Eyewitness testimony, particularly the identification of a suspect, is powerful evidence and can make the difference between a case solved and a case forever cold. However, because it is powerful and so compelling to juries, mistaken identifications are also the most frequent cause of wrongful convictions in the United States (Huff, Rattner, and Sagarin 1996). The problem is that it is difficult for jurors to distinguish between an eyewitness who has correctly identified the perpetrator of the crime and an eyewitness who has identified an innocent person. Both correct and incorrect eyewitnesses may testify at trial with great confidence, detail, and emotion.
Psychologists and legal scholars have, through empirical research, identified a number of factors that affect the accuracy of eyewitnesses. Some of the conditions that reduce the accuracy of memory are conditions over which the police have absolutely no control. Law enforcement cannot control a witness’s opportunity to observe, cannot change the lighting of an event that has already occurred, and cannot encourage the witness to keep calm during the crime.
Here, we focus on factors within the criminal investigation over which the police have either some or complete control, specifically (1) the procedures for questioning and interviewing witnesses, (2) the way the lineup is composed, and (3) the way in which the lineup is presented to the witness. Much has been learned about these aspects of criminal investigations, some just in the last few years, and the research is leading to new insights and new procedures that have been shown to increase the identifications of perpetrators and decrease the misidentifications of innocent people.
This body of research has also led to several publications and training manuals that provide up-to-date guidelines for best practices in obtaining witness statements and eyewitness identification evidence.
Interviewing, Suggestive Questions, and Postevent Information
When interviewing witnesses, one must keep in mind two important properties of memory: Memory retrieval is cue dependent and dynamic.
Cue-Dependent Retrieval
By ”cue dependent” we mean that the information retrieved from memory depends on the cues that are used to probe memory. In simpler terms, the witness report depends not only on the veracity of the witness’s memory, but also on how the questions are asked.
Not only can a suggestive question influence how the witness answers that particular question, it can also influence how the witness answers questions asked later. Consider this example from Loftus and Palmer (1974): Participants were asked about the speed of the cars in a simulated car crash. They were asked one of two questions; either ”About how fast were the cars going when they smashed into each other?” or ”About how fast were the cars going when they hit each other?” Not surprisingly the ”smashed” question produced higher speed estimates than the ”hit” question. More surprising is how these witnesses responded to subsequent questions. When asked if they had seen any broken glass (there was none), the ”smashed” group was twice as likely as the ”hit” group to erroneously state that they had.
From this example, an important lesson is illustrated. We typically think of interviewing as a one-way street in that the flow of information comes from the witness to the interviewer. However, as these experiments illustrate, the interview is actually a two-way street; witnesses can also obtain information from the interviewer.
Memory Is Dynamic
Memory changes over time; that is, it is dynamic. It would be extremely useful if memory were compartmentalized so that we would not confuse what we saw at Time A with what we heard at Time B. Unfortunately, memory does not work that way. When a witness is presented with information from outside sources, that information may find its way into the witness’s subsequent statements and testimony (see Loftus 1992 for a review).
The two-way street of interviewing is one way in which information can be transmitted to, rather than from, the witness. Other sources of postevent information include other witnesses; television and print news media; the witness’s friends, family, and acquaintances; and, of course, the police.
Lineup Composition
One of the major questions in constructing a lineup is how to pick the fillers. In two surveys (Wogalter et al. 1993, 2004), police officers indicated that they typically pick the fillers based on their similarity to the suspect. This certainly sounds, on the surface of it, appropriate.
However, as reasonable as it sounds, the strategy of selecting fillers based on their similarity to the suspect may lead to serious problems. When selecting fillers based on their similarity to the suspect, one question that arises is How similar is similar enough? A conscientious police officer, in trying to be careful to pick only ”good” fillers, may select fillers that are unnecessarily similar to the suspect, making it more difficult to correctly identify the perpetrator (Wells, Rydell, and Seelau 1993).
A second problem of suspect-matched lineups may be more difficult to detect. Consider the case in which an innocent person is suspected of a crime, in part because he fits the description of the perpetrator given by the witness. This innocent suspect is placed in a lineup with five fillers who look similar to that innocent suspect. Given this scenario, one may ask How many people are in the lineup because they fit the description of the perpetrator? The answer is only one—the innocent suspect. The other five people are in the lineup because they look similar to a person who fits the description of the perpetrator. This situation suggests that the person in the lineup who will be the closest match to the witness’s memory is the innocent suspect. It follows further that the innocent suspect is the person in the lineup who is the most likely to be identified. Indeed, laboratory research has confirmed this prediction as well (Clark and Tunnicliff 2001).
Because the biases in a lineup can be very subtle, they may not be immediately apparent. However, mistakes that initially seemed invisible may become glaring once someone else points them out. One such case involved a bank robbery in Chino, California, in 1997. A bank teller described the robber as ”looking more like a college kid than a criminal.” In the lineup shown to the bank tellers, the suspect wore a white t-shirt under a crisp-looking polo shirt, and was surrounded in the photo display by people who looked and dressed more in the style of the stereotypical criminal. This was pointed out to the jury, who quickly acquitted the defendant.
Either way one looks at this outcome, a serious mistake was made. If the defendant was guilty, then his ”walking papers” came as the result of a poorly constructed lineup. If the defendant was innocent, then the criminal investigation went astray due to the poor lineup, an innocent person was forced to stand trial, and the real robber went free.
From the research on lineup composition comes two clear recommendations: (1) It is not necessary to pick fillers who are ”gratuitously” similar in appearance to the suspect, but (2) it is necessary to make sure that the fillers are consistent with the description of the perpetrator given by the witness.
Lineup Presentation
Identification errors can also arise from the procedures used in presenting the lineup to the witness and recording the witness’s response. Eyewitness identification research has pointed to two modifications to traditional identification procedures: that (1) the lineups be administered using a blind administration procedure and (2) lineups be presented sequentially, rather than simultaneously.
Blind Lineup Administration
One of the strongest, clearest recommendations that researchers have made about the presentation of lineups is the use of the blind testing procedure. Blind testing has a long history in the behavioral, social, and medical sciences, much too long to discuss here (see Rosenthal 1966). The main characteristic of the blind procedure, for present purposes, is that it is designed to minimize the influence of the lineup administrator.
The blind testing procedure requires that the police officer who administers the lineup not be involved in the investigation.
The lineup administrator should not know about other evidence in the case, should not know the identity of the suspect, and should not know the position of the suspect in the lineup.
It may seem odd to have the lineup administered by a person who knows nothing rather than a person who is deeply familiar with the case. However, a moment’s reflection illustrates the usefulness of the blind-testing procedure. One way the defense can challenge the reliability of eyewitness evidence is to argue that the witness’s identification was the product of the suggestive behavior of the police officer who administered the lineup (Nettles, Nettles, and Wells 1996). If the police officer becomes overinvolved in the identification with statements such as ”I noticed you paused on number three,” the court and the jury are left to wonder whether the identification was the product of the witness’s memory or the police officer’s comment.
The leaking of information by the police officer need not be deliberate. Even the inadvertent leaking of information from the lineup administrator can taint the witness’s identification. Thus, to minimize the possibility of inadvertently leaking cues, the Source continues, ”Try not to say or do anything during the identification.” In considering the potential advantage of using the double-blind procedure, one need only to consider this: One cannot leak what one does not know.
Simultaneous and Sequential Lineups
In the standard identification procedure, a witness is presented with all of the individuals (or photos) in the lineup at the same time. This is called a simultaneous lineup. Research has provided strong evidence that witnesses, when presented with such a lineup, will often make relative decisions, identifying the person in the lineup who looks more like the perpetrator than anyone else in the lineup. It is not hard to see how such a decision strategy can lead to false identifications.
Researchers reasoned that these false identifications could be reduced if the witness’s tendency to make such relative judgments could be minimized (Lindsay and Wells 1985). One way to do this is to present the lineup members sequentially, rather than simultaneously. In the sequential lineup, the witness must give a yes or no response to each lineup alternative as that person is presented. To see how the sequential lineup can reduce false identifications, consider the case in which an innocent person similar in appearance to the perpetrator is shown as the third person in the lineup. In the sequential lineup the witness may note the high level of similarity but decline to make an identification— waiting instead to see if an even better match might be yet to come. In contrast in the simultaneous lineup, the witness sees all the alternatives, and is more likely to make a positive identification of that very same person (because it is clear that no better alternative is available).
The research shows that the sequential presentation of the lineup results not only in fewer false identifications, but in fewer correct identifications as well. Those who advocate the use of the sequential lineup note that the decrease in false identifications is greater than the decrease in correct identifications.
Summary
Human memory has several properties that have been shown through research to lead to errors: Human memory is incomplete, error prone, cue dependent, and dynamic, allowing memory to be influenced by—and distorted by—outside sources. Moreover, human memory is not self-correcting. A memory stored incorrectly, or distorted by exposure to postevent information, cannot easily be ”fixed,” or undistorted. Finally, witnesses (at least some witnesses) make identification decisions using decision strategies that are error prone. Despite all of these inherent limitations, the likelihood of correct identifications can be increased and false identifications minimized by following procedures that have been developed in recent years.
