issue dated February 25, 2000
Looking Askance at Eyewitness Testimony
By D.W. MILLER
Ronnie Bullock was sentenced to 60 years in jail for
kidnapping and raping a young Illinois girl. Edward Honaker
spent a decade in a Virginia prison for sexually assaulting a
woman at gunpoint. Kirk Bloodsworth was shipped off to
Maryland's death row for raping and strangling a 9-year-old
All three of those men were convicted in part because
eyewitnesses or victims firmly placed them at the scene of the
crime. But not one of them was guilty. They were among the
first convicts to be exonerated by DNA tests proving that
someone else was responsible.
Some psychologists believe that such mistakes happen in
thousands of courtrooms every year. But most crimes leave no
DNA traces to rule out the innocent. For more than two
decades, psychological researchers have asked, How could so
many witnesses be wrong, and what can be done about it? Only
recently have they seen their findings influence the way the
criminal-justice system handles eyewitness testimony.
Psychologists have conducted hundreds of studies on errors in
eyewitness identification. In some cases, of course, witnesses
simply lie. But research has shown that flawed police
procedures and the vagaries of memory often lead witnesses to
identify the wrong person, and that credulous jurors too
easily credit their testimony.
To those familiar with the mountain of evidence about the way
the human mind works, that comes as no surprise. "Why should
people make good eyewitnesses?" asks Gary L. Wells, a
psychologist at Iowa State University who is widely considered
the dean of eyewitness research. In the presence of danger, he
says, "we're wired for fight or flight. What helped for
survival was not a quick recall of details."
The findings of Mr. Wells and his colleagues are finally
gaining currency in the halls of criminal justice. In part
that is due to the gradual acceptance of expert testimony on
Far more crucial, however, is the growing roster of convicts
cleared by DNA evidence. In 1996, the U.S. Department of
Justice released a report on the first 28 known cases of DNA
exoneration. After studying those and 12 subsequent cases, Mr.
Wells discovered that mistaken eyewitness testimony had played
a part in about 90 percent of the convictions.
Concerned about the high rate of eyewitness error in the DNA
cases, U.S. Attorney General Janet Reno invited him to a
meeting in early 1997. As a result of their conversation, the
department's National Institute of Justice asked Mr. Wells and
five fellow scholars to join a panel of law-enforcement
officials, criminal-defense lawyers, and prosecutors created
to write guidelines for handling eyewitness testimony.
The guide, published in October, gave scholars the opportunity
to show that human memory is not a highly reliable tool for
determining guilt in the courtroom. For example, contrary to
popular belief, people under stress remember events no better
than, and often less well than, they do under ordinary
circumstances. Witnesses also perceive time as moving more
slowly during traumatic events. That, in turn, leads them to
overestimate how much time they had to notice details, a key
factor of their credibility in court. And studies have found
that witnesses to a crime are so distracted by the presence of
a weapon -- a phenomenon called "weapon focus" -- that they
remember little else with accuracy.
Researchers cannot ethically recreate the trauma of real
crimes. But plenty of field research suggests that witnesses
are apt to misidentify people.
For example, many studies have tested the ability of
convenience-store clerks and bank tellers to recall customers
they encountered in non-stressful situations. Around a third
of the time, the employees wrongly identified faces from
"lineups" that did not include the person they had actually
In addition, all sorts of factors inhibit our ability to
recognize and recall facial detail. For instance,
psychologists have established that most of us have more
difficulty recognizing people of a different race. And memory
deteriorates very quickly over time.
Elizabeth F. Loftus, a psychologist at the University of
Washington and a pioneer in research on false memory, has
discovered that it's remarkably easy to alter one's
recollection without realizing it. Human beings are highly
susceptible to incorporating "post-event information" --
newspaper articles, comments by police, conversations with
other witnesses -- into their recollections.
Witnesses also have been known to identify as criminals people
they recognized from some other encounter, a process called
"transference." In one bizarre example, an Australian
psychologist and memory researcher named Donald Thomson was
himself once identified by a rape victim as her attacker. Not
only was his alibi airtight -- he was being interviewed on
live television at the time -- but she had mistaken him for
the rapist because she had seen his face on her television
screen during the assault.
Of course, policymakers can't do much to improve the flaws in
our memories. So scholars like Mr. Wells, who wanted to reduce
eyewitness mistakes, began to focus on things that the justice
system can control -- particularly police procedures.
One of the biggest problems with eyewitness identification,
researchers have found, is that uncertain witnesses are often
prompted to finger the person whom police have detained, even
when the suspect is not the same person they spotted at the
scene. Witnesses viewing a lineup tend to assume that police
have caught the person they saw. So they think their job is to
find the face that most resembles the description they gave to
The police sometimes exacerbate that tendency by designing
lineups poorly. Imagine a witness to a liquor-store robbery
who says the robber was white, stocky, and bearded. Based on
that description, the police identify a suspect and ask the
witness to look at a lineup of live individuals or at a spread
of photos (known as a "six-pack").
Too often, say researchers, the "distractor" faces used by
police do not closely match the witness's description, or the
suspect's photo looks different from the others. If the
suspect stands out in any way -- if his is the only color
photo in the six-pack, for instance -- the witness is far more
likely to say, "That's the guy."
Lineups are also fraught with the possibility of mistaken
identity, researchers report, because of our tendency to
overlook differences in facial appearance among people not of
our race. Not only are white witnesses, say, more likely to
mistake one black suspect for another (and vice versa), but
police officers may overestimate the degree to which the
distractors they choose match the suspect's description.
Recently, Mr. Wells has raised the alarm about the way a
witness's confidence can be manipulated. Witnesses are easily
influenced during and after the lineup -- by talking with
other witnesses or police interviewers -- to be more certain
of their choice than their recall warrants. Police
investigators, for example, may praise a witness for "picking
the right guy" out of the lineup.
That taint frequently makes its way to the jury box.
Understandably, jurors put a lot of stock in a witness who can
point to the defendant and say, "He's the one. I'll never
forget his face." But scholars have learned that the degree of
confidence during trial is a poor predictor of a witness's
accuracy. And, they warn, jurors ought to be particularly
skeptical if they learn that a witness professed more
confidence on the witness stand than in the squad room.
Recall, they say, doesn't improve over time.
Until recently, the criminal-justice system made little use of
those findings. Defense lawyers, of course, have embraced and
exploited them at least since the 1980's. But according to
Brian L. Cutler, a psychologist at Florida International
University, they have rarely been able to use the research to
cross-examine eyewitnesses or police.
"Defense lawyers have no special training -- they don't know
what questions to ask," says Mr. Cutler. "If they do ask the
right questions, how well-equipped are jurors to evaluate the
questions?" Unfortunately, jurors cling to a belief that "the
way memory works is a matter of common sense," he says. "It
just isn't so."
"People expect it's like videotape, that we attend equally
well to everything out there," says Roy S. Malpass, a
psychologist at the University of Texas at El Paso who served
on the Justice Department panel. In fact, he says, "we're
No one knows how often eyewitness error leads to false
convictions, but some scholars have taken a stab at the
question. In their book Mistaken Identification: The
Eyewitness, Psychology, and the Law (Cambridge University
Press, 1995), Mr. Cutler and Steven D. Penrod, of the
University of Nebraska at Lincoln, do some courtroom
calculations: If just 0.5 percent of America's yearly 1.5
million convictions for serious crimes are erroneous -- a rate
suggested by some studies -- then other research allows the
authors to infer that well over half of those defendants, or
around 4,500 innocent people, are convicted on false
All that may change now that the nation's top law-enforcement
officials have created new guidelines for police conduct. The
Justice Department report, "Eyewitness Evidence: A Guide for
Law Enforcement," reads like a primer on eyewitness research.
Among other things, it instructs investigators who assemble a
* Select "distractors" that match the witness's description,
even simulating tattoos or other unusual features if
* Remind the witness that the suspect they saw may not even
be in the lineup, and that the lineup is intended to clear the
innocent as much as it is to identify the guilty.
* Avoid any comments that might influence the witness's
* Ask for and record the witness's degree of certainty
* Photograph or film lineups to make the police more
accountable to the defense.
Before they can take their new influence for granted,
psychologists say, there is more to be done. For one thing,
police officers and prosecutors need to be educated about the
guidelines, which do not have the force of law. But Mr. Wells
and others believe that both groups will embrace them once
defense lawyers in the courtroom begin to hold the guidelines
up as the gold standard of diligent police work.
The social scientists didn't win every battle. Despite their
urgings, law-enforcement officials on the Justice Department
panel batted down two key suggestions for improving police
lineups. Research suggests that lineups are more accurate when
they are double-blind -- in other words, when the investigator
in charge doesn't know which person is the suspect -- and
sequential -- when the witness sees faces one at a time.
According to participants, police representatives nixed the
former idea, because logistically it would be difficult to
round up investigators who didn't know who the suspect was.
More important, they said, it would be a tough sell to their
fellow cops, because it smacks of mistrust and requires them
to cede control of an investigation to someone else.
After scholars lost the battle to include double-blind
procedures, participants say, they gave up on demanding
sequential lineups. Without the first precaution, they
explained, sequential lineups might be even more vulnerable to
manipulation than simultaneous lineups are.
John Turtle, a panel member and a psychologist at the Ryerson
Polytechnic Institute, in Toronto, believes that he has a
high-tech solution to all those concerns. He has developed
computer software that purports to take the bias out of the
photo-spread lineups, which constitute about 80 percent of
those in the United States and virtually all of those in
All a police investigator would need to do is scan a photo of
the suspect into a computer and sit the witness down in front
of the screen. The machine would then automatically choose
photos of others who match the witness's description from a
large database, and offer standardized, neutral instructions
that wouldn't nudge the witness toward a particular response.
Psychologists deny they are imputing bad faith to police
investigators. It's human nature, they say, to want your
results to match your expectations. The scholars are simply
urging police officers to treat their procedures for handling
witnesses with all the care of scientific experiments. "Human
memory is a form of trace evidence, like blood or semen or
hair, except the trace exists inside the witness's head," says
Mr. Wells. "How you go about collecting that evidence and
preserving it and analyzing it is absolutely vital."