Forensic Expertise in Facial Image Comparison

Deciding whether two images of unfamiliar faces are the same person or two different people is a difficult task, but one in which forensic facial examiners generally outperform untrained observers, although not with perfect accuracy. Here, the ways in which they perform face matching were compared with forensically-trained and non-expert controls whilst eye movements were recorded. In Experiments 1 and 2, examiners were the most accurate group. In Experiment 3, rating the features prior to the same or different matching decision improved the controls' performance which reduced the examiners' accuracy advantage. Across the experiments, all groups showed similar patterns of responses to the face pairs and similar attention to the features, including a bias towards faces on the left of the screen. The higher overall accuracy of examiners was not accounted for by differences in viewing times, or by a more conservative response to feature rating. Further, examining the performance of individual examiners showed how group accuracy was driven by some high performers, although the same examiners were not consistently the most accurate in all experiments. Overall, this study did not find any differences in the way professionals viewed faces which might explain their high performance as a group. However, as the adoption of a feature comparison strategy improved accuracy for both control groups, this suggests high accuracy for facial experts may be due to their methodological approach to face matching rather than any qualitative differences in their viewing behaviours.