Using Science to Improve Eyewitness Testimony

Five years after a landmark National Academies report on eyewitness identification, a lead author reflects on its impact. 

By Sara Frueh | Jan. 24, 2020

As we move through the world, looking at objects and people, we generally trust that we are accurately perceiving what’s out there. But research has shown that part of what we see sometimes originates in our own minds — that our brains fill in blanks in our vision based on our expectations or past experiences.

While the effects of this tendency are often helpful, they can sometimes be tragic — as when witnesses to a crime mistakenly believe they saw somebody who they did not. In over 70 percent of the 360+ cases in which a person convicted of a crime was later exonerated by DNA testing, at least one mistaken eyewitness identification was involved, according to the Innocence Project.

Now science — and the insights it provides about the pitfalls in our vision and memory — is improving the way eyewitness testimony is taken and used. Over the past few years, police departments in at least 19 states have implemented safeguards that can help reduce the chance of errors, a shift due in part to a report released in late 2014 by the National Academies, Identifying the Culprit: Assessing Eyewitness Identification. 

National Academy of Sciences member Thomas Albright, a neuroscientist at the Salk Institute for Biological Studies, co-led the 14-person committee that developed the report. Albright studies how our eyes and brains process, construct, and remember what we see. He drew upon this knowledge as he and the committee studied how eyewitnesses can get things wrong — and how to improve the odds that they will get things right.

The Shifting, Subjective Nature of Sight 

In his lab at the Salk Institute in California, Albright and his colleagues study how our brains take in and shape what we see. In studies with animals and people, the researchers use instruments to monitor the activity of brain cells as something is viewed — allowing inferences about specific parts of the brain that enable humans to see particular things, such as the color red or the motion of an object. 

Part of the lab’s work has focused on how context affects people’s perceptions. “As a simple example, we’ve known for centuries that the perceived color of a patch of light will depend heavily on the color that surrounds it,” says Albright. “This is partly why if you want to paint your walls at home, it’s difficult to choose a paint color at the paint store and have it come out the way you want it to. The same paint looks different in different contexts.”

Certain contexts — situations where there is dim lighting, or a long viewing distance, or stress on the part of the viewer — lessen our capacity to see things accurately. In these conditions of uncertainty, the brain often fills in the gaps in perceptions, supplying missing information based on that person’s past experiences or expectations.

This capacity evolved over millions of years to aid survival, Albright says. “If you’re on the savannah and you see an ambiguous movement, it’s important that you identify what it is quickly so that you can survive. The visual brain has evolved this capacity to fill in the blanks based on our prior experiences, and most of the time it works.” 

Often it works even if it isn’t precisely accurate: If you’re crossing a street at dusk and catch the glimmer of metal out of the corner of your eye, for example, it doesn’t matter whether your mind fills in the actual blue car or a more-familiar gray one — either image will alert you to leap out of the way.

Sometimes we even knowingly let our eyes play tricks on us, as when we watch a magic act. “Performance magicians will create conditions of uncertainty, and they’ll introduce bias, and they’ll leave you with a very strong impression that you saw something that didn’t actually happen,” says Albright. “And because it’s the nature of the genre, we accept it; we know we’re being tricked. But similar things happen in our normal activities — uncertainties in the context prompt our brains to fill in the wrong information — and we’re completely unaware of it.”

Vision and Memory at a Crime Scene

Crime situations may involve factors that can hinder a viewer’s ability to perceive things accurately, such as quick viewing times and high levels of stress and fear. These conditions, and the uncertainty that results, may trigger the brain’s tendency to fill in the gaps, so that the witness has a coherent interpretation of what happened. But that coherence might come at the expense of accuracy.

Often, for example, a witness will report the presence of a weapon, and it turns out there was no weapon, Albright says. “If you’re witnessing a crime, you might have in your mind that in these kinds of circumstances, the perpetrator often will have a gun. And so you actually perceive a gun, even though there may not be one. The gun gets inserted into your perceptual experience because conditions lead to significant uncertainty, and the gun is what you expect under those kinds of conditions.”

In a related phenomenon, a person may unconsciously insert things into memory from an outside source. For example, a witness might perceive the events of a crime and store that information in memory, and then read in the newspaper that another witness said the perpetrator had a handgun. Even though the original witness didn’t perceive a handgun at the time of the events, that information may get implanted into her memory, without awareness that it didn’t come from her own experience.

“Contamination” of a witness’s memory may also happen in more subtle ways. For example, a police officer who is supervising as a witness picks a suspect out of a lineup may, without realizing it, offer nonverbal cues of affirmation, increasing the witness’s confidence in his identification.

Are most people aware of how error-prone vision and memory can be? “Generally, no,” Albright says. “As we go through the world, we accept what see without question. And so if somebody else tells us that they saw something, we grant them the same benefit of the doubt. But the truth is, if you look at it carefully, you find that individuals very commonly misperceive things. Juries in general, and probably most judges as well, don’t have a good understanding of that.”

To counter that lack of awareness, Identifying the Culprit recommends that courts bring in experts to advise juries about the ways that eyewitness evidence can fail. The report also urges the use of “double blind” lineups, where the police officer administering a lineup does not know who the suspect is — leaving them unable to offer biasing cues to an eyewitness. And it recommends that police officers record a witness’s confidence level upon first identifying a suspect — a time when their assessment of their own confidence is more likely to be accurate, and when their memory is less likely to have been contaminated. 

Impacts in the Justice System and Beyond

In the years since the National Academies' report was released, at least 19 states have passed legislation or adopted rules requiring reforms recommended by the report, such as using double-blind lineup procedures and recording witnesses’ confidence levels. The federal government has acted as well; in 2017 the Department of Justice released new guidelines for how lineups using photographs should be conducted in federal criminal cases, based on procedures recommended in the report. And shortly before that, the International Association of Chiefs of Police issued a new model policy for conducting lineups, also drawing upon the report.

The report has also engaged more researchers to work on the problem of eyewitness identification. “In the scientific community, it’s been incredibly valuable,” says Albright. “There are basic scientists, who have an understanding of how visual perception and memory work, who are now using that information to help figure out better ways of doing lineups, for example.” 

One area of research that has great promise for improving eyewitness identification in the future is the use of modeling to estimate the accuracy of a witness, Albright says. Developed through testing, models could quantify and factor in characteristics that are known to affect perception — the lighting at the crime scene, for example, and the witness’s viewing distance — and offer a numerical estimate of how accurate eyewitness testimony is likely to be. A similar approach is already being used with success in medical diagnosis and prognosis, Albright says. “These are tried and true techniques in a discipline that has lots of parallels to the criminal justice problem, and they could similarly be applied to the criminal justice problem.” 

Has the report and its many impacts affected Albright personally? “I’ve done a lot of things in my scientific career, but there are few things that I’m as proud of as I am of this report. Not only is it an important societal problem, but our work is having an impact on the world in real time,” he says. “At some point I had this realization that there are a lot of things I know about the brain that are relevant to immediate problems in society, and this is one of them. And it’s having — within my lifetime — an effect on people’s lives.”