SPECIAL REVIEW

KOFFKA'S PRINCIPLES OF GESTALT PSYCHOLOGY *

BY EDWIN G. BORING
Harvard University

At last we have a thorough, complete, systematic, representative

Gestalt psychology! For 15 years this vigorous, influential, philo-
sophical and experimental movement has been represented by a large
experimental literature, by the scattered writings of its chief pro-
ponents, or by the less authentic expositions of outsiders or late con-
verts. There is a tradition that the test of a school of psychology lies
in its capacity to produce a systematic text which exhibits the ade-
quacy and advantages of its special point of view throughout the range
of psychology. Wundt, Brentauo, the early Kulpe, James, Titchener,
and Watson met this test. Stumpf, Muller, the Kiilpe of Wiirzburg,
and Angell never did, and their influence on psychology has been less
because their views as applied to many topics remained indeterminate.
Therefore, we owe Koffka a debt of gratitude because he, as one of
three men competent to speak authoritatively of Gestalt psychology,
has exhumed from a mass of literature the proper bones and put
them together so as to show the nature of the Gestalt creature by the
exhibition of a complete skeleton.
The book is large and difficult, but it is also erudite and urbane.
To read it understanding^ is a formidable task for one who is not
already well habituated to at least some of its concepts. On the other
hand, reading yields the reward of an arduous adventure. The action
of the book takes place entirely within a special world of psychological
things, and this for most of us is a strange new land. Heretofore
psychology has had the common dualism of actual experience and
' physical objects'. It has also had the common monism of the
so-called physical world, adjusted to include psychology because
neural and behavioral events can be regarded as physical. It has
also talked about ' reals ' being ' constructs ', with the implication that
psychology might have its own special world of proper constructs;
and particular constructs, like the unconscious and the trace, have
1
Koffka, K., Principles of Gestalt Psychology.- New York: Harcourt,.
Brace and Company, 1935. Pp. xi+720.
59
 60 EDWIN G. BORING
sometimes emerged. However, I have never before read a completed
system that created and kept consistently within such a special world.
There may be such systems, and perhaps psychoanalysis provides one.
In any case the Gestalt world is a new world and, in Koffka's care,
almost a self-sustaining one. The creation of such a system of con-
structs and the maintenance of the exposition within the system is
no mean achievement, and I do homage to the psychology and the
author that have accomplished this feat.
No abstract is possible of a book so replete with information, so
novel in its approach, so strange in its selection of minutiae, so valiant
and meticulous in its polemizing, so scrupulous in its logic, and so
interminably perverse to tradition. It is no wonder that the book is
difficult, although Koffka's persistent candor in exposing (and in
general repairing) the clefts in his armour is refreshing. Let us indi-
cate the general scope and nature of the text before we plunge bravely
into the war that Koffka wages.
It is possible to regard the book as constituted of four parts:
(1) the general argument establishing the fundamental concepts (103
pp.), (2) visual perception (200 pp.), (3) action (117 pp.), and
(4) memory, learning, recognizing, and thinking (225 pp.). There
are also a short and admittedly inadequate chapter on society and
personality (32 pp.), a conclusion (6 pp.), a bibliography (345 titles,
of which 76 are from Psychologische Forschung), and a large index
(18 pp.).
The first part selects the field of battle and ' digs in'. Psychology
has the cultural advantage of combining the study of nature, life, and
meaning, that is to say, it deals with quantification, organization, and
signification. Organization (order) is a subject-matter because psy-
chology undertakes to determine the laws of dynamical interaction
between things and events in the psychological field. Thus the psy-
chologist is always considering totalities (Gestalten) which are molar
(not molecular). However, signification (significance, meaning) is
just as fundamental in the psychological data as in their organization.
The individual organizes its environment into things and things into
• larger structures, and the concept thing carries all the implication and
significance of a meaning. The facts of perceptual constancy support
this view, because then an organism is found recreating and main-
taining a thing, which has not been maintained in the proximal stimu-
lation at the sense-organ, but which has been recreated by the laws of
organization that belong to the organism itself.
The material of psychology, Koffka holds, is molar behavior.
 SPECIAL REVIEW 61
Dynamical organization occurs within the behavioral environment,
and the behavioral environment is the environment as it is to the
individual—not what we might call the ' real' or ' physical' environ-
ment or what Koffka calls the " geographical" environment. This
view sounds like what we are accustomed to call a behaviorism, until
we are confronted by the fact that this behavior is intrinsically mean-
ing. If a rat runs toward food in a food-box, the food is there in
the behavioral environment. It is irrelevant that the experimenter
has removed the ' real' food from the ' geographical' food-box. If I
walk across the thin ice of a snow-covered frozen lake and ' think'
(as the layman would say) that I am walking across a field, then
there is a field in my behavioral environment and no lake. Such a
" behavioral object" as this field takes the place of the perceptual
objects of a simple dualism, and the word behavioral occurs through-
out the text where the naive writer would have written conscious.
The chapters on visual perception form distinctly the ablest por-
tion of the book. In them the reader penetrates into the special psy-
chological world with its atmosphere of meaning. It is possible for
Koffka chiefly to present facts that are derived from experiments and
thus to provide empirical validation of his concepts. He shows us
that primitively the visual field is not even organized into units, but
that organization yields units and, at higher levels, shapes, lines and
points. The point—the molecule of a Wundtian perception—is the
result of a high degree of organization. Figure and ground are a
type of " duo organization ". The one is thing; the other is frame-
work. Contours bound the figure but not the ground, for the ground
is continuous behind the figure. One perceives the table beneath the
book which lies upon it. Wundt's book would have filled a hole in
the table, but not Koffka's. The constancies of shape, size, whiteness,
and color show the organism maintaining things in the face of varia-
tion of proximal stimulation. When one knows what it is that is
invariant when these aspects of behavioral objects are constant in
spite of varying stimulation, then one knows the law of the perception
in question. Visual space is anisotropic: no free transposition of
extensions in space without change of size or shape is possible. The
facts of the visual perception of movement and velocity support this
conception.
The chapters on action dismiss as molecular the reflex and the
stimulus-response concepts. They introduce the Ego, which is a vari-
able thing, expanding now to include its clothes, its country and its
ideals, and contracting again to exclude even its hand. The " bound-
-62 EDWIN G. BORING
ary membrane" of the Ego is determined by the way in which the
Ego as a behavioral object enters into reaction with other behavioral
objects in the psychological field. The Executive (Lewin's Motorik)
is derived from the Ego and aims at a relief of stress which may or
may not result directly in action. In understanding these dynamical
systems we have to realize that things have demand characters, which
they have acquired in the process of their organization, and also
physiognomic characters, which appear in their initial organization.
The newly hatched chick pecks at small grains because of their physi-
ognomic character, but I buy Koffka's book because of its demand
•character. The chief problem of action is the adjustment of behavior,
whereby the individual becomes adequate to his world. Thus we may
revive the old concept of attitude as a general adjustment to the field,
and the concept of attention as a specific adjustment to the field.
Emotion and will are not really special psychological chapters, but
terms which focus interest upon particular systems of behavioral
adjustment. Koffka rejects the classification of emotion as futile and
the physiology of emotion as irrelevant, and deals instead with the
dynamics of " saturation " and subsequent '' explosion ". There are
not yet many experiments extant to support the content of this
section.
The last four chapters might be put under two headings: (1)
Traces and (2) The Communication of Traces. The concept of the
trace is needed to account for the continuity of behavior. Koffka
thinks isomorphically of the trace as a neural condition in the brain,
but he could regard it as a psychological construct for all he gains
by his physiology. The laws of the trace emerge from the experi-
ments on successive comparison. Traces persist. Traces from the
repetition of a given process pile up " in columns ". Such persisting
traces interact dynamically and undergo change, provided they com-
municate with one another. Memory implies communication between
a process and a trace. Forgetting is usually due to lack of communica-
tion. Learning is simply the leaving of a trace by a process, but the
acquisition of skill implies that the deposited traces have interacted
cumulatively and left a new resultant. Associative learning is simply
a kind of associative reorganization of traces (the molecular nonsense
syllable has been excommunicated from psychology), and, since
organization is dynamic, association can thus be regarded as a force.
Recognition requires that a process should communicate with a trace,
and the peculiar Me-ness (feeling of familiarity) of recognition is due
to the communication of the trace with the Ego-object trace. Think-
 SPECIAL REVIEW 63
ing solves problems by reorganizing traces by means of their com-
munications, and the law of effect means that success acts upon the
organization of traces. The experiments cited here are mostly from
Koffka's or Lewin's students, and the dead hand of Ebbinghaus no
longer throttles these problems.
This scant synopsis can at least whet the reader's curiosity about
Koffka's book. It also delineates the universe within which our
further discussion must take place. Having placed on record my
admiration for this thorough, erudite, sophisticated, original, system-
atic treatise, I wish now to devote the remainder of this review to a
criticism of Koffka for his failure to give adequate definition to his
basic concepts. Let me attack his system in this way and let the
reader estimate my success.
The span of Koffka's systematic structure rests upon two pedes-
tals. Let us call them organization and meaning. Organization is the
Gestalt principle that a large dynamic system is more than the sum
of its parts and that such systems are the rule among psychological
phenomena. This principle is the great contribution of Gestalt psy-
chology and I think that no one—certainly no one who has read this
book—will any longer attempt to displace it. Meaning is the principle
that a psychological object has signification (significance, implication,
intentionality) as part of its essence. This principle is not an accepted
fact and dispute about it is bound to arise. In such cases. I think, we
must demand rigid definition.
Here we must pause to ask what a definition is and why we have
a right to insist upon one in such a case. In the old days scientific
disputants could go back to immediate experience. The theory then
was that experience cannot lie, that there must always be agreement
if factual dubiety can be reduced to the basic givens. That was posi-
tivism, which Koffka deplores and I can do without. However, we
have now a more modern panacea for disagreement. Disputation
nearly always arises because some extra attribute or property has got
itself introduced into the ' reals' in the process of reification or con-
ceptualization. The technique by which understanding and agreement
are got lies in defining the things and concepts in terms of the scien-
tific operations by which they are known. I think that Koffka does
not accept this operational procedure. He said of Tolman in 1933:
" I do not agree with him that these criteria can at the same time
serve as definitions, provided we are not dealing with mere fictions "
(Psychol. Bull., 1933, 30, 443, q.v.). It is plain to me from Koffka's
book that he believes firmly in the immediate intuition of scientific
64 EDWIN G. BORING
things and that there lies the reason why he thinks that operational
definitions apply only to ' fictitious' concepts. However, to me it
seems that this very book is the best possible argument against the
validity of such intuition, for with all his care and pains Koffka is still
unable to impart to me many of his intuitions about psychological
things. It is thus that I find my sanction for bringing what is essen-
tially external criticism to bear against him, by asking for definition
and insisting that I know no safe definition in science that is not
operational. If Koffka knows of another way out of this dilemma of
disagreement, the burden lies upon him to show it. I think there is
none.
Let us now return to the problem of the meaning which is intrinsic
in every " behavioral object", and do our best to understand what
Koffka says about it. Let us take the case of the continuity of the
ground behind the figure (cf. pp. 178-181). Can we see the table
beneath the book which lies upon it, or does the book interrupt the
table? Koffka says that we can see the table as continuous. It is a
case of " double representation " in the sense that we see two things
at different distances in the same part of the bidimensional field. The
book is, as it were, transparent, and the table is seen through it. It is
true, he adds, that the table at this region is colorless, but then so is
the phi-phenomenon in perceived movement. As soon as we get over
thinking that we cannot see a thing unless it is black, gray, white
or some hue, we shall have no trouble in perceiving the table beneath
the book.
Koffka admits the existence of alternative points of view in this
case. He imagines counsel confounding a witness who has sworn
that he saw the table through the book, and notes that " what the
counsel means by seeing is not the same as what we mean by i t "
(p. 180). So that is the point. What does Koffka mean by seeing?
Not what Titchener meant. Not sensory phenomenon characterized
by quality, extension, and the other traditional attributes. He means
that the witness knows that the table is there, and that knowing and
perceiving are not different. Knowing is in such a case not a conscious
or unconscious inference; plenty of evidence can be marshalled to show
that the knowing is too immediate for that. We simply perceive the
table as continuous.
Now there is no use in this case of going back to the Titchenerian
alternative. Koffka admits it as a possible point of view, but, although
it may still have uses in psychology, I do not want to champion the
concept of immediate experience in its traditional sense. Moreover,
 SPECIAL REVIEW 65
Koffka has put us in the presence of a fact: we do know at once that
the table is behind the book. He asks us to espouse a psychology that
can take account of such a fact. What are we to do ?
Obviously we must go with him. A fact is a fact. However, I
think, we must also go far beyond him. We have here as a behavioral
object a table, which is continuous, even though in the ' real' table
some psychologist may have cut a hole beneath the book. We simply
must have a definition of a behavioral object.
It seems pretty clear that the meaning which belongs to a behav-
ioral object is some sort of knowledge. However, to identify knowl-
edge with meaning is not to define either, for knowledge is a notori-
ously equivocal concept. Knowledge may be actual in the sense that
it determines some present functional relation. Or it may be potential
in that it would enter into some relation if other terms were given.
It may be conscious in the sense that it is an object of observation
to the person who has it. Or it may be unconscious in the sense that
it is an object of observation only to some other person or is the
result of scientific inference. It is in the face of such equivocality that
a definition may rightfully be demanded, and I insist that there is
nothing for us to do but to go to the operations by way of which such
' knowledge ' becomes known.
Scientific insight, of course, comes first from instances, although
we have to transcend the instances to gain a definition. If we want
to know what exactly is a behavioral object, we must inquire con-
cerning the nature of the evidence that the object exists in a given
instance {e.g. for the table as ground) and in all other instances.
How do we know that a person sees the table as continuous ? Because
he acts as if it were continuous. How do we know that the man sees
the snow-covered frozen lake as a snow-covered field? Because he
acts as if it were a field. How do I know that I saw the lake as a
field? Because I acted in movement and in emotion as if it were a
field. How do I know that this red and this green look alike to you
who are color-blind ? Because you say they are alike. We define the
object by its effects.
As we have seen, Koffka believes in the immediate intuition (my
phrase; he has none) of a behavioral object. That is, however, a
retreat to immediate experience as the final common ground. We
ought not to be offered this technique by the chief opponents of
those who used it (from Wundt to Titchener) and who found that
they could not compel agreement by it. All through this book I have
been puzzled by what seems to me to be a persistent contradiction:
66 EDWIN G. BORING
Koffka's explicit avoidance of immediate experience and his implicit
dependence upon it as the ultimate. Has he not perhaps repressed
immediate experience because he hates positivism, therefore admitting
it only in disguise?
At this point of the argument Koffka seems to me to have a
choice. He may come out frankly and embrace immediate experience
as the ultimate court of appeal. Then I say to him that I am not satis-
fied. The verdict of immediate experience has never been clear.
Koffka remembers Wiirzburg. The unanschauliche Bewusstheit
came to the world out of immediate experience. Is it in his book?
On the other hand, he ma)' repudiate the appeal to immediate experi-
ence and what he said to Tolman, perhaps too rashly, in 1933. He
may accept our argument and wish to go along with us. Where shall
we take him ?
It would be kind if we could keep him in the land of undefined
behavioral objects, but I see no way to do it. If the operational path
is the only one open to us, then we shall find ourselves soon concerned
with that behavior of organisms which indicates the existence of the
behavioral object, and which provides the defining criteria of it. It
does not bother me that the behavioral object may thus appear to him
to be a "fiction". Science is the great cultural artifact and its ' reals '
are fictions. Koffka himself is at pains to show that a thing is made
and not given. This matter seems to me to be so commonly under-
stood that I shall not stop to argue the point. We must take the
behavioral object as defined by its criteria, else it is meaningless.
However, the criteria are the data of the observation—of the
experiment when the observation is controlled. Such things exist
for the observer and the experimenter and they belong to the " geo-
graphical " environment. Thus from Koffka's esoteric world of
psychological things we emerge upon the familiar ground of the real
world, and Koffka's special ' behavioralism' becomes a common
behaviorism. Most persons will say that in it we deal with real,
common-sense objects, but—whether they are right or whether they
are wrong—our criteria are identical in nature with the criteria that
the physicist uses to define his concepts. We are no longer in a special
world, and we have mastered the mystery of meaning by requiring a
definition of it.
Thus I am ready to go all the way with Koffka if he will go all
the way with me. No one knows so well as he how changed his
system will be if he accepts my invitation.
It has been worth our while to argue the case of meaning at
 SPECIAL REVIEW 67
length, first, because meaning is a corner-stone of this systematic
structure and insecurity there affects the entire edifice, and, secondly,
because the argument serves as the type of criticism that can be-
directed against other concepts. Let us consider briefly three other
concepts.
The first of these concepts is the demand character. What is a
demand character? Since I cannot find a definition in Koffka, I
venture this on his behalf: A demand character is that property of a
field object by virtue of which it attracts or repels an individual. I
think that this is exactly what Koffka means, and I note with pleasure
that the definition is operational and that the demand character is-
just as much of a fiction as other scientific ' reals '. However, I think
that Koffka has betrayed his own genius in attempting to localize the
demand character in the object. He discusses the question as to
whether the demand character in the object is identical with the need
in the individual (pp. 345f., 355ff.), and he localizes the force in the
object, because objects have certain properties, like movement, which
compel attention. However, I think that, if Koffka had undertaken
to define a demand character, to establish its criteria, he would have
discovered that what he is considering is a property of the larger
dynamic system that includes both the field object and the Ego. That
is the lesson of Gestalt psychology, that properties can apply to sys-
tems, and we have here a property of a dynamic relation between a
field object and the individual. That Koffka fails to eat the bread
of his own baking is due, I suggest, to the fact that he is not yet
weaned from immediate experience.
(Here I must digress briefly to say that operational definition is
a process that has no logical end, but an actual end in scientific agree-
ment. If Koffka says of my definition of the demand character that
the word attract requires definition, I should cheerfully comply. A
call for definition is ipso facto a need for it. But this regress is not in
practice infinite. It stops in the face of agreement, and the scientist
never is confronted with the problem of defining the meaning of every
word without the use of words.)
The most important concept for the intellectual and memorial-
processes is the trace. In this case the operational definition is clear,
because the trace is defined by the events of the exact experiments on
successive comparison. The concept of trace can therefore readily be
made to meet my requirements, although, of course, Koffka is not
trying thus to define it. Instead he appeals to isomorphism—the only
place in the entire book where he makes serious use of this concept
68 EDWIN G. BORING
which he introduced in Chapter 2—and he localizes the trace in the
brain. Why should he do this since he can offer no operations to
prove that the trace is neural? There is no more reason for his
appealing to isomorphism here than anywhere else. His vague specu-
lations about the neural basis of the trace do not help the argument,
and the trace is quite self-sustaining without the brain. He has placed
a weak neural prop against a sturdy conceptual structure. Again I
think the cause must lie in his need to reify the truly fictional trace
as he reified the demand character.
The concept of communication between traces is not so happy as
the concept of trace. It is the crucial concept in the discussion of
learning, forgetting, recognizing, and thinking. Traces communicate
when they affect each other, entering into a reorganization. Some-
times they do not communicate. If we could say when they communi-
cate and when they do not, we should probably have the data for an
operational definition. Koffka's appeal to similarity as the cause of
communication (pp. 598ff.) will not suffice, for, of all the concepts
which the psychologist uses, similarity is probably the one most in
need of operational definition and least capable of entering into a
convincing argument when it lacks such definition. The consequence
is that most of this latter discussion in the book seems to me very
little more than an elaborate and sophisticated juggling of words. I
cannot pin the argument down to empirical fact, not even by way of
Harrower's excellent experiments. We need, I think, to know some-
thing about an invariant of communication before we know what
communication is.
My notes call for the discussion of the concepts molar, iso-
morphism, unit, shape, anisotropy, Ego, and similarity. The list
could be trebled and still include no trivial concept. However, space
and the reader's patience must be growing short together. A thorough
review of so thorough a book is impossible; and the reviewer must be
content to choose this and that according to his insight of the moment,
say his say, and stop.
What my criticism comes to is this. I think that Koffka is on
one of the two best roads toward psychology's goal. The other road
leads the physiological psychologists through the " geographical
environment ". Koffka's path through the " behavioral environment"
has its culs-de-sac and I have shown where I think some of them
are. He needs, I have argued, not to go back, but to go farther; and
I have ventured the prediction that, if he keeps on under the drive
of his superb determination to be rigorous, he will walk right out of
 SPECIAL REVIEW 69
the behavioral environment into the region where there are no
a priori meanings and all scientists travel together, amply sustained
by their fabricated conceptions.
Thus it appears that Gestalt psychology, as Koffka presents it,
must be peculiarly transitory because we can already see beyond it.
The ultimate test of a system must be its univocality. The pragmatic
test, upon which, I suspect, Koffka relies, will not serve, for any
self-consistent system seems to work when one gets habituated to its
use, until an outsider does to it what the Gestalt psychologists have
done to the Wundtians. A system whose terms can, on call, be rigidly
defined has the most certain chance of a long life, and in this respect
Gestalt psychology has great need to alter its complexion.