No abstract

Two provocative claims about eyewitness confidence have recently been advanced in the eyewitnessidentification literature: (a) suspect identifications made with high confidence are highly accurate and (b) high-confidence suspect-identification accuracy is unaffected by variations in memory strength. Several recent publications have reiterated these claims in spite of providing weak or no support. We present four criteria that can be used to evaluate the empirical support for these claims. Regarding the claim that high confidence implies high accuracy, it is necessary to consider whether (a) high-confidence suspect identifications are in fact highly accurate and (b) high-confidence suspect-identification accuracy is dependent on the assumption of a perfectly fair lineup. Results of a base-rate analysis show that lowering the threshold for the claim that high confidence implies high accuracy undermines the claim substantially. Likewise, relaxing the assumption of a perfectly fair lineup may also undermine the claim. In regard to the claim that high-confidence suspect-identification accuracy is unaffected by variations in memory strength, it is necessary to demonstrate that (a) the conditions under comparison actually differ in memory strength, and (b) there is evidence that high-confidence suspect-identification accuracy is equivalent across conditions. We conclude with discussion of whether laboratory experiments have the capacity to provide valid estimates of high-confidence suspect-identification accuracy in real cases. We believe laboratory experiments have the capacity to estimate high-confidence suspect identification accuracy at the time of an identification procedure, but due to a prominent selection bias, we are skeptical of their capacity to estimate high-confidence suspect-identification accuracy at trial.

When following scientific best-practice recommendations, the simultaneous lineup is effective at demonstrating guilt. The simultaneous lineup is less effective at demonstrating innocence. A critical problem is that when a witness identifies a filler or indicates the culprit is not present, confidence does not measure the strength of match between the suspect and the witness's memory for the culprit. We propose a novel rule-out procedure as a potential remedy. After making an identification decision and expressing their confidence, participants indicated for each person they did not identify, how confident they were this person was not the culprit. The rule-out procedure better discriminated guilty suspects from innocent suspects than did the simultaneous lineup. This improvement was strictly attributable to increased potential to rule out the innocent. Interestingly, both witnesses who made rejections and witnesses who mistakenly identified fillers possessed additional memorial information that was useful for ruling out the innocent.

Visual recognition memory has a remarkable capacity to discriminate between previously seen and novel items. Yet, research on eyewitness lineups suggests that memory is useful for detecting culprit presence, but less useful for detecting culprit absence. We show that this asymmetry is predicted by the equal-variance signal-detection model. When witnesses reject lineups, they provide a global confidence rating that none of the lineup members is the culprit. These ratings do not scale match-to-memory for the suspect and are low in diagnostic value. Consequently, the equal-variance signal-detection model predicts that a one-person showup will have better discriminability than a six-person lineup. A largescale experiment (N = 3281) supported that prediction. However, a modified lineup in which participants were asked to follow categorical identification decisions by assigning a confidence rating to each lineup member had better discriminability than both the showup and the standard simultaneous lineup. We call this modified lineup the rule out procedure. Results also revealed a relatively weak confidenceaccuracy relation for global rejections of lineups, but a much stronger confidence-accuracy relation for rejections of individual faces. Past failures to detect suspect innocence with lineups should be attributed to flawed design, not to limitations of visual recognition memory.

No abstract