what current in psychological a decade or two ago. …...when a lie becomes memory's truth:...
TRANSCRIPT
What follows is a selection of short review papers from the journal Current Directions in Psychological
Science. The first two articles are older articles that appeared in that journal a decade or two ago. The
second two, are articles from the same labs (The Wells and Loftus labs) that appeared in the February
2011 Special Issue on Psychology and the Law of that journal. If you are interested in psychology and
the law, this special issue may be of interest to you (http://cdp.sagepub.com/content/20/1).
The papers that follow are:
Loftus, E. F. (1992). When a lie becomes memory's truth: Memory distortion after exposure to
misinformation. Current Directions in Psychological Science, 1(4), 121‐121‐123. doi:10.1111/1467‐
8721.ep10769035
Wells, G. L., Olson, E. A., & Charman, S. D. (2002). The confidence of eyewitnesses in their identifications
from lineups. Current Directions in Psychological Science, 11(5), 151‐151‐154. doi:10.1111/1467‐
8721.00189
Brewer, N., & Wells, G. L. (2011). Eyewitness identification. Current Directions in Psychological Science,
20(1), 24‐24‐27. doi:10.1177/0963721410389169
Frenda, S. J., Nichols, R. M., & Loftus, E. F. (2011). Current issues and advances in misinformation
research. Current Directions in Psychological Science, 20(1), 20‐20‐23. doi:10.1177/0963721410396620
CURRENT DIRECTIONS IN PSYCHOLOGICAL SCIENCE 121
When A Lie Becomes Memory's Truth:
Memory Distortion After Exposure to Misinformation1
Elizabeth F. Loftus
What happens when people wit
ness an event, say, a crime or acci
dent, and are later exposed to new
information about that event? Two
decades of research have been de
voted to the influence of new infor
mation on the recollections of such
witnesses. An all-too-common find
ing is that after receipt of new infor
mation that is misleading in some
way, people make errors when they
report what they saw. New,
postevent information often be
comes incorporated into a recollec
tion, supplementing or altering it, sometimes in dramatic ways. New
information invades us, like a Trojan
horse, precisely because we do not
detect its influence. Understanding how we become tricked by revised
data about a witnessed event is a
central goal of this research.
Current research showing how
memory can become skewed when
people assimilate new data utilizes a
simple paradigm. Participants first
witness a complex event, such as a
simulated violent crime or automo
bile accident. Subsequently, half the
participants receive new, mislead
ing information about the event. The
other half do not get any misinfor
Elizabeth F. Loftus is Professor of
Psychology at the University of
Washington. Her most recent book is Witness for the Defense: The Ac
cused, the Eyewitness, and the Ex
pert Who Puts Memory on Trial, coauthored with K. Ketcham (St.
Martin's Press, 1991). Address
correspondence to E. Loftus, De
partment of Psychology, Univer
sity of Washington, Seattle, WA
98195; e-mail: eloftus(a milton.
u.washington.edu.
mation. Finally, all participants at
tempt to recall the original event. In
a typical example of a study using this paradigm, participants saw a
simulated traffic accident. They then
received written information about
the accident, but some people were
misled about what they saw. A stop
sign, for instance, was referred to as
a yield sign. When asked whether
they originally saw a stop or a yield
sign, participants given the phony information tended to adopt it as
their memory; they said they saw a
yield sign.2 In these and many other
experiments, people who had not
received the phony information had
much more accurate memories. In
some experiments, the deficits in
memory performance following re
ceipt of misinformation have been
dramatic, with performance differ ences as large as 30% or 40%.
This degree of distorted reporting has been found in scores of studies,
involving a wide variety of materi
als. People have recalled seeing nonexistent items, such as broken
glass, tape recorders, and even
something as large and conspicuous as a barn (in a bucolic scene that
contained no buildings at all), and have recalled incorrect traits for items they did see, such that a clean shaven man developed a mustache,
straight hair became curly, a stop
sign became a yield sign, and a
hammer became a screwdriver. In
short, misleading postevent informa
tion can alter a person's recollection in a powerful, even predictable,
manner.
The change in report arising after
receipt of misinformation is often re
ferred to as the misinformation ef
fect.3 Four questions about the mis
Information effect have occupied the
attention of researchers:
1. When are people particularly
susceptible to the damaging in
fluence on recollection of mis
leading information, and when
are people particularly resistant?
2. What groups of people are par
ticularly prone to having their
recollections be modified, and
what groups are resistant?
3. Does misinformation actually im
pair a person's ability to remem
ber details of an event? Put an
other way, what happens to the
original memory after exposure to misinformation?
4. Do people genuinely believe in
the misinformation?
WHEN ARE PEOPLE SUSCEPTIBLE TO
MISINFORMATION?
A growing body of studies reveals
the conditions that make people par
ticularly susceptible to the influence
of misinformation. For example,
people are particularly prone to hav
ing their memories modified when
the passage of time allows the orig inal memory to fade. Put another
way, with a long interval between
the event and the misinformation, the injection of misinformation be comes relatively easy. In its weak
ened condition, memory?like the disease-ridden body?becomes es
pecially vulnerable to repeated as
saults on its very essence. This find
ing leads us to a principle, the
discrepancy detection principle, for
determining when changes in recol
lection will occur:
Recollections are more likely to
change if a person does not immediately detect discrepancies between postevent information and memory for the original
event.
Other lines of research fit well
with the discrepancy detection prin
ciple. For example, if people are ex
posed to misinformation that is sub
Copyright ? 1992 American Psychological Society
122 VOLUME 1, NUMBER 4, AUGUST 1992
tie, they are more likely to be
influenced than if the misinforma tion is not subtle. Consider the sim
ple interrogative sentence "Was the
mustache worn by the tall intruder
light or dark brown?" This sentence
not so subtly suggests the existence
of a mustache. By comparison, "Did
the intruder who was tall and had a
mustache say anything to the profes sor?" is more subtle in its suggestion
of the mustache, having embedded
this idea in a relative clause. People are more likely to falsely claim that
they saw a mustache when exposed to the more subtle version.
Another line of research that fits
well with the discrepancy detection
principle involves explicit warnings. If people are warned prior to a
postevent narrative that the narrative
may be misleading, they are better
able to resist its influence than if they are not warned. In these various
lines of research, the subject's detec
tion of discrepancies between the
original memory and the postevent
passage (or failure to detect discrep ancies) appears to be crucial. With a
long interval between event and
misinformation, and with misinfor
mation that is subtly embedded, the
ability of subjects to detect a dis
crepancy is minimized. In contrast, when subjects are warned about the
likelihood of incorrect information,
they scrutinize the postevent infor
mation, and the likelihood of detec
tion of a discrepancy is enhanced. It is
also true that people are particularly
susceptible if they can be induced to
repeat the misinformation as fact.
WHO IS SUSCEPTIBLE TO MISINFORMATION?
The majority of the studies of the
misinformation effect have been
conducted with college students, and few individual difference vari
ables have emerged. Where group differences do emerge is in misinfor
mation studies using children as sub
jects. It is common (although not
universal) to find that young children are especially susceptible to these
manipulations.4 The largest study of individual dif
ferences was recently conducted with nearly 2,000 people who were
attending a science museum in San
Francisco.5 The experiment was one
of the interactive exhibits at the mu
seum, which means that subjects
provided data for the experiment while learning from the exhibit. All subjects watched a short film clip and later answered a series of ques tions about it. Some subjects were
exposed to misleading questions but
others were not, so that the impact of misinformation could be as
sessed. The most important demo
graphic variable was the age of the
subject, which varied between 5
and 75. Memory performance rose
as a function of age up to the 20s, leveled off, and then fell sharply for
subjects over age 65. Moreover, the
youngest and the oldest groups showed large misinformation ef
fects. Put another way, the very
young and the elderly were signifi
cantly more accurate when not mis
informed than when misinformed, a
result that is consistent with other
age effects in the literature on epi sodic memory. The article describ
ing the study also reviews relevant literature on individual differences in susceptibility to misinformation.5
WHAT HAPPENS TO THE ORIGINAL MEMORY?
An important issue that has been
debated is whether misinformation
actually impairs a person's ability to
remember details of an event. Put
another way, are memory traces al
tered by postevent misinformation?
There are several ways in which mis
information could impair memory.
First, misinformation could cause
trace impairment; that is, it could
update or alter the previously formed memory. New information
could combine with earlier traces to
change the representation. Second, misinformation could cause retrieval
impairment; that is, misinformation could make the original memory trace less accessible without altering it.6 Impairment of some sort is im
plied by either the trace impair ment or the retrieval impairment mechanism.
Some theorists have rejected the notion that misinformation impairs
memory. McCloskey and Zaragoza7 disagreed with the idea that the mis
information effect is due to recoding processes or updating of previously stored memories or arises because inhibition or suppression renders the
older memory less accessible. Mc
Closkey and Zaragoza argued in
stead that the misinformation does not affect memory at all, but merely influences the reports of subjects
who have never encoded (or do not
recall) the original event. Instead of
guessing blindly, these subjects use
the misinformation to decide what to
report as their memory. Misinforma tion effects could also be obtained if
subjects remember both sources of information but select the mislead
ing information because they con
clude it must be correct.
Several lines of evidence support the notion that misinformation occa
sionally does impair the ability to re member original details, however. One kind of evidence involves stud
ies using tests that do not permit the
misinformation option. Say a subject
originally saw a stop sign, but it was
later referred to as a yield sign. Sup pose we now give the subject a test
that does not permit the selection of the yield sign (e.g., the choice is be tween a stop sign and a no-parking
sign). If the misinformation has im
paired memory for the stop sign, then the misinformed subjects
would be less likely to remember the
stop sign than the control subjects. If
there has been no memory impair ment due to misinformation, then
misled subjects would be expected to be as accurate as control subjects on a test of this type. Although some
Published by Cambridge University Press
CURRENT DIRECTIONS IN PSYCHOLOGICAL SCIENCE 123
studies do show equal performance, there are several published demon
strations of deficits in performance with this restrictive type of test. One
study4 presented preschool children
with stories and found impairment
following misinformation. Another
study8 presented adult subjects with
visual scenes (e.g., nature scenes in
cluding ponds, flowers, mountains) and then provided similar visual
scenes as postevent information.
Subjects who received misinforma
tion were less able than control sub
jects to discriminate the original scenes from novel distractors.
A second line of work supporting a memory impairment interpretation involves the use of a yes-no test.9
Belli showed subjects a simulated
crime via slides and then fed them
some misinformation via a postevent narrative that they read under a pre tense. Finally, subjects were pre sented with a series of statements, each dealing with a critical item
from the crime. Subjects said "yes" if they saw the item in the slides and
"no" otherwise. Compared with
memories for control items, there
was a large reduction in accurate
memory for the items about which
subjects had received misinforma
tion. The large reduction was not
offset by the small improvement in
memory for completely novel items.
Other lines of research that are
consistent with memory impairment involve implicit memory testing10
and logic-of-opposition procedures (described in the next section). Al
though any of these findings might be readily explained by alternative
interpretations, taken together, these
studies support the idea that misin
formation can impair a subject's
ability to remember original details.
DO PEOPLE GENUINELY BELIEVE IN
THE MISINFORMATION?
One reason to think that subjects
truly believe in their misinformation
memories is that they often express these memories with a great deal of
confidence. But how can we rule
out the possibility that subjects re
port misinformation memories to ap
pear observant or prove they are
"good" subjects? The logic-of-opposition para
digm, developed by Jacoby and ap
plied to the study of misleading sug
gestions by Lindsay,11 provides an
ideal means of assessing what sub
jects really believe. By instructing his subjects that any information
contained in the postevent narrative
was wrong and should not be re
ported on the test, Lindsay set the
tendency to report suggested details
in opposition to the ability to re
member the details from the
postevent narrative. Put another
way, he tried to offset subjects' ten
dency to want to report an item they remembered reading by harshly
warning the subjects not to report
anything they remembered from the
reading. If subjects continued to
base their test responses on sug
gested items, despite explicit in
structions against doing so, Lindsay could conclude that misled subjects truly believe that they saw the sug
gested details at the time of the ini
tial event. In fact, Lindsay obtained
such results. He first showed an
event via slides, and later provided misinformation about some items.
Misinformed subjects who saw the
slides and read the narrative in the same session claimed to have seen
the misinformation in the slides 27%
of the time, compared with 9%
for subjects who had not been
misinformed.
CONCLUDING REMARKS
Misleading information can turn a
lie into memory's truth. It can cause
people to believe that they saw
things that never really existed, or
that they saw things differently from
the way things actually were. It can
make people confident about these
false memories and also, apparently,
impair earlier recollections. Once
adopted, the newly created memo
ries can be believed as strongly as
genuine memories. If handled skill
fully, the power of misinformation is
so enormous and sufficiently con
trollable that a colleague and I re
cently postulated a not-too-distant
"brave new world" in which misin
formation researchers would be able
to proclaim: "
'Give us a dozen
healthy memories . . . and our own
specified world to handle them in.
And we'll guarantee to take any one
at random and train it to become any
type of memory that we might select . . . regardless of its origin or the
brain that holds it.' "12
The implica tions of these findings for the legal field, for advertising, for political persuasion, and for clinical settings are far-reaching.
Notes
1. The author's research described in this article was supported by grants from the National Institute of Mental Health, the National Science Foundation, and the U.S. Department of Transportation.
2. E.F. Loftus, Eyewitness Testimony (Harvard
University Press, Cambridge, MA, 1979). 3. E.F. Loftus and H.C. Hoffman, Misinforma
tion and memory: The creation of new memories, Journal of Experimental Psychology: General, 118, 100-104(1989).
4. S.J. Ceci, D.F. Ross, and M.P. Toglia, Sug gestibility of children's memory: Psycholegal impli cations, Journal of Experimental Psychology: Gen
eral, 7 76,38-49(1987). 5. E.F. Loftus, B. Levidow, and S. Duensing,
Who remembers best? Individual differences in
memory for events that occurred in a science mu
seum, Applied Cognitive Psychology, 6, 93-107
(1992). 6. J. Morton, R.H. Hammersley, and D.A. Be
kerian, Headed records: A model for memory and its failures, Cognition, 20, 1-23 (1985).
7. M. McCloskey and M. Zaragoza, Misleading postevent information and memory for events: Ar
guments and evidence against memory impairment hypotheses, Journal of Experimental Psychology: General, 114, 1-16(1985).
8. C.C. Chandler, How memory for an event is influenced by related events: Interference in modi fied recognition tests, Journal of Experimental Psy chology: Learning, Memory, and Cognition, 17, 115-125(1991).
9. R.F. Belli, Influences of misleading postevent information: Misinformation interference and ac
ceptance, Journal of Experimental Psychology: Gen
eral, 7 78,72-85(1989). 10. E.F. Loftus, Made in memory, in The Psy
chology of Learning and Motivation, Vol. 27, G.
Bower, Ed. (Academic Press, Orlando, FL, 1991). 11. D.S. Lindsay, Misleading suggestions can
impair eyewitnesses' ability to remember event de
tails, Journal of Experimental Psychology: Learning, Memory, and Cognition, 76, 1077-1083 (1990).
12. Loftus and Hoffman, note 3, p. 103.
Copyright ? 1992 American Psychological Society
CURRENT DIRECTIONS IN PSYCHOLOGICAL SCIENCE
The Confidence of Eyewitnesses in Their Identifications From Lineups Gary L. Wells,1 Elizabeth A. Olson, and Steve D. Charman
Psychology Department, Iowa State University, Ames, Iowa
Abstract
The confidence that eyewit nesses express in their lineup identifications of criminal sus
pects has a large impact on
criminal proceedings. Many convictions of innocent people can be attributed in large part to confident but mistaken eye
witnesses. Although reason
able correlations between
confidence and accuracy can
be obtained under certain con
ditions, confidence is governed
by some factors that are unre
lated to accuracy. An under
standing of these confidence
factors helps establish the con
ditions under which confi
dence and accuracy are related
and leads to important practi cal recommendations for crimi
nal justice proceedings.
Keywords
eyewitness testimony; lineups;
eyewitness memory
Mistaken identification by eye witnesses was the primary evi
dence used to convict innocent
people whose convictions were
later overturned by forensic DNA
tests (Scheck, Neufeld, & Dwyer, 2000; Wells et al, 1998). The eye
witnesses in these cases were very
persuasive because on the witness
stand they expressed extremely
high confidence that they had iden
tified the actual perpetrator. Long before DNA exoneration cases be
gan unfolding in the 1990s, how
ever, eyewitness researchers in
psychology were finding that con
fidence is not a reliable indicator of
accuracy and warning the justice
system that heavy reliance on eye witness's confidence in their iden
tifications might lead to the convic
tion of innocent people. Studies have consistently dem
onstrated that the confidence an
eyewitness expresses in an identifi
cation is the major factor determin
ing whether people will believe
that the eyewitness made an accu
rate identification. The confidence
an eyewitness expresses is also en
shrined in the criteria that the U.S.
Supreme Court used 30 years ago
(and that now guide lower courts) for deciding the accuracy of an eye
witness's identification in a land
mark case. Traditionally, much of
the experimental work examining the relation between confidence
and accuracy in eyewitness identi
fication tended to frame the ques tion as "What is the correlation be
tween confidence and accuracy?" as
though there were some single, true
correlation value. Today, eyewitness researchers regard the confidence
accuracy relation as something that
varies across circumstances. Some
of these circumstances are outside
the control of the criminal justice
system, but some are determined
by the procedures that criminal jus tice personnel control.
A GENERAL FRAMEWORK FOR CONFIDENCE
ACCURACY RELATIONS
It has been fruitful to think
about eyewitness accuracy and
eyewitness confidence as variables
that are influenced by numerous
factors, some of which are the same
and some of which are different.
We expect confidence and accuracy to be more closely related when the
variables that are influencing accu
racy are also influencing confi
dence than when the variables in
fluencing accuracy are different
from those influencing confidence.
Consider, for instance, the variable
of exposure duration (i.e., how
long the eyewitness viewed the
culprit while the crime was com
mitted). An eyewitness who
viewed the culprit for a long time
during the crime should be more
accurate than one who had only a
brief view. Furthermore, the longer view could be a foundation for the
eyewitness to feel more confident
in the identification, either because
the witness has a more vivid and
fluent memory from the longer du
ration or because the witness infers
his or her accuracy from the long
exposure duration. Hence, the cor
relation between confidence and
accuracy should be higher the more variation there is in the expo
sure duration across witnesses
(Read, Vokey, & Hammersley, 1990). Suppose, however, that
some eyewitnesses were reinforced
after their identification decision
(e.g., "Good job. You are a good witness."), whereas others were
given no such reinforcement. Such
postidentification reinforcement does nothing to make witnesses
more accurate, but dramatically in
flates their confidence (Wells &
Bradfield, 1999).
Eyewitness confidence can be
construed simply as the eyewit ness's belief, which varies in de
gree, about whether the identifica
tion was accurate or not. This belief can have various sources, both in
ternal and external, that need not
be related to accuracy. Shaw and
his colleagues, for example, have
shown that repeated questioning of
eyewitnesses about mistaken mem
ories does not make the memories
Copyright ? 2002 American Psychological Society 151
152 VOLUME 11, NUMBER 5, OCTOBER 2002
more accurate but does inflate the
eyewitnesses' confidence in those
memories (Shaw, 1996; Shaw &
McLure, 1996). Although the pre cise mechanisms for the repeated
questioning effect are not clear
(e.g., increased commitment to the
mistaken memory vs. increased
fluidity of the response), these re
sults illustrate a dissociation be
tween variables affecting confi
dence and variables affecting
accuracy.
It is useful to think about broad
classes of variables that could be
expected to drive confidence and
not accuracy, or to drive accuracy and not confidence, or to drive
both variables. It is even possible to
think about variables that could
decrease accuracy while increas
ing confidence. Consider, for in
stance, coincidental resemblance.
Mistaken identifications from line
ups occur primarily when the ac
tual culprit is not in the lineup.
Suppose there are two such line
ups, one in which the innocent sus
pect does not highly resemble the
real culprit and a second in which
the innocent suspect is a near clone
(coincidental resemblance) of the
real culprit. The second lineup will
result not only in an increased rate
of mistaken identification com
pared with the first lineup, but also
in higher confidence in that mis
take. In this case, a variable that de
creases accuracy (resemblance of an
innocent suspect to the actual cul
prit) serves to increase confidence.
THE CORRELATION, CALIBRATION, AND
INFLATION OF CONFIDENCE
Although many individual stud
ies have reported little or no rela
tion between eyewitnesses' confi
dence in their identifications and
the accuracy of their identifica
tions, an analysis that statistically combined individual studies indi
cates that the confidence-accuracy correlation might be as high as
+ .40 when the analysis is restricted
to individuals who make an identi
fication (vs. all witnesses; see
Sporer, Penrod, Read, & Cutler,
1995). How useful is this correla
tion for predicting accuracy from
confidence? In some ways, a corre
lation of .40 could be considered
strong. For instance, when overall
accuracy is 50%, a .40 correlation
would translate into 70% of the
witnesses with high confidence be
ing accurate and only 30% of the
witnesses with low confidence be
ing accurate. As accuracy deviates
from 50%, however, differences in
accuracy rates between witnesses
with high and low confidence will
diminish even though the correla
tion remains .40.
Another way to think about a
.40 correlation is to compare it with
something that people experience in daily life, namely the correlation
between a person's height and a per son's gender. Extrapolating from
males' and females' average height and standard deviation (69.1, 63.7, and 5.4 in., respectively; Department of Health and Human Services,
n.d.) yields a correlation between
height and gender of +.43. Notice
that the correlation between height and gender is quite similar to the
correlation between eyewitnesses' identification confidence and accu
racy. Thus, if eyewitnesses' identi
fications are accurate 50% of the
time, we would expect to encoun
ter a highly confident mistaken eye witness (or a nonconfident accu
rate eyewitness) about as often as
we would encounter a tall female
(or a short male).
Although the eyewitness-identi fication literature has generally used correlation methods to ex
press the statistical association be
tween confidence and accuracy, it
is probably more forensically valid
to use calibration and overconfi
dence/underconfidence measures
rather than correlations (Brewer,
Keast, & Rishworth, 2002; Juslin, Olson, & Winman, 1996). In effect, the correlation method (specifically,
point-biserial correlation) expresses the degree of statistical associa
tion by calculating the difference in
confidence (expressed in terms of
the standard deviation) between
accurate and inaccurate witnesses.
Calibration, on the other hand, as
sesses the extent to which an eye witness's confidence, expressed as
a percentage, matches the probabil
ity that the eyewitness is correct.
Overconfidence reflects the extent
to which the percentage confidence
exceeds the probability that the eye witness is correct (e.g., 80% confi
dence and 60% probability correct), and underconfidence reflects the
extent to which the percentage con
fidence underestimates the proba
bility that the eyewitness is correct
(e.g., 40% confidence and 60%
probability correct). Juslin et al.
pointed out that the confidence
accuracy correlation can be quite low even when calibration is high.
Work by Juslin et al. (1996) indi
cates that eyewitnesses can be well
calibrated at times, but recent ex
periments (Wells & Bradfield,
1999) illustrate a problem that can
arise when trying to use percent
age confidence expressed by wit nesses to infer the probability that
their identifications are accurate. In
a series of experiments, eyewit nesses were induced to make mis
taken identifications from lineups in which the culprit was absent
and were then randomly assigned to receive confirming "feedback"
telling them that they identified the
actual suspect or to receive no
feedback at all. Later, these wit
nesses were asked how certain
they were at the time of their iden
tification (i.e., how certain they were before the feedback). Those
who did not receive confirming feedback gave average confidence
ratings of less than 50%, but those
receiving confirming feedback
gave average confidence ratings of
Published by Blackwell Publishing Inc.
CURRENT DIRECTIONS IN PSYCHOLOGICAL SCIENCE 153
over 70%. Because all of these eye witnesses had made mistaken
identifications, even the no-feed
back witnesses were overconfi
dent, but the confirming-feedback witnesses were especially overcon
fident. Confidence inflation is a dif
ficult problem in actual criminal cases because eyewitnesses
are
commonly given feedback about
whether their identification deci
sions agree with the investigator's
theory of the case. In these cases, it
is the detective, rather than the eye witness, who determines the confi
dence of the eyewitness.
Confirming feedback not only inflates confidence, thereby induc
ing overconfidence, but also harms
the confidence-accuracy
correla
tion. When eyewitnesses are given
confirming feedback following their identification decisions, the
confidence of inaccurate eyewit nesses is inflated more than is the
confidence of accurate eyewit nesses, and the net result is a re
duction in the confidence-accuracy correlation (Bradfield, Wells, & Ol
son, 2002). Hence, although the
confidence of an eyewitness can
have utility if it is assessed inde
pendently of external influences (e.g., comments from the detective, learn
ing about what other eyewitnesses have said), the legal system rarely assesses confidence in this way.
IMPACT ON POLICIES AND PRACTICES
What impact has research on the
confidence-accuracy problem had on the legal system? Until rela
tively recently, the impact has been almost nil. However, when DNA
exoneration cases began unfolding in the mid-1990s, U.S. Attorney
General Janet Reno initiated a
study of the causes of these miscar
riages of justice. More than three
fourths of these convictions of in nocent persons involved mistaken
eyewitness identifications, and, in
every case, the mistaken eyewit nesses were extremely confident
and, therefore, persuasive at trial
(Wells et al., 1998). A Department of Justice panel used the psycho
logical literature to issue the first
set of national guidelines on col
lecting eyewitness identification
evidence (Technical Working
Group for Eyewitness Evidence,
1999). One of the major recommen
dations was that the confidence of
the eyewitness be assessed at the time of the identification, before
there is any chance for it to be in
fluenced by external factors.
The state of New Jersey has
gone even further in adopting the
recommendations of eyewitness researchers. Based on findings from the psychological literature,
guidelines from the attorney gen eral of New Jersey now call for
double-blind testing with lineups. Double-blind lineup testing means
that the person who administers
the lineup does not know which
person in the lineup is the suspect and which ones are merely fillers.
Under the New Jersey procedures, the confidence expressed by the
eyewitness will be based primarily on the eyewitness's memory, not on the expectations of or feedback
from the lineup administrator.
There is growing evidence that
the legal system is now beginning to read and use the psychological literature on eyewitnesses to for
mulate policies and procedures. The 2002 report of Illinois Gover nor George Ryan's Commission on
Capital Punishment is the latest ex
ample of this new reliance on the
psychological literature. The com
mission specifically cited the litera ture on the problem with confi
dence inflation and recommended
double-blind testing and explicit
recording of confidence state ments at the time of the identifica tion to prevent or detect confidence
inflation (Illinois Commission on
Capital Punishment, 2002).
NEW DIRECTIONS
Although the psychological lit
erature on eyewitness identifica
tion has done much to clarify the
confidence-accuracy issue and
specify some conditions under
which confidence might be predic tive of accuracy, research has
started to turn to other indicators
that might prove even more pre dictive of accuracy. One of the
most promising examples is the re
lation between the amount of time an eyewitness takes to make an
identification and the accuracy of
the identification. Eyewitnesses who make their identification deci sion quickly (in 10 s or less) are
considerably more likely to be ac
curate than are eyewitnesses who
take longer (e.g., Dunning & Per
retta, in press). Confidence is a self
report that is subject to distortion
(e.g., from postidentification feed
back), whereas decision time is a
behavior that can be directly ob
served. Hence, decision time might prove more reliable than confi
dence as an indicator of eyewitness
accuracy. Yet another new direc
tion in eyewitness identification re
search concerns cases in which there are multiple eyewitnesses.
Recent analyses show that the be
haviors of eyewitnesses who do not identify the suspect from a
lineup can be used to assess the
likely accuracy of the eyewitnesses who do identify the suspect from a
lineup (Wells & Olson, in press). The future of eyewitness identifica tion research is a bright one, and the legal system now seems to be
paying attention.
Recommended Reading
Cutler, B.L., & Penrod, S.D. (1995). Mistaken identification: The eyewit ness, psychology, and the law. New York: Cambridge University Press.
Scheck, B., Neufeld, P., & Dwyer, J.
(2000). (See References)
Copyright ? 2002 American Psychological Society
154 VOLUME 11, NUMBER 5, OCTOBER 2002
Wells, G.L., Malpass, R.S., Lindsay,
R.C.L., Fisher, R.P., Turtle, J.W., &
Fulero, S. (2000). From the lab to the police station: A successful ap
plication of eyewitness research.
American Psychologist, 55,581-598.
Note
1. Address correspondence to Gary L. Wells, Psychology Department, Iowa State University, Ames, IA 50011; e-mail: [email protected].
References
Bradfield, S.L., Wells, G.L., & Olson, E.A. (2002). The damaging effect of confirming feedback on the relation between eyewitness certainty and identification accuracy. Journal of Applied
Psychology, 87,112-120.
Brewer, N., Keast, A., & Rishworth, A. (2002). Im
proving the confidence-accuracy relation in
eyewitness identification: Evidence from cor
relation and calibration. Journal of Experimental Psychology: Applied, 8,44-56.
Department of Health and Human Services, Na tional Center for Health Statistics, (n.d.). Na tional Health and Nutrition Examination Survey.
Retrieved May 22, 2002, from http:// www.cdc.gov/nchs/about/major/nhanes/ datatblelink.htm#unpubtab
Dunning, D., & Perretta, S. (in press). Automatic
ity and eyewitness accuracy: A 10 - to -12 sec
ond rule for distinguishing accurate from inaccurate positive identifications. Journal of
Applied Psychology.
Illinois Commission on Capital Punishment. (2002,
April). Report of the Governor's Commission on
Capital Punishment. Retrieved April 23, 2002, from http://www.idoc.state.il.us/ccp/ccp/ reports / commission_report/
Juslin, P., Olson, N., & Winman, A. (1996). Calibra tion and diagnosticity of confidence in eyewit
ness identification: Comments on what can
and cannot be inferred from a low confidence
accuracy correlation. Journal of Experimental Psychology: Learning, Memory, and Cognition, 5, 1304-1316.
Read, J.D., Vokey, J.R., & Hammersley, R. (1990).
Changing photos of faces: Effects of exposure duration and photo similarity on recognition and the accuracy-confidence relationship. Journal of Experimental Psychology: Learning,
Memory, and Cognition, 16,870-882.
Scheck, B., Neufeld, P., & Dwyer, J. (2000). Actual innocence. New York: Random House.
Shaw, J.S., III. (1996). Increases in eyewitness con
fidence resulting from postevent questioning. Journal of Experimental Psychology: Applied, 2, 126-146.
Shaw, J.S., III, & McClure, K.A. (1996). Repeated postevent questioning can lead to elevated lev els of eyewitness confidence. Law and Human
Behavior, 20,629-654.
Sporer, S., Penrod, S., Read, D., & Cutler, B.L.
(1995). Choosing, confidence, and accuracy: A
meta-analysis of the confidence-accuracy rela tion in eyewitness identification studies. Psy chological Bulletin, 118,315-327.
Technical Working Group for Eyewitness Evi dence. (1999). Eyewitness evidence: A guide for law enforcement. Washington, DC: U.S. Depart
ment of Justice, Office of Justice Programs.
Wells, G.L., & Bradfield, A.L. (1999). Distortions in
eyewitnesses' recollections: Can the postiden tification-feedback effect be moderated? Psy chological Science, 10,138-144.
Wells, G.L., & Olson, E.A. (in press). Eyewitness identification: Information gain from incrimi
nating and exonerating behaviors. Journal of Experimental Psychology: Applied.
Wells, G.L., Small, M, Penrod, S., Malpass, R.S., Fulero, S.M., & Brimacombe, C.A.E. (1998).
Eyewitness identification procedures: Recom mendations for lineups and photospreads. Law and Human Behavior, 22,603-647.
Evolutionary Psychology of Facial Attractiveness Bernhard Fink1 and Ian Penton-Voak
Ludwig-Boltzmann-Institute for Urban Ethology, Vienna, Austria (B.F.), and Department of Psychology, University of Stirling, Stirling, Scotland,
United Kingdom (I.P.-V.)
Abstract
The human face communi
cates an impressive number of
visual signals. Although adults' ratings of facial attrac
tiveness are consistent across
studies, even cross-culturally, there has been considerable
controversy surrounding at
tempts to identify the facial
features that cause faces to be
judged attractive or unattrac
tive. Studies of physical attrac
tiveness have attempted to
identify the features that con
tribute to attractiveness by
studying the relationships be
tween attractiveness and (a)
symmetry, (b) averageness, and (c) nonaverage sexually di
morphic features (hormone
markers). Evolutionary psy
chology proposes that these
characteristics all pertain to
health, suggesting that humans
have evolved to view certain
features as attractive because
they were displayed by healthy individuals. However, the
question remains how single features that are considered at
tractive relate to each other, and
if they form a single ornament
that signals mate quality. Moreover, some researchers
have recently explained attrac
tiveness preferences in terms
of individual differences that
are predictable. This article
briefly describes what is cur
rently known from attractive
ness research, reviews some
recent advances, and suggests areas for future researchers' at
tention.
Keywords face; attractiveness; mate choice;
evolutionary psychology
An obsession with beauty is not
unique to modern Western culture
but can be found around the world
in almost all societies that have
been studied. Several studies have
shown that members of different
ethnic groups share common attrac
tiveness standards, suggesting that
the constituents of beauty are neither
Published by Blackwell Publishing Inc.
Eyewitness Identification
Neil Brewer1 and Gary L. Wells2
1 Flinders University and 2 Iowa State University
AbstractEyewitness identifications play an important role in many police investigations and courtroom decisions. Identification decisionaccuracy is shaped not only by the quality of a witness’s memory but also by social-influence variables. Some variables can be categor-ized as general impairments, whereas others produce biases against a specific suspect. We review some of the key variables in eachcategory and consider postidentification indicators of identification accuracy. Finally, we highlight what we think are some of the majordirections for future research. These include addressing some of the significant limitations of past research, examining variables that arenot directly related to memory or social influence, and developing some radical new directions for identification tests.
Keywordseyewitness identification, eyewitness memory, confidence
Witnesses to crimes are sometimes asked to view a police
lineup to see if they can identify the culprit. Using experimen-
tally created events, psychological researchers have long
warned that eyewitness identification evidence is less reliable
than people seem to believe. Corroborating the concerns of
psychologists, since the advent of forensic DNA testing in
the 1990s, 258 people convicted by juries in the United States
have been freed based on exculpatory DNA tests, and 200 of
these were cases of mistaken eyewitness identification (Inno-
cence Project, 2010). Examination of the reasons for these
mistaken identifications has provided rich avenues of investi-
gation guided by cognitive and social perspectives. Here we
focus on (a) variables that produce general impairments of
identification accuracy, (b) postidentification indicators of
identification accuracy, and (c) variables that result in biases
against the suspect.
General Impairments of IdentificationPerformance
Numerous variables have been shown to shape (a) whether
witnesses make positive or negative lineup decisions (i.e.,
choices or rejections) and (b) the accuracy of those decisions.
Not surprisingly, witnesses are likely to assume that the culprit
is in the lineup; when explicitly warned that the lineup may or
may not contain the culprit, witnesses are less likely to make a
selection (Brewer & Wells, 2006). Identification accuracy is
impaired under encoding conditions likely to undermine mem-
ory strength, such as divided attention, short exposure duration,
and long viewing distance (e.g., Lindsay, Semmler, Weber,
Brewer, & Lindsay, 2008; Palmer, Brewer, McKinnon, &
Weber, 2010). Some conditions, such as identifying a culprit
of a different race or one who was wearing a disguise (e.g.,
Meissner & Brigham, 2001), undermine encoding and/or
lineup discrimination performance. Other conditions such as
lengthy retention intervals are associated with diminished
memory strength (Deffenbacher, Bornstein, McGorty, & Pen-
rod, 2008).
Indicators of Identification Accuracy
Because an identification decision is often the key evidence
against a suspect, characteristics of identification decisions that
might discriminate accurate from inaccurate decisions have
been explored. Decision confidence (Brewer & Wells, 2006),
latency (Weber, Brewer, Wells, Semmler, & Keast, 2004) and
phenomenological reports (Palmer et al., 2010) have all been
found to discriminate for positive decisions but not for lineup
rejections. Highly confident decisions, rapid decisions, and
decisions accompanied by relevant recollection (i.e., recall of
contextual information relevant to discriminating the culprit)
are more likely to be accurate than are decisions made with low
confidence, slowly, or without relevant recollection.
Although we cannot specify absolute latencies or amounts
of relevant recollection associated with accurate decisions,
Corresponding Author:
Neil Brewer, School of Psychology, Flinders University, GPO Box 2100,
Adelaide, S. Aust 5001, Australia
E-mail: [email protected]
Current Directions in PsychologicalScience20(1) 24-27ª The Author(s) 2011Reprints and permission:sagepub.com/journalsPermissions.navDOI: 10.1177/0963721410389169http://cdps.sagepub.com
at MICHIGAN STATE UNIV LIBRARIES on August 22, 2011cdp.sagepub.comDownloaded from
eyewitness confidence can provide a valuable pointer to the
accuracy of an individual identification decision. It is not
uncommon for psychologists to express the view that there is
no meaningful relation between confidence and accuracy for
identifications, a position based on the usually modest (at best)
confidence–accuracy (CA) correlations that have been
reported. However, recent research has shown that the CA cor-
relation does not provide a comprehensive picture of the CA
relation. Researchers who have used a calibration approach
(which involves charting the proportion of accurate responses
for each confidence level) to assess the CA relation across a
variety of stimuli, exposure and attention conditions, and reten-
tion intervals, have shown that, when measured immediately
after an identification, confidence does provide a meaningful
guide as to the likely accuracy of decisions made by adult (but
not child) witnesses (Brewer & Wells, 2006; Keast, Brewer, &
Wells, 2007; Sauer, Brewer, Zweck, & Weber, 2010)—a find-
ing that is at odds with oft-stated positions in the literature.
These research outcomes signal that police investigators should
attend carefully to witness confidence when evaluating
whether an identified suspect warrants continued investigation
or whether they should perhaps target other possible suspects.
Confidence is not, however, an infallible index of accuracy.
The calibration research summarized above indicates that,
under many conditions, very high levels of confidence may
exceed the probability of an accurate identification, with confi-
dence levels of 90% to 100% often associated with lower accu-
racy rates around 75% to 90%. Further, as we discuss later,
confidence breaks down as a marker of accuracy under certain
conditions.
Variables Known to ProduceSpecific Suspect Biases
Eyewitness researchers have found it useful to distinguish
between variables that impact general performance (as in the
previous section) and variables that create a specific bias
against the suspect (Wells & Loftus, 2003). Poor lighting con-
ditions or cross-racial identification situations, for example,
impair eyewitness identification performance, but no more so
for one member of a lineup (e.g., one of the fillers) than for
another (e.g., the suspect). Of course, the suspect might or
might not be the perpetrator, so any factors that bias witness
responses toward the suspect are of great concern. Psycholo-
gists have long called for double-blind lineups to prevent the
lineup administrator from inadvertently cueing the witness
toward the suspect (Wells, 1988), but only recently have
experiments more carefully teased apart these cueing dynamics
(Clark, Marshall, & Rosenthal, 2009; Greathouse & Kovera,
2009). A powerful suspect-bias influence can also occur after
the witness makes an identification if the witness receives con-
firming feedback (e.g., ‘‘Good, you identified the suspect’’).
Confirming postidentification feedback dramatically inflates
witnesses’ reports of their certainty, view, attention, and other
variables (Semmler, Brewer, & Wells, 2004; Wells, Olson, &
Charman, 2003). Of course, an innocent suspect can stand out
in a lineup for a variety of reasons, including the presence of
fillers who do not fit the description of the culprit. But there are
other variables that create bias against a suspect, such as mis-
attributed familiarity. Misattributed familiarity can occur
because of repeated identification procedures such as having
seen the person in a mugshot search prior to a lineup or confus-
ing a bystander with the perpetrator.
Understanding suspect-bias variables is an important direc-
tion in eyewitness identification research. In virtually every
trial involving contested eyewitness identification, the to-be-
explained issue is not merely why the witness has a weak mem-
ory or whether witnesses are unreliable. Instead, the question
is, ‘‘Why did the witness choose the suspect, rather than one
of the fillers, from the lineup?’’ General impairment variables
play an important role, but only suspect-bias variables answer
that question. Research examining this latter category of vari-
ables has yielded many practical guidelines for lineup conduct
(Wells, Memon, & Penrod, 2006).
Some Major Directions
We identify four main research directions that we believe can
advance this field. The first is not at all exciting, as it will
involve researchers going back over some old ground. There
is a tendency in this field to speak with certainty about the vari-
ables that explain the variations in identification performance.
Yet we are not convinced that the knowledge base is as robust
as is sometimes assumed. Traditionally, eyewitness identifica-
tion experiments yield one data point per subject from either a
culprit-present or a culprit-absent condition. Even with what
may appear to be large sample sizes for psychology experi-
ments, statistical power is a major, and often underestimated,
issue, as has been clearly demonstrated by Brewer, Weber, and
Semmler (2005). Additionally, the levels sampled for the inde-
pendent variables are necessarily restricted, as is the range of
stimuli sampled. Meta-analyses can address some of these
issues, but when studies that have employed a same–different
face recognition paradigm (to provide stimulus variability and
more data points) are set aside, there are very few identification
test studies examining specific variables (e.g., exposure dura-
tion). These limitations mean that we do not have detailed
knowledge about the influence of individual variables or the
likely complex interactions between variables, a point illu-
strated by Lindsay et al.’s (2008) field study, using multiple sti-
muli, of the effects of viewing distance (and other variables) on
identification performance. One impediment to the sorts of
studies we are calling for might appear to be the capacity to
access sufficiently large participant samples. Several recent
field studies (e.g., Lindsay et al., 2008; Sauer et al., 2010)
reveal some effective and relatively inexpensive solutions to
this problem.
Second, there is a relative dearth of work examining the
interactions between general-impairment variables and
suspect-bias variables. An emerging theoretical view is that
suspect-bias variables have a more powerful influence when
general-impairment variables are present (Charman & Wells,
Eyewitness Identification 25
at MICHIGAN STATE UNIV LIBRARIES on August 22, 2011cdp.sagepub.comDownloaded from
2006). In recent years, computational modeling has been
applied to lineup identification behaviors, and this has been
useful in fleshing out the assumptions behind some of the
‘‘mini-theories’’ that have been used to explain eyewitness
identification errors (Clark, 2008). With a better data base of
how general impairment variables and suspect-bias variables
interact, these computational models could become more
sophisticated and lead to better theories.
Third, there is an emerging type of research that is very
important to eyewitness identification evidence in the real
world that is not directly related to memory or to social influ-
ence yet is being conducted by psychologists. One example
of such research concerns the influence of the base rate for
which the actual culprit is in the lineup (e.g., does the culprit
appear in 90% of lineups or 50% of lineups?). Psychologists
have drawn attention to the fact that this base rate is an
important factor in the chances of mistaken identification.
Two other examples of important nonmemory variables rele-
vant to mistaken identification have also been identified
recently. One is the problem of estimating likelihoods of guilt
based on the consistent and inconsistent behaviors of multiple
witnesses to the same event. Clark and Wells (2008) used
data from a wide variety of experiments to estimate how
probabilities of guilt rise and fall as a function of agreement
and disagreement among witnesses in their lineup identifica-
tions. More work needs to be done to take account of nonin-
dependence among witnesses. Another example is the
‘‘pleading effect,’’ which results in vastly different chances
of mistaken identifications to be expected at the lineup versus
in the courtroom. Wells et al. (2006) noted that 85% of guilty
suspects in the United States plead guilty. Therefore, suppose
95% of suspects who are identified from lineups are guilty
and 5% are innocent. The pleading effect means that 85%of the 95% guilty will not go to trial whereas almost 100%of the 5% innocent will go to trial rather than plead guilty.
Hence, the proportion of identified suspects going to trial
who are innocent would be greater than 33%. These are not
memory variables, but they are important to study because
they have a great impact on our understanding of how to cal-
culate the chances of mistaken identifications surfacing at
various junctures in the justice system.
Fourth, with the exception of the development of sequential
lineup (in which the witness views one lineup member at a
time; Lindsay & Wells, 1985), the eyewitness identification
research paradigm has seldom departed from the traditional
simultaneous lineup (i.e., all lineup members appear together)
used by police in criminal investigations. The extant lineup
paradigm demands that witnesses either choose from among
the members of the lineup or reject them, a decision that is
influenced by an array of social and metacognitive variables
independent of the witness’ memory strength and the degree
of match between their memory and the lineup members.
Although research has identified a number of procedural vari-
ables that can reduce error in the traditional lineup, experimen-
tal psychologists should be able to develop alternative
procedures that provide a more sensitive index of the likelihood
that the suspect is indeed the culprit. One example of such a
procedure is Sauer, Brewer, and Weber’s (2008) use of patterns
of witnesses’ confidence judgments to indicate the lineup
member who most resembles the culprit and whether that per-
son is the offender. Classification algorithms exploiting the
confidence judgments assigned to each lineup member were
used to identify a confidence criterion that optimized the clas-
sification of witnesses’ responses as accurate or inaccurate.
This approach produced culprit-present and culprit-absent
accuracy rates that exceeded the accuracy of the traditional bin-
ary identification test decision.
A likely reaction to such radical approaches is that police
and the courts would never accept a form of identification evi-
dence that doesn’t actually involve the witness picking, or
rejecting, the suspect. Our response is that any procedure that
reduces the likelihood that culprits go free and innocent people
are convicted warrants serious attention from a research per-
spective and from the perspective of giving away psychological
science.
Recommended Reading
Brewer, N., & Weber, N. (2008). Eyewitness confidence and latency:
Indices of memory processes not just markers of accuracy. Applied
Cognitive Psychology, 22, 827–840. A position paper arguing for a
focus on eyewitness decision confidence and latency in theory
development about eyewitness memory.
Brewer, N., Weber, N., & Semmler, C. (2005). (See References).
A chapter that reviews the eyewitness identification literature and
highlights some key methodological issues.
Brewer, N., & Wells, G.L. (2006). (See References). An article apply-
ing the confidence–accuracy calibration approach to eyewitness
identification.
Sauer, J.D., Brewer, N., & Weber, N. (2008). (See References). An
article examining a radical alternative to the traditional eyewitness
identification task.
Wells, G.L., & Quinlivan, D.S. (2009). Suggestive eyewitness identi-
fication procedures and the Supreme Court’s reliability test in light
of eyewitness science: 30 years later. Law and Human Behavior,
33, 1–24. An extensive review of suggestive influences on eyewit-
ness identification and how the findings call into question the U.S.
Supreme Court’s 33-year-old law on how courts should evaluate
eyewitness identification evidence.
Declaration of Conflicting Interests
The authors declared that they had no conflicts of interest with respect
to their authorship or the publication of this article.
Funding
This research was supported by Australian Research Council Grants
DP0556876 and DP1093210 to the first author and National Science
Foundation Grant SES0850401 to the second author.
References
Brewer, N., Weber, N., & Semmler, C. (2005). Eyewitness identifica-
tion. In N. Brewer & K.D. Williams (Eds.), Psychology and law:
An empirical perspective (pp. 177–221). New York, NY: Guilford.
26 Brewer, Wells
at MICHIGAN STATE UNIV LIBRARIES on August 22, 2011cdp.sagepub.comDownloaded from
Brewer, N., & Wells, G.L. (2006). The confidence-accuracy
relationship in eyewitness identification: Effects of lineup instruc-
tions, foil similarity and target-absent base rates. Journal of
Experimental Psychology: Applied, 12, 11–30.
Charman, S.D., & Wells, G.L. (2006). Applied lineup theory. In
R.C.L. Lindsay, D.F. Ross, J.D. Read & M.P. Toglia (Eds.), Hand-
book of eyewitness psychology (Vol. 2, pp. 219–254). Mahwah, NJ:
Erlbaum.
Clark, S.E. (2008). The importance of computational modelling for
eyewitness identification research. Applied Cognitive Psychology,
22, 803–813.
Clark, S.E., Marshall, T.E., & Rosenthal, R. (2009). Lineup adminis-
trator influences on eyewitness identification decisions. Journal of
Experimental Psychology: Applied, 15, 63–75.
Clark, S.E., & Wells, G.L. (2008). On the diagnosticity of
multiple-witness identifications. Law and Human Behavior,
32, 406–422.
Deffenbacher, K.A., Bornstein, B.H., McGorty, E.K., & Penrod, S.
(2008). Forgetting the once-seen face: Estimating the strength of
an eyewitness’s memory representation. Journal of Experimental
Psychology: Applied, 14, 139–150.
Greathouse, S.M., & Kovera, M.B. (2009). Instruction bias and
lineup presentation moderate the effects of administrator
knowledge on eyewitness identification. Law and Human Beha-
vior, 33, 70–82.
Innocence Project (2010). Retrieved August 16, 2010, from http://
www.innocenceproject.org/
Keast, A., Brewer, N., & Wells, G.L. (2007). Children’s metacogni-
tive judgments in an eyewitness identification task. Journal of
Experimental Child Psychology, 97, 286–314.
Lindsay, R.C.L., Semmler, C., Weber, N., Brewer, N., &
Lindsay, M.R. (2008). Eyewitness identification accuracy from a
distance: Why there should not be a 15 m ‘‘rule.’’ Law and Human
Behavior, 32, 526–535.
Lindsay, R.C.L., & Wells, G.L. (1985). Improving eyewitness
identifications from lineups: Simultaneous versus sequential
lineup presentation. Journal of Applied Psychology, 70, 556–564.
Meissner, C.A., & Brigham, J.C. (2001). Thirty years of investigating
the own-race bias in memory for faces: A meta-analytic review.
Psychology, Public Policy, and Law, 7, 3–35.
Palmer, M.A., Brewer, N., McKinnon, A.C., & Weber, N. (2010).
Phenomenological reports diagnose accuracy of eyewitness identi-
fication decisions. Acta Psychologica, 133, 137–145.
Sauer, J.D., Brewer, N., & Weber, N. (2008). Multiple confidence esti-
mates as indices of eyewitness memory. Journal of Experimental
Psychology: General, 137, 528–547.
Sauer, J., Brewer, N., Zweck, T., & Weber, N. (2010). The effect of
retention interval on the confidence-accuracy relationship for eye-
witness identification. Law and Human Behavior, 34, 337–347.
Semmler, C., Brewer, N., & Wells, G.L. (2004). Effects of postidentifi-
cation feedback on eyewitness identification and nonidentification
confidence. Journal of Applied Psychology, 89, 334–346.
Weber, N., Brewer, N., Wells, G.L., Semmler, C., & Keast, A. (2004).
Eyewitness identification accuracy and response latency: The
unruly 10-12 second rule. Journal of Experimental Psychology:
Applied, 10, 139–147.
Wells, G.L. (1988). Eyewitness identification: A system handbook.
Toronto, Ontario: Carswell Legal Publications.
Wells, G.L., & Loftus, E.F. (2003). Eyewitness memory for people
and events. In A. Goldstein (Ed.), Comprehensive handbook of
psychology, Vol. 11: Forensic psychology. New York, NY: John
Wiley and Sons.
Wells, G.L., & Memon, A, & Penrod, S. (2006). Eyewitness evidence:
Improving its probative value. Psychological Science in the Public
Interest, 7, 45–75.
Wells, G.L., Olson, E., & Charman, S. (2003). Distorted retrospective
eyewitness reports as functions of feedback and delay. Journal of
Experimental Psychology: Applied, 9, 42–52.
Eyewitness Identification 27
at MICHIGAN STATE UNIV LIBRARIES on August 22, 2011cdp.sagepub.comDownloaded from
Current Issues and Advances inMisinformation Research
Steven J. Frenda, Rebecca M. Nichols, and Elizabeth F. LoftusUniversity of California, Irvine
AbstractEyewitnesses are often called upon to report information about what they have seen. A wealth of research from the past centuryhas demonstrated, however, that eyewitness memory is malleable and vulnerable to distorting influences, including the effects ofmisinformation. In this article, we review recent developments in research related to the misinformation effect, includingindividual differences in susceptibility, neuroimaging approaches, and protective interview procedures that may better elicitaccurate event details. We conclude with a section on related false memory research.
Keywordseyewitness, memory, law, misinformation
Twenty-five people died when a Metrolink commuter train
collided with a Union Pacific freight train near Los Angeles
in September of 2008 (Steinhauer, 2008). With millions of
dollars in lawsuit payouts at stake, federal accident authorities
began an investigation of the deadly crash and had to decide a
key issue: Did the conductor pass legally through a green light,
as four eyewitnesses maintained? Or did he sail through a red
light, distracted by sending and receiving text messages? The
conductor died in the crash, so he could not be asked. If he were
at fault, the railroad company that was responsible for hiring
and supervising him would be liable. If the signal malfunc-
tioned, another company would be on the hook. After an exten-
sive investigation, the authorities decided the eyewitnesses
were wrong. The signal was red, and the engineer’s text messa-
ging was a major contributor to the accident. Is it possible that
four eyewitnesses—including a conductor, a security guard,
and two railroad enthusiasts—were all mistaken about such a
crucial detail? The answer is yes. Eyewitnesses make mistakes,
multiple eyewitnesses can all be wrong, and their erroneous
testimony can have enormous consequences.
How is it possible that so many witnesses could all be so
wrong? Eyewitnesses are called upon not only to remember
details of events but also to describe what people look like and
to decide how confident they are in the accuracy of their mem-
ories. They are often asked to remember things they saw in
extremely stressful circumstances, sometimes months or even
years after the fact. They are frequently bombarded with infor-
mation following the event they witnessed, such as other wit-
nesses’ reports, investigator feedback, leading questions, and
pressures to be both accurate and helpful. In the face of these
challenges, eyewitnesses misremember. In a recent discussion
of the distorting effect witnesses have on the memory of other
witnesses, Wright, Memon, Skagerberg, and Gabbert (2009)
proposed three accounts of why eyewitnesses come to report
incorrect information. First, a witness’s report may be altered
due to normative social influence. That is, a witness may decide
that the cost of disagreeing with law enforcement—or with
other witnesses—is too high, and so he adjusts his report
accordingly. A second possibility is that through informational
social influence processes, a witness comes to endorse a ver-
sion of events that is different from what he remembers because
he believes it to be truer or more accurate than his own mem-
ory. Finally, a witness’s memory can become distorted, some-
times as the result of being exposed to incorrect or misleading
information. This third possibility, known as the misinforma-
tion effect, is the focus of the current review. Advances in mis-
information research concerning individual differences,
neurophysiological correlates, cognitive interviewing, and
related research paradigms are reviewed.
What Is the Misinformation Effect?
In the wake of more than 30 years of research, an ever-growing
literature continues to demonstrate the distorting effects of mis-
leading postevent information on memory for words, faces, and
Corresponding Author:
Elizabeth Loftus, Department of Psychology & Social Behavior, University of
California, Irvine, CA 92697
E-mail: [email protected]
Current Directions in PsychologicalScience20(1) 20-23ª The Author(s) 2011Reprints and permission:sagepub.com/journalsPermissions.navDOI: 10.1177/0963721410396620http://cdps.sagepub.com
at MICHIGAN STATE UNIV LIBRARIES on August 22, 2011cdp.sagepub.comDownloaded from
details of witnessed events (see Loftus, 2005, for a review of
the misinformation effect). In a typical misinformation experi-
ment, research subjects are shown materials (e.g., photographs)
and are then exposed to deliberately misleading information
about what they saw. In a final testing phase, many subjects
will inadvertently incorporate elements from the misleading
information into their memory for the original source material.
For example, Stark, Okado, and Loftus (2010) showed subjects
a series of photographs that depicted a man stealing a woman’s
wallet and hiding it in his jacket pocket. Later, subjects heard
recorded narratives describing the slides. Embedded in the nar-
ratives were several pieces of misleading information (e.g.,
‘‘Then the man hid the wallet in his pants pocket’’). Finally,
subjects were asked questions about details from the photo-
graphs, such as ‘‘Where did the thief hide the woman’s wal-
let?’’ A substantial number of those subjects not only
reported that the thief hid the wallet in his pants pocket but they
also reported that they remembered that information from the
photographs, not the narratives.
Who Is Vulnerable?
Nobody is immune to the distorting effects of misinformation.
Building on the adult literature, misinformation effects have
been obtained in myriad subject samples, including infants
(Rovee-Collier, Borza, Adler, & Boller, 1993), and even ani-
mals (e.g., Schwartz, Meissner, Hoffman, Evans, & Frazier,
2004). Nonetheless, there is evidence that certain types of peo-
ple are especially vulnerable to misinformation effects. For
instance, very young children and the elderly are more suscep-
tible to misinformation than adolescents and adults (see Davis
& Loftus, 2005). Also especially vulnerable are subjects who
report lapses in memory and attention (Wright & Livingston-
Raper, 2002). What do these findings tell us about the underly-
ing mechanisms driving the misinformation effect? One argu-
ment is that a poverty of cognitive resources necessitates an
increased reliance on external cues to reconstruct memories
of events. As Loftus (2005) points out, misinformation effects
are easier to obtain when subjects’ attentional resources are
limited. Similarly, people who perceive themselves to be for-
getful and who experience memory lapses may be less able
(or willing) to depend on their own memories as the sole source
of information as they mentally reconstruct an event.
Recently, two major studies containing more than 400 sub-
jects have explored cognitive ability and personality factors as
predictors of susceptibility to misinformation. In each study,
subjects viewed slides of two crimes and later read narratives
of the crimes that contained misinformation. Those subjects
who had higher intelligence scores, greater perceptual abilities,
greater working memory capacities, and greater performance
on face recognition tasks tended to resist misinformation and
produce fewer false memories (Zhu et al., 2010a). Certain
personality characteristics were also shown to be associated
with false memory formation, particularly in individuals with
lesser cognitive ability. Specifically, individuals low in fear
of negative evaluation and harm avoidance, and those high in
cooperativeness, reward dependence, and self-directedness
were associated with an increased vulnerability to misinforma-
tion effects (Zhu et al., 2010b). In other words, it seems that
personality variables may be helpful in understanding the pro-
cesses underlying memory distortion following exposure to
misinformation but less so in individuals with superior cogni-
tive ability. These interactions may help explain why individ-
ual difference results have not always replicated in false
memory research.
Misinformation and Neuroimaging
Relatively new but increasingly popular tools for exploring the
effects of postevent information on memory include a set of
highly specialized neuroscientific methods which include func-
tional magnetic resonance imaging, or fMRI. In a typical
fMRI-based behavioral experiment, subjects undergo tradi-
tional experimental procedures in an MRI scanner, during
which functional images of oxygenated blood flow in the brain
are collected. The resulting images can be analyzed and inter-
preted as differential brain activation associated with particular
tasks. Functional MRI, therefore, is a useful and noninvasive
tool for examining the neurobiological correlates of behavior.
Scientists have begun to investigate brain activity associated
with the misinformation effect. In a recent study (Stark et al.,
2010), subjects were shown a series of photographs and later
listened to an auditory narrative describing them, which
included misleading information. Soon afterward, they were
placed into an MRI scanner and given a test of their memory
for the photographs. Functional neuroimaging data revealed
similar patterns of brain activity for true and false memories,
but the true memories (formed by visual information) showed
somewhat more activation in the visual cortex while the false
memories (derived from the auditory narrative) showed some-
what more activation in the auditory cortex. As the researchers
noted, these results are congruent with the sensory reactivation
hypothesis (Slotnick & Schacter, 2004, 2006), which in part
proposes that the same sensory regions activated in the brain
during encoding will be reactivated during retrieval. These
results suggest that there may be differing brain activation pat-
terns for true and false memories when they are encoded in dif-
ferent sensory modalities.
Research that involves neuroimaging and other neuroscien-
tific measurement techniques are promising for discoveries
about the effects of misinformation on memory: They can pro-
vide glimpses into how different neurological processes under-
lie true and false memories. At the present time, however, it
would be wise to err on the side of caution in the application
of these findings. Although some differences were found, the
patterns of brain activation associated with true and false mem-
ories in Stark et al.’s (2010) study were not reliably distinct,
and other small differences in brain activation (unrelated to the
sensory reactivation hypothesis) were not fully accounted for.
Furthermore, data from fMRI studies are often averaged both
within and across participants, which makes interpretation at
the individual level of analysis difficult. Although functional
Misinformation Research 21
at MICHIGAN STATE UNIV LIBRARIES on August 22, 2011cdp.sagepub.comDownloaded from
neuroimaging is elaborate and cutting edge, it has yet to pro-
vide a sure-fire way to confidently judge whether or not a par-
ticular person’s memory is accurate.
Protecting Against Misinformation Effects
Not surprisingly, some effort has been focused on ways to pro-
tect against the distorting effect of misinformation. One tech-
nique for improving the accuracy and completeness of an
eyewitness’s recollection is known as the cognitive interview,
a set of rules and guidelines for interviewing eyewitnesses (see
Wells, Memon, & Penrod, 2006, for a review). The CI recom-
mends, for example, the use of free recall, contextual cues, tem-
poral ordering of events, and recalling the event from a variety
of perspectives (such as from a perpetrator’s point of view).
Also, the CI recommends that investigators avoid suggestive
questioning, develop rapport with the witness, and discourage
witnesses from guessing. In one recent study, subjects viewed
an 8-minute film depicting a robbery (Memon, Zaragoza,
Clifford, & Kidd, 2009). Later, subjects were given either a
CI or a free-response control interview, followed by suggestive
questioning about events not depicted in the film. Results indi-
cated that, consistent with earlier findings, the CI produced
more correct details than did the free-response procedure.
One week after the interview procedure, subjects were given
a recognition test for items in the video, and subjects incorpo-
rated details from the suggestive questioning into their memory
for the event. Results showed that the CI deterred the effects of
suggestion, but only when it came before the suggestive inter-
view. Though the investigative process would ideally be free of
all suggestive influence, a properly implemented cognitive
interview may help protect the integrity of an eyewitness’s
memory.
Related Lines of Research
In addition to the classic misinformation paradigm, researchers
have developed other ways to demonstrate that even the
subtlest suggestions can produce astonishing false witness
reports. For instance, a handful of studies have emerged in
which subjects are simply asked if they have seen video footage
of well-known news events, when in fact no such video footage
exists. One study found that 40% of a British sample was will-
ing to report having seen nonexistent footage of a bus explod-
ing in the 2005 London terrorist attacks (Ost, Granhag, Udell,
& Hjelmsater, 2008). Of the subjects who claimed they saw the
footage, 35% described memories of details that they could not
have seen. Another study (Sjoden, Granhag, Ost, & Hjelmsater,
2009) found that 64% of a Swedish sample claimed to have
seen nonexistent video footage of an attack on the Swedish for-
eign minister, and 19% went on to describe details in the form
of written narratives. The ease with which these studies elicited
blatantly false memory reports is striking.
Research has also shown that suggestion can also shape
autobiographical memory. Beginning with Loftus and
Pickrell’s Lost in the Mall study (1995), a series of studies have
successfully used personalized suggestion (or other suggestive
techniques) to plant false memories of traumatic childhood
events (Porter, Yuille, & Lehman, 1999), receiving a painful
enema (Hart & Schooler, 2006), and even impossible events
such as meeting Bugs Bunny—a Warner Brothers charac-
ter—at Disneyland (Braun, Ellis, & Loftus, 2002). These lines
of research represent a broad area in their own right, with con-
troversies and applications that are beyond the scope of this
paper. However, they show that misleading postevent informa-
tion has implications beyond merely mistaking a green traffic
light for a red one or misremembering where a pickpocket hid
a woman’s wallet. If suggestion can cause us to remember
experiences that never occurred, what does this say about the
reliability of eyewitness evidence in general? If merely asking
people if they have seen events they could not possibly have
witnessed represents a strong enough suggestion to cause such
staggering errors, what are the implications for witnesses who
were present at a crime scene but never saw a perpetrator’s
face, only to hear it described later? Researchers continue to
investigate what conditions lead to memory distortion, which
types of people are most susceptible, and how best to prevent
the distorting effects of postevent information. Unfortunately,
in spite of recent scientific advances, many eyewitness errors
continue to go undetected and can have devastating
consequences.
Declaration of Conflicting Interests
The authors declared that they had no conflicts of interest with respect
to their authorship or the publication of this article.
Recommended Reading
Brainerd, C.J., & Reyna, V.F. (2005). The science of false memory.
New York: Oxford University Press. A book, over 500 pages long,
that describes nearly everything a reader would want to know
about false memories.
Loftus, E.F. (2005). (See References). Summarizes some of the key
studies that have contributed to more than three decades’ worth
of research on the misinformation effect.
References
Braun, K.A., Ellis, R., & Loftus, E.F. (2002). Make my memory: How
advertising can change our memories of the past. Psychology &
Marketing, 19, 1–23.
Davis, D., & Loftus, E.F. (2005). Age and functioning in the legal sys-
tem: Perception memory and judgment in victims, witnesses and
jurors. In I. Noy & W. Karwowski (Eds.), Handbook of forensic
human factors and ergonomics. London, England: Taylor &
Francis.
Hart, R.E., & Schooler, J.W. (2006). Increasing belief in the
experience of an invasive procedure that never happened: The role
of plausibility and schematicity. Applied Cognitive Psychology,
20, 661–669.
Loftus, E.F. (2005). Planting misinformation in the human mind:
A 30-year investigation of the malleability of memory. Learning
and Memory, 12, 361–366.
22 Frenda et al.
at MICHIGAN STATE UNIV LIBRARIES on August 22, 2011cdp.sagepub.comDownloaded from
Loftus, E.F., & Pickrell, J.E. (1995). The formation of false memories.
Psychiatric Annals, 25, 720–725.
Memon, A., Zaragoza, M., Clifford, B.R., & Kidd, L. (2009). Inocula-
tion or antidote? The effects of cognitive interview timing on false
memory for forcibly fabricated events. Law and Human Behavior,
34, 105–117.
Ost, J., Granhag, P., Udell, J., & Hjelmsater, E.R. (2008). Familiarity
breeds distortion: The effects of media exposure on false reports
concerning media coverage of the terrorist attacks in London on
7 July 2005. Memory, 16, 76–85.
Porter, S., Yuille, J.C., & Lehman, D.R. (1999). The nature of real,
implanted, and fabricated memories for emotional childhood
events: Implications for the recovered memory debate. Law and
Human Behavior, 23, 517–537.
Rovee-Collier, C., Borza, M.A., Adler, S.A., & Boller, K. (1993).
Infants’ eyewitness testimony: Effects of postevent information on
a prior memory representation. Memory and Cognition, 21, 267–279.
Schwartz, B.L., Meissner, C.M., Hoffman, M.L., & Evans, S., &
Frazier, L.D. (2004). Event memory and information effects in a
gorilla. Animal Cognition, 7, 93–100.
Sjoden, B., Granhag, P.A., Ost, J., & Hjelmsater, E.R. (2009). Is the
truth in the details? Extended narratives help distinguishing false
‘‘memories’’ from false ‘‘reports.’’ Scandinavian Journal of
Psychology, 50, 203–210.
Slotnick, S.D., & Schacter, D.L. (2004). A sensory signature that
distinguishes true from false memories. Nature Neuroscience, 7,
664–672.
Slotnick, S.D., & Schacter, D.L. (2006). The nature of memory
related activity in early visual areas. Neuropsychologia, 44,
2874–2886.
Stark, C.E.L., Okado, Y., & Loftus, E.F. (2010). Imaging the recon-
struction of true and false memories using sensory reactivation
and the misinformation paradigms. Learning and Memory, 17,
485–488.
Steinhauer, J. (2008, September 12). At least 18 killed as trains collide
in Los Angeles. The New York Times. Retrieved from http://
www.nytimes.com/2008/09/13/us/13crash.html
Wells, G.L., Memon, A., & Penrod, S.D. (2006). Eyewitness
evidence: Improving its probative value. Psychological Science
in the Public Interest, 7, 45–75.
Wright, D.B., & Livingston-Raper, D. (2002). Memory distortion and
dissociation: Exploring the relationship in a non-clinical sample.
Journal of Trauma & Dissociation, 3, 97–109.
Wright, D.B., Memon, A., Skagerberg, E.M., & Gabbert, F. (2009).
When eyewitnesses talk. Current Directions in Psychological Sci-
ence, 18, 174–178.
Zhu, B., Chen, C., Loftus, E.F., Lin, C., He, Q., Chen, C., et al.
(2010a). Individual differences in false memory from misinforma-
tion: Cognitive factors. Memory, 18, 543–555.
Zhu, B., Chen, C., Loftus, E.F., Lin, C., He, Q., Chen, C., et al.
(2010b). Individual differences in false memory from misinfor-
mation: Personality characteristics and their interactions with
cognitive abilities. Personality and Individual Differences, 48,
889–894.
Misinformation Research 23
at MICHIGAN STATE UNIV LIBRARIES on August 22, 2011cdp.sagepub.comDownloaded from