Skip to main content
Find a Lawyer
Sherry F. Colb

The Perils of Eyewitness Identification: A Personal Account


Wednesday, March 18, 2009

A considerable body of research shows that eyewitness identifications of a criminal defendant – the gold standard in courtroom theatrics – are, in reality, quite unreliable. The courtroom setting itself offers a very strong "suggestion" to witnesses about who the perpetrator might be – the man (or woman) sitting with his (or her) lawyer at the defendant's table. More importantly, the identifications that happen prior to trial, when witnesses look at live lineups or at pictures from "mug books," can also be quite suggestive, in ways that might not be immediately obvious.

Professor Gary Wells of Iowa State University has for many years studied the role of identification procedures in producing erroneous eyewitness identifications. Wells has performed experimental research that demonstrates just how suggestive the seemingly neutral lineup method really is. A recent, high-profile example also proves the point: In their book, Picking Cotton: Our Memoir of Injustice and Redemption, Jennifer Thompson-Cannino and Ronald Cotton talk about the former's erroneous lineup identification of the latter as her rapist. Cotton was sentenced to life imprisonment and served over a decade of that time before DNA evidence revealed the identity of the real perpetrator, Bobby Poole. Cotton ultimately spent eleven years in prison for a rape he did not commit, in part because of a lineup procedure that we know produces errors.

In this column, I will examine the implications of Wells's research through the prism of my own experiences, first as a witness observing a lineup and then as a subject in Wells's most recent experiment – in which the reader can participate as well here (click where it says "video") before reading on.

My Own Experience of Armed Robbery in Los Angeles

When I was a third-year law student at the U.S.C. Law Center, I and my then-fiancé visited an automatic teller machine on a Saturday night to get some cash. Upon our return to our parked vehicle, a young man in a leather jacket approached and asked what time it was. I told him the time, and he exhibited his blue-steel revolver and directed us to empty our wallets. We complied, after which the young man told my fiancé to return to the machine and withdraw more money while I remained behind as a hostage.

Though our robber was disappointed by the amount of money we were able to withdraw (only $200, if I remember correctly), he let us go, and we drove home. The next day, we visited the L.A.P.D. to report the crime and to give a description of the perpetrator. "Wait, wait don't tell me!," exclaimed one of the officers, smiling, "a black man, right?" I told him that yes, the robber was African-American, but I was not amused by his racist presumption.

A few months later, after accepting the likelihood that no one was going to track down our robber, I received a phone message inviting me and my fiancé to appear at a lineup. We dutifully showed up at the appointed time to try to identify our robber. Despite the unprofessional behavior of the L.A.P.D. when we first reported the crime, the lineup I experienced was extremely professional.

Victims of many different crimes were seated in a large auditorium, two seats apart from one another. As it turned out, every person in the two lineups we viewed – through a one-way mirror – was suspected of a distinct, individual crime, though they all more or less matched the description that I (and presumably that the other victims as well) had given to the police. It seemed that the crime rate in Los Angeles in the early 1990's was so high that it yielded impressive economies of scale when it came to lineups.

The first noteworthy aspect of the lineup procedure was that there were two arrays of people, not just one, for us to examine. Why does this matter? It effectively conveyed to all of us in the auditorium that when we viewed a given lineup, it was quite possible that we would not see the person who committed the crime. To drive the point home, we were told explicitly that the perpetrator might not be in either of the lineups. If there had been only one lineup, by contrast, the inclination to identify someone in the group would have been much greater. After all, we knew we had been called in because the police had arrested the guy, and it was our job to help things along by getting the identification "right."

We were then given blank sheets of paper and pens and told that if we thought we saw our individual perpetrator, we should not say anything but simply write down the number on the person's chest and indicate how strongly we believed that he was "the one." Even if there was a vague resemblance, we were informed, it was worth writing that down to facilitate later investigation. If we wanted one of the suspects to repeat the words our assailants had used during the crime (for voice-identification purposes, as the Supreme Court upheld as consistent with the Fifth Amendment right against compelled self-incrimination in United States v. Wade), we were to ask that all suspects say those words. And we were instructed, in walking across the stage near the one-way mirror, to spend the same amount of time examining each of the men in the lineup.

All of these procedures helped avoid giving our fellow victims the impression – even subconsciously – that any particular man in the lineup must be the perpetrator because people were spending more time in front of him or because he was the one who kept saying "Give me the money or I'll shoot you."

Because I understood clearly that it would not be a failure (or at least not my failure) if I could not positively identify one of the people in the lineup as my robber, I did not feel pressured into making a precipitous choice. In the first lineup, I saw one man who vaguely resembled the perpetrator, and I wrote something like "It is not #4, but he does look similar to my robber." I figured the police could use this information in their continuing search for the person who had actually robbed me and who remained at large.

When the second lineup group was illuminated by the overhead fluorescent light, I looked at the face of one of the men, and my heart began to race. As Professor Wells has said, recognition – real recognition – is immediate and does not result from the contemplative study of a person's appearance. I saw my robber and wrote this on the piece of paper in front of me. Though the man had lost weight and changed his facial hair (not to mention his clothes) since the night of the robbery, I recognized him immediately. About a year later, I testified against this man who, after robbing me, had apparently gone on to commit a robbery at an ATM adjacent to an L.A.P.D. station. He was convicted and sentenced to 27 years for my robbery and for several additional robberies as well as one count of attempted murder.

Gary Wells's Experiment

Having had the experience I described above, and having read about the problem of suggestive lineups, I felt confident in my ability to recognize faces, as well as in my appreciation for the possibility that a given lineup would not necessarily contain an actual perpetrator. Then I served as a subject in the experiment on Professor Wells's web site.

The experiment presents the subject with a video recording of a young man on the roof of a building. He is in the process of placing something onto the roof (which we're told is a bomb) and then running away. The subject has a very brief opportunity to see the man's face from various angles, both on the roof and as he runs to escape. The lineup is then displayed.

As I watched the staged "crime," I believed I could positively identify the perpetrator from any lineup. I would prove the naysayers wrong, at least in my case. I then looked at the six faces in the lineup and became concerned. None of them ignited the sort of immediate recognition that my robber had inspired. I viewed the lineup again, just to make sure. That's when two or three of the faces began to look like possibilities. The hairstyle of one of the men looked like that of the perpetrator, though the face looked different. I went to the choice list again, hoping that there was a "none of the above" slot, but there wasn't. I picked #6, hoping that I had somehow "gotten it right."

I did not. No one in the lineup was the right man, as it turned out. Because of my prior experience as a witness and my familiarity with the suggestiveness problem, I looked for a "none of the above" option. Not finding one, however, I selected someone – despite my sense that none of the men was "it" – and I even hoped that my powers of recognition had prevailed, despite my misgivings. Participating in this experiment demonstrated to me just how strong the impulse is to choose the correct answer in a multiple-choice situation, if we are not given the clear message – in the particular case – that the correct answer might be "none of the above."

For a witness viewing a conventional lineup, saying the perpetrator is not present requires her to think "outside the box." Though most of us are capable of such thinking from time to time, our default setting will be to stay within the box and thus to choose the best available alternative (just as we are told to pick the best response on multiple choice exams, even when we believe that none of the answers is correct). If choosing the best alternative means picking the person who most closely resembles the actual perpetrator, then we are likely to choose the person whom the police suspect (especially if we gave the police a detailed description or a composite sketch when first reporting the crime). This all raises the odds of an erroneous identification.

The Broader Lessons of Gary Wells's Research on Eyewitness Identification

Implementing Wells's findings should not be difficult, and in fact, some courts and police departments have (finally) begun to do so. The status quo, however, tends to have great staying power, and people in law enforcement may be loath to acknowledge that procedures which they have previously utilized might have led to erroneous identifications and wrongful convictions. Sometimes it is easier to engage in denial and continue to do the wrong thing than to change and thereby confront the harm that one has already done.

Over time, one hopes, the trend will be toward improved lineup procedures, including the use of sequential lineups – which require witnesses to look at one suspect at a time (thus improving by several degrees on the benefits of my two-lineup experience).

A broader lesson that emerges from Gary Wells's research concerns the sorts of crimes that juries are best equipped to handle. Many of the crimes that go to trial – particularly in the rape context – are crimes committed against strangers, people who have no previous relationship with their victims. Some commentators have even resisted the legitimacy of prosecuting offenses between people who know each other (such as acquaintance rape), on the theory that the jury must then confront an insoluble "he said/she said" situation, which is ostensibly more challenging for the fact-finder than the garden-variety "whodunit" crime.

Eyewitness identification research shows us, however, that in stranger offenses, unlike date rapes and other crimes among acquaintances, the eyewitness is quite likely to get it wrong. Once an eyewitness makes a misidentification, moreover, her confidence in that selection will tend to increase as the chosen suspect's face comes to replace the vague and fading face of the actual perpetrator in her mind. "He said/she said" contests, on the other hand, most often involve crimes in which both the defendant and the victim know what happened, who did what, and which of the two acted wrongfully.

In such cases, rather than somehow figuring out whether a confident eyewitness is in reality mistaken, the jury's job is to serve as a group of human lie detectors, to determine who is and who is not telling the truth. According to Professor Wells, this is a job that juries can perform relatively well. One of the takeaway messages from the frailties of eyewitness identification may therefore be that the most appropriate cases for a jury are precisely those of the much-derided "he said/she said" variety.

Sherry F. Colb, a FindLaw columnist, is Professor of Law and Charles Evans Hughes Scholar at Cornell Law School. Her book, When Sex Counts: Making Babies and Making Law, is available on Amazon.

Was this helpful?

Copied to clipboard