STANFORD JOURNAL OF LEGAL STUDIES 25
The Problem with Eyewitness Testimony
Commentary on a talk by
George Fisher and Barbara Tversky*
Laura Engelhardt
The bedrock of the American judicial process is the honesty of witnesses in trial.
Eyewitness testimony can make a deep impression on a jury, which is often exclusively
assigned the role of sorting out credibility issues and making judgments about the truth of
witness statements.1 Perjury is a crime, because lying under oath can subvert the
integrity of a trial and the legitimacy of the judicial system. However, perjury is defined
as knowingly making a false statement�merely misremembering is not a crime.2
Moreover, the jury makes its determinations of witness credibility and veracity in secret,
without revealing the reason for its final judgement.3 Recognizing the fallibility of
witness memories, then, is especially important to participants in the judicial process,
since many trials revolve around factual determinations of whom to believe. Rarely will
a factual question result in a successful appeal�effectively giving many parties only one
chance at justice. Arriving at a just result and a correct determination of truth is difficult
* This commentary was written in response to a talk given by George Fisher, Professor, Stanford Law
School, and Barbara Tversky, Professor of Psychology, Stanford University. The presentation was given
on April 5, 1999 and was sponsored by the Stanford Journal of Legal Studies. In this presentation, George
Fisher placed Barbara Tversky�s research on memory fallibility into the context of police investigations and
jury verdicts, discussing the relevance of such research to our system of justice.
1 See generally George Fisher, The Jury�s Rise as a Lie Detector, 107 YALE L.J. 575 (1997).
2 See 18 USC �1623(a) (1998):
Whoever under oath (or in any declaration, certificate, verification, or statement under
penalty of perjury as permitted under section 1746 of title 28, United States Code) in any
proceeding before or ancillary to any court or grand jury of the United States knowingly
makes any false material declaration or makes or uses any other information, including
any book, paper, document, record, recording, or other material, knowing the same to
contain any false material declaration, shall be fined under this title or imprisoned not
more than five years, or both.
The statute then goes on to list an affirmative defense: �It shall be a defense to an indictment or information
made pursuant to the first sentence of this subsection that the defendant at the time he made each
declaration believed the declaration was true.� Id., �1623(c).
3 See FED. R. EVID. 606(b) (preventing usage of juror testimony to impeach a jury�s verdict); Act of Aug.
2, 1956, ch. 879, � 1, 70 Stat. 935 (18 U.S.C. �1508 (1984 & Supp. 1996)) (criminalizing the recording of
jury deliberations in federal court).
Vol. 1:1 STANFORD JOURNAL OF LEGAL STUDIES 26
enough without the added possibility that witnesses themselves may not be aware of
inaccuracies in their testimony.
Several studies have been conducted on human memory and on subjects�
propensity to remember erroneously events and details that did not occur. Elizabeth
Loftus performed experiments in the mid-seventies demonstrating the effect of a third
party�s introducing false facts into memory.4 Subjects were shown a slide of a car at an
intersection with either a yield sign or a stop sign. Experimenters asked participants
questions, falsely introducing the term �stop sign� into the question instead of referring to
the yield sign participants had actually seen. Similarly, experimenters falsely substituted
the term �yield sign� in questions directed to participants who had actually seen the stop
sign slide. The results indicated that subjects remembered seeing the false image. In the
initial part of the experiment, subjects also viewed a slide showing a car accident. Some
subjects were later asked how fast the cars were traveling when they �hit� each other,
others were asked how fast the cars were traveling when they �smashed� into each other.
Those subjects questioned using the word �smashed� were more likely to report having
seen broken glass in the original slide. The introduction of false cues altered participants�
memories.
Courts, lawyers and police officers are now aware of the ability of third parties to
introduce false memories to witnesses.5 For this reason, lawyers closely question
witnesses regarding the accuracy of their memories and about any possible �assistance�
from others in the formation of their present memories. However, psychologists have
long recognized that gap filling and reliance on assumptions are necessary to function in
our society. For example, if we did not assume that mail will be delivered, or that the
supermarkets will continue to stock bread, we would behave quite differently than we do.
We are constantly filling in the gaps in our recollection and interpreting things we hear.
For instance, while on the subway we might hear garbled words like �next,� �transfer,�
and �train.� Building on our assumptions and knowledge, we may put together the actual
4 See Elizabeth F. Loftus & J.C. Palmer, Reconstruction of Automobile Destruction: An Example of the
Interaction Between Language and Memory, 13 J. OF VERBAL LEARNING & VERBAL BEHAVIOR 585
(1974); Elizabeth F. Loftus, D.G. Miller, & H.J. Burns, Semantic Integration of Verbal Information into a
Visual Memory, 4 J. OF EXPERIMENTAL PSYCH, 19 (1978).
5 See Krist v. Eli Lilly and Co., 897 F.2d 293, 297 (7th Cir. 1990), (listing the findings of various
psychological studies):
Accuracy of recollection decreases at a geometric rather than arithmetic rate (so passage
of time has a highly distorting effect on recollection); accuracy of recollection is not
highly correlated with the recollector's confidence; and memory is highly suggestible �
people are easily �reminded� of events that never happened, and having been �reminded�
may thereafter hold the false recollection as tenaciously as they would a true one.
See also DAVID FRANK ROSS, J. DON READ & MICHAEL P. TOGLIA, EDS., ADULT EYEWITNESS TESTIMONY
(1994); Elizabeth F. Loftus, Eyewitness Testimony: Psychological Research and Legal Thought, 3 CRIME
AND JUSTICE 105 (1981). C.f. Lea Brilmayer & Lewis Kornhauser, Quantitative Methods and Legal
Decisions, 46 U. CHI. L. REV. 116, 135�48 (1978).
Vol. 1:1 STANFORD JOURNAL OF LEGAL STUDIES 27
statement: �Next stop 53rd Street, transfer available to the E train.� Indeed, we may even
remember having heard the full statement.
So what is an �original memory?�6 The process of interpretation occurs at the
very formation of memory�thus introducing distortion from the beginning.
Furthermore, witnesses can distort their own memories without the help of examiners,
police officers or lawyers. Rarely do we tell a story or recount events without a purpose.
Every act of telling and retelling is tailored to a particular listener; we would not expect
someone to listen to every detail of our morning commute, so we edit out extraneous
material. The act of telling a story adds another layer of distortion, which in turn affects
the underlying memory of the event. This is why a fish story, which grows with each
retelling, can eventually lead the teller to believe it.
Once witnesses state facts in a particular way or identify a particular person as the
perpetrator, they are unwilling or even unable�due to the reconstruction of their
memory�to reconsider their initial understanding. When a witness identifies a person in
a line-up, he is likely to identify that same person in later line-ups, even when the person
identified is not the perpetrator. Although juries and decision-makers place great reliance
on eyewitness identification, they are often unaware of the danger of false memories.
Experiments conducted by Barbara Tversky and Elizabeth Marsh corroborate the
vulnerability of human memory to bias.7 In one group of studies, participants were given
the �Roommate Story,� a description of incidents involving his or her two fictitious
roommates. The incidents were categorized as annoying, neutral, or socially �cool.�
Later, participants were asked to neutrally recount the incidents with one roommate, to
write a letter of recommendation for one roommate�s application to a fraternity or
sorority, or to write a letter to the office of student housing requesting the removal of one
of the roommates. When later asked to recount the original story, participants who had
written biased letters recalled more of the annoying or �cool� incidents associated with
their letters. They also included more elaborations consistent with their bias. These
participants made judgements based upon the annoying or social events they discussed in
their letters. Neutral participants made few elaborations, and they also made fewer errors
in their retelling, such as attributing events to the wrong roommate. The study also
showed that participants writing biased letters recalled more biased information for the
character they wrote about, whereas the other roommate was viewed neutrally.
Memory is affected by retelling, and we rarely tell a story in a neutral fashion. By
tailoring our stories to our listeners, our bias distorts the very formation of memory�
even without the introduction of misinformation by a third party. The protections of the
judicial system against prosecutors and police �assisting� a witness�s memory may not
6 See, e.g., James Marshall, Evidence, Psychology, and the Trial: Some Challenges to the Law, 63 COLUM.
L. REV. 197, 197 (1963) (�For the law, the basic problem of ascertaining truth does not arise so much from
the villainy of perjurers and suborners of perjury as from the unreliability of personal observation.�).
7 See Barbara Tversky & Elizabeth J. Marsh, Biased Retellings of Events Yield Biased Memories
(forthcoming). [Hereinafter Tversky-Marsh study].
Vol. 1:1 STANFORD JOURNAL OF LEGAL STUDIES 28
sufficiently ensure the accuracy of those memories. Even though prosecutors refrain
from �refreshing� witness A�s memory by showing her witness B�s testimony, the mere
act of telling prosecutors what happened may bias and distort the witness� memory.
Eyewitness testimony, then, is innately suspect.
Lawyers place great import on testimony by the other side�s witness that favors
their own side�s case. For example, defense attorneys make much of prosecution
witnesses� recollection of exonerating details. In light of psychological studies
demonstrating the effect of bias on memory, the reliance and weight placed on such
�admissions� may be appropriate, since witnesses are more apt to tailor their stories�and
thus their memories�to the interests of the first listeners. An eyewitness to a crime is
more inclined to recount, and thus remember incriminating details, when speaking to a
police officer intent on solving the crime. If later the eyewitness still remembers details
that throw doubt on the culpability of the suspect, such doubts should hold greater weight
than the remembrance of incriminating details.
In another part of the Tversky-Marsh study, participants were asked to play
prosecutors presenting a summation to the jury.8 Participants first read a murder story,
where two men were suspects. Participants were then asked either to prepare a neutral
recounting of all they remembered about one suspect, or to prepare a summation to the
jury about one suspect. Later, participants were asked to recall the original story.
Participants who wrote summations recalled more incriminating details and wrongly
attributed details among suspects more often than participants who originally wrote a
neutral recounting.
Bias creeps into memory without our knowledge, without our awareness. While
confidence and accuracy are generally correlated, when misleading information is given,
witness confidence is often higher for the incorrect information than for the correct
information. This leads many to question the competence of the average person to
determine credibility issues. Juries are the fact-finders, and credibility issues are to be
determined by juries. The issue then arises whether juries are equipped to make these
determinations. Expert testimony may not be helpful. Indeed, since the very act of
forming a memory creates distortion, how can anyone uncover the �truth� behind a
person�s statements? Perhaps it is the terrible truth that in many cases we are simply not
capable of determining what happened, yet are duty-bound to so determine. Maybe this
is why we cling to the sanctity of the jury and the secrecy of jury findings:
We can put such questions before the jury entirely without fear of
embarrassment, because the way the jury resolves the questions and, in all
likelihood, the soundness of its answers will remain forever hidden.
Perhaps the allure of the black box as a means toward apparent certainty in
8 See Tversky, supra note 7 at 22�28.
Vol. 1:1 STANFORD JOURNAL OF LEGAL STUDIES 29
an uncertain world has tempted us to entrust the jury with more and harder
questions than it has the power to answer.9
The courts� reliance on witnesses is built into the common-law judicial system, a reliance
that is placed in check by the opposing counsel�s right to cross-examination�an
important component of the adversarial legal process�and the law�s trust of the jury�s
common sense. The fixation on witnesses reflects the weight given to personal
testimony. As shown by recent studies, this weight must be balanced by an awareness
that it is not necessary for a witness to lie or be coaxed by prosecutorial error to
inaccurately state the facts�the mere fault of being human results in distorted memory
and inaccurate testimony.
