Eyewitness Memory

Eyewitness memory plays a pivotal role in many criminal trials. A substantial body of psychological research on eyewitness memory has developed over the years. This research paper examines various types of eyewitness memory, factors that influence eyewitness memory, methods of improving eyewitness memory, and how eyewitness memory is evaluated in the course of investigations and criminal trials.

History of Research on Eyewitness Memory

The advent of psychological research related to the legal system can be traced to 1908, when Alfred Binet demonstrated that a person’s response to questioning could be influenced by the way in which the question was asked. Although his work did not have a profound influence on the legal system at the time, it was the beginning of empirical research involving witness testimony. Soon thereafter, William Stern actually applied this research directly to eyewitness testimony. He was able to demonstrate that eyewitnesses are susceptible to error and that witnessing variables, such as emotions at the time of the event, can serve to affect the accuracy rate. Around the same time, Hugo Munsterberg released his book, On the Witness Stand, which examined problems associated with eyewitness memory as well as jurors’ inability to accurately assess eyewitness testimony. Munsterberg’s research was met with quick criticism from John Henry Wigmore, who stated that psychological research was not of a nature that the legal system could use. It is fair to say that Munsterberg’s research was not up to present-day methodological standards, but even with this caveat, the importance of his work cannot be diminished. He was the first researcher to examine issues related to eyewitness memory in a systematic and scientific manner.

It was not until the 1970s that eyewitness research was again brought to the forefront, this time by Elizabeth Loftus. She demonstrated, using realistic stimuli such as videotaped and live events, that memory in general, including eyewitness memory, could be altered simply by the way in which the interviewer asked the question. Because of her rigorous methodological controls, she was able to both examine the quantity of eyewitness memory and assess the accuracy and quality of the remembered information. Her research spurred interest in the topic among her students and colleagues. This included Robert Buckhout, who demonstrated the prevalence of eyewitness identification errors. Although there was still some skepticism as to the use of eyewitness research in the legal field, the research gained some ground in 1978, when Gary Wells distinguished between estimator variables and system variables. Establishing this dichotomy made it possible for critics to comprehend how psychological research could contribute to the legal system and allowed researchers to focus their efforts on issues that the legal system could implement.

Types of Eyewitness Memory and Factors Affecting Eyewitness Memory

Broadly speaking, eyewitness memory can be divided into two general classes: eyewitness recall and eyewitness identification, corresponding to the traditional recall-and-recognition distinction pervasive in the cognitive psychological research on human memory. Eyewitness recall often plays an important role in the investigation of crimes. When a crime occurs, police officers responding to the crime interview the eyewitnesses regarding their memories associated with the crime, including descriptions of the perpetrator(s) and the crime itself. The interviewee may be interviewed numerous times throughout the investigation. Some of the details recalled by the eyewitness, such as a description of a weapon or description of clothing worn by the perpetrator, may become important later in the investigation or even at trial.

Research on eyewitness recall has examined factors that influence the accuracy of eyewitness descriptions, such as levels of stress experienced by the eyewitness or the presence of a weapon. One of the most widely studied factors, witness questioning, relates to the information that is given to witnesses after they experience the event and the way in which the witnesses are questioned about the event. It has been repeatedly demonstrated that the wording and intonations of questions can lead eyewitnesses to provide incorrect information. In this research, participants who witnessed an event are questioned in a way that induces subsequent reports containing false details. For example, participant witnesses were asked either “Did you see a broken headlight?” or “Did you see the broken headlight?” Even though only one word is different between the two conditions, participants who heard the word “the,” rather than the word “a,” were more likely to indicate that they had seen a broken headlight. The majority of research discussed thus far has involved adults. However, research has also demonstrated sizeable effects of postevent information on both older adults and children alike. In fact, children below the age of 3 to 4 and adults over the age of 65 seem to be the groups that are most likely to fall prey to postevent suggestion.

Not only can memory of an event be altered by the way in which the witness is questioned, but the act of repeatedly questioning a witness can also have profound effects on the witness’s memories of the event. For example, repeatedly asking college students to think about events that were plausible but did not occur in their childhood (e.g., knocking over a punch bowl at a party) led them to accept the events as true. The most famous example is the “Lost in the Mall” demonstration. In this demonstration, an adolescent boy was asked to remember when he was lost in a mall as a young boy. The boy, who initially indicated he did not remember the event, was asked to simply think about the event and write down his memories of the event each night. This is a therapeutic technique called journaling. After 2 weeks of journaling, not only did he remember the event, an event that in fact never occurred, but he also provided specific details of the event, such as the color of the man’s shirt who found him as well as the coarseness of the man’s hand. Critics of false memory research argue that lab-created false memories are plausible and fairly benign. They dispute the generalizability of the results to legal and therapeutic settings in which the memories recalled are often more traumatic and more atypical (e.g., childhood sexual abuse).

Eyewitness Recognition

Eyewitness identification of perpetrators can play a central role in the investigation of a crime and in resolving the case, whether by trial or through plea bargaining. Eyewitness identification can occur spontaneously, as is the case when a crime victim encounters her perpetrator in public and calls the police. Eyewitness identification can also occur through identification tests, such as showups, photo arrays, and live lineups. These identifications appear very persuasive and compelling to jurors.

Research on eyewitness identification has examined a large array of factors that are thought to influence identification accuracy. These factors include the conditions under which the crime occurred, exposure time to the perpetrator, stress experienced by the witness, the presence of a weapon, disguises worn by the perpetrator, and the time between the crime and the identification.

One such factor that has received a significant amount of attention in the psychological literature is the cross-race effect or own-race bias (ORB). That is, the more an eyewitness’s race is congruent with the race of the perpetrator the more likely the witness will make an accurate identification. In contrast, when the witness and perpetrator are from different races, identification accuracy is impaired. Although there are some differences in false identifications (specifically, White participants demonstrate a larger ORB effect compared with Black participants), the results of accurate identifications indicate no differences in the ORB effect between participants. One theory of ORB posits that the extent to which the ORB effect occurs is dependent on how much interracial experience a person has with the target race. This theory has been supported in that those participants who live in areas that allow for more interracial experience do not demonstrate the typical ORB effect compared with those who do not have this experience.

Methods of Improving Eyewitness Memory

The aforementioned distinction between eyewitness recall and identification accuracy is useful for explaining how psychological research has been used to develop methods for improving eyewitness memory. Practical recommendations from research on eyewitness recall have focused on how to form questions that do not mislead the eyewitness and how to avoid implanting false memories. Research has also examined whether hypnosis can be used to improve eyewitness recall, but the conclusions from this research are pessimistic.

One of the great success stories from research on improving eyewitness recall is the cognitive interview. The cognitive interview was derived from three basic processes: memory/cognition, social dynamics, and communication. The process begins by directing the eyewitness to close his or her eyes and mentally reconstruct the event. Although not always feasible, this can also be done by having the eyewitness revisit the crime scene. The interviewer should not interrupt the witness and should only ask open-ended questions. Witnesses should be encouraged to describe the event from multiple perspectives and should be asked to respond, “I don’t know” rather than guess when unsure. The interviewer should establish a rapport with the witness to balance issues of authority and encourage active participation on the part of the witness. After the eyewitness describes the event, the interviewer should use probing questions to exhaust the memory. Research has demonstrated that careful and thorough use of this procedure can lead to an increase in memory for the event without causing increases in incorrect information.

Research on improving eyewitness identification has likewise yielded impressive gains. There are various tests that are used to identify a suspect. Two of the most common of these tests are the lineup and the showup. A lineup can be conducted either live (the witness views actual people) or by using a photo spread (the witness views a series of photos). In general, a lineup usually contains several fillers, people in the lineup that are known to be innocent, and one suspect. Lineups can contain more than one suspect, but for a variety of reasons, it is not recommended.

The various aspects of lineup administration have been researched extensively. For example, the instructions that are given before the administration of a lineup can affect the likelihood of the witness choosing from the lineup; this effect occurs independently of whether or not the lineup contains the perpetrator. Furthermore, it is recommended that the witness be told that the perpetrator may not be present in the lineup, which therefore encourages the witness not to pick from a lineup that they feel does not contain the culprit. Equally important in the lineup instruction and administration are the fillers that accompany the suspect. The fillers serve as a control for guessing, and if chosen, the administrator will be aware that the eyewitness has made a mistaken identification. It is equally important that the fillers are also picked with some consideration for the description of the suspect given by the witness. To the extent that the fillers are similar in appearance to the witness’s description of the culprit, it is ensured that the witness’s subject choice was not based solely on logical deduction. The presentation of the lineup has also been researched. The two most researched presentation types are the simultaneous lineup, in which the witness views all lineup members at once, and the sequential lineup, in which only one lineup member is shown at a time. Research has repeatedly demonstrated that sequential lineup presentations produce fewer false identifications than simultaneous lineups when the culprit is not actually present in the lineup. However, correct identification rates do not differ between the two lineup presentation modes when the culprit is present in the lineup. If at all possible, a double-blind lineup procedure should always be used. The double-blind procedure refers to a lineup in which the administrator does not know the identity of the suspect. If the lineup administer is unaware of the identity of the suspect, then he or she cannot unwittingly relay information about the identity of the suspect to the witness.

The improvement in identification accuracy gained by these procedures, coupled with large numbers of DNA exonerations in recent years, has led many states to implement these reforms to ensure the fairest and most unbiased lineup identification procedures. For example, in New Jersey and North Carolina, police departments and prosecutor offices are now required to conduct sequential lineups. Similarly, Santa Ana, California, and several counties in Minnesota have opted for sequential lineups. In Clinton, Iowa, the arresting officer is not permitted in the room during the identification procedure. Many cities, such as New York and Seattle, have started using computerized programs to present photo arrays. In Chicago, as well as parts of Wisconsin and Minnesota, committees have been developed for the purpose of investigating identification procedures to reduce false identifications.

Evaluating Eyewitness Memory

In some sense, all estimator variables could be considered postdictors of eyewitness accuracy. Although research has focused on eyewitness recall, testimony, and identification, research on the postdictors has almost exclusively been limited to eyewitness identification. One of the most widely studied postdictor variables is eyewitness confidence. This is most likely the case because jurors seem to find confident eyewitnesses extremely persuasive and believable. This perception of confident eyewitnesses is understandable; intuitively, it seems as though there should be a strong positive relationship between witness confidence and accuracy. This belief is underscored by the fact that the court has suggested that jurors may employ witness confidence as an indicator of the accuracy of the witness. Unfortunately, psychological research has found unequivocally and repeatedly that the relationship between confidence and eyewitness accuracy is, at best, a weak one. Furthermore, this weak relationship deteriorates as the time interval between the event and the confidence statement increases. The reason for the lack of relationship between confidence and accuracy may be that witnesses often rely on misleading information as the basis for their confidence estimates. For example, it has been shown that confirmatory feedback increases participants’ confidence in their eyewitness identification. Simply telling a witness that they have chosen correctly increases the witness’s confidence in the accuracy of his or her identification relative to participants who are not given any feedback. This confidence inflation is especially prominent when the participants are inaccurate.

Just as eyewitness confidence serves an important function in the prosecution phase, eyewitness descriptions of the perpetrator serve an extremely important function in the investigation process. Investigators may use the descriptions to locate the suspect. The problem is that descriptions are generally incomplete and nondescript, such as “White male, 5 feet 9 inches to 6 feet, about 18 to 24 years old.” It should be noted that while the descriptions are generally incomplete, the information that is collected is usually accurate. This is especially the case when witnesses are simply asked to describe the assailant without any prompts from the investigator. The key question is whether description accuracy and length of description correlate with eyewitness identification accuracy. It could be assumed that witnesses who give more complete and detailed descriptions of the culprit would be more accurate in their identification. In fact, the U.S. Supreme Court has employed the witness description accuracy as an indicator of the witness’s reliability. However, it has generally been found that the quality and quantity of the witness descriptions are not related to identification accuracy.

Not only do witness descriptions poorly predict accurate identifications, but they may also in fact harm later identification accuracy. This is what happens in the case of verbal overshadowing. Verbal overshadowing refers to the impairment in lineup identification accuracy when the witness is asked to describe the suspect’s face prior to the lineup administration. Although witness descriptions and identifications are weakly correlated, two factors are inconsistent with this finding. First, witnesses who are particularly adept at describing faces are likely to benefit from the description. Second, the correlation between witness description and identification accuracy is improved when the culprit’s face is especially easy to describe, such as when there is a tattoo on the face or a facial disfigurement.

One estimator variable that is thought to be predictive of identification accuracy is the speed of the identification. The rationale is that witnesses with a good view of the culprit and a vivid memory of the crime should identify the witness quickly and without hesitation. For the most part, the research on speed of identification has been consistent with this logic. Furthermore, it has been noted that witnesses who indicate that the face “popped out” at them when viewing the lineup tend to be more accurate in their identification than witnesses who indicated that they had to take more time to make their decisions. Thus, there is a strong relationship between accuracy and speed of identification, with faster identification resulting in more accurate identification.

Eyewitnesses in the Courtroom

Considerable research has addressed how jurors evaluate eyewitness memory and whether their evaluations can be improved. Using trial simulation methods, research demonstrates that jurors are often insensitive to the factors that are known to influence eyewitness memory: stress experienced by the witness, weapon focus, and the influence of suggestive identification procedures. Furthermore, jurors tend to rely on factors that are known not to strongly predict identification accuracy. Witness confidence, for example, has a strong impact on jurors’ evaluations of eyewitness identifications. In these studies, highly confident witnesses are persuasive, and jurors tend to convict perpetrators on the basis of testimony by highly confident witnesses.

The apparent inability of jurors to evaluate eyewitness identification procedures in a manner consistent with the research on eyewitness identification has led some psychologists to conclude that knowledgeable psychologists should testify in court as expert witnesses. The purpose of the expert testimony is not to make a judgment as to the accuracy of the identification but rather to provide all the relevant information so that the jury can make an informed decision when assessing the reliability of the witness. Not all those in the legal system agree on the efficacy or use of eyewitness testimony during trials. Some argue that the testimony may bias the witness; others indicate that the testimony is superfluous because issues of memory are of common knowledge (a conclusion that is inconsistent with research findings). Although there has been research investigating the impact of expert testimony on factors that influence eyewitness accuracy, the results are mixed with respect to its impact on jurors.


  1. Brewer, N., Weber, N., & Semmler, C. (2005). Eyewitness identification. In N. Brewer & K. D. Williams (Eds.), Psychology and law: An empirical perspective (pp. 177-221). New York: Guilford Press.
  2. Lindsay, R., Ross, D., Read, J., & Toglia, M. (Eds.). (2007). The handbook of eyewitnesspsychology: Vol. 2. Memory for people. New York: Lawrence Erlbaum.
  3. Neuschatz, J. S., & Cutler, B. L (in press). Eyewitness identification. In H. L. Roediger (Ed.), Learning and memory: A comprehensive reference. New York: Lawrence Erlbaum.
  4. Wells, G. L., & Olson, E. (2003). Eyewitness identification. Annual Review of Psychology, 54, 277-295.

See also: