1953).
Favorable outcomes have been achieved in the few

instances where counterconditioning procedures
were applied to emotional behavior exhibited by
individuals diagnosed as schizophrenic (Cowden &
Ford, 1962; Zeisset, 1968).
The ease with which symbolic desensitization

is achieved will depend in part on secondary
factors such as clients’ attentiveness,

cooperativeness in visualizing verbally presented
scenes, and facility in inducing anxiety-inhibiting
responses, in addition to variations in their
conditionability. Cowden & Ford (1962), for
example, encountered considerable difficulty in
persuading their clients to carry out relaxation
exercises. It is possible, therefore, that grossly

deviant personalities who present severe
cognitive deficits, inadequate attending behavior,
and irrelevant associations to verbal stimuli,
would prove less responsive to exclusively
symbolic forms of counterconditioning therapy.

Instances in which these types of limiting factors
are operative may require, at least in early stages
of treatment, greater reliance upon graduated
reexposure to actual threats along with stress-
reducing stimuli that can be externally controlled.
The mere presence of anxiety and avoidance
behavior does not necessarily mean that

conditioned emotionality is the central problem. A
person who lacks requisite behaviors for coping
effectively with the social, intellectual, and
vocational demands of his environment will be
repeatedly subjected to punishing experiences.
Under these circumstances, feared situations are,
in fact, aversive and hence the emotional arousal is

not unrealistic. In cases of this type behavioral
deficits constitute the major problem, whereas the
emotional component is a secondary consequent.
A treatment devoted solely to extinguishing
emotional responses would be at best a

temporarily effective means of producing a
relaxed incompetent. On the other hand, a
treatment that established behavioral
competencies would substantially decrease the
punitiveness of the client’s social environment and

thus achieve stable reductions in fearfulness.
A response induction program may be

necessary even when anxiety disorders do not
originate in behavioral deficits. Persons who suffer
from unrealistic or disproportionate fears tend to
avoid engaging in fear-provoking activities. This
often results in a spiraling process where fear and
avoidance prevent further elaboration of
interpersonal skills, and marginal competencies, in

turn, render threatening situations even more
anxiety-arousing. In most cases, therefore, a
combined treatment aimed at extinguishing
unwarranted fears and at instilling capabilities
would yield best results.
Since removal of unwarranted avoidance

behavior enables people to participate in
potentially rewarding activities, approach
responses, once they have been restored, are likely

to be effectively maintained by their favorable
consequences. However, the initial behavioral
changes are sometimes impeded by the existing

advantages derived from disabling phobias. As a
result of their phobic conditions, people may gain
exemptions from certain responsibilities, they may
reduce unpleasant vocational and familial
demands, and they may achieve considerable
control over the behavior of others. Where such
hindering contingencies exist, unless the rewards

that contribute to the maintenance of avoidance
behavior are withdrawn, desensitization alone is
unlikely to bring about much behavioral
improvement.
Identification of the Stimulus Determinants

of Emotional Behavior
Procedures that effect behavioral changes
primarily through response consequences
ordinarily do not present any major diagnostic
problems. An agent of change who has at his

disposal an effective means for evoking the
desired behavior and sufficiently powerful
incentives can, given adequate control over the

environment, achieve substantial behavioral
modifications by overriding undetermined
maintaining conditions. In contrast, stimulus-
oriented treatments, especially in their present
stage of development, require considerable
diagnostic ingenuity. In order to achieve any

measure of success with classical extinction
procedures the primary stimulus determinants of
emotional behavior must be accurately identified
and neutralized. In current practice the selection
of anxiety sources is based upon informally

collected data, from interviews, case histories, and
various personality tests, most of which were
originally constructed for entirely different
purposes. Although no reliability studies have
been conducted in which different therapists

select from the same protocols what they consider
to be the critical sources of anxiety, it would come

as no surprise to find low consensus, particularly
in cases involving multiform problems.
If the efficacy of counterconditioning methods
is to be maximized, the present informal
assessment approaches must be replaced by more
efficient and reliable procedures. In many cases, of
course, the eliciting stimuli are readily identifiable.

In others, however, the occurrence of emotional
responses is dependent upon stimuli that lack
distinctiveness, or upon particular patterns and
sequences of events which are difficult to discern.
However, individuals are rarely continuously

anxious; rather, they tend to display such
reactions only at certain times and under
particular circumstances. Covariations between
stimulus events and emotional responding can,
therefore, be best identified through careful

analysis of regular variations in the onset and
magnitude of emotional behavior. After the

pattern in anxiety responses has been determined,
one can isolate common features in situations in
which the behavior typically occurs. The major

controlling stimuli cannot always be identified
solely through systematic examination of the
objective characteristics of environmental events
which may, in fact, be highly dissimilar. Rather, the

common determinants are often revealed in
detailed accounts of the thoughts and subjective

reactions that clients experienced in anxiety-
producing situations.
New assessment approaches are particularly

needed for isolating stimulus determinants when
only gross covariant relationships are noted.
Information of this kind can be obtained by
systematic behavioral measurement of emotional
reactions, both in the presence and in the absence

of specific contextual and social cues which appear
to be regularly correlated with variations in the
observed anxiety responses. Because of the
countless and complex varieties of learning
histories represented by clinic populations, a
highly flexible stimulus exploration procedure is
required. When utilization of physical events is too
cumbersome and impractical, suspected fear-
provoking situations can be easily presented in

verbal or pictorial forms. Emotional responses to
these potential threats might be measured
behaviorally, physiologically, through verbal
reports, or a combination of these methods.
Before turning to other issues of stimulus

specification, it is necessary to clarify several
common misconceptions about the range of
applicability of counterconditioning procedures.
In most polemical discussions of psychotherapy,

behavioral and psychodynamic approaches are
usually presented as rival methods of treatment
suitable for different types of anxiety conditions.

Advocates of psychodynamic methods typically
assume that desensitization is essentially limited
to simple “monosymptomatic” disorders under the
control of clear-cut stimuli, whereas
psychodynamic procedures are suitable to more
complex and pervasive anxiety problems.
Behaviorally oriented therapists, on the other

hand, contend that desensitization procedures are
applicable to any dysfunction involving anxiety,
and view psychodynamic approaches as having
proved ineffective in modifying either simple or
complex conditions.
Greatest progress would be achieved in

developing efficacious treatment approaches if
these ill-defined partisan labels were retired from
further use. Much time has been spent fruitlessly

in attempts to define what constitutes “behavior
therapy” and “psychotherapy.” A more productive
and less confusing approach to the understanding

of social influence processes is to focus on the
basic mechanisms through which behavioral
changes are produced. These mechanisms are
undoubtedly brought into play to varying degrees
by conditions created either deliberately or
unwittingly by change agents in influence
attempts arbitrarily designated behavior therapy,

psychotherapy, counseling, reeducation, or some
other appellation. In each of the foregoing
enterprises change agents model certain attitudes
and response patterns. One might, therefore,
analyze these various activities in terms of the

behavior the change agents are modeling, its
functional value for the recipients, and the extent
to which conditions that facilitate modeling are
present. Similarly, one might examine
reinforcement influences, which are operative in

all social situations, to determine what behavior is
being reinforced, with what frequency, and by

what means in different systems designed to
modify psychological functioning.
Because of the ambiguities and erroneous
impressions associated with the terms “behavior
therapy” and “psychotherapy,” their continued use
beclouds most of the fundamental issues in this
field. For example, the rate at which a given

behavior can be extinguished depends upon the
nature of its maintaining conditions rather than
upon whether it is “monosymptomatic” or part of
a multiform problem. In fact, many so-called
simple disorders, such as tics, that are

immediately self-reinforcing because their
occurrence is tension-reducing, are unusually
resistant to change. Therefore, some apparently
simple behaviors fail to respond well to treatment,
whereas many complex disorders are readily

modifiable (Lazarus, 1963a; Meyer & Crisp, 1966).
Moreover, whether emotional responses are

limited or diffuse is often determined by the
pervasiveness of emotion-arousing cues rather
than by conditions existing within the individual.

Those for whom snakes are threatening will
exhibit a highly circumscribed phobic disorder
because reptiles are rarely or never encountered
in urban settings. By contrast, when anxiety has

been conditioned to stimuli that appear in a
variety of frequently encountered situations,

persons will experience pervasive or “free-
floating” anxiety (Wolpe, 1958). Similarly, if our
unperturbed snake-phobics were to become
residents in a reptile-infested locale, they too

would exhibit diffuse and disabling anxiety
disorders.
Discussions of desensitization often create the
impression that this procedure is principally

applicable to stimulus events that are easily
specifiable and nonsocial in character; whereas
anxieties arising from aggression, dependency,
sex, and other interpersonal sources are made to
seem the exclusive domain of psychodynamic
approaches. In actuality, desensitization methods
are not restricted, either on theoretical or practical
grounds, to any particular set of emotion-arousing
stimuli. Indeed, a counterconditioning form of

treatment could be employed to neutralize the
negative valence of Oedipal fantasies in clients for
whom this might constitute a problem.
It would perhaps be more accurate to say that
the applicability of desensitization treatment is

mainly limited by therapists’ ingenuity in
identifying sources of anxiety, particularly when
the crucial stimulus determinates are obscure.
This task is complicated by the absence of any

objective criteria for determining the appropriate
events for treatment. Let us consider, for example,
a female agoraphobic who is unable to venture

outside the household. Should one desensitize her
to progressively farther anxiety-arousing
excursions from the home? One might argue that
her phobic behavior arises from a morbid fear of
sexual encounters, apprehensions about
abandonment and helpless exposure to crowds, or
some other sources, and it is these contents that

must be emphasized in the treatment. To take
another example, should a snake-phobic be
desensitized to progressively closer interactions
with reptiles or to genital concerns on the
assumption that “the sight of snakes provokes

penis emotions [Fenichel, 1945, p. 48]”?
Laboratory studies of desensitization furnish some
evidence that claustrophobia can be successfully
eliminated by neutralizing individuals’ emotions
to cues of increasing space constriction (Lazarus,

1961) without focusing on their fears of being “left
alone with dangerous impulses and fantasies

[Cameron, 1963, p. 286]”; acrophobics have lost
their fear of heights through utilizing elevation
hierarchies (Lazarus, 1961) rather than fears of

“falling in self-esteem” or “self-destructive
impulses [Cameron, 1963, p. 280]”; and countless
snake-phobics have been cured by being
desensitized to reptile rather than phallic stimuli.

In view of these interesting findings, it would be
most instructive to compare the degree to which

phobic behavior is extinguished when
desensitization is directed toward either the
phobic stimuli themselves or the hypothesized
internal threats. A desensitization procedure

combined with objective measurement of changes
in avoidance behavior provides an excellent
means of testing different theories regarding the

stimulus determinants of emotional response
patterns.
As illustrated in the preceding examples,
psychodynamic formulations assume that
anxieties are internally generated by arousal of
unconscious impulses which are then displaced
and projected onto environmental objects.
External phobic cues are therefore regarded as
pseudoevocative stimuli. Contrary to this
interpretation, successful neutralization of

emotional responses to phobic stimuli not only
produces stable decreases in avoidance behavior
without the emergence of new deviant responses,
but it is often accompanied by reductions in
anxiety in other areas of functioning (Bandura,

Blanchard, & Ritter, 1969; Lang, Lazovik, &
Reynolds, 1965; Paul, 1967). The latter findings
make the view that avoidance behavior is
controlled by displaced and projected stimulus
valences appear of questionable validity, or

alternatively, suggest that neutralization of
external projective stimuli is one of the most

powerful means currently available for
extinguishing the arousal properties of
unconscious internal events.
The above discussion is not meant to imply
that stimulus determinants are always inferable
from the content of deviant responses. In many
cases emotional behavior is under multiple

stimulus control in which some of the evocative
cues, because of peculiar conditioning histories,
may be thematically remote. Also, complex social
behaviors characteristically depend upon
interrelated activities, each governed by

somewhat different stimuli. A given performance
may consequently be inhibited or disrupted by
anxiety arising from thematically different
component functions that are not readily evident.
The operation of these more intricate stimulus

determinants is best illustrated by applications of
desensitization procedures to the modification of

diverse sexual disorders.
According to Bond & Hutchison (1960) the
most frequent classes of exposure-eliciting stimuli
for sexual exhibitionists are stress experiences
provoking inadequacy and females who possess
certain physical characteristics that have been
endowed by the exhibitionist with unusually high

sexual valence. The authors therefore employed
both sexual and devaluation hierarchies in
treating a 25-year-old male who presented a long
history of persistent genital exposure leading to
24 charges of indecent exposure including eleven

prison convictions. The client had undergone a
variety of treatments without benefit, including
individual and group therapy, carbon dioxide
abreaction therapy, moralistic exhortations under
hypnosis; finally, in desperation, he resorted to a

specially designed chastity belt that his wife
locked in the morning and unlocked at night. Even

these physical restraints failed to control the
client’s behavior as he was once again arrested for
indecent assault as he attempted to grasp the legs

and breasts of a young woman while wearing his
chastity belt.
Three hierarchies of exposure-provoking

stimuli were constructed for the desensitization
treatment. One set of stimuli was graded on the
basis of the age and physical appearance of the

females, ranging from older women who
minimally provoked exposure to young attractive
females. These stimuli were presented in each of

four settings in which genital exposure frequently
occurred (i.e., in department stores, on beaches, on
sidewalks, and in automobiles). In addition, a
separate hierarchy was constructed on the basis of
washroom situations, since they served as the

most potent contextual stimuli for exhibitionism.
The third stimulus dimension contained social
situations giving rise to feelings of inadequacy.
These sets of eliciting stimuli were then
progressively paired with hypnotically induced
relaxation over a period of 30 sessions. The client
was also instructed to practice relaxation and to
initiate this chain of responses by the word “relax.”
As treatment progressed the client became less

emotionally aroused by provocative females, his
exhibitionistic urges and sexual fantasies
diminished in frequency and intensity, and he
displayed increasing voluntary control over his
exposure behavior on occasions when he

experienced some degree of emotional arousal. As
the client made continued improvement, he was
able to participate in group activities involving
close heterosexual contacts without experiencing

any tension or urges to expose himself. He showed
no exhibitionistic behavior for a period of 13
months following the termination of therapy

(Bond & Hutchison, 1964). Subsequently, the
client exposed himself on a few occasions to
women in washrooms in response to severe
financial and vocational stresses, but provocative
females in public places such as parks, streets, and
department stores no longer elicited sexual
exhibitionism.
It is of interest that in the above case deviant
sexual behavior was partly controlled by
nonsexual stress events. Wolpe (1958) similarly
reports the successful treatment of a pharmacist
who suffered from impotence by desensitizing him

to essentially nonsexual cues. This client, who had
experienced satisfactory sexual relationships with
several different girls, suddenly became impotent
when he attempted intercourse with a virgin

girlfriend who yielded reluctantly to his insistent
pressures. During this unsuccessful seduction the
client found himself thinking about a disturbing

childhood event in which he overheard his parents
having intercourse; the mother’s protestations and
weeping had apparently succeeded in conditioning
aggressive and brutal meanings to the sexual act.
This conditioning was undoubtedly facilitated by
the marked hypersensitivity to aggressive and
physical injury cues that the boy exhibited long

before witnessing the parental sexual episode. By
the time the client had entered psychotherapy he
was sexually incapacitated, not only by
generalization from the parental situation to the
virginal girlfriend, but also by a secondary

generalization from her to a subsequent girlfriend
who bore a physical resemblance to her. On the
basis of a learning analysis of the impotence, the
anticipated pain and physical injury cues
occurring during defloration were considered to

be the critical events producing the sexual
inhibitions. The client was therefore desensitized

to several stimulus dimensions involving physical
injury, vocalization of pain and suffering, and
violent verbal interchanges resembling the

parents’ endless arguments. Complete sexual
responsiveness was restored after aggressive and
pain cues lost their capacity to evoke anxiety.
Further suggestive evidence that in some cases
inhibition of heterosexual behavior may be
maintained primarily by anxiety attached to social

and physical contact cues rather than to sexual
stimuli per se is provided by Stevenson & Wolpe
(1960) in the treatment of a pedophile and two

homosexuals. All three clients displayed markedly
passive, submissive, and withdrawn behavior that
apparently stemmed from authoritarian control by
tyrannical parents. As a consequence of this
aversive social training a broad class of

interpersonal responses was inhibited except
toward little girls in one case, and toward
nurturant nonthreatening male companions in the
other two. The treatment strategy in all three
cases consisted essentially of training in socially
assertive behavior. The fact that these clients
readily adopted exclusively heterosexual patterns
of behavior and experienced them as positively
reinforcing after assertive responses had been

developed suggests that the deviant sexuality was
a function of generalized interpersonal anxieties,
rather than of specific sexual origin.
The covariant relationship between
assertiveness and sexuality noted in the preceding

cases is convincingly demonstrated by Kahn
(1961) in a laboratory experiment with
infrahuman subjects. Two groups of mice were
trained in either socially aggressive or submissive

behavior. Both groups were then tested for sexual
responsiveness toward virgin females known to be
in estrus. The social training in aggression and

submission had a striking differential effect on the
mating responses of the males. Whereas the
aggressively trained animals immediately pursued
the females, copulated with them, and remained
sexually active throughout the test session, not a
single submissive animal did so. On the infrequent
occasions when the submissives approached their

female companions, they did so very hesitantly
and then quickly reverted to the avoidant
response pattern.
The findings of the latter experiment may be
interpreted in several ways. To the extent that

aggression is an important component of sexual
behavior (Ford & Beach, 1951), then any increase
or inhibition of aggressive responses would be
associated with a corresponding enhancement or

inhibition of sexuality. Since, however, the animals
failed to initiate any sexual responses, including
preparatory ones, any anxiety-evoking stimuli

accompanying either aggressive or sexual
behavior could not have been generated in the test
situation. Therefore, a second and perhaps more
likely explanation of these findings is that the
previously acquired physical contact anxieties
motivated and reinforced generalized avoidance
responses that precluded the occurrence of any

sexual behavior, even if such responses were only
weakly inhibited.
The discussion thus far has illustrated how
anxiety arising from nonsexual sources can
control the inhibition of appropriate heterosexual

behavior. Actually, it is possible to delineate
several different stimulus determinants of sexual
deviance, each requiring a somewhat different
treatment strategy. First, there is the behavioral

syndrome, to which reference has already been
made, wherein social and physical contact arouse
anxiety reactions, but sexual stimuli per se may be

positively valenced. In these cases, a program of
treatment utilizing modeling and reinforcement
procedures designed to foster and disinhibit
interpersonal approach tendencies is likely to
result in a corresponding enhancement of
heterosexual behavior.
A second frequently encountered pattern is

one in which a person has little or no
apprehension about close social interactions, but
sexual cues, particularly those associated with
coital performance, are negatively valenced.
Impotence, frigidity and other specific sexual

inhibitions are common complaints associated
with the latter syndrome. For problems of this
sort, the neutralization of sex-related stimuli,
through some form of desensitization procedure,

would be the method of choice. An example is
provided by Lazarus & Rachman (1957), who
successfully treated a case of impotence of recent

origin by eliminating anxiety to pre-coital scenes.
In instances where the sexual inhibitions are less
strongly established, impotence can be effectively
modified by having the client follow a self-
administered program of desensitization in actual
sexual situations (Wolpe, 1958). The latter
procedure is described more fully in a later section

of this chapter.
Individuals who are more severely
incapacitated because they respond with strong
anxiety to close interpersonal contact as well as
specific cues would benefit most from a treatment

program combining both desensitization of
interpersonal and sexual cues and training in
interpersonal competencies.
Finally, it should be realized that deviant

sexual behavior is sometimes maintained by
substantial positive reinforcement, whereas
anxiety mechanisms, which may have figured
prominently in the genesis of the disorder,
currently play a minor role. Many homosexual
patterns of behavior are, in fact, sustained by the
positive rewards deriving from homosexual
alliances (Henry, 1941; Hooker, 1961). If the
person should desire to change his sexual

orientation, differential reinforcement procedures
designed to reduce the strong positive valence of
homoerotic stimuli and to enhance responsiveness
to heterosexual cues would constitute the

appropriate therapeutic strategy in such cases.
These methods are discussed in some detail in the
next chapter.
The necessity for identifying the controlling
stimuli and determining the functional value of
behavioral dysfunctions before selecting the
method of treatment is well illustrated in a clinical

study reported by Lazarus (1963b). Sixteen
women with chronic frigidity were administered

the standard desensitization procedure on the
assumption that frigidity represents conditioned

avoidance maintained by sexually generated
anxiety. The desensitization was conducted along
stimulus dimensions of increasing intimacy and
other sexually inhibiting stimuli that were unique

to each case. Significant improvements in sexual
relationships, as reported by the women and
corroborated by their husbands, were achieved in
9 of the 16 women for whom frigidity appeared to
be determined by sexual anxieties. The majority of

the others, most of whom displayed intense and
generalized hostile attitudes toward men,
terminated therapy after several interviews. This
subgroup of women evidently required a
treatment program aimed at reducing hostile

behavior. Where hostility toward men results
from a sense of inadequacy and submission, a

program of assertion training would not only
reduce exploitation, which a submissive person is
likely to incur, but at the same time increase

feelings of self-esteem and self-worth. In addition
to establishing new response orientations toward
men through modeling and positive
reinforcement, desensitization procedures might

be employed to decrease excessive hostility to
inevitable evocative situations. To achieve greatest

gains it may also be necessary to reduce hostility-
generating behavior on the part of the husband.
ANXIETY RESPONSE DECREMENTS AS A FUNCTION

OF NEUTRALIZING PRIMARY AND
GENERALIZATION STIMULI
The introductory chapter discussed how
emotional responses elicited by a particular
stimulus tend to generalize spontaneously to a

wide variety of cues falling on the same physical or
semantic stimulus dimension. It has also been
demonstrated in numerous laboratory studies
(Bass & Hull, 1934; Hoffeld, 1962; Hovland, 1937)
that extinction effects also generalize to stimuli at
all points on the generalization gradient. Whether
extinguishing emotional responses to the primary
conditioned stimulus is more effective in
neutralizing the entire range of similar cues than

desensitizing the individual to generalization
stimuli represents a question of considerable
theoretical and practical importance.
Many theories assume that stable and

widespread behavioral changes will result only if
the prototypic conditioning involving the primary

stimulus object is modified. Consequently, a
considerable amount of time is typically devoted
to diagnostic exploration and reconstruction of the
client’s social history before any therapeutic
interventions are attempted. If it were found that

neutralizing a generalization stimulus had about
as much effect on the extinction gradient as would

result from extinguishing emotional responses to
the original conditioned stimulus, then it would

make little difference at which point on the
stimulus continuum the change agent began the
counterconditioning process. Unfortunately,
laboratory experiments necessary to resolve this

issue have not yet been conducted. A few
investigations have been reported, however, in
which the magnitude of emotional responses to
the CS originally employed in the establishment of
conditioned responses is assessed as a function of

extinguishing emotional reactions to
generalization stimuli located at various distances
from the CS. Findings from these studies (Bass &
Hull, 1934; Hoffeld, 1962; Hovland, 1937)
consistently demonstrate that neutralizing any

relevant stimulus, whether adjacent to, or remote
from, the CS, has the effect of reducing somewhat

the emotional responsiveness to cues at all points
on the generalization gradient. The anxiety
decrement, however, becomes progressively

smaller the farther the test stimuli are removed
from the stimulus selected for extinction.
The foregoing findings suggest that positive

outcomes can be achieved to some degree by
deconditioning any stimulus possessing anxiety-
arousing properties, but that the greatest benefits

will be derived from focusing on the particular
events that the change agent wishes to neutralize,
regardless of whether they constitute the original

or the generalized stimuli. On the basis of these
findings there is no reason to expect that
desensitizing the primary conditioned stimulus
would have a more widespread effect on
generalization cues than neutralizing the latter

stimuli directly. Thus, for example, if a given
individual’s anxiety responses to persons in
authority primarily represent generalization from
earlier punishing experiences with his parents,
more substantial and rapid benefits would be
derived from neutralizing emotional responses to
authority figures whom he currently fears than to
his parents.
BASIC STIMULUS DIMENSIONS
After the stimuli to be counterconditioned
have been selected, they must be scaled in terms
of their emotion-eliciting potential if one employs

a graduated approach. In setting up the stimulus
hierarchies, the potency of aversive cues can be

varied on several dimensions. When emotional
responses are elicited by nonsocial events they
can be ordered in terms of physical proximity to

the feared objects. The use of a proximity
dimension is illustrated in Wolpe’s (1962)
treatment of a woman suffering from a severe
automobile phobia that originated from a collision
at a crossroad. An anxiety hierarchy was
constructed involving highway scenes in which
cars made progressively closer advances to the
client’s automobile as it approached a highway
intersection.
Stimulus hierarchies have been constructed in

terms of a temporal dimension for treating fears of
public speaking (Paul, 1966), separation anxieties
(Lazarus, 1960), and apprehensions about

examinations (Emery & Krumboltz, 1967). A
symbolic-reality dimension is frequently employed
in setting up the anxiety stimulus continuum. A

claustrophobic series may range from reading
about others being confined in small enclosures to
imagining oneself “trapped” in a stalled elevator
for progressively longer periods of time (Wolpe,
1961); a snake-phobic series may vary from

writing the word “snake” to handling plastic
specimens, to holding a live nonpoisonous reptile

(Lazovik & Lang, 1960).
Another effective way of grading cues is by
varying the number of emotion-provoking elements
in the total stimulus complex. In Wolpe’s
treatment of the automobile phobia described

above the anxiety- arousing value of traffic scenes
was controlled by varying the speed of the
automobiles, the trustworthiness of drivers in the
approaching cars, the presence or absence of

traffic signs and signals, and the characteristics of
the person driving the client’s car.
For many persons who seek treatment,
relatively complex social stimuli or interpersonal
responses themselves serve as the primary
sources of anxiety. In scaling the emotion-arousing
properties of social cues, the nature and intensity

of the behavior of others can be utilized as the basis
for graduating threats. Thus, for example, a person

who was greatly disturbed by displays of
aggression was gradually desensitized to a

stimulus hierarchy which ranged from a situation
in which men engaged in a mild argument to one
in which the participants physically attacked each
other (Wolpe, 1958); similarly, a graded series of

devaluation situations was set up for a
gynecologist who was extremely sensitive about
any criticism or rejection from others (Wolpe,
1962). An intensely jealous male was treated by
utilizing a series of jealousy-provoking

hierarchies, each of which depicted varying
degrees of friendly interactions between the
client’s fiancée and several rival males (Wolpe,
1958).
In the illustrations presented thus far the

determinants of emotional responses are
primarily external social or situational cues. If a

person has been repeatedly punished for
displaying a particular form of behavior, the
tendency to perform these social responses
becomes, through their association with
punishment, a stimulus for anxiety. Thus, for
example, if punishment has occurred frequently in
conjunction with aggressive behavior, its

expression will elicit anticipatory emotional
reactions. Similarly, if sexual behavior is
associated with punishment, sex responses
gradually acquire anxiety-evoking properties.
Both of the emotional reactions considered are

response-correlated or self-generated. The
stimulus hierarchies in the treatment of such
classes of anxiety disorders would contain
increasing intensities of the negatively valenced
social responses, ranging from attenuated forms

that are likely to elicit relatively mild anxiety to
more forceful expressions capable of arousing

emotional responses of high magnitude.
Although the preceding discussion has
highlighted individual dimensions on which
emotion-eliciting cues can be ordered, in many
cases the stimulus hierarchies are constructed by
varying cues on several dimensions
simultaneously. Moreover, many psychological

problems are multiform in character and,
consequently, a number of different stimulus
hierarchies containing both environmental or
response-produced cues may have to be
constructed for a given individual in order to

encompass the full range of his anxiety responses.
The more generalized the emotional behavior, the
greater the need for multiple hierarchies.
Neutralization of Threats in Symbolic or

Realistic Forms
For reasons of ease, economy, and flexibility,

counterconditioning is typically directed toward
symbolic representations of actual threats.

Symbolic presentation obviates the practical
inconvenience and encumbrance of graded
physical presentations; it also allows the client to
terminate fear-arousing stimuli without making
actual avoidance responses by simply thinking of
something else. As illustrated in the preceding

section, persons may be desensitized to imagery of
aggressive behavior, sexual intimacy, social
rejection, or any other type of emotion-provoking
situation. The desensitization procedure devised
by Wolpe, therefore, represents a form of cognitive

counterconditioning where both the aversive
events and the opposing positive condition are
verbally induced and sustained through covert
self-stimulation rather than through paired
presentation of the physical stimuli themselves.

Unlike direct forms of deconditioning in which, for
example, a feared rabbit appears in temporal

conjunction with positive consummatory
responses (Jones, 1924), the subject is instructed
to visualize the rabbit and to imagine himself

eating a delicious culinary treat. In view of the
heavy reliance of Wolpe’s method upon symbolic
processes, it is surprising to find this approach
described critically as being concerned only with

peripheral skeletal processes (Murray, 1963;
Shoben, 1963).
The utilization of symbolized aversive events is
predicated on the assumption that stimuli in this
form possess emotion-arousing properties

analogous to their real-life counterparts. If this
were not the case, the symbolic method would
afford little opportunity to decrease the arousal
potential of actual threats, and hence, there would
be no appreciable treatment effects to transfer

from imagined to real situations. Results of several
studies demonstrate that thoughts do have arousal
capabilities. Miller (1950) found that emotional
responses conditioned to overt verbalizations
generalized extensively to their cognitive
equivalents so that thoughts of the negatively
valenced events generated strong GSR
responsiveness, whereas thoughts of responses
that had never been punished failed to elicit any

emotional arousal. Barber & Hahn (1964)
measured subjective discomfort and physiological
responses (heart rate, frontalis muscle tension,
and reduction in skin resistance) in subjects who
received either a painful cold stressor or merely

imagined the unpleasant experience. They found
that imagined painful stimulation produced
subjective distress and physiological responses
similar to those induced by actual painful
stimulation. In a study directly relevant to the

desensitization procedure, Grossberg & Wilson
(1968) found that instructions to visualize fearful

scenes generated significantly more autonomic
arousal than instructions to imagine neutral
situations.
Individuals who are unable, for one reason or
another, to visualize threatening stimuli vividly, or
for whom imagined scenes fail to evoke emotional
reactions, will most likely derive little benefit from

an exclusively cognitive form of
counterconditioning treatment. It is not entirely
clear why, in some cases, symbolic stimuli have
not spontaneously acquired, through
generalization, some emotion-arousing potential

from their actual counterparts to which
conditioned responses were originally established.
This phenomenon may partly reflect the outcome
of a particular form of discrimination training that
markedly influences the generalization gradient.

Under conditions where thoughts, feelings and
verbalizations are accepted or even encouraged

but corresponding overt actions are punished,
emotional conditioning is apt to be confined to the
actual activities. This type of differential

reinforcement of verbalizations and actions is, in
fact, often recommended in popular books on
child- rearing (Baruch, 1949), and widely
practiced socially.
Since therapists can exercise only indirect
control over client’s self-stimulation, requisite

treatment conditions are sometimes difficult to
create and to manage in a cognitively mediated
desensitization. Weinberg & Zaslove (1963), for

example, report that individuals occasionally
attenuated the aversiveness of situations
presented for their visualization by incorporating
protective elements. Lazovik & Lang (1960)
similarly found that a phobic subject, who derived

little benefit from desensitization, tended to
modify the presented scene by simultaneously
visualizing herself in a comfortable situation.
Emotional responses can be successfully protected
from extinction by introduction of discriminative
safety cues (Solomon, Kamin, & Wynne, 1953). In
addition to stimulus-attenuating alterations,
individuals sometimes generate excessively high
levels of arousal by adding unintended aversive

elements to a presented scene. In an effort to
increase the affective value of nonarousing
imagined stimuli, and to minimize stimulus
modifications in undesired directions, clients who
present these types of problems are often

instructed to verbalize aloud what they are
visualizing (Wolpe, 1958).
Aversive stimuli can be more precisely
controlled, the potency of counterconditioning

methods can be augmented, and problems of
transfer of extinction effects to real-life situations
can be largely obviated by utilizing actual anxiety-

provoking objects or situations. These benefits
accrue because emotional responsiveness is
extinguished to the actual stimuli that exercise
strong control over avoidance behavior under
naturalistic conditions rather than to symbolic
events that may possess weaker arousal potential
because of their dissimilarity to the primary

instigators. In keeping with this assumption,
Strahley (1966) demonstrated that phobic
subjects who were required to interact with the
feared object achieved greater reduction in fear
and avoidance behavior than subjects who

received symbolic desensitization.
A few studies have been reported in which a
graded series of tangible stimuli was employed in
the desensitization paradigm. Clark (1963) treated

a 31-year-old woman who, for more than 25 years,
had exhibited a seriously incapacitating phobia of
feathers and birds. Although she was able to

visualize scenes involving birds without displaying
much affective arousal, she was unable to venture
out of doors to situations where there was any
possibility that birds might be encountered (e.g.,
parks, zoos, outdoor walks, or the seaside); she
responded with marked anxiety to down pillows,
cushions, and feathered hats, and suffered from

persistent “anxiety dreams of people throwing
feathers and birds swooping.” The treatment
sessions were conducted in the following manner:
After feelings of calm relaxation were hypnotically
induced the therapist first presented a feather at a

distance and gradually brought it closer so long as
the client showed neither subjective disturbance
nor GSR deflections. Repeated stimulus
presentations were interspersed with suggestions
of relaxation and calmness. In this manner the

woman was deconditioned to a wide variety of
increasingly disturbing physical stimuli that

included feathers of all shapes and sizes, bags full
of feathers, stuffed birds with wings folded and
outstretched, and finally, caged live birds.
As the client displayed increasing tolerance of
feathers, she was encouraged to perform a
corresponding graded series of tasks in real-life
situations to further augment the deconditioning

and generalization process. After 20 treatment
sessions she was “completely undisturbed by
sleeping on feather pillows, could have handfuls of
feathers flung at her, could plunge her hands into a
bag of down and no longer feared going outdoors

or birds in the garden [p. 65].”
Freeman & Kendrick (1960) similarly
employed a physical stimulus dimension in
treating a woman who was terrified of cats and

responded anxiously toward a wide variety of
furry objects. The hierarchical items included
pieces of material graded in texture and

appearance which ranged from velvet to catlike
fur, toy kittens, pictures of cats, a live kitten, and
finally a large full-grown cat. In addition to adults
who are unable to produce emotion-arousing
imagery, it is also difficult to employ symbolic
desensitization with young children.
Consequently, applications of these methods to

young age groups typically involve carefully
graded exposures to actual feared objects (Bentler,
1962; English, 1929; Jones, 1924; Lazarus, 1960).
Aversive stimuli have occasionally been
presented in other tangible forms when imaginal

procedures were either ineffective or inapplicable.
Friedman (1966) successfully treated a deaf mute
who was incapacitated due to a severe dog phobia
by employing pictorial stimuli of dogs arranged in

increasing size and ferocity. Results of a study by
Leon (1967) suggest that avoidance behavior may
be more durably eliminated by neutralizing

aversive stimuli in pictorial than in imaginal
forms. Seager & Brown (1967) extinguished a
severe wind phobia by altering the fear-provoking
capacity of auditory stimuli. In the latter case, tape
recordings of wind noises progressing from mild
breezes to blustery storms were carefully adjusted
in volume and turbulence to the client’s emotional

responses, as monitored physiologically
throughout treatment.
Counterconditioning of emotional responses
poses no serious problems as long as physical
stimuli can be managed with ease. Even if the

actual objects cannot be introduced into the
treatment situation because of their size and
complexity the desensitization can be conducted
in naturalistic settings in which the critical stimuli

regularly occur. The procedural problems become
considerably more difficult, however, in cases
where emotional arousal is primarily generated by

complex social situations or by the person’s own
behavior. It would be of considerable interest in
this connection to experiment with graded
pictorial stimuli (Bandura & Menlove, 1968) or
tape-recorded social interaction sequences.
Preliminary evidence (Bandura, Blanchard, &
Ritter, 1969) indicates that these more tangible

modes of stimulus presentation, particularly if
combined with modeling cues, can extinguish
anxiety more thoroughly than when subjective
threats are cognitively reinstated. These findings
suggest that it would be advantageous to devise

graduated film sequences for objects and social
situations which are common sources of anxiety.
SELF-ADMINISTERED DESENSITIZATION IN
NATURALISTIC SITUATIONS
In the treatment strategies discussed thus far
the change agent manages the presentation of
both the emotion-provoking and the anxiety-
competing stimuli so that responses to the latter

cues prevail over the former. To the extent that a
person can be trained to manage skillfully these

two sets of events in his everyday experiences, he
can achieve some degree of self-directed
desensitization.
Graded performance tasks have been

successfully employed to some extent in
conjunction with symbolic counterconditioning.
Indeed, because individuals often show

insufficient improvements in performance
following complete symbolic desensitization,
Meyer (1966) has adopted a procedure in which

clients are required to perform behaviors under
optimal real-life circumstances after anxiety has
been thoroughly extinguished to the
corresponding imagined situations. The relative
superiority of this type of approach is

corroborated empirically by Garfield, Darwin,
Singer, & McBrearty (1967). These authors found

that avoidance behavior was more extensively
reduced by symbolic desensitization combined

with graded performance tasks than by
desensitization alone.
A further illustration of how performance
extinction can supplement even

counterconditioning involving realistic stimuli is
provided by Clark (1963) in the treatment of the
bird phobia described earlier. After a particular
aversive stimulus had been successfully
neutralized (e.g., a single feather, a bag full of

feathers, a stuffed bird) the client took the objects
home. Similarly, when emotional reactions to
stuffed birds were extinguished, visits were
arranged to an aviary and a museum containing a

varied array of stuffed birds. In later stages of
treatment, following desensitization to a live bird,
the client visited, with the reassuring support of

her family, a park full of domesticated ducks and
other fowl. As a precaution against possible
negative reconditioning by premature exposure or
the occurrence of unanticipated threats, the client
was instructed to retire from the situation or to
terminate approach behavior if it should become
emotionally disturbing.
Considering the transfer decrements from
imagined to real-life situations, carefully selected
and well-timed performance tasks should be
included as an integral part of desensitization. In
this type of treatment strategy the formal

desensitization is principally used to reduce
anxiety reactions sufficiently to enable clients to
perform desired responses in previously feared
situations where the major extinction of emotional

responding takes place.
Self-regulated desensitization can serve not
only as an important supplement to symbolic
desensitization but as a method of treatment in its
own right. Hutchison (1962) successfully treated
an electronics technician with a long history of
exhibitionism in 26 sessions by training the client
to perform a set of relaxation responses
immediately following the occurrence of stimuli

that typically preceded genital exposure. The first
few interviews were devoted to identifying the
essential covariations between social events and
the exhibitionistic responses. In this particular

case, the critical determinants involved criticism
from his supervisor or his wife, and feelings of
inadequacy connected with his vocational and
personal achievements. The client received
training in relaxation until he was capable of

inducing rapid and deep muscular relaxation.
Thus, by performing the relaxation responses

immediately following the occurrence of
experiences eliciting exposure, he was able to gain

full control over his exhibitionism. A follow-up
study conducted one year after the termination of

therapy revealed that genital exposure had been
completely eliminated.
Wolpe (1958) reports considerable success

with self-conducted desensitization of impotence
for males who are sexually responsive, but for
whom the anxiety produced by coital performance

gives rise to premature ejaculations, or an inability
to achieve and maintain an erection. Briefly, the
procedure is as follows. The individual first

receives training in progressive relaxation as an
aid in counteracting anxieties elicited in the sexual
situation. In addition, to ensure beneficial
outcomes he is advised to engage in sexual
behavior only when he has a strong positive desire

to do so, and under the most favorable
circumstances. He is then instructed to lie in bed
with his partner in a relaxed way, but to confine
the sexual activity initially to caresses and
preliminary love play. In order to avoid any
possible reinforcement of anxiety, no attempt at
intercourse is made until sexual inhibitions have
been sufficiently reduced. As deconditioning
proceeds, the client is likely to exhibit a gradual

increase in sexual responsiveness, and eventually
coitus can be attempted after adequate erections
have been achieved. Additional examples of the
use of relaxation by individuals in the self-
management of chronic anxiety reactions are

provided by Jacobson (1964), and by Haugen,
Dixon, & Dickel (1958).
The extent to which a stable change in
behavior can be produced by deliberate utilization

of self-induced responses that compete with
anxiety and supplant it depends upon whether
they serve primarily to reduce distress or to

neutralize the crucial eliciting stimuli. In
applications of relaxation by Haugen, Dixon, &
Dickel (1958), for example, people are simply
instructed to practice muscular relaxation, but
otherwise the therapists seem to show little
interest in the stimulus determinants of emotional
responses. Thus, a client who achieves sufficient

relaxation to counteract anxiety may gain
temporary relief, but if the eliciting cues are
absent during this process he will still remain
vulnerable to the disturbing stimuli because their
arousing properties have in no way been altered.

Any deconditioning that may result from an
exclusively response-directed program of
relaxation will, therefore, depend on the fortuitous
contiguous occurrence of eliciting stimuli and
anxiety-competing responses. On the other hand,

in the prescribed desensitization programs of
Hutchison (1962) and Wolpe (1958), clients are

encouraged to induce deep muscular relaxation or
other competing responses while exposed to the
crucial emotion-provoking stimuli. Under these

temporal conditions, the motivational properties
of stimulus events can be significantly modified.
If the controlling stimuli for emotional

responses have not been identified,
deconditioning effects may still be achieved to
some degree provided the competing responses

occur in close temporal relationship. The onset of
emotional arousal can serve as a cue signifying the
presence of the enigmatic eliciting events.

Provided the individual is able to discriminate
changes in his arousal and to induce positive
responses of sufficient strength immediately at the
onset of arousal, the self-produced incompatible
responses may coincide with the stimuli eliciting

anxiety, thereby ensuring the temporal
prerequisites for reconditioning. By contrast,
when persons merely engage in relaxation for a
given period of time, usually at the end of the day,
the eliciting and neutralizing events are essentially
uncorrelated.
In most social situations people obviously
cannot lapse into complete muscular relaxation.
This problem, however, does not impose serious

limitations on the utility of relaxation, since a
certain degree of stress reduction can be achieved

through the selective relaxation of muscle groups
that are not in use at any given moment (Jacobson,
1964; Wolpe, 1958). Furthermore, positive
imagery and pleasant activities that can be easily

engaged in may serve as even more effective
stress reducers.
Under free-responding circumstances, the use
of positive events in a fear-arousing situation may

hasten elimination of avoidance responses
because the positive cues enable the individual to
expose himself to threats for longer durations,
rather than because of their direct
counterconditioning effects. Nelson (1966) found
that animals willingly entered a feared situation
twice as frequently and remained there
approximately three times as long when food was
present in the situation as they did when it was

absent. However, animals that were confined in
the feared compartment for an equivalent period
of time without food showed a similar amount of
fear reduction. The influential role of duration of
exposure on extinction is further corroborated by

Proctor (1968) who presented motion pictures of
snakes of either 5 or 20 seconds duration to
snake-phobic subjects in a desensitization
paradigm. The longer exposure produced the
greater reduction in avoidance behavior. The

comparative data reported by Nelson conflict with
Davison’s (1968) finding that students who

received only exposure to aversive stimuli
experienced more distress and less avoidance
extinction than subjects who had the same

exposure paired with relaxation. Positive events
most likely serve both as incentives for self-
exposure to aversive situations and as fear
reducers.
Self-administered desensitization has several
important limitations as a sole method of

treatment. In the first place, individuals often do
not have sufficient control over the incidence and
intensity of aversive stimuli and the social context

in which they occur (Wolpe, 1958). On certain
occasions it may therefore be difficult to marshal
competing responses of sufficient strength to
contravene adverse emotional experiences.
Second, people who exhibit strong avoidance

tendencies are inclined to shun feared situations
even though objectively they may be relatively
innocuous. Finally, in cases involving severe and
widely generalized inhibitions, the extinction of
anxiety may have to commence, under controlled
conditions, at the symbolic end of the stimulus
generalization continuum. After emotional
responses to imagined threats have been
substantially reduced, the individual is in a more

favorable position to attempt previously inhibited
behavior in progressively more fearful situations.
Antagonistic Activities in
Counterconditioning
In order to achieve counterconditioning,
aversive stimuli that ordinarily evoke emotional

responses are introduced in the presence of
incompatible activities. Although relaxation has
received greatest attention, a wide variety of
operations have been employed for inducing

antagonistic response tendencies. The earliest
application of this learning principle (Jones, 1924)
relied upon appetizing foods. This particular case

involved a young boy who exhibited severe
anxiety responses of unknown origin toward
animals, and a host of furry objects including fur
coats, cotton wool, fur rugs, and feathers. In
testing the boy’s reactions to varied fear-
provoking stimuli a rabbit elicited the strongest

emotional response, and it was therefore selected
as the stimulus to be neutralized.
Counterconditioning was achieved by feeding
the boy in the presence of initially weak, but
gradually increasing, anxiety-arousing stimuli.

While he was eating his favorite food, a rabbit in a
cage was placed in the room at a sufficient
distance not to arouse emotional reactions more
powerful than the positive consummatory

activities. Each day the caged rabbit was brought
nearer to the table without evoking the customary
anxiety responses, and eventually, released from

the cage. During the final stage of treatment, the
boy not only displayed no fear at having the rabbit
placed on the feeding table or even in his lap, but
spontaneously verbalized a fondness for the
animal which previously had terrified him. Further
objective tests revealed that the anxiety extinction
effects had generalized to all the other furlike

objects that he had previously feared. Some
therapists (Bettelheim, 1950), working within a
psychodynamic framework, have made extensive
use of appetizing foods in counteracting anxiety
responses of emotionally prone children

whenever they are about to be exposed to
potentially fear-arousing situations.
That food might serve as an effective anxiety
neutralizer gains support from the suggestive

evidence cited earlier (John, 1961) that the
reticular formation possesses reciprocally
inhibitory arousal systems that mediate

conditioned defensive and alimentary activities.
During alimentary activation, response to aversive
stimuli is essentially eliminated. It is generally
assumed that counterconditioning procedures
employing food are limited in applicability to
children in whom emotional arousal cannot be
reduced through verbal means. Actually, adults

could easily conduct an entire course of
desensitization on their own by systematically
pairing real, pictorial, or imaginal aversive stimuli
with gratifying eating activities. Such
contingencies may, in fact, be more easily self-

administered than relaxation-based
desensitization. Whether this type of approach has
counterconditioning capabilities remains to be
demonstrated.
It has been shown that relaxation, which is

most often employed as the stress reducer in
desensitization of adults, increases tolerance of

aversive stimuli and can facilitate the elimination
of avoidance behavior. However, its mode of
influence is not well understood. Because of the
frequent equation of anxiety with autonomic
reactivity, the research in this area has been
almost exclusively concerned with the autonomic
effects of relaxation. It has been reported that

deep muscular relaxation decreases blood
pressure (Jacobson, 1939), pulse rate (Jacobson,
1940), and GSR responsiveness (Clark, 1963). In a
more systematic evaluation of the relative
physiological effectiveness of relaxation,

Grossberg (1965) compared changes in heart rate,
palmar conductance, and forehead muscle tension
in college students who either performed
relaxation exercises to tape- recorded
instructions, listened to music they considered

relaxing, or were simply asked to relax as best
they could without any external aids. No

significant differences were obtained among the
three groups. The author suggests, however, that
these results be accepted with reservation because

active muscle contraction during the training
exercises elevates heart rate and lowers skin
resistance, which confound the autonomic effects
of relaxation.
Paul (1969) conducted a series of studies in
which the autonomic effects of muscular

relaxation are compared with those accompanying
hypnotic suggestions of drowsiness and relaxation
and a control condition in which subjects were

simply told to relax. A variety of physiological
changes are measured including tonic forearm
muscle tension, heart rate, respiratory rate, skin
conductance, and anxiety differential. Both
relaxation training and hypnotic suggestion

reduce physiological responsiveness to stressful
imagery, but relaxation achieves somewhat
greater decrements.
The above studies demonstrate that relaxation
induction procedures can decrease physiological
responsiveness. It has been further shown by
Grings & Uno (1968), by means of a “compound
stimulus transfer” design, that the presentation of

an aversive stimulus in conjunction with
relaxation reduces the arousal capabilities of the
threatening cue. Subjects were separately trained
to respond emotionally to a colored light and to

relax whenever the word “now” was projected on
a screen. In subsequent tests subjects displayed
the strongest autonomic responses to the colored
fear cue alone, the weakest response when the
fear and relaxation cues were presented

simultaneously as a compound stimulus, and
emotional response of intermediate magnitude to

a compound stimulus containing the fear cue and a
neutral word cue. Paul (1968b) reports a positive

relationship (r = .50) between a composite
physiological index of degree of relaxation and

reduction in stress response. However, a tape-
recorded relaxation procedure was less effective
than a socially administered one, both in inducing
relaxation and in attenuating arousal to stressful

imagery.
A muscular relaxation procedure includes at

least three distinct components: calming
suggestions of tranquility and relaxation, positive
imagery, and tensing and relaxing various muscle

groups. The effects of these components need to
be studied separately to determine whether the
muscular activity per se is a significant contributor
to reductions in arousal level. It seems unlikely
that physiological research will clarify behavioral

counterconditioning effects to any great extent
until a viable theory is advanced regarding the
nature and locus of the mechanisms which control
emotional behavior. Since evidence strongly
supports a central rather than a peripheral view of
anxiety, it would be especially important to study
the effects that relaxation, or any other anxiety-
neutralizing activity that might be employed in
counterconditioning, have upon central arousal

mechanisms.
Lazarus & Abramovitz (1962) have
occasionally relied upon positive imagery for
modifying fearful behavior of children for whom
relaxation was not feasible. This procedure is

identical with the standard method of
desensitization except that the graded aversive
stimuli are presented in the context of strong
positive ideations. The child is interviewed on his

areas of interest and his idols, usually drawn from
television, movies, fiction, or the child’s own
imagination. “The child is then asked to close his

eyes and told to imagine a sequence of events
which is close enough to his everyday life to be
credible, but within which is woven a story
concerning his favorite hero [p. 192].” After a
sufficient degree of positive affect has been
created, the therapist introduces into the narrative
the lowest item in the hierarchy and the child is

instructed to signal if he feels afraid, unhappy or
uncomfortable. When the child registers
disturbance, the threatening element is
immediately withdrawn, and the positive imagery
is further enhanced. This procedure is continued

until the most phobic item has been neutralized. In
most cases arousal-reducing imagery can be
presented and controlled more effectively in
discrete conditioning trials rather than in the form
of a continuing narrative. Pleasant imagery and

mollifying thoughts are often used in this manner
with adults to enhance the tranquilizing effects of

relaxation procedures. No attempts have been
made, however, to assess the physiological effects
of positive imagery, or to determine whether it

accelerates extinction of avoidance behavior.
Those who often mediate pleasant experiences
or the reduction of discomfort in others are likely
to acquire positive properties; consequently, the

mere presence of such an individual will elicit
positive affective responses that can serve as

anxiety neutralizers. Frequent social contact, even
though unaccompanied by nurturant functions,
may also endow others with positive valence
(Cairns, 1966; Homans, 1961). That familiar social

stimuli can function as anxiety reducers has been
clearly shown with both humans and infrahuman
subjects. Mason (1960) found that responses

indicative of emotional disturbance were
exhibited less frequently to stressful situations by

monkeys in the presence of peers than in the
company of adult monkeys (whom they had rarely
seen since birth), other animals, or when they

were left alone in the situation. The influence of
familiarity on social stress reduction receives
further support from a study by Kissel (1965)
conducted with college students. A friend’s
company was found to be more effective in
diminishing autonomic arousal to induced failure

than that of a stranger, whose presence had no
distress-reducing value.
There is reason to expect from the above
laboratory findings that relationship-induced

responses can serve to mitigate emotional arousal
to some extent. Wolpe (1958), in fact, contends
that favorable outcomes achieved by traditional
interview methods primarily derive from
inadvertent counterconditioning of anxiety by

positive responses evoked in the client-therapist
relationship. This interpretation is consistent with
Shoben’s (1949) conceptualization of the
treatment process as one in which symbolically
reinstated anxiety is counterconditioned through
association with comfort reactions made by the
therapeutic relationship. Although this type of
deconditioning is possible, the outcome data
reviewed in the introductory chapter suggest that

it does not occur with any degree of consistency.
The high dropout rates and frequent displays of
avoidance behavior by clients who continue in
treatment suggest that therapists are more likely
to arouse anxiety than comfort reactions. Many of

the widely prescribed therapist-role behaviors
would, in fact, be clearly contraindicated if judged
primarily in terms of their comfort-inducing
function. Therapists of psychoanalytic persuasion,
for example, strive hard to maintain a high degree

of ambiguity on the assumption that ambiguity
facilitates and intensifies transference responses.

Controlled studies (Bordin, 1958; Dibner, 1958),
on the other hand, demonstrate that the amount of
anxiety experienced by a client varies positively

with the ambiguity of the therapist. Thus, while
ambiguity may facilitate generalization of
inappropriate response patterns, it is antithetical
to counterconditioning.
The fact that strength of relationship-produced
responses cannot be easily controlled or rapidly

increased if necessary to counteract strong
emotional arousal places additional limitations on
the extent to which relationship variables alone

can create the requisite conditions for successful
desensitization. Positive relationship-induced
responses may thus serve as an important adjunct
to, but not a reliable substitute for, more powerful
anxiety-neutralizing procedures. Even if the

necessary competing comfort reactions were
strongly established, desensitization outcomes
would remain unpredictable if the introduction of
emotionally disturbing contents was primarily left
to the vagaries of clients, rather than carefully
regulated by psychotherapists.
Some of the procedures employed in
counterconditioning treatments, such as
relaxation and pleasant imagery, for the explicit

purpose of inducing strong comfort reactions, can
also enhance the anxiety-reducing cue value of the

therapist. An early hypnosis study conducted by
Estabrooks (1930) demonstrated that even a
person who, in the experiences of subjects, had
been indirectly associated with relaxation

acquired deactivation properties. During early
phases of this experiment, subjects’ level of
autonomic arousal decreased only as hypnosis

was induced. However, after they had been
hypnotized a number of times, the mere presence

of the apparatus operator before the induction of
hypnosis produced much the same decrease in
arousal level. As would be expected, this

phenomenon did not occur with a subject who had
had a negative experience with the operator
sometime prior to the experimental session.
The rate at which emotional behavior is

extinguished by therapeutic agents who vary in
their stress-reducing properties has not been
systematically investigated. Informal observations

of cases where parents serve as therapists in
eliminating their children’s phobic behavior
(English, 1929) have suggestive value and are in

accordance with experimental findings. A 7-
month-old child developed marked fear of a
stuffed cat on the basis of a sudden aversive
experience. In an effort to extinguish this fear a
therapist presented the stuffed animal repeatedly,

but each time the child responded with
withdrawal, trembling, and frightened wails.
When the cat was later offered in the presence of
both parents, she accepted it hesitantly but
continued to exhibit some apprehensiveness. A
brief experience during which the child handled
the toy while she was held in her mother’s arms
effectively eliminated the residual fear responses;
thereafter, the child readily accepted the stuffed

animal and played happily with it. The differential
distress shown by the child to the feared object
when the parents were present and absent is
similar to Liddell’s (1950) findings with
infrahuman subjects that the presence of a mother

increases her offspring’s tolerance of stressful
stimuli. Although the relative efficacy of different
persons for mitigating emotional disturbance was
not explored, Bentler (1962) reports a case in
which a mother completely extinguished an

aquaphobia in her infant daughter by reexposing
her to progressively larger amounts of water in

the context of close maternal contact
supplemented by a highly prized flotilla of toys.
Most parents similarly function as effective

anxiety reducers in modifying their children’s
fears of loud noises, heights, darkness, animals,
and other common fear-provoking situations
(Jersild & Holmes, 1935).
Pharmacologic agents that decrease emotional
arousal have also been occasionally employed in

cases where psychological procedures for
inducing competing activities have proved either
ineffective or not feasible for various reasons

(Friedman, 1966; Walton & Mather, 1963b). Brady
(1966) has made extensive use of short-acting
barbiturates (e.g., metho-hexitone sodium) in
conjunction with relaxation instructions as a quick
means of producing deep muscular relaxation.

This method was applied with notable success in
the treatment of severe frigidity in women who
rarely engaged in coitus because it caused them
considerable physical pain, revulsion and anxiety.
After they were desensitized to individualized sex-
anxiety hierarchies they not only became
considerably more sexually responsive but found
intercourse erotically pleasurable. In a later paper
Brady (1967) reaffirms the efficacy of

desensitization based on drug-induced relaxation
with a variety of anxiety disorders. Friedman
(1968) reports similarly high success rates in the
treatment of impotence by drug-assisted
desensitization.
The foregoing results are sufficiently

promising to warrant systematic comparative
evaluations of the degree to which, if any, drug-
produced effects facilitate the counterconditioning

process. There is some laboratory evidence to
indicate that the beneficial tranquilizing effects of
drugs may be partially offset by their retarding

effects on learning (Cole & Gerard, 1959; Mitchell
& Zax, 1959; Schneider & Costiloe, 1957). Not only
can conditioning be impeded but, if results from
animal experimentation are applicable to humans,
changes induced during a drugged state may have
limited transfer value. Barry, Etheredge, & Miller
(1965) found that varying dosages of amobarbital

sodium enabled animals to resume a food-
producing response that was previously inhibited
by fear conditioning, but the reduction of
behavioral inhibition failed to transfer to the
normal nondrugged state. A study by Sherman

(1967) suggests, however, that the transfer
decrement might be obviated by a treatment
procedure involving progressive reduction in
dosage of the fear-reducing drug. Amobarbital
sodium restored previously inhibited behavior in

animals during the sedated state; however,
subjects who experienced abrupt withdrawal of

the drug exhibited a precipitous decrement in
performance to the level of the saline control
group, whereas those who continued to receive

the drug in progressively smaller amounts showed
a monotonic increase in responsiveness.
Apparently, a gradual drug-withdrawal treatment
can facilitate permanent extinction of fear-

mediated behavior although, as the author notes,
the relative superiority of this approach cannot be

determined without data from a group that is
deconditioned under a constant drug dose for an
equivalent period of time before the drug is
abruptly withdrawn.
The question remains as to why organisms that
repeatedly perform feared responses, which are
intermittently rewarded during a drugged state,
fail to display some degree of permanent fear

extinction. One interpretation, favored by Barry,
Etheredge, & Miller (1965), assumes that lasting
extinction does occur, but transfer of therapeutic
effects is impeded by marked stimulus change
resulting from the shift from sedated to normal
states. Considering that avoidance behavior is
extensively controlled by external stimuli, it seems
improbable that changes in internal stimulation
adequately explain why animals may show a

dramatic loss from 200 rewardable responses
while under the influence of the drug to only 3
scattered responses in the same situation under
nondrugged conditions (Sherman, 1967). An
alternative interpretation would involve learning

rather than generalization deficits. Drugs,
particularly in higher dosages, may produce not
only transient deactivation but also impairment of
learning functions. Behavioral changes which are
primarily chemically induced would not be

expected to persist after pharmacologic recovery.
On the other hand, optimal drug dosages that

produce beneficial tranquilizing effects without
adversely affecting learning processes might
augment extinction of the anxiety-arousing

potential of aversive stimuli. Even if comparative
evaluations showed this to be the case,
pharmacologic aids should be primarily confined
to persons who do not benefit from exclusively

psychological procedures. In view of the
unpleasant effects of intravenous injections and

the potential dangers of drug dependency with
repeated usage, the extinction gains would have to
be substantial to justify frequent use of drugs as
adjuncts to standard counterconditioning

procedures.
It should also be noted in passing that persons
who routinely consume “tranquilizing” drugs will
not necessarily experience progressive extinction

of emotional responses. Achievement of lasting
deconditioning outcomes requires both the
presence of a tranquil or positive emotional state
of sufficient strength, and judicious reexposures to
crucial anxiety-evoking stimuli. These learning
requisites are rarely obtained in experiences of
everyday life. On the contrary, persons are often
prematurely confronted with excessively
threatening situations that result in the

reinforcement of avoidant behaviors rather than
in their extinction. The more severely
incapacitated cases, on the other hand, are usually
too heavily sedated to be capable of much
reconditioning.
Physiological Accompaniments of
Emotional Behavior
Theories of personality and psychotherapy
generally differentiate among types of “impulses”
or emotional states as though they represented

distinct forms of physiological arousal. Thus, in
one case a person is assumed to be suffering from
“repressed hostile impulses” and is therefore

encouraged to express verbal or physical
aggression designed to discharge the troublesome
hostile affective state. In another case, “anxiety”
may be regarded as the major emotional condition
that presumably reflects a somewhat different
form of physiological arousal. If emotional

behaviors which are designated as anger, fear, or
euphoria were controlled by separate
physiological events, then different types of
treatments might be required to extinguish
diverse types of emotional behavior.
Physiological studies disclose that the varied

array of emotions that people experience
phenomenologically are not accompanied by a
corresponding diversity in physiological response

patterns. In the procedures commonly used
individuals are subjected to fear- or anger-
provoking stimulation during which changes in

numerous physiological responses are
simultaneously recorded. Interpretation of these
findings is complicated by lack of independent
evidence that the two stimuli are of comparable
aversiveness. It is consequently difficult to
determine whether obtained differences are
attributable to dissimilarities in the qualitative

features or in the relative intensities of the
aversive stimuli. Moreover, as Duffy (1962) has
noted, unless it has been shown that dissimilar
stimuli within the same emotional class produce
identical patterns of physiological arousal, the

generality of findings yielded by a single fear
stimulus and a single anger stimulus is open to
question.
Based on manipulations in which subjects

experienced increasing shock while the
experimenter expressed alarm about a dangerous
high-voltage short circuit in the apparatus, and in

which they also received rude sarcastic treatment
by an assistant, Ax (1953) found some subtle
differences in physiological reactions to the fear
and anger provocation. Data reported by
Schachter (1957) using analogous manipulations
disclose essentially similar cardiovascular
responsiveness to fear and anger stimulation, but

they both differ significantly from reactions to
pain induced by a cold pressor test. However, on
the basis of subjective categorization of the data
the author extracts greater physiological
specificity than the actual findings warrant. It

seems unlikely that small differences in an
otherwise identical pattern of physiological
reactivity are sufficiently discriminable, if at all, to
serve as the cues for differentiating among
different emotional states.
Results of both physiological and psychological
studies support the conclusion that a common

diffuse state of physiological arousal mediates
diverse forms of emotional behavior and that
different emotional states are identified and
discriminated primarily in terms of external
stimuli rather than internal somatic cues. Among
the situational cues that help to label a given state
of physiological arousal, the inducing stimulus

conditions undoubtedly play an influential role.
Thus, for example, visceral arousal that is
generated by threatening stimulation is likely to
be interpreted as fear or anxiety; arousal
produced by obstacles and thwarting activities of

frustrating agents will tend to be experienced as
anger; and arousal resulting from highly
pleasurable stimulation will be identified as joy or
euphoria. In a study designed to identify the
characteristics of situations that might serve as

cues for differentiating among emotions Hunt,
Cole, & Reis (1958) found that students were

inclined to rate environmental events as
provoking fear when they were threatening, as
anger when frustrating agents figured

prominently, and as sorrow when desired objects
were irretrievably lost.
Even the same inducing operation and its

associated physiological arousal may result in
differential emotions depending on the form of
affective modeling cues which serve to define the

appropriate emotional behavior under conditions
of ambiguity. According to the theory of emotion
advanced by Schachter (1964) when a person

experiences a state of physiological arousal and
cannot clearly identify its source, the same
emotional condition may be interpreted as anger,
euphoria, anxiety, or some other type of feeling
depending upon the nature of external influences.

The interaction between modeling, cognitive, and
physiological determinants of emotional state is
revealed in an experiment by Schachter & Singer
(1962) which proceeded in the following manner.
One group of college students received injections
of epinephrine, a sympathetic stimulant, and was
at the same time correctly informed of its physical
side effects. A second group of subjects was
similarly administered epinephrine but was

uninformed about its side effects, while other
students received a placebo injection of saline
solution. Immediately after the experimental
manipulation of physiological arousal, al subjects
were sent to a room where they were exposed to

the experimenter’s confederate, supposedly
another subject, who displayed considerable anger
and verbal aggression toward the experimental
procedure. Subjects in the epinephrine-
uninformed group displayed more anger than

students in either the epinephrine-informed or
placebo condition, which did not differ from each

other. In another phase of this experiment, four
treatments were employed, the three described
above and one in which subjects were injected

with epinephrine and deliberately misinformed
concerning its side effects so they had no adequate
explanation for their aroused state. The
confederate in this phase behaved in an

extraordinarily euphoric manner, for example,
flying paper airplanes, hoola-hooping, and playing

basketball with equipment contained in the room.
Subjects who experienced physiological arousal
and were misinformed or uninformed concerning
the basis for their reactivity displayed a great deal

of euphoria, whereas subjects who were equally
aroused but for whom an accurate explanation
was available, and the nonaroused placebo group,

were little affected by the behavior of the
confederate.
In a related experiment, Schachter & Wheeler
(1962) extended the range of autonomic arousal
by administering either epinephrine, a placebo
saline solution, or chlorpromazine, a sympathetic
depressant, to different groups of subjects. After
receiving their injections all subjects viewed a
slapstick comedy film. The epinephrine-injected
group showed a greater amount of amusement,

assessed both by self-reports and behavioral
ratings, than placebo subjects, while the
chlorpromazine-injected group was least affected
by the comical displays.
Further research conducted by Nisbett &

Schachter (1966) showed that emotional states
induced by environmental stimuli are manipulable
to some extent, as was demonstrated with drug-
induced arousal. Students were administered

either weak or severe electric shocks after
receiving a placebo pill. Half the subjects within
each condition were led to believe that the side

effects accompanying the drug were similar to
shock- produced emotional reactions, whereas the
remaining subjects were correctly informed that
shock evokes arousal symptoms such as
palpitations and tremors. Students who received
weak shock and attributed their autonomic
arousal to the pill tolerated more painful

stimulation and reported less pain than those who
interpreted their arousal as due to shock.
However, ascribing arousal to an artificial source
had no influence on pain tolerance when the
shocks were severe. The latter findings indicate

that arousal states are less susceptible to
relabeling when the controlling stimuli are
apparent and powerful.
Taken together, the studies demonstrate that

emotional states are partly a function of the
degree of physiological arousal, but that social and
cognitive variables may play a crucial role in

determining both the nature and intensity of the
emotions experienced, particularly when
individuals cannot accurately label the source of
their aroused condition. Thus, the same state of
physiological arousal can be experienced as
euphoria, anger, or some other type of emotional
condition depending upon defining cognitions and

the affective reactions of others to arousal-
inducing situations.
The preceding findings have several important
therapeutic implications. It is not necessary to
reduce specific types of physiological arousal in

order to modify different forms of affective
behavior. To the extent that counterconditioning
procedures can successfully neutralize the arousal
potential of valenced stimulus events, then the

method should be applicable not only to problems
of anxiety but also to emotions experienced as
hostility, jealousy, grief, or by some other name. It

should also be possible to decrease positively
labeled emotions by this means if the theory of
nonspecific visceral control is valid. Finally, some
emotional problems might result from mislabeling
of feeling states; given this, the person would need
to be taught to discriminate accurately the
determinants of his arousal states.
People often experience anxiety arousal
without being able to identify the evocative
stimuli. In interpretive therapies such emotional
responses are frequently attributed to
hypothetical causes operating at an unconscious

level. If emotional arousal can be reduced to some
extent by misattributing the reactions to non-
emotional sources, then it is conceivable that
neutral stimuli could become invested with fear-

arousing properties if they are erroneously
interpreted as the source of anxiety reactions.
Summary
In this chapter the principle of
counterconditioning has been discussed in
relation to the modification of emotional behavior
by neutralizing the arousal potential of
threatening stimuli. The reconditioning process is
achieved by inducing activities incompatible with

emotional responses in the presence of anxiety-
arousing stimuli. This mode of behavior change is
based on the fact that classically conditioned
effects can exert mediating control, principally

through central mechanisms, over instrumentally
learned behavior.
Three sets of variables, some necessary and
others merely facilitative, have been singled out as
especially relevant to counterconditioning
processes. First, one must select an anxiety-
neutralizing stimulus capable of eliciting

competing events of sufficient strength to
predominate over responses characteristically

evoked by emotion-arousing cues. In practice,
muscular relaxation, appetizing foods, positive

imagery, relationship-induced affective responses,
and pharmacological agents that decrease
emotional arousal have been employed as anxiety
neutralizers. These types of competing activities

have been shown to increase tolerance of aversive
stimuli, accelerate the rate of desensitization, and
generally facilitate extinction of severe avoidance
behavior.
The second class of variables pertains to

emotion-provoking events. The issues here
concern accurate identification of the stimulus
determinants of emotional behavior, and the
forms and intensities in which arousal stimuli are

neutralized. Counterconditioning treatments are
typically directed toward symbolic
representations of actual threats because, in the

latter form, they can be easily controlled and the
method can be applied to an almost infinite variety
of anxiety sources. However, in cases where
symbolic stimuli lack arousal capacity or requisite
conditions for achieving desensitization cannot be
adequately induced through verbal means, actual
threats are presented in physical, pictorial, or

auditory modes.
In most applications of the
counterconditioning principle, the aversive events
are initially administered in attenuated forms so
that the emotional responses to be counteracted

are relatively weak and can therefore be readily
extinguished. As weak items lose their anxiety-
producing value, progressively more threatening
stimuli, which presumably have been weakened

through generalization of extinction effects, are
gradually introduced. Although stimulus
graduation is not a necessary condition for

achieving desensitization, it permits greater
control over the change process and produces
minimal anxiety elicitation as compared to
approaches involving repeated confrontation with
stimuli having high threat value.
The third variable pertains to the temporal
prerequisites for the occurrence of

counterconditioning outcomes. Both the anxiety-
neutralizing and the aversive events must be
contiguously associated. The mechanism of
conditioning is conceived as including both
mediational and association processes.
Numerous laboratory investigations and
controlled individual studies utilizing the symbolic

desensitization paradigm with relaxation show
this approach to be effective for extinguishing the

emotional arousal capacity of aversive stimuli and
for reducing avoidance behavior. Moreover,
generalized improvements in behavioral
functioning often result from specifically induced
changes. However, more refined analyses of
degree of transfer of extinction effects from
symbolic stimuli to real-life situations disclose
some generalization decrement. Not only is the
number of approach responses that persons can

perform behaviorally generally less than the
number that have been successfully neutralized in
symbolic form, but restored approach behavior is
usually accompanied by moderate anxiety when

first performed. This transfer decrement partly
reflects the inherent limitations of working
exclusively with symbolic counterparts which
rarely encompass the full range of aversive
elements contained in real-life situations.

Counterconditioning alone is also likely to effect
limited behavioral improvements in conditions

where anxious responding is a realistic
consequence of behavioral deficits, or where the

rewards associated with restored functioning are
outweighed by the advantages of remaining

behaviorally incapacitated.
Symbolic desensitization might primarily serve
to decrease the anxiety-producing value of
aversive stimuli below the threshold for activating

avoidance behavior, thus enabling persons to
engage, though somewhat anxiously, in new

approach behavior. This provides the opportunity
for further extinction of residual anxiety and
avoidance behavior in naturalistic settings. Also, in
cases involving severe anxiety disorders,

deconditioning may have to commence with
symbolic stimuli that are far enough removed
from real threats to evoke less intense reactions.

After emotional responses to imaginal stimuli
have been substantially reduced, the individual is

better prepared for encounters with the
corresponding real-life situations. Emotional
behavior can be most thoroughly extinguished by

supplementing symbolic desensitization with
graded performance tasks, positive reinforcement
of desired approach behavior, and appropriate
modeling procedures. It is possible that the
efficacy of desensitization itself can be further
enhanced, and transfer problems reduced, by the

use of more tangible threats along with more
powerful competing activities.
The behavioral changes achieved by
desensitization operations cannot be attributed

solely to the conditioning of competing responses
to fear-arousing stimuli through repetitive paired
association. Other mechanisms are also operative.
Some reduction in avoidance behavior
undoubtedly results from selection of explicit

behavioral objectives and positive reinforcement
of progressive advances toward the chosen goal.
The resultant changes also partly reflect the
influence of exposure to aversive stimuli
independently of the effects of the explicitly
programmed competing activities. For this reason,
the multiform procedure combining graduated
exposure, anxiety-neutralizing events, and
positive reinforcement is generally more effective

in extinguishing avoidance behavior than the
separate components alone.
Although counterconditioning methods have
been primarily employed to extinguish “anxiety,”
evidence that diverse emotional behaviors are

mediated by a common diffuse state of
physiological arousal indicates that this approach
may be applicable to other emotional conditions
as well. Furthermore, counterconditioning

procedures can be utilized not only to neutralize
aversive events but also to attach negative
valences to positive stimuli which are potentially

harmful. The principles governing these aversive
forms of counterconditioning are discussed in the
following chapter.
REFERENCES
Agras, W. S. Transfer during systematic

desensitization therapy. Behaviour Research and
Therapy, 1967, 5, 193-199.
Ashem, B. The treatment of a disaster phobia by

systematic desensitization. Behaviour Research
and Therapy, 1963, 1, 81-84.
Ax, A. F. The physiological differentiation between

fear and anger in humans. Psychosomatic
Medicine, 1953, 15, 433-442.

13, 173-199.

Barber, T. X., & Hahn, K. W., Jr. Experimental studies in

“hypnotic” behavior: Physiological and subjective
effects of imagined pain. Journal of Nervous and
Mental Disease, 1964, 139, 416-425.
Barry, H., in., Etheredge, E. E., & Miller, N. E.
Counterconditioning and extinction of fear fail to
transfer from amobarbital to nondrug state.
Psychopharmacologia, 1965, 8, 150-156.
Baruch, D. W. New ways in discipline. New York:

McGraw-Hill, 1949.
Bass, M. J., & Hull, C. L. The irradiation of a tactile

conditioned reflex in man. Journal of Comparative
Psychology, 1934, 17, 47-65.
Bentler, P. M. An infant’s phobia treated with

reciprocal inhibition therapy. Journal of Child
Psychology and Psychiatry, 1962, 3, 185-189.
Bettleheim, B. Love is not enough. Glencoe, Ill.: Free

Press, 1950.
Beyme, F. Hyperesthesia of taste and touch treated by

reciprocal inhibition. Behaviour Research and
Therapy, 1964, 2, 7-14.

Black, A. H. Heart rate changes during avoidance

learning in dogs. Canadian Journal of Psychology,
1959,13, 229-242.
Black, A. H. Transfer following operant conditioning in
the curarized dog. Science, 1967,155, 201-203.
Bond, I. K., & Hutchison, H. C. Application of reciprocal

inhibition therapy to exhibitionism. Canadian
Medical Association Journal, 1960, 83, 23-25.

inhibition therapy to exhibitionism. In H. J.
Eysenck (Ed.), Experiments in behaviour therapy.
New York: Pergamon, 1964. pp. 80-86.
Bordin, E. S. Inside the therapeutic hour. In E. A.

Rubinstein & M. B. Parloff (Eds.), Research in
psychotherapy. Washington, D. C.: American
Psychological Association, 1958. pp. 235-246.
Bower, G., & Grusec, T. Effect of prior Pavlovian

discrimination training upon learning an operant
discrimination. Journal of Experimental Analysis
of Behavior, 1964, 7, 401-404.
Brady, J. P. Brevital-relaxation treatment of frigidity.

Behaviour Research and Therapy, 1966, 4,71-77.
Brady, J. P. Brevital-relaxation treatment of neurosis.

Proceedings of Symptomatic Higher Nervous
Activity, Madrid, Spain, September 9, 1967.
Cairns, R. R. Attachment behavior of mammals.

Cameron, N. Personality development and
psychopathology: A dynamic approach. Boston:
Houghton Mifflin, 1963.
Chapel, J. L. Treatment of a case of school phobia by

reciprocal inhibition. Canadian Psychiatric
Association Journal, 1967, 12, 25-28.
Clark, D. F. The treatment of monosymptomatic

phobia by systematic desensitization. Behaviour
Cohen, S. I., & Reed, J. L. The treatment of “nervous

diarrhoea” and other conditioned autonomic
disorders by desensitization. British Journal of
Psychiatry, 1968, 114, 1275-1280.
Cole, J. O., & Gerard, R. W. (Eds.), Psychopharmacology:

Problems in evaluation, Washington, D. C.:
Research Council, 1959.
Cooke, G. The efficacy of two desensitization

procedures: An analogue study. Behaviour
Cooke, G. Evaluation of the efficacy of the components

of reciprocal inhibition psychotherapy. Journal of
Cooper, J. E. A study of behaviour therapy in thirty
psychiatric patients. Lancet, 1963, 23,411-415.

162-166.
Cowden, R. C., & Ford, L. I. Systematic desensitization

with phobic schizophrenics. American Journal of
Psychiatry, 1962, 119, 241-245.
Davison, G. C. Systematic desensitization as a

counterconditioning process. Journal of Abnormal
Psychology, 1968, 73, 91-99.
Dibner, A. S. Ambiguity and anxiety. Journal of

174.
Donner, L. Effectiveness of a pre-programmed group

desensitization treatment for test anxiety with
and without a therapist present. Unpublished
doctoral dissertation, Rutgers University, 1967.
Duffy, E. Activation and behavior. New York: Wiley,

1962.
Efran, J. S., & Marcia, J. E. The treatment of fears by

expectancy manipulation: An exploratory
investigation. Proceedings of the 75th Annual
Convention of the A.P.A. Washington, D. C.:
239-240.
Emery, J. R., & Krumboltz, J. D. Standard versus

individualized hierarchies in desensitization to
reduce test anxiety. Journal of Counseling
Psychology, 1967,14, 204-209.
English, H. B. Three cases of the “conditioned fear

response.” Journal of Abnormal and Social
Psychology, 1929, 24, 221-225.
Epstein, R., & Liverant, S. Verbal conditioning and sex-

role identification in children. Child Development,
196.3, 34, 99-106.
Estabrooks, G. H. The psychogalvanic reflex in

hypnosis. Journal of General Psychology, 1930, 3,
150-157.
Farber, I. E. Response fixation under anxiety and non-

anxiety conditions Journal of Experimental
Psychology, 1948, 38, 111-131.
Fenichel, O. The psychoanalytic theory of neurosis. New

York: Norton, 1945
Folkins, C. H., Lawson, K., Opton, E. M., Jr., & Lazarus,

R. S. Desensitization and the experimental
reduction of threat. Journal of Abnormal
Psychology 1968, 73, 100-113.
Ford, C. S., & Beach, F. A. Patterns of sexual behavior.
New York: Harper 1951.
Freeman, H. L., & Kendrick, D. C. A case of cat phobia:

Treatment by method derived from experimental
psychology. British Medical Journal 1960, 2, 497-
502.
Friedman, D. Treatment of a case of dog phobia in a

deaf mute by behavior therapy. Behaviour
Research and Therapy, 1966, 4, 141.
Friedman, D. The treatment of impotence by brevital

relaxation therapy. Behaviour Research and
Therapy, 1968, 6, 257-261.
Gale, D. S., Sturmfels, G., & Gale, E. N. A comparison of

reciprocal inhibition and experimental extinction
in the psychotherapeutic process. Behaviour
Garfield, Z. H., Darwin, P. L., Singer, B. A., & McBrearty,

J. F. Effect of “in vivo” training on experimental
desensitization of a phobia. Psychological Reports,
1967, 20, 515-519.
Geer, J. H., & Katkin, E. S. Treatment of insomnia using

a variant of systematic desensitization. Journal of
Abnormal Psychology, 1966, 7,161-164.
Geer, J. H., & Silverman, I. Treatment of a recurrent
nightmare by behavior- modification procedures.
190.
Gelder, M. G., & Marks, I. M. Desensitization and

phobias: A cross-over study. British Journal of
Psychiatry, 1968,114, 323-328.
Gray, B. B., England, G., & Mohoney, }. L. Treatment of

benign vocal nodules by reciprocal inhibition.
193.
Grings, W. W., & Lockhart, R. A. Galvanic skin response

during avoidance learning. Psychophysiology,
1966, 3, 29-34.
Grings, W. W., & Uno, T. Counterconditioning: Fear

and relaxation. Psychophysiology, 1968, 4, 479-
485.
Grossberg, J. M. The physiological effectiveness of

brief training in differential muscle relaxation.
Unpublished manuscript, San Diego State College,
1965.
Grossberg, J. M., & Wilson, H. Physiological changes

accompanying the visualization of fearful and
neutral situations. Journal of Personality and
Social Psychology, 1968,10, 124-1.33.
Harper, 1935.
Hain, J. D., Butcher, R. H. G., & Stevenson, I. Systematic

desensitization therapy: An analysis of results in
twenty-seven patients. British Journal of
Psychiatry, 1966, 112,295-307.
Hartup, W. W. Friendship status and the effectiveness

of peers as reinforcing agents. Journal of
Haslam, M. T. The treatment of an obsessional patient

by reciprocal inhibition. Behaviour Research and
Therapy, 1965, 2, 213-216.
Haugen, G. B., Dixon, H. H., & Dickel, H. A. A therapy for

anxiety tension reactions. New York: Macmillan,
1958.
Henry, G. W. Sex variants. New York: Hoeber, 1941. 2

vols.
Hoenig, J., & Reed, G. F. The objective assessment of

desensitization. British Journal of Psychiatry,
1966, 112, 1279-128.3.
Hoffeld, D. R. Primary stimulus generalization and

secondary extinction as a function of strength of
conditioning. Journal of Comparative and
Homans, G. C. Social behavior: Its elementary forms.
New York: Harcourt, Brace & World, 1961.
Hooker, E. The homosexual community. Proceedings

of the XIV International Congress of Applied
Psychology. Copenhagen: Munksgoord, 1961.
Hovland, C. I. The generalization of conditioned

responses: I. The sensory generalization of
conditioned responses with varying frequencies
of tone. Journal of General Psychology, 19.37, 17,
125-148.
Howe, E. S. GSR conditioning in anxiety states,

normals, and chronic functional schizophrenic
subjects. Journal of Abnormal and Social
Psychology, 1958, 56, 183-189.
Hunt, J. McV., Cole, M. W., & Reis, E. E. S. Situational

cues distinguishing anger, fear, and sorrow.
American Journal of Psychology, 1958, 71, 136-
151.
Hutchison, H. C. Reciprocal inhibition therapy as an

alternative to psychotherapy in the treatment of
certain neurotic conditions. Paper read at the
meeting of the American Psychological
Association, St. Louis, 1962.
Jacobson, E. Variation of blood pressure with skeletal
muscle tension and relaxation. Annals of Internal
Medicine, 1939, 12, 1194-1212.
Jacobson, E. Variation of blood pressure with skeletal

muscle tension and relaxation. II. The heart beat.
Annals of Internal Medicine, 1940, 13, 1619-1625.
Jacobson, E. Anxiety and tension control. Philadelphia:

Lippincott, 1964.
Jersild, A. T., & Holmes, F. B. Methods of overcoming

children’s fears. Journal of Psychology, 1935,1, 75-
104.
John, E. R. High nervous functions: Brain functions

and learning. Annual Review of Physiology, 1961,
23, 451-481.
Jones, M. C. A laboratory study of fear: The case of

Peter. Pedagogical Seminary, 1924, 31, 308-315.
Kahn, M. W. The effect of socially learned aggression

or submission on the mating behavior of C57
mice. Journal of Genetic Psychology, 1961, 98,211-
217.
Kelley, H. H. The warm-cold variable in first

impressions of persons. Journal of Personality,
1950,18,431-439.
Kimble, G. A., & Kendall, J. W., Jr. A comparison of two
methods of producing experimental extinction.
90.
Kissel, S. Stress-reducing properties of social stimuli.

2, 378-384.
Kraft, T., & Al-Issa, I. The application of learning

theory to the treatment of traffic phobia. British
Journal of Psychiatry, 1965, 111, 277-279. (a)
Kraft, T., & Al-Issa, I. Behaviour therapy and the recall

of traumatic experiences—a case study.
(b)
Kraft, T., & Al-Issa, I. Alcoholism treated by

desensitization: a case report. Behaviour Research
and Therapy, 1967, 5, 69-70. (a)
Kraft, T., & Al-Issa, I. Behavior therapy and the

treatment of frigidity. American Journal of
Psychotherapy, 1967, 21, 116-120. (b)
Krapfl, J. E. Differential ordering of stimulus

presentation and semiautomated versus live
treatment in the systematic desensitization of
snake phobia. Unpublished doctoral dissertation,
University of Missouri, 1967.
Lacey, J. I. Individual differences in somatic response
patterns. Journal of Comparative and
Lacey, J. I. Somatic response patterning and stress:

Some revisions of activation theory. In M. H.
Appley & R. Trumbull (Eds.), Psychological stress:
Issues in research. New York: Appleton-Century-
Crofts, 1967. pp. 14-42.
Lang, P. J. Fear reduction and fear behavior: Problems

in treating a construct. In J. M. Shlien (Ed.),
Research in psychotherapy, Vol. III. Washington, D.
C.: American Psychological Association, 1968. pp.
90-102.
Lang, P. J., & Lazovik, A. D. Experimental

desensitization of a phobia. Journal of Abnormal
Lang, P. J., Lazovik, A. D., & Reynolds, D. J.

Desensitization, suggestibility, and pseudo-
therapy. Journal of Abnormal Psychology, 1965,
70, 395-402.
Lazarus, A. A. The elimination of children’s phobias by

deconditioning. In H. J. Eysenck (Ed.), Behavior
therapy and the neuroses. New York: Pergamon,
1960. pp. 114-122.
Lazarus, A. A. Group therapy of phobic disorders by
systematic desensitization. Journal of Abnormal
Lazarus, A. A. The results of behaviour therapy in 126

cases of severe neurosis. Behaviour Research and
Therapy, 1963, 1, 69-79. (a)
Lazarus, A. A. The treatment of chronic frigidity by

systematic desensitization. Journal of Nervous
and Mental Disease, 1963, 136, 272-278. (b)
Lazarus, A. A. Crucial procedural factors in

Therapy, 1964, 2, 65-70.
Lazarus, A. A., & Abramovitz, A. The use of “emotive

imagery” in the treatment of children’s phobias.
Journal of Mental Science, 1962, 108, 191-195.
Lazarus, A. A., & Rachman, S. The use of systematic

desensitization in psychotherapy. South African
Lazovik, A. D., & Lang, P. J. A laboratory

demonstration of systematic desensitization
psychotherapy. Journal of Psychological Studies,
1960, 11,238-247.
Leitenberg, H., Agras, W. S., Barlow, D. H., & Oliveau, D.

C. Contribution of selective positive
reinforcement and therapeutic instructions to
systematic desensitization therapy. Journal of
Leon, PI. V. Reciprocal inhibition: An evaluation of

group procedures with “normal” snake phobic
subjects. Unpublished doctoral dissertation,
University of Tennessee, 1967.
Liddell, PI. Some specific factors that modify tolerance

for environmental stress. Research Publication of
the Association for Nervous and Mental Disease,
1950, 29, 155-171.
Madsen, C. H., Jr., & Ullmann, L. P. Innovations in the

desensitization of frigidity. Behaviour Research
and Therapy, 1967, 5, 67-68.
Mann, J., & Rosenthal, T. L. Vicarious and direct

counterconditioning of test anxiety through
individual and group desensitization.
Unpublished manuscript, University of Arizona,
1969.
Marder, M. The effects of experimenter attractiveness

and negative reinforcement on verbal
conditioning. Dissertation Abstracts, 1961, 22,
1243.
Marks, I. M., & Gelder, M. G. A controlled retrospective
study of behaviour therapy in phobia patients.
British Journal of Psychiatry, 1965, 111, 561-573.
Mason, W. A. Socially mediated reduction in emotional

responses of young rhesus monkeys. Journal of
104.
Melamed, B., & Lang, P. J. Study of the automated

desensitization of fear. Paper read at the meeting
of the Midwestern Psychological Association,
Chicago, May, 1967.
Melvin, K. B., & Brown, J. S. Neutralization of an

aversive light stimulus as a function of number of
paired presentations with food. Journal of
58, 350-353.
Meyer, V., & Crisp, A. H. Some problems in behaviour

therapy. British Journal of Psychiatry, 1966,112,
367-381.
Migler, B., & Wolpe, J. Automated self-desensitization:

a case report. Behaviour Research and Therapy,
1967, 5, 133-135.
Miller, H. R. The role of control of aversive stimulus

termination in the systematic desensitization of
snake phobic subjects. Unpublished doctoral
dissertation, University of Missouri, 1967.
Miller, N. E. Learnable drives and rewards. In S.

psychology. New York: Wiley, 1950. pp. 435-472.
Mitchell, L. E., & Zax, M. The effects of chlorpromazine

on GSR conditioning. Journal of Abnormal and
Moore, N. Behaviour therapy in bronchial asthma: A

controlled study. Journal of Psychosomatic
Research, 1965, 9, 257-276.

Murray, E. J. Learning theory and psychotherapy:

Biotropic versus sociotropic approaches. Journal
of Counseling Psychology, 1963, 10, 250-255.
Nelson, F. Effects of two counterconditioning

procedures on the extinction of fear. Journal of
62, 208-213.
Nisbett, R. E., & Schachter, S. Cognitive manipulation

of pain. Journal of Experimental Social Psychology,
1966, 2, 227-236.
Notterman, J. M., Schoenfeld, W. N., & Bersh, P. J. A
comparison of three extinction procedures
following heart rate conditioning. Journal of
677.
O’Connor, N., & Rawnsley, K. Two types of

conditioning in psychotics and normals. Journal
161.
Patterson, G. R., & Anderson, D. Peers as social

reinforcers. Child Development, 1964,35, 951-
960.
Paul, G. L. Insight vs. desensitization in psychotherapy.

Paul, G. L. Insight versus desensitization in

psychotherapy two years after termination.
Journal of Consulting Psychology, 1967, 31, 333-
348.
Paul, G. L. Two-year follow-up of systematic

desensitization in therapy groups. Journal of
Abnormal Psychology, 1968,73, 119-130. (a)
Paul, G. L. Personal communication. 1968. (b)
Paul, G. L. Physiological effects of relaxation training

and hypnotic suggestion. Journal of Abnormal
Psychology, 1969, 74, 425-437.
Paul, G. L., & Shannon, D. T. Treatment of anxiety
through systematic desensitization in therapy
groups. Journal of Abnormal Psychology, 1966, 71,
124-135.
Poppen, R. L. Counterconditioning of conditioned

suppression. Unpublished doctoral dissertation,
Prince, A. I. Relative prestige and the verbal

conditioning of children. American Psychologist,
1962,17, 378 (abstract).
Proctor, S. Duration of exposure to items and

pretreatment training as factors in systematic
desensitization therapy. Unpublished
manuscript, Indiana University, 1968.
Rachman, S. The treatment of anxiety and phobic

reactions by systematic desensitization
psychotherapy. Journal of Abnormal and Social
Psychology. 1959, 58, 259-263.
Rachman, S. Studies in desensitization—I: The

separate effects of relaxation and desensitization.
251
Rachman, S. Studies in desensitization—III: Speed of

generalization. Behaviour Research and Therapy,
1966, 4, 7-15.
Rachman, S. The role of muscular relaxation in

Therapy, 1968, 6, 159-166.
Ramsay, R. W., Barends, J., Brenker, J., & Kruseman, A.

Massed versus spaced desensitization of fear.
207.
Rescorla, R. H., & Solomon, R. L. Two-process learning

theory: Relationships between Pavlovian
conditioning and instrumental learning.
Rifkin, B. G. The treatment of cardiac neurosis using

systematic desensitization. Behaviour Research
and Therapy, 1968, 6, 239-241.
Rosenthal, R. Experimenter effects in behavioral

research. New York: Appleton- Century-Crofts,
1966.
Rosenthal, T. L. Stimulus modality and aerophobia:

cautions for desensitization therapy. American
Journal of Clinical Hypnosis, 1967, 9, 269-274.
Rosenthal, T. L. Severe stuttering and maladjustment

treated by desensitization and social influence.
130.
Schachter, J. Pain, fear, and anger in hypertensives

and normotensives: A psychophysiological study.
Psychosomatic Medicine, 1957, 19, 17-29.
Schachter, S. The interaction of cognitive and

physiological determinants of emotional state. In
L. Berkowitz (Ed.), Advances in experimental
social psychology, Vol. I. New York: Academic
Press, 1964. pp. 49-80.
Schachter, S., & Singer, J. E. Cognitive, social, and

physiological determinants of emotional state.
Schachter, S., & Wheeler, L. Epinephrine,

chlorpromazine, and amusement. Journal of
128.
Schneider, R. A., & Costiloe, J. P. Effects of centrally

acting drugs on conditioning in man; the
inhibiting and facilitating effects of
chlorpromazine, amobarbital, and
methylphenidylacetate on the conditioned
galvanic skin response. American Journal of the
Medical Sciences, 1957, 233, 418-423.
Schubot, E. D. The influence of hypnotic and muscular
relaxation in systematic desensitization of phobic
behavior. Unpublished doctoral dissertation,
Seager, C. P., & Brown, B. H. An indicator of tension

during reciprocal inhibition. British Journal of
Psychiatry, 1967, 113, 1129-1132.
Sherman, A. R. Therapy of maladaptive fear-motivated

behavior in the rat by the systematic gradual
withdrawal of a fear-reducing drug. Behaviour
Shoben, E. I., Jr. Psychotherapy as a problem in

learning theory. Psychological Bulletin, 1949, 46,
366-392.
Shoben, E. J., Jr. The therapeutic object: Men or

machines? Journal of Counseling Psychology,
1963, 10, 264-268.
Smith, E. W. L., & Dixon, T. R. Verbal conditioning as a

function of race of the experimenter and
prejudice of the subject. Journal of Experimental
Solomon, R. L., & Brush, E. S. Experimentally derived

conceptions of anxiety and aversion. In M. R.
1956. pp. 212-305.
Solomon, R. L., Kamin, L. J., & Wynne, L. C. Traumatic

avoidance learning: The outcomes of several
extinction procedures with dogs. Journal of
302.

69, 202-219.
Spence, K. W., & Taylor, J. A. The relation of

conditioned response strength to anxiety in
normal, neurotic, and psychotic subjects. Journal
of Experimental Psychology, 195.3, 45, 265-272.
Stevenson, H. W. Social reinforcement of children’s

behavior. In L. P. Lipsitt & C. C. Spiker (Eds.),
Advances in child development and behavior. Vol.
2. New York: Academic Press, 1965. pp. 97-126.
Stevenson, I., & Wolpe, J. Recovery from sexual

deviations through overcoming of non-sexual
neurotic responses. American Journal of
Psychiatry, 1960, 116, 737-742.
Strahley, D. F. Systematic desensitization and
counterphobic treatment of an irrational fear of
snakes. Dissertation Abstracts, 1966, 27, 973B.
Terrace, H. S. Stimulus control. In W. K. Honig (Ed.),

Crofts, 1966. pp. 271-344.
Trapold, M. A., & Winokur, S. Transfer from classical

conditioning and extinction to acquisition,
extinction, and stimulus generalization of a
positively reinforced instrumental response.
517-525.
Valins, S., & Ray, A. A. Effects of cognitive

desensitization on avoidance behavior. Journal of
350.
Vinogradov, N. V. Ob okhranitel’ nom i “zastoinom”

tormozheinii u bol’ nykh shizofreniei. [Protective
and “stagnant” inhibition in schizophrenic
patients.] Zhurnal vysshei nervnoi diatel’ nosti
imeni, 1962, 12, 426-431.

Walton, D., & Mather, M. D. The application of learning
principles to the treatment of obsessive-
compulsive states in the acute and chronic
phases of illness. Behaviour Research and
Therapy, 1963, 1, 163-174. (b)
Weinberg, N. H., & Zaslove, M. “Resistance” to

systematic desensitization of phobias. Journal of

Wolpe, J. The systematic desensitization treatment of

neuroses. Journal of Nervous and Mental Disease,
1961, 132, 189-203.
Wolpe, J. Isolation of a conditioning procedure as the

crucial psychotherapeutic factor: a case study.
Journal of Nervous and Mental Disease, 1962, 134,
316-329.
Wolpe, J. Behaviour therapy in complex neurotic

states. British Journal of Psychiatry, 1964, 110,
28-34.
Wolpe, J., & Lazarus, A. A. Behavior therapy techniques.

New York: Pergamon, 1966.
Wolpin, M., & Raines, J. Visual imagery, expected roles

and extinction as possible factors in reducing fear
and avoidance behavior. Behaviour Research and
Therapy, 1966, 4, 25-37.
Wynne, L. C., & Solomon, R. L. Traumatic avoidance

learning: Acquisition and extinction in dogs
deprived of normal peripheral autonomic
function. Genetic Psychology Monographs, 1955,
52, 241-284.
Zeisset, R. M. Desensitization and relaxation in the

modification of psychiatric patients’ interview
behavior. Journal of Abnormal Psychology, 1968
73, 18-24.
8
Aversive Counterconditioning
In the preceding chapter it was shown how
threatening events can be neutralized by
associating them with positive experiences. Some
forms of psychological dysfunctions reflect a
converse problem, resulting from the fact that
certain activities or objects, which are potentially
harmful or socially prohibited, have acquired

unusually potent reinforcing value for the
individual. These deviations generally take the
form of addiction to various drugs or intoxicants,

or marked sexual attraction to inappropriate
stimuli as manifested in transvestism, fetishism,
exhibitionism, homosexuality, and other aberrant

sexual expressions. Attempts are sometimes made
to control such behavior by developing
conditioned aversion to the positively reinforcing
stimuli by pairing them contiguously with
negative experiences. A variety of methods has

been devised for producing conditioned aversion.
These have been applied mainly to persons who

wish to gain control over intractable behavior
which can produce serious long-term
consequences for them.
The classical conditioning approach to the

elimination of aberrant response patterns should
be distinguished from aversive instrumental
conditioning, discussed in Chapter 5, in which
response tendencies are inhibited by having
punishing consequences follow the occurrence of

the behavior. In the classical conditioning
paradigm, aversive events are stimulus correlated
for the purpose of altering the valence of stimuli,
whereas in the instrumental procedure, negative
outcomes are response correlated so as to inhibit
the performance of deviant responses.
Punishment that is made contingent on the

appearance of undesired behavior may
temporarily suppress its occurrence, but if the
positive stimuli that evoked the behavior are
absent during the punishment they are likely to
retain their attractiveness. Although the classical

and the instrumental approaches can be easily
differentiated operationally, the latter procedure
often produces some classical conditioning effects.
When a given behavior is punished, stimuli arising
from the punished response itself, and

environmental events present at the time, may
become endowed with negative properties. Thus,

for example, in treatment programs in which
aversive stimulation is presented to alcoholics in
the act of drinking alcoholic beverages, and to

transvestites while they are putting on ladies’
undergarments, the designation of the procedure
is somewhat arbitrary.
Development of Conditioned Aversion and

Avoidance
The basic process involved in the development
of conditioned aversion and avoidance has been
discussed in some detail in Chapter 5. Briefly, an

object or an activity that is repeatedly associated
with aversive experiences acquires some of the
negative properties of the aversive stimulus. As

long as the negatively conditioned stimuli retain
their aversive effects, the individual will be

inclined to avoid them.
In most laboratory investigations of
conditioned aversion, formerly neutral stimuli are
endowed with negative properties. Of greater
relevance to issues of behavioral change, however,
are studies showing that aversion can be
established to strongly preferred stimuli (Garcia,
Kimeldorf, & Koelling, 1955; Masserman, 1943;
Peacock & Watson, 1964). Although the

phenomenon has been demonstrated, there has
been no systematic research to determine the
optimal conditions for creating persistent
aversions.
UNCONDITIONED STIMULI IN AVERSIVE

COUNTERCONDITIONING
Until recent years, most applications of
aversive counterconditioning have utilized
nauseous pharmacological agents to create the

requisite negative conditions. The specific
procedures followed in this form of therapy are
well illustrated by Lavin, Thorpe, Barker,
Blakemore, & Conway (1961) in their treatment of

a male transvestite. The client, a 22-year-old
married truck driver, frequently dressed in

women’s clothing, a pattern of behavior that began
in early childhood when he occasionally wore his
sister’s dresses. After puberty he experienced
orgasms when dressed in female apparel, and

masturbated to the accompaniment of transvestite
fantasies. Cross-sex dressing had become a
powerful sexual reinforcer, but the fear of serious
legal consequences, since he occasionally

appeared in public in feminine attire, and
pressures from his wife, led the client to seek
treatment.
This case illustrates careful application of
several of the requirements for successful
counterconditioning. In selecting the conditioned
stimulus it is necessary to determine what specific

aspects of the stimulus events are most positively
valenced. In this particular case, for example, it

was noted that the texture or feel of women’s
clothing produced no sexual excitement, but

viewing himself in a mirror dressed in feminine
apparel was highly stimulating. It was decided,
therefore, to employ a series of colored slides of
the client in various stages of female dress,

ranging from panties only to fully clothed, in order
to condition negative reactions to all aspects of
cross-sex dressing. In addition, a tape recording of
the client describing the transvestite behavior
depicted (e.g., “I have now put on and am wearing

a pair of ladies’ panties”) accompanied each slide.
The auditory cues were primarily designed to
augment the conditioned stimulus, to ensure the
presence of the stimulus when the client might not
be attending to the pictorial displays, and to

facilitate generalization effects.
Intramuscular injections of apomorphine

hydrochloride served as the unconditioned
stimulus, although occasionally emetine
hydrochloride was substituted after the client’s
tolerance to apomorphine had increased. As soon
as he reported feeling nauseous, a slide was
projected onto the screen and simultaneously the
tape recorder was started; the pictorial and

auditory stimuli were promptly terminated after
emesis. The client received 66 emetic trials, one
every two hours, for six days. As aversive
conditioning progressed, transvestism assumed
sufficiently unpleasant properties that the client

no longer derived any erotic satisfaction from
cross-sex dressing. A series of follow-up,
interviews with both the client and his wife
revealed complete cessation of transvestism, and
continued indifference to clothes that had

previously excited him.
Later sections of this chapter review numerous

studies in which emetic drugs, usually
apomorphine or emetine, have been employed to
create aversion. There are, however, many
disadvantages in pharmacological procedures that
place certain limitations on their utility and
applicability.
Stimuli contiguous with the onset of aversive

experiences gradually acquire negative properties,
whereas stimuli associated with reduction or
termination of discomfort may actually acquire a
palliative function (Mowrer, 1960). One must,
therefore, exercise precise control over the timing

and sequence of stimulus variables in order to
ensure that positively reinforcing events are
accruing negative, rather than discomfort-relief,
value. The optimal temporal requirements are

difficult to achieve with the use of biochemical
agents because of the gradual, and often
unpredictable, onset of physiological reactions. As

a consequence, the presentation of conditioned
stimuli is typically delayed until some time after
persons begin to exhibit nauseous reactions or
other signs of mounting discomfort. In addition to
the temporal issue, it has been shown (Fromer &
Berkowitz, 1964) that aversive stimuli with a
gradual onset produce significantly weaker

aversion responses than those with a sudden
onset. Since there is no effective way of
terminating abruptly drug- induced aversive
states, the nauseous reactions are needlessly
prolonged. Moreover, because they also tend to

subside gradually, the therapist lacks reliable
criteria for timing the withdrawal of conditioned
stimuli to prevent their association with the
reduction of discomfort.
Apart from the problems created by

inadequate control over the rate of onset,
duration, intensity, and recovery time of drug

activity, undesirable physiological side-effects are
sometimes produced, requiring the administration
of additional medicinals. Some of the actions of
drugs that accompany, but are unrelated to, the
effects for which they are being administered may,
of course, impair the conditioning process itself.
Central depressants have been shown to decrease

conditioning, whereas stimulant drugs facilitate
the formation of conditioned responses (Franks,
1966). Emetine is, therefore, generally favored
because apomorphine has sedative effects. When
the latter drug is employed, stimulants such as

dexamphetomine sulphate (Lavin et ah, 1961) or
caffeine (Freund, 1960) are usually administered
to counteract its depressant action. The ideal
pharmacological agent would be a stimulant drug
that produces brief and readily controllable emetic

responses and is free of adverse side effects.
Although emetine is more suitable since it does

not possess depressant properties, its frequent
administration may result in cardiovascular
sequelae (Barker, Thorpe, Blakemore, Lavin, &

Conway, 1961). Finally, the use of pharmacological
methods necessitates hospitalization and imposes
certain limitations on the manner in which the
conditioned stimuli can be presented.
Aversive electrical stimulation has been
increasingly employed in aversion treatment,

mainly because it permits precise control over
conditioning variables (Rachman, 1965). Aversive
shocks can be presented and terminated abruptly;

they can be easily varied in duration and intensity;
also, except in the case of cardiac disorders, they
produce no adverse physiological effects. This
increased control makes it possible not only to
arrange optimal temporal relationships between

conditioned and unconditioned stimuli but also to
conduct numerous aversion trials within the same
session. The treatment can even be self-
administered in the naturalistic situations in
which the problem behavior is evoked.
During conditioning trials moderately severe
shocks are administered either to the forearm
(Kushner & Sandler, 1966) or to the feet

(Blakemore, Thorpe, Barker, Conway, & Lavin,
1963) of the client in conjunction with the elicitors

of the undesired behavior. A number of shock
trials are presented during each of the sessions,
which are scheduled over a period of a week or
two.
It is sometimes difficult to introduce into
treatment the stimulus events for deviant
behavior in the forms and intensities in which they
are ordinarily encountered in everyday situations.

Consequently, aversion reactions may be
conditioned to verbal, pictorial, or imaginal
representations of the actual stimulus objects,
with the hope that sufficient transfer will occur to
inhibit approach tendencies to the real-life
counterparts. McGuire & Vallance (1964) have
devised a portable electric stimulation apparatus
that makes it possible for the client, after some
preliminary training, to carry out his own

counterconditioning in naturalistic settings
whenever he feels impelled to perform the deviant
behavior.
The utilization of the self-conditioning
procedure is illustrated in the treatment of a 25-

year-old graduate student who had been
masturbating to fetishistic fantasies with
considerable guilt about three times a day for 10
years. The client began aversion therapy after

having participated in conventional treatment,
which had reduced neither his fetishistic
masturbatory behavior nor the attendant guilt

feelings. In the initial phase of treatment, the client
was asked to produce the usual fantasies and to
signal by raising his hand when the fetish objects
were clearly visualized, at which time a shock was
administered. At later sessions, when the client
was unable to conjure up the sexually provocative
fantasies, photographs of persons attired in the

fetishistic clothing were employed as the
conditioned stimuli. In addition, he was
encouraged to use the conditioning apparatus at
home whenever he felt instigated to masturbate.
The fetishistic fantasies were completely

eliminated within a short period, the incidence of
masturbatory behavior was substantially reduced,
and on the occasions when the client did
masturbate, for the first time in his life this
behavior was accompanied by heterosexual

fantasies. The authors report favorable outcomes
in the use of the self-conditioning procedure for

reducing obsessional ruminations, obesity,
smoking, and alcoholism. Similarly, Wolpe (1965)
was able to achieve temporary control over drug-

addictive behavior in a physician by having him
administer to himself a shock from a portable
apparatus whenever craving for the drug
appeared.
In the methods discussed thus far, aversive
counterconditioning is accomplished through

repeated association of pharmacologically created
nausea or unpleasant electrical stimulation. A
much more interesting approach, that has

numerous advantages over noxious physical
stimuli, involves symbolically induced aversion. In
this form of counterconditioning, positively
valenced events are repeatedly paired with strong
feelings of nausea and emetic responses which are

verbally induced. The negative verbal contents are
usually drawn from disagreeable, painful, or
revolting experiences that have previously arisen
either in connection with the pleasurable objects
and activities or in other contexts. As in the other
paradigms, the conditioning trials are continued
until the formerly positive stimuli alone elicit
feelings of revulsion.
Miller (1959, 1963) provides several

illustrations of the successful use of symbolic
aversion methods in modifying homosexuality and
alcoholism. Most homosexuals have experienced
specific disgust reactions, at one time or another,
in intimate relationships with certain male

partners. During treatment hypnotically revivified
nauseous reactions, which the client has
experienced in previous homosexual contacts, are
repeatedly associated with visualized homosexual

practices involving current male companions. One
client, for example, had felt strong revulsion to the
smell and taste of urine and stale perspiration

while performing fellatio with an uncircumcised
male. These past experiences were, therefore,
employed as aversive verbal stimuli in the
treatment. The author reports that after several
sessions the client became nauseated by his male
lovers and eventually broke off all homosexual
contacts.
Avoidance responses established in this
manner are typically reinforced by supplementary
conditioning trials at monthly intervals for the
first year. In addition to attaching negative valence
to homoerotic responses and love objects,

symbolic counterconditioning procedures are also
employed to enhance the positive reinforcing
value of heterosexual stimuli. In the preceding
illustrative case the client participated

concurrently in a number of sessions in which
certain feminine attributes that he found desirable
were hypnotically augmented and associated with

women. Although on the few occasions when the
client had formerly dated girls he had selected
masculine types, following the
counterconditioning treatment he was attracted
to, and dated, women possessing notable feminine
attributes.
In aversion treatment of alcoholism, which is

reviewed later, verbally revivified hangover
experiences are contiguously associated with the
smell and taste of alcoholic beverages. Essentially
similar symbolic aversion techniques have been
applied on a limited basis to the treatment of

obesity (Cautela, 1966), alcoholism (Abrams,
1964; Anant, 1967), addictions, and sexual
perversions (Kolvin, 1967). The major advantages
of this technique are that it has no adverse side

effects, it is highly adaptable, and people can be
taught to administer the treatment to themselves
in the naturalistic situations in which their

problem behavior is apt to arise.
PROCESS GOVERNING CONDITIONED AVERSION
The manner in which aversive procedures are
employed and the durability of resultant aversions
are likely to be influenced to a considerable extent
by one’s conceptualization of the mechanism
through which aversive stimulation produces its

effects. Most traditional accounts of
counterconditioning convey the impression that,

as a result of paired association with negative
experiences, formerly positive stimuli directly and
automatically evoke aversive reactions. The
temporal relationship between stimulus events is

therefore considered to be a major determinant of
the strength and durability of conditioned
aversions. As will be shown in the following
chapter, this nonmediational view is at variance
with certain experimental findings. It has been

demonstrated, for example, that both conditioned
autonomic and avoidance responses promptly
disappear when shock electrodes are removed or
subjects are merely informed that a given stimulus
will no longer be accompanied by painful
stimulation. Moreover, conditioned emotional
responses can be established cognitively without

the immediate support of external aversive
stimulation. Since in aversion treatments the

negative experiences associated with pleasurable
activities are arbitrarily rather than intrinsically
related to the behavior, individuals can readily
discriminate that in everyday life the same

activities will not only be unaccompanied by
unpleasant consequences but may, in fact, prove
highly rewarding. Given cognitive control over

conditioning effects and markedly different
situational contingencies, one might expect

conditioned aversions to be readily extinguishable
and to show little transfer from treatment to real-
life situations. On the other hand, there is

considerable evidence that established revulsions
usually generalize across situations and that they
can be relatively long lasting.
An alternative view of counterconditioning

effects is that external stimuli acquire the capacity
to activate a self-stimulation mechanism which, in
turn, produces the aversive reactions. Thus, for

example, after a person has repeatedly
experienced strong nausea in conjunction with
alcoholic beverages the mere sight or smell of

alcohol leads him to revivify his past nauseous
experiences. In this conceptualization aversive
reactions are, in large part, self-induced rather
than automatically evoked. If the aversive self-
stimulation established through

counterconditioning is potent enough, a person
may be able to counteract the disposition to
engage in deviant behavior by symbolically
reinstating nauseous reactions whenever the need
arises.
The manner in which counterconditioning is
applied will differ in several important respects
depending upon whether it is viewed as a form of

automatic immunization or as a technique of self-
control. In the former case, methods are favored

that permit precise management and split-second
timing of stimulus events. In the latter approach,
on the other hand, the procedures are designed to
develop strong and readily recreatable aversions

to certain objects or activities. For this purpose,
verbal and pharmacological agents may
regenerate more natural and symbolically

reproducible aversions than electric shock, the
physiological manifestations of which are

relatively subtle. An optimal procedure might
initially involve the combined use of verbal
induction and either emetic drugs or shocks to

create vivid aversive reactions. In subsequent
sessions, however, verbal stimuli alone would be
used as the conditioning agent although they
might be paired occasionally with emetic drugs or

shock to preserve their potency. After an aversive
self-stimulation system had been established in
sufficient strength, individuals would be

instructed to avoid engaging in the deviant
behavior by deliberately inducing nauseous

reactions. When conditioned aversions are
regarded as self-induced reactions rather than as
automatic products of stimulus pairings, the
change agent assumes greater responsibility for
arranging positive incentives to ensure that

individuals utilize this potentially effective means
of self-control. In a comprehensive treatment

program this practice would, of course, be used in
conjunction with other methods of self-control, as
well as procedures designed to reduce the
instigation to engage in the deviant behavior.
When some persons who have undergone
aversive treatment later revert to their deviant
activities these outcomes are often attributed to
deficiencies inherent in the conditioning

procedure itself. Faulty timing and sequencing of
stimulus events, selection of inadequate aversive
agents, and conditionability deficits are typically
invoked as explanatory factors. Similarly,
recommendations for enhancing the efficacy of

this method single out conditioning variables.
These include the use of intermittent schedules of
reinforcement, continuation of the trials for a
sufficient period to ensure overlearning, and
inclusion of booster treatments at periodic

intervals after the formal program has been
discontinued (Eysenck, 1963). There exists

suggestive evidence (Voegtlin, Lemere, Broz, &
O’Hollaren, 1942) that conditioned aversion and
avoidance can be successfully maintained through

periodic reconditioning trials. Portable devices
that permit self-administration of aversive stimuli
and the judicious use of symbolically generated
consequences would undoubtedly also reduce the

disposition to engage in deviant behavior.
However, there is little reason to expect that

intermittent reinforcement would increase the
durability of aversive reactions. Partial
reinforcement retards the rate of extinction by
reducing discriminability of the occasions when

customary consequences will or will not occur
(Spence, 1966). Administration of reinforcements
in an unpredictable manner may produce stable

conditioned responses during treatment, but the
conditions of reinforcement prevailing in

treatment situations and in everyday life differ
markedly and are easily distinguishable. The
situational change would ordinarily result in a

rapid decrement in responsiveness unless
cognitive functions were utilized in a self-
reinforcing system that could endure long after
conditioning influences were discontinued.
Induced aversions are generally less durable
than approach behavior that has been restored by
eliminating inhibitions and anxieties. This

differential reversion rate is most likely
attributable to the different maintaining
contingencies associated with these two forms of

behavior. In the case of behavioral conditions
involving fear, removal of avoidance responses
enables people to engage in many potentially
rewarding activities which, in turn, further
strengthen the newly established behavior. Initial

behavioral changes are thus apt to set in motion a
positive reciprocal influence process. By contrast,
sexual, appetitive, and addictive behavior produce
powerful immediate reinforcement, whereas any
aversive consequences that do occur are typically
long delayed. These types of controlling conditions
result in behavior that is not only refractory to
change but also easily reinstatable because of the
positive reinforcing effects that accompany its

recurrence. Aversion treatments attempt to
forestall the recurrence of deviant behavior by
developing immediate aversive reactions to the
stimuli that typically evoke the behavior.
The major value of aversion procedures is that

they provide a means of achieving control over
injurious behavior for a period of time during
which alternative, and more rewarding, modes of
response can be established and strengthened.

Used by itself, this method may bring about only
temporary suppression of deviant tendencies. The
answer to the reduction of reversion rates,

therefore, lies in the scope of the treatment
program employed rather than in variables
operating within the conditioning paradigm. This
issue will be discussed more fully in the context of
the modification of specific behavior disorders.
STIMULUS EVENTS IN AVERSION TREATMENTS
The consequences of selecting inappropriate
stimulus events for aversion treatment are
considerably more serious than those for
desensitization approaches. In the latter case,

deconditioning an irrelevant stimulus merely
prolongs treatment, whereas the former

procedures can establish needless aversions and
avoidant behaviors. Although the risks associated
with aversive counterconditioning may be greater,

in the types of behavior disorders that are
customarily treated by these methods the
rewarding objects are easily identified (e.g.,
alcoholic beverages, specific drugs, fetishistic
objects, cross-sex apparel, homoerotic stimuli).
The major decisions, therefore, concern the
variety of conditioned stimuli and the forms in
which they will be presented.
If inadequately applied, aversive procedures
can have generalized inhibitory effects. Control of

generalization is therefore an important issue. A
change agent is faced with the task of presenting
aversive stimuli sufficiently intense to arouse

strong aversion, while simultaneously confining
generalization to a particular class of activities. To
achieve this dual objective, the stimulus events

must be distinctive and carefully delimited.
Applications of aversion therapy to the
modification of homosexuality will serve to
illustrate the latter point. Typically, pictures of
nude, semi-nude, and fully clothed males are

contiguously associated with unpleasant
experiences. In these instances, aversions are

established to a general stimulus class so that
some undesired generalization may occur, but the

most relevant stimulus events (i.e., specific
homosexual practices) are not included in the
conditioning process. Hence, there is no assurance
that the negative conditioning will necessarily

transfer extensively from pictures of men to
specific homosexual responses toward them.
Clearly, stronger but circumscribed transfer would
be achieved if the stimulus presentations
contained all aspects of homosexuality that the

treatment is designed to eliminate.
The problem of negative transfer can be even
more complicated in cases where the existing
behavior is entirely satisfactory, but the evoking

stimuli are so bizarre and disturbing to others that
attempts are made to nullify their arousal
capabilities through aversive procedures. This

situation arises most frequently in the treatment
of heterosexual fetishists and transvestites who, in
order to obtain an erection and to engage in sexual
intercourse, must wear their wives’ clothing
(Blakemore et al., 1963), rubberized mackintoshes
(Oswald, 1962), be kicked with rubber boots or
high-heeled shoes (Marks, Rachman, & Gelder,

1965). Such unique modes of erotic arousal create
serious marital conflicts, even conditioning
revulsions in sexual partners. A similar problem
exists when efforts are made to negate the sex-
arousing value of self-flagellating, fetishistic, and

sadistic fantasies that arouse masturbatory or
heterosexual behavior without suppressing the
behavior itself. It is of considerable import that in
many of the published cases more appropriate
sexual fantasies emerge as the arousal potential of

bizarre elicitors is eliminated, and desirable sexual
behavior is either maintained at its original level

or further enhanced (Blakemore et al., 1963;
Cooper, 1963; Kushner & Sandler, 1966; Marks &
Gelder, 1967; McGuire & Vallance, 1964; Oswald,

1962; Raymond, 1956).
The direction and extent of generalization of
aversion effects can be regulated through a
program of differential reinforcement in which

undesirable events are repeatedly associated with
negative experiences, while the desired ones are

paired with either rewarding or no adverse
consequences. Verbal labeling can also be utilized
effectively both to delimit what is being negatively
conditioned and to enhance the most relevant

elements in the stimulus complex. To continue
with the above example, pictorial stimuli could be
supplemented with descriptions of the client,

recorded in his voice, engaging in anal and
interfemoral intercourse, oral-genital relations,

mutual manual masturbation, and other forms of
erotic responsiveness to males.
It seems reasonable to expect that, as in the
case of desensitization, the more inclusive the
representation of relevant stimuli in the aversion
program, and the more similar they are to real-life
events, the more potent will be the
counterconditioning outcomes. This practice is

generally followed in the treatment of fetishism,
transvestism, and the various addictions because,
in each of these disorders, the attractive stimuli
can be easily included in the treatment
procedures. To further enhance the visual stimuli,

tape recordings are sometimes played in which
clients describe themselves engaging in the
deviant behavior, or they are instructed to develop
mental imagery involving the undesired activities.
In aversion treatment of compulsive gambling

behavior, Barker & Miller (1966) and Goorney
(1968) employed newspaper items, sound

broadcasts, televised presentations, and actual
gambling devices that ordinarily served as the
evocative stimuli.
There is some evidence based on the
modification of alcoholism (Lemere & Voegtlin,
1940; Quinn & Henbest, 1967) and sexual
deviations (Marks & Gelder, 1967) that when a

single stimulus is employed in aversive
counterconditioning, the resultant aversion
reactions may be highly specific to that particular
stimulus class. Consequently, a wide sampling of
stimuli is employed when, as in the treatment of

alcoholism, a generalized aversion is required.
Sexual Deviance
There exists considerable cross-cultural
evidence (Ford & Beach, 1951) that in societies in

which homosexuality and transvestism are
socially disapproved, sexual inversions rarely
occur; by contrast, cross-sex dressing, anal
intercourse, oral-genital contacts, and mutual
masturbation are regularly practiced by most
members of societies in which such behavior is
fully sanctioned and positively reinforced.
Whereas in our own society inflicting pain on a
sexual partner is regarded as a sadistic perversion,

in other cultures pain cues, resulting from
aggression that normally accompanies coital
activities, serve as powerful sexual reinforcers
(Holmberg, 1946; Malinowski, 1929). There are
also wide cultural variations in the physical

attributes and adornments that become culturally
conditioned sexual stimuli. What has been
endowed with erotic arousal properties in one
society—corpulence, skinniness, upright
hemispherical breasts or long pendulous ones,

shiny white teeth or black pointed ones, deformed
ears, nose, or lips, broad pelvis and wide hips or

narrow pelvis and slim hips, light or dark skin
color—may be neutral or highly repulsive to
members of another social group. These cross-

cultural data showing the range of preferred
sexual reinforcers are striking testimony of the
influential role of social learning in the
development of sexual behavior that may be

judged deviant by some social group.
Although our society imposes severe social and

legal prohibitions against some forms of behavior
that are believed to have sexual implications,
certain members may nevertheless experience

unusual reinforcement and modeling influences
serving to promote and to maintain deviant sexual
behavior. A major obstacle to the understanding of
human sexual deviance is that, for ethical reasons,
experimentation designed to identify the

conditions governing sexual phenomena cannot be
conducted. Consequently, the search for the
relevant controlling variables must rely on
naturalistic studies. A number of clinical reports
have been published containing data that illustrate
the social learning processes whereby culturally
inappropriate stimuli and responses acquire
unusually strong sexual reinforcing properties.
Litin, Giffin, & Johnson (1956) describe the

development of transvestism in a young boy who
continually dressed up in his mother’s clothes,
including cosmetics and jewelry, exhibited almost
complete feminine-role behavior, and even
adopted a girl’s name which his mother suggested.

Cross-sex dressing first appeared following an
episode in which the mother, in response to her
son’s comment that she looked pretty in her new
shoes, hugged him and offered him her old shoes.

He wore these shoes daily, eliciting considerable
maternal approbation. The mother continued to
encourage and reward sex-inappropriate behavior

with demonstrations of affection and approval,
while the grandmother and neighbors
supplemented the mother’s training in
transvestism by supplying the boy with generous
quantities of old shoes, hats, purses, bridal veils,
and other female apparel. When the boy’s inverted
behavior met with disapproval from other

persons, the mother attempted some
discrimination training by informing her son, “You
must never dress like that in front of company,
only in front of the family.” In a study of the
mothers and wives of 32 transvestites, Stoller

(1967) found that the subjects were initiated into
transvestism by being dressed in girls’ clothing or
highly rewarded whenever they dressed
themselves in feminine apparel. The transvestite
behavior was further elaborated by mothers and

wives who devoted many hours to teaching the
subjects how to dress in feminine clothing, how to

apply cosmetics, and how to behave as women.
Litin, Giffin, & Johnson (1956) depict how a
mother actively conditioned voyeuristic behavior
in her son by sleeping with him, and by being
physically and verbally seductive while appearing
nude before him. When the boy was six years old,
the mother had shown him her vagina a number of

times, but she later discontinued her physically
seductive behavior after the son suggested that
they engage in sexual intercourse. The boy’s
strongly established voyeuristic behavior had
generalized to the maid and other persons;

eventually he was apprehended by the police
while peeping from a ladder into neighborhood
bedrooms.
Generalization of strongly reinforced

homoerotic responses is illustrated in the case of a
16-year-old girl whose mother would lie in bed
with her, encouraging mutual stroking of the

breasts, and other erotic play. The mother sought
psychiatric advice when she became jealous of her
daughter’s homosexual attachment to a teacher.
Homosexual patterns of behavior are not always
fostered in such a blatant fashion. In many
instances, the father serves as an inadequate
model for masculine behavior and the distribution

of social power within the family constellation
promotes cross-sex modeling (Whitener & Nikelly,
1964). Sex-appropriate behavior is nonrewarded
and, when the parents encourage peer
relationships, homosexual associates tend to be

favored (Fleck, 1960; Kolb & Johnson, 1955).
A mother’s active reinforcement of deviant
sexual behavior is again evident in a 17-year-old
exhibitionist described by Giffin, Johnson, & Litin

(1954). The mother often showered with her son,
engaged in endless sex discussions, enjoyed
exhibiting herself to him, and delighted at looking

at the boy’s nude body, particularly his “beautiful
masculine endowment.” A dress fetish was
similarly conditioned in a ten-year-old boy by a
mother who responded demonstratively
whenever the son stroked her dress or
complimented women on their attire (Johnson,
1953).
That erotic experiences can endow formerly
neutral stimuli with sexual arousal properties is
supported by results of an interesting study by
Rachman (1966), designed to create a mild fetish
under laboratory conditions. After a photograph of

women’s boots was repeatedly associated with
slides of sexually stimulating nude females,
subjects exhibited sexual arousal (as measured by
penile volume changes) to the boots alone, and

generalized the conditioned sexual responses to
other types of black shoes. In accord with these
findings, McGuire, Carlisle, & Young (1965) report

that deviant sexuality often develops through
masturbatory conditioning in which aberrant
sexual fantasies are endowed with strong erotic
value through repeated association with
pleasurable experiences from masturbation. The
details of their interpretive scheme, and its
therapeutic implications, are discussed more fully

later.
The foregoing cases represent a small sample
of those documented in the reports to which
reference has been made. Three social learning
variables emerge from these naturalistic studies as

important determinants of deviant sexual
behavior. The first consideration involves the
degree to which parents themselves model
homosexual, transvestite, fetishistic, or

exhibitionistic behavioral patterns in either
blatant or attenuated forms. Second, once the
responses are elicited, either by direct instigation

or modeling, they are endowed with exaggerated
sexual significance and strong positive valence.
This may derive from repeated association with
relatively intense, affectionate demonstrativeness,
with close physical intimacy, or from
masturbatory conditioning. Indeed, for many of
the children, inappropriate cues and responses

had acquired strong positive valence, and in some
cases had resulted in well-developed patterns of
sexual behavior long before the onset of
pubescence. Third, parents tend to maintain
children’s deviant sexual responses on an

instrumental basis over a long period, both
through direct and vicarious reinforcement.
Sexual responses that have acquired strong
positive valence can also become self-reinforcing

through their stress-reducing capabilities. The
diminution of aversive emotional states by
engaging in sexual behavior may reflect the

operation of two somewhat different
psychological processes. First, sexual activities can
produce sufficiently intense pleasurable
experiences to contravene feelings of
apprehension or frustration. Secondly,
performance of sexual behavior also changes the
stimulus situation by temporarily directing the

person’s attention away from stress-provoking
events. This attentional shift may in itself produce
substantial anxiety relief.
Evidence of the stress-reducing function of
deviant sexual responses is disclosed in a report

published by Cooper (1963). The client
characteristically experienced heightened sexual
arousal in response to tactual sensations from
silken garments. Dressed in women’s clothing, he

would frequently masturbate to the point of
orgasm by straddling a chair. Although initially the
transvestite behavior served a sexual function, it

later acquired, through an accidental contingency,
generalized anxiety-terminating value. One day
while experiencing considerable apprehension
regarding a scholastic examination, the client
discovered that cross-sex dressing resulted in an
abrupt diminution in anxiety. Thereafter, a wide
range of stressful situations elicited transvestite

behavior. Several other authors (Bond &
Hutchison, 1964; Conn, 1954) have noted a similar
behavioral sequence in which mounting tension is
abruptly reduced by the execution of deviant
sexual responses. These observational data would

seem to indicate that deviant sexual behavior of an
unusually persistent character is probably
maintained not only by sexual reinforcement but
also by its tension-reducing effects.
In applications of aversive counterconditioning

to sexual disorders, attempts are made to reverse
the sexual arousal value of appropriate and

inappropriate stimuli through differential
conditioning procedures.
EFFICACY OF CONDITIONED AVERSION

APPROACHES WITH SEXUAL DISORDERS
There have been no controlled studies of the
relative efficacy of aversion treatments for
modifying deviant sexual behavior, but numerous

case studies are available which have some
evidential value. These reported outcomes, while

very interesting, must be accepted with
reservation for several reasons. Many of the
treatment programs involve a combination of
methods, which makes it difficult to assess the

relative contribution of the aversion procedures.
Moreover, changes in sexual response patterns are
typically measured in terms of clients’ self-reports,

although substantiating information is obtained
from spouses and close associates whenever

possible.
Several objective tests of sexual arousal have
been developed which are beginning to be
employed to measure progress during treatment,
the degree of alteration in sexual arousal patterns
at the completion of therapy, and their stability
over time. One of the laboratory procedures,
originally devised by Freund (1963; Freund,

Sedlacek, & Knob, 1965), measures, by means of a
transducer, penile volume changes in response to
pictures of males and females of varying ages, or
to other erotic objects. Several validational studies
(Freund, 1967a, b; McConaghy, 1967) have shown

that this measure can successfully differentiate
between persons with homo- or hetero-erotic
preferences for adults, adolescents, and children.
Another quantitative index of the attraction

value of visual stimuli is provided in terms of
changes in pupillary size. In this technique (Hess,
1968) a subject’s eye is filmed at the rate of two

frames per second while he views test items
alternated with control patterns equated for
brightness. The film is later magnified and the
diameter of the pupil size is measured manually
frame by frame or electronically with photocells. It
has been shown that pupillary dilation occurs in
response to stimuli that have high interest value,

whereas negatively valenced stimuli produced
pupillary constriction. Findings of a pilot study
(Hess, Seltzer, & Shlien, 1965), in which pupillary
responses of homosexuals and heterosexuals were
measured to pictures of males and females, verify

that this procedure can differentiate sexual
preferences with considerable accuracy. All
heterosexuals displayed larger responses to
pictures of women than to pictures of men,
whereas homosexuals showed the reverse pattern.

The authors suggest that standard sets of pictorial
stimuli could be devised to provide reliable

measures of sexual attraction. However, aversive
conditioning may be difficult to evaluate through
pupillometric measures because pupillary

responses could represent either sexual interest
or anxiety arousal conditioned to sexual stimuli.
This interpretive problem does not arise with
respect to erectile responses.
Although it is customary to question the value
of case studies, because sexual deviations are

exceedingly resistant to change, favorable results
cannot be casually dismissed. Deviant sexual
patterns rarely change “spontaneously,” and they

have proved equally unresponsive to the planned
efforts of psychotherapists employing varied
strategies (Curran & Parr, 1957; Woodward,
1958). The fact that most persons receiving
aversion therapy have undergone traditional

forms of treatment for extended periods of time
without any degree of success makes it unlikely
that the behavioral modifications accompanying
aversive counterconditioning are attributable to
interpersonal and cognitive variables that are
prominent in conversational therapies. The nature
and brevity of the treatment further limit the
opportunity for social factors to exert a strong
influence on the outcomes.
The individual case method can be effectively
employed to assess the functional relationship
between treatment variables and behavioral
changes. It was previously shown how
intrasubject replication furnishes a means of

evaluating the role of reinforcement variables in
behavioral processes. In their detailed
measurement of changes accompanying aversive
counterconditioning, Marks & Gelder (1967;

Gelder & Marks, 1969) provide an excellent model
that other researchers would do well to emulate. A
large number of transvestites and fetishists

received aversion treatment in which shocks were
administered while they carried out their deviant
behavior or imagined themselves performing the
same activities. In each case, the process of
treatment was studied by recording changes in
sexual arousal through the use of a penis
transducer. The clients’ attitudes to the objects of

their deviation, to sexual intercourse, to
themselves, and to the therapist were also
measured by the evaluative scales of the semantic
differential. In addition, progressive changes in the
latency of sexual images and accompanying

erectile responses were recorded. Different
aspects of the clients’ deviant behavior were
modified one at a time to determine whether the
observed changes reflected the operation of
general factors present in any treatment or the

specific counterconditioning procedures.
Figure 8-1 depicts changes during the course

of aversion therapy in both the frequency and
latency of erectile responses of a transvestite to
different feminine garments that he regularly used
in cross-dressing. At the beginning of treatment he
1646
Figure 8-1. Changes in the frequency and latency of erectile responses of a transvestite to
different feminine garments as they were included serially in aversive
counterconditioning. Marks & Gelder, 1967.
was highly aroused sexually by all of the items.
The aversive procedures were then applied to one
garment at a time beginning with panties. After

approximately 20 trials the client no longer
showed any erectile responses to panties, but he
responded with undiminished sexual excitement
to the remaining articles of clothing. As the other
items were serially counterconditioned they also

lost their sex- arousing capacity. It is important to
note, however, that the client maintained high
sexual responsiveness to appropriate heterosexual
stimuli after deviant sources of sexual arousal
were eliminated. The specific sequence of

behavioral changes obtained by repeated
application of aversive stimulation provides

convincing demonstrations that the alteration in
sexual arousal was indeed produced by the
conditioning procedure. In the cases studied,

sexual reconditioning was followed by
corresponding elimination of deviant sexual
desires and activities.
The findings of the above study not only attest
to the efficacy of aversive conditioning, but they
also help to clarify the stimulus function of

symbolic processes and their modification. Prior
to treatment, imagery of the deviant behavior
elicited strong erectile responses. As these

fantasies were repeatedly associated with
unpleasant experiences the latency of the images
increased until eventually they could be produced
only with considerable difficulty (Figure 8-1).
Moreover, the erections which originally

accompanied the fantasies diminished gradually
with successive trials and eventually disappeared
even when the images could be produced. The
selective effect of the treatment is again shown in
the fact that clients were able to conjure up
arousing heterosexual imagery with ease.
The pattern of attitudinal change
corresponded closely to the modifications
achieved in sexual arousal. Sexually attractive

garments were selectively devalued as the specific
objects were negatively conditioned, whereas self-

attitudes, and general attitudes toward sexual
intercourse and toward therapists, changed little
during treatment. The consistent selective changes
achieved in erectile responses, in the arousal

properties of deviant fantasies, in attitudes, and in
overt sexual behavior, would indicate that in cases
where aversive conditioning is adequately applied

the outcomes are at least partly attributable to the
conditioning experiences.
Findings of controlled individual studies by
Barlow, Leitenberg, & Agras (1969), in which a
pedophile and a homosexual were treated with
symbolic aversion in a replicative design, lend
further support to the contribution of contiguous
experiences to observed changes. The clients’
autonomic and evaluative responses to deviant
sexual stimuli and the frequency of their sexual

urges were measured during periods when
sexually arousing imagery was successively paired
with verbally induced nausea or occurred alone.
Deviant sexual arousal declined sharply during the
aversive conditioning phase, it gradually increased

when sexually arousing scenes were repeatedly
presented without any aversive experiences, and it
was virtually eliminated after the aversive
contingency was reinstated.
Aversion therapy has produced many marked

and enduring changes in transvestism (Blakemore
et ah, 1963; Cooper, 1963; Glynn & Harper, 1961;

Lavin et al., 1961; Marks & Gelder, 1967;
Morgenstern, Pearce, & Rees, 1965); in fetishism
(Kushner & Sandler, 1966; Marks, Rachman, &
Gelder, 1965; McGuire & Vallance, 1964; Oswald,
1962; Raymond, 1956; Raymond & O’Keeffe, 1965;
Thorpe, Schmidt, Brown, & Castell, 1964); in
exhibitionism (Evans, 1967; Kushner & Sandler,

1966); and in homosexuality (Costello, 1963;
Freund, 1960; James, 1962; Max, 1935; Miller,
1963; Thorpe et al., 1964). In other cases involving
similar disorders, aversion therapy effected
temporary cessations or reductions in deviant

sexual behavior (Clark, 1963a, b; Freund, 1960;
Oswald, 1962; Thorpe, Schmidt, & Castell, 1963).
Still other clients have derived little benefit from
this mode of treatment (Freund, 1960; Solyom &
Miller, 1965).
It is difficult to identify the factors responsible
for the differential efficacy of aversion methods

because the cases involve variations in motivation
for change, in aversive stimuli, in the extent to
which heterosexual attractions are established, in
length of follow-up, in degree and duration of
homosexual behavior, and in the extent to which
favorable conditions for heterosexual activities
exist within the environment. The influential role

of some of these factors is revealed in studies
containing a sufficient number of cases to compare
outcome rates. Freund (1960), for example, found
that for homosexuals who were self-referred,
aversion therapy produced enduring heterosexual

orientations in approximately 45 percent of the
cases, and 16 percent adopted predominantly
heterosexual patterns for at least some period of
time. By contrast, only 6 percent of those who
were coerced into treatment by legal authorities

and relatives became exclusively heterosexual.
These data, in accord with the self-arousal

interpretation of counterconditioning presented
earlier, show how motivational factors may
counteract the effects that customarily result from

aversive stimulus pairing. Reluctant individuals
can impede the development of aversions by
failing to attend to the attractive stimuli and to
produce accompanying imagery. They can give

fake signals at crucial points in the procedure
where the therapist usually depends upon

guidance from the client. They could even reverse
the direction of counterconditioning by conjuring
up heterosexual imagery while undergoing
unpleasant stimulation. Finally, if aversion

reactions were established, they could be easily
extinguished by clients’ repeatedly engaging in
homosexual activities. Whatever the reasons

might be, the differential outcomes associated
with the desire to modify one’s sexual orientation

reaffirm the view that unless individuals are
committed to the selected objectives, their
behavior is likely to nullify the effects of change

programs.
McGuire, Carlisle, & Young (1965) advance the
interesting thesis that in some cases deviant
sexual preferences are developed through
masturbatory conditioning. According to the

authors, three factors usually figure prominently
in this form of sexual learning. First, as a result of
unpleasant heterosexual experiences or feelings of
physical and social inadequacy, the person comes
to believe that he cannot achieve a normal sex life.

Second, the person usually has a sexual experience
that is not sufficient by itself to establish deviant
erotic preferences, but it stimulates a fantasy for
later masturbation. The major conditioning is
assumed to occur in relation to symbolic

representations. As the person repeatedly
masturbates to the fantasy as his exclusive sexual

outlet, the pleasurable experiences from
masturbation endow the deviant fantasy with
increasing erotic value. This is essentially the

same mechanism through which Rachman (1966)
conditioned sexual excitement to shoes except
that instead of seductive photographs, the
arousing events are more powerful orgasmic

experiences.
The authors document their thesis with

extensive case data obtained from a large number
of sexual deviants. In one of these cases, for
example, a 17-year-old male was highly sexually

stimulated at seeing a girl dressed only in her
underwear. Thereafter, he frequently masturbated
to the mental imagery of the scantily clad girl.
Eventually the memory of the girl’s characteristics
faded, but advertisements and shop-window

displays of women’s undergarments continued to
serve as strong masturbatory fantasies. After a
period of several years the erotic potential of these
fetishistic objects had increased to the point
where he no longer showed interest in girls, but
rather derived his sexual stimulation almost
entirely from women’s underwear which he
bought or stole. This helps to explain how
outlandish fantasies can acquire powerful sexual

valence through contiguous association with
masturbatory experiences (Marks, Rachman, &
Gelder, 1965; Mees, 1966) and, once established,
why they are so refractory to extinction. Other
case reports by McGuire and his associates depict

a similar process in which aberrant sexual
fantasies are selectively reinforced; they
eventually become able to provoke corresponding
homosexual, exhibitionistic, and voyeuristic
behavior. The prevalence of masturbatory

conditioning in aberrant sexuality is further
shown by Evans (1968), who found that among a

large group of sexual deviates 79 percent used
deviant fantasy while masturbating.
In instances where erotic fantasies serve as
evocative stimuli for deviant sexual activities,
control over the behavior may be achieved by
eliminating either the aberrant fantasies or their
arousal properties. Evans (1967) reports

surprisingly high success in the treatment of seven
exhibitionists by aversive conditioning of imaginal
stimulus events. In this procedure, clients are
presented with image-producing phrases that
depict either normal heterosexual behavior or

exhibitionistic activities. Vivid imagery of the
deviant behavior is associated with shock, which
the client can terminate by changing to a slide
describing normal sexual responses. Of the seven
exhibitionists who finished treatment, five no

longer experience any urge to expose themselves
and have completely ceased exhibitionistic

behavior, while the remaining two have reduced
their frequency of genital exposure from a pre-
therapy high of 28 per month to 2 episodes per

month. In a subsequent study Evans (1968) found
that the speed of aversive conditioning was
related to the content of masturbatory fantasies.
Exhibitionists who utilized normal masturbatory

fantasies discontinued exposure behavior within
about 4 weeks of treatment, whereas those who

engaged in exhibitionistic masturbatory fantasies
required approximately 24 weeks before they
ceased the deviant activities. Results obtained by
Mees (1966) in modifying a sadistic fantasy

suggest that aberrant erotic fantasies of long
standing can be most effectively eliminated if, in
addition to aversive conditioning, normal

heterosexual imagery is induced and reinforced.
DIFFERENTIAL CONDITIONING OF SEXUAL

RESPONSIVENESS
In many of the cases cited above, therapists

have tried not only to create aversions to
inappropriate objects, but also to develop
attraction to heterosexual stimuli. A variety of
differential conditioning techniques has been
employed for this purpose. In one such approach,
which is probably minimally effective,

homosexuals, in addition to receiving aversive
conditioning trials, are shown pictures of nude or
scantily clad females several hours after
administration of testosterone (Freund, 1960;

James, 1962). Gonadal hormones may increase
sexual arousal, but they do not determine its
quality or direction. Indeed, attempts to treat
homosexuals by administrations of large amounts
of androgen simply increased their homoerotic

desires and behavior (Ford & Beach, 1951; Perloff,
1965). It remains a question, therefore, whether

any positive conditioning is achieved by this
method.
Masturbatory conditioning has also been
employed as a means of increasing the erotic

arousal properties of heterosexual stimuli. As part
of an aversion-relief method, Thorpe, Schmidt,
Brown, & Castell (1964) had clients masturbate to
photographs of attractive females and utilize these

images in masturbatory fantasies. In addition,
shock stimulation was presented in conjunction
with phrases describing deviant sexual practices,
whereas descriptions of heterosexual behavior
occurred with shock termination. The authors

report that heterosexual fantasies of high arousal
value can be strongly established in this manner.
In an earlier study, Thorpe, Schmidt, & Castell
(1963) found that masturbatory conditioning
alone did not eliminate homosexual fantasies, but

a method combining positive and negative
conditioning eventually replaced homoerotic

fantasies with heterosexual ones. In the absence of
comparative data of the sexual responses of
individuals receiving either aversive trials alone or

the combined procedure, there is no way of
determining the degree to which the positive
component facilitated the change in sexual
interest.
A differential conditioning strategy was
similarly employed by Davison (1968) in treating

a college student whose sexual activities were
confined to masturbation evoked by fantasies of
inflicting injury on women. The client was

instructed initially to utilize the sadistic fantasy to
induce sexual arousal but to masturbate while
looking at pictures of captivating belles from
Playboy magazine. After conventional sexual
stimuli had acquired sex-arousing value the

sadistic fantasy was paired with nauseous imagery
and supplanted by normal masturbatory fantasies.
Consistent with the interpretation of
counterconditioning as partly reflecting self-
control processes, the client was later able to
reinstate and to eliminate sadistic fantasies at will
through masturbatory conditioning with sadistic
or normal erotic fantasies.
In a third approach a differential conditioning

procedure is employed in which presentations of
pictures of nude males are accompanied by
electric shock, while looking at projected pictures
of nude females is reinforced by termination of
continuous electrical stimulation (Solyom & Miller,

1965). This particular method is predicated on the
assumption that stimuli associated with aversion
relief will acquire positive properties, it is unclear,
however, how pain-relief experiences can endow

related stimuli with sex arousal value.
Nevertheless, Solyom and Miller found, during
treatment of a group of homosexuals, that

plethysmograph responses to female pictures
became progressively greater, while responses to
male pictures remained essentially unchanged.
This suggests that the differential conditioning
produced a change from predominantly
homosexual to bisexual responsiveness. However,
these findings are difficult to interpret because, as

the authors themselves acknowledge, their
plethysmograph measure does not differentiate
between sexual and anxiety arousal. Erotic arousal
can be most validly assessed in terms of penile
erectile responses as demonstrated by Bancroft,

Jones, & Pullan (1966), who measured changes in
several object preferences in a pedophile
throughout the course of aversion therapy. This
measure makes it possible to conduct systematic
investigations of the relative efficacy of different

conditioning procedures for altering erotic
preferences. It also provides an objective criterion

for determining the duration of treatment, thus
safeguarding against either premature
termination, or needless prolongation, of the

conditioning sessions.
It should be emphasized here that conditioning
of sexual attraction to appropriate objects
constitutes only part of a broader treatment

objective. Persons who have engaged in deviant
sexual practices for a long time must develop not

only heterosexual attraction but also intricate
patterns of heterosexual behavior. This may
require, among other things, acquisition of new
speech patterns, dress styles, courtship behaviors,

modes of sexual stimulation that are closely
associated with heterosexual coitus, and many
other aspects of sex-role behavior. To the extent

that such behavioral changes enable persons to
engage in rewarding intimate interactions, the

resultant positive experiences will exert a
powerful influence on the further development of
heterosexual feelings and preferences.
CONDITIONING OF AVERSIONS AND

REINFORCEMENT OF ALTERNATIVE MODES OF
BEHAVIOR
The existence of desirable modes of sexual
behavior is probably a major determinant of the
durability of changes induced through aversive

counterconditioning. Enduring modification of
sexual deviance through aversion methods will

therefore be governed, not so much by the
magnitude of negative properties conditioned to
previously attractive stimuli but rather by the
availability of alternative means of securing sexual

gratification. Because the sexual taboos that
prevail in the culture extend to treatment
situations as well, it is exceedingly difficult to

produce am to reinforce desired patterns of sexual
behavior. Consequently, treatment interventions

are mainly directed toward reducing the incidence
of deviant behavior, but the appearance and
continued maintenance of heterosexual responses

are left to fortuitous circumstances.
Treatment by aversion methods exclusively
generally achieves favorable outcomes with
persons who display bisexual responsiveness or
engage in heterosexual intercourse, even through

they rely upon fetishistic objects for erotic arousal.
On the other hand, aversion treatment alone
appears to be much less successful in cases where
the deviant behavior constitutes the sole means of
obtaining sexual gratification. This differential

efficacy is well illustrated by Oswald’s (1962)
treatment of two rubberized-clothing fetishists. In
both cases aversion to rubberized garments was
successfully established, but one of the clients
subsequently reverted to the deviant behavior.
In the first case, rubberized garments
primarily served as arousal stimuli for exclusively

heterosexual patterns of behavior. The client was
prompted to seek treatment because of marital
conflicts arising from his wife’s refusal to wear her
rubberized mackintosh in bed. The erotic arousal
properties of the fetishistic stimuli were rapidly
eliminated through a series of counterconditioning
trials, in each of which the client donned a

mackintosh while experiencing apomorphine-
induced nausea supplemented by a recorded tape
suggesting similar reactions. A 21-month post-
therapy evaluation is summarized as follows: “He
feels quite indifferent to rubberized clothes and

finds it hard to believe how he could ever have had
this interest in them. His career has prospered
extremely well by his own efforts and talents, and
his wife confirms that they are normal and happy
in their general and sexual life [p. 201].”
The second case, a 32-year-old military recruit,
experienced heightened sexual arousal whenever

he tied himself up tightly in black shiny rubber.
This fetishistic behavior apparently originated in
an early experience in which a group of boys
seized the client, tied a groundsheet over his head
and masturbated him. “Since then he had made a
practice of tying himself up with rubber
groundsheets, a rubber hood and ropes. He came

for treatment partly because he feared he might
encompass his own death, as he had recently had
difficulty in releasing himself [p. 201].” Except for
one occasion when the client had sexual
intercourse at a specialized brothel equipped with

rubber strait-jackets, hoods, thongs, and rubber
zip suits, masturbation served as his predominant
form of sexual behavior. The counterconditioning
sessions proceeded along lines similar to the case
cited above. Following the administration of

apomorphine, the client tied himself up in
groundsheets or donned a frogman’s suit. In

addition, on two successive days he received
several injections of testosterone and was
furnished with books featuring photographs of

females.
It is apparent from follow-up reports that
rubberized garments had temporarily lost their
erotic value, but the reinstatement of fetishistic

behavior would be expected from the limited
scope of the treatment, aside from the fact that the

deviant behavior was instrumental in eventually
gaining the client a military discharge.
He tried out his rubber articles a week after

leaving hospital, found they held no interest
for him and discarded them. He went out to
dances and other social events for the first
time in years. After 6 months he relapsed
and a further 4 months later made known his
deviation to the Service authorities and was
invalided. He told me at that time that he
intended to live in London where there were
others who shared his interests. He had been
back to the brothel which, he pointed out,
advertises in a well-known week-end
publication available at any bookstall, under
the guise of a rubber-clothing store. He had
formed a friendship with a male homosexual
(but not had sexual relations with him),
whom he had first noticed wearing a black,
shiny, rubber mackintosh in Hyde Park one
fine summer’s evening [p. 202].
The conditioning of aversion to inappropriate
sexual stimuli can assist in promoting

heterosexual behavior provided these alternative
responses already exist in the client’s repertoire,

and sufficient positive reinforcement is available
in everyday situations to maintain them. The latter
conditions were clearly present in the first case

reviewed, but the second client exhibited only a
weakly developed heterosexual repertoire, which
evidently was further extinguished during the

period following treatment when he became more
active socially.
The mere absence of heterosexual behavior in
itself does not necessarily indicate a behavioral

deficit. A person may have developed some
capability and desire for culturally approved
forms of sexuality, but these tendencies are
strongly inhibited because of heterosexual
anxieties. It is therefore important to distinguish
between developmental deficits and inhibitory

effects in devising supplemental treatment
programs to aversion therapy. The inclusion of
anxiety extinction procedures is particularly
important in modifying deviant sexual behavior

that is, in part, maintained by fear of heterosexual
involvements. Cooper (1963) reports a case of this
type that illustrates the combined use of aversive
therapy and self-directed desensitization in the
treatment of a 25-year-old pharmacist whose

central problems featured transvestism and
impotence.
The discussion of the necessity for
supplemental programs has dealt thus far with

conditions in which heterosexual repertoires are
either minimal or strongly inhibited. The scope of

the interventions may also need to be extended
when changes in behavior cannot be readily
achieved and sustained due to inadequate sexual
reinforcement. This problem is revealed in the

treatment of a transvestite reported by Glynn &
Harper (1961). After a series of aversion trials in
which the client experienced apomorphine-
induced nausea while attired in women’s clothes,
he exhibited revulsion at the sight of the female

clothing, he no longer felt any desire to wear it,
and he declined a request to do so. While aversive
counterconditioning successfully eliminated the
transvestite behavior, the client’s marital
relationship provided absolutely no sexual

gratifications. Although he had been married for
four years, the marriage had never been

consummated, due largely to his wife’s frigidity.
The wife’s marked sexual inhibitions were
therefore treated by the standard desensitization

procedure with considerable success, as evidenced
by the fact that “she is now pregnant and the
marriage is happily stabilized.” A seven-month
follow-up study disclosed no transvestism nor any

desire on the client’s part to engage in cross-sex
dressing. Had the treatment in this case been

confined to negative conditioning of transvestite
responses, it is highly probable that any attempted
heterosexual behavior would have been rapidly
extinguished and transvestism might have

regained its sexually rewarding function.
When deviant sexual behavior is less strongly
established, it may be possible to dispense with
aversive counterconditioning by substituting a

treatment program based on extinction of
heterosexual anxieties coupled with positive
reinforcement of desired alternative behaviors.
Existing sociosexual repertoires can in this way be
further developed and strengthened until
eventually they become more rewarding than the
deviant tendencies. It will be recalled from clinical
studies reviewed earlier (Bond & Hutchison, 1964;
Stevenson & Wolpe, 1960) that successful

outcomes were, in fact, achieved with
exhibitionists and homosexuals treated by
desensitization methods alone. These findings
suggest the need for systematic comparative
studies of the relative efficacy of aversive

counterconditioning, desensitization, and
reinforcement procedures, utilized singly or in
combination, with persons exhibiting varying
dispositions for sexually deviant behavior.
Modification of Symbolic Activities
Although aversive counterconditioning has
been most extensively employed in the treatment
of alcoholism and sexual deviations, a number of
investigators have also attempted to eliminate
unusually persistent ruminations by this method.
Perturbing fantasies containing sexual, aggressive,
and other elements have proved exceedingly
refractory to change. Because of their limited

accessibility, these types of ideational phenomena
are not readily amenable to experimental
investigation. As a consequence, the conditions
governing the occurrence and persistence of
emotionally disturbing thoughts remain obscure.

There exists some research evidence (Eriksen &
Kuethe, 1956) to suggest, however, that thoughts
can be brought under control by applying aversive
contingencies. In this particular experiment,
students were instructed to produce associations

to fifteen words, and their associations to five of
these words were followed by shock. Associations

that were accompanied by the aversive experience
declined rapidly, while the frequency of
nonpunished associations remained unchanged

over succeeding trials (Figure 8-2). The decrement
Figure 8-2. Percentage of first-trial responses repeated on
succeeding trials as a function of punishment. Eriksen
& Kuethe, 1956.
in associations occurred whether or not subjects
were aware of the punishment contingency.
It is, of course, easier to eliminate relatively
neutral ideations than the highly valenced forms
appearing in treatment. Wolpe (1958) reports

varying degrees of success in removing
obsessional ruminations by associating their
occurrence with electric shock. Employing

essentially the same procedure, McGuire &
Vallance (1964) successfully treated a 29- year-old
teacher who was plagued by intrusive thoughts
concerning his wife’s fidelity, dating back to an
ambiguous remark originally made by his mother.

Although he realized that these thoughts had no
basis in reality, nevertheless this knowledge was
of little avail in controlling the obsessional
ruminations. In treatment, shocks were
contiguously associated with thoughts about his
mother making the uncomplimentary remark and
its implications. During the second and third
sessions, the client controlled the timing and
intensity of shock administration and, thereafter,

he continued the conditioning process at home
with a portable apparatus. Within a relatively brief
time the obsessional ruminations were eliminated,
a change that was accompanied by a general
diminution of anxiety.
The elimination of disturbing thoughts in the

course of aversion therapy is best revealed by
Marks, Rachman, & Gelder (1965), who
demonstrated that with successive aversive trials

the latency of perverted thoughts increased until
eventually they could not be produced at all. In the
previously reviewed studies involving more

extensive measurement of changes Marks &
Gelder (1967) found that during reduction of
deviant sexual fantasies, erotic arousal
accompanying the imagery also diminished
progressively. In approximately half the cases,
however, aversive conditioning reduced the
arousal potential of deviant sexual fantasies, but it

did not eliminate the mental imagery. The reasons
for the differential susceptibility to inhibition are
not apparent. At this point it is worth noting that
aversion therapy provides an excellent
opportunity for studying the extent to which

thought processes are amenable to reinforcement
control. It also furnishes a means of investigating
symbolic control of overt behavior.
In a number of studies, strongly established

behavior has been enduringly eliminated by
aversive conditioning of symbolic events alone.
Evans (1968) and Thorpe and his associates

(1964), for example, modified sexual disorders by
endowing verbal and imaginal representations of
the deviant activities with negative qualities
through association with shock. Agras (1967) also
successfully applied the symbolic conditioning
paradigm to aggressively destructive behavior in a
chronic schizophrenic, who had to be physically

restrained because of uncontrollable tendencies to
smash any glass in sight. The client participated in
a series of sessions in which he was asked to
visualize himself breaking glass, whereupon he
was administered a painful electric shock. As the

treatment progressed the latency of the
destructive imagery increased and eventually he
lost all urge to smash glass. A follow-up study
revealed that, except for one minor incident, the
destructive behavior never appeared again.
Imaginal counterconditioning may be of
greatest value in modifying disorders in which

symbolic events possessing high arousal potential
exert substantial control over behavior. If the
internal elicitors are eliminated the related actions
should decrease in frequency. Some investigators
have assumed, however, that symbolic stimuli can
be substituted for real events without sacrificing
effectiveness. If this assumption were valid, then

the method would have wide applicability, since
any event, no matter how complex, can be easily
visualized. On the basis of generalization
principles one would expect actual stimuli to
produce better results than imagined ones.
OTHER ADDICTIVE BEHAVIORS
Another behavioral problem to which aversive
forms of treatment have been applied is cigarette

addiction. Preliminary case studies report
relatively high abstinence rates accompanying
treatments in which the act of smoking is
contiguously paired with aversive electrical
stimulation (McGuire & Vallance, 1964), drug-
induced nausea (Raymond, 1964), or a
disagreeable mixture of smoke and hot air (Wilde,
1964). These favorable outcomes contrast with
those of Koenig & Masters (1965), who compared
changes in smoking behavior in groups of students

who received either desensitization, aversive
counterconditioning, or supportive counseling.
Very few of their subjects discontinued the habit,
and degree of reduction in smoking was found to

be unrelated to the type of treatment
administered.
In an experimental study employing
appropriate controls, Stollak (1968) had little
success in modifying obesity by pairing
descriptions of fattening foods with shock
stimulation. It would appear from the enduring

weight reductions obtained through self-
management of contingencies by Harris (1969)

and Stuart (1967), that a broad program of self-
control can be highly effective, whereas aversive

conditioning alone is likely to yield unimpressive
results.
Alcoholism
A wide variety of “neurotic personality

disturbances” have been proposed as the
underlying determinants of chronic alcoholism.
Among the more widely accepted interpretations

are those advanced by psychoanalytic theory,
according to which alcoholism derives from latent
homosexuality related to fixations on “passive-
narcissistic aims.” Oral dependent needs and

characterological structures are thus frequently
invoked as the decisive predisposing factors in the

excessive use of alcohol. Self-destructive drives,
feelings of inferiority, unconscious needs to
dominate, and a host of other factors including
excessive mothering, insufficient mothering,
emotional immaturity, and introverted
psychoneuroses have also been proposed as
determinants of alcoholism.
In contrast to the widespread claim for a
prealcoholic personality, comparative studies of

alcoholics and nonalcoholics (Sutherland,
Schroeder, & Tordella, 1950; Syme, 1957) have
generally failed to identify any specific personality
traits or “underlying dynamics” that would clearly
differentiate alcoholics from other deviant groups,

or for that matter, from persons judged to be
“normal.” Even if some consistent personality
correlates of alcoholism had been obtained, it
would be impossible, without longitudinal studies,

to determine whether the given personality
patterns represented the cumulative
consequences or the causes of chronic

intoxication. It is nevertheless evident from a large
body of empirical findings and knowledge of
behavior maintenance mechanisms that the search
for personality dynamics that supposedly control
excessive drinking is a futile pursuit. Just as
persons who differ markedly in personality
attributes can learn to use tobacco excessively, so,

given appropriate social-learning conditions,
persons who possess diverse personality
characteristics can be taught heavy drinking of
alcoholic beverages. In fact, it has been shown
repeatedly that no matter what deviant behavior

is singled out for study, it is usually found in a
wide variety of personality types. A much more
fruitful approach to the understanding of
alcoholism would be to investigate the learning
contingencies specifically associated with drinking

behavior and the reinforcement mechanisms
maintaining self-intoxication.
EFFECTS OF ALCOHOL ON EMOTIONAL AROUSAL

AND REACTIVITY
Psychodynamic theories have generally

emphasized the symbolic value of alcohol in
gratifying “oral” or “passive-dependent” needs, but
remarkably little attention has been paid to the
pharmacological properties of ethanol which,

under certain conditions, make it a powerful
positive reinforcer.
One set of experiments that has direct bearing

on the reinforcing qualities of ethanol is concerned
with its effects on both autonomic arousal and
reactivity. In these studies subjects’ physiological

responses are measured prior to and following the
ingestion of alcohol, with basal conductance level
and magnitude of GSR responses to specific
stressor stimuli typically serving as indices of
emotional responsiveness. The findings generally

show that alcohol in small doses has no consistent
effects (Docter & Perkins, 1961; McDonnell &

Carpenter, 1959), but it can produce substantial
reduction in affective arousal when taken in

moderate to large quantities (Carpenter, 1957;
Greenberg & Carpenter, 1957). A comparison of
alcohol, meprobamate, and a placebo by Lienert &
Traxel (1959) further reveals that alcohol and the

tranquilizer are equally effective in reducing GSR
responses to disturbing verbal stimuli. Moreover,
subjects who had exhibited high emotionality, as
assessed several weeks prior to the experimental
session, were tranquilized by alcohol to a greater

degree than those who had previously displayed
low arousal.
It is sometimes mistakenly assumed (Chafetz &
Demone, 1962) that reinforcement principles

cannot adequately account for alcoholism because
the devastating social and physical consequences
of chronic drinking far outweigh its temporary

relief value. This argument overlooks the fact that
behavior is more powerfully controlled by its
immediate, rather than delayed, consequences,
and it is precisely for this reason that persons may
persistently engage in immediately reinforcing,
but potentially self-destructive, behavior. Future
adverse consequences, reinstated symbolically in

the present, may be sufficiently strong to inhibit
drinking behavior when instigation for escape is
relatively weak. On the other hand, it is
unreasonable to expect thoughts of future effects
to exert much of an inhibitory influence in persons

who experience a high level of aversive
stimulation and who present a well-established
stress-alcoholism response pattern.
EFFECTS OF ALCOHOL ON AVOIDANCE AND

ESCAPE RESPONSES
Further evidence for the stress-reducing
properties of alcohol is provided in experiments
with animals designed to study disinhibitory

effects and extinction of avoidance responses.
Suggestive findings regarding the fear-reducing

effects of alcohol were originally reported by
Masserman & Yum (1946) in a study in which cats
that had learned to perform complex
manipulations to secure food subsequently

inhibited the instrumental manipulatory and
approach responses after they had been shocked
at the goal. Administration of small doses of
alcohol, however, promptly restored the approach

manipulations designed to obtain the food
rewards. In addition, the cats developed a
preference for milk cocktails containing 5 percent
alcohol to plain milk during the series of shock
trials, but reverted to their original preference for

nonalcoholic drinks after the aversive stimulation
had been discontinued and emotional responses

were completely extinguished. In a partial
replication of the Masserman and Yum study,

Smart (1965) further confirmed the anxiety-
mitigating effects of alcohol.
In order to test whether alcohol reduces
punishment-induced avoidance behavior or
increases approach tendencies in an approach-
avoidance conflict, Conger (1951) designed an

experiment in which he trained one group of
animals to approach the lighted end of the alley for
food, and a second group to avoid the lighted end
of the alley to escape electric shocks. Compared to
the behavior of sober controls, which had been

administered placebo injections, the avoidance
responses of subjects that had received alcohol
injections showed a substantial reduction in
strength, but approach responses seemed
unaffected.
A few experiments involving human subjects
have also demonstrated the disinhibiting effects of

alcohol on verbal expressions of sexual and
aggressive behavior in social drinking situations
(Bruun, 1959; Clark & Sensibar, 1955). Among
humans, however, the same dose of alcohol may
have diverse effects because individuals differ in
the types of responses inhibited, the strength of
inhibitions, and variations in social conditions

which, in part, serve to define and to control
appropriate behavior.
Numerous studies have been concerned with
the influence of ethanol on escape and avoidance
responses tested in a variety of aversive

conditioning situations involving no rewarded
competing responses. In these experiments
animals are initially taught to perform responses
which either avert the onset of aversive

stimulation or terminate it after its occurrence.
Changes in the rate of avoidance and escape
responses as a function of the administration of

ethanol are then assessed relative to the
performances of control groups, which receive
either water or solutions containing other types of
drugs. Ethanol in moderate doses produces more
rapid extinction of fear-mediated responses
(Kaplan, 1956; Pawlowski, Denenberg, & Zarrow,
1961), and it reduces the rate of responses

designed to postpone the occurrence of aversive
stimuli (Hogans, Moreno, & Brodie, 1961; Sidman,
1955). Moreover, the capacity of alcohol to reduce
emotional behavior is similar to that of other
drugs possessing central depressant properties

(Korpmann & Hughes, 1959).
The withdrawal of positive reinforcers
following a period of reward generally produces
aversive effects that lead to the suppression of

associated responses. Further support for
interpreting the behavioral effects of ethanol in
terms of emotion-reducing processes is furnished

by experiments concerned with the reinstatement
of responses following their inhibition through
frustrative nonreward. Under these conditions,
animals administered ethanol are more persistent
than those given a placebo in performing
nonreinforced behavior (Barry, Wagner, & Miller,
1962), and they increase their rate of response in

the presence of stimuli signifying nonreward
(Blough, 1956; Miller, 1961).
The experimental data reviewed so far, based
on the forced administration of moderate doses of
ethanol, strongly indicate that alcohol can produce

significant decrements in both autonomic arousal
and emotional behavior generated by aversive
environmental conditions. Investigations
concerned with variables governing the voluntary

intake of alcohol also contribute to an
understanding of the development and
maintenance of self-intoxication. This research is

reviewed next.
DETERMINANTS OF VOLUNTARY ALCOHOL

CONSUMPTION
The self-selection method has been extensively
employed in studies designed to identify the
stimulus determinants of alcohol consumption. In

these investigations animals are typically provided
with a choice between an alcoholic solution and

one or more other liquids; the base level of
voluntary alcohol intake is then compared with
the amount consumed under various
environmental conditions.
Findings of studies utilizing a forced alcohol
regimen, in which the animals’ entire liquid intake
is restricted over a period of time to solutions
containing various concentrations of ethanol,

reveal that alcohol per se has no strong inherently
reinforcing properties. Under such conditions,
animals consume small, nonintoxicating amounts
of alcohol, but they readily revert to drinking
other liquids when these later become available
(Korman & Stephens, 1960; Richter, 1956). The
positive reinforcing value of alcohol, as inferred
from increases in alcohol consummatory
responses, can be substantially enhanced,

however, by certain physiological and
psychological conditions. Although animals which
metabolize ethanol rapidly consume greater
amounts of alcohol in free-choice situations than
those who exhibit lower metabolic rates, studies of

intraindividual variations reveal that nutritional
deficiencies, endocrinal factors such as insulin,
and drugs which produce liver damage increase
voluntary ethanol intake (Mardones, 1960;
Rodgers & McClearn, 1962). However, studies of

human alcoholism, though complicated by
ambiguities regarding cause and effect, have failed

to yield any reliable differences between
alcoholics and nonalcoholics in genetic and
endocrinological characteristics (Lester, 1966).
Findings of laboratory studies comparing
voluntary intake of alcohol before, during, and
after aversive stimulation throw some light on one
of the possible mechanisms underlying alcohol

consumption. It will be recalled that Masserman &
Yum (1946) found that animals who had initially
preferred plain milk to an alcoholic milk solution
developed a preference for alcohol during periods
of shock-induced stress, but reverted to

nonalcoholic drinks following termination of
aversive stimuli and fear extinction. Clark & Polish
(1960) measured the intake of water and a
solution of 20 percent alcohol by monkeys before,
during, and after avoidance training in which each

response briefly postponed the occurrence of
electric shocks. Although there was relatively little

change in water intake across phases, alcohol
consumption increased during, and decreased
following, the avoidance conditioning sessions.
The effects of aversive stimulation on alcohol
consumption are apt to be more prolonged when
punishment is administered on a noncontingent
and unpredictable basis. Casey (1960), for

example, studied the relative intake of water and
an alcohol solution as a function of aversive
shocks programmed according to a variable
interval schedule. Under such conditions of
uncertainty, the animals drank somewhat larger

amounts of alcohol for the period of stress, but the
greatest increments in voluntary alcohol
consumption occurred during the following
month, after the shocks had been discontinued. On
the other hand, in a second group of animals

provided with a free choice of water or a solution
of reserpine (which has long-delayed effects), the

same experimental manipulations failed to
increase the attraction of the latter drug. These
differential findings suggest that the relatively

rapid absorption of alcohol, and the attendant
reduction in aversive arousal, may partly
contribute to its effectiveness as a positive
reinforcer under conditions of aversive

stimulation. Moreover, generalized emotionality
may further enhance its reinforcing effects (Korn,

1960).
SOCIAL LEARNING OF DRINKING BEHAVIOR
The research discussed above indicates that

excessive alcohol consumption is maintained
through positive reinforcement deriving from the
central depressant and anesthetic properties of
alcohol. Persons who are repeatedly subjected to

environmental stress are, consequently, more
prone to consume anesthetic doses of alcohol than
those who experience less stress and for whom,
therefore, alcohol has only weak reinforcing value.
In many cases, also, excessive drinking may

primarily serve to relieve the aversive effects of
boredom.
Prolonged and heavy use of alcoholic
beverages produces alterations in the metabolic

system which provide the basis for a second
maintaining mechanism that is quite independent
of the original functional value of alcohol. That is,

withdrawal of alcohol elicits exceedingly aversive
physiological reactions consisting of
tremulousness, nausea, vomiting, marked

weakness, diarrhea, fever, hypertension, excessive
perspiration, and insomnia (Isbell, Fraser, Wikler,
Belleville, & Eiseman, 1955; Mendelson & La Dou,
1964). After the person thus becomes physically
dependent on alcohol, he is compelled to consume

large quantities of liquor both to alleviate
distressing physical reactions and to avoid their

recurrence. Since the ingestion of intoxicants
promptly terminates physiologically generated

aversive stimulation, drinking behavior is
automatically and continuously reinforced. After
pharmacological addiction occurs, the major part
of the alcoholic’s time and resources becomes

devoted to maintaining a continuously high level
of self-intoxication.
Although aversion reduction and other
positive reinforcements which typically
accompany social drinking may account

adequately for the maintenance of inebriety, an
adequate theory of alcoholism must invoke
additional social-learning variables since,
obviously, most people who are subject to

stressful experiences do not become alcoholics. It
has been customary to summon internal
determinants in the form of neurotic personality

disturbances and underlying pathologies as the
differentiating antecedent variables. The
inadequacy of theories of alcoholism, which
emphasize the role of personality traits and
internal dynamics, becomes readily evident in the
marked cultural and subcultural differences in the
incidence of alcoholism. Indeed, based on the

theory that a “neurosis” is instrumental in the
development of chronic alcoholism, one would be
forced to conclude that Jews, Mormons, Moslems,
Italians, Chinese, and members of other cultural
groups which present exceedingly low rates of

addictive drinking are lacking in neuroses, oral
deprivations, self-destructive tendencies, latent
homosexuality, indulgent mothering, and
inadequacy feelings, whereas the latter pernicious
conditions must be highly prevalent among the

Irish, who surpass all other ethnic groups in
chronic alcoholism (Chafetz & Demone, 1962;

McCarthy, 1959; Pittman & Snyder, 1962).
Perhaps the most striking evidence that
alcoholism primarily represents a learned pattern

of behavior, rather than a manifestation of a
particular type of predisposing underlying
pathology, is provided by the extraordinarily low
rates of alcoholism among Jews, who experience

no less, and in all probability more, psychological
stress than members of other ethnic groups noted

for their drinking (Glad, 1947; Snyder, 1958).
These ethnic and subcultural differences in the use
of intoxicants point to the importance of the
prealcoholic social learning of drinking behavior

in the development of alcoholism.
The social-learning variables take several
forms. At the most general level they are reflected
in the cultural norms that define the

reinforcement contingencies associated with the
use of alcohol. There is considerable evidence that
the consumption of alcohol is significantly
influenced by the drinking mores of given social
groups. Members of cultures that are highly
permissive toward the use of intoxicants, or even
consider drinking to be emulative behavior,
display a higher incidence of drunkenness than
individuals reared in cultures that, for religious

and other reasons, demand sobriety. Similarly,
within heterogeneous cultures such as our own,
the prevalence of chronic intoxication varies as a
function of the types of social-learning conditions
that are associated with class status, religious

affiliation, racial and ethnic background,
occupation, and urban or rural residence.
Although cultural and subgroup mores
obviously play an influential role in determining

the extent of excessive drinking, normative
injunctions alone do not explain either the
relatively low incidence of addictive drinking in

social groups that positively sanction the use of
alcoholic beverages, or the occurrence of chronic
alcoholism in cultures prohibiting intoxicants.
Cultural and subgroup mores are to a large
extent transmitted through the modeling behavior
of socializing agents; consequently, one cannot

assume that members of a particular class
undergo equivalent learning experiences. Studies

of the family background of alcoholics generally
reveal an unusually high incidence of familial
alcoholism (Fort & Porterfield, 1961; Lemere,
Voegtlin, Broz, O’Hollaren, & Tupper, 1942b; Wall,

1936). It might be argued that these data support
a genetic interpretation of alcoholism, but the
pattern of drinking behavior being modeled and

the range of circumstances in which it occurs are
of greater importance than exhibition of some

drinking or complete abstinence by family
members. For example, in Italian and Jewish
households, use of dilute alcoholic beverages,

particularly wine, is approved under clearly
circumscribed conditions but negatively
sanctioned if consumed in either intoxicating
amounts or inappropriate situations. When the

use of alcohol is thus restricted primarily to
mealtimes or forms an integral part of religious
ceremonies and other social and familial practices,

alcohol consumption may be brought under
sufficiently narrow stimulus control to ensure

moderation (Bales, 1946; Glad, 1947; Snyder,
1958). On the other hand, in familial situations
where alcohol is consumed extensively in a large
variety of circumstances and is a preferred
response to monotony or stress, a similar type of

drinking pattern is likely to be transmitted to
growing offspring. Although drinking behavior is

initially most often acquired under nonstress
conditions, a habitual social drinker will
experience stress reduction on many occasions.
Once alcohol consumption is thus intermittently

reinforced, it will be readily elicited under
frustrative or aversive conditions. Therefore,
alcoholism typically results from habituation after
prolonged heavy social drinking acquired within
the context of familial alcoholism.
The relationship of stress to alcoholism is

perhaps strongest among alcoholics who are
members of subcultural groups that negatively
sanction the consumption of intoxicating

beverages, and whose parents have practiced total
abstinence. The social-learning history of

alcoholism under these conditions has never been
adequately documented, but there is some
evidence to suggest that in these cases the

drinking pattern is originally acquired under
highly stressful conditions and then generalizes to
less acute emotional circumstances (Fort &
Porterfield, 1961). In addition, outside the family,

peer models who drink may play an influential
role in transmitting drinking behavior (Skolnik,
1957).
In the foregoing analysis of alcoholism,
aversive stimulation and its quick reduction

through the depressant action of alcoholic
beverages were assigned a central role in the
development and maintenance of addictive
drinking. It should be emphasized, however, that
conflict, boredom, frustration, and other stressful

conditions may elicit a wide variety of reactions
including aggression, dependency, withdrawal,
somatization, regression, apathy, autism,
inebriety, or constructive coping behaviors.

Persons who exhibit the latter stress response
pattern will typically be judged “normal”; in
contrast, “neuroses,” “deep-seated personality

disturbances,” and other disease processes are
frequently invoked as explanatory factors when
persons have acquired one or more of the former
patterns of coping behavior. These assumed
pathologies represent essentially pseudo-
explanations since the main evidence for their
existence is the behavior that they are called on to

explain.
From a social-learning point of view, alcoholics
are people who have acquired, through differential
reinforcement and modeling experiences, alcohol
consumption as a widely generalized dominant

response to aversive stimulation. Therapeutic
attention would therefore be most profitably
directed toward reducing the level of aversive
stimulation experienced by individuals, and

toward eliminating alcohol stress responses either
directly or, preferably, by establishing alternative
modes of coping behavior. Given more effective

and rewarding means of dealing with
environmental demands, individuals will have less
need to resort to self-anesthetization against
everyday experiences.
It is entirely possible that the stress
component in alcoholism has been assigned
excessive weight largely because investigations

involving psychological variables have been
essentially confined to aversive conditioning

procedures, to the relative neglect of other
potentially significant determinants of alcohol
consumption. Moreover, although physiological
conditions and environmental stress have been

shown to increase alcohol intake, the amount
consumed seldom exceeds the animals’ oxidative
capacity. In contrast, Lester (1961) found that

animals on a variable interval schedule of positive
reinforcement maintained a steady and prolonged

state of self-intoxication accompanied by signs of
overt drunkenness, behavioral impairment, and
the development of metabolic tolerance analogous

to human alcoholism. Since intermittent food
reward can hardly be considered a highly
distressing situation, evidently factors other than
stress reduction were primarily responsible for
the self-maintained inebriation. The findings from

the latter study point to the need for experimental
investigations of other psychological variables that

may be expected eventually to produce addictive
drinking.
Since alcoholism often arises in rewarding

social interactions, operant drinking, in which
alcohol consumption primarily serves an
instrumental rather than a reinforcing function,
warrants detailed examination. In this process a
person drinks in order to obtain a variety of

reward; deriving from social interactions with
imbibing companions. Prolonged heavy drinking
leads to the development of physiological
tolerance and dependence on alcohol which, in
turn, necessitates increased alcohol intake. Thus,
in advanced stages biochemical, stress-reduction,
and social-reinforcement mechanisms may
contribute to the maintenance of addictive
drinking.
Regardless of the theory of alcoholism and
psychotherapy to which one may subscribe, the
elimination or drastic modification of alcoholic
behavior is obviously an objective of considerable
import. One of the behavioral approaches to this

problem has relied on the conditioning of aversive
properties to alcoholic beverages. In the following
sections the value and limitations of this mode of
therapy, and the conditions under which

alternative or supplementary procedures are
essential for the successful modification of
alcoholism, are considered in detail.
CONDITIONED-AVERSION THERAPY
The first systematic application of aversive
counterconditioning to the modification of
alcoholism was reported by Kantorovich (1934).
Twenty alcoholics participated in 5 to 18 sessions
in which cards containing the names of drinks, and
actual sight of bottles of vodka, wine, and beer,

and the smell and taste of these various alcoholic
beverages were successively paired with electric

shocks. A control group of 10 alcoholics received
hypnotic suggestions and medication. Of the 20
clients in the experimental group, 17 acquired
stable aversion reactions to alcohol, and 14

remained totally abstinent when subsequently
evaluated at periods ranging from 3 weeks to 20
months. In contrast, all but one of the controls
reverted to their customary alcoholic ways within
a few days following their discharge from the

hospital.
Although Kantorovich’s procedure aroused
little interest, aversion therapies employing
pharmacological agents have been widely applied
to the treatment of alcoholism. Except for minor
variations, the conditioning procedures are
generally patterned after the methods originally

devised by Voegtlin and his associates (Lemere,
Voegtlin, Broz, O’Hollaren, & Tupper, 1942a;
Voegtlin, 1940) at a sanitarium devoted
exclusively to the treatment of alcoholism. The
treatment consists essentially of associating the

sight, smell, taste, and thought of alcohol with
drug-induced nausea in 4 to 7 brief sessions
distributed over a period of about ten days.
On the morning preceding the treatment, the

client is administered only liquids and a stimulant
drug (e.g., benzedrine sulphate) designed to
augment the conditioning process. The sessions

are conducted in a semi-darkened, soundproof
room from which all extraneous auditory, visual,
and olfactory stimuli have been excluded. In front
of the client’s chair is a table containing a varied
array of liquors including bourbon, scotch, gin,
brandy, rum, beer, and wine, spotlighted so as to
focus attention on the liquors. The client is first

given a glass of tepid water containing oral
emetine, and immediately thereafter an injection
of an emetine-pilocarpine-ephedrine mixture.
Emetine is utilized as the aversion-producing
agent primarily because its emetic action is more

sustained and it does not have the sedative effect
of apomorphine.
Just prior to the onset of nausea the client is
poured an ounce of straight whiskey and asked to

smell it, to sip it, and to taste it thoroughly. This
same procedure is repeated several times with
whiskey either taken straight or mixed with warm

water to afford easy emesis. The rationale for
relying on whiskey exclusively in the initial
session and at the beginning of each subsequent
session is that it produces greater gastric irritation
than beer or wine and therefore serves to facilitate
the emetic reaction. Kant (1944) has seriously
questioned the wisdom of using the conditioned

stimulus in order to enhance the unconditioned
response, since this procedure runs a high risk of
reinforcing drinking behavior. If 4 to 6 ounces of
alcohol are ingested before emesis occurs, large
quantities of alcohol are likely to be absorbed.

Under these conditions, the immediate reinforcing
effects of alcohol may reduce, or even outweigh,
the effectiveness of subsequent aversive
experiences. Considering the difficulties in
precisely timing the onset of emetic responses,

this factor may partly account for variations in
strength of conditioned aversion developed by

different investigators supposedly using the same
method. While Voegtlin has taken necessary
precautions to avoid alcohol absorption by the

client during treatment, it is not clear whether
other therapists have paid as close attention to
this important point in the technique.
The procedural changes recommended by Kant
(1944, 1945) obviate the alcohol absorption
problem without detracting from the efficacy of

the treatment. During the first two sessions just
prior to and during nausea, the client is asked to
look at, smell, and taste the different alcoholic

beverages, but then to spit them out. In
subsequent sessions, the client is requested to
drink some alcohol at the height of nausea. Only
during the terminal sessions, when alcohol itself
has acquired the capacity to produce rapid

emptying of the stomach, is the client encouraged
to take several drinks.
It is important to include all varieties and types
of alcoholic beverages as conditioned stimuli in
order to establish the most stable and generalized
aversion responses. Lemere & Voegtlin (1940), for
example, report several cases in which aversion
persisted to the class of beverages originally

counterconditioned, but the client began to drink
intoxicants toward which aversion reactions had
never been established. Subsequent treatment
involving these beverages produced total

abstinence. Quinn & Henbest (1967) report a
similar specificity of aversion in most cases in
which negative properties were conditioned to
whiskey alone. Although beer, wine, and whiskey
are used, greatest attention is usually directed

toward the particular type of intoxicant that the
client most prefers. The conditioning trials are

continued until the alcoholic stimuli alone elicit
nauseous reactions, and ingestion of the different

varieties of liquor produces prompt emesis. At the
conclusion of the treatment the client is instructed

that he must, in the future, abstain totally from all
alcoholic beverages.
There have been some variations in the

conditioning procedure originally developed by
Voegtlin. Miller, Dvorak, & Turner (1960) report
that excellent aversions to alcohol can be

developed by group applications of this method.
The authors report that simultaneous presence of
several persons undergoing treatment frequently

produces contagious emesis, thereby facilitating
the negative conditioning process.
Many European therapists have employed a

counterconditioning method devised by Feldmann
(DeMorsier & Feldmann, 1950), in which
apomorphine serves as the UCS, and treatment
sessions continue at 2 to 4 hour intervals until
complete aversion has been achieved toward all
varieties of alcoholic beverages.
The search continues for an unconditioned
stimulus which would have sufficiently strong
aversion-producing properties without
disagreeable side effects, but which at the same
time permitted the administration of

counterconditioning methods on an out-patient
basis. Miller (1959), for example, reports

considerable success in the treatment of
alcoholism with hypnotically induced aversion.
The client is hypnotized and instructed to
reexperience vividly his worst hangover, including

the general malaise, headache, nausea, and
vomiting. With the onset of nausea and emesis the
client is asked to smell and taste alcoholic
beverages including whiskey, wine, and beer. In
addition to the conditioning trials, the client is also

given direct suggestions that, in the future, the
taste or smell of alcohol will promptly evoke
disagreeable nauseous feelings. Both Miller and
Strel’chuk (1957), who has similarly experimented
with verbal induction methods under hypnosis,
maintain that hypnotically induced aversion
responses are more stable than those produced by

emetic drugs, although no comparative data are
furnished to support this claim.
EFFICACY OF AVERSION THERAPY
Numerous outcome studies have been

reported indicating the rates of abstinence from
alcoholic beverages that follow the application of
aversive counterconditioning. In evaluating the
outcomes of any form of treatment for alcoholism,
it is important to bear in mind several

qualifications. Precise assessment of a person’s
alcohol intake would necessitate continuous
monitoring of his activities. Since this is obviously
both ethically objectionable and impractical,
outcomes are typically measured in terms of self-

reports of drinking behavior; ratings by persons
who are well acquainted with the client; public
records of intoxication; and various indirect
indices based on adequacy of social, physical, and

occupational functioning (Hill & Blane, 1967).
It has been widely agreed, on the assumption
that alcoholics can never resume a controlled and

less excessive pattern of drinking, that complete
abstinence from alcoholic beverages should be the
main objective of any remedial program.

Consequently, the efficacy of behavioral
approaches is typically evaluated in terms of
duration of sobriety achieved among its clients. In
recent years several investigators (Davies, 1962;
Kendell, 1965) have reported that a small

percentage of alcoholics with a long history of
addictive drinking have been able to drink in

moderation after treatment. If one employed a
measure of change in drinking behavior rather

than the stringent criterion of total abstinence, the
proportion of cases deriving some benefit from a
therapeutic program would be somewhat higher.
However, the validity of more refined

improvement ratings is open to question,
considering that assessment of pretreatment
drinking behavior is usually based on
retrospective reports rather than on direct
measurement of amount and pattern of alcohol

consumption. Mello & Mendelson (1965) have
developed a sensitive measure of drinking
behavior that could be employed to study changes
in alcoholic intake. Participants are given free
access to an operant conditioning device on which

they can work for either alcohol or monetary
reinforcement. After performing a certain number

of responses a small amount of alcohol is
dispensed or money points are recorded on a
counter, depending upon which of these

reinforcers is selected. This procedure permits
detailed examination of the pattern of alcohol
consumption over a specified period.
Although a major purpose of treatment should
be to modify the problem behavior for which
clients seek help, the efficacy of a given method of

treatment can be best evaluated in terms of its
total consequences. This is particularly true of
chronic alcoholism, which has profound adverse

effects on social, marital, occupational, and other
areas of functioning. However, in emphasizing the
value of measuring multiple outcomes, therapists
too often discount the relevance of the abstinence
criterion (Hill & Blane, 1967). This reordering of

criteria is often accompanied by enumeration of
hazardous consequences that can result from
cessation of drinking. Assessment of drinking
behavior is therefore largely neglected in favor of
inferred psychological changes that can make any
form of therapy look good even though it has
failed to achieve its intended objective. To
safeguard against perpetuation of weak methods
on the basis of extraneous criteria, evaluation

research should include assessment of drinking
behavior, regardless of whatever other outcome
criteria one wishes to invoke.
Table 8-1 Abstinence Rates Obtained by Aversion Therapy

Investigator Number Aversive Complete Duration of
of Cases Stimulus Abstinence Follow-up
(%)
Edlin, 63 Emetine 30 3-10
Johnson, months
Hletko, &
Heilbrunn
(1945)
Kant (1945) 31 Emetine 80 Unspecified
Lemere & 4096 Emetine 51 1–10 years
Voegtlin
(1950)
Miller, 10 Emetine 80 8 months
Dvorak, &
Turner
(1960)
Shanahan & 24 Emetine 70 9 months
Hornick
(1946)
Thimann 275 Emetine 51 3-7 years
(1949)
Wallace 31 Emetine 42 4-17
(1949) months
DeMorsier 150 Apomorphine 46 8-31
& Feldmann months
(1950)
Mestrallet & 183 Apomorphine 41
Lang (1959)
Ruck Apomorphine 50 1.5 years
(1956)
Kantorovich 20 Electric 82 3 weeks-20
(1934) shock months
Blake 25 Electric 23 12 months
(1967) shock
37 Electric 48 12 months
shock with
relaxation
training
Miller 24 Verbally 83 9 months
(1959) induced
aversion
Anant 26 Verbally 96 8-15
(1967) induced months
aversion
Ashem & 15 Verbally 40 6 months
Donner induced
(1968) aversion
Table 8-1 summarizes the percentages of
complete abstinence obtained by different

investigators employing aversive
counterconditioning. The figures in the table
generally do not include cases whose drinking
status was unknown because they could not be

located in subsequent follow-up studies. These
methods, of course, are not applied under
psychologically sterilized conditions. The
conditioning events are socially administered,

clients are undoubtedly given some practical
suggestions for more constructive means of coping
with their life situations, and they are probably
socially reinforced for maintaining sobriety. It is
also undoubtedly true that by the time alcoholics

appear for aversion therapy they have been
recipients of considerable wise counsel,

impassioned appeals by significant people in their
lives, repeated admonitions, rewards, and a

variety of remedies, to no avail. Treatment
outcomes are frequently attributed to common

social influences as though these were
encountered for the first time in the treatment
situation.
The reported variability in abstinence rates
most likely reflects the differential time intervals
at which the various investigators conducted their

follow-up evaluations. In general, abstinence rates
are extremely high in the period immediately
following treatment; the incidence of reversions to

drinking is greatest between 6 and 12 months;
thereafter abstinence declines gradually with
increasing duration. The fact, however, that some
variation in outcomes is found even when similar
procedures and follow-up intervals are involved,

suggests that the differences may also be partly
attributable to inadequate implementation of
requisite conditioning procedures, differences in
sample characteristics, and variations in the extent
to which environmental contingencies are
unfavorable for maintaining sobriety. The
conspicuous absence of any controlled
experimentation in this area makes it impossible
to determine the degree to which treatment

outcomes may be differentially affected by the
nature of the aversive stimulus, the number and
distribution of conditioning sessions, clients’
resources for alternative modes of response to
stress, and environmental contingencies

associated with drinking behavior.
The outcome data reported by Voegtlin and his
associates merit some discussion since they reflect
the most judicious and extensive application of the

principle of counterconditioning to the treatment
of alcoholism. Except for instances where the
therapy was contraindicated for physical reasons

(4 percent) and cases who refused to accept
treatment after detoxication (5 percent), aversion
therapy was offered to all applicants without
further selection. Consequently, an extremely wide
age range, varied socioeconomic levels, and
practically all occupational groups are
represented. Statistical analyses of follow-up data

from approximately 3000 cases treated over a
period of ten years reveal numerous significant
correlates of abstinence (Voegtlin & Broz, 1949).
Clients under 25 contributed the lowest rates of
sobriety (23 percent), whereas the incidence of

abstinence increased with each succeeding age
interval. Although occupational status, in itself, did
not appear to represent an important source of
variance, clients who presented a history of
unemployment and frequent job changes proved

to be considerably less responsive (21 percent)
than those with relatively stable employment

histories (71 percent). Similarly, a considerably
smaller proportion of charity cases (20 percent)
remained abstinent compared to middle class (49

percent) or wealthy (62 percent) participants.
Continued association with drinking companions
accounted for a number of reversions to
alcoholism. It is not surprising, therefore, that

clients who joined abstinence clubs maintained
much higher sobriety rates (87 percent) than

those who refused to associate with such groups
(40 percent).
Of particular interest is the finding that clients’

willingness to, and actual participation in, periodic
reconditioning sessions during the year
immediately following treatment (when most
alcoholic reversions occur) significantly increased
the probability of continued abstinence (Voegtlin

et al., 1942). Of a total number of 155 clients who
initially agreed to participate in the post-therapy
program, 91 percent remained abstinent during
the year of the study, whereas the corresponding
figure for 73 clients who refused to volunteer for
follow-up sessions was 71 percent. In order to
provide a further comparison group, every fourth
case was not offered the opportunity to receive
additional conditioning trials following the

completion of the basic treatment. The latter
control group yielded a 70 percent abstinence
rate.
It is difficult to determine from these data the
influence of motivation to change per se, since

some of the clients who initially volunteered for
additional sessions failed to return; conversely, an
unspecified number of control cases, who
subsequently learned of the program, voluntarily

requested and received the supplemental
treatment. It is clear from the within-groups
analysis, however, that abstinence rates are

positively related to the number of supplementary
conditioning sessions (Table 8-2). Based on the
overall findings of this study, an alcoholic who is
favorably disposed toward continued periodic
treatment has an 86 percent chance of remaining
abstinent for at least one year.
Table 8-2 Percentage of Abstinence as a Function of the

Number of Supplementary Conditioning Sessions
Number of Number of Cases Percentage of Abstinence
Supplementary
Sessions
None 88 74
One 113 80
Two 57 95
Three 20 90
Four or more 7 100
Aversive counterconditioning is thus a simple,

brief, economical, and relatively effective method
for producing aversion to alcohol for at least a
limited period, and for continued total abstinence

in approximately 50 percent of the clients. The
aversion form of therapy offers the additional
advantages of ready acceptance by clients and

wide applicability. Contraindications primarily
include certain physical disorders such as
gastrointestinal ulceration or hemorrhage, hernia,
hepatic cirrhosis, cardiac conditions, and coronary
disease. Despite this relatively favorable showing

aversion treatment of alcoholism has never been
widely accepted, and with the advent of
disulfiram, applications of counterconditioning
procedures have further declined. Although

disulfiram therapy relies upon aversive
contingencies for the maintenance of sobriety, as
we shall see next, this approach is aimed primarily
at suppressing drinking behavior and does not
necessarily alter the positive valence of alcoholic

beverages.
DISULFIRAM REGIMEN
In 1948, Hald & Jacobsen reported

experiments in which they found that persons
who had ingested Antabuse or disulfiram

(Tetraethylthiuram disulphide) for a period of
time experienced intensely aversive physiological
reactions when they subsequently drank even
small amounts of alcohol. The disulfiram-alcohol

reactions (Bowman, Simon, Hine, Macklin, Crook,
Burbridge, & Hanson, 1951) consist initially of
disagreeable warmth or flushing in the face,
conjunctiva and tachycardia occurring

approximately 5 to 20 minutes after the intake of
alcohol, followed during the next 20 to 50 minutes
by headache, dyspnea, dizziness, nausea and
vomiting, chest pains, physical weakness, pallor,
and hangover symptoms. These reactions, which

usually persist for 1 to 2 hours, apparently result
from the action of disulfiram in blocking the

oxidation of alcohol at the stage of acetaldehyde.
Because of the slow elimination of the disulfiram

substance, a single dose can render a person
physiologically sensitized to alcoholic beverages

for a relatively long time. Reports of encouraging
results obtained through this method, along with
its simplicity, led to the widespread use of
Antabuse treatment for alcoholism.
The therapeutic regimen generally proceeds
along the following lines: On the first day

following detoxication the client is orally
administered 1 to 2 grams of disulfiram, and is
given progressively diminishing dosages on the

next three days. After the primary intolerance for
alcohol has been established, the client is then
given one or more test trials of alcohol in order to
determine the optimum maintenance dosage of
Antabuse. The dosage is adjusted individually to

the level where the characteristic, unpleasant side
effects of the drug are reduced to a minimum, but
the dosage is still adequate to produce sufficiently
intense reactions to deter further drinking. The
reactions to the test doses also serve to impress
upon the client the serious physical consequences
of ingesting even small amounts of alcohol while
on disulfiram. Following discharge, the client is
placed on a maintenance dosage which usually

varies from &frac18; to &frac12: gram tablet of
disulfiram taken each day either before breakfast
or in the evening (Bowman et al., 1951; Child,
Osinski, Bennett, & Davidoff, 1951).
Because of the violent physiological reactions

that can be elicited by alcohol when disulfiram is
present in the body, the primary intolerance to
alcohol and the maintenance dose are generally
established during a brief period of hospitalization

with the client under careful observation.
However, Martensen-Larsen (1953), who has
written authoritatively about this mode of

therapy, describes a therapeutic regimen that may
be conducted on an outpatient basis. The initial
dosage of disulfiram, calculated on the basis of 15
milligrams per kilogram of body weight, is
administered to the client during his first
consultation visit. After the first treatment, the
size of the optimum maintenance dose is regulated

during a period of several months in accordance
with the individual’s response to alcohol test trials
and reported side effects.
Considerable statistical data are available
related to the efficacy of the disulfiram regimen. In

general, the abstinence rates associated with
follow-up periods of varying duration are
essentially of the same magnitude as those
obtained by aversive counterconditioning

(Bourne, Alford, & Bowcock, 1966; Bowman et al.,
1951; Brown & Knoblock, 1951; Child et al., 1951;
Epstein & Guild, 1951; Hoff & McKeown, 1953;

Jacobsen, 1950; Shaw, 1951). Despite the
voluminous statistical data relating to the
counterconditioning and disulfiram approaches,
there is a paucity of comparative investigations in
which the relative efficacy of different modes of
therapy is systematically assessed within the same
experimental design.
The problem of evaluating different treatment
approaches is further complicated by the fact that
psychotherapists employing traditional interview
procedures generally confine their reports to
prescriptive statements concerning the proper

conduct of therapy or elaborate accounts of
psychodynamic processes, but they generally fail
to cite objective data regarding the efficacy of this
type of psychotherapeutic enterprise.

Questionnaire surveys further reveal that
psychotherapists are reluctant to treat alcoholics
because of their aversive disruptive behavior

(Hollingshead, 1956; Robinson & Podnos, 1966).
When interview methods are applied, the clinical
reports convey the impression that successful
outcomes are disappointingly low. Although there
are no adequate data available for estimating
precisely the base rates of change in addictive
drinking without psychotherapeutic interventions,

the average figures quoted are generally on the
order of 10 to 15 percent.
Comparative experimental studies of
conditioned aversion and disulfiram therapies are
particularly essential, because these two

approaches have proved to be most efficacious for
modifying and controlling chronic drinking
behavior. However, unlike counterconditioning
methods which involve few risks and

contraindications, potentially serious physical
effects may result from the use of disulfiram
should the client ingest moderate or large

amounts of alcohol while on the drug. Apart from
the physiological reactions to alcohol, a number of
unpleasant side effects of disulfiram have also
been noted; these include drowsiness, nausea,
headache, unpleasant taste and body odor,
gastrointestinal disturbances, and sometimes
decreased sexual potency. These attendant

reactions may, in themselves, be sufficiently
disturbing to lead clients to terminate medication.
There is some evidence, however, that such side
effects can be substantially reduced by decreasing
the daily maintenance dose (Child et al., 1951;

Martensen-Larsen, 1953). The disulfiram
treatment regimen is also generally
contraindicated for clients suffering from
cardiovascular disorders, cirrhosis of the liver,
nephritis, diabetes, epilepsy, advanced

arteriosclerosis, and in cases of pregnancy.
It is possible that eventually an effective anti-

alcohol agent will be discovered that produces few
unpleasant side effects. Ferguson (1956), for
instance, reports a drug, citrated calcium
carbimide (CCC), whose action is similar to that of
disulfiram in inhibiting acetaldehyde metabolism,
but which is free of some of its disagreeable
features. A preliminary experiment in which

different groups of alcoholics were treated with
CCC and with disulfiram revealed that fewer
subjects in the CCC group discontinued medication
of their own accord (Armstrong & Kerr, 1956).
In view of the possible physical manifestations

associated with disulfiram and the inconvenience
of continuous self-medication, the selection of this
mode of therapy over the shorter, safer, and more
economical counterconditioning methods would

be justified only if the pharmacological approach
were shown to yield significantly higher rates of
successful outcomes. In a study comparing the

relative efficacy of aversive counterconditioning,
Antabuse, group hypnotherapy, and milieu
therapy, Wallerstein (1957) found that Antabuse
was most efficacious according to an aggregate
rating based on degree of abstinence, general
social adjustment, “subjective feelings of
difference,” and changes in “personality structure.”

Results for the conditioning group, however, are at
such marked variance with those achieved by
other investigators that the findings of this
experiment must be accepted with reservation.
Yanushevskii (1959) analyzed the follow-up data

on 2000 alcoholics who had received either
medication, psychotherapy, hypnosis,
apomorphine-counterconditioning, or disulfiram
in a Moscow clinic. Conditioned aversion and
disulfiram proved superior to the other

procedures, but both of these therapeutic
approaches produced essentially similar

abstinence rates. Since there is no way of
determining what selective criteria were
employed in assigning cases to the different

treatment groups, these results have only
suggestive value. In view of the limited and
conflicting findings, any conclusion regarding the
relative efficacy of the methods under discussion

must be deferred until adequate empirical data
become available.
It should be emphasized that modifications in
drinking behavior produced by aversive
counterconditioning and by anti-alcohol drugs are

achieved through entirely different mechanisms.
In the case of disulfiram, abstinence is maintained
on a chemical basis. As long as the pills are taken
regularly, the potent physiological consequences
of drinking serve as a powerful deterrent.

However, the conditioning of aversive properties
to alcoholic beverages is precluded by the
relatively long temporal interval between the
ingestion of alcohol on the one hand, and the onset
of the aversive consequences on the other.
Consequently, alcohol retains its positive value
and the client is able to drink within several days
after terminating medication. Many alcoholics, in
fact, will take disulfiram intermittently and go on

drinking sprees during periods when their
physiological tolerance for alcohol has been
restored. The duration and degree of abstinence is,
therefore, contingent on the duration and
regularity with which medication is used

(Jacobsen, 1950).
The temporal prerequisites for aversive
conditioning are also absent from methods in
which nauseants are added to alcoholic beverages.

Under these conditions a person will refrain from
drinking emeticized cocktails but retain his strong
attraction to unmedicated alcoholic drinks. In

addition to physiologically induced restraints
against the use of alcohol, physical prevention
methods were also employed as a means of
ensuring sobriety in the early history of the
treatment of alcoholism. Persons were given
posthypnotic suggestions that they would develop
arm paralysis whenever they attempted to drink

alcoholic beverages. This mode of therapy not only
resulted in a considerable amount of spilt liquor,
but also fostered the acquisition of highly
ingenious drinking styles.
Unlike the preceding approaches, by creating

aversion reactions to the smell, taste, and thought
of alcohol, counterconditioning procedures
directly reduce the positive value of intoxicants
and, therefore, do not require the continuance of

externally imposed deterrents to drinking.
MULTIFORM TREATMENT OF ALCOHOLISM
Aversive counterconditioning alone has

proved most successful with alcoholics who have
developed their habituation by way of prolonged
heavy social drinking, and who possess sufficient
personal resources to derive adequate
gratifications from sober behavior (Thimann,
1949; Voegtlin & Broz, 1949). It is sometimes
erroneously assumed by critics of treatment

programs aimed at the direct modification of
drinking behavior that these approaches are based

on the premise that alcohol is the sole problem of
the alcoholic. Quite to the contrary, they assume
that psychological functioning involves a
reciprocal influence process in which the

characteristics of behavior are important
determiners of the way the environment responds
to it; as a person changes, so does his
environment. Sustained abstinence is therefore
largely ensured not by the fact that liquor has

been endowed with negative properties, but
because elimination of drinking behavior removes
the adverse consequences of chronic inebriation
and creates new reinforcement contingencies with
respect to a broad range of behavior. The
restoration of physical well-being and the positive
experiences derived from improved social,

marital, and financial functioning can reinforce
sobriety and reduce aberrant tendencies. For this

reason, “neuroses” and grossly deviant behaviors
often disappear after alcoholism has been brought
under control (Jellinek, 1962; Thompson &
Bielinski, 1953). Evidently, the risks of aversion

therapy, even when employed as the sole method
of treatment, are minor compared to the hazards
of chronic inebriety.
The fact that 40 to 60 percent of the alcoholics

who receive aversion therapy eventually resume
excessive drinking after a period of abstinence
clearly reveals that, in certain cases, this method
must be supplemented with, or replaced by, other
programs if sobriety is to be maintained.
Traditional interpretive therapy is generally
assumed to be best suited for such cases, a belief
that persists despite evidence that, of the various
treatments available, interview approaches have

proved least successful in modifying chronic
drinking behavior.
Development of Behavioral Competencies.
Alcoholics whose drinking behavior has been
temporarily controlled are unlikely to remain

abstinent for long if they lack the behavioral
competencies for securing gratifications while
sober. Individuals deficient in educational
achievements and satisfying vocational skills and

those who have failed, for one reason or another,
to acquire interpersonal adroitness will be
subjected to considerable negative experiences. As

alcohol is increasingly used to provide escape
from an unrewarding existence, the alcoholic’s
initially minimal competencies typically undergo
further deterioration, resulting in even greater
aversive experiences and avoidance. Therefore,
alcoholics who present behavioral deficits require
a treatment program employing modeling and

reinforcement procedures designed to establish
behavioral competencies. It is assumed that as
positively reinforcing behaviors are developed
they will compete with, and eventually replace,
alcoholic escape and avoidance.
In a pilot project Narrol (1967) employed

reinforcement principles to promote vocational
activities in chronic, hospitalized alcoholics. A
simulated economy was devised in which points

earned for performance of work assignments were
used to purchase commissary items, clothing,
hospital leaves, recreational opportunities, and

room and board on wards that varied in the
comforts and freedom they provided. All members
in this project devoted approximately twice as
much time to their work assignments as did
alcoholics for whom the hospital privileges were
not made contingent upon work performance. In
accord with previous findings, this project

demonstrates that a simulated economy will
control behavior as long as the contingencies
remain in effect. To test the therapeutic efficacy of
this type of program, reinforcement practices
would need to be applied over a long period of

time, and extended to a wide range of behaviors,
including drinking behavior under conditions
where alcoholic beverages are readily available.
Modification of Self-Reinforcement Patterns.

There is a third class of alcoholics who experience
a great deal of aversive stimulation, not because of
behavioral deficits or unavailability of rewarding

resources, but because they impose exceedingly
severe performance demands on themselves.
Events and accomplishments considered worthy
of self-approval by most persons are viewed by
alcoholics who have set themselves high standards
for self-reinforcement as marginal or inadequate
performances. Consequently, such persons not

only deny themselves warranted gratifications but
they also engage in a great deal of self-castigation,
from which they periodically escape through
alcoholic intoxication. The primary objective in the
treatment of alcoholics who are escaping from

self-generated aversive consequences would
involve the lowering of standards for self-
reinforcement, rather than the elimination of
behavioral deficits.
Desensitization of Stress-Provoking Situations.

Under conditions where drinking behavior is
strongly controlled by relief from aversive

stimulation, a desensitization form of treatment
would constitute the method of choice. Kraft and
Al-Issa (1967a, b) report success in modifying
alcoholism by desensitizing clients to stressful
interpersonal situations that typically provoked
them to drink. Although the sample sizes are too
small to draw reliable conclusions, Blake (1967)

found that alcoholics who received aversion
therapy combined with relaxation training were
better able to maintain complete abstinence over a
one-year period than those who were
administered aversive counterconditioning alone.

Had relaxation been deliberately employed to
neutralize sources of tension and anxiety the
differences might have been even more marked.
The discussion thus far has focused on the

individual methods best suited for modifying
different conditions which may exert control over
drinking behavior. In many cases, achievement of

stable changes in alcoholism requires a
combination of treatment procedures in which
people are desensitized to situations they find
stressful; they acquire rewarding patterns of
behavior which will become prepotent over
alcoholic self-anesthetization; they develop
aversions to alcoholic beverages; and they are

taught other self-control techniques for
forestalling drinking in their natural environment.
Social-Systems Approach to the Treatment of
Widespread Alcoholism. Counterconditioning, or
any other individual treatment approach, is of

little value in modifying the drinking behavior of
Skid Row alcoholics. The impoverished personal
resources of these disadvantaged persons, and the
deviant reinforcement contingencies existing

within the Skid Row milieu, serve as powerful
influences in shaping a more or less irreversible
alcoholic destiny (Pittman & Gordon, 1958).
The Skid Row subculture provides social aliens
a refuge from an otherwise demoralizing and
unrewarding existence, in which few demands are
made of its members, anonymity is respected,
failures and deviant tendencies are not
condemned, and achievement strivings and other

behaviors that are actively reinforced in the larger
society are either nonrewarded or disapproved
(Jackson & Connor, 1953). Within this milieu, in
addition to the low level of aversive control and

the adoption of anti-achievement standards, the
major positive reinforcements center around
drinking behavior. Social prestige is largely
contingent on being a good drinking partner and
on the ability to get enough to drink without

having to resort to gainful employment. Similarly,
most interpersonal rewards occur during the

mutual sharing of alcohol in drinking cliques or
“bottle gangs.” These patterns of reinforcement

not only promote continuous inebriation, but by
attaching negative values to societal norms and

demands, they also establish barriers to reentry
into the larger community.
Except for repeated incarceration for public

intoxication and conspicuous violation of other
legal codes, Skid Row alcoholics receive little or no
constructive attention. For the most part, they are

left to themselves to control their own behavior
according to their own deviant norms. Therefore
the rehabilitation of Skid Row alcoholics must

involve an extensive resocialization process that
can be achieved only in a markedly different
environment. If they are to be successful
participants in the larger society, alcoholics must
acquire, among other things, a new set of

incentives and behavioral norms, a wide variety of
social competencies, and esteem- and income-
producing vocational skills. Social organizations
such as Alcoholics Anonymous (1952) may
provide some of the learning conditions necessary
for the attainment of resocialization objectives,
but these programs fail to reach those who, on the
basis of an extended extinction history, have little
reason to share the societal values, and who are

therefore unlikely to seek rehabilitation
voluntarily. Since any radical change in the
alcoholic’s social activities will inevitably lose him
the gratifications associated with the drinking
subculture, willingness to undergo relatively

extensive behavioral modifications cannot be
obtained without the provision of more rewarding
alternatives.
To accomplish fundamental changes in the

behavior of persons from a deviant subculture, it
is necessary to create social systems that provide
the necessary conditions for learning new styles of

life. Such a system must teach new skills; it must
furnish exemplary role models; and it must
embody a set of reinforcement contingencies that
will counteract deviant activities and promote
more constructive modes of behavior. It is
interesting to note, as dramatically illustrated by
the Synanon approach to the treatment of drug

addiction (Yablonsky, 1965), that these types of
social systems often prove most successful when
they are developed by persons for their own
rehabilitation.
Ethical Considerations in Aversion Therapy
The use of aversive procedures in the
modification of human behavior generally meets

with either a cool or a hostile reception on the part
of professional psychotherapists. In some cases

aversive techniques are applied in an ethically
objectionable manner that justifies censorious
reactions. For example, exceedingly noxious
procedures are occasionally employed even
though they produce no greater changes than
stimuli in much weaker intensities (Campbell,
Sanderson, & Laverty, 1964; Clancy, Vanderhoof, &
Campbell, 1967; Hsu, 1965). Oswald (1962) has
promulgated an ill-conceived procedure in which

clients not only receive massed aversion trials, but
are also subjected to disparaging personal
remarks played endlessly on a recorded tape. The
rationale for these needless personal assaults,
some of which are recorded in the therapist’s

voice, is apparently based on Sargent’s (1957)
impressions that social conversions are facilitated
by intense emotional crises. Considering that the
verbalizations were “designed deliberately to
disturb the patient emotionally,” it is

understandable that a number of the clients
exhibited suspicious animosity, unplugged tape

recorders, and refused to submit themselves to
this inexorable ordeal. It should be emphasized
that the brainwashing prescription, which is

antithetical to practices derived from learning
principles, is better designed to instill marked
aversion to therapeutic agents than to stimulus
events that evoke deviant behavior in clients.
Many of the applications of
counterconditioning based on nauseous drugs

employ massed aversion trials in which the
procedures are continuously administered at two-
hour intervals over a period of several days.

Raymond, who originally devised this method, has
subsequently questioned the necessity for such a
rigorous regimen (Raymond, 1964). Apart from
ethical considerations, massed aversive
experiences can produce many undesirable side

effects that seriously obstruct progress. With
repeated administrations of pharmaceuticals,
physical tolerance develops and drugs become less
effective. Therapists are, therefore, forced to use
increasingly larger dosages or less desirable
emetic mixtures in order to induce sufficient
emesis (Cooper, 1963). Moreover, persons who
are repeatedly subjected to unpleasant
experiences without any rewarding respites and

opportunities for positive relationships with the
therapists administering the aversive procedures
are likely to develop resentment, antipathy to the
entire treatment situation, and escape behavior.
Many clients who have completed a course of

aversion therapy may need to return occasionally
for supplementary sessions if they find themselves
reverting to their formerly deviant activities. One
effect of a massive aversive experience is to
reinforce strong avoidance of the treatment

situation itself on future occasions, even though a
limited reconditioning experience might yield

highly beneficial results. For these reasons, a
substantial amount of positive reinforcement
should be incorporated into aversive conditioning

procedures.
If the major purpose of aversion experiences is
to provide clients with a means of exercising
control over harmful behavior, then clients should

play an active role in practicing self-control
techniques in the presence of progressively

stronger evocative stimuli, rather than serving
merely as passive recipients of stimulus pairings.
Thus, for example, in the treatment of alcoholism,
after persons have been taught how to self-induce

nauseous feelings, they should be exposed for
increasingly longer periods to social and stress
situations that involve high instigation to drinking

behavior. In this type of program self-control
techniques are developed, repeatedly tested, and

adequately reinforced.
The fact that some applications of aversive
procedures contain objectionable features does
not warrant a blanket indictment of the
responsible use of aversion therapy (Allchin,
1964; Matthews, 1964) under special
circumstances demanding a more drastic form of
therapeutic intervention, any more than one

would be justified in condemning the use of
surgical and dental procedures with patients
willing to undergo a brief painful experience in
order to alleviate more deleterious and long-
lasting suffering. The brief discomfort occasioned

by a program of aversion therapy is minor
compared to the repeated incarceration, social
ostracism, serious disruption of family life, and
self-condemnation resulting from uncontrollable
injurious behavior. It is a therapist’s responsibility

to provide clients with information about the
treatment alternatives available to them and

which outcomes are most likely to result from
each choice. Given this knowledge, it should be the
client’s right to decide what types of treatment, if

any, he wishes to undergo.
As noted earlier, aversion therapy has proved
least effective with sexual deviants who are
coerced into treatment in an attempt to change

their behavior in the direction of conformity with
more conventional practices. In cases where their

conduct threatens the welfare of others, they have
the choice of either altering their injurious
behavior or having their freedom revoked. There
are other forms of sexual activities, however,

which are also legally prohibited, such as cross-sex
dressing and homosexuality involving consenting
adults, that generally have no adverse

consequences for others. As sexual mores undergo
further changes private consensual sexual

behavior between adults will eventually be
legalized. Nevertheless, deviant sexual practices
will continue to be subjected to social ridicule and,

hence, to serve as a source of emotional
disturbance. After threat of criminal sanctions is
removed, persons who seek to alter their sexual
orientation will do so under motivational
conditions that are more favorable for achieving

behavioral changes.
Summary
This chapter is primarily concerned with

classical aversion treatment of exceedingly
persistent behavior that is maintained by

inappropriate, potentially harmful, or culturally
prohibited positive reinforcers. This stimulus-
oriented approach attempts to establish control

over behavior by endowing formerly attractive
stimuli or symbolic representations of deviant
activities with negative properties through
contiguous association with aversive experiences.
These negative experiences are typically induced
by administering nauseous pharmacological
agents or unpleasant shocks, or they may take the
form of symbolically revivified feelings of
revulsion.
Aversive counterconditioning is not regarded

as a process in which aversive reactions become
directly and automatically attached to formerly
positive stimuli. Rather, the counterconditioning
procedure establishes an aversive self-stimulation
mechanism which enables persons to counteract

the disposition to engage in deviant behavior by
symbolically reinstating nauseous reactions
previously experienced in treatment. The most
direct evidence that conditioned aversions

represent, in large part, self-induced reactions,
rather than automatic products of stimulus
pairings, is provided by laboratory studies

demonstrating that classically conditioned
responses are amenable to symbolic control.
Viewed from this perspective, aversive
counterconditioning creates a means of self-
control rather than automatic immunity to
addictive or rewarding stimuli.
Aversive procedures have been applied most

extensively to harmful addictive behaviors and to
various types of sexual aberrations. Treatment
programs aimed at modifying sexual disorders
usually involve differential conditioning of sexual
responsiveness, wherein aversion is developed to

fetishistic, transvestite, or homoerotic stimuli,
while concurrently erotic arousal properties are
conditioned to heterosexual stimuli. The summary
of results, based mainly on single case studies,

indicates that this mode of therapy can, in addition
to eliminating deviant sexuality, assist in
promoting heterosexual behavior, provided that

these alternative responses already exist in the
person’s repertoire and that environmental
conditions are favorable for maintaining them. On
the other hand, an exclusively aversive form of
therapy is much less effective in cases where
heterosexual behaviors are either lacking or
strongly inhibited. In the latter conditions,

aversive counterconditioning must be
supplemented with procedures designed to
develop behavioral patterns that will enable
persons to engage in rewarding heterosexual
interactions.
Similarly variable outcomes have been

obtained in the counterconditioning treatment of
chronic alcoholism. Aversive procedures alone
have proved most successful with alcoholics who

possess sufficient personal resources to derive
adequate gratifications from sober behavior. In
most cases, however, the treatment must also be

directed toward the conditions that control
drinking behavior. This might involve
development of behavioral competencies to the
point where sober behavior is sufficiently
reinforcing to predominate over anesthetic
avoidance; the lowering of standards for self-
reinforcement that result in self-generated

aversive consequences; and, in cases where
excessive drinking is controlled by relief from
interpersonal stresses, the desensitization of
primary sources of tension and anxiety.
Of particular interest are studies

demonstrating that symbolic events, which may
serve as important internal elicitors of deviant
behavior, are modifiable through aversive
conditioning. When imagery possessing affective

value is repeatedly paired with negative
experiences, the symbolic events not only lose
their arousal potential but they are less frequently

self-generated. The imaginal counterconditioning
is generally accompanied by reductions in the
corresponding behavior.
The major value of aversive procedures is that
they provide a rapid means of achieving control
over injurious behavior for a period during which

alternative, and more rewarding, modes of
behavior can be established. A treatment which

addresses itself to both stimulus and response
events is most likely to yield uniformly favorable
results because it not only alters the valence of
stimuli that evoke deviant behavior, but also

creates reinforceable response patterns.
REFERENCES
Abrams, S. An evaluation of hypnosis in the treatment

of alcoholics. American Journal of Psychiatry,
1964, 120, 1160-1165.
Agras, W. S. Behavior therapy in the management of

chronic schizophrenia. American Journal of
Psychiatry, 1967, 124, 240-243.
Alcoholics Anonymous. Twelve steps and twelve

traditions. New York: Alcoholics Anonymous
Publishing Co., 1952.
Allchin, W. H. Behaviour therapy. British Journal of
Psychiatry, 1964, 110, 108.
Anant, S. S. A note on the treatment of alcoholics by a

verbal aversion technique. Canadian Psychologist,
1967, 8, 19-22.
Armstrong, J. D., & Kerr, H. T. A new protective drug in

the treatment of alcoholism. Canadian Medical
Association Journal, 1956, 74, 795-797.
Ashem, B., & Donner, L. Covert sensitization with

alcoholics: a controlled replication. Behaviour
Bales, R. F. Cultural differences in rates of alcoholism.

Quarterly Journal of Studies on Alcohol, 1946, 6,
480-499.
Bancroft, J. H. J., Jones, H. G., & Pullan, B. R. A simple

transducer for measuring penile erection, with
comments on its use in the treatment of sexual
disorders. Behaviour Research and Therapy, 1966,
4, 239-241.
Barker, J. C., & Miller, M. Aversion therapy for

compulsive gambling. Lancet, 1966,1, 491-492.
Barker, J. C., Thorpe, J. G., Blakemore, C. B., Lavin, N. I.,

& Conway, C. G. Behaviour therapy in a case of
transvestism. Lancet, 1961, 1, 510.
Barlow, D. H., Leitenberg, H., & Agras, W. S. The
experimental control of sexual deviation through
manipulation of the noxious scene in covert
sensitization. Journal of Abnormal Psychology,
1969, 74, 596-601.
Barry, H., Wagner, A. R., & Miller, N. E. Effects of

alcohol and amobarbital on performance
inhibited by experimental extinction. Journal of
55, 464-468.
Blake, B. G. A follow-up of alcoholics treated by

behaviour therapy. Behaviour Research and
Therapy, 1967, 5, 89-94.
Blakemore, C. B., Thorpe, J. G., Barker, J. C., Conway, C.

G., & Lavin, N. I. The application of faradic
aversion conditioning in a case of transvestism.
Behaviour Research and Therapy, 1963,1, 29-34.
Blough, D. S. Technique for studying the effects of

drugs on discrimination in the pigeon. Annals of
the New York Academy of Sciences, 1956, 65, 334-
344.

inhibition therapy to exhibitionism. In H. J.
Eysenck (Ed.), Experiments in behaviour therapy.
New York: Pergamon, 1964. pp. 80-86.
Bourne, P. G., Alford, J. A., & Bowcock, J. Z. Treatment
of Skid Row alcoholics with disulfiram. Quarterly
Journal of Studies on Alcohol, 1966, 27, 42-48.
Bowman, K. M., Simon, A., Hine, C. H., Macklin, E. A.,

Crook, G. H., Bur- bridge, N., & Hanson, K. A
clinical evaluation of tetraethylthiuram
disulphide (Antabuse) in the treatment of
problem drinkers. American Journal of Psychiatry,
1951, 107, 832-838.
Brown, C. T., & Knoblock, E. C. Antabuse therapy in the

Army. A preliminary report of fifty cases. U. S.
Armed Forces Medical Journal, 1951, 2, 191-202.
Bruun, K. Significance of role and norms in the small

group for individual behavioral changes while
drinking. Quarterly Journal of Studies on Alcohol,
1959, 20, 53-64.
Campbell, D., Sanderson, R. E., & Laverty, S. G.

Characteristics of a conditioned response in
human subjects during extinction trials following
a single traumatic conditioning trial. Journal of
6.39.
Carpenter, J. A. Effects of alcoholic beverages on skin

conductance: An exploratory study. Quarterly
Casey, A. The effect of stress on the consumption of
alcohol and reserpine. Quarterly Journal of
Studies on Alcohol, 1960, 21, 208-216.

16, 33-41.
Chafetz, M. E., & Demone, H. W., Jr. Alcoholism and

society. New York: Oxford University Press, 1962.
Child, G. P., Osinski, W., Bennett, R. E., & Davidoff, E.

Therapeutic results and clinical manifestations
following the use of tetraethylthiuram disulfide.
American Journal of Psychiatry, 1951,107, 774-
780.
Clancy, J., Vanderhoof, E., & Campbell, P. Evaluation of

an aversive technique as a treatment of
alcoholism. Quarterly Journal of Studies on
Alcohol, 28, 476-485.
Clark, D. F. Fetishism treated by negative

conditioning. British Journal of Psychiatry,
1963,109, 404-407. (a)
Clark, D. F. Treatment of fetishism by negative

conditioning—A further note. British Journal of
Psychiatry, 1963,109,695-696. (b)
Clark, R., & Polish, E. Avoidance conditioning and
alcohol consumption in rhesus monkeys. Science,
1960, 132, 223-224.
Clark, R. A., & Sensibar, M. R. The relationship

between symbolic and manifest projections of
sexuality with some incidental correlates. Journal
334.
Conger, J. J. The effects of alcohol on conflict behavior

in the albino rat. Quarterly Journal of Studies on
Alcohol, 1951,12, 1-29.
Conn, J. H. Ilypno-synthesis. IV. Hypnotherapy of the

sex offender. Journal of Clinical and Experimental
Hypnosis, 1954, 2, 13-26.
Cooper, A. J. A case of fetishism and impotence treated

by behaviour therapy. British Journal of
Psychiatry, 1963, 109, 649-652.

162-166.
Curran, D., & Parr, D. Homosexuality: An analysis of

100 male cases seen in private practice. British
Davies, D. L. Normal drinking in recovered alcohol

addicts. Quarterly Journal of Studies on Alcohol,
1962, 23, 94-104.
Davison, G. C. Elimination of a sadistic fantasy by a

client-controlled counterconditioning technique:
A case study. Journal of Abnormal Psychology, 73,
84-90.
DeMorsier, G., & Feldmann, H. Le traitement

biologique de l’alcoolisme chronique par
l’apomorphine. Etude de 200 cas. (The biological
treatment of chronic alcoholism with
apomorphine. Study of 200 cases.) Schweizer
Archiv fur Neurologie und Psychiatrie, 1950, 65,
472-473.
Docter, R. F., & Perkins, R. B. The effects of ethyl

alcohol on autonomic and muscular responses in
humans. I. Dosage of 0.5 milliliter per kilogram.
374-386.
Edlin, J. V., Johnson, R. H., Hletko, P., & Heilbrunn, G.

The conditioned aversion treatment of chronic
alcoholism. Preliminary report. Archives of
Neurology and Psychiatry, 1945, 53, 85-87.
Epstein, N. B., & Guild, J. Further clinical experience

with TEDT in the treatment of alcoholism.
366-380.
Eriksen, C. W., & Kuethe, J. L. Avoidance conditioning
of verbal behavior without awareness: A
paradigm of repression. Journal of Abnormal and
Evans, D. R. An exploratory study into the treatment

of exhibitionism by means of emotive imagery
and aversive conditioning. Canadian Psychologist,
1967, 8,162.
Evans, D. R. Masturbatory fantasy and sexual

deviation. Behaviour Research and Therapy, 1968,
6, 17-19.
Eysenck, H. J. Behaviour therapy, extinction, and

relapse in neurosis. British Journal of Psychiatry,
1963, 109, 12-18.
Ferguson, J. K. W. A new drug for alcoholism

treatment. 1. A new drug for the treatment of
alcoholism. Canadian Medical Association Journal,
1956, 74, 793-795.
Fleck, S. Family dynamics and origin of schizophrenia.

Psychosomatic Medicine. 1960, 22, 333-344.
Ford, C. S., & Beach, F. A. Patterns of sexual behavior.

New York: Harper, 1951.
Fort, T., & Porterfield, A. L. Some backgrounds and

types of alcoholism among women. Journal of
Health and Human Behavior, 1961, 2, 283-292.
Franks, C. M. Conditioning and conditioned aversion

therapies in the treatment of the alcoholic.
International Journal of Addiction, 1966, 1, 61-96.
Freund, K. Some problems in the treatment of

homosexuality. In H. J. Eysenck (Ed.), Behaviour
therapy and the neuroses. New York: Pergamon,
1960. pp. 312-325.
Freund, K. A laboratory method for diagnosing

predominance of homo- or hetero-erotic interest
in the male. Behaviour Research and Therapy,
1963, 1, 85-93.
Freund, K. Diagnosing homo- or heterosexuality and

erotic age-preference by means of a
psychophysiological test. Behaviour Research and
Therapy, 1967, 5,209-228. (a)
Freund, K. Erotic preference in pedophilia. Behaviour

Research and Therapy, 1967, 5,209-348. (b)
Freund, K., Sedlacek, F., & Knob, K. A simple

transducer for mechanical plethysmography of
the male genital. Journal of the Experimental
Fromer, R., & Berkowitz, L. Effect of sudden and

gradual shock onset on the conditioned fear
response. Journal of Comparative and
Garcia, J., Kimeldorf, F. J., & Koelling, R. A. Conditioned

aversion to saccharin resulting from exposure to
gamma radiation. Science, 1955, 122, 157-158.
Gelder, M. G., & Marks, I. M. Aversion treatment in

transvestism and transsexualism. In R. Green
(Ed.), Transsexualism and sex reassignment.
Baltimore: Johns Hopkins Press, 1969 (in press).
Giffin, M. W., Johnson, A. M., & Litin, E. M. Antisocial

acting out: 2. Specific factors determining
antisocial acting out. American Journal of
Orthopsychiatry, 1954, 24, 668-684.
Glad, D. D. Attitudes and experiences of American-

Jewish and American-Irish male youth as related
to differences in adult rates of inebriety.
406-472.
Glynn, J. D., & Harper, P. Behaviour therapy in

transvestism. Lancet, 1961, 1, 619.
Goorney, A. B. Treatment of a compulsive horse race

gambler by aversion therapy. British Journal of
Psychiatry, 1968, 114, 329-333.
Greenberg, L. A., & Carpenter, J. A. The effect of
alcoholic beverages on skin conductance and
emotional tension. I. Wine, whisky, and alcohol.
190-204.
Hald, J., & Jacobsen, E. A drug sensitizing the organism

to ethyl alcohol. Lancet, 1948, 2, 1001-1004.

Psychology, 1969 (in press).
Hess, E. H. Pupillometric assessment. In J. M. Shlien

(Ed.), Research in psychotherapy. Washington, D.
C.: American Psychological Association, 1968. pp.
573-583.
Hess, E. H., Seltzer, A. L., & Shlien, J. M. Pupil response

of hetero- and homosexual males to pictures of
men and women: A pilot study. Journal of
Hill, M. J., & Blane, H. T. Evaluation of psychotherapy

with alcoholics. Quarterly Journal of Studies on
Alcohol, 1967, 28, 76-104.
Hoff, E. C., & McKeown, C. E. An evaluation of the use

of tetraethylthiuram disulfide in the treatment of
560 cases of alcohol addiction. American Journal
of Psychiatry, 1953,109, 670-673.
Hogans, A. F., Moreno, C. M., & Brodie, D. A. Effects of
ethyl alcohol on EEG and avoidance behavior of
chronic electrode monkeys. American Journal of
Physiology, 1961, 201,434-436.
Hollingshead, A. B. Views on the etiology of

alcoholism. In H. D. Kruse (Ed.), Alcohol as a
medical problem. New York: Harper, 1956. pp. 56-
67.
Holmberg, A. R. The Siriono. Unpublished doctoral

dissertation, Yale University, 1946.
Hsu, J. J. Electroconditioning therapy of alcoholics. A

preliminary report. Quarterly Journal of Studies
on Alcohol, 1965, 26, 449-459.
Isbell, H., Fraser, H. F., Wikler, A., Belleville, R. E., &

Eiseman, A. J. An experimental study of the
etiology of “Rum fits” and delirium tremens.
1-33.
Jackson, J. K., & Connor, R. The skid road alcoholic.

468-486.
Jacobsen, E. Biochemical methods in the treatment of

alcoholism, with special reference to Antabuse.
Proceedings of the Royal Society of Medicine, 1950,
43, 519-526.
James, B. Case of homosexuality treated by aversion

therapy. British Medical Journal, 1962,1, 768-770.
Jellinek, E. M. Phases of alcohol addiction. In D. J.

Pittman & C. R. Snyder (Eds.), Society, culture, and
drinking patterns. New York: Wiley, 1962. pp.
356-368.
Johnson, A. M. Factors in the etiology of fixations and

symptom choice. Psychoanalytic Quarterly, 1953,
22, 475-496.
Kant, F. Further modifications in the technique of

conditioned reflex treatment of alcohol addiction.
229-232.
Kant, F. The use of conditioned reflex in the treatment

of alcohol addicts. Wisconsin Medical Journal,
1945, 44, 217-221.
Kantorovich, N. V. An attempt of curing alcoholism by

associated reflexes. Novoye Refleksologii nervnoy i
Fiziologii Sistemy, 1928, 3, 436-445. Cited by
Razran, G. H. S., Conditioned withdrawal
responses with shock as the conditioning
stimulus in adult human subjects. Psychological
Bulletin, 1934, 31,111-143.
Kaplan, H. S. The effects of alcohol on fear extinction.
Dissertation Abstracts, 571-572.
Kendell, R. E. Normal drinking by former alcohol

addicts. Quarterly Journal of Studies on Alcohol,
1965, 26, 247-257.
Koenig, K. P., & Masters, J. Experimental treatment of

habitual smoking. Behaviour Research and
Therapy, 1965, 3, 235-243.
Kolb, L. C., & Johnson, A. M. Etiology and therapy of

overt homosexuality. Psychoanalytic Quarterly,
1955, 24, 506-515.
Kolvin, I. “Aversive imagery” treatment in

adolescents. Behaviour Research and Therapy,
1967, 5, 245-248.
Korman, M., & Stephens, H. D. Effects of training on

the alcohol consummatory response in rats.
Korn, S. J. The relationship between individual

differences in the responsivity of rats to stress
and intake of alcohol. Quarterly Journal of Studies
on Alcohol, 1960, 21, 605-617.
Korpmann, E., & Hughes, F. W. Potentiating effect of

alcohol on tranquilizers and other central
depressants. AM A Archives of General Psychiatry,
1959, 1,7-11.
Kraft, T., & Al-Issa, I. Alcoholism treated by

desensitization: A case report. Behaviour
Kraft, T., & Al-Issa, I. Desensitization and reduction in

cigarette consumption. Journal of Psychology,
1967, 67,323-329. (b)
Kushner, M., & Sandler, J. Aversion therapy and the

concept of punishment. Behaviour Research and
Therapy, 1966, 4, 179-186.
Lavin, N. I., Thorpe, J. G., Barker, J. C., Blakemore, C. B.,

& Conway, C. G. Behaviour therapy in a case of
transvestism. Journal of Nervous and Mental
Disease, 1961, 133, 346-353.
Lemere, F., & Voegtlin, W. L. Conditioned reflex

therapy of alcoholic addiction: Specificity of
conditioning against chronic alcoholism.
California and Western Medicine, 1940, 53, 268-
269.
Lemere, F., & Voegtlin, W. L. An evaluation of the

aversion treatment of alcoholism. Quarterly
Lemere, F., Voegtlin, W. L., Broz, W. R., O’Hollaren, P.,

& Tupper, W. E. Conditioned reflex treatment of
chronic alcoholism: VII. Technic. Diseases of the
Nervous System, 1942, 3, 243-247. (a)
Lemere, F., Voegtlin, W. L., Broz, W. R., O’Hollaren, P.,

& Tupper, W. E. The conditioned reflex treatment
of chronic alcoholism: VIII. A review of six years’
experience with this treatment of 1526 patients.
Journal of the American Medical Association, 1942,
120, 269-270. (b)
Lester, D. Self-maintenance of intoxication in the rat.

223-231.
Lester, D. Self-selection of alcohol by animals, human

variation, and the etiology of alcoholism.
395-438.
Lienert, G. A., & Traxel, W. The effects of

meprobamate and alcohol on galvanic skin
response. Journal of Psychology, 1959, 48, 329-
334.
Litin, E. M., Giffin, M. E., & Johnson, A. M. Parental

influence in unusual sexual behavior in children.
Psychoanalytic Quarterly, 1956, 25, 37-55.
McCarthy, R. (Ed.) Drinking and intoxication. Glencoe,

Ill.: Free Press, 1959.
McConaghy, N. Penile volume change to moving
pictures of male and female nudes in
heterosexual and homosexual males. Behaviour
McDonnell, G. J., & Carpenter, J. A. Anxiety, skin

conductance, and alcohol. A study of the relation
between anxiety and skin conductance and the
effect of alcohol on the conductance of subjects in
groups. Quarterly Journal of Studies on Alcohol,
1959, 20, 38-52.
McGuire, R. J., Carlisle, J. M., & Young, B. G. Sexual

deviations as conditioned behaviour: A
hypothesis. Behaviour Research and Therapy,
1965, 2, 185-190.
McGuire, R. J., & Vallance, M. Aversion therapy by

electric shock: A simple technique. British
Malinowski, B. The sexual life of savages in north-

western Melanesia. New York: Harcourt, Brace,
1929.
Mardones, J. Experimentally induced changes in the

free selection of alcohol. International Review of
Neurobiology, 1960, 2, 41-76.
Marks, I. M., & Gelder, M. G. Transvestism and
fetishism: Clinical and psychological changes
during faradic aversion. British Journal of
Psychiatry, 1967,113,711-729.
Marks, I. M., Rachman, S., & Gelder, M. G. Methods for

assessment of aversion treatment in fetishism
with masochism. Behaviour Research and
Therapy, 1965, 3, 253-258.
Martensen-Larsen, O. Five years’ experience with

disulfiram in the treatment of alcoholics.
406-418.
Masserman, J. H. Behavior and neurosis. Chicago:

Masserman, J. H., & Yum, K. S. An analysis of the

influence of alcohol on experimental neurosis in
cats. Psychosomatic Medicine, 1946, 8, 36-52.
Matthews, P. C. Behaviour therapy. British Journal of

Psychiatry, 1964, 110, 108.
Max, L. W. Breaking up a homosexual fixation by the

conditioned reaction technique: A case study.
Psychological Bulletin, 1935, 32, 734 (Abstract).
Mees, H. L. Sadistic fantasies modified by aversive

conditioning and substitution: A case study.
320.
Mello, N. K., & Mendelson, J. H. Operant analysis of

drinking patterns of chronic alcoholics. Nature,
1965, 206, 43-46.
Mendelson, J. H., & La Dou, J. Experimentally induced

chronic intoxication and withdrawal in
alcoholics. Part 2, Psychophysiological findings.
Quarterly Journal of Studies on Alcohol, 1964,
Supplement No. 2, 14-39.
Mestrallet, A., & Lang, A. Indications, techniques et

resultats du traitement par l’apomorphine de
l’alcoolisme psychiatrique. (Indications,
technique, and results in apomorphine therapy of
psychiatric alcoholism). Le Journal de Medecine de
Lyon, 1959, 40, 279-285.
Miller, E. C., Dvorak, A., & Turner, D. W. A method of

creating aversion to alcohol by reflex
conditioning in a group setting. Quarterly Journal
of Studies on Alcohol, 1960, 21, 424-431.
Miller, M. M. Treatment of chronic alcoholism by

hypnotic aversion. Journal of the American
Medical Association, 1959, 171, 1492-1495.
Miller, M. M. Hypnotic-aversion treatment of

homosexuality. Journal of the National Medical
Association, 1963, 55, 411-415.
Miller, N. E. Some recent studies of conflict behavior

and drugs. American Psychologist, 1961,16, 12-24.
Morgenstern, F. S., Pearce, J. F., & Rees, W. L.

Predicting the outcome of behaviour therapy by
psychological tests. Behaviour Research and
Therapy, 1965, 2, 191-200.

Narrol, H. G. Experimental application of

reinforcement principles to the analysis and
treatment of hospitalized alcoholics. Quarterly
Oswald, I. Induction of illusory and hallucinatory

voices with considerations of behaviour therapy.
Journal of Mental Science, 1962, 108, 196-212.
Pawlowski, A. A., Denenberg, V. H., & Zarrow, M. X.

Prolonged alcohol consumption in the rat. 2.
Acquisition and extinction of an escape response.
232-240.
Peacock, L. J., & Watson, J. A. Radiation-induced

aversion to alcohol. Science, 1964,143, 1462-
1463.
Perloff, W. H. Hormones and homosexuality. In J.
Marmor (Ed.), Sexual inversion. New York: Basic
Books, 1965. pp. 44-69.
Pittman, D. J., & Gordon, C. W. Revolving door: A study

of the chronic police case inebriate. Glencoe, Ill.:
Free Press, 1958.
Pittman, D. J., & Snyder, C. R. (Eds.) Society, culture,

and drinking patterns. New York: Wiley, 1962.
Quinn, J. T., & Henbest, R. Partial failure of

generalization in alcoholics following aversion
therapy. Quarterly Journal of Studies on Alcohol,
1967, 28, 70-75.
Rachman, S. Aversion therapy: Chemical or electrical?

299.
Rachman, S. Sexual fetishism: An experimental

analogue. Psychological Record, 293-296.
Raymond, M. J. Case of fetishism treated by aversion

therapy. British Medical Journal, 1956, 2, 854-
857.
Raymond, M. J. The treatment of addiction by aversion

conditioning with apomorphine. Behaviour
Raymond, M., & O’Keeffe, K. A case of pin-up fetishism
treated by aversion conditioning. British Journal
of Psychiatry, 1965, 111, 579-581.
Richter, C. P. Production and control of alcoholic

craving in rats. In Neuropharmacology,
Proceedings of the third conference. New York:
Josiah Macy, Jr. Foundation, 1956. pp. 39-146.
Robinson, L. H., & Podnos, B. Resistance of

psychiatrists in treatment of alcoholism. Journal
of Nervous and Mental Disease, 1966, 143, 220-
225.
Rodgers, D. A., & McClearn, G. E. Mouse strain

differences in preference for various
concentrations of alcohol. Quarterly Journal of
Ruck, F. Alkoholentziehungskur mit Hilfe eines

bedingten Reflexes
(Apomorphinentziehungskur). (Conditioned
reflex treatment of alcoholism.) Psychiatrie,
Neurologie, und medizinische Psychologie, 1956, 8,
88-92.
Sargent, W. Battle for the mind: A physiology of

conversion and brainwashing. London:
Heinemann, 1957.
Shanahan, W. M., & Hornick, E. J. Aversion treatment
of alcoholism. Hawaii Medical Journal, 1946, 6,
19-21.
Shaw, I. A. The treatment of alcoholism with

tetraethylthiuram disulfide in a state mental
hospital. Quarterly Journal of Studies on Alcohol,
1951, 12, 576-586.
Sidman, M. Technique for assessing the effects of

drugs on timing behavior. Science, 1955,122, 925.
Skolnik, J. H. The stumbling block. A sociological study

of the relationship between selected religious
norms and drinking behavior. Unpublished
doctoral dissertation, Yale University, 1957.
Smart, R. G. Effects of alcohol on conflict and

avoidance behavior. Quarterly Journal of Studies
on Alcohol, 1965, 26, 187-205.
Snyder, C. R. Alcohol and the Jews: A cultural study of

drinking and sobriety. Glencoe, Ill.: Free Press,
1958.
Solyom, L., & Miller, S. A differential conditioning

procedure as the initial phase of the behaviour
therapy of homosexuality. Behaviour Research
and Therapy, 1965, 3, 147-160.
Stevenson, I., & Wolpe, J. Recovery from sexual

deviations through overcoming of non-sexual
neurotic responses. American Journal of
Psychiatry, 1960, 116, 737-742.
Stollak, G. E. Weight loss obtained under various

experimental procedures. Unpublished
manuscript, Indiana University, 1968.
Stoller, R. J. Transvestites’ women. American Journal

of Psychiatry, 1967, 124, 333-339.
Strel’chuk, I. V. C novykh sovremennykh metodakh

lecheniya bol’nykh alkogolizmom. [New
contemporary methods of treating patients with
alcoholism.] Sovietskaia Meditsina, 1957, 21 (5),
26-33.

365.
Sutherland, E. H., Schroeder, H. G., & Tordella, C. L.

Personality traits and the alcoholic. A critique of
existing studies. Quarterly Journal of Studies on
Alcohol, 1950, 11, 547-561.
Syme, L. Personality characteristics and the alcoholic.
A critique of current studies. Quarterly Journal of
Thimann, J. Conditioned-reflex treatment of

alcoholism. II. The risks of its application, its
indications, contraindications, and
psychotherapeutic aspects. New England Journal
of Medicine, 1949, 241, 406-410.
Thompson, G. N., & Bielinski, B. Improvement in

psychosis following conditioned reflex treatment
for alcoholism. Journal of Nervous and Mental
Disease, 1953, 117, 537-543.
Thorpe, J. G., Schmidt, E., Brown, P. T., & Castell, D.

Aversion-relief therapy: A new method for
general application. Behaviour Research and
Therapy, 1964, 2, 71-82.
Thorpe, J. G., Schmidt, E., & Castell, D. A comparison of

positive and negative (aversive) conditioning in
the treatment of homosexuality. Behaviour
Research and Therapy, 196.3, 1, 357-362.
Voegtlin, W. L. The treatment of alcoholism by

establishing a conditioned reflex. American
Journal of Medical Science, 1940, 199, 802-810.
Voegtlin, W. L., & Broz, W. R. The conditioned reflex

treatment of chronic alcoholism. X. An analysis of
3125 admissions over a period of ten and a half
years. Annals of Internal Medicine, 1949, 30, 580-
597.
Voegtlin, W. L., Lemere, F., Broz, W. R., & O’Hollaren, P.

Conditioned reflex therapy of chronic alcoholism.
IV. A preliminary report on the value of
reinforcement. Quarterly Journal of Studies on
Alcohol, 1942, 2, 505-511.
Wall, J. H. A study of alcoholism in men. American

Journal of Psychiatry, 1936, 92, 1389-1401.
Wallace, J. A. The treatment of alcoholism by the

conditioned-reflex method. Journal of the
Tennessee Medical Association, 1949, 42, 125-128.
Wallerstein, R. S. (Ed.) Hospital treatment of

alcoholism; a comparative, experimental study.
New York: Basic Books, 1957.
Whitener, R. W., & Nikelly, A. G. Sexual deviation in

college students. Journal of Orthopsychiatry,
1964, 34, 486-492.
Wilde, G. J. S. Behaviour therapy for addicted cigarette

smokers: A preliminary investigation. Behaviour

Wolpe, J. Conditioned inhibition of craving in drug
addiction: A pilot experiment. Behaviour Research
and Therapy, 1965, 2, 285-288.
Woodward, M. The diagnosis and treatment of

homosexual offenders. British Journal of
Delinquency, 1958, 9, 44-59.
Yablonsky, L. The tunnel back: Synanon. New York:

Macmillan, 1965.
Yanushevskii, I. K. Effektivnost’ protwoolykogol’ nogo

techeniya po dannym katamneza. [The
effectiveness of antialcohol treatment according
to follow-up data.] Zhurnal nevropatologii i
psikhiatrii, 1959, 59, 693-696.
9
Symbolic Control of
Behavioral Changes
Both complex behaviors and even relatively
simple performances that have generally been
assumed to represent direct linkages between
external stimuli and overt responses are
extensively controlled by symbolic processes.
These higher level activities involve, among other

things, strategic selection of the stimuli to which
attention is directed, symbolic coding and
organization of stimulus inputs, and acquisition,

through informative feedback, of mediating
hypotheses or rules which play an influential role
in regulating response selection.
The introductory chapter considered in some
detail questions regarding variables that govern
the occurrence of symbolic events, the different
forms that mediators may take, and the conditions

under which they exercise cue function in
directing overt actions. The present chapter is
focused on the extent to which behavioral change

processes are affected by persons’ awareness of
stimuli impinging upon them, responses that they
are exhibiting, consequences that their behavior

incurs, and the contingencies that exist among the
latter events. In addition, relationships between
attitudinal and behavioral changes are reviewed

with particular reference to the development of
self-regulatory mechanisms.
Role of Awareness of Contingencies in

Behavioral Change
A number of different theories have been

suggested concerning the functional role of
symbolic activities, which is usually subsumed
under the general term “awareness,” in the
behavioral change process. The main alternative
views are depicted in a simplified form in Figure

9-1, following the schematization of Farber (1963)
and Spielberger & DeNike (1966).
According to the nonmediational theory of
learning (Skinner, 1953; Thorndike, 1933),
reinforcing consequences act directly and

automatically to strengthen preceding overt
responses. While learning occurs independently of
awareness, a person may eventually recognize the
reinforcement contingencies from the high output

of correct responses. In this view, however,
awareness is a resultant rather than a
precondition of change.
The independent response systems theory
(Verplanck, 1962), which is a more recent version
of the foregoing position, treats awareness as
Figure 9-1. Schematization of the functional relationship
between awareness and response change. Dashed lines
represent temporally contiguous events, arrows denote
causal relationships, and plus signs designate the
magnitude of response change.
merely a verbal operant rather than as a factor
that controls performance. Since verbal and
nonverbal classes of behavior represent

independent response systems, almost any type of
relationship can be obtained between these two
sets of events, depending upon the manner in
which their governing reinforcement

contingencies are arranged. Thus, verbalizations
and actions will be congruent under conditions
where the same contingencies are applied to
verbal statements and the corresponding
instrumental responses. On the other hand,

verbalizations and actions can be made to diverge
through the application of conflicting

reinforcement to these two forms of response.
Since reinforcing stimuli are assumed to exert

automatic control over behavior independently of
their effects upon awareness, this theoretical

position also represents a nonmediational model
of behavioral change.
According to the cognitive view (Dulany, 1962,

1968; Spielberger & DeNike, 1966), which stands
in marked contrast to the preceding formulations,
awareness is considered a prerequisite for

learning and improvement in performance. The
information conveyed by reinforcing stimuli,
rather than their response-strengthening effects,

is highlighted in this point of view. It is assumed
that, in the course of observing the differential
consequences associated with different types of
responses emitted, subjects test various
hypotheses about the required response class and

eventually figure out what they are supposed to
do. The acquired information, in turn, gives rise to
intentions or self-instructions to produce the
correct responses, the strength of the tendency
depending upon subjects’ valuation of the
contingent incentives. The magnitude of
performance gains is thus primarily a function of
the accuracy of the guiding hypothesis and the
incentive value of the consequences. However, no

conditioning can presumably occur without either
partial or correct symbolic representation of the
reinforcement contingencies.
The reciprocal interaction theory (Farber,
1963; Postman & Sassenrath, 1961), on the other

hand, assumes that awareness is both a
consequent and a condition of behavioral change.
According to this view, a certain amount of
learning can take place from the automatic action

of aftereffects, independently of subjects’
understanding of the basis on which
reinforcements are administered. During the

learning process, however, subjects not only make
overt responses, but they also develop thoughts or
hypotheses about the responses required to
obtain reinforcement. These self-generated rules
serve as discriminative stimuli for directing
instrumental actions in essentially the same way
as external stimuli control behavior. Accurate

hypotheses are likely to be accompanied by
correct overt responses, whereas erroneous
hypotheses tend to coincide with inappropriate
performances. Consequently, symbolic events are
selectively strengthened, maintained, or

extinguished by the differential reinforcements
administered for the more distally occurring overt
behavior. The emergence of awareness may, of
course, be facilitated by performance gains which
make the contingencies more obvious. Once the

correct hypothesis is established, it can result in a
substantial increase of appropriate responding,

given adequate incentive conditions.
The acquisition of rules and their functions in
regulating performance are typically studied in
concept identification and other forms of
discrimination learning. In these paradigms
subjects must categorize different stimuli on the
basis of some common property that the

experimenter has arbitrarily selected as relevant
for the classification. In more complex situations
the correct responses are defined in terms of a
combination of attributes rather than a single
common element. Under these circumstances,

subjects must abstract the relevant stimulus
dimensions and formulate a rule about how the
different attributes combine to specify the
appropriate behavior (Bourne, 1966; Shepard,
Hovland, & Jenkins, 1961).
Numerous experiments have been conducted
in which subjects are asked to state the rules they

employed for making responses. When hypotheses
are measured after subjects have performed and
experienced the outcome, it is difficult to
determine whether the responses were derived
from a rule or the rule was inferred from the
correct responses. This interpretive problem does
not arise when rule statements are recorded prior

to performance and measures are obtained of the
degree to which they control trial-by-trial
responding. Studies of this type show that rules
can be strengthened either by direct
reinforcement or indirectly through response

outcomes and that they serve as a primary
determinant of overt behavior (Dulany &
O’Connell, 1963; O’Connell & Wagner, 1967).
However, under conditions where stimuli are
more complex and verbal control over responding

is not explicitly encouraged, accurate performance
often occurs in the absence of adequate verbalized

rules (Hislop & Brooks, 1969).
Several different approaches, some of which
were discussed earlier, have been employed in
experimental analyses of the role of symbolic
activities in behavioral change processes. These
have included investigations of verbal
conditioning rates as a function of awareness of

response-reinforcement contingencies,
mediational control of classical conditioning and
extinction, the occurrence of semantic
generalization in which a common cognitive
associate of heterogeneous stimuli provides the

basis for generalization, covert verbal control of
problem-solving activities, and the influence of
recognition and discrimination of weak stimuli
upon discriminative nonverbal behavior.
The methods of assessing symbolic activities

have been equally varied. In some studies
awareness is manipulated instructionally through

explicit descriptions of the response-
reinforcement contingencies given prior to the
conditioning series. More frequently, however,
awareness is inferred from observations about the
experiment reported by subjects either at periodic
intervals during the course of conditioning or in
interviews conducted immediately following the

experimental session. In other cases, awareness is
manipulated indirectly by arranging conditions
that either facilitate or impede recognition of
contingencies during the acquisition process.
Verbal Conditioning as a Function of

Awareness
Innumerable studies employing paradigms of
verbal conditioning have been conducted in order
to determine whether response consequences
increase performance primarily by effecting

voluntary symbolic control over available
responses or through a process of automatic
response strengthening. Although the issue of

whether learning can occur without awareness is
by no means settled (Farber, 1963; Kanfer, 1968;
Postman & Sassenrath, 1961), most experiments
fail to obtain performance gains in the absence of
accurate or at least partially correct hypotheses
regarding the reinforcement contingencies

(Adams, 1957; Dulany, 1962; Spielberger &
DeNike, 1966). Subjects who are able to describe
the responses required for reinforcement
generally show a substantial increase in
appropriate responses, whereas exposure to

reinforcement contingencies is relatively
ineffective in modifying the behavior of subjects
who remain unaware.
Experiments utilizing post-acquisition

measures of awareness furnish inconclusive
results, since it is entirely possible that subjects
may initially condition without awareness and

later recognize the reinforcement principle
employed when it is made more apparent by the
increased output of correct responses. In order to
establish whether awareness precedes or follows
behavioral change it is therefore necessary to
assess subjects’ speculations about the
experimental contingencies at periodic intervals

during the acquisition process. DeNike (1964), for
example, asked college students, who were
reinforced for human noun responses in a word-
naming task, to write down their “thoughts about
the experiment” after each block of 25 words

during conditioning. On the basis of the written
reports, approximately a third of the subjects were
judged to have gained awareness of the
contingency at different points in the series, while
the remainder were categorized as unaware. A

control group of subjects, who were reinforced on
a random basis for 10 percent of their responses,

was also included. As can be seen from Figure 9-2,
aware subjects displayed a substantial increase in
human noun responses, whereas unaware

subjects, like the control group, showed no
Figure 9-2. (A) Mean percent human noun responses given
by aware, unaware, and control groups in the verbal
conditioning task. (B) Mean percent of correct
responses given by subjects in the aware group prior to
and after verbalization of the reinforcement
contingency. Spielberger & DeNike, 1966.
performance gains whatsoever. Of considerably
greater interest, however, is the temporal
relationship obtained between the emergence of

awareness and the occurrence of large
performance increments. Subjects displayed no

significant rise in the number of critical responses
before they were able to report the reinforcement
contingency, but they markedly increased their
output of reinforced responses after they had

discerned the contingency governing the
administration of social rewards (Figure 9-2).
Considering, however, that each block contained

25 trials, it is entirely possible that awareness may
still have resulted from behavior change during

the block in which the contingency was discerned.
A stringent test of mediational control of
performance changes would require a trial-by-

trial inquiry.
Results of experiments conducted by
investigators who differ widely in their views
regarding the role of awareness in learning
consistently demonstrate that symbolic

representation of the conditions of reinforcement
has a strong facilitative effect upon overt
performance. The empirical data diverge,
however, on the question of whether any learning
can take place without symbolic mediation. The

findings of Dulany, DeNike, & Spielberger may be
contrasted with those of Hirsch (1957), Philbrick
& Postman (1955), and Sassenrath (1962), who
likewise have analyzed performance curves as a
function of the temporal appearance of awareness,

and find small but significant improvements in
performance prior to correct statement of

contingencies, particularly among subjects who
eventually developed complete awareness (Figure
9-3). Evidence of verbal conditioning without

awareness is usually dismissed by staunch
Figure 9-3. Average number of correct responses given on
blocks of words at varying distances from the point at
which the principle was first verbalized correctly.
Philbrick & Postman, 1955.
adherents to cognitive explanations as probably
either the result of insensitive methods for
gauging awareness, or the operation of partially

correct hypotheses.
Although virtually all investigators subscribe,

albeit uneasily, to the same operational definition
of awareness (i.e., correct verbalization of
response-reinforcement contingencies), a number

of factors may lead to inaccuracies and
inconsistencies in composing aware and unaware
groups of subjects. First, awareness is usually
treated as an all-or-none phenomenon, when in
fact it may vary in accuracy from a correct

determination, through partially correlated
hypotheses, to highly misleading notions about
why the subject is being rewarded. As Adams
(1957) has noted, partially correct formulations
(e.g., a subject believes that the experimenter is
interested in comments about people when
actually familial references is the correct response
class) can produce some increases in performance.
Because criteria of awareness are somewhat

arbitrary, the types of relationships obtained
between events are partly dependent upon the
stringency of definitions employed by different
investigators.
A second major complication in the assessment

of awareness arises from the fact that a number of
other variables, quite apart from the amount of
relevant information possessed by the subject,
may contribute to the types of verbal reports that

are obtained.
Awareness is frequently inferred from
responses to a series of progressively more
suggestive interview questions. Therefore, the
number of subjects judged to be aware is
determined to some extent by the number and
nature of informative cues conveyed by the
interview probes. The more intensive the
questioning the less the likelihood that partially

aware subjects will be erroneously categorized as
unaware, but also the greater the danger that the
assessment procedure itself may induce
recognition of the correct contingency that did not

exist at the time of conditioning. Thus, Levin
(1961) found evidence of learning without
awareness when subjects’ discernment of
contingencies was estimated from a brief
interview, whereas categorization of the same

subjects on the basis of their replies to a more
extended specific inquiry yielded a peculiar set of

results, in which subjects who were unaware of
both the contingency and the reinforcer displayed

as much conditioning as subjects who were fully
aware, and a higher rate of response than a group

that was only aware of the reinforcing stimulus.
While this irregular pattern of relationships does
not support the common assumption that
comprehensive assessments furnish more valid

measures, it should be noted that early studies of
verbal conditioning, which reported relatively

high incidences of learning without awareness
(Krasner, 1958), generally relied upon brief
examinations that may not have been sufficiently
sensitive to detect partial degrees of awareness.

The unreliability of post-acquisition reports of
awareness is further underscored by Weinstein &
Lawson (1963), who found that interviews of the

kind widely employed in this line of research
yielded complete awareness in only half the

subjects who had been fully informed, midway
through the experiment, of the contingencies and
the entire purpose of the study. Based upon

criteria utilized in earlier experiments,
approximately half the sample would have been
misclassified and their improved performances
interpreted as evidence for learning without
awareness.
In addition to the influence of the type of
measurement procedures employed, the

likelihood that subjects will report awareness is
reduced if they are examined by a female or a
person of low status (Krasner, Ullmann, Weiss, &

Collins, 1961), if a negative valence is attached to
the reinforced class of responses (Krasner &
Ullmann, 1963), if they feel hostile toward the
experimenter (Weiss, Krasner, & Ullmann, 1960),
and if they obtained information about the

contingencies spuriously (Levy, 1967). Moreover,
there is some evidence (Rosenthal, Persinger,
Vikan-Kline, & Fode, 1963) that experimenters
who are biased to expect a high incidence of
awareness obtain it more frequently than those
who assume it to be a relatively uncommon
phenomenon.
It is possible to control, to some extent, for the
operation of experimenter bias by relying upon

written reports rather than interview procedures,
which provide greater opportunities for
inadvertent influence of respondents’ accounts.
Also, reluctance to disclose provisional judgments
about the experiment, and intentional distortions,

might be effectively counteracted by the provision
of positive incentives which would maximize
verbalization of the information that subjects do
possess. It is evident from the data cited above

that if much importance is to be attached to
studies of symbolic mediation of learning based
upon information provided by subjects, then

extensive research is needed to identify the
variables influencing reported awareness with a
view toward further improving the accuracy of
such measures.
Because of the numerous problems associated
with phenomenological data, some researchers
have recommended that awareness be relegated

to the status of a dependent variable and treated
as a verbal operant. This type of approach

decisively solves a technical problem at the
expense of a potentially influential independent
variable which, under some conditions, can exert
more powerful discriminative control over

behavior than reinforcement variables (Ayllon &
Azrin, 1964; Dulany, 1968; Kaufman, Baron, &
Kopp, 1966).
Granted all the deficiencies of verbal reports as

indicants of subjects’ level of awareness, findings
based on the temporal relationship between
awareness and performance nevertheless indicate
that one can predict with considerably greater
accuracy performance increments during the
course of conditioning by taking into account
subjects’ hypotheses than if self-instructional
influences are disregarded.
The host of methodological and interpretive

problems associated with post-acquisition
measures of awareness can be easily avoided by
the use of research procedures in which
knowledge about reinforcement contingencies is
experimentally induced rather than inferred from

subjects’ verbal reports. Experiments in which
subjects are informed of the appropriate
responses and their consequences prior to
conditioning disclose substantial symbolic

regulation of overt performances.
Situations of social conditioning contain a
number of different elements of which a person
might become aware. These separable events
include the environmental cues eliciting his
behavior, the class of responses considered
appropriate to the situation, the occurrence and
scheduling of reinforcements, and the contingent
relation between the latter classes of events. Some

of the experiments in which awareness is
manipulated experimentally have been specifically
designed to compare the relative efficacy of insight
into the different aspects of the behavioral

influence process. Dulany (1962), for example,
found that students who were informed about the
correct response-reinforcement contingency and
those who received only response instructions
markedly increased their output of correct

responses compared both to their baseline rates
and to the performance of control groups who

were apprised only of the reinforcing events or
given no information. The latter groups, in fact,

exhibited no significant improvements in
performance. In this study, knowledge of the

desired behavior was the critical determinant,
since the addition of reinforcing consequences in
the form of either tones, “Um-hmms,” or omission
of electric shocks did not augment the change

process.
In a subsequent study Dulany (1968) had

students perform a verbal conditioning task in a
chamber kept at a temperature of 110° and
utilized a contingent stream of air of either 70°,

100°, or 150° as positive, neutral, or negative
reinforcers for different groups of subjects.
Students within each of the reinforcement
conditions were also given different reinforcement
instructions: that the stream of air signified a

correct response, an incorrect response, or had
nothing to do with their performance. Dulany
found that reinforcement instructions exercised
greater control over conditioning performance
than did the nature of the reinforcing
consequences.
The findings of Dulany are, in large part,
corroborated by Kaufman, Baron, & Kopp (1966),
who provided students with complete or minimal

information about the required response, and
either accurate or erroneous knowledge

concerning the schedule according to which
rewarding consequences would be administered.
One group of students was accurately informed
that rewards would be forthcoming each minute

on the average (variable interval schedule),
whereas other groups were misled into believing
that their behavior would be reinforced either on a

fixed interval of one minute, or after they had
performed 150 responses on the average (variable

ratio schedule).
Inspection of Table 9-1 reveals that knowledge
about the required behavior markedly increased
subjects’ rate of responding. Even more
impressive, however, is the finding that illusory
schedules governed students’ responsiveness in
much the same way as they do in reality: Fixed
interval instructions produced very low rates,

variable ratio instructions maintained an
extremely high response output, and variable
interval instructions generated intermediate rates
of response. Alleged schedules thus outweighed
the influence of the program of reinforcement that

was actually imposed on students’ behavior. A
further study disclosed that reinforcement had
little effect upon performance rate under
conditions of minimal response specification, but
students who had received variable ratio

instructions combined with monetary rewards
were approximately twice as productive as those

who were furnished the same schedule
information without any reinforcing
consequences. Instructional influences can be

equally powerful in regulating responsiveness
under aversive contingencies. Simply exposing
subjects to punishing contingencies proved to be
an extremely inefficient way of altering behavior,

whereas instructions about the appropriate
behavior and its consequences immediately

produced stable and discriminated avoidance
behavior (Baron & Kaufman, 1966; Scobie &
Kaufman, 1969).
Table 9-1 Median Number of Responses Performed per

Minute as a Function of Information about the
Required Response and Alleged Schedules of
Reinforcement (adapted from Kaufman, Baron, & Kopp,
1966)
Response Rate
Experimental Condition First 30 Min. Last 30 Min.
No schedule information
Minimal response instruction 17 17
Complete response instruction 161 161
Schedule information
Variable interval 88 43
Fixed interval 5 7
Variable ratio 250 269
It is noteworthy that among persons who have
achieved complete awareness the extent to which
this knowledge continues to govern their behavior

during extinction depends upon the type and
schedule of reinforcement applied during the

acquisition process (Hirsch, 1957).
The experiments of Ayllon & Azrin (1964), in
which highly persistent behavior of psychotics

was modified in naturalistic settings, reveal that
when discrepancies exist between verbally
represented contingencies and actual
consequences, instructional influences lose their
potency over time, and behavior comes more

extensively under the control of the prevailing
conditions of reinforcement. These data
underscore the necessity for exercising
considerable caution in generalizing about the
relative efficacy of cognitive and reinforcement
variables on the basis of exceedingly brief
experiments conducted with cooperative college
sophomores who are offered trivial rewards for
performing inconsequential responses that

require little expenditure of effort, under
conditions where they are expected to participate
in experiments in partial fulfillment of course
requirements.
The findings of investigations in which

awareness is experimentally induced are in accord
with those furnished by studies relying upon
report-inferred measures. We return now to the
major controversial issue: Is awareness a

prerequisite for behavioral change? Proponents of
cognitive theories have been unable to find any
evidence of verbal conditioning in the absence of

correct or correlated hypotheses, whereas
Postman and his colleagues report, on the basis of
experiments involving more complex
reinforcement contingencies, that a significant
amount of learning can take place prior to
verbalization of the basis for reinforcement. These
divergent conclusions do not appear to be

attributable to any major differences in the
definition and assessment of awareness. Nor can
they be accounted for in terms of the operation of
partially correct hypotheses, since the use of
partially relevant hypotheses in the concept

learning task did not facilitate performance
(Hirsch, 1957; Postman & Sassenrath, 1961), and
the phenomenon is evident even when awareness
is defined to include partially correct
verbalizations (Sassenrath, 1962).
Some additional suggestive evidence of
behavioral change without awareness is furnished

by investigations involving more complex tasks,
such as probability learning, in which persons
predict alternative events or outcomes that vary in
their frequency of occurrence. In these situations,
persons’ choice behavior gradually adjusts to the
event probabilities even though the vast majority
of subjects are not only unable to state the

probability rules, but frequently entertain quite
erroneous hypotheses (Goodnow & Postman,
1955).
Discrepancies in results may, in part, result
from the complexity of the principle governing the

administration of reinforcement and the response
restrictions imposed by the nature of the learning
task. Studies in which verbalization is
accompanied by dramatic performance gains have

generally involved relatively explicit response
classes or simple discrimination tasks in which
subjects are asked to construct sentences by

selecting one of several personal pronouns or
verbs printed on cards. When the critical response
class is unambiguous and the response
alternatives are severely curtailed, both
awareness and “learning” are most likely to occur
as a one-trial event rather than as an incremental
process.
Considering the feeble and inconsequential
nature of the reinforcers employed in most verbal
conditioning experiments, one might seriously
question whether reinforcement processes, which
presumably govern automatic response-

strengthening effects, are even operative in most
of the studies that have been reviewed. This issue
is, of course, of little or no concern to researchers
who are quite content with a circular empirical

law of effect. Nevertheless, it is important to
distinguish between the informational and
incentive properties of feedback stimuli, both of

which may enhance correct responding (Keller,
Cole, Burke, & Estes, 1965). Flashing fights, buzzer
noises, and ambiguous guttural sounds can convey
to subjects adequate information for altering their
behavior, but it is exceedingly doubtful that such
feedback events have much rewarding value as
determined independently of response changes on

the conditioning task.
In view of the fact that verbal conditioning
procedures are designed to augment the
production of existing responses under conditions
where both response and reinforcing events are

highly salient, the findings may have more
relevance to issues of social conformity than to
conditioning. Although the verbal conditioning
paradigm is adequate to demonstrate the

facilitative role of awareness upon performance, it
is poorly suited to throw much light on the more
basic theoretical issue of whether awareness is a

prerequisite for learning or performance change.
The question of whether learning must be
consciously mediated can be answered most
decisively by studies in which the reinforced
responses are not observable to the performer, or
the reinforcing events are devoid of informational

cues but of sufficient rewarding value to activate
mechanisms regulating reinforcement effects.
Either of the latter conditions would effectively
preclude recognition of the contingency employed.

The studies discussed earlier, in which correct
responding in animals was significantly increased
by intravenous presentation of nutritive solutions,
would seem to dispute the radical cognitive view.
Such reinforcing events are not observable and,

therefore, convey no information to the subject.
There is evidence that covert responses, such

as invisibly minute thumb contractions, can be
successfully conditioned in adult humans without

their observation of the rewarded responses
(Hefferline & Keenan, 1963; Hefferline, Keenan, &

Harford, 1959; Sasmor, 1966). These studies
provide further convincing demonstrations of how
persons learn to respond in accordance with
reinforcement contingencies without appropriate

symbolic mediation. In the latter experiments
subjects are equipped with several sets of
electrodes, ostensibly to measure their ability to
relax. A visually imperceptible thumb contraction
of a preselected magnitude, detected by the

experimenter through electromyographic
amplification, is then selected for modification
through reinforcement either in the form of
monetary rewards or termination of aversive
stimulation. Responses in the chosen amplitude

category increase substantially during
reinforcement and decline abruptly after

reinforcement is withdrawn. As might be
expected, none of the subjects could identify the
response that produced reinforcement.
Apart from the laboratory findings, it is
difficult to believe that planaria, goldfish, and
other lower organisms, which lack the anatomical
structures for adequate symbolic representation

of environmental events, are totally unaffected by
response consequences until they have accurately
cognized their experimenter’s contingencies.
Implicit mediators would, of course, assume an
important role in governing performance in tasks

that require response on the basis of relatively
complicated principles or rules.
The overall evidence would seem to indicate
that learning can take place without awareness,

albeit at a slow rate, but that symbolic
representation of response-reinforcement
contingencies can markedly accelerate

appropriate responsiveness. The validity of this
view, which assumes a reciprocal interaction
between awareness and performance gains, seems
even more probable when one realizes the
limitations of paradigms of verbal conditioning for
elucidating the role of symbolic activities in
behavioral change processes.
Interactive Effects of Cognitive and

Incentive Variables
Although awareness of environmental
contingencies is usually accompanied by some

gains in performance relative to baseline rates, the
absolute level of responsiveness after the

contingencies have been either ascertained or
divulged remains comparatively low. That is,
performance increases are generally in the order
of 20-30 percent, which can hardly be considered

a massive outpouring of correct responses. It is
also extremely likely that if the experiments were
extended beyond the usual single session,

symbolic control, in the absence of supporting
incentives, would decrease over time and the
desired behavior might eventually return to its
original level. Moreover, even in short-term
situations embodying high demand
characteristics, a significant number of aware

subjects never do show any change in their
behavior (Farber, 1963). Hence, the findings of
verbal conditioning studies, rather than
demonstrating the potency of symbolic control, in
fact illustrate the limitations of approaches that

rely primarily upon cognitive variables to effect
behavioral changes. The experiments do provide
considerable evidence, however, that awareness
combined with incentive-related variables can
exert a powerful influence over behavior.
Spielberger, Bernstein, & Ratliff (1966)
compared the response rate of aware and

unaware subjects during an initial phase of the
experiment in which “Mm-hmm” served as the
reinforcer, and after an effort was made to bolster
the incentive value of the utterance by challenging
subjects to produce as many “Mm-hmms” as they
possibly could. Students who remained unaware
throughout both phases of the study showed no

evidence of conditioning; those who discovered
the contingency prior to the incentive
manipulation displayed moderate improvements
in performance followed by extremely high
response rates under the heightened motivational

conditions; a third group of subjects, who became
aware after the incentive manipulation, exhibited
an intermediate level of responding (Figure 9-4).
Studies in which reinforcing properties of the

feedback events are evaluated by subjects
(Spielberger, Berger, & Howard, 1963;
Spielberger, Levin, & Shepard, 1962) rather than

being varied independently likewise disclose that,
among aware subjects, those who prize the
reinforcers show a high output of criterion
behavior. By contrast, aware subjects who are
Figure 9-4. Mean percent of correct responses given during
the baseline (A), low-incentive (B), and high-incentive
(C) phases of the experiment by subjects who either
discovered the reinforcement contingency prior to or
after the incentive manipulation or remained unaware
throughout the experiment. Spielberger, Bernstein, &
Ratliff, 1966.
indifferent to, or annoyed by, experimenters’
guttural utterances may perform some correct
responses in order to confirm their speculations,

but otherwise they are about as unproductive as
their unaware counterparts.
In addition to the influence of incentives
specifically associated with the desired behavior,
more generalized sets may determine the extent to

which persons will behave in accordance with
their knowledge of social contingencies. Farber
(1963) and Holmes (1967) found that aware
subjects with cooperative dispositions displayed a

sharp increase in reinforced responses, but
nonconforming aware subjects showed relatively
little change in behavior and did not differ in this

respect from subjects who remained unaware.
While awareness typically facilitates
performance, where the correct responses carry
negative connotations, awareness may exert
inhibitory effects upon performance, as shown in
the study by Ekman, Krasner, & Ullmann (1963).

Among aware subjects, those who were led to
believe that the verbal conditioning task exposed
personal debilities exhibited fewer responses
during reinforced trials compared to their baseline

rates, whereas those who were informed that the
task measured empathy and warmth toward
people showed a substantial increment in
reinforced responses. On the other hand, groups of
unaware students who received the same negative

and benign sets displayed relatively small
response gains and did not differ from each other.
It is apparent from the research thus far
reviewed that awareness of reinforcement

contingencies has greater behavioral
consequences under laboratory conditions than

appears to be the case in naturalistic or
psychotherapeutic situations. This difference may
be attributable to several factors. Based upon the
findings previously discussed, one would not

expect development of insight into social
contingencies to produce much change in behavior
if the customary incentives are weak, delayed, or
only sporadically applied, as is often true in
realistic circumstances. Second, the responses

chosen in experimental studies (e.g., plural
responses, verbs, personal pronouns, emotion-
arousing words) are readily available within
subjects’ repertoires, and the task is primarily a
matter of response selection rather than response

acquisition. In most behavioral change programs,
on the other hand, individuals must develop

behavior requisite to bring them into contact with
prevailing contingencies, rather than merely
gaining information about what one would have to

do in order to obtain reinforcement. Acquired
insights, no matter how valid they may be, have
limited utility for individuals who lack the
necessary performance skills. The case is

analogous to informing English monolingual
students in verbal conditioning experiments that

the criterion responses are Hindustani adjectives.
Symbolic Control of Classical Conditioning

Phenomena
Consistent with findings from studies of
instrumental conditioning, results of numerous

investigations of classical conditioning (Grings,
1965) reveal extensive mediational control of

conditioned autonomic responses. This process
has been demonstrated in several different ways.

In one experimental approach to the problem,
subjects are informed that the CS will sometimes
be followed by shock; they are then given a sample
shock or a single confirmation trial during the
acquisition series when autonomic responses to
the CS are measured. The studies (Bridger &
Mandel, 1964; Cook & Harris, 1937; Dawson &
Grings, 1969) show that autonomic responses are

readily conditioned through association of
stimulus events with anticipatory emotional
responses.
Data of this sort have led to suggestions that a
distinction be drawn between genuine

conditioning and perceptual or relational learning
(Grings, 1965; Razran, 1955). Implicit in this
dichotomy is the assumption that conditioned
emotional responses established through

instructional manipulations represent a “pseudo-
conditioning” phenomenon. An alternative
conceptual scheme, which has the potential of

elucidating the process without proliferating
unnecessary varieties of learning, would hold that
conditioning based upon the occurrence of actual
versus imagined events mainly involves
differences in the locus of the emotion-producing
stimulus rather than in the governing mechanism.
It is extremely unlikely that verbal association

of events alone is sufficient to establish
conditioned responses, since a stimulus capable of
activating the autonomic responses is also
required. There is consistent evidence from
studies in which autonomic responses are

continuously recorded during desensitization
sessions (Clark, 1963; Mackay & Laverty, 1963),
and from controlled laboratory investigations
(Barber & Hahn, 1964), that imagined aversive

events can produce emotional effects analogous to
the actual Occurrence of aversive stimulation.
These findings indicate that subjects can acquire

conditioned responses in the absence of an
externally administered UCS to the extent that
prior instructions lead them to generate fear-
producing thoughts in conjunction with the CS. On
the other hand, verbal instruction alone would not
be expected to produce any conditioning in
subjects who did not engage in appropriate covert

self-stimulation. According to this self-arousal
formulation, instructional conditioning may
represent a variant of the basic conditioning
paradigm in which autonomic responses are
cognitively induced rather than directly elicited by

aversive stimuli under the experimenter’s control.
The foregoing conceptualization implies that,
rather than representing a simple process in
which external stimuli are directly and

automatically connected to overt responses,
classical conditioning is partly mediated through
symbolic activities. In the mediational

interpretation, the CS elicits covert symbolic
activities that produce autonomic responses. Some
suggestive evidence for the influential role of self-
stimulation in instructional conditioning is
provided by Dawson (1966), who found that the
degree of belief in instructions that shock would
follow a certain signal and the amount of

anticipation of shock were positively correlated
with extent of autonomic conditioning.
The influential role of mediational variables in
classical conditioning is also supported by other
lines of evidence. Fuhrer & Baer (1965), for

example, demonstrated that subjects who
recognized the CS-UCS relationship while
undergoing differential conditioning of GSR
responses demonstrated considerable autonomic

conditioning. In contrast, subjects who remained
unaware of the stimulus contingencies did not
respond differently to auditory stimuli that were

associated with shock and to those never paired
with aversive stimulation. Dawson & Grings
(1969) have likewise shown that masked CS-UCS
pairings, which impeded recognition of the
stimulus contingency, were not sufficient to
condition discriminative autonomic responses.
In order to ascertain if covariations in symbolic

behavior and conditioned responses do involve a
causal relationship, Chatterjee & Eriksen (1962)
conducted an experiment in which awareness of
the stimulus contingencies was manipulated in
advance. One group of subjects was informed that

a shock would only follow one particular word in a
chain association task, but that aversive
stimulation would be discontinued at a clearly
designated point in the experiment. A second

group was told that although shock would always
follow a particular word in the list, each of the
remaining words would be paired once with

aversive stimulation and eventually shocks would
cease altogether. A third group was instructed that
a certain number of shocks would be administered
during the experiment without implying a regular
contingency. Subjects in the first two groups, all of
whom discerned the correct word-to-shock
relationship, displayed conditioned heart rate

responses, whereas subjects who received
minimal information and remained unaware
evidenced no conditioning. The controlling power
of symbolic events is further shown by evidence
that aware subjects exhibited strong autonomic

responses to the critical stimulus but they did not
generalize these responses inappropriately along
either semantic or physical dimensions. Moreover,
those who were informed when the extinction
phase commenced showed a prompt and virtually

complete loss of conditioned responses before
experiencing any nonreinforced presentations of

the conditioned stimulus.
In accord with the above finding, the most
striking evidence of symbolic control of classically
conditioned responses is provided by studies in
which extinction of autonomic responsiveness is
compared in subjects who are told that there will
be no further aversive stimulation and in those

who remain uninformed. Induced awareness of
the change in stimulus contingencies generally
results in rapid and practically complete
disappearance of conditioned responses to the CS
(Cook & Harris, 1937; Grings & Lockhart, 1963;

Wickens, Allen, & Hill, 1963). The decrement is
most sudden and dramatic when subjects who,
despite assurances to the contrary, suspect that
they might continue to be shocked are excluded
from the analysis (Bridger & Mandel, 1965). On

the other hand, under circumstances where
arousal level is maintained and the operation of

cognitive factors is curtailed by disguising the
conditioning procedures (Spence, 1966),
extinction proceeds at a comparatively slow rate

after reinforcement has been discontinued.
Although a strong causal relationship has been
established between cognitive variables and rate
of classical conditioning and extinction, it should

not be concluded that all conditioned responses
are necessarily consciously mediated. It seems

extremely unlikely that in the case of interoceptive
conditioning either the conditioned stimulus (e.g.,
intestinal distention) or the unconditioned
stimulus (e.g., chemical stimuli presented

internally) are symbolically represented. In these,
and other experiments employing internal
stimulative procedures (Razran, 1961), the

contingencies are undoubtedly operating below
the threshold of awareness.
There is also some evidence to indicate that
the strength of symbolic control partly depends on
the conditions under which emotional behavior
was originally acquired. Bridger & Mandel (1964)
found autonomic conditioning was similar
regardless of whether the CS was associated with
threat of shock alone or with threat and shock
stimulation (Figure 9-5). However, emotional

responses established on the basis of actual
painful experiences were less susceptible to
cognitive control. Whereas threat- conditioned
responses were promptly abolished by removal of
electrodes and information that shocks would be

discontinued, emotional responses produced by
painful stimulation were much more resistant to
extinction. The latter findings lend support to the
view that conditioned responses typically contain
dual components (Bridger & Mandel, 1965). One

of the component parts, which is produced by the
self-arousal mechanism, is readily manipulable by

varying emotion-provoking cognitions. On the
other hand, the nonmediated component is
directly evoked by external stimulus events and

requires disconfirming experiences for its
1853
Figure 9-5. Mean GSR responses during acquisition and extinction for groups of subjects for whom the
CS was associated with either threat of shock alone or threat plus shock. The pseudo-conditioning
curves show the subjects’ GSR responses to control stimuli that were never paired with either
threat or shock. The latter data provide a control for the effects of general arousal and orienting
mechanisms. Bridger and Mandel, 1964.
extinction.
A study by Mandel & Bridger (1967) of the

interaction between cognitive influences and
stimulus contingencies lends further validity to
the view that conditioning outcomes typically

reflect the operation of both associative processes
and symbolic generative mechanisms. Subjects
who were informed that no further negative

reinforcement would occur showed marked
decrements in conditioned autonomic responding;
nevertheless the rate with which the response
extinguished differed depending on the order in
which conditioned and unconditioned stimuli

were presented during the acquisition period and
on the temporal interval between these stimulus
events.
The nonmediational theory of classical
conditioning assumes that, in order for
conditioning to occur, the associated stimulus
events must at least be registered in the nervous
system of the organism. Therefore, in studies

assessing the role of awareness in conditioning, it
would be of considerable value to obtain evidence
that there has, in fact, been input from the
conditioned stimulus. It is not inconceivable that

in experiments employing masking procedures, in
which subjects’ attention is diverted to irrelevant
features of the task, the conditioned stimuli may
not be registered in a sufficiently consistent
manner to produce stable conditioned responses.

Hernandez-Peon, Scherrer, & Jouvet (1956)
provide evidence, based upon neurophysiological

studies, that attention focused on a particular
stimulus simultaneously reduces afferent signals

activated by other sensory stimuli. The evoked
auditory potential in the cochlear nucleus of eats

to a loud auditory stimulus was virtually
eliminated when they gazed at mice, attentively
sniffed fish odors, or received electric shocks that
distracted their attention. Horn (1960) has

demonstrated a similar diminution of neural
responses to a light flash during active attention to

other visual and auditory cues. Although there is
some dispute about whether the attenuation of
sensory signals results from mechanisms
operating at the periphery or at more central

levels, there is no doubt that neural responses to
afferent input can be substantially reduced by
attending behavior directed toward irrelevant

stimulus events.
Even in the absence of an experimentally

induced diverting set, some subjects may choose
to attend closely to extraneous stimuli, and thus
fail to achieve appropriate registration,
recognition, or conditioning. Under these
circumstances, the absence of learning may be
erroneously attributed to lack of conscious
recognition when, in fact, it derives from deficient
stimulus registration. The most decisive

demonstration that awareness is a necessary
precondition for learning would require evidence
that, despite adequate stimulus registration,
classical conditioning does not occur without
awareness of the stimulus contingency.
Implications of Symbolic Control for

Behavioral Modification
The therapeutic potential of symbolic
processes has not been fully exploited although,
contrary to common belief, behavioral therapies

rely heavily upon effects which are cognitively
produced. This is particularly true of
desensitization treatments in which imagined

stimulus events are characteristically employed to
evoke emotional responses that ordinarily occur
to the actual stimuli. In some variants of this
approach emotion-neutralizing responses are
likewise induced symbolically. It is true that the
therapist exercises a certain degree of

instructional control over the timing and duration
of clients’ implicit activities, but no deconditioning
effects are likely to ensue unless clients engage in
appropriate cognitive representation of the
suggested sequence of events. As noted in an

earlier chapter, most desensitization methods
represent a form of cognitive counterconditioning
in which either the anxiety responses, the anxiety-
neutralizing responses, or both sets of events are
in large part symbolically controlled.
A similar higher-order conditioning process is
involved in the modification of addictive or

compulsive forms of behavior by means of aversive
cognitive counterconditioning (Cautela, 1966;
Miller, 1959, 1963). In the application of this
procedure, individuals typically visualize the
objects to which they are markedly attracted and
immediately revivify strong nauseous reactions.
Preliminary results based upon individual case

studies, which were reviewed in the preceding
chapter, indicate that conditioned aversions and
avoidance responses can be established in this
manner.
The material discussed earlier attests to the

fact that the most rapid and enduring changes in
instrumental behavior are achieved when
knowledge of contingencies is supplemented with
appropriate reinforcing consequences. In

interview approaches interpretations of probable
contingencies and suggestions for preferable
courses of action are offered repeatedly, but

favorable outcomes are rarely arranged. On the
other hand, practitioners utilizing reinforcement
procedures carefully plan the necessary
behavioral consequences, but often fail to specify
the basis for the reinforcement. It is apparent from
the influential role played by cognitive variables in
change processes that in an optimal treatment

program change agents should designate the
conditions of reinforcement in addition to
arranging the requisite response consequences.
There is a further potential application of
knowledge of symbolic control that is well worth

exploring. It has been amply demonstrated that
behavior is partly regulated by its immediate
consequences. Extensions of this principle to the
phenomenon of self-regulation (Bandura & Perloff,

1967; Ferster, Nurnberger, & Levitt, 1962) provide
evidence that people can exercise a certain degree
of control over their own behavior by arranging

favorable contingencies for themselves. Extending
this notion of self-management a step further, it is
entirely possible that individuals may be able to
control and alter their behavior by symbolically
produced consequences.
Many forms of behavior that eventually create
adverse social or physiological effects are strongly

maintained by their immediately reinforcing
effects. If the remote consequences could be
moved forward, or if other types of negative
outcomes were applied to the incipient precursors
of the behavior, its occurrence might be

significantly reduced. In most cases, this
rearrangement of consequences is difficult to
achieve by manipulating actual reinforcing events.
However, there is some suggestive evidence that

symbolized outcomes possess reinforcing
properties that are similar to their physical
equivalents. Weiner (1965) found that both

imagining aversive consequences and the actual
occurrence of the same negative events reduced
responding compared to a condition involving no
feedback, although imagined outcomes produced
somewhat weaker reductive effects. These
findings add credence to the view that overt
behavior can be partly regulated by covert self-

reinforcement operations.
Possible applications of this form of behavioral
control are discussed briefly by Homme (1965) in
a paper concerned with the conditioning of
implicit responses. In dealing with such problems

as excessive cigarette smoking, overeating, and
other activities which produce immediate and
automatic reinforcing effects, the individual
selects numerous aversive consequences of the

behavior which can be employed as covert
negative reinforcers. When he wishes to smoke,
for example, he immediately symbolizes the

aversive effects of smoking, or revivifies other
negative experiences. To the extent that
sufficiently strong aversive consequences of
smoking can be created by anti-smoking ideations,
smoking may be significantly reduced. Such
implicitly produced consequences, if effective at
all, are likely to exert greatest controlling power

when applied to weaker incipient forms of the
behavior than when the response tendency is
quite compelling, or after the behavior has already
been performed.
In the above examples cognitive activities are

employed in a contingent manner as reinforcing
events to reduce the incidence of overt behaviors.
Often certain trains of thought produce strong
emotional responses or disruptive effects upon

behavior, in which case the problem becomes one
of controlling the symbolic events themselves. In
its less extreme but more prevalent form, this type

of dysfunction tends to hamper persons’ efficiency
and productivity. As Dollard & Miller (1950) point
out, productive and creative work requires, among
other things, sustained attention to the task at
hand. Any interruptions from without, or intrusion
by irrelevant thoughts from within, temporarily
halt ongoing activities. Whereas the physical

environment can be arranged so as to minimize
external distractions, attainment of control of
one’s thoughts presents a much more challenging
problem.
The most direct and effective means of

reducing emotion-arousing intrusive thoughts is
to modify their eliciting conditions by the types of
procedures discussed in preceding chapters.
Nevertheless, ruminations about upsetting

experiences inevitably occur in everyday life and
hence persons must develop effective means of
ideational self-control which can be utilized

whenever the need arises. By far the most
prevalent, as well as the most futile, approach to
this problem is to advise others simply to banish
disturbing thoughts from their minds.
If a person is to achieve greater voluntary
control over his thought processes he must
manipulate variables which are capable of eliciting

and sustaining competing ideational activities.
This can be achieved in several different ways. The

simplest is by effecting changes in attentional
responses. That is, perturbing trains of thought
can be promptly turned off by directing one’s
attention to absorbing events which elicit

superseding cognitive activities. Indeed, this form
of self-control, in which thought-produced arousal
is diminished by engrossment in absorbing

literary material, televised programs, vocational
and avocational pursuits, and other engaging

projects, is widely practiced inadvertently.
Individuals could undoubtedly exercise greater
self-control over their thought processes and

attendant emotional responses through more
deliberate use of prepotent activities which are
kept in reserve for this very purpose.
Although the discussion thus far has

highlighted the possible tranquilizing effects of
attentional changes, they can serve a preventive
function as well. In many cases, a relatively weak

external stimulus may elicit a particular train of
thought which, through its associative
connections, activates further ideational contents

capable of generating strong emotional responses.
By interrupting this associative sequence in its
early stages, the occurrence of thought-produced
arousal may be forestalled altogether.
Assuming that symbolic activities obey the
same psychological laws as overt behavior, it
should be possible to influence significantly the
nature, incidence, and potency of covert events.
The difficulties in detecting the presence of
implicit responses present a major obstacle to
their control by reinforcement practices if one
adheres to the conventional paradigm in which an
external agent monitors the occurrence of the

desired behavior, imposes the contingencies, and
administers the reinforcers. However, as Homme
(1965) points out, the occurrence or absence of
covert events can be easily and reliably detected

by the person doing the thinking. Consequently,
such responses are most readily conditioned
through self-reinforcement operations. In this type
of approach implicit responses are self-monitored,
the contingencies are self-prescribed, and the

consequences self-produced.
Homme suggests that Premack’s (1965)

differential-probability hypothesis (i.e., any highly
preferred activity has reinforcing capabilities)

might be utilized to good advantage in the
contingency arrangement and selection of self-

reinforcers. That is, the strength and incidence of
certain classes of thoughts can be increased by
making preferred activities contingent upon their
occurrence. Presumably depressive, infuriating,

and other vexatious ruminations could be
displaced by self-reinforcement of more
constructive lines of thought. However, in view of
the dearth of information regarding the
conditionability of implicit events, development of

efficacious treatment methods must await
thoroughgoing analysis of thought control
processes as influenced by external or self-
monitored reinforcement operations, by
attentional changes, and by other self-manipulable

factors.
Discrepancy between Response Systems

and the Unconscious
The discussion thus far has emphasized
cognitive control of autonomic and instrumental
behavior under conditions where both the
environmental stimulus supports and the
reinforcing events are easily recognizable. Of
equal interest, particularly for theories of

personality, is the observation that persons
sometimes display discriminative autonomic or
motor responsiveness without conscious
recognition of eliciting stimuli that are too weak
for reliable identification. Traditional explanations

of such phenomena have tended to invoke a
potent psychic entity in the form of an
“unconscious mind” which supposedly possesses
sensitive discriminating capacities. According to
this interpretation, the unconscious mind readily

perceives threatening stimuli which occur below
the level of awareness, and the ego mobilizes

various defensive mechanisms to cope with them
and keep them out of awareness.
Considerable research has been conducted
over the years, principally by Eriksen (1958,
1960), to evaluate the empirical status of
subliminal discrimination and conditioning.
Results of these rigorously designed

investigations, along with other findings, lend
support to an alternative conception of the
characteristics of unconscious processes.
In the typical paradigm subjects are presented
a series of neutral and negatively valenced visual

stimuli at about threshold value, and observers’
nonverbal responses (autonomic or motor), as
well as verbal responses to the stimuli, are
recorded concurrently. A major controversy arose

in connection with evidence originally reported by
Lazarus & McLeary (1951) that subjects
frequently displayed conditioned autonomic

responses to aversive stimuli in the absence of
correct verbal recognition of the stimulus
sequence. The authors interpreted these data as
demonstration of unconscious discrimination.
Although the findings were not disputed, their
explanation as support for the existence of a
“subception” process was vigorously challenged

by Eriksen (1958) and Goldiamond (1958) on
methodological grounds.
In studies of this sort, the definition and
method of establishing thresholds provide one
major source of error. In most instances the

threshold is defined as the stimulus value at which
verbal recognition is about 50 percent correct.
Hence, a subject can, in fact, discriminate at better
than chance level a stimulus below the arbitrary

statistical threshold. Therefore, occurrence of
conditioned autonomic responses at slightly better
than chance expectancy does not represent a

compelling demonstration of unconscious
discrimination. Of much greater significance is
evidence (Eriksen, 1960) that subliminal
conditioning or discrimination rarely occurs when
the relevant stimuli are below recognition level.
A second methodological problem arises
because verbal reports can be influenced by

nonsensory factors. Subjects are generally
reluctant to admit the presence of a faint stimulus
when they are unsure, a negative response bias
that is likely to raise the verbal threshold
artifactually. A cautious person would obtain a

verbal threshold much higher than is, in fact, the
case, thus producing a large, spurious subliminal
effect. Feigned nonrecognition is most likely to
occur when subjects are presented socially

censured content such as taboo sexual words.
In addition to the influence of propriety and
subjective confidence upon verbal reports, verbal
responses do not convey the finer discriminations
made by subjects when they are required to
classify their perceptual experiences in terms of a
few discrete verbal categories. Pseudo-subliminal
effects may arise from the differential accuracy of
continuous autonomic responses and discrete

categorical responses at the verbal level. The fact,
however, that discrepancies in response systems
exist even when subjects are allowed a more
refined set of verbal responses, and motor

responses are substituted for an autonomic
measure (Eriksen, 1957), indicates that
discriminative behavior can occur without
accurate verbal labeling of the relevant stimulus
events.
As Eriksen points out, the question of whether
a more sensitive discriminatory mechanism exists

at an unconscious level can be answered most
directly by comparing the thresholds of different

response systems. This procedure involves
conditioning an autonomic response to a

supraliminal stimulus and then comparing the
incidence of accurate verbal recognition and
concurrent autonomic responsiveness to the
stimulus presented above and below threshold.

Using this method, along with a verbal response
measure that controlled for negative reporting
bias, Dulany & Eriksen (1959) found that
physiological and verbal response systems were
equally insensitive to low stimulus values, but

verbal response was the superior discriminator at
intermediate and high levels of stimulation.
Although the above findings provide no
evidence for unconscious responding to

threatening stimuli too weak to produce
awareness, nevertheless discrepancies between
simultaneous responses to environmental cues do

occur and hence require explanation. The latter
phenomenon, rather than being attributed to the
regulating influence of egos, superegos, or other
psychic agents operating within an unconscious
mind, can perhaps be more plausibly and
parsimoniously conceptualized in terms of
discrepancy between partially independent

response systems, a view for which Eriksen
provides considerable empirical support.
When two responses appear more or less
concomitantly with a series of sensory stimuli, the
nature of the relationship between these events

remains obscure. It may be that conscious
recognition of threatening stimuli evokes the
autonomic responses; the occurrence of
discriminative autonomic responses may produce

correct verbal recognition; or the two modes of
response may be elicited independently by the
environmental stimulus. In an effort to unravel the

relationships among variables in “subception”
studies, Eriksen conducted a series of experiments
in which verbal recognition responses and
concurrent motor or autonomic responses were
each correlated with the eliciting stimuli, with the
influence of the concomitant response variable
partialled out statistically. By this method of

analysis it is possible to determine what
relationship, if any, exists between autonomic
responses and sensory stimuli when differences in
awareness are eliminated. Results of these
experiments consistently show that verbal and

autonomic modes of behavior represent parallel
response systems that are both reactive to sensory
stimulation and are partially independent of each
other. The findings furthermore provide no
evidence that persons respond autonomically or

motorically in a more sensitive and accurate
manner than they do at the conscious verbal level.
Variables that affect symbolic, physiological,
and motor response systems differentially would
be expected to reduce the degree of correlation
between different classes of reactions. To the
extent that certain factors impair the accuracy of
measurement of one class of behavior without
affecting another, response disparities will obtain

even though the two systems are, in fact, highly
congruent. Extraneous variables are most likely to
exercise a high degree of control when persons are
responding to stimuli that are too weak to
override the effects of more salient characteristics

of the situation.
Genuine disparities among different classes of
behavior can be produced through the application
of differential reinforcement. Thus, if hostile

thoughts and verbalizations are approved or
permissively accepted but overt aggressive actions
are consistently punished, persons will readily

verbalize aggressive feelings without exhibiting
any of their motor equivalents. Similarly, by
reversing the reinforcement contingencies one
could effectively inhibit cognitive representations
of behavioral manifestations. It should likewise be
possible to produce other patterns of correlations
by varying systematically the contingencies of

reinforcement into which the responses enter.
Most of the material in the preceding sections
treats unconscious phenomena as internal and
environmental events which are not represented
in awareness. Some theories, however, consider it

important to distinguish between the
preconscious and the unconscious. In this
distinction the preconscious comprises mental
elements which are readily accessible to

consciousness by directing the subject’s attention
to them. On the other hand, the unconscious
includes elements which are relatively

inaccessible to awareness and which can be made
conscious only by removing strong resistance,
preferably through interpretive treatment.
In view of the questionable validity of
psychotherapists’ speculations about unconscious
contents, accessibility to consciousness, as

inferred from degree of resistance, is an
exceedingly unreliable criterion upon which to

build a structural theory of mental functioning. If
an individual refuses to acknowledge certain
thoughts or impulses suggested to him by his
therapist, the question remains whether the

resistance reveals repressed contents or justifiable
incredulity in the face of erroneous
interpretations. For example, it would be a

mistake to conclude that a person who vigorously
opposed interpretations that certain of his deviant

behaviors represented derivative manifestations
of a clandestine and powerful “zoognick,” did in
fact possess an unconscious “zoognick” held in

check by strong repressive forces. As noted earlier,
when the interpretive process is viewed from a
framework of social influence, the degree of
opposition shown by clients, and the amount and

type of elements which eventually emerge from
their hypothesized unconscious depend, in large
part, upon psychotherapists’ prestige, credibility,
belief systems, and other extraneous factors.
Although one might seriously dispute whether
the existence of unconscious psychic events can be

reliably established through psychotherapeutic
methods, there is no doubt that thoughts and
other implicit activities can be effectively
inhibited. Thought inhibition is traditionally
attributed to the mechanism of repression, which

is believed to operate largely at unconscious
levels. It is further assumed, particularly in
psychoanalytic formulations, that repressed
elements not only maintain a dynamic life of their
own, but are charged with a cathexis of drive
energy which continuously presses for discharge
either directly or in derivative form. Opposing
psychic agents which assume defensive and
prohibitive functions maintain the repression by

expending a portion of the psychic energy at their
disposal.
In behavior theory the phenomenon ascribed
to repression is conceptualized in terms of
processes of avoidance conditioning. If certain

thoughts are repeatedly associated with painful
experiences, they gradually become endowed with
aversive properties. Since thoughts are private
events, they can be affected only indirectly by

reinforcing operations. One way in which this
might come about is through the process of
response generalization. For example, the

experiment by Miller (1951), which was discussed
earlier, demonstrated that anxiety attached to a
spoken word tends to generalize to the thought of
the word. Also, since thoughts typically precede or
accompany overt responses, implicit events may
acquire the capacity to arouse anxiety through
their close temporal proximity to punished

instrumental acts.
Once certain thoughts come to function as
conditioned aversive stimuli, their occurrence
generates anxiety and their elimination allays it.
As shown in the experiment by Eriksen & Kuethe

(1956), inhibition of anxiety-provoking thoughts is
usually achieved through simultaneous
occurrence of competing ideational activities.
Students were asked to associate to a series of

words, and they were punished by electric shock
every time they responded with associations given
initially to five arbitrarily selected words.

Punished associations declined rapidly, whereas
nonpunished word associations remained
unchanged as items were repeated. Some
indications that punishment eliminated responses
at the implicit level rather than merely produced
deliberate word substitution are shown by the fact
that replacement of initially dominant associates

occurred in students who were unaware of the
basis for shocks or that their associations had
changed; the inhibitory effects also persisted on a
somewhat different association task in which the
threat of aversive stimulation was removed.
In social-learning interpretations of
repression, incompatible responses rather than
psychic agents are considered to be the inhibitory
forces. It should also be noted that the various

mechanisms of defense are defined, for the most
part, by the characteristics of the behavior that
competes with and supersedes negatively

valenced tendencies. For example, if a person who
is instigated to hostility exhibits positive thoughts
and actions toward the thwarting agent he is
engaging in reaction formation; if he should
become irritated at a blameless person he is
exhibiting displacement; if he avoids unpleasant
thoughts by becoming preoccupied with

competing ideational content and attendant rituals
he is resorting to obsessive-compulsive defenses.
Social-learning theory not only makes no
appeal to prohibitive psychic agents in accounting
for the inhibitory process, but it differs from

psychodynamic approaches also in the
assumptions made about the nature of inhibited
response tendencies. While formerly punished
responses may retain their capacity to generate

emotional effects if they are activated, it is not
assumed that they lead a dynamic existence within
an unconscious mind, that they possess a drive

energy which must be reduced periodically, that
they press continuously for discharge in one guise
or another, or that they require unceasing
restraint to confine them in the unconscious
domain. Rather, it is assumed that inhibited
behavior propensities remain inert unless
activated by appropriate stimulation. Under

circumstances where the incompatible controlling
responses to the same stimuli are clearly
dominant and therefore readily evoked, the
punished behavior is unlikely to reach even the
incipient level. On the other hand, when punished

responses are strongly established and competing
tendencies are not completely dominant, inhibited
responses may be aroused to the point where they
generate conflict and anxiety. A repressed
element, according to this view, does not have a

qualitatively different nature from any response
that has been superseded by an alternative

pattern of behavior.
Apart from its more secure empirical status,
the concurrent response system model has many
advantages over formulations which assume the
existence of an unconscious mind. The former
conceptualization does not lend itself to pseudo-
explanations in which a descriptive label for

response disparities is reified and assigned causal
properties—for example, discrepancy between
symbolic and motor or physiological responses to
the same stimulus events is given the descriptive
label of “unconscious,” which is then converted

into an internal agency that exercises powerful
control over behavior. The multiple-response-
systems interpretation of phenomena designated
as unconscious also encourages systematic
exploration of the variables which give rise to

response disparity. Finally, the theory has
important treatment implications. Given the

existence of independent but partially correlated
modes of response to significant stimuli, the
psychotherapist can profitably concentrate his

efforts on direct modification of the classes of
response that the treatment is designed to alter,
rather than embarking on a protracted search for
unconscious causative agents that one can predict

in advance will prove to be highly concordant with
the therapist’s particular theoretical predilections.
COGNITIVE EFFECTS OF SUBLIMINAL STIMULATION
The studies reviewed in the preceding section

investigated unconscious phenomena mainly in
terms of discrimination of weak stimuli. Klein and
his collaborators (Klein, Spence, Holt, &
Gourevitch, 1958; Spence, 1964; Spence &
Holland, 1962) have employed a somewhat

different approach to the problem. The effects of
subliminal stimulation are measured, not on
autonomic or recognition responses, but in terms
of indirect indicators of cognitive changes.
In most of these experiments neutral rather

than threatening stimuli are presented at levels
which apparently are sufficiently intense to
register and evoke some response in the central
nervous system, but the sensory input is either too
faint or too brief to produce full recognition or

correct identification of the stimulus events. In
other studies a verbal stimulus, which is presented

briefly well above identification level, is imbedded
in other competing stimuli and, therefore, may not
be recalled even though it has been fully
registered. Numerous indirect measures of

unconscious activation have been employed, such
as the inclusion of similar stimulus elements in
fantasy production (e.g., stories given to projective
tests, drawing, dreams); increased tendency to
give related responses to items in a word-

association test; cognitive elaboration of a neutral
stimulus in terms of the affective tone of preceding
subliminal cues; and arousal of responses
belonging to the same associative network.
Two conflicting hypotheses have been
proposed regarding the characteristics of
subliminal stimulation. According to the continuity

view of perceptual processes, the effects of
subliminal stimuli are similar to those of

recognized stimuli; since the magnitude of
influence varies with the intensity of input,
however, the former are less potent. The opposing
theory, on the other hand, assumes that the effects

of subliminal stimulation are more diffuse and
pronounced than those produced by supraliminal
stimuli because, in the former case, the restricting

effects of consciousness upon thought are
removed. This formulation, which was originally

advanced by Freud (1953), assumes that
preconscious and conscious influences obey
different laws, with stimuli which operate outside

of awareness being less bound by logical and
reality-oriented controls.
These alternative views cannot be evaluated
empirically because it has yet to be reliably
demonstrated that subliminal stimulation does in

fact influence cognitive activities to any
appreciable degree. Research of this type is

plagued by even more difficult methodological
problems than the approaches to unconscious
phenomena reviewed earlier. As was true of the
research on “perceptual defense,” the question of

whether one is dealing with a genuine
phenomenon or with a methodological artifact is
of continuing concern. Interpretations of findings

always remain in doubt because subjects who
display subliminal effects may be partially aware

of the stimulus patterns before reporting them so
that the stimulation is actually at or above
recognition threshold. Since subliminal stimuli

rarely appear in cognitive performances in direct
form, investigators must search for indirect,
distorted, or remote symbolic representations in
order to obtain at least a few scorable responses
for most subjects. To complicate matters further,
no objective criteria are available for identifying

distorted or transformed representations, and
high inter-scorer reliability is often difficult to

achieve.
Even if one includes the altered guises that

unrecognized sensory input presumably takes, the
effect of subliminal stimulation upon cognitive
functioning is exceedingly weak and, in some
cases, it is wiped out completely when sources of
artifact such as associative clustering effects are

controlled (Worell & Worell, 1966), and base rates
of indirect manifestations are obtained from
subjects who have never been exposed to the
subliminal stimuli ( Johnson & Eriksen, 1961). It
would appear from the elusive and scanty yield of
research in this area that subliminal activation
must play a relatively inconsequential role in
regulating human behavior. Whereas recognizable
stimuli assume a powerful behavior-directing

function, nonrecognized stimuli have, at best,
weak, inconsistent, and fragmentary psychological
effects.
Attitudinal Consequences of Behavioral and

Affective Changes
RELATIONSHIP BETWEEN ATTITUDINAL AND

BEHAVIORAL CHANGES
In most instances change agents who strive to
alter people’s attitudes are not interested in
attitudes per se. Although the influence
procedures are designed to change evaluations
and preferences concerning certain persons,
commodities, or activities, the principal aim is to
modify behavior. Thus, for example, the intent of

persuasive efforts is to get people to buy
particular brands of merchandise, to vote for
certain candidates, to engage in or to abandon
smoking, or to pursue various other courses of
action. The attitude-change approach is selected as

a means of influencing behavior when the desired
behavior cannot be directly elicited and reinforced
for practical or other reasons.
It is widely assumed that attitudes are

important determinants of overt actions and
consequently that any changes brought about in

the attitudinal domain will have widespread
effects upon subsequent behavior. It is further
believed that altered response patterns that are
accompanied by correspondingly altered attitudes
will be more stable over time than behavior that is

induced directly without cognitive supports. For
these and other reasons, the development of

beneficial attitudes is often regarded as a major
objective of social change endeavors. If it is

demonstrated that attitudinal changes do, in fact,
have significant behavioral consequences, it would
be of considerable import to devise efficacious
procedures that might be specifically directed

toward the modification of attitudes.
A voluminous literature has accumulated over
the years concerning attitude change through
exposure to persuasive communications, which,
because of their extensive mass application,

constitute the prevailing influence technique
explored in this field. Considering the multitude of
publications and the powerful controlling
functions assuredly conferred upon attitudes, it is

surprising to find that a thorough search of this
literature by Festinger (1964) yielded a dearth of
studies in which the influence of attitude change

upon behavior had been specifically investigated.
Nor has the yield of pertinent studies increased
noticeably in more recent years. The available
data, though admittedly meager, disclose that
changes in attitudes produced by persuasive
communications generally have little or no effect
upon overt actions (Fleishmann, Harris, & Burtt,

1955; Levitt, 1965; Maccoby, Romney, Adams, &
Maccoby, 1962). In contrast to these findings,
Greenwald (1965a) reports a positive, but low,
correlation between attitudinal and behavioral
changes. However, for subjects who express a

prior commitment counter to the influence
attempt, persuasive communications alter
attitudes but produce no significant change in
behavior (Greenwald, 1965b). These apparently
conflicting results may partly reflect differences in

the nature and importance of the behavior
undergoing change, the time elapsing between

assessment of changes in attitudes and behavior,
and the order in which these two sets of events are
measured. There is some reason to expect (Cohen,

1964) that changes in these two forms of response
will be more highly correlated when attitudes are
measured following, rather than prior to,
performance of discrepant behavior.
The failure to obtain consistent and high
relationships may be interpreted in several ways.

It is possible that responses to questionnaires and
self-ratings, which constitute the dependent
variables in most social- psychological research,

are unreliable indicants of people’s actual opinions
and attitudes. It has been shown (Schanck, 1932),
for example, that privately held attitudes often
differ markedly from those that are publicly
espoused. Persuasive communications may thus

elicit compliance to implied situational demands
without significantly affecting people’s private
attitudes.
A second explanation, suggested by Festinger
(1964), is that attitudinal changes resulting from
persuasive influences are relatively unstable and
will, therefore, disappear unless the
corresponding overt behavior is sustained by

adequate consequences. This view assumes that
attitudes produce temporary performance
changes; however, when environmental
contingencies do not support the new activities

individuals revert to their old behavior and the
newly established attitudes are similarly altered to
coincide with the actions.
The relative modifiability of attitudes and
actions, and the degree of correspondence
obtained between changes in these two sets of
events, may vary with the affective and social

consequences accompanying the behavior. A given
social influence might produce analogous changes

in both attitude and action when persons are
indifferent to, or favorably disposed toward,

performing the advocated activities. Most
attempts to control consumer behavior through
persuasive communications would fall in this
category. Thus, for example, a person who is

considering purchasing a television set may be
prompted to buy an expensive color console after
being convinced by persuasive commercials that it
provides more pleasurable viewing than a black-
and-white set. The process is more complicated,

however, when persons resist advocated behavior
that they can perform because it results in self-
devaluation. This is illustrated by situations in
which people are induced to behave in ways that
conflict with their beliefs. Obstacles to change also

arise when individuals are amenable to engaging
in desired activities but are unable to do so

because of strong fears and inhibitions. In the
latter instances, a weak method may alter
responses that are readily susceptible to change,

such as verbal evaluations, but fail to modify overt
behavior that is rendered intractable by its
adverse consequences. A relatively powerful
influence would be required to achieve correlative

changes in different types of response systems.
Unlike the inconsistent effects of persuasive

communications, desensitization and modeling
approaches are capable of producing correlated
changes in refractory behavior and attitudes.
These positive relationships range from r = .39

when changes in attitudes are measured by the
semantic differential technique to r = .59 and r =
.72 for change scores based on a variety of attitude

scales (Bandura, Blanchard, & Ritter, 1969;
Blanchard, 1969).
Correlated changes, when they do occur, may
reflect the operation of several different
mechanisms. According to most contemporary
attitude theories, some of which will be reviewed
later, there is a drive to maintain consistency
among beliefs, feelings, and actions. A change in
any one of the components will, therefore,
engender congruous modifications in the other

constituents. In these consistency models, changes
in attitudes or behavior are treated, not as
consequent events, but as causal variables that
affect other forms of behavior. An alternative
interpretation is that environmental influences

have similar but independent effects on feelings,
beliefs, and behavior. In this view, consistencies of
belief and behavior represent correlated coeffects
rather than outcomes of a process in which
modification of one type of behavior produces

changes in other forms of responding.
Definitive tests of the parallel effects and

consistency explanations of change processes are
precluded by the absence of a methodology that
would permit simultaneous measurement of
beliefs, affect, and actions. If incongruity creates
an internal stimulus for psychological change then
a sequential testing procedure confounds the
effects of external influences and the consistency

drive. Conversely, a given environmental influence
could have similar consequences on different
classes of response that would be erroneously
ascribed to the operation of a consistency drive.
These alternative formulations perhaps should be

regarded as complementary rather than
conflicting. Under most conditions, powerful
stimulus events produce diverse psychological
changes, and performance of new behavior is
likely to have additional cognitive and emotional

consequences.
In many respects, the question of whether

attitudes regulate overt behavior might be
considered a pseudo-issue created by arbitrary
distinctions between different types of response.
An attitude is variously defined as a disposition to
behave favorably or unfavorably toward a given
object (Brown, 1965); an organization of valenced
cognitions combined with a predilection for

motive arousal (Newcomb, Turner, & Converse,
1965); an affective evaluative response toward an
object (Rosenberg, 1960); or an implicit
anticipatory mediating response (Doob, 1947).
Like most implicit tendencies, attitudes are

characteristically inferred from various forms of
overt behavior rather than identified by some
independent criterion. Consequently, if self-ratings
were treated as a class of behavior rather than
assigned special status as indicants of an internal

mediator which is given substance and endowed
with influential regulatory powers, then the issue

of the relationship of attitude to behavior might be
more meaningfully conceptualized as a problem of
correlation between different response systems.

From this point of view, there exists no intrinsic
relationship between these two sets of responses,
since they can be either highly correlated or
dissociated depending upon their respective

contingencies of reinforcement.
The differentiation between attitudes and

overt actions disappears completely when the
former are primarily inferred from nonverbal
behavior, as is generally the case in

interpretations of naturalistic phenomena. For
example, a person who displays antagonistic
responses or actively avoids members of a given
ethnic group is believed to have a negative
attitude, whereas he is assumed to possess a

positive attitude if he exhibits approving amicable
reactions. In such circumstances, the issue of
whether attitudes influence behavior reduces to
the meaningless question of whether a particular
response pattern determines itself!
It should be emphasized that the foregoing
discussion is not intended to minimize the role of
symbolic mediators in human functioning, but

rather to call attention to the conceptual
difficulties created by a lack of independent

measures of covert tendencies designated as
attitudes. As a result, it is often difficult to
ascertain whether investigators are dealing with
superfluous abstractions from behavior, with

coeffects of operations of social influence, or with
causally related events.
Much of the ambiguity that prevails in this field
might be reduced and the theoretical issues more

precisely delineated if the indeterminate concept
of “attitude” were abandoned altogether. Rather,
the basic issue posed earlier might be rephrased
as follows: To what degree, and under what
conditions, do changes brought about in either
cognitive, affective, or motor classes of behavior
have reciprocal effects? It might be argued that
there would be some value in retaining the
attitude construct, apart from its shorthand

labeling function, if it were used to represent the
higher organization processes that are inferred
from specific behavioral manifestations. However,
in view of the substantial evidence (Bandura &
Walters, 1963; Mischel, 1968) that human

behavior is markedly specific and extensively
regulated by discriminative cues, reinforcement
contingencies, and other external events, there is
reason to question the utility of theoretical
formulations that invoke unitary mediators that

cannot possibly serve as determinants of
heterogeneous responses that are not

intercorrelated to any appreciable degree.
In an effort to account for behavioral
specificity within an attitude theory framework,
Rokeach (1966) has advanced the view that social
behavior is determined by sets of interacting
attitudes—one activated by the attitude objects,
and the other elicited by the situation. He

rightfully argues that in many instances situational
influences, which have been neglected in attitude
theory, may outweigh the response-eliciting
effects of the attitude object. Hence, individuals
often display quite different attitudes toward the

same attitude object in different social situations.
A theory that predicts attitudinal responses on the
basis of both subject and situational variables
would undoubtedly have greater predictive power
than one relying solely on subjects’ evaluations of

the attitude object in undefined contexts. The
controlling properties of situations primarily

reflect differences in reinforcement contingencies
as they apply to attitudinal behaviors expressed in
diverse social contexts. One might achieve even

greater predictive efficacy by treating attitudes as
evaluative responses that are under reinforcement
and discriminative stimulus control, as is any
other form of behavior.
Strategies of Attitude Change
BELIEF-ORIENTED APPROACH
Three general approaches can be employed

either singly or in various combinations to induce
attitudinal changes. The informational or belief-
oriented approach attempts to effect modifications
in people’s attitudes by altering their beliefs about
the attitude object through exposure to various

forms of persuasive communications. It is
assumed that people can be induced to change
their evaluations of an attitude object by
presenting them with new information about its
characteristics.
Most of the research generated by this
informational approach (Cohen, 1964; Hovland &
Janis, 1959; Hovland, Janis, & Kelley, 1953;
Rosenberg et al., 1960) has been expressly
designed to isolate the conditions under which a
given communication will have its maximal effect

upon recipients’ attitudes. Three general sets of
variables, namely the nature of the communicator,

the communication, and the recipients, have been
most extensively investigated. Studies of the
persons being influenced have generally been
concerned with their personality characteristics,

the level of their intelligence or sophistication, the
nature of their pre-existing attitudes, and the
strength of their commitment to a given position.
The effects of communicator variables in
enhancing attitudinal modification are typically

analyzed in terms of attributes such as expertness,
trustworthiness, prestige, impartiality, social
power, and concealment of the persuader’s
manipulative or propagandistic intent. The form
and organization of persuasive arguments, which
can also significantly influence attitude formation
and change, involves such matters as the optimal

order of presenting weak and major arguments,
the sequence of supporting and opposing

arguments, the degree of explicitness with which
conclusions are stated, the amount of repetition,
the degree of discrepancy between the subject’s
views and the ones advocated, the affective

properties of the contents, and whether the
influence program relies upon a one-sided
presentation or also includes some consideration

and refutation of counterarguments. Research
findings show that the effects of these different

variables rarely produce simple effects; rather
their direction and magnitude are dependent in
part upon the simultaneous influence of other

factors. For example, the amount of attitude
change may increase as a direct function of degree
of discrepancy of opinions advocated by a highly
respected source, whereas a less credible
persuader may exert a decreasing influence the

more divergent his opinions (Aronson, Turner, &
Carlsmith, 1963; Bergin, 1962). To complicate

matters further, a given variable may have
differential immediate and long-term effects upon
attitudes. With the passage of time, relevant
content may be retained but the source forgotten,

thus reducing initial credibility influences
(Hovland & Weiss, 1951; Hovland, Lumsdaine, &
Sheffield, 1949; Kelman & Hovland, 1953).
Most of the preceding investigations of

persuasive communications have been primarily
guided by a set of empirical principles rather than
a systematic theory. However, these principles are
organized around the basic assumption that
attitude change is governed to a large extent by
anticipations conveyed through communications
of rewarding and punishing consequences for
certain courses of action. A competent or
prestigious communicator is generally more

influential than a less competent one because the
former’s behavioral recommendations, if executed,
are more likely to result in favorable outcomes. As
noted below, the content of communications often
includes incentive references or is expressly

designed to alter the valence of the attitude object.
Although belief changes can be induced by
exposure to communication stimuli, there is little
evidence that mere presentation of information

about the attitude object alters people’s behavior
toward it to any great extent. Higher-order
conditioning processes are therefore frequently

employed to augment the potency of persuasive
communications. One method, which relies upon
the phenomenon of vicarious reinforcement,
increases the likelihood that an observer will
respond in the recommended manner by depicting
reinforcing consequences accruing to a
performing model. In positive appeals,

performance of the behavior suggested by the
communicator results in a host of rewarding
effects. Thus, smoking a certain brand of cigarettes
or using a particular hair lotion wins the loving
admiration of voluptuous belles, enhances job

performance, masculinizes one’s self-concept,
actualizes individualism and authenticity,
tranquilizes irritable nerves, invites social
recognition and amicable responsiveness from
total strangers, and arouses affectionate reactions

in spouses. Laboratory studies (Bandura, 1968)
disclose that, according to their nature, depicted

consequences to a performer not only facilitate or
inhibit response tendencies, but their effects may
outweigh the previously acquired value system of

the viewers (Bandura, Ross, & Ross, 1963).
Negative appeals, on the other hand, portray
the adverse consequences which result from
failure to comply with a communicator’s

behavioral recommendations. Although vicarious
punishment may inhibit existing response

dispositions to some extent, it is a less reliable
procedure for producing desired attitudes and
corresponding patterns of behavior. Display of
noxious or revolting outcomes tends to arouse

strong emotional responses which may give rise to
avoidance of disturbing material and associated
recommendations (Janis, 1967) or endow the

attitude object itself with negative valence.
Belief changes achieved by persuasive

arguments may temporarily increase the
likelihood of advocated courses of action, but it is
doubtful that this type of approach can by itself
produce enduring effects unless favorable
incentive conditions, which govern persistence of
induced changes, are arranged as well. That is, if a
person were to act on his beliefs, the effects of
adverse outcomes would eventually negate the

influence of persuasive communications. An
uncomely brunette, for example, who has been
persuaded that “blondes have more fun” may dye
her hair a flaxen tint, but if her dismal dating
plight should remain unchanged she is likely to

discard the belief and revert to her natural hue.
Some research indicates that susceptibility to
counterinfluence and rate of extinction of newly
established beliefs can be temporarily attenuated

by preparatory communications (Janis & Herz,
1949, cited in Hovland, Janis, & Kelley, 1953;
McGuire, 1964). These serve to stimulate

rehearsal of refutations of opposing arguments or
to instill expectations in recipients that although
they will at first encounter failure experiences and
other adverse outcomes, if they adhere to their
convictions they will eventually achieve success. A
person who has been led to anticipate unfavorable
initial outcomes may discount subsequent failures

for a time. However, where discrepancies exist
between assumed and actual schedules of
reinforcement, both his behavior and his beliefs
probably will gradually adjust to existing
reinforcement conditions.
Because of ease of application, persuasion

techniques presented through verbal or pictorial
devices are widely employed on a mass basis in
efforts to control consumer behavior, to influence

voting choices, and to indicate either positive or
negative evaluative responses toward particular
attitude objects and social issues. The efficacy of

mass persuasion methods is often diminished,
however, by the limited control that influence
agents can exercise over people’s attention to
communication stimuli, and by lack of direct
means of immediately reinforcing the audience
members for performing the recommended
behavior. On the other hand, under conditions

where selected communication stimuli are capable
of attracting and holding viewers’ attention and
the advocated actions do, in fact, result in
favorable consequences, mass appeals may initiate
lasting changes in people’s beliefs and behavior.
AFFECT-ORIENTED APPROACH
A second general strategy for inducing
attitudinal changes involves an affect-oriented

approach. In this paradigm, both evaluations of,
and behavior toward, particular attitude objects
are modified by altering their affective properties.
These emotional changes are typically achieved
through procedures based upon the principle of
classical conditioning. As shown in preceding
chapters, attitudinal and behavioral reversals can
be produced by contiguous association of objects
that are highly positive in valence with noxious
experiences in aversive forms of

counterconditioning, or by pairing subjectively
distressing stimuli with positively reinforcing
events in desensitization operations. The most
convincing demonstrations of transfer effects of

emotional reconditioning are furnished by studies
in which the affective properties of attitude
objects are independently measured, usually in
terms of appropriate physiological indices, with
adequate controls for nonspecific social influences

(Marks & Gelder, 1967).
Although the use of association principles to

facilitate attitudinal changes has been widespread,
there has been surprisingly little research into the

effectiveness of this approach. There is some
evidence that evaluative responses can be altered

by presenting persuasive messages or objects
contiguously with appetizing foods (Janis, Kaye, &
Kirschner, 1965; Razran, 1938), unpleasant odors
(Razran, 1940), or sexually arousing stimuli

(Smith, 1968). In an effort to determine whether
extraneous gratification facilitates attitudinal
change by a conditioning mechanism or by
creating a positive attitude toward the donor,
Dabbs & Janis (1965) compared the attitudinal

effects of food consumption occurring
contiguously or noncontiguously with exposure to
persuasive messages under two different
endorsement conditions. For half the subjects the
experimenter positively endorsed the messages

while for the remaining subjects he personally
disagreed with the conclusions advocated by the

communication. Neither the contiguity nor the
endorsement variable alone produced a significant
effect, but contiguous food combined with positive

endorsement increased acceptance of unpopular
opinions. These findings, however, must be
accepted with reservation because attitude
measures are of questionable validity when

obtained by the same person who positively or
negatively endorses the opinions being rated.
In naturalistic influence situations the method
that is most frequently employed to induce
changes in the affective value of an object involves

higher-order associations of symbolic stimuli. In
this procedure, the names and attributes of
attitude objects are paired with verbal stimuli or
pictorial presentations likely to evoke in listeners
strong emotional responses on the basis of prior

first-order conditioning. In several laboratory
investigations of this learning process (Insko &
Oakes, 1966; Staats & Staats, 1957), formerly
neutral nonsense syllables have been contiguously
associated with emotionally toned adjectives. The
syllables take on negative valence through
repetitive pairings with adjectives having negative
connotations (e.g., ugly, dirty), whereas these
same items are evaluated as pleasant after they

have been associated with positively conditioned
words such as beautiful, tasty, and happy. Pre-
existing attitudinal responses toward familiar
names of persons and nations have also been
significantly altered through conditioning

methods utilizing emotional words as the
evocative stimuli (Staats & Staats, 1958).
A study by Das & Nanda (1963) further reveals
that developed conditioned evaluative responses

tend to generalize along previously established
associative networks, thus resulting in widespread
effects. After nonsense syllables had been

contiguously associated with the names of two
aboriginal tribes, favorable and unfavorable
attitudes were developed toward the syllables. In
a subsequent test subjects ascribed positive and
negative attributes to the tribes in accordance
with the evaluative responses conditional to their
corresponding nonsense syllables.
It should be noted in this context that, unlike
laboratory analogues of attitudinal learning in
which single emotional words are presented in
discrete trials, in real-life situations considerably
more intense emotional reactions are typically

elicited in audiences by the cumulative impact of
long series of emotionally toned descriptions or
pictorial presentations.
The above studies, though relevant to the issue

of attitudinal modification through affective
manipulations, would have greater implications
had they included more extensive assessment of
emotional changes. Of much greater import would
be evidence that exposure to communication
stimuli does, in fact, endow attitude objects with
emotion-arousing properties, and that alterations
in the affective domain are associated with
corresponding changes in individuals’ overt

behavior toward the objects in question.
Another method of inducing affective changes
that have considerable behavioral consequences

relies upon modeling processes (Bandura, 1968).
This outcome is achieved by associating attitude
objects or their descriptions with affective

modeling cues capable of arousing in viewers
analogous emotional responses. Attitudinal
modification through modeling is illustrated in an
ingenious experiment by Duncker (1938). In an
initial test of food preferences children chose

powdered chocolate with a pleasant lemon flavor
over a very sweet sugar with a disagreeable

medicinal taste. Later, a story was read to the
children in which a stalwart astute hero abhorred

a sour-tasting foodstuff similar to the children’s
preferred food and enthusiastically relished a
sweet-tasting substance. The reactions of the
admired hero reversed the children’s initial food

preference, as measured immediately after the
story session and in six successive tests in which
the children chose between powdered chocolate
and medicated sugar. Moreover, brief recall of the
story reinstated the experimentally induced

preferences that had declined gradually over time.
More recently, Carlin (1965) found that young
children showed a greater preference for deferred
gratification after they saw an adult model display
positive affective reactions while waiting for

delayed rewards than they did after they observed
the model express negative emotional reactions

and devalue the goal object during the imposed
delay period.
In the foregoing studies both evaluative
judgments and emotional responses were
modeled. The observed changes therefore cannot
be attributed solely to the influence of affective
modeling cues. There is some reason to believe

from evidence provided by Culbertson (1957) that
the modeling of preferences and beliefs without
strong affective displays can alter attitudes.
Observers who witnessed others express
favorable attitudes toward integration

subsequently exhibited a decrease in prejudicial
attitudes.
The potency of modeling for inducing
attitudinal changes is further demonstrated in the

experiment by Bandura, Blanchard, & Ritter
(1969) that was previously described. Snake-
phobic subjects were administered eight

evaluative dimensions of the semantic differential
technique, and six attitude scales on which they
rated how much they would like or loathe
different types of encounters with reptiles.
Subjects were then given factual information
about the characteristics and habits of snakes in
order to control and to assess the possible

influence of incidental information before any
treatment procedures were applied. After the test
for snake avoidance behavior, the attitude
measures were again administered. In the next
phase of the experiment subjects received either

systematic desensitization, symbolic modeling,
live modeling combined with guided participation,
or no treatment. Following completion of the
treatment series the attitude measures were again
administered prior to, and immediately after, the

snake avoidance test.
The results are summarized graphically in

Figure 9-6. Subjects’ loathing of reptiles was not
altered in the slightest by factual information and
exposure to the test snake. The refractory quality
of these negative attitudes is further shown by the
1926
Figure 9-6. Attitudinal changes for subjects who received either one of the three treatment procedures or served
as untreated controls. The numeral 1 indicates subjects’ attitudes prior to the behavioral test,
and the numeral 2 shows their attitudes immediately after the test of avoidance behavior. Bandura,
Blanchard, and Ritter, 1969.
control subjects, whose evaluative reactions
remained unchanged across repeated
assessments. Both symbolic modeling and

desensitization, which successfully extinguished
negative emotional responses to snake stimuli,
produced extensive attitudinal changes. The
treatment condition that neutralized the anxiety-
arousing properties of snakes and enabled

subjects to interact with the repugnant attitude
object without any adverse consequences
achieved the greatest modification in attitudinal
behavior. In a study designed to assess the relative
influence of information, modeling, and guided

contact in the latter method, Blanchard (1969)
found that modeling accounted for approximately

80 percent of the attitude change. Information, on
the other hand, increased subjects’ emotional
arousal to modeling displays and had, if anything,

a slightly adverse effect.
The attitudinal consequences of affective
change are also disclosed in desensitization
studies involving more general attitudes dealing

with sex, aggression, and other interpersonal
contents. These findings indicate that the cognitive

evaluative component of attitudes can be
substantially modified through direct
manipulation of the affective properties of the
attitude object without involving informational

references of a favorable or unfavorable sort.
Essentially similar results are reported by
Rosenberg (1960), who has shown that a negative

affect induced through post-hypnotic suggestions
produces a corresponding change in beliefs about

the attitude object.
BEHAVIOR-ORIENTED APPROACH
The third approach to the modification of

attitudes, which is frequently employed in
experimental social psychology (Brehm & Cohen,
1962; Festinger, 1957), relies upon a behavior-
oriented strategy. Change programs conducted

within a social-learning framework likewise favor
this type of approach, although they receive little
mention in discussions of attitude theory because
until recently the cognitive consequences that

undoubtedly accompany behavioral modifications
have rarely been systematically assessed. Before
specific experimental findings bearing on
behavioral approaches are discussed, the
conceptual scheme underlying most of this

research will be presented briefly.
Investigations of the process of attitude change

have, in large part, been guided by various models
of cognitive consistency. Among the more

prominent theoretical positions are those of
congruity (Osgood & Tannenbaum, 1955), balance

(Abelson & Rosenberg, 1958; Heider, 1958), and
cognitive dissonance (Festinger, 1957). Although
these formulations differ somewhat in the types of
events that are interrelated and the methods used

to disrupt internal equilibrium, they have in
common the view that a person’s cognitions about
himself and his environment are organized into an
internally consistent system. It is further assumed,
albeit implicitly, that there exists a strong drive for

self-consistency. Consequently, the introduction of
new information that contradicts existing
attitudes or beliefs creates an aversive
motivational state that instigates the individual to
eliminate or reduce it by making cognitive

adjustment designed to achieve a new mental
equilibrium. These consistency doctrines thus

assume that disruption of internal congruity
between cognitive elements constitutes a basic
determinant of attitude change.
In laboratory investigations the requisite
cognitive disequilibrium is usually created
through exposure to persuasive communications
which counter subjects’ initial attitudes. Research

stimulated by the cognitive dissonance model is
especially relevant to the issue under discussion
because, unlike the other consistency enterprises,
the method most often employed to induce
attitudinal changes involves getting a person to

engage in attitude-discrepant behavior under
conditions of minimal external inducement.
There are several reasons for selecting change
in behavior as a primary mode of attitude change.

First, it is much easier to arrange reinforcement
contingencies for altering specific overt actions
than for changing personal convictions, which

have a more private character and are often more
difficult to define. By skillful management of
incentives a person can be induced to take
progressively more favorable actions toward
attitude objects. Second, diverse opinions usually
exist about possible effects of engaging in certain
forms of behavior. Consequently, such beliefs are

more readily modifiable than cognitive
representations of the behavior itself, which,
because of its objective status, is more firmly fixed.
Thus, for example, it is easier to alter one’s
opinions about the effects of smoking than to deny

that one is, in fact, smoking, or to discontinue
smoking altogether. Third, in many cases behavior
is so powerfully maintained by its immediate
consequences that any induced cognitive
modification is likely to exert, at most, weak and

transitory influence upon corresponding actions. A
psychotherapist, for example, who contracted to

cure chronic alcoholism or debilitating compulsive
rituals by exposing his clients to discrepant
information about the physiological hazards of

excessive drinking or the irrationality of needless,
arduous compulsions would, in a short time, suffer
insolvency. Obviously, in instances where
behavior is highly resistant to change,

modification of response consequences is essential
for effecting enduring alterations in performance

to which attitudes would eventually be expected
to adapt.
In the prototypic dissonance experiment,

subjects’ attitudes toward a particular issue or
object are assessed through self-ratings, after
which they are prompted, in one way or another,
to engage in behavior which contradicts their
private views. The same rating scales are later

readministered, and the change scores are taken
to represent the degree of attitude alteration.
These studies (Brehm & Cohen, 1962; Cohen,
1964; Festinger, 1957) demonstrate that induced
behavioral changes typically produce a
corresponding modification in subjects’ attitudes.
After it was demonstrated that behavioral change
has attitudinal consequences, subsequent research
was primarily concerned with identifying the

variables governing the amount of attitudinal
change effected by performance of discrepant
behavior. The conditions selected for investigation
are based upon the general assumption that the
less compelling the reasons for engaging in the

contradictory behavior, the greater the dissonance
and hence, the more attitude change is required to
reduce it. Thus, persons who engage in attitude-
discrepant behavior because of large rewards or
strong coercive pressures have ample external

justification for their actions and presumably,
therefore, experience little dissonance and change

of attitude. On the other hand, it is assumed that
those who behave contrary to their private
opinions under conditions of minimal external

inducement are obliged to discover new
attractions in the disagreeable activity to justify to
themselves their voluntary performance of
inconsistent actions.
According to dissonance theory, inconsistent
action will produce the greatest amount of attitude

change under conditions where small incentives,
just sufficient to get the person to comply, are
employed; there are minimal threats or coercive

inducements; few reasons are given for taking the
discrepant stand; the person receives a high
degree of choice in committing himself to the
counterattitudinal performance; there is high
expenditure of effort in the attainment of the goal

object or in the enactment of the discrepant
behavior; the inducing agent is viewed
unfavorably; and the person being influenced
displays high self-esteem. It should be noted here
that in naturalistic situations it is ordinarily no
easy task to get people to perform personally
repugnant actions for any length of time under
such unfavorable incentive conditions.
Evaluation of the major theoretical issues and

voluminous empirical findings bearing on
dissonance-arousing variables goes beyond the
scope of this book. For the interested reader,
detailed reviews are available elsewhere (Abelson,
Aronson, McGuire, Newcomb, Rosenberg &

Tannenbaum, 1968; Chapanis & Chapanis, 1964;
Elms, 1967; Feldman, 1966). The empirical studies
have generally yielded conflicting results;
consequently, the precise conditions under which

induced discrepant performance will have
greatest effect on attitudes still remain somewhat
obscure. A major difficulty in verifying derivations

from dissonance theory and in drawing
conclusions from experimental data arises
because there exists no independent measure of
the degree to which the postulated state of
dissonance has been aroused by a given
procedure. Since the induction operations
typically involve a complex set of events,

experimental outcomes are open to numerous
alternative explanations, which complicate
interpretation. For a graphic illustration of the
ambiguity concerning the independent variables
in forced compliance studies, the reader is

referred to the spirited debate between Aronson
(1966) and Rosenberg (1966) who interpret the
same experimental manipulation as having
created opposite amounts of cognitive dissonance!
One theoretical issue, because of its obvious

relevance to the role of incentives in change
processes, warrants discussion in this context. It is

widely believed that experimental findings
concerning the effects of incentives upon attitude
change brought about by divergent behavior
contradict derivations from “conventional
reinforcement theory.” In fact, because of
inadequate application of incentives, results of
those studies are of limited relevance to

reinforcement principles. Moreover, as will be
discussed later, contrary to common belief both
dissonance and reinforcement theory offer the
same nonobvious implications.
In these experiments subjects are induced to

write essays, enact prescribed roles, or otherwise
publicly espouse a set of opinions that contradict
their private feelings and beliefs. Some subjects
are offered small monetary incentives (15¢, 50¢)

for assuming the discrepant position, while others
are promised more generous rewards ($5, $20).
Several experiments (Brehm & Cohen, 1962;

Festinger & Carlsmith, 1959) report an inverse
relationship between size of monetary incentive
and attitude change; other investigations have
yielded both positive and inverse relationships
(Carlsmith, Collins, & Helmreich, 1966; Linder,
Cooper, & Jones, 1967), no incentive effects of
statistically significant magnitude (Elms & Janis,

1965; Janis & Gilmore, 1965; Nuttin, 1966), or
evidence that higher monetary incentives produce
greater degrees of attitude change (Collins, 1969;
Rosenberg, 1965).
The conditions governing the relationship

between incentives and attitude change cannot be
reliably identified unless data are presented for
two other critical relationships, namely, the
amount of counterattitudinal behavior engaged in

as a function of different magnitudes of reward,
and the degree to which variations in amount of
discrepant behavior are associated with extent of

attitude change. In experiments where the amount
and quality of counterattitudinal performance are
measured, outcomes are often uninterpretable
because the material rewards, which supposedly
serve as external inducements for discrepant
behavior, are applied in such a loose contingency
that their incentive function is virtually

obliterated. Rewards are offered for any
performance subjects choose to display, but
otherwise the incentives are not explicitly made
contingent upon the number, persuasiveness, and
elaborated quality of arguments.
Even experiments conducted by proponents of

incentive principles have limited bearing on
incentive theory because rewards are offered
without specific performance requirements. If

incentives facilitate attitude change because they
motivate individuals to generate positive
arguments counter to their own beliefs (Janis,

1968), then subjects should be rewarded on the
basis of the number of favorable arguments that
they produce. An adequate test of the predictive
efficacy of incentive theory would also require
independent evidence that variations in reward
actually produce a differential number and variety
of arguments. When rewards are provided without

regard to response output there is no reason to
expect them to have any consistent behavioral or
attitudinal effects. This is borne out by the actual
findings. A number of limiting conditions under
which the consistency theories hold have been

proposed, including freedom of choice,
commitment, public or private performance,
anticipated consequences of influencing others in
the counterattitudinal direction, and self-
devaluative consequences. However, none of these

explanations adequately reconciles all the
divergent results.
The positive influence of incentives is also
frequently nullified in dissonance experiments by
introducing monetary rewards in the context of
inordinate social pressures upon subjects to
perform the disagreeable task. In the procedure
most often employed, a flustered experimenter
explains to an unwitting subject that an

unexpected emergency has arisen because the
regular assistant has just phoned saying that he
will be unable to conduct the study with the next
subject, who has already arrived and is waiting for
his scheduled session. Would the subject be

willing, for a small or a large fee, to substitute for
the absent assistant by informing the waiting
subject that a boring task is interesting and
enjoyable? It is hardly surprising that, given such
compelling reasons, the same amount of

discrepant behavior is enacted regardless of
incentive size (Carlsmith, Collins, & Helmreich,

1966; Festinger & Carlsmith, 1959), and subjects
are willing to perform the disagreeable task even
without any monetary rewards at all (Nuttin,

1966). Indeed, given this “sudden, unexpected,
and pressing” crisis, and urgent appeals to the
subject to help the experimenter “out of a jam,”
none of the experimental conditions, regardless of

the appended fee, can be considered as providing
insufficient justification for compliance. By

contrast, when the inordinate social pressures are
absent and the monetary rewards serve as the
main justification for developing
counterattitudinal arguments, as in essay-writing

situations, increased incentives often produce
increasing amounts of attitude change (Carlsmith,
Collins, & Helmreich, 1966). Other investigators

(Elms, 1967; Janis & Gilmore, 1965; Rosenberg,
1966) have therefore attributed the effects of

differential payments to arousal of resentment,
suspicion, and other interfering emotional
responses rather than to their intended positive

incentive value.
Although there is abundant evidence that
performance of counterattitudinal behavior can
be a highly efficacious means of altering existing
attitudes, divergent findings regarding

contributory conditions suggest that more than
one mediating process is probably involved. Some
of these alternative mechanisms are discussed
below.
Individuals undergo considerable social

training to be logical and consistent in their
beliefs. To the extent that contradictory beliefs
engender critical reactions from others and other
negative consequences, inconsistency may become

an aversive condition that instigates emotional
arousal and cognitive modifications designed to
remove the source of discomfort. Hence,

dissonance processes may be involved to some
extent under conditions where people have
voluntarily committed themselves to perform
disagreeable behavior with weak external
inducement and are therefore compelled to
modify their beliefs to justify their contradictory
actions. Because of the many limiting conditions

under which dissonance effects are believed to
occur, the phenomenon could not be highly
prevalent. Dissonance reduction must, therefore,
be only one of several processes activated by
counterattitudinal performance.
Whenever a given action has been rewarded,

reinforcement effects tend to generalize across
similar classes of behavior, with the result that the
incidence of corresponding verbal responses is

likewise increased to some degree (Lovaas, 1961).
Cognitive equivalents of the reinforced overt
behavior are also affected in a similar manner

(Miller, 1951) even though they have never been
directly involved in the reinforcement
contingency. Thus, in situations where counter-
attitudinal behavior is contingently rewarded,
analogous changes in the cognitive domain may
partly reflect a response generalization process.
A third interpretation of role-enactment

effects, advanced by Janis & King (1954), Janis &
Gilmore (1965), and Rosenberg (1965),
emphasizes the self-persuasive consequences of
recalling and developing numerous positive
arguments. According to this point of view,

favorable incentive conditions are likely to
produce a greater amount of improvisation and
more persuasively elaborated arguments
upholding the opposed point of view. In the course

of role-playing the person becomes influenced by
the merits of his own convincing arguments.
Although there is some evidence that degree of
attitude change is positively related to amount and
quality of counterattitudinal behavior, incentive
size alone has no consistent effects upon either
improvisation or attitude change (Janis & King,
1954; Janis & Gilmore, 1965; Kelman, 1953;
Rosenberg, 1965). However, Janis has shown that

large incentives furnished by a favorable source
produce better quality of performance and greater
modification in attitudes than do small monetary
rewards offered by an unfavorable sponsor for

taking a contradictory stand. Rosenberg (1966)
also offers the interesting proposition that self-
persuasive consequences of behavioral rehearsal
may depend upon the performer’s psychological
set. As in the case of influences from external

sources, a person who labels his
counterattitudinal advocacy as manipulative and

deceptive may be considerably more resistant to
his own persuasive arguments than if he

undertakes the task with a positive self-searching
orientation. This factor, if operative, might account

for some of the conflicting results. Bern (1967)
similarly argues that the self-persuasive effects of
observing one’s own behavior may be partly
determined by the stimulus conditions under

which it occurs.
To the extent that attitudinal changes are

partly governed by the amount of discrepant
behavior engaged in, the selection of incentive
magnitude as the critical variable for testing

predictions from dissonance and reinforcement
theory was an unfortunate choice, because
variations in amount of reward have no consistent
effects upon performance by human subjects
(Bruning, 1964; Elliott, 1966; Lewis & Duncan,

1961). This is analogous to manipulating a
variable that has no uniform effect upon the
amount of dissonance arousal. In order to furnish
a critical test of reinforcement theory, it is
necessary to vary an incentive property that has
reliable behavioral consequences, since the only
reason for employing rewards is to alter the
incidence of the crucial behavior. In view of
evidence that variable, intermittent reinforcement

results in higher performance than the same
rewards administered on a fixed schedule, a more
appropriate incentive variable, from the
standpoint of reinforcement theory, would be the
pattern in which counterattitudinal behavior is

rewarded. For most incentive characteristics, the
supposedly rival theories predict the same
outcome, though for different reasons. Consider,
for example, situations in which counterattitudinal
behavior is generously reinforced on a fixed-

interval schedule in one case, and much less
frequently on a variable-ratio schedule in a

contrasting treatment. The less favorable incentive
condition would be expected to produce more
attitudinal change because, according to

dissonance theory, it provides less justification
and hence greater dissonance, and according to
reinforcement theory, because it generates more
self-persuasive behavior.
It should also be noted in passing that, as far as
behavior change programs are concerned, there

are no reinforcement theories that prescribe the
use of excessive rewards. On the contrary, as
explained in Chapter 4, it is most advantageous for

several reasons to employ incentive conditions
just sufficient to elicit the desired behavior. First,
the aim is to produce enduring alterations in
behavior, and partially reinforced behavior is most
resistant to extinction. Second, in a well-designed

program artificial, external inducements, initially
required to elicit the desired responsiveness, are
gradually reduced as the behavior produces
natural and self-evaluative reinforcing
consequences. Since reductions in incentives
generate disruptive emotional effects, a change
agent would be inviting unnecessary trouble by
the use of needlessly large rewards.
The discussion thus far has focused on internal

readjustments prompted by inconsistent action
and alternative mediating processes that might
account for the phenomenon. A fourth mechanism
mediating role-enactment effects—an
experimental consequences process—highlights

the fact that a change in behavior provides a
person with a variety of new experiences with the
attitude object. Information gained from these
new social interactions and observations can, in

itself, produce substantial reorganization of
attitudes (Kelman, 1961). Thus, for example, a
prejudiced person who has been induced to

behave positively toward members of a minority
group may adopt a more favorable attitude not so
much because of stress created by intrapsychic
inconsistency, but because close positive
associations with minority groups furnish
additional evaluative knowledge and rewarding
outcomes for the participants. Direct experimental

consequences of behavioral change, depending
upon their nature, may far outweigh the influence
of intraspychic tensions in initiating and
maintaining attitudinal changes.
Another important aspect of this process

concerns the effects of induced behavioral
modifications upon the social environment of a
monitoring membership group. If a person
behaves in a discrepant manner he may be

virtually forced into association with the outgroup
through ostracism. Under these circumstances
“consistency” may be enforced and maintained

through external, social mediation rather than
intrapsychic compromises. Experiential
consequences of behavioral change are likely to
play a major role in determining how long induced
attitudinal changes will endure.
MODIFICATION OF SELF-ATTITUDES
Of particular relevance to social approaches for
developing and modifying attitudes is the evidence
provided by Breer & Locke (1965) that task
experiences can exert strong influence upon

performers’ attitudes. In these studies individuals
are either differentially rewarded or experience

differential success for performing tasks in two
different ways. After the performance experiences,
subjects’ preferences for similar activities and

more abstract values only indirectly related to the
tasks themselves are measured. The overall
results, based upon numerous investigations of
attitudes toward individualism, equalitarianism,
theism, and achievement, show that significant
attitudinal changes can be induced by providing
individuals with successful task experiences. For
example, college students who worked better in
groups than alone became more collectivistic in
their attitudes, whereas subjects who experienced

greater success when performing tasks
independently adopted a more individualistic
orientation. These studies also provide some
evidence that attitudes induced by success tend to

generalize to related types of activities and to
abstract preferences.
Change agents are often concerned not only
with altering individual’s evaluations of different
forms of behavior but in modifying their self-
attitudes as well. Indeed, in some schools of
psychotherapy, such as the client-centered

approach (Rogers, 1959), self-concept changes are
routinely selected as one of the primary treatment

objectives. According to this point of view, self-
attitudes can be modified most effectively through

intrapsychic exploration under conditions where
the change agent displays empathy, noncontingent
positive regard, and genuineness. The individual’s
difficulties presumably stem from the fact that

experiences that are incompatible with his faulty
self-conception are consistently denied or
inadequately symbolized. Self-examination in a
positive, non- evaluative relationship will lead him
to attend to warded-off experiences and accept

them as part of himself; this, in turn, produces
increased feelings of self-worth, self-acceptance,
and greater freedom of action. This approach is
predicated on the basic assumption that the
person already has developed highly competent

repertoires of behavior, most of which are
inherently satisfying, but which are neither

accepted nor actualized because of the faulty self-
evaluative contingencies that he has adopted from
misguided socialization agents.
Undoubtedly many competent people do
experience a great deal of self-generated distress
and many self-imposed constraints as a result of
adherence to ill-advised or excessively high

standards of self-reinforcement. To the extent that
a change agent differentially reinforces realistic
standard-setting behavior and elicits emulation of
more lenient self- evaluative standards as
conveyed through his comments and actions, the

client’s habitual self-attitudes are likely to undergo
change. However, results of outcome studies
presented earlier indicate that this objective may
not be too readily achieved on the basis of the
types of conditions prescribed by the client-

centered approach.
In many cases, of course, unfavorable self-

attitudes stem from behavioral deficits and are
repeatedly reinforced through failure experiences
occasioned by the person’s inability to meet
realistic cultural expectations. It is obvious that for
such persons no amount of self-exploration will
yield esteem-producing vocational skills, academic
capabilities, interpersonal competencies, and

rewarding avocational proficiencies which would
support realistic positive self-evaluations. Here
the primary concern must be with self-
development rather than self-exploration.
Evidence that attitudes are significantly influenced

by rewarding performance feedback indicates that
enduring positive self-evaluations can be most
effectively achieved by arranging optimal
conditions for the individual to acquire the
requisite competencies. On the other hand, the

likelihood is exceedingly small that favorable self-
attitudes, however induced, could survive in the

face of discontinuing performance experiences.
“Internalization” and Persistence of

Behavioral Changes
It is generally assumed that when a change in
behavior is accompanied by a set of congruent
attitudes the behavior has become effectively
internalized. After this state of integration is

achieved, a person’s conduct is presumably guided
by inner values rather than by compliance with
external demands and outcomes. As a result of
reciprocal support of attitudes and behavior,
internalized response tendencies are presumed to

be more stable and enduring over time, even
under relatively unfavorable conditions of
external reinforcement, than compliant behavior
without personal conviction. This view, if true,
would appear on casual inspection to dispute the

principle that behavior is regulated by its
consequences. This apparent contradiction arises

because the latter proposition has often been
interpreted, by both its ardent proponents and its
critics, to mean that behavior is governed by

situational contingencies. In fact, as will be
explained later, even so-called internalized
behavior remains under reinforcement feedback
control, although it may be relatively independent

of externally occurring consequences.
Before discussing alternative mechanisms that

have been advanced to account for phenomena
subsumed under the term internalization, it is
appropriate to question what, if anything, is

internalized in the organism. It is perhaps
misleading to talk of behavior being internalized
since, after response patterns have been acquired,
it is doubtful that they can undergo any further
interiorization. The major issues, therefore, are

less concerned with the locus of behavior than with
the nature of its controlling conditions.
In evaluating theories of internalization and
self-regulating processes, it is important to
distinguish between the reinforcing and the
discriminative functions of stimuli. Behavior is
controlled not only by its reinforcing
consequences, but also by environmental stimuli

which signify the types of outcomes that are likely
to follow certain courses of action. A great deal of
human behavior that appears to be internally
directed is, in fact, under the control of such

discriminative cues. Sometimes the controlling
stimulus events can be readily identified because
of their distinctive properties, as in the case of a
motorist who waits patiently at a red signal light
on a deserted street without an automobile,

pedestrian, or traffic officer in sight. While this
motorist is exhibiting remarkable control,

nevertheless his behavior is clearly externally
regulated. This example, incidentally, illustrates

some of the problems inherent in definitions of
internalization which ignore the behavior-

directing function of cues, and in which the
primary criterion is the occurrence of behavioral
control in the absence of social surveillance.
In most instances the controlling external
stimuli are not as easily identified, and
consequently internal controlling agents tend to

be invoked as explanatory factors. The case of a
dog who refrains from lounging on an inviting sofa
though unobserved, which is sometimes cited as

an example of internalized control, might serve to
illustrate the latter point. Some time ago we
trained our spaniel through differential
reinforcement to keep clear of all lounge chairs
except one old recliner that soon became the dog’s

semi-permanent abode. Our conditionable spaniel
exhibited a finely developed “superego” until one
day when my wife rearranged the furniture. Upon
entering the living room I was greeted by the
tranquil scene of our socialized canine snoozing
contentedly on a new chair located in the area
previously occupied by the threadbare recliner. It
suddenly became apparent that our dog’s behavior
was regulated by irrelevant spatial cues rather

than by an internal governor.
It has been repeatedly demonstrated in
research with infrahuman organisms and with
human subjects that cues regularly correlated
with reinforcement eventually gain control over

the associated behavior. Hence, in situations
where reliable discriminative stimuli are present
it is reasonable to question what has been
internalized, and why it is necessary to invoke an

internal agency which supposedly regulates the
observed behavior. The fact that other types of
subtle stimuli such as temporal cues (Ferster &

Skinner, 1957; Sidman, 1966) can exert
considerable discriminative control over the form,
rate, and pattern of responding makes it
exceedingly difficult to rule out external stimulus
determinants.
Although there are extensive stimulus
supports for behavior, one must nevertheless

explain differences in the responses of individuals
to what appear to be essentially the same cues. To
return to the example of our lonesome motorist, a
systematic field study would undoubtedly disclose
that under deserted traffic conditions some

motorists would wait dutifully for the green light,
others might stop momentarily and then continue
on their way, while still others are likely to
disregard the traffic signal completely. A

noninternalist would most likely argue that the
signal light, by itself, does not adequately define
the total controlling environment. Whether or not

a particular individual will transgress depends
upon a large number of other stimulus variables
(e.g., restraining influence of other passengers,
time pressures on the motorist, his subjective
estimates of the probability of being caught and
the gravity of the consequences that might follow,
etc.) each of which may exercise some degree of

control over his behavior. It is conceivable that in
many cases consideration of all the relevant
stimulus events operative in a given situation
would disclose that behavior appearing to be
internally governed is, in large part, under the

control of complex patterns of multiple stimuli,
which are rarely identical either over time or
across individuals.
Systematic social analyses would undoubtedly

also reveal that persons often persist in behavior
that receives little or no social support (Bateson,
1961), they forego rewarding activities and

objects which are readily available and socially
permitted, they impose upon themselves highly
unfavorable performance demands (Bandura &
Perloff, 1967), and their actions may be highly
refractory even to severe external consequences
(Farber, Harlow, & West, 1957). These and other
data indicate that regulatory mechanisms can be

established that render behavior partially
independent of specific situational contingencies
and outcomes.
One can differentiate several different types of
“intrinsic” reinforcement control. As noted in

Chapter 4, response patterns can be effectively
maintained without social or natural support by
their intrinsic sensory consequences. Artificial
incentives and a great deal of social surveillance

may initially be required, for example, to induce
children to acquire the necessary skills to play the
piano; but after proficiency is achieved keyboard

performances are likely to be engaged in for their
melodic feedback. Other activities may similarly
be self-reinforced through their intrinsic sensory
feedback. It should be noted, however, that the
reinforcing values of most visual or auditory
stimuli generated by behavioral sequences must
themselves be developed through a process of

differential reinforcement. There is nothing
inherently rewarding about a skillfully executed
Wagnerian aria, an abstract painting, or a tuba
solo.
Response patterns may also be partly

maintained by anticipatory consequences. Studies
reported earlier show that behavior can be
sustained by imagined rewards or punishments.
This process is also vividly illustrated in the case,

cited in Chapter 1, of the patient who tenaciously
performed arduous bizarre rituals designed to
forestall dreadful, hellish torture, even though his

atonement rituals were consistently and severely
punished by ward personnel. In this instance
imagined aversive consequences had such
overpowering effects on the patient’s behavior
that it became relatively autonomous of external
reinforcements.
The third mechanism by which behavior can

become largely independent of situational
contingencies and outcomes involves a process in
which response patterns are largely controlled by
their self-evaluative consequences. As discussed in
detail elsewhere, people adopt certain standards

of behavior and generate self-rewarding or self-
punishing consequences depending upon how
their behavior compares to their self-prescribed
demands. It is this self-imposition of contingencies

that probably serves as the basis for the notion
that values govern conduct. Under conditions
where self-evaluative and externally occurring

consequences conflict, as when a given pattern of
behavior is socially rewarded but personally
devalued, the inhibiting effects of anticipatory self-
criticism may prevail over external rewards.
Conversely, positive self-reinforcement may
maintain, in some strength, behavior that is
nonrewarded or negatively sanctioned by societal

agents whose behavioral standards are
repudiated.
Although in both of the “internalization”
mechanisms discussed above behavior is
internally regulated by self-generated

consequences, the types of outcomes produced
differ in at least one important respect. In the first
case, behavior is controlled by anticipatory
representation of response consequences

administered by external agents. Consequently, in
situations which involve little risk that
transgressive behavior will be detected, or when

anticipated aversive outcomes are mild, people
may readily transgress. In the second case, a
person is deterred from behaving counter to his
standards of conduct by anticipatory self-
punishing responses. Since the person’s own self-
demands and self-respect serve as his main guides
and deterrents, behavior that is under the latter

form of self-control is apt to be less affected by
variations in specific situational contingencies. It
might be noted in passing that the distinction
drawn between types of self-generated
consequences is similar to the common

differentiation of fear-controlled and conscience-
controlled behavior.
The preceding remarks are not intended to
imply, of course, that self-reinforcement standards

do not require some degree of social support.
Persons tend to affiliate with others who share
similar behavioral norms and mutually reinforce

adherence to the standards they have adopted.
Those who choose a small, select reference group
that does not share the values of the vast majority
may appear “inner-directed,” whereas in actuality
they are very much dependent on the real or
imagined approval and disapproval of a few
individuals whose judgments are highly valued.
Stabilization of Behavioral Changes through

Development of Self-Regulatory Functions
By far the most important but most neglected
aspect of behavioral change processes is the

appropriate generalization of established patterns
of behavior to new situations and their persistence

after the original controlling conditions have been
discontinued. The generalization and persistence
of behavior can be facilitated by three different
means. These include transfer training, alteration

of the reinforcement practices of the social
environment, and the establishment of self-
regulatory functions. In cases where newly

established or disinhibited response patterns
either relieve subjective distress or are favorably
received within the naturalistic environment, the
altered behavior will be adequately sustained
without special maintenance procedures.
Occasional negative experiences in the context of

many neutral or rewarding experiences with
formerly threatening events are usually ineffective
in reinstating fear responding. On the other hand,
where behavior that is rewarding and positively
self- evaluative must be counteracted, and where

customary environmental contingencies provide
only weak support for alternative modes of
behavior, the development of self-regulatory
functions is essential if induced behavioral
changes are to transfer and to endure to any

significant degree. This issue is best exemplified in
the treatment of antisocial personalities.
The most fundamental changes would clearly
be accomplished by altering the contingency
structure and reinforcement practices prevailing
in the deviant subculture. This would require
modification of the behavior of individuals who
exert a strong controlling influence over their
fellow members in the social system. However,

owing to the individualistic bias of treatment
approaches and the inordinate difficulties
encountered in efforts to gain adequate control
over antisocial groups, the common procedure is
to remove a transgressor from his usual

environment and to subject him to some type of
social influence.
Severe antisocial behavior can be controlled in
residential centers through differential

reinforcement. Moreover, the resultant
conforming behavior is likely to persist as long as
the institutional sanctions remain in effect. The

residents may, in fact, come to behave
irreproachably and even to perform obligingly
whatever behavior is expected of them in order to
make conditions in the institution as pleasant as
possible and to expedite their release. A beneficent
incentive system in a treatment center may thus
extract considerable prosocial behavior from

delinquents, but such persons often revert to their
usual antisocial patterns whenever supervisory
staff members are no longer present. The
attraction of the deviant subculture can be
reduced by having members acquire alternative

rewarding patterns of behavior and adopt new
standards for self- evaluation.
Findings of studies reviewed in preceding
sections and earlier chapters suggest several

procedures that might be successful in developing
self-regulatory functions. First, the desired
patterns of behavior and standards for self-

reinforcement should be adequately exemplified
by change agents. Second, an explicit set of
performance requirements linked with a graded
system of incentives should be instituted, such
that progressive adoption of more advanced
behavior brings increased freedom, privileges, and
access to rewarding activities. However, provision

of exemplary models and positive inducements for
behavioral change may not alone be sufficient for
transmitting self-monitoring reinforcement
systems to avowedly antisocial persons.
After participants adopt new patterns of

behavior on the basis of their utilitarian value, the
next phase in the program may require direct
training in self-reinforcement. This is achieved by
gradually transferring evaluative and reinforcing

functions from change agents to the individual
himself. Rewards are now made contingent not
only upon occurrence of desired behavior, but also

accurate evaluation of one’s own performances.
Although at this stage the person judges when his
behavior warrants reward according to the
prevailing contingency structures, others still
serve as the reinforcing agents. After accurate self-
evaluative behavior is well established, the
reinforcing function is likewise transferred so that

the individual both evaluates his own behavior
and reinforces himself accordingly. In addition, the
artificial material rewards are gradually reduced
as the person’s behavior is brought increasingly
under the control of self-administered and

symbolic consequences. The ultimate aim of the
training in self-reinforcement is to produce a level
of functioning at which participants can control
their own behavior with minimum external
constraints and artificial inducements.
Another means of instilling self-regulatory
functions is to provide ample opportunities for

participants to enact role behaviors toward peers
that are ordinarily performed by regular change
agents. Specifically, this entails delegating
progressively more of the standard-setting,
evaluative, and reinforcement functions to
members of the group as they progress in the
program. The members themselves, with staff

guidance, thus become the contingency managers.
In order to enhance participants’ willingness to
adopt new role behaviors, increased privileges
and rewards are associated with increased
responsibility for guiding member behavior.

Active participation in decision-making,
application of rewards and sanctions for
regulating the behavior of peers in accord with
institutional standards, and performance of other
counterattitudinal behaviors would be expected to

exert greater influences on values and preferences
than a program in which contingencies are simply

imposed on covertly resistant members. It might
also be supposed that those who willingly
implement reinforcement contingencies

advocated by a social agency for modifying the
behavior of their associates will similarly alter
their own standards of self-reinforcement in the
practiced direction.
When individuals function as change agents for
members of their own group they not only achieve

modifications that might otherwise be strongly
resisted, but they also advance the treatment of
their subordinates by providing models for

desirable modes of behavior. One might expect
peers to be emulated to a much greater extent
than staff because peers are less socially distant,
have more opportunity to exhibit desired
response patterns, and are likely to evoke weaker

resistive tendencies. Additionally, individuals are
less apt to be ostracized for adopting the behavior
of their peers.
Just as self-regulatory functions are socially
transferable and conditionable, they are also
extinguishable unless given a sufficient amount of
social support. Various social groups, which differ
considerably in their behavioral standards and

reinforcement practices, are potentially available
to individuals. The groups with which they choose
to affiliate largely determine the role models and
contingency system to which they will be exposed

and, consequently, the direction in which their
behavior will be further modified. Therefore,
attention to factors governing the selection of
reference groups is of critical importance,
particularly in cases where induced behavioral

changes are discordant with the conduct
advocated and reinforced by the individuals’

former associates. Generalization and
maintenance of personality changes can, therefore,

be best ensured if the program instills in
participants behavioral competencies and self-

reinforcement standards that are likely to exert
decisive influence on association preferences.
After a person has adopted new standards for

self-evaluation, a group’s pressures for conformity
to conflicting behavioral requirements are likely to
be resisted. Instead, when the advocated conduct

is not in accord with self-prescribed standards, the
individual may attempt to alter the demands, he
may remain a marginal member, or, if the rewards

for membership are insufficient, he may
discontinue his association with the group.
If persons are to affiliate with new social

groups, they must acquire at least some of the
requisite behaviors for securing approval and
recognition that will be necessary to sustain their
active involvement. Otherwise, they will be unable
to meet successfully the demands of their new
social environment and will either eventually
withdraw from or be rejected by the group. Many
rehabilitative programs, for example, concentrate
on producing radical changes in offenders’
behavior that will lose them the social and

material rewards associated with a deviant career,
but relatively little attention is devoted to
providing them the means for obtaining adequate
substitute gratifications.
The way in which affiliation processes govern
the course of behavior change is revealed in

studies (Ellis & Lane, 1963; Krauss, 1964)
investigating the sources of high educational
aspirations among lower-class children. In the
families of such children the parents cannot
themselves provide satisfactory models of class-

typed habits of speech, customs, and social skills
required to win acceptance from upper middle-

class peers. The parents characteristically initiate
the upward mobility process by attaching positive

value to educational achievements; admired
teachers further reinforce, by approval and
example, college ambitions in lower-class youth;
and selective association with college-oriented

peers provides the social-learning conditions for
the gradual acquisition of attitudes, belief systems,
and complex behavioral repertoires necessary for
achieving the desired socioeducational status.
Bandura & Walters (1959) have similarly shown

that adolescents tend to choose close associates
who share similar value systems and behavioral
norms; peer group members, in turn, serve to
reinforce and to uphold the standards of behavior
that the boys adopted.
Summary
Several theories have been proposed
concerning the role of symbolic processes in the
regulation of behavior. These range from
nonmediational views that assume that
reinforcing consequences modify behavior directly
and automatically, to cognitive formulations that
consider symbolic representation of contingencies

a prerequisite for learning and performance
change. A reciprocal-interaction theory seems best
able to order the divergent findings bearing on
this issue. According to this view, reinforcing
consequences can alter behavior independently of

awareness, but individuals eventually infer, from
observation of their behavior and its differential
outcomes, the correct reinforcement rules which
partly control subsequent responding.
In studies of both instrumental and classical

conditioning persons who discern the
contingencies governing the administration of

rewards and punishments typically display
significant increments in learning or performance,
whereas unaware subjects generally show few or
no conditioning effects. Interpretation of these
findings, however, remains unclear because the
studies lack the data necessary for determining
whether conditioning failures in the absence of

awareness are attributable to inadequate central
registration of sensory inputs or to lack of
recognition of contiguous stimulus events that
have registered and evoked neural responses.
Experiments designed so that subjects cannot

observe either their own responses or the
occurrence of reinforcing events—thus precluding
recognition of contingencies—reveal that
conditioning can occur, albeit much less reliably,
on a nonmediated basis. The overall findings seem

to indicate that awareness is a powerful
facilitative factor, but it may not be a necessary

and certainly is not a sufficient condition for
behavioral change. Awareness in itself is unlikely
to produce response changes unless persons

possess the necessary performance skills and
unless adequate incentives are provided to elicit
and to sustain appropriate responsiveness.
Symbolic activities not only augment the
efficacy of reinforcement operations, but they are
also increasingly employed to generate emotional

effects that constitute the major reinforcing
consequences in behavioral modification
programs. In symbolic desensitization both the

aversive stimuli and the emotion-neutralizing
responses are in large part symbolically induced.
Similarly, in aversive cognitive
counterconditioning, avoidance responses toward
addictive objects are established by contiguous

association of symbolic representations of
positively valenced stimuli with thought-produced
nauseous reactions. Imagined consequences may
also be employed instrumentally as covert
reinforcers either to strengthen or to reduce the
incidence of overt behavior. Perturbing trains of
thought often disrupt psychological functioning, in
which case the problem becomes one of
controlling symbolic events themselves. Self-

control of thought processes can be achieved by
redirecting attention to absorbing activities that
elicit competing cognitions and by self-
reinforcement of more constructive lines of
thought.
The issue of mediational control of behavior is

also frequently raised in the context of attitude
theory. Although it is commonly assumed that
attitudinal changes have widespread and

stabilizing influences upon overt actions, induced
alterations in attitudes in fact generally have few
enduring effects upon behavior unless they

receive sufficient reinforcement support. On the
other hand, direct modification of the affective
properties of attitude objects and performance of
attitude-discrepant behavior produce stable
corresponding changes in attitudes. This process
of cognitive accommodation to affective and
behavioral changes has been variously attributed

to striving for cognitive consistency, to response
generalization of reinforcement effects, to the self-
persuasive influence of counterattitudinal
behaviors, and to new experiential consequences
resulting from the induced behavioral changes. It

still remains to be established whether
environmental influences have similar but
independent effects on feelings, beliefs, and
behavior, or whether a change in one of these
components engenders congruous modifications

in the other constituents.
When regulatory symbolic processes are

combined with self-generated consequences,
behavior may become “internalized” or partially
independent of situational contingencies and
outcomes. Several different types of “intrinsic”
reinforcement control can be distinguished.
Behavior may be sustained by its inherent sensory
feedback, by anticipatory outcomes, or by self-

evaluative consequences.
The establishment of self-monitoring
reinforcement systems is essential if induced
behavioral changes are to generalize and endure,
particularly where social environments provide

either weak support for new modes of behavior or
conflicting patterns of reinforcement. Stabilization
of changes is most likely to be ensured when the
standards adopted for self-reinforcement result in

selective association with persons who share
similar behavioral norms, thus providing social
support for one’s own system of self-evaluation.
REFERENCES
Abelson, R. P., Aronson, E., McGuire, W. J., Newcomb,

T. M., Rosenberg, M. J., & Tannenbaum, P. H.
Theories of cognitive consistency: A sourcebook.
Chicago: Rand McNally, 1968.
Abelson, R. P., & Rosenberg, M. J. Symbolic psycho-

logic: A model of attitudinal cognition. Behavioral
Science, 1958, 3, 1-13.
Adams, J. K. Laboratory studies of behavior without

awareness. Psychological Bulletin, 1957, 54, 393-
405.
Aronson, E. The psychology of insufficient

justification: An analysis of some conflicting data.
In S. Feldman (Ed.), Cognitive consistency. New
York: Academic Press, 1966. pp. 109–133.
Aronson, E., Turner, J. A., & Carlsmith, J. M.

Communicator credibility and communication
discrepancy as determinants of opinion change.
67, 31–36.

instructions with mental patients. Journal of the
331.

McNally, 1968.
13, 173-199.
Bandura, A., & Perloff, B. Relative efficacy of self-

Psychology, 1967, 7, 111-116.
Bandura, A., Ross, D., & Ross, S. A. Vicarious

reinforcement and imitative learning. Journal of
607.


Barber, T. X., & Hahn, K. W., Jr. Experimental studies in

“hypnotic” behavior: Physiological and subjective
effects of imagined pain. Journal of Nervous and
Mental Disease, 1964, 139, 416-425.
Baron, A., & Kaufman, A. Human, free-operant

avoidance of time out from monetary
reinforcement. Journal of the Experimental
Bateson, G. (Ed.) Perceval’s narrative: A patient’s

account of his psychosis, 1830-1832. Stanford
Bem, D. J. Self-perception: An alternative

interpretation of cognitive dissonance
phenomena. Psychological Review, 1967, 74, 183-
200.
Bergin, A. E. The effect of dissonant persuasive

communications upon changes in a self-referring
attitude. Journal of Personality, 1962, 30, 423-
438.
Blanchard, E. B. The relative contributions of

modeling, informational influences, and physical
contact in the extinction of phobic behavior.
Unpublished doctoral dissertation, Stanford
University, 1969.
Bourne, L. E., Jr. Human conceptual behavior. Boston:

Allyn & Bacon, 1966.
Breer, P. E., & Locke, E. A. Task experience as a source

of attitudes. Homewood, Ill.: Dorsey, 1965.
Brehm, J. W., & Cohen, A. R. Exploration in cognitive

dissonance. New York: Wiley, 1962.
Bridger, W. H., & Mandel, I. J. A comparison of GSR
fear responses produced by threat and electric
shock. Journal of Psychiatric Research, 1964, 2,
31-40.
Bridger, W. H., & Mandel, I. J. Abolition of the PRE by

instructions in GSR conditioning. Journal of
Brown, R. W. Social psychology. New York: Free Press,

1965.
Bruning, J. L. Effects of magnitude of reward and

percentage of reinforcement on a lever
movement response. Child Development, 1964,
35, 281-285.
Carlin, M. T. The effects of modeled behavior during

imposed delay on the observer’s subsequent
willingness to delay rewards. Unpublished
doctoral dissertation, Stanford University, 1965.
Carlsmith, J. M., Collins, B. E., & Helmreich, R. L.

Studies in forced compliance: I. The effect of
pressure for compliance on attitude change
produced by face-to-face role playing and
anonymous essay writing. Journal of Personality
16, 33–41.
Chapanis, N. P., & Chapanis, A. Cognitive dissonance:

Five years later. Psychological Bulletin, 1964, 61,
1-22.
Chatterjee, B. B., & Eriksen, C. W. Cognitive factors in

heart rate conditioning. Journal of Experimental
Psychology, 1962, 64, 272-279.
Clark, D. F. The treatment of monosymptomatic

phobia by systematic desensitization. Behaviour
Cohen, A. R. Attitude change and social influence. New

York: Basic Books, 1964.
Collins, B. E. The effect of monetary inducements on

amount of attitude change produced by forced
compliance. In A. C. Elms (Ed.), Role playing,
reward, and attitude change. Princeton, N. J.: Van
Nostrand, 1969, 209-223.
Cook, S. W., & Harris, R. E. The verbal conditioning of

the galvanic skin reflex. Journal of Experimental
Psychology, 1937, 21, 202-210.
Culbertson, F. M. Modification of an emotionally held

attitude through role playing. Journal of Abnormal
Dabbs, J. M., Jr., & Janis, I. L. Why does eating while
reading facilitate opinion change? An
experimental inquiry. Journal of Experimental
Das, J. P., & Nanda, P. C. Mediated transfer of attitudes.

66, 12-16.
Dawson, M. E. Comparison of classical conditioning

and relational learning. Unpublished master’s
thesis, University of Southern California, 1966.
Dawson, M. E., & Grings, W. W. Comparison of classical

conditioning and relational learning. Journal of
Experimental Psychology, 1969 (in press).
De Nike, L. D. The temporal relationship between

awareness and performance in verbal
1964, 68, 521-529.
Dollard, J., & Miller, N. E. Personality and

psychotherapy. New York: McGraw-Hill, 1950.
Doob, L. W. The behavior of attitudes. Psychological

Review, 1947, 54, 135–156.
Dulany, D. E. The place of hypotheses and intentions:

An analysis of verbal control in verbal
conditioning. In C. W. Eriksen (Ed.), Behavior and
awareness—a symposium of research and
interpretation. Durham, N. C.: Duke University
Press, 1962. pp. 102-129.
Dulany, D. E. Awareness, rules, and prepositional

control: A confrontation with S-R behavior
theory. In T. R. Dixon & D. L. Horton (Eds.), Verbal
behavior and general behavior theory. Englewood
Cliffs, N. J.: Prentice-Hall, 1968. pp. 340-387.
Dulany, D. E., Jr., & Eriksen, C. W. Accuracy of

brightness discrimination as measured by
concurrent verbal responses and GSRs. Journal of
Abnormal arid Social Psychology, 1959, 59, 418–
423.
Dulany, D. E., Jr. & O’Connell, D. C. Does partial

reinforcement dissociate verbal rules and the
behavior they might be presumed to control?
Journal of Verbal Learning and Verbal Behavior,
1963, 2, 361-372.
Duncker, K. Experimental modification of children’s

food preferences through social suggestion.
33, 489-507.
Ekman, P., Krasner, L., & Ullmann, L. P. Interaction of

set and awareness as determinants of response
to verbal conditioning. Journal of Abnormal and
Elliott, R. Reaction time and heart rate as functions of

magnitude of incentive and probability of
success: A replication and extension. Journal of
Experimental Research in Personality, 1966, 1,
174-178.
Ellis, R. A., & Lane, W. C. Structural supports for

upward mobility. American Sociological Review,
1963, 28, 743-756.
Elms, A. C. Role playing, incentive, and dissonance.

Elms, A. C., & Janis, I. L. Counter norm attitudes

induced by consonant versus dissonant
conditions of role-playing. Journal of
60.
Eriksen, C. W. Prediction from and interaction among

multiple concurrent discriminative responses.
353-359.
Eriksen, C. W. Unconscious processes. In M. R. Jones

(Ed.) Nebraska symposium on motivation. Lincoln:
University of Nebraska Press, 1958. pp. 169-227.
Eriksen, C. W. Discrimination and learning without
awareness: A methodological survey and
evaluation. Psychological Review, 1960, 67, 279-
300.
Eriksen, C. W., & Kuethe, J. L. Avoidance conditioning

of verbal behavior without awareness: A
paradigm of repression. Journal of Abnormal and
Farber, I. E. The things people say to themselves.

American Psychologist, 18, 185-197.
Farber, I. E., Harlow, H. F., & West, L. J. Brainwashing,

conditioning, and DDD (debility, dependency, and
dread). Sociometry, 1957, 20, 271-285.
Feldman, S. Cognitive consistency. New York:


control of eating. Journal of Mathetics, 1962,1, 87-
109.

Crofts, 1957.
Festinger, L. A theory of cognitive dissonance.

Evanston, Ill.: Row, Peterson, 1957.
Festinger, L. Behavioral support for opinion change.
Public Opinion Quarterly, 28, 404-417.
Festinger, L., & Carlsmith, J. M. Cognitive

consequences of forced compliance. Journal of
210.
Fleishmann, E., Harris, E., & Burtt, H. Leadership and

supervision in industry: An evaluation of a
supervisory training program. Columbus: Ohio
State University, Bureau of Educational Research,
1955.
Freud, S. The interpretation of dreams. London:

Hogarth, 1953.
Fuhrer, M. J., & Baer, P. E. Differential classical

conditioning: Verbalization of stimulus
contingencies. Science, 1965, 150, 1479-1481.
Goldiamond, I. Indicators of perception: I. Subliminal

perception, subception, unconscious perception:
An analysis in terms of psychological indicator
methodology. Psychological Bulletin, 1958, 55,
373-412.
Goodnow, J. J., & Postman, L. Probability learning in a

problem-solving situation. Journal of
Greenwald, A. G. Behavior change following a
persuasive communication. Journal of Personality,
1965, 33,370-391. (a)
Greenwald, A. G. Effects of prior commitment on

behavior change after a persuasive
communication. Public Opinion Quarterly, 1965,
29, 595-601. (b)
Grings, W. W. Verbal-perceptual factors in the

conditioning of autonomic responses. In W. F.
Prokasy (Ed.), Classical conditioning: A
symposium. New York: Appleton-Century-Crofts,
1965. pp. 71-89.
Grings, W. W., & Lockhart, R. A. Effects of “anxiety-

lessening” instructions and differential set
development on the extinction of GSR. Journal of
Hefferline, R. F., & Keenan, B. Amplitude-induction

gradient of a small-scale (covert) operant. Journal
307-315.
Hefferline, R. F., Keenan, B., & Harford, R. A. Escape

and avoidance conditioning in human subjects
without their observation of the response.
Science, 1959,130, 1338-1339.
Heider, F. The psychology of interpersonal relations.
New York: Wiley, 1958.
Hernández-Péon, R., Scherrer, H., & Jouvet, M.

Modification of electric activity in cochlear
nucleus during “attention” in unanesthetized cats.
Science, 1956,123, 331-332.
Hirsch, J. Learning without awareness and extinction

following awareness as a function of
reinforcement. Journal of Experimental
Psychology, 1957, 54, 218-224.
Hislop, M. W., & Brooks, L. R. Suppression of concept

learning by verbal rules. Journal of Experimental
Psychology, in press.
Holmes, D. S. Verbal conditioning, or problem-solving

and cooperation? Journal of Experimental
Research in Personality, 1967, 2, 289-294.

Horn, G. Electrical activity of the cerebral cortex of the

unanesthetized cat during attentive behavior.
Brain, 1960, 83, 57-76.
Hovland, C. I., & Janis, I. L. (Eds.), Personality and

persuasibility. New Haven: Yale University Press,
1959.
Communication and persuasion: Psychological
studies of opinion change. New Haven: Yale
Hovland, C. I., Lumsdaine, A. A., & Sheffield, F. D.

Experiments on mass communication. Princeton,
N. J.: Princeton University Press, 1949.
Hovland, C. I., & Weiss, W. W. The influence of source

credibility on communication effectiveness.
Public Opinion Quarterly, 1951, 15, 635-650.
Insko, C. A., & Oakes, W. F. Awareness and the

“conditioning” of attitudes. Journal of Personality
Janis, I. L. Attitude change via role-playing. In R. P.

Abelson, E. Aronson, W. J. McGuire, T. M.
Newcomb, & P. H. Tannenbaum (Eds.), Theories of
cognitive consistency: A sourcebook. Chicago:
Rand McNally, 1968.
Janis, I. L. Effects of fear arousal on attitude change:

Recent developments in theory and experimental
research. In L. Berkowitz (Ed.), Advances in
experimental social psychology. Vol. III. New York:
Janis, I. L., & Gilmore, J. B. The influence of incentive
conditions on the success of role playing in
modifying attitudes. Journal of Personality and
Janis, I. L., Kaye, D., & Kirschner, P. Facilitating effects

of “eating-while- reading” on responsiveness to
persuasive communications. Journal of
Personality and Social Psychology, 1965, 1, 181–
186.
Janis, I. L., & King, B. T. The influence of role playing

on opinion change. Journal of Abnormal and
Johnson, H., & Eriksen, C. W. Preconscious perception:

A re-examination of the Poetzl phenomenon.
62, 497-503.

1968. pp. 254-290.
Kaufman, A., Baron, A., & Kopp, R. E. Some effects of

instructions on human operant behavior.
Psychonomic Monograph Supplements, 1966, 1,
243-250.
Keller, L., Cole, M., Burke, C. J., & Estes, W. K. Reward
and informational values of trial outcomes in
paired-associate learning. Psychological
Kelman, H. C. Attitude change as a function of

response restriction. Human Relations, 1953, 6,
185-214.
Kelman, H. C. The induction of action and attitude

change. Proceedings of the XIV International
Congress of Applied Psychology. Copenhagen:
Munksgaard, 1961.
Kelman, H. C., & Hovland, C. I. Reinstatement of the

communicator in delayed measurement of
opinion change. Journal of Abnormal and Social
Psychology, 1953, 48, 327-335.
Klein, G. S., Spence, D. P., Holt, R. R., & Gourevitch, S.

Cognition without awareness: Subliminal
influences upon conscious thought. Journal of
266.
Krasner, L. Studies of the conditioning of verbal

170.
Krasner, L., & Ullmann, L. P. Variables affecting report
of awareness in verbal conditioning. Journal of
Psychology, 1963, 56, 193-202.
Krasner, L., Ullmann, L. P., Weiss, R. L., & Collins, B. J.

Responsivity to verbal conditioning as a function
of three different examiners. Journal of Clinical
Psychology, 1961, 17,411-415.
Krauss, I. Sources of educational aspirations among

working-class youth. American Sociological
Review, 1964, 29, 867-879.
Lazarus, R. S., & McCleary, R. A. Autonomic

discrimination without awareness: A study of
subception. Psychological Review, 1951, 58, 113-
122.
Levin, S. M. The effects of awareness on verbal

1961, 61, 67-75.
Levitt, T. Industrial purchasing behavior: A study of

communications effects. Boston: Harvard
Levy, L. H. Awareness, learning, and the beneficent

subject as expert witness. Journal of Personality
Lewis, D. J., & Duncan, C. P. Effects of variable

magnitude of reward on a lever-pulling response.
203-205.
Linder, D. E., Cooper, J., & Jones, E. E. Decision freedom

as a determinant of the role of incentive
magnitude in attitude change. Journal of
254.
Lovaas, O. I. Interaction between verbal and

nonverbal behavior. Child Development, 1961, 32,
329-336.
Maccoby, N., Romney, A. K., Adams, J. S., & Maccoby, E.

E. “Critical periods” in seeking and accepting
information. Stanford: Stanford University
Institute for Communication Research, 1962.
Mackay, H. A., & Laverty, S. G. GSR changes during

therapy of phobic behavior. Unpublished
manuscript, Queen’s University, Ontario, 1963.
Mandel, I. J. & Bridger, W. H. Interaction between

instructions and ISI in conditioning and
extinction of the GSR. Journal of Experimental
Psychology, 1967, 74, 36-43.
Marks, I. M., & Gelder, M. G. Transvestism and

fetishism: Clinical and psychological changes
during faradic aversion. British Journal of
Psychiatry, 1967,113, 711-729.
McGuire, W. J. Inducing resistance to persuasion. In L.

Berkowitz (Ed.), Advances in experimental social
psychology. Vol. I. New York: Academic Press,
1964. pp. 191-229.
Miller, M. M. Treatment of chronic alcoholism by

hypnotic aversion. American Medical Association
Journal, 1959,171, 1492-1495.
Miller, M. M. Hypnotic-aversion treatment of

homosexuality. Journal of the National Medical
Association, 1963, 55, 411-415.
Miller, N. E. Learnable drives and rewards. In S. S.

psychology, New York: Wiley, 1951. pp. 435-472.
Mischel, W. Personality and assessment. New York:

Wiley, 1968.
Newcomb, T. M., Turner, R. H., & Converse, P. E. Social

psychology: The study of human interaction. New
York: Holt, Rinehart & Winston, 1965.
Nuttin, J. M., Jr. Attitude change after rewarded

dissonant and consonant “forced compliance.”
International Journal of Psychology, 1966, 1, 39-
57.
O’Connell, D. C., & Wagner, M. V. Extinction after
partial reinforcement and minimal learning as a
test of both verbal control and PRE in concept
learning. Journal of Experimental Psychology,
1967, 73, 151-153.
Osgood, C. E., & Tannenbaum, P. H. The principle of

congruity in the prediction of attitude change.
Philbrick, D. B., & Postman, L. A further analysis of

“learning without awareness.” American Journal
of Psychology, 1955, 68, 417-424.
Postman, L., & Sassenrath, J. The automatic action of

verbal rewards and punishments. Journal of
General Psychology, 1961, 65, 109-136.
Premack, D. Reinforcement theory. In D. Levine (Ed.),

Nebraska symposium on motivation: 1965.
Lincoln: University of Nebraska Press, 1965. pp.
123-180.
Razran, G. PI. S. Conditioning away social bias by the

luncheon technique. Psychological Bulletin, 1938,
35, 693.
Razran, G. H. S. Conditioned response changes in

rating sociopolitical slogans. Psychological
Bulletin, 1940, 37, 481.
Razran, G. Conditioning and perception. Psychological
Review, 1955, 62, 83-95.
Razran, G. The observable unconscious and the

inferable conscious in current soviet
psychophysiology. Psychological Review, 1961,
68, 81-147.
Rogers, C. R. A theory of therapy, personality, and

interpersonal relationships, as developed in the
client-centered framework. In S. Koch (Ed.),
Psychology: A study of a science. Vol. III.
Formulations of the person and the social context.
New York: McGraw-Hill, 1959. pp. 184-256.
Rokeach, M. Attitude change and behavioral change.

Public Opinion Quarterly, 1966, 30, 529-550.
Rosenberg, M. J. An analysis of affective-cognitive

consistency. In M. J. Rosenberg, C. I. Hovland, W. J.
McGuire, R. P. Abelson, & J. W. Brehm (Eds.),
Attitude organization and change: An analysis of
consistency among attitude components. New
Haven: Yale University Press, 1960. pp. 15-64.
Rosenberg, M. J. When dissonance fails: On

eliminating evaluation apprehension from
attitude measurement. Journal of Personality and
Rosenberg, M. J. Some limits of dissonance: Toward a
differentiated view of counter-attitudinal
performance. In S. Feldman (Ed.), Cognitive
consistency. New York: Academic Press, 1966. pp.
135-170.
Rosenberg, M. J., Hovland, C. I., McGuire, W. J., Abelson,

R. P., & Brehm, J. W. Attitude organization and
change: An analysis of consistency among attitude
components. New Haven: Yale University Press,
1960.
Rosenthal, R., Persinger, G. W., Vikan-Kline, L. I., &

Fode, K. L. The effect of experimenter outcome-
bias and subject set on awareness in verbal
conditioning experiments. Journal of Verbal
Learning and Verbal Behavior, 1963, 2, 275-283.
Sasmor, R. M. Operant conditioning of a small-scale

muscle response. Journal of the Experimental
Sassenrath, J. M. Transfer of learning without

awareness. Psychological Reports, 1962, 10, 411-
420.
Schanck, R. L. A study of a community and its groups

and institutions conceived of as behavior of
individuals. Psychological Monographs, 1932, 43
(2, Whole No. 195).
Scobie, S. R., & Kaufman, A. Intermittent punishment
of variable interval and variable ratio responding.
1969, 12,137-147.
Shepard, R. N., Hovland, C. I., & Jenkins, H. M. Learning

and memorization of classifications. Psychological

Crofts, 1966. pp. 448-498.

Macmillan, 1953.
Smith, G. H. The influence of a female model on

perceived characteristics of an automobile. Paper
read at the meeting of the American
Psychological Association, San Francisco, August,
1968.
Spence, D. P. Conscious and preconscious influences

on recall: Another example of the restricting
effects of awareness. Journal of Abnormal and
Spence, D. P., & Holland, B. The restricting effects of

awareness: A paradox and an explanation.
64, 163-174.

Spielberger, C. D., Berger, A., & Howard, K.

Conditioning of verbal behavior as a function of
awareness, need for social approval, and
motivation to receive reinforcement. Journal of
246.
Spielberger, C. D., Bernstein, I. H., & Ratliff, R. G.

Information and incentive value of the
reinforcing stimulus in verbal conditioning.
31.
Spielberger, C. D., & De Nike, L. D. Descriptive

behaviorism versus cognitive theory in verbal
operant conditioning. Psychological Review, 1966,
73, 306-326.
Spielberger, C. D., Levin, S. M., & Shepard, M. C. The

effects of awareness and attitude toward the
reinforcement of the operant conditioning of
verbal behavior. Journal of Personality, 1962, 30,
106-121.
Staats, A. W., & Staats, C. K. Attitudes established by
classical conditioning. Journal of Abnormal and
Staats, C. K., & Staats, A. W. Meaning established by

classical conditioning. Journal of Experimental
Psychology, 1957, 54, 74-80.
Thorndike, E. L. A theory of the action of the after-

effects of a connection upon it. Psychological
Review, 1933, 40, 434-439.
Verplanck, W. S. Unaware of where’s awareness:

Some verbal operants—notates, monents, and
notants. In C. W. Eriksen (Ed.), Behavior and
awareness. Durham, N. C.: Duke University Press,
1962. pp. 130-158.
Walters, R. H., & Parke, R. D. Influence of response

consequences to a social model on resistance to
deviation. Journal of Experimental Child
Psychology, 1964, 1, 269-280.
Weiner, H. Real and imagined cost effects upon

human fixed-interval responding. Psychological
Reports, 1965, 17, 659-662.
Weinstein, W. K., & Lawson, R. The effect of

experimentally-induced “awareness” upon
performance in free-operant verbal conditioning
and on subsequent tests of “awareness.” Journal
of Psychology, 1963, 56, 203-211.
Weiss, R. L., Krasner, L., & Ullmann, L. P. Responsivity

to verbal conditioning as a function of emotional
atmosphere and pattern of reinforcement.
Wickens, D. D., Allen, C. K., & Hill, F. A. Effects of

instruction and UCS strength on extinction of the
conditioned GSR. Journal of Experimental
Psychology, 1963, 66, 235-240.
Worell, L., & Worell, J. An experimental and

theoretical note on conscious and preconscious
influences on recall. Journal of Personality and

Principles of Behavior Modification Albert Bandura

Uploaded by

Copyright:

Available Formats

Principles of Behavior Modification Albert Bandura

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Principles of Behavior Modification Albert Bandura

Uploaded by

Copyright:

Available Formats

Principles of

e-Book 2019 International Psychotherapy

All Rights Reserved

This e-book contains material protected under

Created in the United States of America

Social Learning as a Reciprocal Influence Process

Efficacy of Conventional Methods of Behavioral

2. Value Issues and Objectives

Factors Impeding Specification of Objectives

Decision Processes in the Selection of Objectives

3. Modeling and Vicarious Processes

Establishment of New Response Patterns

Vicarious Conditioning of Emotional

Inhibitory and Disinhibitory Effects of Vicarious

Response Facilitating Effects of Modeling

Utilization of Modeling Principles in Planned

Essential Components of Reinforcement

Ethical Implications of Reinforcement Practices

Applications of Contingency Systems

Social Organizational Applications of

Applications of Aversive Contingency Systems

Extinction of Positively Reinforced Behavior

Extinction of Defensive Behavior

7. Desensitization through Counterconditioning

Identification of the Stimulus Determinants of

Neutralization of Threats in Symbolic or Realistic

Antagonistic Activities in Counterconditioning

Physiological Accompaniments of Emotional

Ethical Considerations in Aversion Therapy

9. Symbolic Control of Behavioral Changes

Verbal Conditioning as a Function of Awareness

Interactive Effects of Cognitive and Incentive

Symbolic Control of Classical Conditioning

Implications of Symbolic Control for Behavioral

Discrepancy between Response Systems and the

Attitudinal Consequences of Behavioral and

Strategies of Attitude Change

“Internalization” and Persistence of Behavioral

Stabilization of Behavioral Changes through

principles governing human behavior within the

years an impressive body of knowledge about the

mechanisms through which behavior is acquired

and modified has been accumulated. But despite

this vigorous growth of research on human

that are highly influential in human functioning

have been overlooked or only partially

investigated. This volume reviews the recent

theoretical and experimental advances in the field

important roles played by vicarious, symbolic, and

little notice even in most contemporary theories of

The worth of a psychological theory must be

judged not only by how well it explains laboratory

modification procedures that it produces. Recent

years have witnessed widespread applications of

to the modification of important social phenomena

in familial, educational, clinical, and other social

settings. By requiring clear specification of

outcomes, the social-learning approach presented

in this book contains a self-corrective feature that

distinguishes it from change enterprises in which