Memory Illusions

JOURNAL OF MEMORY AND LANGUAGE 35, 76–100 (1996)
ARTICLE NO. 0005
Memory Illusions
HENRY L. ROEDIGER III

Rice University
Memory illusions may be defined as cases in which a rememberer’s report of a past event seriously
deviates from the event’s actual occurrence. This article introduces the special issue of the Journal
of Memory and Language that is devoted to memory illusions by grounding their study in the context
of perceptual illusions. Perceptual illusions have been investigated since the 1850s, whereas memory
illusions have been systematically investigated only since the late 1960s or early 1970s (despite some
pioneering research and writing before this time). I suggest possible reasons for this discrepancy in
research activity, sketch a brief history of the study of memory illusions, and then consider the variety
of memory illusions that are studied in contemporary psychology. The papers composing the special
issue are introduced during this brief cataloging of memory illusions. Related areas of research are
discussed in the concluding remarks. q 1996 Academic Press, Inc.
Illusions have fascinated philosophers, recall of a past experience.’’ Even though both
physicists, and psychologists throughout re- perceptual and memory illusions were men-
corded history. Whenever our experience of tioned in this definition, by 1964 there had
events is shown to be at odds with the actual been hundreds of studies of perceptual illu-
nature of the events, our curiosity is piqued. sions but, even counting generously, there
The word illusion is derived from the Latin probably existed fewer than 30 publications
word illudere, meaning to mock. Illusions directed specifically at memory errors, with
mock our belief that what we perceive, re- the term memory illusion rarely arising at all.
member, and know is in perfect accord with One exception occurred in Titchener’s (1928)
the state of the external world. Textbook of Psychology, where he devoted a
Despite long-standing interest in this topic, section to ‘‘Illusions of Recognition and
defining exactly what an illusion is can be a Memory.’’ He wrote that ‘‘Illusory memories
difficult exercise. When psychologists see the and recognitions are of two kinds. We may
word illusion, it is usually modified by visual, remember or recognize something that is re-
perceptual, or optical. The term memory illu- ally, objectively, unfamiliar to us, and we may
sion is much less common. In a 1964 diction- fail to recognize or remember something
ary of psychology, Drever included the fol- which once formed part of our common expe-
lowing entry under Illusion. ‘‘In the case of rience. Both types of illusion are quite com-
sense perception, ‘a subjective perversion of mon’’ (p. 424). Titchener discussed these mat-
the objective content,’ or actual sense data; in ters for only one and a half pages and alluded
the case of memory a subjective falsification to the existence of other memory illusions but
by addition, omission, or substitution, in the did not describe them. The two illusions to
which Titchener referred are usually called by
their French names, déjà vu and jamais vu,
Preparation of this article was supported by Grant and have been little studied experimentally,
F49620-92-J-0437 from the Air Force Office of Scientific
Research. The article benefited from comments by Marcia
until lately when experimental analogs have
Johnson, Doug Nelson, Kathleen McDermott, Kerry Rob- been developed (as discussed below).1
inson, Dan Schacter, and Endel Tulving. I thank Kerry
1
Robinson for help in preparing the manuscript. Corre- It is clear from Titchener’s examples that he did not
spondence should be addressed to the author at Depart- mean that forgetting should be considered a memory illu-
ment of Psychology, MS 25, Rice University, Houston, sion. Rather, the failure to recognize something quite fa-
TX 77005-1892. E-mail: roddy@rice.edu. miliar (such as a well-known word) was intended.
0749-596X/96 $18.00 76
Copyright q 1996 by Academic Press, Inc.
All rights of reproduction in any form reserved.
AID JML 2467 / a002$$$101 04-10-96 17:23:17 jmla AP: JML

MEMORY ILLUSIONS 77
torical overview of their discovery and impor-

tance to the study of perception. Due to the
longer history of serious scientific inquiry, a
consideration of the topic of perceptual illu-
sions seems an appropriate way to ground the
study of memory illusions. I then turn to the
topic of memory illusions and survey some of
the major milestones in their study, as well. I
provide a brief catalog of some of the main
types of memory illusions that have been stud-
FIG. 1. A shape illusion. The shaded tops of the two ied, and in the process I introduce the set of
figures appear quite different in shape, but are actually papers that constitute the special issue.
congruent. Reprinted by permission of Roger Shepard and
Lawrence Erlbaum, Inc. PERCEPTUAL ILLUSIONS
Illusions occur in all sense modalities, al-
We may ask why, historically, this imbal- though those occurring in vision have been
ance has existed between interest in perceptual studied in greatest detail. The ancient Greeks
illusions and in memory illusions. One answer were quite aware of visual illusions. The Par-
probably lies in the immediacy of perceptual thenon was constructed to overcome two com-
illusions relative to memory illusions. Shown mon visual illusions that would have caused it
in Fig. 1 is the remarkable shape illusion that to look imperfect had steps not been taken in
Shepard (1981) created. The darkened parts its design and construction to counteract their
(the tops of the figures) appear quite different effects (see Coren & Girgus, 1978). Aristotle
in shape, but they are actually congruent. described the moon illusion—the fact that it
(Trace one on a sheet of paper and lay it over looks larger at the horizon than when over-
the other, if you need to be convinced.) Such head—and also described the illusion of touch
a remarkable demonstration provides immedi- that still bears his name (Benedetti, 1985).
ate and convincing evidence of a fascinating Closer to our own times, scientific interest
process at work, one worthy of serious and in perceptual illusions began in the early
thoughtful study. Most visual illusions have 1800s and had exploded by the turn of the
just this quality. Even if memory illusions oc- century. In 1832 Louis Albert Necker, a crys-
cur routinely in life—and I think they do— tallographer, noted a curious effect he had dis-
the evidence for their occurrence is quite dif- covered in gazing at crystals: Line drawings of
ferent. Usually the events in question have crystals would spontaneously reverse in depth
long since disappeared from life’s stage and when an observer stared at them. This obser-
we can believe that our memories are right, vation gave rise to the familiar Necker cube.
despite claims to the contrary that may be pro- In 1854, J. J. Oppel reported the curious obser-
vided by others. If we simply look at the two vation that a certain distance of linear extent
shapes in Fig. 1, we are not aware of the illu- appeared longer if it were filled (say, with
sion at all—one shape is simply perceived as lines) than if it were unfilled. Coren and Gir-
quite different from the other. It is only if gus (1978) estimated that over 200 scientific
we are told to check that we can convince papers appeared on perceptual illusions in the
ourselves of the difference. This critical 50 years after publication of Oppel’s (1854)
checking operation usually is not possible for paper. Many illusions identified during this
remembered events. period still bear the names of their discoverers:
The purpose of this essay is to introduce Zöllner, Müller-Lyer, Ponzo, Ebbinghaus, and
the special issue of the Journal on the topic so on. One important theoretical contribution
of memory illusions. First I consider the topic from this era came from Hermann von Helm-
of perceptual illusions, providing a brief his- holtz, who offered a general approach to per-

78 HENRY L. ROEDIGER III
ception still found in many theories today. As holtz’s theme in this quote reverberates to the
Coren and Girgus (1978, page 9) put it, ‘‘He present day through the writings of scholars
proposed that, in general, perception in the studying perception, but it is not a uniform
adult observer is based on unconscious infer- view. Kolers (1964, p. 98) noted that ‘‘Illu-
ences that the mind makes about the pattern sions have not always had a good standing
of stimulation in the nervous system. Since among the investigators of perception. When
most stimulus arrays are ambiguous, the ob- perceptual illusions were introduced as a topic
server must interpret the sensory stimulation of study in the nineteenth century, the prevail-
arriving in the brain in light of his knowledge ing attitude about them was that they were
of the environment. Thus perception is an in- mere parlor tricks—minor imperfections or
ductive process wherein the observer uses his errors’’ in the workings of perception. (Kolers
experience to interpret the patterns of excita- went on to discredit this view.)
tion in his receptors.’’ Helmholtz called this The study of illusions has occupied center
constructive process unconscious inference stage in the study of perception since the turn
because when people create their perceptual of the century and has even given rise to entire
world, they are unconsciously weaving to- theories that developed largely to explain their
gether data from the senses with prior knowl- occurrence (e.g., the transactional approach to
edge about the world. A classic example of perception proposed in the late 1940s and
unconscious inference occurs in depth percep- early 1950s by Ames, Ittelson, and Kilpatrick
tion, wherein the two-dimensional display on and based on the remarkable demonstrations
the retina gives rise to the experience of a they produced (e.g., Ittelson, 1952; Ittelson &
three-dimensional world through the use of Kilpatrick, 1951). In 1978 Coren and Girgus
many cues that are combined without our estimated that over 1000 articles had been
awareness. The role of the active perceiver published on the topic of perceptual illusions.
appears throughout the history of the study of Now, 18 years later, hundreds more have ap-
perception (Neisser, 1968). Rather than the peared and whole books are devoted to special
external world merely being projected onto illusions, such as the edited volume by Hers-
the retina and faithfully transmitted to the henson (1989) on the moon illusion.
brain, most perceptual theorists have empha- In the next section I turn to considering
sized that top-down, constructive processes why illusions and errors have not, until lately,
help determine what we perceive. In The played a central role in research and theory
Principles of Psychology, William James about human memory. Before leaving the
described his general law of perception, topic of perception, however, consider one
which he believed was well substantiated: more quote noting the importance of studying
‘‘. . .whilst part of what we perceive comes how systems fail in helping to understand how
through our senses from the object before us, they normally operate. Gregory and Gombrich
another (and it may be the larger part) always (1973) wrote that ‘‘Illusions are also tools for
comes . . . from our own head’’ (1890, Vol. discovering processes in perception. In medi-
II, p. 108). cine, in engineering, and very frequently in
The topic of illusions plays a large role in biology, the abnormal and surprising lead to
these constructive views of perception (e.g., key ideas and facts for understanding the nor-
Gregory, 1970). Helmholtz (1881) wrote, mal. So here we may expect abnormal percep-
‘‘The study of what are called illusions of the tions (deviations from truth) to give insights
sense is, however, a very prominent part of and data for understanding normal (correct)
the psychology of the senses; for it is just perception’’ (p. 7). We may hope that studying
those cases which are not in accordance with illusion and error can play the same role in
reality which are particularly instructive for advancing our understanding of human mem-
discovering the laws of those processes by ory. The study of perceptual illusions may
which normal perception originates.’’ Helm- need to become an integral part of the study

MEMORY ILLUSIONS 79
of memory illusions, as they both reveal con- this sort of procedure is that the investigator
structive errors in the overall cognitive sys- typically has no reliable way to check the ac-
tem. Later I consider some illusions that could curacy of a reported recollection’’ (p. 5). We
be classified as either perceptual or memory may add that this difficulty—lack of control
illusions, illustrating that no firm boundaries over the original events—plagues many mod-
exist between these phenomena. ern treatments of the question of accuracy and
inaccuracy in memory. The experimentalist’s
MEMORY ILLUSIONS credo is that unless one can be sure as to the
Perceptual illusions reveal the constructive course of original events, one cannot make
nature of perception: We can misperceive a strong claims and statements about the accu-
stimulus while it appears before us under lei- racy or inaccuracy of later recollections.
surely and optimal viewing conditions, as in There is a sense in which the study of mem-
Fig. 1. If perceiving is conceived as an active ory errors is quite old. After all, every time
process, one full of inference and hence at subjects provide intrusions in recall tasks or
least occasional error, then it is hardly surpris- make false alarms in recognition tests, these
ing that remembering would be so considered, errors could be interpreted as memory illu-
as well (Neisser, 1967, 1968). After all, if the sions. However, in the history of experimental
cognitive system can err in misrepresenting studies of memory, errors were rarely of inter-
objects when they are present before the eyes, est in their own right. The emphasis has tradi-
the opportunities for error when a person later tionally been on correct responding and errors
tries to recreate happenings of the past must were typically considered a methodological
be even greater. nuisance, reflecting guessing or criterion
The origins of formal memory research in shifts. Errors were most commonly used to
Ebbinghaus’s (1885/1964) great experiments correct accurate recall or recognition for
probably helped keep early researchers from guessing and other biases, as in signal detec-
examining errors and illusions. Ebbinghaus tion theory. Exceptions to this general meth-
learned and relearned his series of nonsense odological attitude concerning errors are de-
syllables and measured the savings during re- scribed below, but it should be borne in mind
learning. The relearning and savings method that (at least until the 1970s), mainstream ex-
was ingenious and set the field off on a scien- perimental psychologists were not usually in-
tific and sure path, but like any method, it was terested in illusions of remembering that
better adapted to asking some questions than might be reflected in errors.
others. The method was generally unsuitable The first experiments showing the inaccu-
for investigating memory errors and illusions, racy of memory were conducted with children
despite the fact that Ebbinghaus had a keen by Binet (1900) in France and by Stern (1910)
interest in the topic of perceptual illusions. in Germany (see Ceci & Bruck, 1993). Both
Nonetheless, one of Ebbinghaus’s method- investigators exposed children to objects or
ological contributions is critically important events and later tested their memories with a
even for the study of memory illusions. As series of misleading questions. The results of
Schacter (1995) noted in this context, both studies revealed memory distortions in
‘‘. . .Ebbinghaus introduced a methodologi- the children. At about this same time, Mun-
cal innovation that has proved to be essential sterberg (1908) reviewed evidence of the un-
to evaluating whether a memory is true or reliability of eyewitness testimony in his book,
false: he gained control over the input to the On the Witness Stand.
memory system. In earlier philosophical and In a totally different tradition, Freud (1895)
clinical treatises on memory, discussion cen- began exploring memory distortions in the
tered on introspective recollections of past ex- 1890s and continued to revise his statements
periences by the writers themselves or by pa- on the subject throughout his life. His most
tients whom they observed. The problem with famous idea was that of repression—the no-

tion that painful early memories are banished ences of a variety of psychical forces. Thus
to an unconscious state where they may be the ‘childhood memories’ of individuals come
revealed only indirectly, through dreams, in general to acquire the significance of
slips, and symptoms of psychopathology. Er- ‘screen memories’ and in so doing offer a re-
delyi (1985, Chapter 5) describes the many markable analogy with the childhood memo-
changes in Freud’s concept of repression, ries that a nation preserves in its store of leg-
from people simply not wanting to think about ends and myths.’’ However, without the meth-
bad events of their lives at one extreme (Erde- odological stricture of the experimental
lyi & Goldberg, 1979) to a deep and mysteri- school—knowledge of the original events and
ous process of psychological banishment of how they occurred—Freud’s various discus-
memories to an unconscious state at the other. sions of whether childhood memories are ac-
One corollary of the idea that memories of curate or inaccurate is based more on conjec-
threatening experiences could be removed or ture than on empirical evidence.
banished to an unconscious state is that they These two streams of study of memory illu-
could somehow reside in this state relatively sions I have briefly described—one based on
intact and then be recalled (through therapy) experimental methods and the other rooted in
at some later time in life. However, in his later the psychoanalytic tradition—have guided
writings Freud abandoned his earlier belief discussion of memory errors to the present
that childhood memories obtained during ther- day. However, as previously noted, within ac-
apy should be treated as genuine and empha- ademic psychology the main issues in memory
sized instead the processes of distortion that research lay elsewhere. Researchers from
could occur during therapy. In 1910 he wrote early in the century worried about the course
that childhood memories are ‘‘. . .only elic- of forgetting and its causes (retroactive inter-
ited at a later age when childhood is already ference and retroactive inhibition), but rarely
past; in the process they are altered and falsi- worried about error and illusion. There are two
fied, and are put in the service of later trends, prominent exceptions to this generalization,
so that generally speaking they cannot be dis- however. The first is represented in the Gestalt
tinguished from phantasies’’ (p. 83). tradition of memory research in which it was
Two other types of distortions that Freud believed that memories changed over time in
(1903) analyzed were paramnesias (or false the directions that were in line with the Gestalt
recollections of forgotten events that took on laws of organization. Briefly, people would
subjective certainty), and screen memories (or remember events in a more organized fashion
cases when one memory substituted for an- (the events would represent good patterns)
other and screened it out of consciousness). when they were remembered at increasing in-
Freud interpreted both in terms of defense tervals from the original event. For example,
mechanisms to protect against psychic threat Wulf (1922) reported that visual forms were
and analyzed several cases that seemed to fit remembered as being more regular and sym-
this mold. The difficulty with interpretation of metrical over time, and many other research-
screen memories, in particular, is that there is ers conducted similar studies. The topic re-
no rule to decide which may be the real mem- mained of sporadic interest in experimental
ory and which the screen memory, except for psychology, but Riley’s (1964) piercing re-
ad hoc supposition (Schacter, 1995). Nonethe- view of the methodological difficulties and the
less, Freud (1903, p. 68) likened memories of inconsistencies in results in this area quelled
early childhood to screen memories: ‘‘One is continued interest. Most commentators have
thus forced by various considerations to sus- concluded more recently that nothing much
pect that in the so-called earliest childhood was gained from this study (Baddeley, 1976;
memories we possess not the genuine mem- Schacter, 1995). However, as Loftus (1979b)
ory-trace but a later revision of it, a revision pointed out, Riley’s (1964) analysis of re-
which may have been subjected to the influ- search from the Gestalt tradition did empha-

MEMORY ILLUSIONS 81
size the effect of verbalization on memory. while violating most other strictures that care-
The words that people used to label visual ful experimentalists might prefer (Kintsch,
figures, for example, determined how they 1995). His data were in the form of sample
were later remembered (e.g., Carmichael, Ho- protocols produced by some of his subjects
gan & Walters, 1932). The importance of ver- and showed numerous omissions, with the sto-
balization on memory continues to be studied ries becoming shorter over time. More inter-
to the present day, as represented in the current estingly, the subjects made errors of commis-
issue in Melcher and Schooler’s experiments sion in which they added material to make the
on verbal overshadowing. story more rational and consistent. Supernatu-
The second, and more important, contribu- ral elements dropped out and in general the
tion during the 1930s to the study of memory students seemed to convert the story more to
distortions was publication of Bartlett’s the form of an English fairytale, sometimes
(1932) great book, Remembering: A Study in even tacking on a moral. Bartlett noted that
Experimental and Social Psychology. Bartlett this property of rationalization in the protocols
took it as axiomatic that remembering was brought the story in line with schemas with
imperfect and even argued that it would be which the students were more familiar.
unnatural for people to remember events ver- A few years before his death, Donald
batim. Before turning to remembering, he de- Broadbent (Bartlett’s student) related to me an
voted a chapter to discussing constructive pro- interesting story about rationalization in ‘‘The
cesses in perceiving. In one demonstration, he War of the Ghosts.’’ Briefly, it did not start
showed people a painting for a brief period of with Bartlett’s subjects, but with Bartlett him-
time and noted that their reports of what it self, when he created the materials. One of
contained were quite different and seemed to the most memorable parts of ‘‘The War of the
depend more on the background and proclivit- Ghosts’’ occurs at the end, when an Indian
ies of the observers than on the content of falls down, something black comes from his
the painting. Further, when he exposed the mouth, and he dies. Bartlett’s version put it
painting repeatedly, but with quite brief expo- this way: ‘‘Something black came out of his
sures, his subjects would often cling to their mouth.’’ Broadbent told me that Bartlett had
earlier interpretations and fail to perceive the revised this part of the story, omitting an of-
actual scenes portrayed, a finding presaging fensive phrase. Sure enough, in checking
that of Bruner and Potter (1964) years later. Franz Boas’ (1901) original version, the sen-
Bartlett argued that both perceiving and re- tence read ‘‘Something black came out of his
membering were constructive processes mouth and blood came out of his anus.’’
guided by schemas, or cognitive frameworks Broadbent opined that this version was too
that people brought to the events and through vivid for Bartlett to present to his refined Cam-
which they interpreted them. bridge students! In fairness, Bartlett seems to
Bartlett’s most famous experiments were have rewritten the entire story to make it flow
those in which college students read and later better than Boas’ original literal translation,
recollected an Indian folktale, ‘‘The War of but this is the only case where an important
the Ghosts.’’ In one type of experiment, which change of content occurred.
he called repeated reproduction, subjects re- Bartlett’s (1932) book, so important today
called the story from memory on at least two in hindsight, actually seems to have had little
occasions, without intervening study. Typi- impact on the questions experimental psychol-
cally, Bartlett required a first recall 15 min ogists asked at the time, as reflected in the
after his subjects (Cambridge University stu- mainstream literature of the 1930s, 1940s, and
dents) had read the story twice. Then he tested 1950s. However, as noted below, the book
them at various later points in time. His exper- received renewed attention in the late 1960s
imental methods were casual, to say the least, and early 1970s when Bartlett’s approach to
but he did at least control the input material remembering gained renewed interest and fa-

vor. The legend of Bartlett’s repeated repro- ‘‘There are probably many ways to do this
duction experiments lives on today in most recoding, but probably the simplest is to group
textbooks of psychology (introductory, cogni- the input events, apply a new name to the
tive, and memory texts alike). However, to group, and then to remember the new name
my knowledge no one has successfully repli- rather than the input event’’ (1956b, page 93).
cated his repeated reproduction experiments Recoding is an active process of transforma-
on materials like ‘‘The War of the Ghosts’’ tion and therefore the possibility of errors in
when they used instructional sets requesting recoding and later decoding may arise. Miller
that people remember the story. Gauld and (1956a) wrote:
Stephenson (1967) found errors like Bartlett’s
when they told people to make up a story like A . . . more general technique for organizing our
the one they had heard, but not when they experience into convenient units is provided by
asked them to remember the story. Wheeler language. . . When you witness a scene or hear a
story that you want to remember, you try to trans-
and Roediger (1992) gave subjects repeated late it ’into your own words,’ into the linguistic
tests on both ‘‘The War of the Ghosts’’ and units that will fit into your own cognitive hierar-
on another prose passage and found that they chy. This highly schematic, verbalized abbrevia-
actually recalled more on the second test than tion is remembered. Then when you try to recall
on the first, at least with short delays between you must decode. Since the fit of words to experi-
ence is seldom as tight as the fit of laws to data,
tests. It seems odd that experimental observa- the decoding process often goes astray. You supply
tions could live on for so long in textbooks details by secondary elaboration that are consistent
when the original methods were so casual and with your coded memory. Often these details are
when no one has successfully replicated the wrong. (p. 132)
work.2 The probable reason lies in the fact
that most observers thought Bartlett’s (1932) Miller (1956a, 1956b) paved the way to the
basic story about memory rang true, regardless information processing approach to remem-
of his methods and results, and if so, that as- bering, which has largely guided research for
sessment still seems accurate today. the past 40 years. As Koriat and Goldsmith
(in press) have noted, research conducted in
THE MODERN ERA this tradition has largely been devoted to is-
The information processing approach to sues of how much (how many units) could be
cognition began in the 1950s as researchers remembered, and errors have been given
carried concepts and methods from communi- rather short shrift. However, nothing within
cations theory and electrical engineering into the information processing approach itself or-
psychology. In the 1950s Donald Broadbent dained this emphasis, as the quotes from
(1958) produced the first information flow dia- Miller indicate; and some researchers did at-
grams and George Miller (1956a, 1956b) pro- tend to error analyses. For example, Conrad
duced two papers that began a revolution in (1964) studied errors of short-term memory
the study of human memory. One of Miller’s and concluded that they had a phonological
key ideas was the concept of recoding, basis, whereas other researchers found seman-
wherein information is received from the out- tic confusion errors in studies of long-term
side world in one form but is transformed and memory (e.g., Underwood, 1965). However,
represented in a different way internally. most researchers influenced by Miller (1956a,
1956b) studied how material was organized
so as to be remembered better. For example,
2
Bartlett’s (1932) serial reproduction experiments, in Tulving (1962, 1964) studied how subjects or-
which one person reads and recalls a passage, the second ganized random word lists into subjective
person reads the first person’s recollections and then re-
calls the information, etc., also induce systematic distor- units in multitrial free recall so as better to
tions. These experiments have been successfully repli- remember the words. Glanzer and Clark
cated (e.g., Paul, 1959). (1963) developed the verbal loop hypothesis,

MEMORY ILLUSIONS 83
seeking to show that when people are faced hypothesis of remembering (pages 284–286).
with visual information, they recode it into Perceiving and remembering were seen to be
a verbal description and that this description amenable to similar interpretation: ‘‘The anal-
guides later recall. ogy being offered asserts only that the role
Many different trends in the modern era which stored information plays in recall is like
launched psychologists on the road to study- the role which stimulus information plays in
ing errors and distortions of memory. I review perception’’ (page 285). In neither case can
some of the main trends in the field and, while the information to be used directly enter con-
doing so, show how articles in the current sciousness or be reproduced in behavior in
issue fit in. a literal manner; rather, constructive activity
(top-down processing) operates on the infor-
Neisser’s Cognitive Psychology mation and the perceiver and rememberer be-
In 1967 Ulric Neisser published his great come conscious only of the product of this
book, which gave an entire field its name and constructive activity. Therefore, ‘‘. . .stored
set many of the topics that were to be the information is not revived, but simply used,
center of attention in the new field: iconic in the constructive activity of recall’’ (page
memory, echoic memory, reading, imagining, 289). Just as constructive activity can give rise
and so on. Traditional memory research in the to illusions in perceiving, the same is true in
Ebbinghaus tradition was not much in evi- remembering.
dence, but more importantly for present pur- Although it is not possible to pinpoint a
poses, Neisser capped his book with a chapter cause and effect relationship with certainty, it
on remembering and thinking that harked back seems likely that Neisser’s famous book, and
to Bartlett’s (1932) idea of remembering as a especially its final chapter, may have helped
constructive process. Neisser emphasized the launch the renaissance of experiments in the
unity of cognitive processes and noted that 1970s that examined errors in prose retention
perceiving, remembering, and thinking were in the Bartlett (1932) tradition.
all active and interdependent processes. It
seems obvious that remembering depends on Memory for Prose: The Influence
perceiving, but less so that higher cognitive of Schemas
processes usually studied as issues of problem
solving and thinking are also involved in re- In the early 1970s several groups of re-
membering, at least to the extent that remem- searchers began studying errors in prose reten-
bering is conceived as a constructive activity. tion that implicated the active role of the re-
Neisser’s (1967) assumptions were stated memberer in bringing his or her knowledge
clearly: to bear in recoding information. Bransford and
Franks (1971) presented subjects sentences of
It is assumed that remembering and thinking are varying complexity that created a coherent
analogous to adaptive movement and to motor idea. Later they tested subjects’ memories for
skill; they also resemble the synthetic processes
sentences that had been presented, ones that
of visual memory and speech perception. Stored
information consists of traces of earlier construc- had not been presented but were consistent
tive acts, organized in ways that correspond to the with the complex idea, and ones that were
structure of those acts. However, the ’traces’ are unrelated to the idea. Subjects could easily
not dormant copies of earlier experiences, some- reject the unrelated sentences, but they failed
how aroused into consciousness from time to time.
to distinguish between presented and not pre-
Stored information is never aroused, it is only
used, just as stimulus information is used in the sented sentences so long as they were consis-
act of perception. (page 279) tent with the overall idea. Subjects’ confidence
in recognition was directly related to the num-
Neisser (1967) developed the idea in the ber of propositions in the test sentence that
last sentence just quoted into the utilization were consistent with the overall idea, regard-

less of whether they had actually been pre- nition tests composed in part of lures that bear
sented in the study phase. various semantic relations to the studied
In a later study, Bransford, Barclay, and words. Underwood (1965) gave subjects a
Franks (1972) showed that people would continuous recognition test in which they de-
falsely remember events as having occurred cided if each presented word had been studied
when the events had only been implied by previously in the list. When subjects encoun-
prose passages but not actually stated (see too tered a word that bore an associative relation
Johnson, Bransford, & Solomon, 1973). For to a previously studied word (e.g., to decide
example, subjects studied sentences such as if chair had been previously presented, when
‘‘Three turtles rested on a floating log, and a in fact table had been), they were more likely
fish swam beneath them.’’ Later subjects were to provide false alarms to the words (to chair
asked if the sentence ‘‘Three turtles rested on in this example) than if there were no associa-
a floating log and a fish swam beneath it’’ had tive connection between the test word and
appeared in the passage and a large percentage prior words. Others replicated this effect (e.g.,
said yes. Other research also shows that infor- Anisfeld & Knapp, 1968; Paul, 1979), but it
mation implied, but not actually stated, is of- is sometimes rather small or even nonexistent
ten remembered as if it had actually occurred (Gillund & Shiffrin, 1984). However, others
(Brewer, 1977; Harris, 1974). have shown that the false recognition effect
Much work in this tradition showed how increases with the number of previously pre-
what a person already knew, or the context in sented words that are semantically related to
which information was presented, determined the test word or lure (Hall & Kozloff, 1973;
whether and how it would be remembered. Hintzman, 1988; Shiffrin, Huber, & Marinelli,
Sulin and Dooling (1974) showed that sub- 1995). Underwood (1965) interpreted the false
jects would falsely recognize a statement recognition phenomenon in this paradigm as
about a person if an appropriate schema had being caused by implicit associative responses
been invoked. For example, a week after read- during encoding; when studying table, chair
ing a story about a troubled girl, subjects were might be aroused and therefore later recog-
more likely to recognize the sentence ‘‘She nized. With increasing numbers of related
was deaf, dumb, and blind’’ if they had been words, the chances of subjects producing the
told the story was about Helen Keller than if critical associates might be increased, thereby
they had been told it was about Carol Harris. accounting for the more powerful effect ob-
Many other experiments of similar form have served in later studies.
demonstrated the power of schema (and The great bulk of the evidence about mem-
scripts) in organizing incoming information, ory distortion relies on recognition paradigms,
in line with Bartlett’s (1932) and Miller’s the primary measure in the studies described
(1956b) ideas, cited above (see Bower, in this section and the previous one being false
Black, & Turner, 1979; Owens, Bower, & alarms. Studies examining false recall, either
Black, 1979; and Spiro, 1980, for some further in the schema/prose tradition or the more stan-
examples and Alba & Hasher, 1983, and dard word list tradition are much rarer, espe-
Brewer & Nakamura, 1984, for reviews). cially with the failures to replicate Bartlett’s
(1932) findings (but see Brewer, 1977;
Relatedness Effects Hasher & Griffin, 1978; and Spiro, 1980, for
The experiments described in the previous exceptions in the prose tradition). Deese
section show that people will recognize sen- (1959) reported results of a paradigm that
tences or passages as having occurred if they showed robust levels of false recall following
are similar in meaning to previously studied presentation of word lists, at least for certain
prose materials. Similar phenomena have been lists. Deese (1959) constructed his lists from
uncovered in other paradigms in which sub- word association norms by selecting words
jects study word lists and then are given recog- associated to one target word. For example,

MEMORY ILLUSIONS 85
hill, valley, climb, summit, and top are all represented by two different speakers during
lated to mountain. Deese (1959) presented lists study and subjects were asked on a later recog-
of 12 related words and then measured the intru- nition test to make judgments about which
sion of the critical nonpresented word from speaker had said words that were recognized
which the lists were derived, such as mountain. as old or studied. Payne et al. found that sub-
For some lists (ones in which there was a high jects were quite willing to indicate which of
probability of generating the critical nonpre- the two people had spoken words that had
sented target from list members, as measured actually never been presented in either voice.
by free association norms), the level of intru- The conclusion that they reached is similar to
sions, or false recalls, was quite high. Cramer that derived from the remember/know judg-
(1965) reported similar observations, but neither ments: subjects apparently experience the rec-
study received widespread attention. ollection of these events that never happened
Roediger and McDermott (1995) revived as quite real, as real as the recall of the word
Deese’s (1959) paradigm, replicated the phe- events that actually had occurred. ‘‘False
nomenon of high levels of false recall follow- memories’’ may be a misnomer, at least from
ing presentation of related word lists, and ex- the subjects’ viewpoint, because the experi-
tended the findings in several directions. For ence of recollecting the critical nonpresented
example, they showed that false recognition words appears as real to the subjects as their
of the target words (the false alarm rate) recollections of the presented words.
equaled the hit rate in some conditions and Schacter, Verfaille, and Pradere used the
that the act of recall later enhanced both accu- same paradigm to study development of false
rate and false recognition. Further, using Tul- memories in amnesic patients. Interestingly,
ving’s (1985) paradigm in which subjects they found that patients were less susceptible
were asked to judge whether they actually re- to this memory illusion than were control sub-
membered the occurrence of words (rather jects. They argued that this outcome indicates
than knowing that they occurred, but not spe- that development of false memories in this
cifically remembering the moment of presen- paradigm can be interpreted with the Reyna
tation), subjects frequently remembered the and Brainerd (1995) fuzzy trace theory (as did
words that were never presented. Payne et al., too). In particular, Schacter et al.
Three papers in this issue use Roediger and interpreted the finding of less false recall and
McDermott’s (1995) paradigm. McDermott recognition in amnesic patients as indicating
replicates the basic phenomenon and adds sig- that the gist representations that normally sup-
nificant new information. For example, she port recollective processes are impaired in the
shows that multiple study and test opportuni- patients, relative to normal controls.
ties cause the levels of false recall to decline, The paradigms introduced by Deese (1959)
but that they are still substantial after five and Underwood (1965) and their variations
study/test trials. In addition, blocked presenta- serve as tractable experimental manipulations
tion of words leads to higher levels of false for studying the development of false memo-
recall than does random presentation and, over ries. As noted by Roediger and McDermott
time, accurate recall declines, whereas false (1995), the robust levels of false recall and
recall actually becomes somewhat greater. false recognition are remarkable, because they
Payne, Elie, Blackwell, and Neuschatz also occur under conditions that often promote
replicate the basic phenomena reported by good verbatim (or reproductive) recall: simple
Roediger and McDermott (1995) while adding word list materials, free recall tests that dis-
important new insights. For example, the false courage guessing, short retention intervals,
memory effect grew over repeated testing (un- and the use of metamemory judgments (re-
der conditions, unlike those of McDermott, in member/know judgments, modality judg-
which there were no additional study opportu- ments) that direct subjects’ attention to the
nities). In another experiment, words were bases of veridical judgment.

Effects of Interference and Misleading present in the scene, so these responses consti-
Information tuted errors. Loftus and Palmer (1974) argued
that the question involving the powerful verb
Perhaps the most studied way of systemati- smashed caused subjects to recode the acci-
cally distorting memory is presenting other dent differently and consequently to be more
events or information to interfere with reten- likely later to remember broken glass where
tion of a critical target event of interest. Müller none existed.
and Pilzecker (1900) first identified the role The Loftus and Palmer (1974) paradigm
of interference in forgetting and it has been eventually developed into what has come to
studied, in various traditions, ever since. The be called the misinformation effect paradigm.
critical roles of both retroactive interference In the standard case (e.g., Loftus, Miller, &
(McGeoch, 1932) and proactive interference Burns, 1978) subjects view a scene (for exam-
(Underwood, 1957) in producing forgetting ple, a traffic accident in which an automobile
have long been recognized. Although error pulls into an intersection where there is a stop
analyses played a role in the classic study of sign) and then later they are asked questions
interference—for example, the curious pat- (or read a passage) that contain information
tern of intrusions as a function of trials of about the original scene. Some subjects are
interpolated learning convinced Melton and told that the sign was a yield sign (the misin-
Irwin (1940) that a Factor X (unlearning) formation condition), whereas it is referred to
needed to be added to response competition in a neutral manner (a traffic sign) or not men-
to explain forgetting—the primary interest of tioned at all for subjects in the control condition.
the times was not in how interfering informa- Typically, memory for the information in the
tion created memory distortions and illusions. original scene is worse when misinformation has
Rather, the interest was in the forgetting of been given relative to performance in the control
studied information. condition, although the magnitude of the effect
At about the historic point when interfer- depends on many factors.
ence theory had peaked and a decline in re- This paradigm has produced a huge amount
search interest was occurring (usually dated of research (see Loftus, 1979a, 1993b, for re-
from the Postman and Underwood (1973) pa- views) showing how information coming after
per), Loftus and Palmer (1974) introduced a an event can shape and mold remembrance of
paradigm that bore formal similarity to a retro- the event. Interest centers both on forgetting
active interference paradigm, but that was of the original event and, more importantly for
used to ask different questions with different present purposes, on the case where subjects
materials. They showed subjects a videotape come to remember the suggested event as ac-
of an automobile accident and later gave them tually having occurred (Lindsay & Johnson,
a questionnaire in which a critical question 1989; Weingardt, Toland, & Loftus, 1994).
was embedded: ‘‘About how fast were the two Two papers in the special issue are devoted
cars going when they ___ each other?’’ The to this problem. Mitchell and Zaragoza exam-
verb was varied for different groups of sub- ine the effects of repeating the misinformation
jects. For example, when hit was used, sub- on memory for the target event and for the
jects estimated the speed at 34 miles per hour misinformation itself. Repeating the mis-
but when smashed into was used, the estimate leading statements makes subjects less likely
grew to 41 miles per hour. More interestingly to recall the original events accurately. Roe-
for present purposes, when subjects were later diger, Jacoby, and McDermott examine the
asked the question ‘‘Did you see broken effects of repeated testing on recall of events
glass?’’ 14% of those who had received the in this paradigm. In particular, they provided
question with the verb hit answered yes, conditions in which subjects were encouraged
whereas 32% of those who had received to produce the misleading information on a
smashed answered yes. No broken glass was first test, but then examined subjects 2 days

MEMORY ILLUSIONS 87
later under strict instructions to recall only negative effects of verbalization on remem-
events that they had actually witnessed during bering. In line with Glanzer and Clark’s
the first phase of the experiment, in the slides. (1963) verbal loop hypothesis, people remem-
Subjects who had provided the misinformation ber the events as they described them to them-
on the first test were still quite likely to do selves, not as they actually happened.
so again on the second test (relative to control
groups). Further, subjects in this condition Illusions of Reality Monitoring and Source
claimed to remember the actual occurrence of Monitoring
the suggested event, even though it had never In the 1970s Marcia Johnson and her col-
happened. The misinformation paradigm intro- leagues began a program of research that has
duced by Loftus continues to play a prominent identified a variety of interesting and im-
role in studies of memory errors (see the edited portant facts about accuracy and inaccuracy
volume by Ross, Read, and Toglia (1994) for of remembering. In the original work (e.g.,
recent reviews of these issues). Johnson, Taylor, & Raye, 1977; Johnson &
Raye, 1981), interest was in how subjects dis-
Verbal Overshadowing tinguish between information derived from
A situation related to that of Loftus’ para- external sources (events that really happened)
digm was developed by Schooler and En- and those that were generated by sources in-
gstler-Schooler (1990), which they called ver- ternal to the person (such as events that were
bal overshadowing. Carmichael, Hogan, and only generated or imagined), or, to use the
Walter (1932) showed long ago that ambigu- question that formed the title of a related paper
ous visual stimuli were remembered in differ- by Anderson (1984): ‘‘Did I do it or did I only
ent ways depending on the verbal description imagine doing it?’’ Early experiments were
given to the stimuli. Of course, in many situa- concerned with estimating the frequency with
tions verbalization can aid memory, and many which events were presented externally or
studies have revealed positive effects of re- were internally generated. For example, sub-
hearsal and verbal elaboration. However, jects show better relative frequency judgments
Schooler and Engstler-Schooler (1990) argued for internally than externally produced events.
that verbalization may harm memory when More interestingly for present purposes, the
the events to be remembered are difficult to more subjects imagined an event, the greater
capture in words. They have demonstrated this the impact on judgments of frequency that the
in a number of studies using stimuli from vari- event actually occurred (Johnson et al., 1977;
ous domains. For example, when subjects Johnson & Raye, 1981). Subjects confused the
viewed faces and had to describe them, mem- imagined events with those that actually oc-
ory for the faces was worse than when subjects curred in estimating their frequency, a failure
did not describe the faces. to monitor real events accurately, or a failure
In their article in this issue, Melcher and of reality monitoring, as Johnson and Raye
Schooler extend their verbal overshadowing (1981) called it.
paradigm to recognition of the taste of wine. Johnson and Raye (1981) proposed that
Untrained wine drinkers were impaired in people distinguish between internal and exter-
their recognition of wines when they engaged nal events in memory both by the types of
in verbalization during their initial tasting, rel- information encoded and by decision rules
ative to the case in which they participated in used when tested. Encoded representations of
an unrelated verbal activity. Consistent with external events are especially rich in percep-
their predictions, this verbal overshadowing tual information (sensory features or attri-
effect did not occur in trained wine drinkers butes) and contextual information (such as
who had considerable expertise in translating time and place of occurrence). On the other
their tastes into words. We can expect to see hand, encodings of internally generated events
many more experiments on the positive and are assumed to include considerable informa-

tion about cognitive operations that were used reading about the accident, as actually having
to encode the event. People are assumed to occurred at the original scene (Lindsay &
use these differences in information when de- Johnson, 1989; Zaragoza & Lane, 1994).
ciding on whether something ‘‘really hap- In the present issue, Johnson, Nolde, and
pened’’ to them. However, confusions can DeLeonardis add important new information
arise in some cases, such as when an imagined about the impact of subjects’ emotional focus
event may have been vividly imaged in a par- (either on themselves or on another person)
ticular time, at a particular location, and with on their abilities to accurately monitor the
considerable sensory detail. source of information. Johnson et al. asked
Another critical aspect of Johnson’ theory, people to focus either on how they personally
as developed in later writings (e.g., Johnson, felt about statements that were read to them
1983, 1995) is the distinction between two gen- or about how they thought the speakers of the
eral types of judgment processes (similar to statement felt when reading them. Although
those proposed by others; Atkinson & Juola, focusing on their own reactions to statements
1973; Jacoby & Dallas, 1981; Mandler, 1980). made people recognize the statements better
Briefly, subjects can use either a relatively on an old/new recognition test, this self-focus
quick, nondeliberative heuristic process in mak- actually reduced their ability to monitor the
ing judgments or they can use more systematic source (to tell which of two people had spoken
processes that are slower and more deliberate. the statement). The difference in self versus
The heuristic process is quick and dirty, used other focus was eliminated, however, in an
to make snap judgments and resulting in errors experiment when subjects were instructed to
when the snap judgment leads to the conclusion focus on how they felt about the speakers
(for example) that something actually happened rather than (as in the studies described above)
when it only seems familiar because the person when subjects focused on how they thought
thought about it (or generated it internally). Sys- the speakers felt about what they were saying,
tematic judgment processes bring other features a finding in line with predictions made from
to bear on the decision as to whether something the source monitoring framework.
actually happened and are therefore less likely The contribution by Hyman and Pentland
to be error prone. in this issue also can be interpreted within
A large variety of evidence has been col- Johnson’s source monitoring framework.
lected that supports Johnson’s general ap- They use the technique of having people
proach, which has been broadened to a general imagine events that may or may not have oc-
source monitoring framework (Johnson, curred during their childhoods and show that
Hashtroudi, & Lindsay, 1993). Information imagining the events sometimes induces peo-
can arise from many sources that vary in mo- ple to believe that the events did happen. The
dality (auditory, visual, verbal, pictorial), time events are, on the face of it, fairly improbable
of occurrence, place of occurrence, and so on, ones. This outcome agrees with others in
rather than simply being external or internal. showing the powerful role of imagination in
A primary type of data used to support the creating false memories (Johnson, Foley,
theory is confusion that subjects show among Suengas, & Raye, 1989; see too Garry, Man-
sources, leading to interesting illusions of ning, Loftus, & Sherman, in press). Imagining
memory. The source monitoring framework seems to involve brain mechanisms similar to
has assumed increased importance in the field, those used in perceiving (e.g., Farah, 1989),
as researchers have interpreted other phenom- so it is not surprising that imagining events
ena under its umbrella. For example, the im- can make people later remember that they had
pact of misleading postevent information can happened (even when they did not). However,
be conceived as a failure of source monitoring; the positive effects of imagery have usually
subjects may remember the stop sign, which been emphasized by psychologists interested
was only suggested by a question or from in memory—how imagining studied events

MEMORY ILLUSIONS 89
makes them more memorable—so the role of (if the subjects’ attention is focused on per-
imagery in undermining memory may at first forming some other task) this fluency may be
glance seem surprising. However, Johnson’s misattributed to some other source besides the
framework permits us to understand how im- past events. For example, Jacoby, Allan, Col-
agery can both make studied events more lins, and Larwill (1988) tested subjects’ judg-
likely to be correctly recalled, and imagined ments of the loudness of background noise in
‘‘nonevents’’ more likely to be falsely re- which words were embedded when some of
called. In the former case, imagining adds ad- the words had been previously heard and oth-
ditional cognitive operations to the encoded ers had not. When subjects had previously
event that serve to enhance later memory for heard the words, they judged the noise to be
the event’s occurrence; in the latter case, the less loud than when nonstudied words were
cognitive operations help make the ‘‘non- judged, despite the fact that the objective noise
event’’ appear real by specifying perceptual level was the same in the two cases. The facili-
detail, which usually signifies an event that tating effects of prior experience on hearing
actually occurred (Johnson et al., 1989). Other the words through noise were misattributed
illusions on the border between perceiving and to differences in the noise levels rather than
remembering are considered below, in the accurately attributed to memory for prior ex-
next two sections. perience of the words.
Jacoby and Whitehouse (1989) manipulated
Fluency Illusions: Misattributions the ease with which words were perceived on
of Memory a recognition test by preceding them with a
Jacoby, Kelley, and their colleagues have briefly presented prime. Subjects judged the
developed an attributional view of memory words that were primed as more familiar on
that is similar in some ways to Johnson’s a recognition test, presumably mistaking the
(1983) source monitoring approach (Jacoby, fluency of processing for familiarity. Jacoby
Kelley, & Dywan, 1989; Kelley & Jacoby, and Whitehouse proposed that such fluency
1990). They borrow from attribution theory in may be the cause of déjà vu experiences (such
social psychology, where (for example) emo- as described by Titchener in the opening para-
tional states are thought to arise from a partic- graphs). The enhanced fluency of an experi-
ular combination of physiological arousal and ence from a prior event that is not remembered
appropriate cognitions induced by the situa- may create a strange sense of familiarity, with
tion (e.g., Schachter & Singer, 1962). Simi- no obvious source to which it should be attrib-
larly, Jacoby, Kelley, and their colleagues uted. Whittlesea, Jacoby, and Girard (1990)
conceive of remembering as a combination of and Whittlesea (1993) reported illusions of
fluent processing of an event with the mental immediate memory that were similarly caused
set that attributes the fluency to past experi- by manipulations of perceptual fluency.
ence, or to remembering. As in Johnson’s Two papers in the current issue draw on
framework, Jacoby and Kelley distinguish dif- Jacoby and Kelley’s attributional framework.
ferent bases of judgment: a more intuitive, Lindsay and Kelley show that enhanced flu-
rapid, and nonanalytic basis on the one hand, ency can drive know judgments in Tulving’s
and a more deliberate, reflective, and analytic (1985) remember/know paradigm. In their ex-
basis on the other. For the experience of re- periments, either easily solved word fragment
membering there needs to be fluent processing cues, or ones that were difficult to solve, were
of information (of it readily coming to mind) presented as test cues following study of a
and an attribution of the fluency as arising word list. Subjects were told that the cues
from past experience. should remind them of list words and that they
Several interesting illusions support Jacoby were to report the word and to judge whether
and Kelley’s attributional framework. First, they remembered details of its list presenta-
past events may lead to fluent processing and tion, or rather simply knew that it had been

on the list. Most studied words were repre- we can look at our writing more as someone
sented by fragments appearing on the test, just coming to it, without being quite so ego-
leading subjects to believe that the fragments centric. Kelley and Jacoby studied these fasci-
were reliable retrieval cues for the list items. nating processes of judgment with an anagram
Lindsay and Kelley’s primary interest was in solution task. When subjects had studied ana-
judgments for a small group of words that had grams, they solved them faster on a later test
not been presented in the study phase. In this (a priming effect), but they misattributed this
case, making the solution easy led subjects to ease to properties of the anagram rather than
report the word as having been presented, but to their previous experience. Subjects’ order-
to judge that they knew it had been on the ing of the general difficulty of the anagrams
list (but did not remember the details of its was different when the anagrams had been
occurrence). This interesting illusion of know- studied than when they had not; presumably,
ing depends a bit on assumptions used to ana- when subjects had studied solutions to the
lyze know responses. In addition, recall that anagrams, they partly used fluency to deter-
Roediger and McDermott (1995) reported that mine the ease of arriving at the anagram’s
subjects would often provide remember re- solution, or they relied more on nonanalytic
sponses for nonpresented items recognized in bases of judgment. When anagrams had not
their paradigm. The differences between their been studied, subjects probably used more de-
results and those of Lindsay and Kelley may liberate, rule-based judgments. Thus, some-
lie in the tests and the types of processing they what paradoxically, experience within a do-
engender. The Lindsay and Kelley paradigm main may reduce one’s ability to assess how
capitalizes on perceptual (data-driven) factors others assess and comprehend the same infor-
in producing fluent output, with powerful cues; mation, a form of adult egocentrism (as Kelley
this sense of false familiarity arising from per- and Jacoby call it) that doubtless afflicts us
ceptual factors may lead to know judgments. all. All college professors have had the experi-
The Roediger and McDermott (1995) subjects ence of explaining an interesting point, with
produced an erroneous memory conceptually re- what they imagine to be great clarity, only to
lated to other memories and judged them as be confronted with the befuddled looks of
being remembered (rather than known) most of most of the students in the class.
the time. One hypothesis is that facile processing
driven by perceptual fluency may lead to the Illusions of Perception and Memory
illusion of knowing, whereas when material The illusions arising from perceptual flu-
pops to mind that is conceptually related to other ency discussed in the previous section docu-
memories, it may give rise to the illusion of ment the intimate link between perceiving and
remembering. This speculation awaits further re- remembering, which can be easily demon-
search (see Rajaram & Roediger, in press, for strated either during the encoding or retrieval
related observations). phase of memory experiments (as in the work
In another set of studies, Kelley and Jacoby of Jacoby and Kelley). Many illusions seem
studied how memory of past events may inter- to lie on the border of perceiving and remem-
fere with making good subjective judgments. bering. Consider the thought experiment of
Writers are frequently unable to judge whether testing subjects’ retention of the shapes in Fig.
what they have written is comprehensible, and 1. If they were simply shown the figures with
a frequently suggested strategy, once a paper instructions to attend to the shaded space and
is written, is to put it away for a few weeks then asked, a few moments later, which was
before revising it. Presumably this advice cap- wider and shorter, the one on the right or the
italizes on the intuition that our judgment of one on the left, they would doubtless respond
the goodness of our writing will be better once that it was the one on the right. They would
we have forgotten what we wanted to say and of course be wrong, as the shaded portions are
why we wrote what we did; if we wait a while, identical. Is this an illusion of perception? an

MEMORY ILLUSIONS 91
illusion of memory? or is it both? If the origi- what do you remember?). Obviously, the experi-
nal event is wrongly perceived, then usually mental situations blend into one another and it
the resulting memory will be wrong too, un- is impossible to say when perceiving is com-
less some other factor corrects the false per- pleted and remembering takes over in subjects’
ception later (making memory for the event reports.
better, in a sense). One memory illusion at the border between
Another issue: Are sensory memories perceiving and remembering is the phenome-
(iconic and echoic) veridical, or illusory, or non of boundary extension, studied by Intraub
both, in different aspects? Sensory memories and her collaborators. Observers view a pic-
are thought to be quite faithful replicas of the ture and then later remember seeing more of
original experience, at least for their brief life- the scene than was actually shown; it is as
times. On the other hand, the perceiver/re- though the scene’s boundaries have become
memberer still experiences events as being extended when it is remembered. The effect
perceived that have disappeared. So perhaps occurs both in recall and recognition tests (In-
sensory memories have both veridical and il- traub & Richardson, 1989; Intraub, Bender &
lusory aspects. Like the questions above, this Mangels, 1992). In this issue, Intraub, Gottes-
one about sensory memories is difficult to an- man, Willey, and Zuk examine the effects of
swer conclusively. Similarly, the illusory phe- brevity of exposure on the phenomenon, as
nomenon known as the McCollough effect well as testing after very short delays. They
(McCollough, 1965) can last for days. (The found that the effect occurs even when expo-
McCollough effect is an illusion of color that sure durations for the studied pictures are
is contingent on the particular orientation of quite brief (250 or 333 ms) and when the re-
adaptation and test lines.) Is it a phenomenon tention interval is only 1 s. In conjunction with
of erroneous perception, of memory, or both? other evidence, they interpret the illusion as
These questions also illustrate again the diffi- arising just after a picture is perceived. In par-
culty of rigorously defining what an illusion ticular, subjects interpret a picture in terms of
is; such a definition has thus far eluded psy- a perceptual schema that contains information
chologists studying perceiving and it will likely to exist just outside the actual picture.
probably be even more difficult to achieve for When the scene is comprehended, subjects re-
remembering. member information that did not actually exist
Several other types of illusion are also diffi- in the picture, but only in their schema used
cult to classify as perceptual or as memorial and to comprehend the picture. The error is clearly
both labels likely apply. Part of the problem memorial, in one sense, because boundary ex-
is that the operational procedures for studying tension does not occur while subjects are look-
perception and memory are often formally quite ing at the scene, so it is not a perceptual illu-
similar. In many experiments on perception of sion in the usual sense. However, the Intraub
words or pictures, the stimuli are presented very et al. results show that the illusion develops
briefly and subjects are asked to report what very rapidly upon offset of the picture, pre-
they saw (or heard) as soon as possible, or they sumably as it is still being interpreted. There-
are asked to make a rapid decision or judgment fore, the phenomenon falls at the boundary
about the event. In many memory experiments between perceiving and remembering.
words and pictures are shown and the subjects Another illusion falling at this boundary is
are asked, after some retention interval, to recall representational momentum (Freyd & Finke,
what occurred or to make a judgment about the 1984). In the original experiments, subjects
presented events. Often the primary differences received three computer animated rectangles
in the situations are (a) the duration of the origi- (called inducing stimuli) and then later had to
nal display, (b) the amount of time before a judge whether a fourth stimulus, the probe or
report or judgment is required, and (c) the nature test stimulus, occurred in the same position as
of the query (what did you perceive? versus the last inducing stimulus. On some occasions

the three inducing stimuli implied rotation in rather small but Reinitz, Lammers, and Coch-
either a clockwise or a counterclockwise di- ran (1992) and Reinitz, Morrisey, and Demb
rection. The interesting finding was that sub- (1994) replicated it and showed similar results
jects were more likely to respond same to the in encoding of faces (see too Solso & McCar-
probe stimulus if it was slightly forward of thy, 1981). The interpretation of such illusory
the final inducing stimulus in the direction of conjunctions in memory, as in perception, is
implied motion, as if the representation of the that features may not be tightly bound together
motion of the inducing stimuli had momentum into unitized wholes. Rather, features are free-
and was carried forward (hence the name). floating to some degree and, when two fea-
The effect increased with the velocity of the tures are combined from different units, they
implied motion and is not due to eye move- may be falsely recognized.
ments (see Hubbard, 1995, for a review). Two papers in the current issue use the illu-
Again, it is difficult to classify this illusion as sory conjunction paradigm to explore the roles
perceptual, mnemonic, or both. of attention and consolidation in encoding. In-
A third and final case to be considered is terestingly, both sets of researchers also used
that of illusory conjunctions in perceiving, a a neuropsychological approach in which pa-
topic of study introduced by Treisman and tients with memory disorders arising from
Schmidt (1982). Subjects are presented quite brain damage were tested. Reinitz, Verfaellie,
briefly with stimuli that contain several fea- and Milberg compared amnesic patients to
tures, such as letters that differ in color (e.g., control subjects in the basic recognition para-
a blue R, a red M, a green C, and so on). The digm in which subjects are tested with pre-
interesting observation, now replicated many viously studied words, conjunction words (the
times, is that subjects will often report seeing two syllables have been presented, but in dif-
objects that represent illusory conjunctions of ferent words), feature words (one syllable has
features, such as a red R or a blue C in the been studied and the other is new), or com-
above example. Because the displays are so pletely new words. Control subjects showed
rapid in this research, these errors have a per- the pattern expected from prior experiments:
ceptual feel, yet they occur at a short interval The hit rate for studied words was much
after the perceptual display has disappeared. higher than the false alarm rate for conjunc-
However, as we see in the next section, illu- tion words, and in turn this false alarm rate
sory conjunctions also arise in remembering. was higher than that for either feature words
or new words. Amnesic patients showed a dif-
Illusory Conjunctions in Memory ferent pattern in that they were unable to dis-
Underwood and Zimmerman (1973) and criminate between studied words and conjunc-
Underwood, Kapelak, and Malmi (1976) in- tion words. The patients seemed not to have
troduced a way of studying illusory conjunc- formed a global representation of the words,
tions in memory by presenting compound because parts of the words could ‘‘float’’ and
words for subjects to study (e.g., handstand, be erroneously joined together, leading to
shotgun). Later subjects were tested for recog- false alarm rates as high as the hit rates.
nition of previously studied words, words that Kroll, Knight, Metcalfe, Wolf, and Tulving
had one syllable in common with the studied use a variant of the illusory conjunction para-
words (handmaid) or, in the most interesting digm on patients having left, right, or bilateral
case, words in which both syllables had been hippocampal damage. In a running recogni-
previously studied (handgun, in this example). tion task in which subjects judged every pre-
Subjects produced more false alarms to the sented word as old or new, items could repre-
conjunction test items like handgun than to sent either previously studied items, conjunc-
either words with one syllable of overlap with tion items, single feature items, or completely
the studied word or control words that were new items. They found that patients with left
unrelated to the studied words. The effect was hippocampal damage classified more conjunc-

MEMORY ILLUSIONS 93
tion items as old (i.e., they produced more ing evidence that hypnosis leads to illusory
false alarms on these items) relative to normal memories (e.g., Lynn & Nash, 1994; White-
subjects or those with right hippocampal dam- house, Orne, Orne, & Dinges, 1991). Subjects
age. Kroll et al. interpreted these results, and under hypnosis are often encouraged to generate
those from a similar experiment involving pic- everything they can think of to free associate
tures, as evidence for a binding process during about the event they are trying to remember.
consolidation of information occurring via the Much of the information obtained under this
hippocampus. state is erroneous, although some is correct
The illusory conjunction paradigm using (but no more than that retrieved by nonhypno-
words, pictures, or faces may prove to be im- tized control subjects given extra time). The
portant in the study of false memories. It is difficulty comes later when subjects (back in
also worth noting that both papers discussed their normal waking state) try to distinguish
here (and the one by Schacter et al. described what actually happened in the target event
previously) used memory-disordered patients from what they recalled during hypnosis. The
to study memory illusions. Certain patients, available evidence shows that subjects who
especially those with damage to the frontal score high on hypnotic susceptibility scales
lobes and to the areas surrounding the hippo- often confuse events they retrieved under hyp-
campus, display profound memory disorders. nosis with those that actually occurred and
Although amnesic patients—those with dam- that they cannot even distinguish the memo-
age in and around the hippocampus who are ries they retrieved under hypnosis from those
rendered incapable of consciously recollecting that they retrieved prior to the hypnotic ses-
newly presented information—represent an sion (Sheehan, 1988; Lynn and Nash, 1994;
interesting type of disorder, those patients Whitehouse et al., 1991).
with frontal damage who confabulate and These findings of false memories induced
claim to remember events that never happened by hypnosis bear many points of similarity
may turn out to be just as interesting in the with phenomena discussed previously. In a
study of memory illusions (Johnson, 1991; sense, hypnotized subjects are encouraged to
Moscovitch, in press; Schacter, Curran, Gal- provide their own misinformation about
lucio, Milberg, & Bates, in press). events during the hypnotic session and it may
even be that such internally generated misin-
Hypnosis and Guessing Effects formation (in the form of guesses of what
The use of hypnosis in attempts to aid mem- might have happened) could be even more
ory retrieval has a long and checkered history. powerful than externally presented misinfor-
The topic has been examined intensively for mation in creating interference, although this
the past 20 to 30 years and there now seems hypothesis awaits experimental test. At the
general agreement that hypnosis does not very least, subjects who are queried later seem
serve as the key to unlock forgotten memories, to display reality monitoring and source moni-
as some had hoped (see Smith, 1983, and toring confusions, not being able to tell what
Lynn & Nash, 1994, for reviews and discus- actually happened from descriptions that they
sion). Relative to appropriate control groups, produced while hypnotized. In addition, the
hypnotized subjects generally recall no more Jacoby et al. (1989) attributional approach
veridical information than do control subjects may help here, because as Schacter (1995) has
and, on the few occasions when they have noted, ‘‘The evidence suggests that hypnosis
been shown to do so (e.g., Dywan & Bowers, creates a retrieval environment in which peo-
1983), the increased recall comes only at the ple are more willing than usual to call a mental
expense of an increase in errors, suggesting a experience a ‘memory,’ and in which they ex-
general criterion shift. press a great deal of confidence in both true
Although hypnosis does not lead to en- and false memories’’ (p. 17).
hanced accuracy of retrieval, there is mount- Although most researchers attribute the illu-

sions created when subjects are hypnotized to source of information. Are some types of peo-
the state of hypnosis per se, it may be that they ple more susceptible to memory illusions than
are caused in part by the instructions during others? Considerable work has already been
hypnosis that lead people to produce errone- conducted with children in an attempt to deter-
ous information. That is, it may well be that mine how accurately they remember, espe-
nonhypnotized subjects who are instructed to cially for the practical purpose of whether
guess during retrieval will show the same dif- children’s testimony should be admitted in
ficulties in later monitoring their recollections court (see Ceci & Bruck, 1995, for a compre-
that hypnotized subjects show. Studies briefly hensive review). The research activity with
reported in Jacoby et al. (1989, p. 413) and older people has also begun, but is less volu-
Roediger, Wheeler, and Rajaram (1993, pp. minous. For example, Johnson and her collab-
114–122) provide evidence for this hypothe- orators have asked if reality and source moni-
sis (although neither study used groups tested toring is less accurate for children and for
under hypnosis for comparison). For example, older adults than for young adults (e.g., Foley,
Roediger et al. used a forced recall procedure Johnson, & Raye, 1983; Hashtroudi, John-
in which subjects were made to guess during son, & Chrosniak, 1989). Similarly, other
recall. They later asked subjects to rate all studies have shown that people high in imag-
items they had produced as to whether the ery ability are more likely to be confused in
response actually occurred during the study a reality monitoring paradigm after imagining
phase of the experiment. They found that sub- events than are poor imagers (Johnson, Raye,
jects who had guessed often judged that their Wang & Taylor, 1979). As already noted
erroneous responses, presumably produced as above, another rich source of information is
guesses, had actually occurred. Therefore, it that supplied by various patient populations,
may be that the act of guessing while retriev- as in the Kroll et al., Reinitz et al., and
ing information about an event produces false Schacter et al. papers appearing in this issue.
memories even when subjects are not hypno- Finally, people who score high on scales of
tized. However, study of this possibility is just hypnotic suggestibility are more easily led to
beginning. memory illusions than are those who score low.
• Illusions of judging and deciding. Many
CONCLUSIONS
cognitive illusions discovered by psycholo-
Perceptual illusions have been thoroughly gists studying judgment and decision making
studied for nearly 150 years, whereas memory have a memorial basis. For example, the avail-
illusions have been intensively studied for ability heuristic (Tversky & Kahneman, 1974)
(perhaps) 30 years. Like perceptual illusions, refers to the tendency of decision makers to
there are doubtless many types of memory assess the frequency of events by the ease
illusions that are caused by different factors. with which instances come to mind. So, for
We are as unlikely to find unitary explanations example, subjects judge that more people die
of memory illusions as we are of perceptual from shark attacks than from being hit by fall-
illusions. ing parts from airplanes, but actually the
This essay has touched on only some illu- chances of dying from falling airplane parts
sions that can be considered memory illusions. are 30 times higher (Plous, 1993). Hindsight
My survey in the second half of this paper bias and the knew-it-all-along effect may sim-
was motivated more by setting the stage for ilarly be caused by problems in assessing our
articles in the special issue than by an exhaus- own metamemories (Fischoff, 1975; Wood,
tive review of the relevant phenomena. I con- 1978). After something has happened, we can
clude by noting briefly a number of phenom- no longer access our mental state prior to the
ena and issues that deserve attention and that happening, to see how unlikely we might then
have not been emphasized here. have judged the event to be. (The processes
• Individual differences represent a rich may be similar to those studied by Kelley and

MEMORY ILLUSIONS 95
Jacoby, this issue.) In a different arena, people related example, Glenberg, Wilkinson, and
notice and remember streaks of events and Epstein (1982) reported an illusion of know-
may then judge them as representative of a ing in which subjects’ ratings of how well
general rule, even though careful analyses they comprehended a passage failed to corre-
show that these streaks are probably occurring late with how well they could later answer
no more than would be expected by chance. questions about it (but see Weaver, 1991, for
For example, basketball players’ belief in the some limitations on this finding). These exam-
phenomenon of the hot hand—that players ples could be extended (see Nelson, 1992).
suddenly get hot and make streaks of shots— Two of the most famous illusions of mem-
is probably due to this illusion of perceiving ory—déjà vu and jamais vu—are illusions of
and remembering clusters in a random se- metacognition. In the former case, which we
quence (Gilovich, Vallone, & Tversky, 1985). have considered, people suffer the strange
These and other examples of phenomena of feeling that a new situation is very familiar;
bias and illusion in judgment, often studied as noted, Jacoby and Whitehouse (1989) ana-
more by social psychologists than by cogni- lyzed this feeling as possibly arising from mis-
tive psychologists, seem to have their basis in attributed perceptual fluency in the apparently
memory illusions (see Gilovich, 1991; Pia- new situation. Jamais vu is the failure to find
telli-Palmarini, 1994; and Plous, 1993, for sur- familiar something that was recently experi-
veys of these phenomena). Just as there is no enced and should feel familiar. An experimen-
firm dividing line between illusions of per- tal example may be the phenomenon of the
ceiving and remembering, there is no firm line recognition failure of recallable words, re-
between illusions of remembering and those ported by Tulving and Thomson (1973). Sub-
of judging and deciding. Future work may jects fail to recognize in one context that a
seek commonalities that underlie illusions in word has recently been presented, yet they
different domains. can recall that word as a recently experienced
• Other illusions of metacognition. The episode with a different cue. In pathological
study of metacognition—the assessment of cases, amnesic patients fail to recognize as
our own knowledge of how we learn and what familiar events that have been recently experi-
we know—has greatly informed the study of enced and that can be produced in implicit
memory illusions. Studies of source monitor- or indirect tests of memory (Warrington &
ing, of remember/know judgments, and of Weiskrantz, 1968; Graf, Squire, & Mandler,
over- and underconfidence in decisions all rely 1985). However, the concept of jamais vu has
on metamemory judgments. However, the been given very little attention and the two
study of metacognition is rife with other possi- phenomena mentioned here are not usually
bilities (see Nelson, 1992, for an introduction). raised in this context (although they seem to
Whenever our metacognitive monitoring fails fit the definition).
to track our performance accurately (and espe- • The topic of memory illusions has not
cially when our judgments about our perfor- been confined to the psychologists’ labora-
mance are totally at odds with our behavior), tory, but is now a topic of concern in several
our judgments reflect an illusion of knowing. areas of society. The accuracy of eyewitness
For example, Nelson and Leonesio (1988) testimony (Munsterberg, 1908; Loftus, 1979)
showed that subjects would continue to study is of long-standing interest. The problem of
material (under self-paced study conditions) false identification of suspects and false re-
in attempting to master it, even though the ports of details in criminal investigations is of
additional study time produced little or no gain prime concern. Recently, the issue of memo-
in performance. Subjects could not accurately ries that appear to be recovered in therapy
monitor when to call it quits in studying the (often an adult recalling being abused as a
information, which Nelson and Leonesio (1988) child) has been a topic of much debate (see
referred to as the labor-in-vain effect. In a Lindsay & Read, 1994; Loftus, 1993a). Do

these recovered memories represent accurate BADDELEY, A. D. (1976). The psychology of memory.
recollections, or are they false memories in- New York: Harper & Row.
BARTLETT, F. C. (1932). Remembering: A study in experi-
duced during the process of therapy (by imag- mental and social psychology, Cambridge, England:
ining events, by reading about related events, Cambridge Univ. Press.
and from therapist suggestions)? Another BENEDETTI, F. (1985). Processing of spatial tactile infor-
topic of interest outside the lab is how groups, mation with crossed fingers. Journal of Experimental
societies and cultures may systematically mis- Psychology: Human Perception and Performance,
11, 517–525.
remember their past. Such studies cross disci-
BOAS, F. (1901). Kathlamet texts. Smithsonian Institution
plinary boundaries, and sociologists and histo- Bureau of American Ethnology Bulletin, 26, 182–186.
rians are more likely to investigate these topics BOWER, G. H., BLACK, J. B., & TURNER T. J. (1979).
than are psychologists. Kammen (1995) re- Scripts in memory for text. Cognitive Psychology,
ports on some commonly believed distortions 11, 177–220.
of American history, and Schudson (1995) re- BRANSFORD, J. D., BARCLAY, J. R., & FRANKS, J. J.
flects on the dynamics that may underlie dis- (1972). Sentence memory: A constructive versus in-
terpretive approach. Cognitive Psychology, 3, 193–
tortions in collective memories of the past. 209.
All the examples in this section further sup- BRANSFORD, J. D., & FRANKS, J. J. (1971). The abstraction
port the basic premise of this essay and, in- of linguistic ideas. Cognitive Psychology, 2, 331–350.
BREWER, W. F. (1977). Memory for the pragmatic impli-
deed, of the special issue. Memory illusions
cations of sentences. Memory and Cognition, 5, 673–
exist, they are robust, they come in many 678.
types, and they are worthy of systematic BREWER, W. F., & NAKAMURA, G. V. (1984). The nature
study. Just as with perceptual illusions, mem- and function of schemas. In R. S. Wyer & T. K.
ory illusions should be considered an estab- Srull (Eds.), Handbook of social cognition. Hillsdale,
lished topic whose understanding is critical to NJ: Erlbaum.
BROADBENT, D. E. (1958). Perception and communica-
develop a successful understanding of remem-
tion. New York: Pergamon.
bering. Indeed, the past 30 years has seen an BRUNER, J. S., & POTTER, M. C. (1964). Interference in
impressive body of scholarship develop on visual recognition. Science, 144, 424–425.
memory illusions, although often under dispa- CARMICHAEL, L., HOGAN, H. P., & WALTERS A. A.
rate labels and in seemingly unrelated areas (1932). An experimental study of the effect of lan-
of research. The theme of this essay is that we guage on the reproduction of visually perceived form.
Journal of Experimental Psychology, 15, 73–86.
may want to consider interconnections among
CECI, S. J., & BRUCK, M. N. (1993). Suggestibility of
these topics, to permit them to inform one the child witness: A historical review and synthesis.
another. In addition, we should pursue the Psychological Bulletin, 113, 403–439.
cases in which remembering is erroneous to CECI, S. J., & BRUCK, M. (1995). Jeopardy in the court-
inform our theories of memory. As in the room: A scientific analysis of children’s testimony.
study of perception, clues generated from Washington, DC: American Psychological Associa-
tion Press.
studying illusions may serve as a crucible for
CONRAD, R. (1964). Acoustic confusions in immediate
testing theories of memory. memory. British Journal of Psychology, 55, 75–84.
REFERENCES COREN, S., & GIRGUS, J. S. (1978). Seeing is deceiving:
The psychology of visual illusions. Hillsdale, NJ: Erl-
ALBA, J. W., & HASHER, L. (1983). Is memory schematic?
baum.
Psychological Bulletin, 93, 203–231.
ANDERSON, R. E. (1984). Did I do it or did I only imagine CRAMER, P. (1965). Recovery of a discrete memory. Jour-
doing it? Journal of Experimental Psychology: Gen- nal of Personality and Social Psychology, 1, 326–332.
eral, 113, 594–613. DEESE, J. (1959). On the prediction of occurrence of par-
ANISFELD, M., & KNAPP, M. (1968). Association, syn- ticular verbal intrusions in immediate recall. Journal
onymity, and directionality in false recognition. of Experimental Psychology, 58, 17–22.
Journal of Experimental Psychology, 77, 171–179. DREVER, J. (1964). A dictionary of psychology. New
ATKINSON, R. C., & JUOLA, J. F. (1973). Factors influenc- York: Penguin.
ing speed and accuracy of word recognition. In S. DYWAN, J., & BOWERS, K. S. (1983). The use of hypnosis
Kornblum (Ed.), Attention and performance IV (pp. to enhance recall. Science, 222, 184–185.
583–612). New York: Academic Press. EBBINGHAUS, H. (1885/1964). Memory: A contribution

MEMORY ILLUSIONS 97
to experimental psychology. (H. A. Ruger & C. E. nal of Experimental Psychology: Learning, Memory,
Bussenius, Trans.). New York: Dover Publications. and Cognition, 10, 164–178.
ERDELYI, M. H. (1985). Psychoanalysis: Freud’s cogni- GREGORY, R. L. (1970). The intelligent eye. London:
tive psychology. New York: Freeman. Neiderfield & Nicolson.
ERDELYI, M. H., & GOLDBERG, B. (1979). Let’s not sweep GREGORY, R. L., & GOMBRICH, E. H. (1973). Illusion in
repression under the rug: Toward a cognitive psy- nature and art. London: Duckworth.
chology of repression. In J. F. Kihlstrom & F. J. HALL, J. F., & KOZLOFF, E. E. (1973). False recognitions
Evans (Eds.), Functional disorders of memory (pp. of associates of converging versus repeated words.
355–402). Hillsdale, NJ: Erlbaum. American Journal of Psychology, 86, 133–139.
FARAH, M. J. (1989). The neural basis of mental imagery. HARRIS, R. J. (1974). Memory and comprehension of im-
Trends in Neurosciences, 12, 395–399. plications and inferences of complex sentences.
FISCHOFF, B. (1975). Hindsight is not equal to foresight: Journal of Verbal Learning and Verbal Behavior,
The effects of outcome knowledge on judgment un- 13, 626–637.
der uncertainty. Journal of Experimental Psychology: HASHER, L., & GRIFFIN, M. (1978). Reconstructive and
Human Perception and Performance, 1, 288–299. reproductive processes in memory. Journal of Exper-
FOLEY, M. A., JOHNSON, M. K., & RAYE, C. L. (1983). imental Psychology: Human Learning and Memory,
Age-related changes in confusion between memories 4, 318–330.
for thoughts and memories for speech. Child Devel- HASHTROUDI, S., JOHNSON, M. K., & CHROSNIAK, L. D.
opment, 54, 51–60. (1989). Aging and source monitoring. Psychology
FREUD, S. (1895). Project for a scientific psychology. The and Aging, 4, 106–112.
standard edition of the complete psychological works HELMHOLTZ, H. VON. (1903). Popular lectures on scien-
of Sigmund Freud, Vol I (pp. 295–397). London: tific subjects. (E. Atkinson, Trans.). New York:
Hogarth Press. Longmans, Green. [Original work published 1881]
FREUD, S. (1903/1965). The psychopathology of everyday HERSHENSON, M. (1989). The moon illusion. Hillsdale,
life. (J. Strachey, Trans., Ed.). New York: Norton. NJ: Erlbaum.
FREUD, S. (1910). Leonardo da Vinci and memory of HINTZMAN, D. L. (1988). Judgments of frequency and
his childhood. The standard edition of the complete recognition memory in a multiple-trace memory
works of Sigmund Freud, Vol. XI (pp. 63–137). Lon- model. Psychological Review, 95, 528–551.
don: Hogarth Press. HUBBARD, T. L. (1995). Environmental invariants in the
FREYD, J. J., & FINKE, R. A. (1984). Representational representation of motion: Implied dynamics and rep-
momentum. Journal of Experimental Psychology: resentational momentum, gravity, friction, and cen-
Learning, Memory, and Cognition, 10, 126–132. tripetal force. Psychonomic Bulletin & Review, 2,
GARRY, M., MANNING, C. G., LOFTUS, E., & SHERMAN, 322–338.
S. J. (in press). Imagination inflation: Imagining a HYMAN, I. E., & PENTLAND, J. (1996). The role of mental
childhood event inflates confidence that it occurred. imagery in the creation of false childhood memories.
Psychonomic Bulletin & Review. Journal of Memory and Language, 35, 101–117.
GAULD, A., & STEPHENSON, G. M. (1967). Some experi- INTRAUB, H., GOTTESMAN, C. V., WILLEY, E. V., & ZUK,
ments related to Bartlett’s theory of remembering. I. J. (1996). Boundary extension for briefly glimpsed
British Journal of Psychology, 58, 39–49. photographs: Do common perceptual processes result
GILLUND, G., & SHIFFRIN, R. M. (1984). A retrieval model in unexpected memory distortions? Journal of Mem-
for both recognition and recall. Psychological Re- ory and Language, 35, 118–134.
view, 91, 1–67. INTRAUB, H., & RICHARDSON, M. (1989). Wide-angle
GILOVICH, T. (1991). How we know what isn’t so: The memories of close-up scenes. Journal of Experimen-
fallibility of human reason in everyday life. New tal Psychology: Learning, Memory, and Cognition,
York: The Free Press. 15, 179–187.
GILOVICH, T., VALLONE, R., & TVERSKY, A. (1985). The INTRAUB, H., BENDER, R., & MANGELS, J. (1992). Look-
hot hand in basketball: On the misperception of ran- ing at pictures but remembering scenes. Journal of
dom sequences. Journal of Personality and Social Experimental Psychology: Learning, Memory, and
Psychology, 17, 295–314. Cognition, 18, 180–191.
GLANZER, M., & CLARK, W. H. (1963). The verbal loop ITTELSON, W. H. (1952). The Ames demonstrations in
hypothesis: Binary numbers. Journal of Verbal perception. Princeton, NJ: Princeton Univ. Press.
Learning and Verbal Behavior, 2, 301–309. ITTELSON, W. H., & KILPATRICK, F. P. (1951). Experi-
GLENBERG, A., WILKINSON, A. C., & EPSTEIN, W. (1982). ments in perception. Scientific American, 185, 50–55.
The illusion of knowing: Failure in the self-assess- JACOBY, L. L., ALLEN, L. G., COLLINS, J. C., & LARWILL,
ment of comprehension. Memory & Cognition, 10, L. K. (1988). Memory influences subjective experi-
597–602. ence: Noise judgments. Journal of Experimental Psy-
GRAF, P., SQUIRE, L. R., & MANDLER, G. (1984). The chology: Learning, Memory, and Cognition, 14,
information that amnesic patients do not forget. Jour- 240–247.

JACOBY, L. L., & DALLAS, M. (1981). On the relationship lam, and L. E. Sullivan (Eds.), Memory distortion
between autobiographical memory and perceptual (pp. 329–345). Cambridge, MA: Harvard Univ. Press.
learning. Journal of Experimental Psychology: Gen- KELLEY, C. M., & JACOBY, L. L. (1990). The construction
eral, 110, 306–340. of subjective experience: Memory attributions.
JACOBY, L. L., KELLEY, C. M., & DYWAN, J. (1989). Mind & Language, 5, 49–68.
Memory attributions. In H. L. Roediger & F. I. M. KELLEY, C. M., & JACOBY, L. L. (1996). Adult egocen-
Craik (Eds.), Varieties of memory and conscious- trism: Subjective experience versus analytic bases for
ness: Essays in honour of Endel Tulving (pp. 391– judgment. Journal of Memory and Language, 35,
422). Hillsdale, NJ: Erlbaum. 157–175.
JACOBY, L. L., & WHITEHOUSE, K. (1989). An illusion of KINTSCH, N. (1995). Introduction. In Bartlett, F. C. Re-
memory: False recognition influenced by uncon- membering: A study in experimental and social psy-
scious perception. Journal of Experimental Psychol- chology (pp. xi–xv). Cambridge: Cambridge Univ.
ogy: General, 118, 126–135. Press.
JAMES, W. (1890). Principles of psychology. New York: KOLERS, P. A. (1964). The illusion of movement. Scien-
Dover. tific American, 211, 98–108.
JOHNSON, M. K. (1983). A modular model of memory. KORIAT, A., & GOLDSMITH, M. (in press). Memory meta-
In G. H. Bower (Ed.), The psychology of learning phors and the laboratory/real life controversy: Corre-
and motivation: Advances in research and theory spondence versus storehouse views of memory. Be-
(Vol. 17, pp. 81–123). New York: Academic Press. havioral and Brain Sciences.
JOHNSON, M. K. (1991). Reality monitoring: Evidence KROLL, N. E. A., KNIGHT, R. T., METCALFE, J., WOLF,
from confabulation in organic brain disease patients. E. S., & TULVING, E. (1996). Cohesion failure as a
In G. P. Prigatano & D. L. Schacter (Eds.), Aware- source of memory illusions. Journal of Memory and
Language, 35, 176–196.
ness of deficit after brain injury: Clinical and theo-
retical issues (pp. 176–197). New York: Oxford LINDSAY, D. S., & JOHNSON, M. K. (1989). The eyewit-
ness suggestibility effect and memory for source.
Univ. Press.
Memory and Cognition, 17, 349–358.
JOHNSON, M. K. (1995). The relation between memory
LINDSAY, D. S., & KELLEY, C. M. (1996). Creating illu-
and reality. Paper presented at the 103rd Annual
sions of familiarity in a cued recall Remember/Know
convention of the American Psychological Associa-
paradigm. Journal of Memory and Language, 35,
tion: New York, NY.
197–211.
JOHNSON, M. K., BRANSFORD, J. D., & SOLOMON, S. K.
LINDSAY, D. S., & READ, J. D. (1994). Psychotherapy and
(1973). Memory for tacit implications of sentences.
memories of childhood sexual abuse: A cognitive
Journal of Experimental Psychology, 98, 203–205.
perspective. Applied Cognitive Psychology, 8, 281–
JOHNSON, M. K., FOLEY, M. A., SUENGAS, A. G., & RAYE, 338.
C. L. (1989). Phenomenal characteristics of memo-
LOFTUS, E. F. (1979a). Eyewitness testimony. Cambridge:
ries for perceived and imagined autobiographical Harvard Univ. Press.
events. Journal of Experimental Psychology: Gen-
LOFTUS, E. F. (1979b). The malleability of memory.
eral, 117, 371–376. American Scientist, 67, 312–320.
JOHNSON, M. K., HASHTROUDI, S., & LINDSAY, D. S. LOFTUS, E. F. (1993a). The reality of repressed memories.
(1993). Source monitoring. Psychological Bulletin, American Psychologist, 48, 518–537.
114, 3–28. LOFTUS, E. F. (1993b). Made in memory: Distortions in
JOHNSON, M. K., NOLDE, S. F., & DE LEONARDIS, D. M. recollection after misleading information. In D. L.
(1996). Emotional focus and source monitoring. Medin (Ed.). The psychology of learning and motiva-
Journal of Memory and Language, 35, 135–156. tion: Advances in theory and research (pp. 187–
JOHNSON, M. K., & RAYE, C. L. (1981). Reality monitor- 215). New York: Academic Press.
ing. Psychological Review, 88, 67–85. LOFTUS, E. F., MILLER, D. G., & BURNS, H. J. (1978).
JOHNSON, M. K., RAYE, C. L., WANG, A. Y., & TAYLOR, Semantic integration of verbal information into a vi-
T. H. (1979). Fact and fantasy: The role of accuracy sual memory. Journal of Experimental Psychology:
and variability in confusing imaginations with per- Human Learning and Memory, 4, 19–31.
ceptual experiences. Journal of Experimental Psy- LOFTUS, E. F., & PALMER, J. C. (1974). Reconstruction
chology: Human Learning and Memory 5, 229–240. of automobile destruction: An example of the inter-
JOHNSON, M. K., TAYLOR, T., & RAYE, C. L. (1977). action between language and memory. Journal of
Fact and fantasy: The effects of internally generated Verbal Learning and Verbal Behavior, 13, 585–589.
events on the apparent frequency of externally gener- LYNN, S. J., & NASH, M. R. (1994). Truth in memory:
ated events. Memory and Cognition, 5, 116–122. Ramifications for psychotherapy and hypnotherapy.
KAMMEN, M. (1995). Some patterns and meanings of American Journal of Hypnosis, 36, 194–208.
memory distortion in American history. In D. L. MANDLER, G. (1980). Recognizing: The judgment of pre-
Schacter, J. T. Coyle, G. D. Fischbach, M. M. Mesu- vious occurrence. Psychological Review, 87, 252–271.

MEMORY ILLUSIONS 99
MCCOLLOUGH, D. (1965). Color adaptation of edge-detec- PAYNE, D. G., ELIE, C. J., BLACKWELL, J. M., &
tors in the human visual system. Science, 149, 1115– NEUSCHATZ, J. S. (1996). Memory illusions: Recall-
1116. ing, recognizing and recollecting events that never
MCDERMOTT, K. B. (1996). The persistence of false mem- occurred. Journal of Memory and Language, 35,
ories in list recall. Journal of Memory and Language, 261–285.
35, 212–230. PIATELLI-PALMARINI, M. (1994). Inevitable illusions: How
MCGEOCH, J. A. (1932). Forgetting and the law of disuse. mistakes of reason rule our minds. New York: Wiley.
Psychological Review, 39, 352–370. PLOUS, S. (1993). The psychology of judgment and deci-
MELCHER, J. M., & SCHOOLER, J. W. (1996). The misre- sion making. New York: McGraw–Hill. Inc.
membrance of wines past: Verbal and perceptual ex- POSNER, M. I., & KEELE, S. W. (1970). Retention of
pertise differentially mediate verbal overshadowing abstract ideas. Journal of Experimental Psychology,
of taste memory. Journal of Memory and Language, 83, 304–308.
35, 231–245. POSTMAN, L., & UNDERWOOD, B. J. (1973). Critical issues
MELTON, A. W., & IRWIN, J. M. (1940). The influence of in interference theory. Memory and Cognition, 1,
degree of interpolated learning on retroactive inhibi- 19–40.
tion and the overt transfer of specific responses. RAJARAM, S., & ROEDIGER, H. L. (in press). Remembering
American Journal of Psychology, 53, 173–203. and knowing as states of consciousness during recol-
MILLER, G. A. (1956a). Human memory and the storage lection. In J. D. Cohen and J. W. Schooler (Eds.),
of information. IRE Transactions on Information Scientific approaches to consciousness. Cambridge,
Theory, IT-2, 129–137. MA: MIT Press.
MILLER, G. A. (1956b). The magical number seven plus REINITZ, M. T., LAMMERS, W. J., & COCHRAN, B. P.
or minus two: Some limits on our capacity for pro- (1992). Memory-conjunction errors: Miscombination
cessing information. Psychological Review, 63, 81–97. of stored stimulus features can produce illusions of
MITCHELL, K. J., & ZARAGOZA, M. S. (1996). Repeated memory. Memory and Cognition, 20, 1–11.
exposure to suggestion and false memory: The role REINITZ, M. T., MORRISEY, J., & DEMB, J. (1994). The
of contextual variability. Journal of Memory and role of attention in face encoding. Journal of Experi-
Language, 35, 246–260. mental Psychology: Learning, Memory, and Cogni-
MOSCOVITCH, M. (1995). Confabulation. In D. L. Schacter, tion, 20, 161–168.
J. T. Coyle, G. D., & Fischbach, M. M., Mesulam, REINITZ, M. T., VERFAELLIE, M., & MILBERG, W. P.
and L. E. Sullivan (Eds.), Memory distortion (pp. (1996). Memory conjunction errors in normal and
226–251). Cambridge, MA: Harvard Univ. Press. amnesic subjects. Journal of Memory and Language,
MÜLLER, G. E., & PILZECKER, A. (1900). Experimentalle 35, 286–299.
Beitrage zur Lehre vom Gedachtnis. Zeitschrift fur REYNA, & BRAINERD (1995). Fuzzy-trace theory: An in-
Psychologie, 1, 1–300. terim synthesis. Learning and Individual Differences,
MUNSTERSBERG, H. (1908). On the witness stand: Essays 7, 1–75.
on psychology and crime. New York: Clark, Board- RILEY, D. (1964). Memory for Form. In L. Postman (Ed.),
man, Doubleday. Psychology in the making: Histories of selected re-
NEISSER, U. (1968). The processes of vision. Scientific search problems. New York: Knopf.
American, 219, 204–214. ROEDIGER, H. L., JACOBY, J. D., & MCDERMOTT, K. B.
NEISSER, U. (1967). Cognitive Psychology. New York: (1996). Misinformation effects in recall: Creating
Appleton–Century–Crofts. false memories through repeated retrieval. Journal
NELSON, T. O., & LEONESIO, R. J. (1988). Allocation of of Memory and Language, 35, 300–318.
self-paced study time and the ‘‘labor-in-vain effect.’’ ROEDIGER, H. L., & MCDERMOTT, K. B. (1995). Creating
Journal of Experimental Psychology: Learning, false memories: Remembering words not presented
Memory, and Cognition, 14, 676–686. in lists. Journal of Experimental Psychology: Learn-
NELSON, T. O. (Ed.) (1992). Metacognition: Core reading, Memory, and Cognition, 21, 803–814.
ings. Boston, MA: Allyn and Bacon. ROEDIGER, H. L., WHEELER, M. A., & RAJARAM, S.
OPPEL (1854). Uber geometrisch-optische Tauschungen. (1993). Remembering, knowing and reconstructing
Jahresbericht des Frankfurter Vereins 37–47. the past. In D. L. Medin (Ed.), The psychology of
OWENS, J., BOWER, G. H., & BLACK, J. B. (1979). The learning and motivation: Advances in theory and
‘‘soap-opera’’ effect in story recall. Memory and research (Vol. 30, pp. 97–134). New York: Aca-
Cognition, 7, 185–191. demic Press.
PAUL, I. H. (1959). Studies in remembering: The repro- ROSS, D. F., READ, J. D., & TOGLIA, M. P. (1994). Adult
duction of connected and extended verbal material. eyewitness testimony: Current trends and develop-
Psychological Issues, 1, (Monograph 2), 1–152. ments. Cambridge: Cambridge Univ. Press.
PAUL, L. M. (1979). Two models of recognition memory: SCHACHTER, S., & SINGER, J. (1962). Cognitive, social,
A test. Journal of Experimental Psychology: Human and physiological determinants of emotional states.
Learning and Memory, 5, 45–51. Psychological Review, 69, 379–399.

SCHACTER, D. L. (1995). Memory distortion: History and TULVING, E., & THOMSON, D. M. (1973). Encoding speci-
current status. In D. L. Schacter, J. T., Coyle, G. D., ficity and retrieval processes in episodic memory.
Fischbach, M. M., Mesulam L. E., and Sullivan Psychological Review, 80, 352–373.
(Eds.), Memory distortion (pp. 1–43). Cambridge, TVERSKY, A., & KAHNEMAN, D. (1974). Judgment under
MA: Harvard Univ. Press. uncertainty: Heuristics and biases. Science, 185,
SCHACTER, D. L., CURRAN, T., GALLUCIO, L. D., MIL- 1124–1130.
BERG, W. P., & BATES, J. (in press). False recognition UNDERWOOD, B. J. (1957). Interference and forgetting.
and the right frontal lobe: A case study. Neuropsy- Psychological Review, 64, 49–60.
chologia. UNDERWOOD, B. J. (1965). False recognition produced by
SCHACTER, D. L., VERFAELLIE, M., & PRADERE, D. implicit verbal responses. Journal of Experimental
(1996). The neuropsychology of memory illusions: Psychology, 70, 122–129.
False recall and recognition in amnesic patients. UNDERWOOD, B. J., KAPELAK, S. M., and MALMI, R. A.
Journal of Memory and Language, 35, 319–334. (1976). Integration of discrete verbal units in recog-
SCHOOLER, J. W., & ENGSTLER-SCHOOLER, T. Y. (1990). nition memory. Journal of Experimental Psychology:
Verbal overshadowing of visual memories: Some Human Learning and Memory, 2, 293–300.
things are better left unsaid. Cognitive Psychology, UNDERWOOD, B. J., & ZIMMERMAN, J. (1973). The sylla-
22, 36–71. ble as a source of error in multisyllable word recogni-
SCHUDSON (1995). Dynamics of distortion in collective tion. Journal of Verbal Learning and Verbal Behav-
memory. In D. L. Schacter, J. T. Coyle, G. D. Fisch- ior, 12, 701–706.
bach, M. M. Mesulam, and L. E. Sullivan (Eds.), WARRINGTON, E. K., & WEISKRANTZ, L. (1968). Anew
Memory distortion (pp. 346–376). Cambridge, MA: method of testing long-term retention with special
Harvard Univ. Press. reference to amnesic patients. Nature, 217, 972–974.
SHEEHAN, P. W. (1988). Memory distortion in hypnosis. WEAVER, C. A. (1991). Constraining factors in the cali-
International Journal of Experimental and Clinical bration of comprehension. Journal of Experimental
Hypnosis, 36, 296–311. Psychology: Learning, Memory, and Cognition, 16,
SHEPARD, R. N. (1981). Psychophysical complementarity. 214–222.
In M. Kubovy & J. Pomerantz (Eds.), Perceptual WEINGARDT, K. R., TOLAND, H. K. & LOFTUS, E. F.
(1994). Reports of suggested memories: Do people
organization (pp. 279–341). Hillsdale, NJ: Erlbaum.
truly believe them? In D. F. Ross, J. D. Read, and
SHIFFRIN, R. M., HUBER, D. E., & MARINELLI, K. (1995).
M. P. Toglia (Eds.), Adult eyewitness testimony: Cur-
Effects of category length and strength on familiarity
rent trends and developments (pp. 3–26). NY: Cam-
in recognition. Journal of Experimental Psychology:
bridge Univ. Press.
Learning, Memory, and Cognition, 21, 267–287.
WHEELER, M. A., & ROEDIGER, H. L. (1992). Disparate
SMITH, M. C. (1983). Hypnotic memory enhancement of
effects of repeated testing: Reconciling Ballard’s
witnesses: Does it work? Psychological Bulletin, 94,
(1913) and Bartlett’s (1932) results. Psychological
387–407.
Science, 3, 240–245.
SOLSO, R. L., & MCCARTHY, J. E. (1981). Prototype for-
WHITEHOUSE, W. G., ORNE, E. C., ORNE, M. T., &
mation of faces: A case of pseudomemory. British DINGES, D. F. (1991). Distinguishing the source of
Journal of Psychology, 72, 499–503. memories reported during prior waking and hypnotic
SPIRO, R. J. (1980). Accommodative reconstruction in recall attempts. Applied Cognitive Psychology, 5,
prose recall. Journal of Verbal Learning and Verbal 51–59.
Behavior, 19, 84–95. WHITTLESEA, B. W. A. (1993). Illusions of familiarity.
SULIN, R. A., & DOOLING, D. J. (1974). Intrusion of a Journal of Experimental Psychology: Learning,
thematic idea in retention of prose. Journal of Exper- Memory and Cognition, 6, 1235–1253.
imental Psychology, 103, 255–262. WHITTLESEA, B. W. A., JACOBY, L. L., & GIRARD, K.
TITCHENER, E. B. (1928). A text-book of psychology. New (1990). Illusions of immediate memory. Journal of
York: Macmillan. Memory and Language, 29, 716–732.
TREISMAN, A., & SCHMIDT, H. (1982). Illusory conjunc- WOOD, G. (1978). The knew-it-all-along effect. Journal
tions in the perception of objects. Cognitive Psychol- of Verbal Learning and Verbal Behavior, 4, 345–353.
ogy, 14, 107–141. WULF, F. (1922). Uber die veranderung von Vorstellungen.
TULVING, E. (1962). Subjective organization in free recall Psychologische Forschung, 1, 333–373.
of ‘‘unrelated’’ words. Psychological Review, 69, ZARAGOZA, M. S., & LANE, S. M. (1994). Source misattribu-
344–354. tions and the suggestibility of eyewitness memory.
TULVING, E. (1964). Intratrial and intertrial retention: Journal of Experimental Psychology: Learning, Mem-
Notes towards a theory of free recall verbal learning. ory, and Cognition, 20, 934–945.
Psychological Review, 71, 219–237.
TULVING, E. (1985). Memory and consciousness. Cana- (Received December 7, 1995)
dian Psychologist, 26, 1–12. (Revision received December 21, 1995)

Memory Illusions

Uploaded by

Copyright:

Available Formats

Memory Illusions

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

Memory Illusions

Uploaded by

Copyright:

Available Formats

JOURNAL OF MEMORY AND LANGUAGE 35, 76–100 (1996)

ARTICLE NO. 0005

HENRY L. ROEDIGER III

AID JML 2467 / a002$$$101 04-10-96 17:23:17 jmla AP: JML

torical overview of their discovery and impor-

AID JML 2467 / a002$$$102 04-10-96 17:23:17 jmla AP: JML

AID JML 2467 / a002$$$102 04-10-96 17:23:17 jmla AP: JML

AID JML 2467 / a002$$$102 04-10-96 17:23:17 jmla AP: JML

AID JML 2467 / a002$$$103 04-10-96 17:23:17 jmla AP: JML

AID JML 2467 / a002$$$103 04-10-96 17:23:17 jmla AP: JML

AID JML 2467 / a002$$$103 04-10-96 17:23:17 jmla AP: JML

AID JML 2467 / a002$$$103 04-10-96 17:23:17 jmla AP: JML

AID JML 2467 / a002$$$103 04-10-96 17:23:17 jmla AP: JML

AID JML 2467 / a002$$$104 04-10-96 17:23:17 jmla AP: JML

AID JML 2467 / a002$$$104 04-10-96 17:23:17 jmla AP: JML

AID JML 2467 / a002$$$104 04-10-96 17:23:17 jmla AP: JML

AID JML 2467 / a002$$$105 04-10-96 17:23:17 jmla AP: JML

AID JML 2467 / a002$$$105 04-10-96 17:23:17 jmla AP: JML

AID JML 2467 / a002$$$105 04-10-96 17:23:17 jmla AP: JML

AID JML 2467 / a002$$$105 04-10-96 17:23:17 jmla AP: JML

AID JML 2467 / a002$$$105 04-10-96 17:23:17 jmla AP: JML

AID JML 2467 / a002$$$106 04-10-96 17:23:17 jmla AP: JML

AID JML 2467 / a002$$$106 04-10-96 17:23:17 jmla AP: JML

AID JML 2467 / a002$$$106 04-10-96 17:23:17 jmla AP: JML

AID JML 2467 / a002$$$106 04-10-96 17:23:17 jmla AP: JML

AID JML 2467 / a002$$$107 04-10-96 17:23:17 jmla AP: JML

AID JML 2467 / a002$$$107 04-10-96 17:23:17 jmla AP: JML

AID JML 2467 / a002$$$107 04-10-96 17:23:17 jmla AP: JML

AID JML 2467 / a002$$$107 04-10-96 17:23:17 jmla AP: JML

You might also like