chevron-down Created with Sketch Beta.
January 17, 2019 Article 2

Seeing Bugs Bunny at Disney World

The Reliability of Eyewitness Testimony

by Joyce Lacy


Jennifer Thompson, a confident 22-year-old with a bright future, was nearing her college graduation with a perfect 4.0 GPA. She and her boyfriend were talking of getting married soon. Life was good. But one July evening, everything changed. She awoke in the middle of the night to a blade pressed against her throat and a man sitting on her legs. Despite her fear and panic, she forced herself to focus on the identifying features of her assailant. Jennifer survived that night by remaining alert and eventually fleeing to a neighbor’s house. Soon after, she would identify Ronald Cotton as her assailant in a lineup and again in court. The prosecution could not have asked for a more prepared and confident witness. She had studied her assailant’s face, noted his almond-shaped eyes, high cheekbones, and shadowy mustache. Cotton was convicted, and he would serve over 10 years in prison before being exonerated by DNA evidence. How could have Jennifer, the straight-A college student, been wrong?

            It is an unfortunate misperception that memory is like a video recorder, faithfully recording our experiences and preserving these details so that they can be reviewed at any time. Recalling information from our memories is such a mundane experience that we rarely have reason to doubt its accuracy. In cases when our everyday memory does fails us, the repercussions are typically minor or insignificant such as forgetting to run an errand or misremembering the source of a famous quote. However, when it comes to eyewitness testimony in cases of law, the accuracy of memory is paramount. Recalled memories may be the major evidence, sometimes even the only evidence, against a defendant. Additionally, eyewitness testimony is often heavily weighted in court. Research has shown that mock jurors are more likely to vote “guilty” when there is an eyewitness account than in identical cases sans an eyewitness.

            In one study, mock jurors read about a grocery store robbery where the owner was killed. In the control group with no eyewitnesses, only 18% of mock jurors voted guilty. This served as a baseline to show how convicting the evidence was in this scenario. From the 18% of guilty votes, one can see that the evidence was only somewhat convincing. When a different group of mock jurors was given identical evidence plus the addition of a store clerk as an eyewitness, the percentage of guilty votes rose to 72%. The most striking finding of this study was that even when the eyewitness’ testimony was discredited (e.g., the clerk was known to have poor vision), nearly the same percentage of jurors (68%) voted guilty. Thus, even highly questionable eyewitness testimony is convincing. It is important to note that eyewitness testimony is not always faulty, but situations like Ronald Cotton’s show us that memory can be faulty and lead to wrongful convictions. How and why does our memory fail us? More than four decades of psychological research has much to tell us.

Let us rewind to the 1970’s where Elizabeth Loftus, a recent Stanford Ph.D. alumnus, was beginning her research career as a professor at the University of Washington. She had aspirations to move from her roots of researching fundamental concepts of memory towards something more immediately impactful to society. Thus, she began to study how memories could be modified. Modifying memories is not like in the movies Total Recall or Inception where professionals surgically implant fake memories into your brain. Loftus was looking for natural experiences that cause one’s memory of an event to reflect something different than what actually happened. Little did she know that she would become the pioneer of an entirely new field of research around what we now refer to as “false memories.”

Loftus began this work by looking at the effect wording can have on an individual’s memories. She started by showing people films of car accidents and then asking them questions such as, “How fast were the cars going when they smashed into each other?” On average, participants asked this question reported higher speeds than participants shown the same film but asked the question with the word smashed replaced with the word hit. Even more intriguing was that in a follow-up question, participants given the smashed wording were statistically more likely to misremember seeing broken glass in the film (although, there was no broken glass). Additionally, Loftus discovered that using definite articles (e.g., “Did you see the broken headlight?”) versus indefinite articles (e.g., “Did you see a broken headlight?”) was more likely to lead to an affirmative response, even if the affirmative was incorrect.

Wording effects are subtle and perhaps a far cry from misremembering a face, as in the case of Ronald Cotton. However, this nascent field bloomed into a much richer body of research surrounding the impact of questioning, interrogation, and post-identification feedback. There is a large body of psychological literature that suggests when interviewers have preconceived notions of the “correct” answer, they can (intentionally or not) bias an individual’s response. This may be through verbal conversation, tone of voice, or body language. For example, asking an eyewitness after they made an identification, “Are you sure?” with a certain type of inflection may cause a witness to doubt their original choice and pick an lternative. Additionally, post-identification feedback can alter an individual’s subsequent identifications or their perception of other factors surrounding their identification such as their level of confidence. Positive post-identification feedback such as “Good job!,” “You picked the suspect,” or “Another witness chose the same person” increases confidence. Likewise, negative feedback decreases confidence. Sources of post-identification feedback also include crosstalk between witnesses and media coverage. In research studies, participants who received confirming feedback also reported having paid more attention to the perpetrator’s face and having better views of the perpetrator than witnesses who did not receive this feedback. Confidence is generally considered an indication of accuracy, and in many cases this relationship is true, but it is not a perfect relationship. High confidence is not always an indictor of memory accuracy as can be seen in Jennifer’s account. However, confident witnesses are considered more believable and are more willing to testify in court.

            Cotton appealed his conviction and was granted a retrial two years later. In the intervening time, a new inmate arrived at the same prison that Cotton was in. His name was Bobby Poole. They had similar appearances. They looked so alike that a kitchen steward at the prison often confused them with each other. At one point, Poole confessed to another inmate that he was Jennifer’s real assailant, although he denied this in a voir dire examination at Cotton’s retrial. Once again, Jennifer unequivocally identified Cotton as her assailant. DNA evidence would eventually confirm that Poole was the real perpetrator. But how could Jennifer be wrong again, especially with her actual assailant sitting in the courtroom? It was almost as if she had no memory of Poole’s face, as if Poole’s face was replaced with Cotton’s.

Thus far, the psychological research illustrates that memories are malleable and susceptible to suggestion. It is one thing to change minor details about a memory, but is it possible to create an entirely false memory? Loftus began investigating this possibility in the 1990’s. She started with a simple false scenario: being lost at the mall as a young child. The researchers first confirmed with the participants’ caregivers that no such event had actually happened. Caregivers also provided details of three real events that participants experienced as a young child, and participants were presented with all four “memories” without being told that one was false. Not all participants accepted the lost-in-the-mall event as a real memory, but a statistically significant percentage (approximately one fourth) of participants did.

Many studies have replicated this ability to instill whole events into participants’ memories. In research, these are referred to as “rich false memories.”  Although the rate of success varies from roughly 15 to 70%, meta-analyses suggest that on average approximately a third of participants in these research models adopt a false memory. These studies began with relatively innocuous memories—participants have falsely remembered spilling a bowl of punch at a wedding or meeting Bugs Bunny at Disney World (Bugs is a Warner Brothers character). However, other studies involved mild levels of trauma such as nearly drowning or being viciously attacked by an animal. In all of these experiments, researchers reveal to the participants at the conclusion of the study that these memories are indeed false and that no such event happened.

The most recent false memory experiments are the most realistic. For example, one 2015 study attempted to convince college aged participants that they had committed a crime (a theft, assault, or assault with a weapon) between the ages of 11 to 14. By the end of three interviews over the span of approximately two weeks, 70% participants believed they had actually committed the crime and they provided additional details of the event. Other memory studies take advantage of real events such as national tragedies. In one study, Russian participants misremembered seeing wounded animals near a scene of a terrorist bombing that occurred in Moscow a few years prior. Several studies have replicated this phenomenon of how we frequently misremember the details of a tragic national event. After the explosion of the Challenger space shuttle in 1986, participants were asked how they heard the news shortly after the event and then again three years later. More than one third of the reports were inaccurate. For example, participants who initially heard from a friend later misremembered first hearing of the explosion from watching TV. This pattern of results was also seen in studies probing participants’ memories of the 9/11 New York City terrorist attacks.

There are many ways to create or modify memories. A simple method of modification is by providing misinformation after the original event. Loftus discovered this phenomenon in the 1970’s. Once again, she showed participants films of traffic accidents. This time, she discovered she could modify participants’ memories of the original event by implying a bit of misinformation. For example, if participants viewed a car stopping at a stop sign but were later asked about a yield sign, a significant percentage of participants misremember seeing a yield sign in the original film. Post-event misinformation is thought to employ the same mechanism as post-identification feedback previously discussed. Other research models ask participants to recall information of multiple previous events; however, participants are unaware that one of these events is actually false. This is the mechanism by which rich false memories have been instilled. Another method that may be used is asking participants to imagine events. In one study, participants were asked to either perform or imagine a variety of events such as rolling a pair of dice or kissing a plastic frog. At later questioning, a significant number of participants misremembered actually performing some of the imagined events.

When participants adopt rich false memories as their own, they feel confident that they actually happened and they often add their own details, which can include sensory perceptions such as shaking Bugs Bunny’s hand and hearing him say “What’s up, Doc?” They even add their own subjective experiences such as their feelings during the event and their rationale of thinking. Psychological research offers an explanation for how and why participants might fill in these details. We often make use of schemas, or scripts, of typical experiences to help us navigate our day-to-day lives. Thus, when asked to recall information, we fill in details from our expectations of an event. For example, in one study, participants first listened to an audio recording of a mock trial. A significant number of these participants later recalled details from the trial that were not actually stated, but which fit a typical description of that crime (e.g., a robber pulling out a weapon). Interestingly, participants also had difficulty differentiating between statements from the eyewitness and what was implied in misleading questions from the attorneys, that is, some of the details implied in the attorneys’ questions were misremembered as being stated by the witness.

Critics have argued that the mild levels of trauma occasionally employed in research studies are not realistic enough and are not the same as a personally experienced crime. However, an interesting set of experiments in elite military personnel speaks to the accuracy (or lack thereof) of human memory when confronted with a realistic and personally relevant threat. In these studies, participants were undergoing military survival school training and, therefore, could be subjected to objectively high levels of stress during the mock prisoner of war phase of training. Participants endured physical aggression and intimidation tactics during a mock interrogation such as being physically struck and stared down. Cortisol levels during this phase of training were equivalent to jumping out of an airplane for the first time. Despite the elite training of these military individuals, approximately two-thirds of participants later misidentified their interrogator in a lineup even though they had had a clear view of their interrogator’s face for 30-40 minutes.

If psychological studies are not evidence enough that false memories can occur, DNA evidence verifiably proves that false memories do indeed occur. The Innocence Project is a non-profit legal organization that was founded in 1992 with the goal of raising awareness of wrongful convictions and advocating for reform in the criminal justice system. The Innocence Project was involved in Cotton’s case. By the most recent statistics in late 2017, there have been 351 cases of wrongful convictions overturned by DNA evidence. Experts agree that this represents only a fraction of wrongful convictions since most crimes do not have access to viable DNA. The average exoneree served 13 years in prison before being released. Several cases involved inmates on death row. Typically, there is little retribution or support after exoneration. What makes this so relevant to false memory is that eyewitness misidentification is the leading cause of false convictions. In fact, 70% of the convictions overturned by DNA evidence originally involved an eyewitness misidentification.

It may surprise you to hear that Jennifer Thompson and Ronald Cotton are actually close friends today. After Cotton’s release, Jennifer was wracked with guilt about her misidentification. They were both 22 years old when Poole broke into Jennifer’s home. In the ten years since that fateful night, Jennifer moved on, got married, and had kids. All the while, Ronald Cotton was deprived of those opportunities. It took Jennifer two years after Cotton’s release for her to work up the courage to meet him. When she finally did, Jennifer blurted out a tearful apology, and in an amazing display of grace, Cotton’s first words to Jennifer were words of forgiveness. They have since developed a close relationship. They co-authored a memoir together, Picking Cotton: Our Memoir of Injustice and Redemption,” and they travel the nation together advocating for the rights of the wrongfully convicted.

So, where do we go from here? Psychological research has shown that memory is not infallible and alterations to memory are possible. However, this does not mean that we should completely remove the use of human memory in law. In many cases, memory is indeed accurate and still provides valuable evidence. A better approach is to educate law enforcement, legal practitioners, and the general public about the nuisances and limitations of memory and to reform our policies to better reflect the probative value of eyewitness testimony and other forms of memory in court. Lastly, psychological research on aspects of memory pertaining to law is ongoing. Greater collaboration between the fields of psychology and law would help refine and direct research as well as apply its findings to actual court cases.