Skip to main content

Line 'em up and pick 'em out

The eyewitness lineup is an iconic scene in criminal investigations. Just reading that sentence probably summoned up a specific image in your mind: the one way glass, separating two very different rooms. In the first, four to eight (probably) men stand, slightly nervous, staring straight ahead. In the other, an officer stands with an eyewitness, asking them if they recognize any of the people in the other room. Sometimes, the witness is seated at a desk looking at photographs of the men. In either situation, your eyes are drawn to one guy in particular. The witness thinks for a long time, and then picks a person. The officer gives a huge sigh of relief and reassures the witness, “You picked the guy, thanks.”

This stereotype of eyewitness lineups is likely based on movies and media – and it is, in many instances, consistent with how lineups are conducted in real life. Unfortunately, many of the things that happen in lineups like this are problematic can lead to the wrong person being identified as the perpetrator and going to prison for the crime.

Like generating an alibi, an eyewitness lineup is a memory test. The eyewitness has a memory for who they saw commit the crime, and they should rely on that memory to say whether the person they remember is present in the lineup. Unfortunately, memory is suggestible and subject to distortion. This means that eyewitnesses can unintentionally rely on information other than their memory to make a selection. And when that happens, they’re not telling you what actually happened at the crime – they’re giving information based on memory errors, biases, and suggestion. Indeed, the Innocence Project reports that incorrect eyewitness identifications are involved in 70% of DNA exoneration cases1.

That is not to say that eyewitnesses are completely useless. Their accurate memory can be used to identify the guilty perpetrator and be used as valuable evidence at trial. It’s important to note here that by “guilty” we don’t necessarily mean the suspect in the lineup. After all, the police may be investigating the wrong person. If that’s the case, an eyewitness shouldn’t choose him or her out of a lineup, because they shouldn’t recognize them from the crime. Unfortunately, when a lineup is conducted in the way we described above, the suspect may be selected – regardless of whether or not they are guilty.

So how should a lineup be conducted in order to a) increase the rate of the guilty person being chosen and b) decrease the rate that an innocent person is chosen? Several decades of memory research have discovered problems in lineups that can lead to incorrect identifications. Most of the research on eyewitness memory follows a pretty basic structure. First, participants view some event, either via a live event, photographs or videos. After a delay period, participants then view a lineup, and pick the person they think they saw commit the mock crime. The researchers can then test different changes in the lineup procedure to see how it affects the rate at which the participants choose the perpetrator, or one of the other people in the lineup, known as “fillers.” From this, they can estimate the rates of selecting the perpetrator under specific circumstances. Much of the work we discuss here follows this paradigm, and can compare the rate of correct selection for traditional lineups versus possible improvements. In a 2014 report, the National Academy of Sciences summarized much of this research, and explained the risks posed by certain lineup procedures – and offered procedures on how to fix them2.

The first step in ensuring a well-conducted lineup occurs before the eyewitness even sees it – specifically, how the people in the lineup are chosen. Obviously, the police will include their suspect. But what about the other people, or “fillers”? In a well-designed lineup, there’s only one suspect, and all of the fillers are people the police know couldn’t have committed the crime*. After all, the best tests of memory are the ones where there is only one right answer. Imagine the extreme opposite of this; if all six people in the lineup were suspects, then even if the witness just guessed randomly a suspect would still be “identified.” By having known-innocent fillers, the police get an idea of how useful the eyewitness is. If they pick a person who couldn’t have committed the crime, then they probably don’t have very good memory for what they saw.

What the fillers look like is also important. First, all of the fillers should look similar enough to the suspect that no one person stands out. For example, if the eyewitness saw someone with a beard and one of the filler doesn’t have a beard, that inadvertently communicates the filler is…well, a filler3. This can reduce the “effective size” of the lineup – that is, if you had no memory of the crime and just guessed, what are the odds that you would pick the suspect? If fillers are selected correctly, you should have an equal chance of picking any one person. But if some feature makes a person stand out, then a witness is more likely to select them.

How the eyewitness sees the lineup can also influence the accuracy of the identification. One lineup type of procedure is called a simultaneous lineup, because the eyewitness sees the suspect and fillers simultaneously. This is akin to a multiple choice test question. And while we may prefer multiple choice questions on a college exam because they are easier, they are not what we want to use for sending someone to prison. This is because on multiple choice questions, you usually compare the answers to pick the one that looks closest to what you think the right answer is. Eyewitnesses tend to do this as well, when they look at the lineup and pick the person who looks most like what they remember. This is exactly what happened to Ronald Cotton, who was wrongfully convicted in 1985. Jennifer Thompson was attacked by Bobby Poole; when she saw the lineup, which contained Cotton but not Poole, she picked Cotton because he looked sufficiently similar to her memory of her attacker – although not identical.

The live lineup that Jennifer Thompson picked Cotton out of. Cotton is suspect #5.
Image from
Right: Ronald Cotton. Left: Bobby Poole. Poole was the man that
actually attacked Thompson, despite Thompson's
confident memory that it was Cotton. 

So how do we get participants to compare people in the lineup to their memory, rather than to each other? Research has produced two fairly easy solutions. First, witnesses should be shown suspects (or their photographs) one at a time, or sequentially, rather than simultaneously, and make a decision before seeing the next person. This way, witnesses are comparing the person in the lineup only to their memory, rather than to each other. Indeed, research has shown that the simultaneous lineup decreases the chances that a witness picks the perpetrator, but also decreases the chance the witness picks an innocent person at a much greater rate4.

Another solution is to have the police conducting the lineup inform the eyewitness that the person they saw may not be included in the lineup, and that they are not required to pick someone. The fact that the lineup might be “target absent” (meaning the true perpetrator is not included) is, in fact, true; the police’s suspect may not be the actual perpetrator. And while we think we might instinctively be aware of this, eyewitnesses feel pressured to pick someone. In class demonstrations of eyewitness lineups, many students will pick a person from a lineup that does not contain the actual perpetrator; when asked why they chose someone, many will respond that they didn’t realize not picking was an option. This happens in a low-stakes situation such as a class demonstration, but now imagine being an eyewitness, where you are trying to bring someone to justice. The pressure to select someone can drive witnesses to pick someone even when they’re not sure, yet merely reminding them they do not have to pick can drastically reduce false identifications5,6.

What the police say to the witness can affect the quality of the lineup. But it can also decrease it as well. Obviously, if an officer rushes a witness to pick, that can result in an incorrect decision. If an officer tells the witness to focus on someone in particularly, that can increase the chance of picking as well – just like a person who stands out from the rest of the lineup. What the officer says after the witness has picked someone can negatively affect the quality of the lineup, too.

Once an eyewitness has picked someone from a lineup, their role in the investigation likely isn’t over. They may be asked to make another identification, or they may testify in court. This can be difficult if a lot of time passes, or if they talk to other people about what they saw, their memory can be distorted. This same sort of distortion can occur if the police give feedback about the choice. By saying something like, “That’s the guy!” or “You made the right choice,” or even, “good job” the officer is confirming the witness’s choice – which can lead to the witness becoming overconfident.

This overconfidence is problematic for any future involvement. For example, if the witness testifies in court, they may be asked to describe their attacker and say whether they selected them out of a lineup. Research shows that compared to witnesses who don’t receive feedback, a witness that receives affirming feedback from the lineup administrator will likely be more confident about their decision7 – and confidence can be very convincing to the jury8.

Of course, police are likely not giving feedback to intentionally distort the witness’s memory. Instead, it may be an involuntary reaction to calm a very distraught witness, or the case moving forward. So what’s the best way to prevent such feedback? Use a “double-blind” procedure9. Borrowed from medical research, a double-blind procedure requires that neither party involved in the lineup (that is, the witness AND the administrator) know who the suspect is. This can be easily accomplished by asking an officer or detective working on a different case to administer the lineup, and prevents any intentional or unintentional feedback about the witness’s choice.

Unfortunately, confidence in a choice can inflate over time even without police feedback. One potential cause for this is a process called “unconscious transference” – essentially a source monitoring error where you know you’ve seen a face before but mistake where you saw it. After the first lineup (either in person or photo), the witness will correctly remember that they’ve seen the suspect before. However, they might make a mistake and say they saw the suspect during the crime, rather than in a previous lineup10. By the time a witness is testifying in court, they may have seen the suspect several times via lineups, news reports, or even in trial, but incorrectly incorporate the suspect’s face into their memory for the crime** – which can have very severe consequences if the suspect is not the actual perpetrator.

How the courts should deal with witness confidence is a bit of a tricky question. Our lay expectation is that the more confident someone is about their memory, the more accurate the memory is. However, memory research – much of it using eyewitness paradigms – has shown this is rarely the case11. That is, confidence is not indicative of accuracy: very confident people can be right or wrong, and very unconfident people can be right or wrong. As a result, researchers have warned practitioners from putting much stock in confidence.

However, recent research suggests that confidence can be indicative of accuracy, or diagnostic, in certain situations. Specifically, studies show that when a witness makes a very quick decision in which they are very confident, they very likely chose the guilty person. 12. Thus, we should put more stock in decisions that are made almost instantaneously and with high confidence – otherwise, confidence isn’t much help and can be very misleading.
Just because an eyewitness is confident at trial does not mean
they are accurate in their ID. However, jurors tend to be
convinced by high confidence. What's important is that the
witness is confident during the lineup.
However when the decisions take longer than a few seconds, or initial confidence isn’t very high, the chances of picking an innocent person are much higher

So if a movie were to shoot a scene with a good lineup, what would it look like? Before the lineup even starts, the police start a video camera to record the entire lineup. That way, both the prosecution and defense will have a record of whether the lineup was conducted fairly.  You’d have the same two rooms (or one room with a desk and photographs)***, separated by a one-way window. The officer would tell the witness they’re going to see a lineup one person at a time, and need to decide if that person is the perpetrator before continuing the lineup. The officer will also explain that the perpetrator might not be any of the people in the lineup, and that the witness should not pick a person if they’re unsure. Then, the witness will be shown the lineup, one person at a time. The lineup only contains one suspect; the rest of the people in the lineup are known to be innocent. All of the people look fairly similar to each other. No one should stand out as dramatically different from anyone else. Once the witness has said “that’s the person” the officer should thank the witness, but not give any feedback about the selection.

You’ll notice this lineup is…a bit different than what we started with. Fortunately, there is some progress on changing how lineups are conducted. The Department of Justice issued a memorandum for reform of lineup procedures in 201713 based on many of these ideas – however, reform was not mandatory. According to the Innocence Project, 19 states and some large cities have implemented these and other reforms. Unfortunately there is some pushback, such as pointing out eyewitnesses are less likely to choose a person with a sequential lineup  but many researchers and practitioners view the tradeoff acceptable – after all, our justice system is focused on protecting the innocent before punishing the guilty.

While many factors about eyewitness memory are out of the control of the legal system (such as lighting during the crime, how far away the eyewitness was, how long the eyewitness saw the perpetrator, etc.), the way a lineup is conducted and used is something within our control. In fact, it is precisely because so many factors can lead to poor witness memory that we should be as careful in administrating lineups in the best possible way. Research on the topic has uncovered many effective, easy-to-implement reforms that can improve the way lineups are conducted. And while the research continues, it is by working with practitioners to put reforms into practice that we can protect innocent suspects from wrongful conviction.

This post was written by Will Crozier and edited by Timothy Luke


*Perhaps the most famous lineup from pop culture is from The Usual Suspects. Ironically, this is also probably the worst way to conduct a lineup. In the film, a bunch of known bad guys that look nothing alike are rounded up and told to repeat a specific phrase that the witness heard during the crime. While it makes for compelling, profanity-laced entertainment, such a lineup would not serve much use because none of the people in the lineup are fillers, and there is no suspect. As such, it’s more of a fishing expedition than a real lineup.

*Going back to the example of Ronald Cotton, Jennifer Thompson exhibited such unconscious transference. Not only did she see Cotton in the original photo lineup, she also saw him in a subsequent lineup, as well as on the news leading up to her trial and during the trial itself. In fact, Thompson’s memory was so strongly distorted by seeing Cotton so many times that she failed to recognize Bobby Poole, her actual attacker, when he was brought into the court room. Instead, she maintained the belief and memory that Cotton attacked her.  

***Some have made the point that a live lineup provides better cues to memory than a photo lineup, such as being able to see the suspect from different angles and having lineup members say specific phrases. However, as Dr. Neil Brewer effectively summarizes here (, there are advantages to a photo lineup over a live lineup. Most research to date that was discussed here has focused on problems and reforms common to both forms of lineup. And on these issues, researchers are generally in agreement. On some issues, researchers don’t agree, such as live versus photo lineups – but these are relatively minor issues compared to major issues on which we’re in agreement.



[2] National Academy of Sciences (2014). Identifying the culprit: Assessing eyewitness identification. Washington, DC: National Academies Press.

[3] Wells, G. L., Small, M., Penrod, S., Malpass, R. S., Fulero, S. M., & Brimacombe, C. E. (1998). Eyewitness identification procedures: Recommendations for lineups and photospreads. Law and Human Behavior22(6), 603.

[4] Steblay, N. K., Dysart, J. E., & Wells, G. L. (2011). Seventy-two tests of the sequential lineup superiority effect: A meta-analysis and policy discussion. Psychology, Public Policy, and Law17(1), 99.

[5] Malpass, R. S., & Devine, P. G. (1981). Eyewitness identification: Lineup instructions and the absence of the offender. Journal of Applied Psychology66, 482–489.

[6] Steblay (1997). Social influence in eyewitness recall; A meta-analytic review of lineup instruction effects. Law and Human Behavior21, 283–298.

[7] Wells, G. L., & Bradfield, A. L. (1998). " Good, you identified the suspect": Feedback to eyewitnesses distorts their reports of the witnessing experience. Journal of Applied Psychology83(3), 360.

[8] Cutler, B. R., Penrod, S., & Dexter, H. R. (1990). Juror sensitivity to eyewitness identification evidence. Law and Human Behavior14, 185–191.

[9] Phillips, M. R., McAuliff, B. D., Kovera, M. B., & Cutler, B. L. (1999). Double-blind photoarray administration as a safeguard against investigator bias. Journal of Applied Psychology84(6), 940.

[10] Loftus, E. F. (1976). Unconscious transference in eyewitness identification. Law & Psychol. Rev.2, 93.

[11] Sporer, S. L., Penrod, S., Read, D., & Cutler, B. (1995). Choosing, confidence, and accuracy: A meta-analysis of the confidence-accuracy relation in eyewitness identification studies. Psychological Bulletin118(3), 315.

[12] Wixted, J. T., Mickes, L., Clark, S. E., Gronlund, S. D., & Roediger III, H. L. (2015). Initial eyewitness confidence reliably predicts eyewitness identification accuracy. American Psychologist70(6), 515.



Popular posts from this blog

Let's talk about the role of psychology in law

Will and Timothy are joined by guest Dr. Jason Chin, for a chat about the relationship between psychology and law. In this chat format, we gather regular authors and guests in Slack and have a moderated conversation, guided by prompts and questions selected in advance. Participants get to respond to each other's points, make comments, and ask each other questions in real-time. The transcript has been lightly edited. Will Crozier &#x1F419 Welcome to another Exercise in Exceptions chat! We’ve talked a lot about how psychology research can influence the law – but that research needs to make it into the courts to actually make the intended difference. However, it’s never as easy as explaining a study or two to a jury. In this chat, we’re going to discuss this collision a bit – how psychology science is used in the legal system. Timothy and I are joined by Dr. Jason Chin , a lecturer at the TC Beirne School of Law at University of Queensland, Australia. Wel

When investigations go wrong – in science and policework

A story of both a wrongful conviction and scientific fraud We’ve talked about many of the ways police investigations can go wrong, including mistaken eyewitness identifications , memory errors , and false confessions . Often, when people imagine police investigations running afoul, they imagine egregious cases in which police plant evidence or physically torture suspects to get them to produce confessions they know are false. Although situations like that do occur, mistakes in investigations require no intentional wrongdoing. A detective doesn’t need to be trying to get a false confession, for instance, in order to get one ( as our guest writer Fabi Alceste has written about) . Errors happen often without the investigators realizing anything has gone wrong. Similarly, when people imagine bad scientific research happening, they often imagine scientists fabricating data or committing outright fraud. Scientific fraud is a problem, but it’s quite rare. However, there are many questio

Interrogation, coercion, and false confessions: The case of Brendan Dassey

In December 2015, Netflix brought the justice system to the forefront of the public consciousness with its award-winning documentary Making a Murderer . The 10-episode serialization followed the story of Steven Avery – a Wisconsin man who was wrongfully convicted of rape in 1985, spent 18 years in prison before being exonerated, and was then arrested and ultimately convicted for the murder of Teresa Halbach in 2005. While the documentary focuses primarily on Steven Avery as the main suspect, Making a Murderer also highlighted a lesser-known element of the case: the confession and conviction of Brendan Dassey, Avery’s nephew. Although there’s a lot to analyze in this case, we focus here on a nagging question – did Brendan Dassey give a false confession? Dassey’s involvement in the case was little understood until Netflix’s deep dive into the case, but has emerged as one of the most scrutinized and tragic elements. Briefly, the police focused on Avery as a suspect in the disappe