Eyewitness Identification: Psychological Aspects

The scientific eyewitness identification literature has tended to rely on a distinction between estimator variables and system variables (Wells, 1978). Estimator variables are those that affect the accuracy of eyewitness identifications, but cannot be controlled by the criminal justice system. System variables also affect the accuracy of eyewitness identifications, but the criminal justice system can control those variables. Estimator variables tend to revolve around factors involved in the acquisition phase, such as lighting conditions, distance, arousal, the presence of weapons, and so on. System variables tend to revolve around factors involved in the retrieval phase, such as the structure of a lineup, instructions given to witnesses prior to viewing a lineup, and so on.

The methods used in the scientific eyewitness identification evidence typically involve staging live crimes or showing video events to people. Because the events are created by the researchers, it is known with certainty who the actual "perpetrator" was and the performance of eyewitnesses in picking him/her from a lineup can be scored systematically. These eyewitnesses can also be asked to indicate their confidence in the identification decision, thereby permitting analyses of the relation between confidence and accuracy. Systematic manipulations to key variables (e.g., structure of lineup) allows for a causal analysis of variables that affect identification accuracy, eyewitness confidence, and the relation between the two.

Estimator variables. One of the estimator variables that has received considerable attention is the race of the perpetrator relative to the race of the eyewitness (Bothwell, Brigham, and Malpass). A consensus now exists that it is more difficult to identify the face of a stranger from another race than to identify the face of a stranger from one's own race (Meissner and Brigham). There appears to be an element of symmetry to this effect. For instance, white Americans have more difficulty identifying the faces of black Americans than they do of other white Americans, and black Americans have more difficulty identifying the faces of white Americans than they do of black Americans. The precise mechanisms underlying this problem are not fully understood, although most evidence suggests that it is largely a matter of experience rather than prejudice. Another estimator variable that is frequently cited is weapon focus. Experiments suggest that the presence of a weapon draws attention toward the weapon and away from the weapon-holder's face, resulting in less reliable identification performance by eyewitnesses (Steblay, 1992). Stress, fear, and arousal have rarely been studied with regard to identification evidence (as opposed to recall) and the problems with studying these variables in an ecologically valid manner are complex. Gender, intelligence, and personality factors appear to be weakly, if at all, related to the tendency to make correct or mistaken identifications (Cutler and Penrod).

System variables. Scientific understanding of system variables has progressed more rapidly than it has for estimator variables. A primary reason for this is that the "payoff" for understanding system variables may be higher than it is for estimator variables, leading researchers to invest more in system variable research than in estimator variable research. This difference in payoff owes to the fact that an understanding of system variables can inform the criminal justice system about ways to improve the accuracy of eyewitness identification evidence.

System variable research has focused primarily on four factors, namely the instructions to eyewitnesses, the content of a lineup, the presentation procedures used during the lineup, and the behaviors of the lineup administrator. In attempting to understand the importance of these system factors, it is useful to describe briefly the process through which mistaken identifications seem to occur. A dominant account of the process of eyewitness identification that has emerged is the relative judgment process. According to this account, eyewitnesses tend to select the person from the lineup who most closely resembles the perpetrator relative to the other members of the lineup. This process works reasonably well for eyewitnesses as long as the actual perpetrator is in the lineup. When the actual perpetrator is not in the lineup, however, there is still someone who looks more like the perpetrator than the remaining members of the lineup, thereby luring eyewitnesses to pick that person with surprising frequency.

The relative judgment process leads to a rapid understanding of why is it critical to instruct eyewitnesses that the actual perpetrator might or might not be present in the lineup before showing the lineup to eyewitnesses. Experiments show that failure to instruct eyewitnesses in this manner leads to a very high rate of choosing, even when the actual perpetrator is not present (Malpass and Devine, 1981a). Proper instructions warning the eyewitness that the perpetrator might not be present do not eliminate the relative judgment tendency altogether, but they do reduce the magnitude of the problem. Importantly, proper instructions lead eyewitnesses to less often mistakenly pick someone when the perpetrator is not in the lineup, but have little effect on their ability to pick the perpetrator when the perpetrator is in the lineup. The result of proper instructions is a net improvement in eyewitness identification performance (Steblay, 1997).

The relative judgment process also has implications for how investigators should select lineup fillers. A lineup filler is a known-innocent member of a lineup. Normally, a lineup will have one suspect and several (five or more) fillers whose primary purpose is to prevent the eyewitness from simply guessing. If an eyewitness is merely guessing, then odds against selecting the suspect are N:1 (where N is the number of fillers). However, if investigators use fillers who do not fit the general description of the suspect (as provided previously by the eyewitness) whereas the suspect does fit that description, then the lineup is said to be biased against the suspect. As predicted by the relative-judgment process, lineups in which the fillers do not fit the description of the perpetrator lead eyewitnesses toward picking the suspect, even if the suspect is innocent, because the suspect most closely resembles the perpetrator relative to the other lineup members. Making sure that each lineup member fits the general verbal description of the perpetrator does not lead eyewitnesses to fail to recognize the perpetrator when he is in the lineup, but it does help prevent mistaken identifications of the innocent suspect when the actual perpetrator is not in the lineup (Wells, Rydell, and Seelau, 1993).