The Impact of Exposure Duration and Target Prevalence on the Confidence-Accuracy Relationship in Unfamiliar Face Matching

Abstract

Unfamiliar face matching is the process of determining whether different faces of individuals belong to the same or different person, and is crucial in a number of contexts including criminal identification, age identification and border security. A wealth of research however, suggests that it is highly error prone. Unfamiliar face matching is found to be reliant on different processes compared to familiar face matching, which contrastingly is performed quite well. As a result, it is important to enhance understanding of the underlying processes involved in order to determine valuable estimates of performance, and explore key factors which influence these. In line with previous research, this study aims to examine the effects of exposure duration and target prevalence which are vital factors in applied face matching settings. Eighty participants completed a one-to-one face matching task manipulating these variables. Results are analysed and discussed in terms of general face matching performance, as well as a confidence-accuracy calibration approach which is applied to overall, match and mismatch decisions. Exploratory signal detection analysis is also conducted to examine the effects further. The results suggest that neither exposure duration nor target prevalence have an impact on the confidence-accuracy relationship. However, a strong positive-negative asymmetry is observed for decision type, suggesting that confidence would not be a reliable indicator of accuracy for mismatch trials. Further research is recommended in regards to both manipulations, with the suggestion that a higher number of mismatch trials should be used. The implications of these findings are discussed.