Skip navigation
CLN bookstore
× You have 2 more free articles available this month. Subscribe today.

Eyewitness (Mis)Identification in the Criminal Justice System: Powerful, Persuasive, and Problematic

by Christopher Zoukis

“Our procedure has always been haunted by the ghost of the innocent man convicted.”

—Judge Learned Hand (1923)

“Truth isn’t truth.”

—Rudy Giuliani (2018) 

In 1984, Ronald Cotton was arrested and charged with the knifepoint rape of Burlington, North Carolina, college student Jennifer Thompson. Cotton became a suspect in the crime after Thompson picked him out of a photo array, telling detectives, “I think this is the guy.” When Thompson asked the investigators if she did okay, they responded, “You did great.”

Following Cotton’s arrest, detectives showed Thompson a live lineup. Cotton was the only person in the lineup who also appeared in the photo array. Thompson was hesitant but ultimately told detectives that Cotton “looks the most like him.” When Thompson was told that Cotton was the same person she had identified in the photo array, she described feeling a “huge amount of relief.”

At trial, Thompson told the jury that she was “absolutely sure” that Cotton was the assailant. Cotton denied any involvement in the crime.

On the sole basis of Thompson’s identification, Cotton was convicted and sentenced to life in prison, plus 54 years. Nearly 11 years into his sentence, DNA evidence conclusively established a shocking fact: Cotton was not the rapist.

The actual perpetrator was a man named Bobby Poole. At a postconviction hearing on the issue, Thompson was unable to identify Poole as her attacker. When asked, she said of Poole, “I have never seen him in my life.”

Postconviction proceedings established that Cotton was innocent of the crime, and he was exonerated. Thompson initially had serious difficulty accepting that Cotton did not rape her.

“I don’t know,” Thompson said at the time of Cotton’s exoneration. “The DNA tests, the science tells me that we have the wrong guy. It was Bobby Poole. Ronald Cotton says it was not him, it was Bobby Poole. They do look very similar, it is almost frightening how similar they look to each other. . . . I don’t know. I really don’t know. I have to accept the answer that has been given to me and put faith in our system.”

Ronald Cotton was robbed of more than a decade of his life due to one of the most pervasive and intractable problems in the criminal justice system: eyewitness misidentification. According to the Innocence Project, a mistaken eyewitness identification was present in nearly 75 percent of all DNA exonerations. Moreover, the biological material needed for DNA testing is not available in most cases involving an eyewitness identification. As such, it is likely that many individuals who were wrongly convicted on the basis of eyewitness misidentification will never be exonerated.

Error-Prone Memory

Eyewitness identification is powerful and persuasive evidence. As humans, we are innately confident in our perceptions and memory. But modern science has begun to shed light on just how wrong our perceptions and memory can be. Eyewitness identification relies on human sensation and brain systems, and there is significant room for error in both of these areas.

Consider the case of the invisible gorilla. In a now-famous series of experiments, scientists Christopher Chabris and Daniel Simmons illustrated just how error-prone human perception and memory can be. They published their shocking findings in the 2010 book, The Invisible Gorilla: How Our Intuitions Deceive Us.

Participants in the experiment were shown a short film of two teams playing basketball, one team wearing black shirts and the other white. The subjects were instructed to count the number of passes between members of the white-shirt team. Halfway through the video, a woman in a gorilla suit walks onto the court, stops in the middle and beats her chest, and then walks off the court.

In a post-experiment interview, half of the participants reported nothing unusual. Half of the participants did not see the gorilla.

Nobel prize-winning economist Daniel Kahneman wrote about the invisible gorilla in his bestselling book, Thinking Fast and Slow. He said the experiment shows just how wrong we can be about what we perceive and remember.

“The authors [of the experiment] note that the most remarkable observation of their study is that people find its results very surprising,” wrote Kahneman. “Indeed, the viewers who fail to see the gorilla are initially sure that it was not there—they cannot imagine missing such a striking event. The gorilla study illustrates two important facts about our minds: We can be blind to the obvious, and we are also blind to our blindness.”

Scientists now know that human visual, perception and memory systems are far from perfect. Incorporating that knowledge into the criminal justice system has proven to be a slow process that has met resistance along the way. Given what we now know about the many ways that eyewitness identification can go wrong, however, it is a crucially important process.

In 2014, the National Academy of Sciences (“NAS”) released a report that has gone a long way toward improving law enforcement understanding of problems associated with eyewitness identification evidence. The report, “Identifying the Culprit: Assessing Eyewitness Identification” (hereafter “the NAS report”), summarized current scientific understandings and made a series of recommendations designed to limit a significant and pervasive cause of injustice in the justice system: eyewitness misidentification.

Eyewitness Identification as an Investigation Technique

Law enforcement officers view eyewitness identification as a critical tool in the investigation of crime and prosecution of criminal actors. The FBI estimates that law enforcement entities made 521,196 arrests for violent crime in 2012. While there are no exact numbers on how many involved eyewitness identification, one 1989 survey of prosecutors suggests that at least 80,000 eyewitnesses make identifications of criminal suspects each year.

Investigators employ a variety of techniques and procedures when securing eyewitness identification evidence. The three most common procedures are the showup, the photo array, and the live lineup.

A showup is generally used shortly after the commission of a crime, when a suspect is in custody. In most cases, an officer takes the witness to the suspect’s location and seeks an identification. Often the suspect is in handcuffs or otherwise restrained or in the back of a squad car. According to the NAS report, a showup is “designed to promptly clear innocent suspects, thereby sparing them from a prolonged period of detention as the investigation continues.”

The photo array is the most commonly used eyewitness identification tool. In the typical situation, an eyewitness is shown several photographs and asked to identify the perpetrator. Ideally, investigators will use “filler” photographs that resemble the suspect as closely as possible. The photos may be presented all at once (the simultaneous method), or one at a time (the sequential method). When an eyewitness makes an identification, investigators sometimes ask for a “confidence statement.”

A live lineup is similar to the photo array, except that it uses the actual physical suspect and actual physical fillers. Just as with the photo array, fillers are generally chosen for their similarity to the suspect. Live lineups may be displayed simultaneously or sequentially. Given the resources necessary to stage a live lineup, the technique is not the predominant method used for eyewitness identification.

These techniques are age-old and are not rooted in science. According to a report from the criminal justice reform group The Justice Project, methods such as the lineup were developed as “ad hoc” procedures, “created and embraced in the law enforcement community because of [their] intuitive plausibility.” However, a deep dive into the science of human vision and memory illustrates precisely why this is a significant problem in the criminal justice system. It is becoming more and more clear that such an intuitive belief in the accuracy of human sensation and memory may be entirely misplaced.

The Process and Limitations of Eyewitness Identification

The ease with which we see, remember and recall everyday events masks what are a complex series of interactions among distinct biological systems. In order to understand how easily an eyewitness identification can go wrong, it is necessary to understand these systems and interactions, as well as their limitations.

* Vision. The first step in the process is visual sensation. Light strikes the eye, forms an image in the retina, and is “sensed” by the brain. The act of visual sensation is subject to interference—aka “noise”—which can limit accuracy. Examples of noise at this stage include low lighting, limited viewing time, and observer-specific visual defects (myopia, color blindness, etc.). The impact of noise during visual sensation is influenced by other factors, such as the angle of the observer’s gaze and the distance from the observed object.

Most of the light that reaches the eye is filtered out, leaving a small amount that receives attention. Light that is sensed and attended to reaches awareness. Because very little of what is sensed can receive attention, the observer must select what is attended to. Noise plays a major role in what sensations receive attention.

Several aspects of the attention process are highly relevant in the eyewitness identification milieu. Competing interests are capable of hijacking an observer’s attentional focus. The invisible gorilla experiment provides an example of this phenomenon. Scientists refer to what happened to the observers who did not see the gorilla as inattentional blindness.

The phenomenon of inattentional blindness is not limited to unusual events. In one experiment, two strangers engaged in a brief conversation. During the conversation, an opaque door was carried between the participants, and one of them (the experimenter) was replaced with a person who looked and sounded different, and wore different clothing. When the door passed, the conversation continued—and more than half of the participants failed to notice that they were no longer talking to the same person.

Fear is also well known to hijack attentional focus. In the case of an eyewitness, the fear response is often triggered by the presence of a weapon. As the NAS report authors point out, the “weapon focus” is certainly a beneficial adaptation in terms of survival, but it is also “surely detrimental to one’s efforts to bear witness.”

Once the sensation and attention processes are complete, objects and events enter the perception stage. Perception is not a passive action; it is not similar to the recording of an object on film. Instead, according to the NAS report, “visual perception is constructive and entails (1) integrating and segmenting attended attributes of the visual image into objects, (2) complementing and interpreting the product with expectations derived from memory of prior experiences with the world, and (3) assigning meaning and emotional valence by reference to prior knowledge of function and value.” All aspects of the perception process are subject to the interference of noise.

Environmental issues, such as lighting and angle of view, can distort the perception process. Where an object appears in the observer’s field of vision can also alter what is perceived. But there are bigger noise problems associated with the perception process.

Memory and expectations routinely fill in the blanks of what is sensed and attended to during perception. In fact, the observer’s own internal noise—expectations from prior experience—can influence what is perceived as much, or more than, the pattern of light that strikes the eye. The NAS report points out the good reason for this: “the sensory input (the pattern of light received) is often noisy, incomplete, and ambiguous, and memories of what is likely to be out there, given the context, are called on to fill in the blanks, reconcile ambiguities, and leave clear and coherent precepts.”

With this understanding of the perception stage in hand, it is easy to see how eyewitness identification can go wrong. Perception is not a certainty but a probability. And it is not always necessarily “probably” accurate. Indeed, the process whereby blanks are filled in is entirely capable of fabricating content that does not reflect reality—without the observer’s conscious knowledge. The observer will experience the perception—real or not—with 100 percent certainty.

“[L]earned properties of the world are capable of sharply altering our experience and, moreover, reinforcing our convictions about what we have seen, even in the face of countermanding sensory evidence,” the NAS report states. “In view of this inherent dependence of perception on prior experiences and context—and, importantly, the fact that the viewer is commonly none the wiser when perception differs from the ‘ground truth’ of the external world—it appears that accurate eyewitness identification may be difficult to achieve.”

Another source of noise that can interfere with the perception process is the categorical nature of perception itself. We lump objects—including human faces—into categories based on internal associations. When an object such as a face is witnessed for a brief period of time, what is perceived may be more akin to a category prototype than the actual face. The NAS report authors make the point this way: “[A]lthough you may have seen the iconic Marlboro Man countless times on billboards and in magazines, it is unlikely that you could distinguish him in a lineup from other square jawed mustachioed men.”

* Memory. Once the sensation, attention, and perception processes are complete, the resulting visual perceptual experience is stored in memory. Similar to the processes involved with vision, memory processes are all subject to error-inducing noise.

During memory encoding, visual stimuli and perceived objects are stored in either short-term or long-term memory. Short-term memory is by definition of limited duration and is both restricted in capacity and subject to rapid decay. Short-term memory is highly susceptible to the interference of noise, such as subsequent sensory events, emotions, prior knowledge, and expectations. Long-term memory lasts much longer than short-term memory, but it is also subject to interference, mostly at the hands of ongoing experience and the recall process. Long-term memory, in fact, evolves.

After memory is encoded, it heads for storage. The memory storage process is uniquely misunderstood; similar to the perception process, memory storage is nothing like recording an event on film or videotape. Instead, stored memories are subject to continual challenge and modification. According to the NAS report, “We forget, qualify, or distort existing memories as we acquire new perceptual experiences and encode new content and associations into memory.”

And that’s not all. Reactivation of a stored memory is an interactive process. Inconsistencies may be smoothed over, embellishments added, and disturbing content omitted. Memories that are altered during reactivation are stored in an altered state.

Stored memories evolve in other ways. New perceptions and memories impact stored memories. Long-term memories are updated with new content that did not exist when the original memory was stored. Such new content need not be accurate. Scientists have successfully planted fabricated information into the memories of experimental subjects.

In the eyewitness context, the plasticity of stored memories presents special concerns.

“A witness’ inevitable interactions with law enforcement and legal counsel, not to mention communications from journalists, family, and friends, have the potential to significantly modify the witness’ memory of faces encountered and of other event details at the scene of the crime,” the NAS report states. “[This conclusion] has important implications for law enforcement and the legal process and calls into question the validation of in-court identifications and their appropriateness as statements of fact.”

The process of memory retrieval is similarly subject to noise. The categorical nature of memory can lead to “intrusion errors,” whereby content associated with events of a general type is added to what is recalled. As an example, the NAS report suggests that an observer of a robbery may unwittingly incorporate a gun into the retrieved memory, because guns are often associated with robbery.

An observer may also forget how he or she knows something that is recalled. “Source memory failure” can lead an observer to attribute knowledge gained from a completely different source to an earlier experience.

In the eyewitness context, this phenomenon may manifest when an observer attributes something learned from the police or the news to the actual observation.

The processes involved in the encoding, storage, and retrieval of memory are all highly interactive. While memory is commonly understood to be similar to a static recording and playback device, scientists caution that it is more akin to Wikipedia than a dusty set of encyclopedias. Elizabeth F. Loftus, perhaps the leading authority on the psychology of memory, said memory is “more akin to putting puzzle pieces together than retrieving a video recording.”

The many ways in which vision and memory can be distorted present serious problems in the criminal justice system. When it comes to eyewitness identification, the implications are profound. According to the NAS report, the problems associated with eyewitness identification may, in fact, be “insurmountable.”

“Memory is often far from a faithful record of what was perceived through the sense of sight: its contents can be forgotten or contaminated at multiple stages, it can be biased by the very practices designed to elicit recall, and it is heavily swayed by emotional states associated with witnessed events and their recall,” the NAS report cautions. “From this analysis, the committee must conclude that there are insurmountable limits on vision and memory imposed by our biological nature and the properties of the world we inhabit.”

Reform Efforts: Ameliorating the Impact of Biological Limitations

Going all the way back to Harvard professor Hugo Münsterberg’s 1908 book, On the Witness Stand, scientists have been aware of the frailties of eyewitness identification. But it wasn’t until the 1970s that researchers began to focus their attention on finding solutions to the pernicious problem of eyewitness misidentification. The last several decades have seen significant efforts in scientific and criminal justice communities to find and implement reforms.

Scientists and advocates seeking to better understand how to ameliorate the biological limitations on accurate eyewitness identification divide relevant factors into two categories: system variables and estimator variables.

System variables may be defined as those factors that can and should be controlled by the criminal justice system. System variables include the procedures involved in showups, photo arrays, and lineups. Because these variables have a significant impact on the accuracy of eyewitness identification and are subject to modifiable policies, practices, and procedures, many of the recommendations for reform target this category.

Estimator variables are those factors that are beyond the control of law enforcement. Lighting, distance, and angle of view are all estimator variables. Other such variables include the presence (or absence) of a weapon, race (of the observer and the actor) and the level of stress experienced by the observer. While estimator variables cannot be controlled, they can be better understood and accounted for.

The practices and procedures involved in photo arrays and lineups are system variables that are highly amenable to reform. Without scrupulous adherence to carefully crafted procedures, the identifications that result from these methods may have been influenced by the unintentional communication of expectations from law enforcement officer to eyewitness. Bodily gestures, facial expressions, and nonverbal cues—intentional or unintentional—are well known to influence the eyewitness identification process.

Fortunately, this problem can be wholly eliminated in both the photo array and lineup context through the use of double-blind procedures. An integral part of the scientific method, double-blind testing ensures that the test administrator is “blind” to which person in the photo array or lineup is the suspect. Experts recommend the use of double-blind procedures because they reduce the likelihood that an eyewitness identification will be improperly influenced by law enforcement officers.

Some law enforcement agencies push back on the double-blind reform recommendation. Officers tend to believe (incorrectly) that due to their professional training, they do not improperly influence lineups and photo array identifications when they know the identity of the suspect. This belief is misplaced, however; no amount of training or expertise will change what an officer unintentionally or unconsciously does during an eyewitness identification procedure. University of Wisconsin law professor Keith A. Findley made this point in his 2016 article, “Implementing the Lessons from Wrongful Convictions: An Empirical Analysis of Eyewitness Identification Reform Strategies.”

“This recommendation is not based upon any doubts about police integrity,” writes Findley. “[R]ather, it is based on the well-accepted understanding that people are influenced by their own beliefs, and that they can unknowingly leak information, which can influence the subject’s response on the tests and the administrator’s interpretation of the results.”

Law enforcement agencies sometimes cite financial and manpower limitations in rejecting the use of double-blind procedures. Finding an officer who is not aware of the subject’s identity is not always possible. However, it isn’t always necessary. Shielding a test administrator who is “in the know” from observing the lineup is generally good enough. In the case of a photo array, photos can be placed in folders, shuffled, and opened so that the administrator cannot see the photographs, but the eyewitness can.

The integrity of a photo array or lineup identification also can be improved through the appropriate use of fillers. Fillers are the non-suspects included in the photo array or lineup, and experts strongly suggest that they be chosen carefully. Fillers should resemble the eyewitness’ description of the perpetrator, and the actual suspect should look similar to the fillers.

The use of fillers who closely resemble the suspect guards against the eyewitness picking someone solely because that person is the only one who resembles the perpetrator. Fillers who are similar to the suspect also protect against an eyewitness identifying someone solely because that person stands out.

Experts also recommend the use of standardized witness instructions during a photo array or lineup procedure. An eyewitness may feel pressure to pick someone, even if they are not sure that the perpetrator is present. This pressure can be alleviated by instructing the witness that the perpetrator may or may not be in the photo array or lineup, that the test administrator does not know who the perpetrator is, and that the investigation will continue regardless of whether the witness selects a suspect.

Given what we now know about the malleability of human memory, experts further recommend that investigators document the witness’ confidence in his or her identification. Such documentation should be verbatim and taken immediately following an identification. This is crucial because as the NAS report points out, “self-reported confidence at the time of trial is not a reliable predictor of eyewitness accuracy.”

“[T]he committee recommends that law enforcement document the witness’ level of confidence verbatim at the time when she or he first identifies a suspect, as confidence levels expressed at later times are subject to recall bias, enhancements stemming from opinions voiced by law enforcement, counsel and the press, and to a host of other factors that render confidence statements less reliable,” the NAS report states.

Video Evidence

In addition to the confidence statement, the NAS report recommends that investigators videotape the entire witness identification process. Doing so provides a permanent record of what happened at the time of the initial identification. This is not intended to prevent eyewitness misidentification but to allow for later investigation into why an identification may be mistaken.

This video evidence could be used as part of a pretrial judicial inquiry into the reliability of an eyewitness identification, which the NAS report also recommends. Such an inquiry would increase the likelihood that an identification used at trial is reliable, and it would encourage investigators to adopt and utilize best practices.

Reform recommendations designed to ameliorate the impact of estimator variables focus on training and instruction. These recommendations are of no less import than those designed to limit the impact of system variables. Indeed, ensuring that law enforcement investigators, juries, and judges understand the way that environmental and biological factors influence an eyewitness identification may be more important than prescribing procedures designed to limit the impact of system variables.

As such, experts recommend that law enforcement agencies provide their officers with training on vision and memory. If investigators better understand the limitations inherent in these biological systems and processes, they are more likely to obtain accurate eyewitness identification evidence. Such training should be welcome in the law enforcement community, as it is now well established that guilty parties have gotten away with many heinous crimes due to investigative failures during the eyewitness identification process.

In addition to the training of law enforcement officers, experts recommend that juries be educated as to the limitations and pitfalls of eyewitness identification. Given the jury’s role, and the counterintuitive nature of current scientific understanding of vision and memory, most experts, including the NAS report, recommend the use of expert testimony on the relevant precepts of human vision, memory, and identification.

“Expert witnesses can explain scientific research in detail, capture the nuances of the research, and focus their testimony on the most relevant research,” the NAS report states. “Expert witnesses can convey current information based on the state of research at the time of trial. Expert witnesses can also be cross-examined, and limitations of the research can be expressed to the jury.”

Not all courts allow expert testimony as to eyewitness identification, however, and federal courts are divided on the issue. Judge Easterbrook of the U.S. Circuit Court of Appeals for the Seventh Circuit wrote in a 2009 opinion why it makes sense to allow expert testimony on this crucial issue.

“It will not do to reply that jurors know from their daily lives that memory is fallible,” wrote the judge in United States v. Bartlett, 567 F.3d 901 (7th Cir. 2009). “The question that social science can address is how fallible, and thus how deeply any given identification should be discounted. That jurors have beliefs about this does not make expert evidence irrelevant; to the contrary, it may make such evidence vital, for if jurors’ beliefs are mistaken then they may reach incorrect conclusions. Expert evidence can help jurors evaluate whether their beliefs about the reliability of eyewitness testimony are correct.”

In the absence of expert testimony, reform advocates recommend the use of jury instructions on issues related to eyewitness identification. New Jersey courts use a detailed set of instructions emphasizing the need to understand the limitations of eyewitness identifications and to carefully weigh such evidence. Studies have shown, however, that these instructions reduced jury reliance on both strong and weak eyewitness evidence. Thus, expert testimony is superior to jury instructions.

The Current State of Reform Efforts

It is abundantly clear that eyewitness identification evidence presents significant and unique concerns in the criminal justice system. Scientists have known about biological limitations on vision and memory for many decades. Moreover, any questions as to whether these limitations impact the criminal justice system have been answered. Eyewitness identification testimony is the most common feature of DNA-exonerated wrongful convictions.

Reform recommendations abound, but the question remains: Are they being implemented? Law professor Keith Findley set out to answer this question in his 2016 article. Unfortunately, Findley found “an alarming disconnect that has emerged between a growing body of knowledge about wrongful convictions and the steps that can be taken to reduce them, on the one hand, and efforts in the criminal justice system to implement these measures, on the other.”

Findley considered the findings of a 2013 survey conducted by the Police Executive Research Forum (“PERF”) in arriving at this conclusion. The PERF survey found that the overwhelming majority of law enforcement agencies have no written policies on showups (76.9 percent), photo arrays (64.3 percent), or lineups (84 percent). Even worse, over 50 percent of the surveyed agencies had no standardized witness instructions for photo arrays or lineups, 69 percent permitted non-blind photo arrays, and a whopping 92 percent allowed non-blind lineups.


The processes through which knowledge about problems with eyewitness identification evidence are translated into reform may be the culprit. Findley examined four such processes in an effort to identify the method that most effectively results in true reform.

Judicial oversight is the least effective method through which reforms to the use of eyewitness evidence may be actualized. Courts are historically very hesitant to intrude on law enforcement practices, and the Due Process Clause has proven to be an ineffective vehicle for reform. The Supreme Court’s due process standard for identification procedures is nearly half a century old and largely fails to ensure the reliability of such evidence.

The standard was announced in Manson v. Brathwaite, 432 U.S. 98 (1977). The factors considered in making a reliability determination are: “(1) the opportunity of the witness to view the perpetrator at the time of the crime, (2) the witness’ degree of attention at the time of the crime, (3) the accuracy of the witness’ prior description of the perpetrator, (4) the level of certainty demonstrated by the witness, and (5) the time between the crime and identification.”

The Brathwaite factors largely fail to ensure the reliability of eyewitness identification evidence. Moreover, according to Findley, “because [the Brathwaite factors] are mostly self-reported by the witness, [they] are in fact distorted by the very suggestiveness they are meant to overcome.” The NAS report came to a similar conclusion.

“The Brathwaite test evaluates the ‘reliability’ of eyewitness identifications using factors derived from prior rulings and not from empirically validated sources,” the NAS report states. “It includes factors that are not diagnostic of reliability and treats factors such as the confidence of a witness as independent markers of reliability when, in fact, it is now well established that confidence judgments may vary over time and can be powerfully swayed by many factors.”

Judicial oversight as a method of eyewitness identification reform is further hobbled by a significant issue: The vast majority of criminal cases are resolved at the plea bargain stage. As a result, eyewitness identification used to obtain a conviction often doesn’t make it into court.

A few state courts have stepped in to update reliability tests in the absence of Supreme Court action. The Wisconsin Supreme Court went so far as to jettison the whole idea of judicial reliability determinations, instead ruling that eyewitness evidence should be assessed solely on the basis of whether law enforcement efforts were unnecessarily suggestive. Unfortunately, the ruling in State v. Dubrose, 699 N.W.2d 582 (Wis. 2005), has been limited to show-up evidence, despite its applicability to all eyewitness identification evidence.

New Jersey has completely rejected Brathwaite. In State v. Henderson, 27 A.3d 872 (N.J. 2011), the New Jersey Supreme Court found the Brathwaite framework inadequate and instructed courts to consider a number of system variables when ruling on the admissibility of eyewitness evidence. Oregon went even further, relying not on its constitution, but its rules of evidence to impose constraints on the gathering and use of eyewitness evidence. State v. Lawson, 291 P.3d 673 (Or. 2012).

Professor Findley firmly establishes, however, that the court system is not the place to look for reforms. The U.S. court system “remains largely unresponsive to the lessons from social science,” and is “quite sensitive to pressures against moving too quickly in the area of police reform.”

The top-down legislative mandate is a more effective vehicle for reform but faces drawbacks of its own. At least eight states—Colorado, Connecticut, Georgia, Illinois, Maryland, North Carolina, Ohio, and Vermont—have enacted laws that mandate the use of specific procedures in the eyewitness identification context. These legislative mandates largely adopt requirements and best practices rooted in current scientific understandings of eyewitness evidence.

The top-down approach has many advantages. One significant advantage is that it does not leave the keys to reform in the hands of law enforcement officials, who have very little incentive to change anything about the eyewitness identification process. And as the Innocence Project notes, a legislative directive is immediate and unquestionable.

“A clear advantage of a statute is that it assures uniformity and consistency in expectations of practice across a given state and accomplishes this goal promptly, rather than uneven implementation over a protracted period of time,” the group said. “Another benefit legislation can offer is its ability to provide clear direction to the courts about how to consider eyewitness evidence that has been gathered in violation of best practices. Finally, legislation can provide law enforcement with both the resources and direction for necessary training for improved eyewitness identification protocols.”

Findley notes several drawbacks to the top-down, command and control model, however. First, political realities sometimes hamper the imposition of laws that can be seen as doubting the professionalism of law enforcement officers. Second, legislation is not necessarily well-suited to establishing best practices in an area where the science is still evolving. Third, laws tend to be general, and as such, they may be less able to direct procedures and rules that require specificity.

The final limitation on top-down legislative approaches involves police culture. Findley points out that police are “notoriously resistant to criticism and change from outside.” When it comes to mandates that they do not like, police are known to circumvent.

“It is therefore possible that, while the legislative mandates make good ‘law on the books,’ they are not as effective as ‘law on the streets,’” Findley writes. “Indeed, initial anecdotal evidence suggests that, in some places at least, police may not be complying widely with the requirements of such eyewitness identification statutes.”

This is, of course, utterly ridiculous. The police must follow the laws handed down by the legislature, just as other citizens must. However, the concerns are real and lead to the third method through which eyewitness identification reform may be attempted: The ad hoc bottom-up method.

In this approach, police are simply encouraged to sort out eyewitness identification problems themselves. It is not hard to imagine why this doesn’t work, but California serves as an example.

When Governor Arnold Schwarzenegger vetoed a 2007 legislative effort to establish best practices, he said that even voluntary state guidelines would interfere with police operations. That left reform of eyewitness identification procedures and practices entirely up to local agencies.

Findley found that most California counties and police agencies have done nothing. While the California Association of Chiefs of Police has pushed for the establishment and adoption of best practices, as of 2014, only five of California’s 58 counties have done so.

The final and most promising approach to eyewitness identification reform is what Professor Findley calls the experimentalist bottom-up method. This hybrid approach mandates policies on eyewitness identification procedures but leaves it to law enforcement to develop their own policies. Mandating a policy ensures that reform efforts are actualized and leaving the particulars to police agencies alleviates the “us vs. them” mentality, while also giving police officers a stake in the outcome.

In 2005, Wisconsin enacted legislation mandating that law enforcement agencies develop written policies and procedures governing eyewitness identification. Findley collected data on police policies in order to assess the effectiveness of the experimentalist bottom-up method in Wisconsin. He found that many policies were adopted or revised as a result of the statutory mandate.

Unfortunately, the approach has had less than stellar results across the board in Wisconsin. One surprising find in the data was a number of agency admissions that they were unaware of the statutory requirement. Many of those agencies soon developed policies, but Professor Findley pointed out that were it not for his research, they most likely would not have. This failure suggests that the use of an experimentalist bottom-up approach requires a system for ensuring accountability.

Where Are the Reforms?

Regardless of the method employed, there is little doubt that reforms to eyewitness identification policies and procedures are necessary. Scientific understanding of the limitations that are inherent in biological processes, such as vision and memory, has increased exponentially in the last several decades. Unfortunately, criminal justice lags far behind in the practical application of that knowledge.

The constituents of the law enforcement side of the justice system should support such reform. Ensuring the accuracy of eyewitness identification doesn’t just keep the innocent out of prison, it ensures that the guilty party is caught and punished.

Jennifer Thompson is doing her part to get the word out on the perils of faulty eyewitness identification procedures. Despite her initial misgivings, Thompson came to grips with the injustice that resulted from the misidentification of Ronald Cotton as the man who assaulted her. Thompson is now the president of the Washington, D.C.-based nonprofit Healing Justice, which “focuse[s] on recovery and healing for those harmed by wrongful convictions.”

Thompson also befriended Cotton after they met soon after Cotton’s release. The two coauthored a book about their experiences called Picking Cotton. They now travel the country together speaking about the toll of eyewitness misidentification and the role of apology and forgiveness in the criminal justice system.

Thompson told The New York Times that she hoped that something good could come out of the nightmare scenario that she and Ronald Cotton shared.

“I will never forget the day I learned about the DNA results,” Thompson said. “I was standing in my kitchen when the detective and district attorney visited. They were good and decent people who were trying to do their jobs—as I had done mine, as anyone would try to do the right thing. They told me: ‘Ronald Cotton didn’t rape you. It was Bobby Poole.’”

“The man I was so sure I had never seen in my life was the man who was inches from my throat, who raped me, who hurt me, who took my spirit away, who robbed me of my soul,” Thompson continued. “And the man I had identified so emphatically on so many occasions was absolutely innocent. . . . If anything good can come out of what Ronald Cotton suffered because of my limitations as a human being, let it be an awareness of the fact that eyewitnesses can and do make mistakes.”

That statement was made over 18 years ago. Sadly, the last two decades have seen little improvement in the policies and procedures used to obtain eyewitness identification evidence. This leads to an unavoidable conclusion: Undoubtedly, there are many more Ronald Cottons sitting in prison, right now, convicted of crimes they did not commit solely on the basis of an eyewitness identification—evidence that is far less reliable than anyone wants to believe. 


Additional sources: Identifying the Culprit: Assessing Eyewitness Testimony by the National Academy of Science, National Research Council of The National Academies (The National Academies Press, 2014); “Implementing the Lessons from Wrongful Convictions: An Empirical Analysis of Eyewitness Identification Reform Strategies” by Keith A. Findley, Missouri Law Review (Spring 2016); “Eyewitness Identification: A Policy Review” by The Justice Project (2007); Thinking Fast and Slow by Daniel Kahneman (Farrar, Straus and Giroux, 2011); Phillips v. Allen, 668 F.3d 912 (7th Cir. 2009); United States v. Bartlett, 668 F.3d 912 (7th Cir. 2009). 

About the Author: Christopher Zoukis is the author of Federal Prison Handbook, Prison Education Guide, and the forthcoming Directory of Federal Prisons. He is the managing director of the Zoukis Consulting Group, a boutique federal criminal justice consultancy that assists attorneys, criminal defendants, and prisoners with their prison preparation, in-prison, and reentry needs. He can be found online at

As a digital subscriber to Criminal Legal News, you can access full text and downloads for this and other premium content.

Subscribe today

Already a subscriber? Login



Disciplinary Self-Help Litigation Manual - Side
Advertise Here 3rd Ad
The Habeas Citebook Ineffective Counsel Side