In the past 30 years researchers have examined the impact of heightened stress on the fidelity of eyewitness memory. Meta-analyses were conducted on 27 independent tests of the effects of heightened stress on eyewitness identification of the perpetrator or target person and separately on 36 tests of eyewitness recall of details associated with the crime. There was considerable support for the hypothesis that high levels of stress negatively impact both types of eyewitness memory. Meta-analytic Z-scores, whether unweighted or weighted by sample size, ranged from −5.40 to −6.44 (high stress condition-low stress condition). The overall effect sizes were −.31 for both proportion of correct identifications and accuracy of eyewitness recall. Effect sizes were notably larger for target-present than for target-absent lineups, for eyewitness identification studies than for face recognition studies and for eyewitness studies employing a staged crime than for eyewitness studies employing other means to induce stress.

Inasmuch as a completely satisfactory estimate of effect size for the eyewitness accuracy-confidence relation does not exist, we conducted a meta-analysis of 35 staged-event studies. Estimated r = .25 (d = .52), with a 95% confidence interval of .08 to .42. Sampling error accounted for 52% of the variation in r, leaving room for measurement error and possibly moderator variables to account for the remaining variation. Further analysis identified duration of target face exposure as a moderator variable, providing support for Deffenbacher's (1980) optimality hypothesis. When corrected for the attenuating effect of sampling error in the accuracy-confidence correlations, the correlation of exposure duration and the accuracy-confidence correlation was .51: Longer exposures allowed for greater predictability of accuracy from confidence. Even though correction for unreliability in the confidence measure produces a higher estimate of the population correlation of accuracy and confidence, .34, one must be cautious in assessing the utility of confidence for predicting accuracy in actual cases.

In deciding the trustworthiness of eyewitness testimony, the U.S. judiciary employs as one of five criteria the witness' level of confidence demonstrated at the confrontation. A very recent laboratory study has shown that juror perceptions of witness confidence account for 50% of the variance in juror judgments as to witness accuracy. This strong faith in the adequacy of certainty as a predictor of accuracy is not at all supported by the present review of 43 separate assessments of the accuracy/confidence relation in eye-and earwitnesses. Statistical support is provided for the notion that the predictability of accuracy from overtly expressed confidence varies directly with the degree of optimality of information-processing conditions during encoding of the witnessed event, memory storage, and testing of the witness' memory. Low optimal conditions, those mitigating against the likelihood of highly reliable testimony, typically result in a zero correlation of confidence and accuracy. Using the arbitrary criterion of 70% or greater accuracy to define high optimal conditions, seven forensically relevant laboratory studies are identified, with six of them exhibiting significant positive correlations of confidence and accuracy. It is concluded, however, that no really clear criteria currently exist for distinguishing post hoc high from low optimal witnessing conditions in any particular real-life situation. Hence the judiciary should cease their reliance on witness confidence as an index of witness accuracy.

The other-race effect was examined in a series of experiments and simulations that looked at the relationships among observer ratings of typicality, familiarity, attractiveness, memorability, and the performance variables of d' and criterion. Experiment 1 replicated the other-race effect with our Caucasian and Japanese stimuli for both Caucasian and Asian observers. In Experiment 2, we collected ratings from Caucasian observers on the faces used in the recognition task. A Varimax-rotated principal components analysis on the rating and performance data for the Caucasian faces replicated Vokey and Read's (1992) finding that typicality is composed of two orthogonal components, dissociable via their independent relationships to: (1) attractiveness and familiarity ratings and (2) memorability ratings. For Japanese faces, however, we found that typicality was related only to memorability. Where performance measures were concerned, two additional principal components dominated by criterion and by d' emerged for Caucasian faces. For the Japanese faces, however, the performance measures of d' and criterion merged into a single component that represented a second component of typicality, one orthogonal to the memorability-dominated component. A measure of face representation quality extracted from an autoassociative neural network trained with a majority of Caucasian faces and a minority of Japanese faces was incorporated into the principal components analysis. For both Caucasian and Japanese faces, the neural network measure related both to memorability ratings and to human accuracy measures. Combined, the human data and simulation results indicate that the memorability component oftypicality may be related to small, local, distinctive features, whereas the attractiveness/familiarity component may be more related to the global, shape-based properties of the face.For many years, it has been suspectedthat faces of one's own race are recognized more accurately than faces of other races (Feingold, 1914). Indeed, there is abundant empirical evidence for this other-race phenomenon, as two recent metaanalyses of the face recognition literature attest (Bothwell, Brigham, & Malpass, 1989;Shapiro & Penrod, 1986). In addition to the empirical support for this phenomenon, the other-race effect is widely known outside of the laboratory. Deffenbacher and Loftus (1982), Thanks are due June Chance and AI Goldstein for providing the Caucasian and Japanese faces used in the present experiments and simulations and to James C. Bartlett for helpful comments throughout the entire course of the project. We would like to thank also James I. Chumbley, John W. Shepherd, and John R. Vokey for very helpful comments on a previous version of the manuscript. We are grateful also to Barbara Edwards and to Ray Zhu for assistance in testing subjects. Requests for reprints should be sent to A. J. O'Toole, School of Human Development, GR4.1, University of Texas at Dallas, Richardson, TX 75083-0688 (e-mail: otoole@utdallas.edu). -Accepted by previous editor...

Three experiments examined the possibility that eyewitness identifications may be biased because persons may be much better able to recognize a face than to recall where they saw it. In Experiment 1, 14 college-student subjects were asked both to recognize SO facial photographs seen 2 days before and to recall in which of two distinctive rooms they had been seen. Strong rec.ognition and minimal recall were found, Experiments 2 (64 college-student subjects) and 3 (146 college-student subjects) modeled more closely the usual criminal identification situation with mugshot and lineup sessions occurring after the initial encounter with the suspects. Subjects in Experiment 2 were aware they would need to remember the subjects' faces; in Experiment 3, they were not aware of this need. Both experiments provided evidence of considerable confusion in mugshot and lineup identifications as well as a lack of correlation between eyewitness accuracy and confidence. In addition, there were strong mugshot-induced biases in Experiment 3 that could have a bearing on questions of legal procedure and the admissibility of evidence.Recent research suggests that visual recognition of scenes and faces can be strikingly accurate, at least under optimal conditions. Using self-paced presentations, Shepard's (1967) subjects viewed 612 pictures of "things" and "scenes" in a directed-memory task. Recognition accuracy on an immediate test of 68 "old-new" pairs was about 98%. With a one-week delay, accuracy was about 90%. In similar tasks but with controlled presentation times, Standing, Conezio, and Haber (1970) and Standing (1973) reported immediate recognition of about 90% accuracy on old-new pairs even when several thousand pictures of scenes were presented These experiments were supported by the Law Enforcement Assistance Administration Criminal Justice Program Development Grant 72-CD-99-002/73-CD-99-002. Grant administrator Vincent Webb was most helpful. The assistance of David Barrows, Jeffrey Selzle, and Gordon Becker is also gratefully acknowledged, that of Barrows and Selzle for their collection of the data for Experiment 1 and that of Becker for valuable discussion in the design of Experiment 3. Portions of the data for Experiments 1 and 2 were presented at the meeting of the Psychonomic Society in Denver, November 1975.Requests for reprints should be sent to Evan Brown,