Elizabeth Loftus Hypothesis See discussions, stats, and au

See discussions, stats, and author profiles for this publication at: https://www.researchgate.net/publication/247358793

Reconstructing memory: The incredible eyewitness

Article · January 1975

DOI: 10.2307/29761487

CITATIONS

109
READS

5,454

1 author:

Elizabeth F Loftus

University of California, Irvine

386 PUBLICATIONS   23,408 CITATIONS   

SEE PROFILE

All content following this page was uploaded by Elizabeth F Loftus on 03 September 2014.

The user has requested enhancement of the downloaded file.

RECONSTRUCTING MEMORY: THE INCREDIBLE EYEWITNESS
Author(s): Elizabeth F. Loftus
Source: Jurimetrics Journal, Vol. 15, No. 3 (Spring 1975), pp. 188-193
Published by: American Bar Association
Stable URL: http://www.jstor.org/stable/29761487 .
Accessed: 03/09/2014 14:02

Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at .
http://www.jstor.org/page/info/about/policies/terms.jsp

.
JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of
content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms
of scholarship. For more information about JSTOR, please contact [email protected]

.

American Bar Association is collaborating with JSTOR to digitize, preserve and extend access to Jurimetrics
Journal.

http://www.jstor.org

This content downloaded from 128.195.137.154 on Wed, 3 Sep 2014 14:02:13 PM
All use subject to JSTOR Terms and Conditions

RECONSTRUCTING MEMORY: THE
INCREDIBLE EYEWITNESS*

Elizabeth F. Loftust

“I SAW IT WITH MY OWN EYES.” That statement has ended many an

argument, since for most people seeing is believing. But it shouldn’t be.
Between the time you first witness an event and the time you recount it
to someone else, your memory of the event may change drastically.

Many factors can affect the accuracy of your report. I have found that
the questions asked about an event influence the way a witness “remem
bers” what he saw. Changing even one word in a single question can

systematically alter an eyewitness account.
Most previous research on this topic has been directed toward

demonstrating how poor eyewitness testimony is, without exploring why
people make the errors they do. One favorite method of study has been
to stage an incident, then interrogate all the witnesses about what hap
pened. Typically, everyone tells a different story.

In a study conducted at Dartmouth in the 1930s, some students

unknowingly became subjects in such an experiment. While a class was
in session, a man dressed in workman’s overalls entered the room, made
some remarks about the heat, tinkered with the radiator for a minute or

two, and left. About two weeks later he returned with five other men of
similar appearance, and the students were asked to pick him out from a

lineup of all six individuals. Seventeen percent of the students chose the

wrong man.

Another group of students, who had not witnessed the event but
who were told they had seen it, also had to make a selection. Seventy
percent of these subjects reported (correctly) that they could not recall
the incident, but 29 percent did point to one of the men. That is, they
“identified” a man they had never seen.

In a more recent study by Robert Buckhout and his colleagues at
California State University, Hayward, 141 students witnessed a staged

*Reprinted from PSYCHOLOGY TODAY magazine, December 1974. Copyright @ 1974
Ziff-Davis Publishing Company. All rights reserved.

tDr. Elizabeth F. Loftus is Assistant Professor of Psychology at the University of Washing
ton.

188 JURIMETRICS JOURNAL

This content downloaded from 128.195.137.154 on Wed, 3 Sep 2014 14:02:13 PM
All use subject to JSTOR Terms and Conditions

assault on a professor. Seven weeks later they were asked to pick out the
assailant from a group of six photographs. Although the episode had been
a dramatic one that could hardly have gone unnoticed, 60 percent of the
witnesses, including the professor who had been attacked, chose the

wrong man. Twenty-five percent selected an individual who had been at
the scene of the crime, but as an innocent bystander.

This kind of demonstration is rather entertaining, but when some?

thing similar happens in real life, the results can be serious. A few months

ago, the Los Angeles Times reported the erroneous conviction of a man
whom seven witnesses had identified as the robber of a bank. In a similar
case last year, 17 witnesses identified a man charged with shooting a

police officer; later it turned out the man had not even been in the vicinity
of the crime while it was going on. Innocent people have sat in prison
for years on the strength of eyewitness testimony. The witnesses in these
cases probably were all honest people, but they were tragically wrong.

Yet, despite the poor performance of eyewitnesses, judges and juries
continue to place great faith in them. My colleagues and I recently
studied the influence a single eyewitness can have in the courtroom. We
simulated a criminal trial, using 150 students as jurors. The students
received a written description of a grocery store robbery in which the
owner and his granddaughter were killed. They also received a summary
of the evidence and arguments presented at the defendant’s trial. Each

juror had to arrive at a verdict, guilty or not guilty.
We told some of the jurors that there had been no eyewitnesses to

the crime. We told others that a store clerk testified he saw the defendant
shoot the two victims, although the defense attorney claimed he was

mistaken. Finally, we told a third group of students that the store clerk
had testified to seeing the shootings, but the defense attorney had dis?
credited him. The attorney claimed the witness had not been wearing his

glasses on the day of the robbery, and since he had vision poorer than
20/400 he could not possibly have seen the face of the robber from where
he stood.

When we analyzed our results, we found that 82 percent of the jurors
who had not heard about an eyewitness voted for acquittal, while 72

percent of those who thought there was a credible witness voted guilty.
Most important, 68 percent of the jurors who had heard about the dis?
credited witness still voted for conviction, in spite of the defense attor

TRAGIC MISTAKES

THAT’S THE MAN

No Eyewitness

Percentage of Guilty Verdicts

Eyewitness Discredited

Eyewitness
18% 72% 68%

SPRING 1975 189

This content downloaded from 128.195.137.154 on Wed, 3 Sep 2014 14:02:13 PM
All use subject to JSTOR Terms and Conditions

ney’s remarks. It seems that people are convinced by a witness who
declares with conviction, “That’s the man.”

Since eyewitness testimony carries so much weight, it is important
to find out why distortion occurs in a witness’ memory. I would like to
know what goes on in a person’s mind when he is trying to make a

truthful report but makes a false one. To find the answer, one must
consider the nature of human memory.

Studies of memory for sentences and pictures indicate that when we

experience an event, we do not simply file a memory, then on some later
occasion retrieve it and read off what we’ve stored. Rather, at the time
of recall or recognition, we reconstruct the event, using information from

many sources. These include both the original perception of the event
and inferences drawn later, after the fact. Over a period of time, informa?
tion from these sources may integrate, so that a witness becomes unable
to say how he knows a specific detail. He has only a single, unified

memory.
I studied the way leading questions can introduce new information

that alters one’s memory of an event. A leading question is one that by
its form or content suggests to a witness the answer he should give, as
in the classic “When did you stop beating your wife?” We all probably
ask leading questions without realizing we are doing so. Lawyers, though,
have long recognized the usefulness of deliberately asking such ques?
tions. They know that by the time the opposing lawyer objects and the

judge rules the question improper, a suggestion may already have taken
hold in the minds of the jurors.

Police undoubtedly use leading questions too, when they are interro?

gating witnesses to a crime. If they influence a witness to make a false
statement (and that can happen easily, even with well-intentioned police
officers), chances are good he will repeat the error later when a trial

lawyer asks him to “tell in your own words what happened.”
A famous example of police suggestion occurred in the 1921 case of

Nicola Sacco and Bartolomeo Vanzetti, two Italian anarchists tried for
murder and robbery in Massachusetts, at the height of nationwide hys?
teria over radicalism. Five prosecution witnesses identified Sacco at the
trial, yet most of these witnesses had originally told police they could not

identify anyone. Four witnesses identified Vanzetti, although one of
them had earlier told police he had been unable to get a good look at the
robbers. This same witness stated at the trial that he had had a very good
look, after all. In fact, he was able to recall that the gunman had a dark

complexion, high cheek bones, red cheeks, short hair, a trimmed mus?

tache, a high forehead, and a hard, broad face. Sacco and Vanzetti were
convicted of the crime, and eventually executed. A later investigation of
the identification techniques used in this case indicated that witnesses
had been subject to enormous suggestion from the police.

TRAFFIC ACCIDENTS

Such cases, while instructive, do not prove conclusively that leading
questions affect testimony. In order to examine more carefully the influ

190 JURIMETRICS JOURNAL

This content downloaded from 128.195.137.154 on Wed, 3 Sep 2014 14:02:13 PM
All use subject to JSTOR Terms and Conditions

ence of the interrogator’s language on an eyewitness, I took the problem
into the laboratory. My assistants and I conducted several experiments,
using students as eyewitnesses, and films of automobile accidents as the
events they had to remember and report. Since we had a permanent
record of each event, and we asked specially constructed questions, we
were able to pinpoint the sources and types of inaccuracies.

In our first study, we showed 100 students a short film segment
depicting a multiple-car accident. In the film, a car makes a right-hand
turn into the main stream of traffic. The turn causes oncoming cars to

stop suddenly, and there is a five-car, bumper-to-bumper collision. After
our subjects viewed this film, they filled out a 22-item questionnaire
containing 16 fillers and six critical questions. Three of the key questions
asked about items that had appeared in the film, while three others asked
about items that had not actually been present. For half the subjects, the
critical questions began with the words Did you see a, as in “Did you see
a broken headlight?” For the rest, the critical questions began with the
words Did you see the, as in “Did you see the broken headlight?” Thus
the sentences differed only in the form of the article, the or a.

We had a good reason to look at this contrast. A speaker uses the
when he assumes the object referred to exists and may be familiar to the
listener. An investigator who asks, “Did you see the broken headlight?”
essentially says, “There was a broken headlight. Did you happen to see
it?” His assumption may influence the witness. But a requires no such

assumption.

When we tabulated the percentage of “yes,” “no,” and “don’t know”
responses, we found that witnesses who received questions with the were

much more likely to report having seen something that had not really
appeared in the film; 15 percent in the the group said “yes” when asked
about a nonexistent item; while only seven percent in the a group made
that error. On the other hand, witnesses who received questions with a

were more likely to respond “don’t know,” both when the object had
been present and when it had not. We see, then, that even this subtle

change in wording can influence eyewitness reports.

Percentage of “Don’t Know” Responses to Questions With “A” Or “The”

_Item Present_Item
Not Present_

the a the a

23% 51% 13% 38%

SOME GOT SMASHED

We also wanted to know whether the substitution of one word for
another could affect quantitative judgments, e.g., judgments of speed. We
showed 45 subjects seven films of traffic accidents, again varying the form
of the questions we asked after the film. For some of our subject-wit?
nesses, the critical question was “About how fast were the cars going
when they hit each other?” For others we replaced the verb hit with
smashed, collided, bumped or contacted. Although these words all refer

SPRING 1975 191

This content downloaded from 128.195.137.154 on Wed, 3 Sep 2014 14:02:13 PM
All use subject to JSTOR Terms and Conditions

to the coming together of two objects, they differ in what they imply
about speed and force of impact. We wondered if these differences would
affect judgments about velocity.

They did. Our subjects’ estimates varied considerably, depending on
which question they had to answer. Those questioned with contacted
gave the lowest speed estimates, while those questioned with smashed
gave the highest.

Four of our films involved staged crashes, and we knew exactly how
fast the cars had been traveling: one 20 mph, another 30 mph, and two
others 40 mph. The average estimates for these collisions were 37.7, 35.2,
39.7 and 36.1 mph, respectively. These figures bear out previous findings
that people are not very good at judging the speed of a vehicle, and
increase our confidence that our results were due to the way our ques?
tions were worded.

The studies I’ve described so far do not tell us why people are
influenced by leading questions. Perhaps they are merely biased by the
form of the question to give one answer instead of another. For example,
a witness might be uncertain whether to say 30 mph or 40 mph, but the
verb smashed could sway him toward the higher estimate. In that case,
we could not say that his memory of the event had changed, only that
his answer had.

To find out if our subjects were really misremembering, we ran one
more experiment. Again, we showed subjects a short film of a traffic
accident. A third of them answered the question, “About how fast were
the cars going when they smashed into each other?” Another third
answered the same question with hit instead of smashed. The remaining
third, which acted as a control group, did not get a question about
automobile speed. As in our previous study, witnesses who saw smashed

gave higher estimates than those who saw hit.
A week later our subjects returned. Without viewing the film again,

they answered a new series of questions about it. This time, the critical

question asked whether the witness had seen any broken glass, although,
in fact, there had been none in the film. If smashed really influenced

subjects to remember the accident as more severe than it had been, they
might also “remember” details that were not shown but were commensu?
rate with an accident occurring at high speed?like broken glass.

Our analysis showed that more than twice as many subjects queried
with smashed reported seeing the nonexistent glass as those queried with
hit. This result is consistent with our interpretation that memory itself

undergoes a change as a result of the type of question asked.

Average Speed Estimates for Different Verbs

smashed
collided

bumped
hit

40.8 mph
39.3 mph
38.1 mph
34.0 mph
31.8 mph contacted

192 JURIMETRICS JOURNAL

This content downloaded from 128.195.137.154 on Wed, 3 Sep 2014 14:02:13 PM
All use subject to JSTOR Terms and Conditions

MURDER OR SELF-DEFENSE?

Eyewitnesses are inaccurate in estimating not only speed, but also
time and distance. Yet in courts of law they must make quantitative
judgments all the time. Last year I worked with the Seattle Public De?
fender’s office on a case involving a young woman who had killed her boy
friend. The prosecutor called it first-degree murder, but her lawyer
claimed she had acted in self-defense. What was clear was that during an

argument, the defendant ran to the bedroom, grabbed a gun, and shot her

boy friend six times. At the trial, a dispute arose about the time that had

elapsed between the grabbing of the gun and the first shot. The defendant
and her sister said two seconds, while another witness said five minutes.
The exact amount of elapsed time made all the difference in the world
to the defense, which insisted the killing had occurred suddenly, in fear,
and without a moment’s hesitation. In the end the jury must have be?
lieved the defendant; it acquitted her.

I do not know whether leading questions played a role in this case,
but I am sure they have in others. Accident investigators, police officers,
lawyers, reporters and others who must interrogate eyewitnesses would
do well to keep in mind the subtle suggestibility that words carry with
them. When you question an eyewitness, what he saw may not be what

you get.

SPRING 1975 193

This content downloaded from 128.195.137.154 on Wed, 3 Sep 2014 14:02:13 PM
All use subject to JSTOR Terms and Conditions

View publication statsView publication stats