NEUROSCIENCE UPDATE ■

Mistaken Memories: importantly, the theme word is never

presented (Fig. 1). People frequently
claim to remember the theme word
Remembering Events from the learning list, even though it
was not presented, and their level of
That Never Happened confidence in these false memories
can be as high as for their true mem-
BRIAN GONSALVES and KEN A. PALLER ories. Apparently, these false memo-
Department of Psychology and Institute for Neuroscience ries do not reflect a vague sense of
Northwestern University
Evanston, Illinois
familiarity but rather resemble the
full-blown experience of accurate
remembering.
Our memories can be accurate, but they are not always accurate. In line with findings that people
Eyewitness testimony, for example, is notoriously unreliable. Insights can mistakenly claim that a prior
into both veridical and false remembering have come from recent inves- event occurred while remaining
tigations of memory distortion. Behavioral measures have been used highly confident, results from brain
to demonstrate false memory phenomena in the laboratory, and neu- imaging have demonstrated similar
roimaging measures have been used to provide clues about the rele- patterns of brain activity for true and
vant events in the brain that support remembering versus misremem- false memories, using positron emis-
bering. A central category of misremembering results from confusion sion topography (PET), functional
between memories for perceived and imagined events, which may magnetic resonance imaging (fMRI),
result from overlap between particular features of the stored informa- and event-related potential (ERP)
tion comprising memories for perceived and imagined events. NEU- methods for monitoring the brain in
ROSCIENTIST 8(5):391–395, 2002. DOI: 10.1177/107385802236964 action (Schacter and others 1996;
Düzel and others 1997; Schacter and
KEY WORDS Memory, False memory, Neuroimaging, Event-related potentials, ERPs others 1997). Other evidence, how-
ever, indicates that the brain treats
true and false memories differently.
Each one of us has a life story, a col- basis in the brain, as reviewed below, In two studies, veridical memories
lection of personally experienced can provide many insights into this tended to be rated as including more
events, that we can bring to mind on process of reconstruction, thus help- perceptual detail than false memo-
command. These episodic memories ing to clarify the inner workings of ries (Mather and others 1997;
are stored in the brain such that we memory in general, both when it Norman and Schacter 1997). These
can recall an event experienced works well and when it fails. experiential differences suggest that
either a few minutes earlier or many False memories take many forms, brain activity associated with true
years ago, and in doing so re-experi- and false remembering can be pro- and false memories should also dif-
ence multiple aspects of that event, duced in a laboratory setting using fer. Indeed, differential activation of
including a variety of perceptual several different methods. One time- prefrontal cortex was found when
details, thoughts, observations, and honored method for studying human extended periods of false remember-
emotions. The brain networks that memory involves requiring people to ing and of accurate remembering
enable this memory retrieval to hap- learn a list of words. When recogni- were compared, perhaps reflecting
pen, however, do not include exact tion is subsequently tested for those post-retrieval monitoring processes
records of those past events. Instead, words, people occasionally make (Schacter and others 1996; Johnson
episodic remembering is a recon- “false alarms,” indicating that they and others 1997; Schacter and others
structive process, whereby one must remember a word that was not, in 1997). However, no definitive con-
make use of stored information that fact, present on the original list. The nection has been established between
is often limited and incomplete. number of false alarms can be sub- those prefrontal differences and the
Because of this reconstructive nature stantially boosted using a method experiential differences observed in
of retrieval, memories are subject to pioneered by Deese (1959) and behavioral studies. In contrast,
distortion (Bartlett 1932; Schacter extended by Roediger and greater activity in the auditory cortex
1995; Schacter and others 1998). The McDermott (1995), in which the was found during memory retrieval
study of memory distortion and its learning list includes groups of for true than false memories for
words that are each closely associat- heard words, which may reflect the
ed with one theme word. For exam- retrieval of auditory information
Address correspondence to: Brian Gonsalves, about words that were actually heard
Massachusetts Institute of Technology, Depart-
ple, people will be asked to remem-
ment of Brain and Cognitive Sciences, NE20-51, ber a list of words that are all associ- (Schacter and others 1996). In an
Cambridge, MA 02140 (e-mail: [email protected]). ated with the theme word sleep—but fMRI study, both true and false

Volume 8, Number 5, 2002 THE NEUROSCIENTIST 391

Copyright © 2002 Sage Publications
ISSN 1073-8584

Downloaded from http://nro.sagepub.com at NORTHWESTERN UNIV LIBRARY on August 23, 2009

 imagined events tend to differ, there
are similarities. Critically, when
memory records for imaginations
and perceptions extensively overlap
in features, the two tend to be con-
fused, and we may then mistake
imagination for reality.
We recently developed a new
procedure for generating reality-
monitoring errors, which allowed us
to obtain neural correlates of true
and false remembering (Gonsalves
and Paller 2000b). As shown in
Figure 2, subjects in our experiment
viewed a series of object names and
generated a visual image of each cor-
responding object. For half of the
words, a picture of the object was
also presented 2 seconds after the
word, but the pictures required no
response. Subsequently, in a surprise
memory test, subjects heard a series
Fig. 1. An example of the Deese-Roediger-McDermott procedure for inducing false
memories, in which participants hear a series of words associated with a nonpresented
of spoken words. Some of these
theme word. Later, participants often claim to have heard the nonpresented theme words were seen earlier, some were
word. seen with the corresponding picture,
and some were not seen at all. For
each word, subjects decided whether
memories were associated with hip- ory tests. Another arguably more nat- or not they saw a picture of the
pocampal activity, whereas accurate ural way in which false remembering named object earlier. This paradigm
retrieval elicited relatively more can be induced is by leading people thus creates substantial overlap in
activity in the parahippocampal cor- to believe that an event that they memory attributes for imagined
tex, perhaps reflecting the greater imagined had actually occurred. This objects and for perceived objects,
reactivation or re-experiencing of is known as an error in “reality mon- thereby eliciting frequent reality-
perceptual details (Cabeza and oth- itoring” because the individual con- monitoring errors. Subjects claimed
ers 2001). Differences in neural fuses memories for imagined and to have seen pictures of 30% of the
activity associated with true and perceived events. These errors may items that they had only seen as a
false memory were also observed in arise as a consequence of similarities word during encoding, whereas they
several ERP studies (Walla and oth- between how imagined and per- only claimed to have seen 9% of the
ers 2000; Curran and others 2001; ceived events are stored in the brain, new items. Brain activity recorded
Miller and others 2001; Nessler and and between the event features that during the retrieval phase of this
others 2001). In one particularly ele- are reactivated during retrieval experiment differed between true
gant ERP study, word learning was (Johnson and Raye 1981; Johnson and false memories. Specifically,
accomplished using lateralized visu- and others 1993). People are general- from 900 to 1200 ms after word
al presentations such that contralat- ly quite competent at this form of onset, ERP responses at parietal and
eral visual areas were preferentially source discrimination, given that occipital scalp locations were more
engaged (Fabiani and others 2000). several cues can be used to distin- positive for true memories than for
When recognition was subsequently guish between perceived and imag- false memories. In interpreting this
tested using central presentations, ined events. For example, memories effect, we were able to build on our
true memories of presented words for perceived events tend to include previous observation of an ERP cor-
showed appropriately lateralized more perceptual detail than do mem- relate of visual imagery during mem-
ERP activity, whereas falsely recog- ories for imagined events (Johnson ory retrieval (Gonsalves and Paller
nized theme words did not. Only and others 1988; Mather and others 2000a). Participants in this previous
words actually presented during 1997; Norman and Shacter 1997). experiment studied sets of spoken
learning elicited brain activity indica- Memories for imagined events, on words either with or without instruc-
tive of a sensory memory trace. the other hand, often include more tions to generate visual images, and
False memories do not occur only records of the cognitive operations they later recognized those words
when, as in the aforementioned stud- required for internally generating the encoded with visual imagery better.
ies, extensive semantic associations information in question. Although ERPs from occipital scalp locations
between learned words and a theme subjective characteristics of mnemon- differed for these two types of words,
word engender false alarms on mem- ic traces associated with real and and these differences were interpret-

392 THE NEUROSCIENTIST Cognitive Neuroscience of False Memory

Downloaded from http://nro.sagepub.com at NORTHWESTERN UNIV LIBRARY on August 23, 2009

Fig. 2. Procedure used by Gonsalves and Paller (2000b) for generating false memories, as outlined in the text. Brain potentials
observed in the two parts of this experiment are shown on the right. (From Paller 2002, adapted with permission; Figure 1 and the left
side of this figure from de Schipper S. 2000, December 2. Sterke verbeelding: valse herinnering blijkt uithersengolven [Strong
imagination: false memory is evident from brainwaves]. NRC Handelsblad, p. 47. Copyright 2000 by NRC Handelsblad. Adapted by
permission.)

ed as reflecting the enhanced phase items that were subsequently remembering. Preliminary results
retrieval of visual object representa- remembered elicited a more positive from an event-related-fMRI study of
tions for words encoded using visual ERP than did items later forgotten. In false memory using this same design
imagery. Similarly, we interpreted our experiment, we sorted study- suggest that imagery-associated
the ERP difference between true and phase ERP responses to pictures activity in the precuneus, anterior
false memories as reflecting the according to whether they were later cingulate cortex, and inferior parietal
enhanced retrieval of visual percep- remembered or forgotten, and we gyrus likewise promotes later false
tual information about object pictures observed a subsequent memory ERP remembering (Gonsalves and others
for true compared to false memories. effect widely distributed across the 2001). Precuneus activation has been
This interpretation accords well with scalp and likely reflecting the activi- associated previously with memory-
the reality-monitoring framework, ty of multiple brain regions. We also related imagery (Fletcher and others
which predicts that memories for recorded ERP responses to study 1995, 1996), and precuneus, anterior
perceptions should include, on aver- words not followed by a picture and cingulate, and inferior parietal gyrus
age, more perceptual detail than sorted them according to whether the are areas engaged during both visual
memories for imaginations (Johnson person later falsely claimed to have imagery and visual perception
and Raye 1981; Johnson and others seen the picture. ERPs associated (Kosslyn and Thompson 2000).
1988; Johnson and others 1993). with later false remembering were These results are in line with our ear-
In addition to these differences more positive than those associated lier interpretations that more elabo-
between true and false memories, with later correct rejections, most rate or vivid visual imagery during
ERPs recorded during the study reliably from 600 to 900 ms at occip- the imagery task precipitates a later
phase were computed according to ital and parietal scalp locations. false memory in this paradigm.
whether or not items were later rec- Given our prediction that vivid visu- We interpreted these neurophysio-
ognized. Neural differences between al imagery can promote subsequent logical harbingers of false memories
later remembered and later forgotten reality-monitoring errors, such that as reflections of the association
items, so called subsequent memory an item presented only as a word between vivid visual imagery and
effects, have been observed using could later be falsely remembered as later reality-monitoring errors, an
both ERP and event-related fMRI a picture, we interpreted these effects account consonant with prior work
methods (for review, see Paller and as a reflection of strong visual on the effects of imagery. For exam-
Wagner, 2002). Typically, study- imagery that promoted later false ple, when people actively imagine a

Volume 8, Number 5, 2002 THE NEUROSCIENTIST 393

Downloaded from http://nro.sagepub.com at NORTHWESTERN UNIV LIBRARY on August 23, 2009

concocted childhood event, the like- monitoring mechanism may also Gonsalves B, Reber PJ, Gitelman DR, Parrish
lihood that subjects later believe the provide a model for the creation of TB, Mesulam M-M, Paller KA. 2001.
Event-related fMRI reveals brain activity
event had actually happened increas- false childhood memories through at encoding that predicts true and false
es (Hyman and Pentland 1996). suggestive questioning (i.e., Hyman memories for visual objects. Soc Neurosci
Repeatedly imagining an event and Pentland 1996). In this case, Abstr 27: Program No. 236.9.
increases subjective ratings of the when people actively imagine a fab- Hyman IE, Pentland J. 1996. The role of men-
tal imagery in the creation of childhood
likelihood that the event occurred, a ricated childhood event, they some- memories. J Mem Language 35:101–17.
phenomenon called imagination times come to believe that the event Johnson MK, Hashtroudi S, Lindsay SD.
inflation (Garry and others 1996). actually happened. Continuing 1993. Source monitoring. Psychol Bull
Encoding words with an emphasis on research on memory distortion in the 114:3–28.
imagery increases the number of laboratory, using the methods of cog- Johnson MK, Nolde SF, Mather M, Kounios J,
Schacter DL. 1997. The similarity of brain
errors when people later attempt to nitive neuroscience, can be expected activity associated with true and false
remember whether particular items to shed additional light on the brain recognition memory depends on test for-
were seen as words or pictures mechanisms whereby imagined mat. Psychol Sci 8:250–57.
(Durso and Johnson 1980; Lane and events may later become confused Johnson MK, Raye CL. 1981. Reality moni-
Zaragoza 1995). The constructive with actual events, leading to false toring. Psychol Rev 88:67–85.
Johnson MK, Suengas AG, Foley MA, Raye
memory framework of Schacter and memories outside the laboratory. CL. 1988. Phenomenal characteristics of
colleagues (Schacter and others 1998) memories for perceived and imagined
suggests a general neural mechanism References autobiographical events. J Exp Psychol
for these reality-monitoring errors. Gen 117:371–6.
Bartlett FC. 1932. Remembering. Cambridge, Kosslyn SM, Thompson WL. 2000. Shared
This framework takes as a starting mechanisms in visual imagery and visual
UK: Cambridge University Press.
point the idea that event memory Cabeza R, Rao SM, Wagner AD, Mayer AR, perception: insights from cognitive neuro-
traces consist of distributed networks Schacter DL. 2001. Can medial temporal science. In: Gazzaniga MS, editor. The
localized in cortical regions involved lobe regions distinguish true from false? new cognitive neurosciences. Cambridge
in initially encoding an event and An event-related fMRI study of veridical (MA): MIT Press. p 975–85.
and illusory recognition memory. Proc Lane SM, Zaragoza MS. 1995. The recollec-
that these networks are reactivated tive experience of cross-modality confu-
Natl Acad Sci U S A 98:4805–10.
when the memory is retrieved Curran T, Schacter DL, Johnson MK, Spinks sion errors. Mem Cogn 23:607–10.
(Squire 1987; McClelland and others R. 2001. Brian potentials reflect behav- Loftus EF, Feldman J, Dashiell R. 1995. The
1995; Paller 2002). Two highly simi- ioral differences in true and false recogni- reality of illusory memories. In: Schacter
lar events will have a large amount of tion. J Cogn Neurosci 13:201–16. DL, editor. Memory distortion. Cambridge
Deese J. 1959. On the prediction of occur- (MA): Harvard University Press. p 47–68.
overlap in their distributed features, rence of particular verbal intrusions in Mather M, Henkel LA, Johnson MK. 1997.
such that events may be difficult to immediate recall. J Exp Psychol 58:17–22. Evaluating characteristics of false memo-
distinguish from one another during Durso FT, Johnson MK. 1980. The effects of ries: remember/know judgements and
retrieval. This general mechanism orienting tasks on recognition, recall, and memory characteristics questionnaire
modality confusion of pictures and words. compared. Mem Cogn 25:826–37.
can be applied to the occurrence of McClelland JL, McNaughton BL, O’Reilly
J Verbal Learn Verbal Behav 19:416–29.
reality-monitoring errors, and in par- Düzel E, Yonelinas AP, Mangun GR, Heinze RC. 1995. Why there are complementary
ticular the confusion between visual HJ, Tulving E. 1997. Event-related brain learning systems in the hippocampus and
perceptions and visual imaginations. potential correlates of two states of con- neocortex: insights from the successes and
It is well established that visual scious awareness in memory. Proc Natl failures of connectionist models of learn-
Acad Sci U S A 94:5973–78. ing and memory. Psychol Rev 102:419–
imagery activates many of the same Fabiani M, Stadler MA, Wessels PM. 2000. 57.
brain areas active during visual per- True but not false memories produce a Miller AR, Baratta C, Wynveen C, Rosenfeld
ception (Kosslyn and Thompson sensory signature in human lateralized brain JP. 2001. P300 latency, but not amplitude
2000), which suggests that images potentials. J Cogn Neurosci 12:941–9. or topography, distinguishes between true
Fletcher PC, Frith CD, Baker SC, Shallice T, and false recognition. J Exp Psychol Learn
and perceptions can have overlap- Mem Cogn 27:354–61.
Frackowiak RSJ, Dolan RJ. 1995. The
ping memory representations in the mind’s eye—precuneus activation in Nessler D, Mecklinger A, Penney TB. 2001.
cerebral cortex. Particularly vivid memory-related imagery. Neuroimage Event related brain potentials and illusory
visual imagery of an event might 2:195–200. memories: the effects of differential
lead to the formation of a memory Fletcher PC, Shallice T, Frith CD, Frackowiak encoding. Cogn Brain Res 10:283–301.
RSJ, Dolan RJ. 1996. Brain activity during Norman KA, Schacter DL. 1997. False recog-
trace that is indistinguishable from memory retrieval: the influence of nition in young and older adults: exploring
the trace that would have been imagery and semantic cueing. Brain the characteristics of illusory memories.
formed had the event actually been 119:1587–96. Mem Cogn 25:838–48.
perceived. These general mecha- Garry M, Manning CG, Loftus EF, Sherman Paller KA. 2002. Cross-cortical consolidation
nisms might be applied to false SJ. 1996. Imagination inflation: imagining as the core defect in amnesia: prospects
a childhood event inflates confidence that for hypothesis-testing with neuropsychol-
memory phenomena such as the it occurred. Psychon Bull Rev 3:208–14. ogy and neuroimaging. In: Squire LR,
effects of misleading questions on Gonsalves B, Paller KA. 2000a. Brain poten- Schacter DL, editors. The neuropsycholo-
eyewitness testimony, whereby post- tials associated with recollective processing gy of memory. 3rd ed. New York:
event suggestions about what hap- of spoken words. Mem Cogn 28:321–30. Guilford. p 73–87.
Gonsalves B, Paller KA. 2000b. Neural events Paller KA, Wagner Ad. In press. Monitoring
pened during an event lead to distor- that underlie remembering something that the transformation of experience into
tion in people’s accounts of events never happened. Nat Neurosci 3:1316–21. memory. Trends Cogn Sci 6:93–102.
(Loftus and others 1995). This reality-

394 THE NEUROSCIENTIST Cognitive Neuroscience of False Memory

Downloaded from http://nro.sagepub.com at NORTHWESTERN UNIV LIBRARY on August 23, 2009

Roediger HL, McDermott KB. 1995. Creating false recognition: an event-related fMRI veridical and illusory recognition memo-
false memories: remembering words not study. Neuroimage 6:259–69. ry: evidence from positron emission
presented in lists. J Exp Psychol Learn Schacter DL, Norman KA, Koutstaal W. 1998. tomography. Neuron 17:267–74.
Mem Cogn 21:803–14. The cognitive neuroscience of construc- Squire LR. 1987. Memory and brain. New
Schacter DL, editor. 1995. Memory distortion. tive memory. Annu Rev Psychol York: Oxford University Press.
Cambridge (MA): Harvard University Press. 49:289–318. Walla P, Endl W, Lindinger G, Deecke L, Lang
Schacter DL, Buckner RL, Koustaal W, Dale Schacter DL, Reiman E, Curran T, Yun LS, WF. 2000. False recognition in a verbal
AM, Rosen BR. 1997. Late onset of ante- Bandy D, McDermott KB, and others. memory task: an event-related potential
rior prefrontal activity during true and 1996. Neuroanatomical correlates of study. Cogn Brain Res 9:41–4.

Volume 8, Number 5, 2002 THE NEUROSCIENTIST 395

Downloaded from http://nro.sagepub.com at NORTHWESTERN UNIV LIBRARY on August 23, 2009