Why Eyewitnesses Fail
I don’t know about you, but I fly a lot, mostly for work. At the end of many of my flights, I confront this problem. Surely, many of you are familiar with this. I know what my bag looks like. At least I think I do. But the airports remind us that many bags look alike. Indeed, they do. Most are black. Most are about the same size and shape.
Few have any distinguishing characteristics, which is weird. Of course, there are way too many of them on the luggage carousel. And so, we make mistakes. I can’t tell you how many times I’ve lifted someone else’s bag off the carousel, only to realize that it wasn’t mine. Twice I watched in dismay as someone grabbed my bag and disappeared into the crowd.
At La Guardia, one woman once got as far as a taxi line out front before I caught up with her. All which can be pretty distressing, especially when you’re traveling. In the end, it’s just a minor inconvenience. It’s a little bit of rushing around or perhaps waiting for someone to realize they picked up the wrong bag. It’s trivial in the grand scheme of things.
But now I want you to consider a similar type of misidentification problem, but one with severe, tragic consequences. In 2004, a 16-year-old high school student from Lemon Grove, California known as Erika was abducted by a man while walking to a friend’s house. Erika was tossed into the roadside bushes just on the other side of this tunnel formed by the Highway 94 overpass. She was sexually assaulted. Erika punched her assailant repeatedly in the face, eventually broke free, and was rescued by a passerby.
There were two significant witnesses to this crime. Erika and a man by the name of Angel Rivera who happened to be driving by. Both witnesses describe the assailant as white, about 5’10”, medium build, brown hair and a goatee. Erika provided a drawing in the sheriff’s notebook. Erika also reported that she’d seen the assailant driving a pickup truck.
As she put it, it was a crappy-looking pickup truck. Based on these limited clues, the San Diego Sheriff’s Office eventually detained a 25-year-old construction worker from San Diego by the name of Uriah Courtney. Courtney was presented to both of the witnesses in a lineup. Lineups these days typically take the form of a set of photographs, usually six. It’s called a six-pack.
There is the six-pack for Courtney. Courtney is the person on the lower left. Both witnesses picked Courtney out of the lineup. In the end, the prosecution’s case rested entirely on this eyewitness testimony. Not a shred of physical evidence came to light. In 2006, Uriah Courtney was found guilty by a San Diego jury and sentenced to life in prison.
When he was permitted to address the court, Courtney maintained that he could not express remorse for a crime that he did not commit. To his accuser, he simply said, “I am sorry, Erika, but you are mistaken in your identification of me. It was a simple mistake but one with monumental repercussions because now neither of us is receiving the justice we deserve.”
In 2010, the Innocence Project took up Courtney’s case. While DNA from the original investigation had proved inconclusive, a later analysis found DNA on the victim’s clothing that was incompatible with Courtney. Instead, it matched a former convict living in Lemon Grove, not far from the crime scene. Based on this new evidence, Courtney’s conviction was vacated. In 2013, he was released from prison, after having served eight years behind bars.
Sadly, this is not unusual. The Innocence Project has recorded nearly 350 such DNA-based exonerations. In about three quarters of these cases, eyewitness identification, or more properly eyewitness misidentification, was significant evidence leading to conviction.
We should rejoice in Uriah Courtney’s freedom and that of these other people who have suffered a similar miscarriage of justice. But as a civil society, and one with a scientific reckoning, we also have to ask, what the hell happened here? Why do eyewitnesses identify the wrong people?
In 2013, the National Academy of Sciences addressed this question by convening a study group to look at the validity of eyewitness testimony. This committee, which I chaired together with Judge Jed Rakoff of the New York Federal District Court, developed a number of recommendations which are already beginning to reform the use of eyewitness evidence. This report is freely available online.
Today, I simply want to go to the heart of the matter and tell you why eyewitnesses fail. The ability to recognize objects depends upon objects we’ve seen before. It depends upon accurate vision and memory. It’s one of the most fundamental and highly valued of human skills. We take this ability for granted. We apply it in pretty much every possible context. I can find my car in the parking lot every day. So can you. I can locate the book I was reading on the shelf in the library. I can identify my dog in the big pack of dogs at Dog Beach.
Moreover, because we tend to believe our own version of what we’ve seen, we naturally grant the same confidence in others. It must be true, because she saw it with her own two eyes. Of course, we make mistakes. I don’t mean to sound glib but I hope that you see that there’s a similarity between Uriah Courtney’s problem and my misappropriated luggage.
The consequences are vastly different. I chased down a sheepish traveler. Uriah Courtney went to prison. But the root cause is the same. In both of these cases, identifications were made based on the visual experience and memory of human observers. In both cases, those decisions were wrong.
It turns out that the brain systems that underlie visual perception and memory have been the subject of a lot of research over the past few decades. From this science, we know that there are three factors that affect the usefulness of reported experience. Uncertainty, bias and confidence.
Uncertainty refers here to the fact that vision in general is far from perfect. The reason is visual noise. Vision is plagued by noise from many natural sources, including such things as poor optics of the eye, dim illumination or distracting features in the visual scene. In the presence of noise, we’re faced with a lot of uncertainty about what we’re actually looking at, such that any decision we might make has a significant likelihood of being wrong.
If uncertainty can be likened to a breakdown of accurate sensory communication, bias is the patch that holds things together. It’s the patch that helps us to fill in the blanks in the presence of visual noise and uncertainty. We fill in the blanks with what we believe is likely to be out there based on our prior experiences with the world. Confronted with a noisy visual world, as we typically are, our memory-based expectations and biases grant us the perceptual certainty that we need to survive.
But here’s the catch. This same system that grants perceptual certainty in the face of noise is also capable of introducing content that has no bearing on external reality. To put it differently, our misinformed biases cause us to perceive things that don’t exist. To make matters worse, oftentimes, we’re absolutely confident about it. This is, of course, exactly what magicians aim for. They create conditions of uncertainty. They introduce bias, and they leave us with a strong sense of confidence about something that didn’t actually happen. But if the goal is truth, as it is in the criminal justice system, then misinformed biases and overconfidence are the hidden enemies.
I can give you a sense of this interplay between uncertainty, bias and confidence using a simple example. In the next slide, I’m going to show you a stimulus, a very noisy visual pattern. I want you to raise your hand if you see something meaningful. There are a couple of hands coming up. You people think you see something. The rest of you are clueless. It is a pretty noisy pattern.
Now I’m going to introduce a bias. I’m going to give you a little piece of information you didn’t have before. Do you see the face of a man? He looks a little bit like Che Guevara. We’ll call him Che. Now we’re going to go back to the original noisy image.
Again, I want you to raise your hand if you see something meaningful. There are a lot of hands coming up. There are a lot of hands from people who didn’t raise their hands before. For those of you who are raising your hands, I want to point out that this is the same image I showed you a moment ago. The image hasn’t changed. You changed.
I fed you a bias, which is now dictating your perceptual experience of what is undeniably a very noisy visual stimulus. Here’s the worst part. You see Che now, and you’ll always see Che. I didn’t ask your permission but I modified your brains forever. You can’t go back.
I might tell you, “It’s not really Che. It’s a pile of dirty laundry.” Or “It’s a Dalmatian in a snowstorm.” Or “Maybe it’s a map of the Aegean Islands.” But I guarantee you, you’re going to stick with Che because you are now supremely confident about Che. Che makes sense to you now. Of course, there are a lot of reasons for your confidence. The least of which is the fact that everyone else in the room also sees Che.
It’s this unholy trinity of uncertainty, bias and confidence. It’s the primary reason why eyewitnesses fail. Bear that in mind. Let’s take a moment and go back and look at the conditions in Lemon Grove. There are a number of features and viewing conditions there that could have introduced noise and uncertainty on the part of the eyewitnesses.
Things such as emotional distress or perhaps the brief duration of events. A particularly prominent factor is poor illumination since the events occurred near the entrance of this tunnel under the Highway 94 overpass. The witness, Angel Rivera, may well have seen a white man with brown hair and a goatee. But he saw him in a dark tunnel, which means that there is necessarily some uncertainty in his experience.
There are also factors in this case that could have introduced bias on the part of the eyewitnesses. It’s well known, for example, that the ability of people to discriminate faces of their own race is far better than the ability to discriminate between faces of a race different from their own. It turns out, in this case, the witness Angel Rivera was Hispanic. The culprit was white, which means the own-race bias is very likely to have been at play under those circumstances and could easily have affected the outcome.
Another factor in this case that could have introduced bias is the fact that the cop who administered the lineup knew which person in the lineup was the suspect. He wasn’t blinded to the status of the suspect, which opens the door for the possibility of unintentional, unconscious, unwitting communication in the form of subtle signals, like eye movements and body posture, that could have been picked up on by the eyewitnesses and biased the outcome.
Finally, we have confidence, or more properly, overconfidence. This is probably the most pernicious reason why eyewitnesses fail. A witness may be wrong because of uncertainty and bias, but if the witness testifies in court with confidence, it’s very difficult for a jury to discount their version of events.
The big problem here is that witnesses become more confident with the passage of time. They repeatedly tell their stories to others who appear to believe them, which promotes a cycle of confidence inflation. By the time the witness gets to court, they’re absolutely certain. They wouldn’t be there if they weren’t.
When Erika was asked about her confidence in her identification of Uriah Courtney, she replied, “I will never forget what he looks like.” When Angel Rivera was asked whether he had any doubt about his identification of Uriah Courtney, he replied, “Not at all.”
In the end, Uriah Courtney spent eight years of his life in prison. Not because these witnesses were deceitful or malicious, but because the prosecutors and deciders of fact, and pretty much everyone else involved in this case, failed to appreciate how a toxic combination of uncertainty, bias and overconfidence can cause decent, intelligent, well-meaning people to say things that are untrue and identify things they’ve never seen before.
The science of vision and memory is very clear on this point. What are we to do? The recommendations of the National Academy of Science’s report are already beginning to have a positive effect. But, more generally as a society, we need to come to terms with the fact that there are insurmountable limits to visual perception and memory that are imposed by our biological nature and the properties of the world that we inhabit.
It is this baseless fabric of our vision, to paraphrase Shakespeare, that un-wits us and does so far more often than we are aware or that we’re willing to admit. As the old saying goes, seeing is believing, but neither seeing nor believing is equivalent to truth. Thank you.