An eyewitness takes the stand and describes salient aspects of an event that he or she witnessed several months earlier. Then, in the hush of the courtroom, points to the defendant and says “That's him. That's the man I saw.” Simple, clean, and convincing. And therein rests the problem; what appears to be a simple identification is in fact the result of a series of complex and potentially unreliable social and cognitive events that began unfolding several months earlier when the event was originally witnessed.

This chapter, and much of the empirical research on which it is based, operates on an assumption that there are two sources of unreliability in eyewitness accounts. First, there are some inherent limitations in human information processing. These limitations exist at sensory levels (for example, Sperling, 1960), attentional levels (for example, Broadbent, 1958; Deutsch & Deutsch, 1963; Triesman, 1964), and memory levels (for example, Miller, 1956; Atkinson & Shiffrin, 1968). But inaccuracies in eyewitness accounts are not entirely attributable to human imperfections in sensation, perception, and memory. The second source of inaccuracy in eyewitness accounts can be attributed to the methods the justice system uses to obtain information from eyewitnesses. The work of Elizabeth Loftus on the effects of misleading questions serves to make this point (see Loftus, 1979; and this volume). The account one gets from an eyewitness depends very much on the methods used to solicit the information.

The study of how to improve eyewitness accuracy by manipulating the methods used to obtain information from eyewitnesses is known as a systemvariable approach to eyewitness research (Wells, 1978).