Downloaded from [Link].

org on May 17, 2025 - Published by Cold Spring Harbor Laboratory Press

Commentary

Searching for the neurobiology

of the misinformation effect
Elizabeth F. Loftus
University of California–Irvine, Irvine, California 92697, USA

For more than 30 years, I have been studying a phenomenon McDermott 1995). In this procedure, subjects study a list of re-
called the misinformation effect (Loftus and Palmer 1974; Loftus lated words such as sour, candy, bitter, sugar, good, taste, tooth, nice,
1975; Loftus and Hoffman 1989). The studies that we have done honey, soda, chocolate, cake, heart, tart, and pie. When later tested,
show how readily memory can become skewed when people are subjects often think they heard or saw certain non-presented
fed misinformation. They used a deceptively simple procedure. items (sweet, in this case). The procedure reliably led most sub-
Subjects first see a complex event, such as a simulated automo- jects to produce a goodly number of false memories; that is, sub-
bile accident. Next, half the subjects receive misleading informa- jects would describe the falsely remembered word in detail, and
tion about the accident while the others get no misinformation. gave every indication that they were remembering a real event.
Finally, all subjects try to remember the original accident. In one Thus, virtually all of the neuroimaging studies of true and
actual study using this paradigm, subjects saw an accident and false memories used the DRM paradigm, or some variant (e.g.,
later some of them received misinformation about the traffic sign Cabeza et al. 2001). Therefore, most of what we learned about the
used to control traffic. The misled subjects got the false sugges- neurobiology of true and false memories was about the neurobi-
tion that the stop sign that they had actually seen was a yield ology of true and false memories of words, or at least about
sign. When asked later what kind of traffic sign they remembered memories that were relatively pallid in nature.
seeing at the intersection, those who had been given the false Enter the work of Okado and Stark (2005). First, and impor-
suggestion tended to adopt it as their memory, and now claimed tantly, they developed materials that were far closer to the com-
that they had seen a yield sign (see Fig. 1). Those who had not plex visual scenarios used in most of the misinformation re-
received the phony information had much more accurate memo- search. Moreover, their procedure was one in which a decent
ries (see Loftus 1979). Taken together, the hundreds, if not thou- number of false memories could be produced in a given subject,
sands, of similar studies show that misinformation can change thereby allowing meaningful comparisons with the true memo-
an individual’s recollection in predictable, and sometimes very ries. And in the process they have taught us something about the
powerful, ways. In this issue of Learning & Memory, Okado and misinformation effect that can relate to the hundreds of behav-
Stark (2005) shed light on the neural mechanisms underlying the ioral studies that have been done on the topic.
misinformation effect. Okado and Stark (2005) used eight complex vignettes each
Misinformation can contaminate our memories when we of which were shown to participants through 50 color slides. One
talk to other people, when we are interrogated in a suggestive vignette, for example, depicted a man stealing a girl’s wallet and
fashion, or when we see biased media coverage about some event subsequently trying to hide. Thus, the materials were far more
that we may have experienced ourselves. After more than two complex and “realistic” than had been used in the typical neu-
decades exploring the power of misinformation, we have learned roimaging memory study. Subjects were scanned while they
a great deal about the conditions that make people especially watched these original vignettes. Later, during the misinforma-
susceptible to its damaging influence. We have learned, for ex- tion phase, subjects watched the vignettes again, under the belief
ample, that memories are more easily modified when the passage that they were watching the same scenes. However, 12 critical
of time allows the original memory to fade. Those faded, weak- slides were altered in some way. For example, if the original event
ened memories become particularly vulnerable to contamina- showed the man hiding behind a door, the misinformation slide
tion. showed him hiding behind a tree.
When contemporary investigators began using neuroimag- Two days later, subjects took a recognition test on their
ing and other physiological measures to explore true versus false memory for the original scenarios. It included questions such as
memories, I had hoped that one day these would be applied to “Where was the man hiding after he stole the girl’s wallet and
the misinformation paradigm. But in misinformation studies, crossed the street? Behind a door? Behind a tree? Behind a car?”1
there was often only one critical item (the subject was falsely led On a subsequent “source memory” test, subjects indicated the
to remember a stop sign instead of a yield sign in the accident source of their memory for every answer they had given during
scene) and thus a maximum of one false memory per subject. the first recognition test. Thus, if they claimed on the recognition
Occasionally investigators might try to insert a few false items, test that the man was hiding behind a tree, they would indicate
but never very many to prevent subjects from catching on to the that they knew this because they saw it in the original presenta-
trick and rejecting the post-event misinformation wholesale. tions, they saw it in the second set of presentations, they saw it
To do neuroimaging studies of true and false memories, in- in both, they noticed conflict, or they simply guessed. A misin-
vestigators needed to have a decent number of false memories formation error was observed whenever subjects said they saw
per subject to be able to compare them to true ones. An excellent the misinformation item (hiding behind a tree) during the origi-
paradigm for ensuring that a respectable number of false memo- nal presentation (whether only then, or whether they said they
ries were produced was the DRM (after Deese-Roediger- saw this in both presentations).
McDermott, based on work by Deese 1959 and Roediger III and
1
A few other examples of questions used in the research were “After the man
E-mail eloftus@[Link]; fax (949) 824-3002. bumped into her, what part of the girl’s body was hurt? Her neck? Her arm?
Article and publication are at [Link] Her back?” “What did the man, in frustration, do to the car after he slammed
lm.90805. the trunk on his hand(s)? Kick the car? Hit the car? Scratch the car?”

12:1–2 ©2005 by Cold Spring Harbor Laboratory Press ISSN 1072-0502/05; [Link] Learning & Memory 1
[Link]
 Downloaded from [Link] on May 17, 2025 - Published by Cold Spring Harbor Laboratory Press

Loftus

mation effect. Earlier we proposed a discrepancy detection model

to discuss various misinformation results (Loftus 1991). The idea
is that people are likely to reject misinformation and produce the
original true event item when they notice a discrepancy at the
time the misinformation is processed. Any variables that lower
the likelihood that a discrepancy is detected should reduce the
misinformation effect. Therefore, shorter exposure time of the
original event, longer retention intervals between the original
and misinformation phases, and warnings about potential mis-
information prior to the misinformation phase all reduced the
misinformation effect, consistent with the notion of discrepancy
detection. The current neuroimaging results, and especially pat-
terns of activity seen during the misinformation phase, are con-
sistent with these discrepancy notions. The neuroimaging gives
us a window into that misinformation processing that before
now we could only speculate about.
While the specific neuroimaging results reported by Okado
and Stark (2005) are highly detailed, complex, and well discussed
by them in their paper, let it suffice to say that they have pro-
vided here the first research to date that uses neuroimaging to
reveal underlying mechanisms of the misinformation effect. For
more than a half century, investigators have been actively inter-
ested in relating the brain to memory (Hebb 1949; Hull 1952), an
enterprise that has grown far more rapidly than even they might
have foreseen (Cahill et al. 2001). Okado and Stark’s (2005) in-
Figure 1. Images used in a classic misinformation experiment per- novations in determining an elegant way to apply neuroimaging
formed by Loftus et al. (1978). (A) Traffic intersection with stop sign. (B)
Misinformation image with stop sign replaced by yield sign. A detailed to an important memory problem would make their forebears
account of the research can be found in Loftus 1979. proud. It will be a long time before these results have practical
use, for example, in discriminating whether a particular memory
report derives from an original event or from misinformation
How often did subjects make misinformation errors? First,
(something the judicial system might like to see), but the current
subjects endorsed the misinformation items quite often on the
investigation is certainly an important first step.
recognition test, and significantly more often than they en-
dorsed ordinary foil items. In our example, they claimed the man
hid behind a tree. This standard misinformation effect was, of
course, entirely expected. Of course, some of these endorsements References
might not be clear cases of false memories. After all, on subse- Cabeza, R., Rao, S.M., Wagner, A.D., Mayer, A.R., and Schacter, D.L.
quent source memory tests, subjects occasionally classified mis- 2001. Can medial temporal lobe regions distinguish true from false?
information responses as “guesses.” When subjects claimed to An event-related functional MRI study of veridical and illusory
recognition memory. Proc. Natl. Acad. Sci. 98: 4805–4810.
have recognized the misinformation item, they typically said Cahill, L., McGaugh, J.L, and Weinberger, N.M. 2001. Trends Neurosci.
that they remembered the item only from the original event The neurobiology of learning and memory: Some reminders to
(27% of the time) or that they saw the item in both events (20% remember. 24: 578–581.
of the time). Thus, the pictorial version of the misinformation Deese, J. 1959. On the prediction of occurrence of particular verbal
intrusions in immediate recall. J. Exp. Psychol. 58: 17–22.
paradigm readily created false memories. Hebb, D.O. 1949. The organization of behavior. Wiley, New York.
The major contribution of the current work is that the neu- Hull, C.L. 1952. A behavior system. Yale University Press, New Haven,
ral activity during the encoding of the original presentations and CT.
the misinformation presentations was predictive of whether a Loftus, E.F. 1975. Leading questions and the eyewitness report. Cognit.
Psychol. 7: 560–572.
misinformation error would occur. Activity in particular brain . 1979. Eyewitness testimony. Harvard University Press, Cambridge,
regions such as the left hippocampus predicted whether the criti- MA.
cal items from the original presentations or misinformation pre- . 1991. Made in memory: Distortions in recollection after
sentations were later remembered on the recognition test. The misleading information. Psychol. Learn. Motiv. 27: 187–215.
Loftus, E.F. and Hoffman, H.G. 1989. Misinformation and memory: The
reported misinformation items were robust false memories, ones creation of memory. J. Exp. Psychol. General 118: 100–104.
misattributed to the original event during the source test, and Loftus, E.F. and Palmer, J.P. 1974. Reconstruction of automobile
activity in other brain regions such as the prefrontal cortex cor- destruction: An example of the interaction between language and
respondingly showed that for these misinformation errors, con- memory. J. Verbal Learn. Verb. Behav. 13: 585–589.
Loftus, E.F., Miller, D.G., and Burns, H.J. 1978. Semantic integration of
text/source encoding was strong during the original event phase verbal information into a visual memory. J. Exp. Psychol. Hum. Learn.
and poor during the misinformation phase, which set up a vul- 1: 19–31.
nerable situation whereby the misinformation item was easily Okado, Y. and Stark, C.E.L. 2005. Neural activity during encoding
embedded into the context of the original event, creating strong predicts false memories created by misinformation. Learn. Mem. (this
issue).
false memories. Roediger III, H.L. and McDermott, K.B. 1995. Creating false memories:
These findings are important in their own right, but they Remembering words not presented in lists. J. Exp. Psychol. Learn.
also dovetail nicely with behavioral discussions of the misinfor- Mem. Cogn. 21: 803–814.

2 Learning & Memory

[Link]
 Downloaded from [Link] on May 17, 2025 - Published by Cold Spring Harbor Laboratory Press

Searching for the neurobiology of the misinformation effect

Elizabeth F. Loftus

Learn. Mem. 2005, 12:

Access the most recent version at doi:10.1101/lm.90805

References This article cites 8 articles, 1 of which can be accessed free at:
[Link]

License

Email Alerting Receive free email alerts when new articles cite this article - sign up in the box at the
Service top right corner of the article or click here.

Cold Spring Harbor Laboratory Press