Matching the Faces of Robbers Captured on Video

Recent research has shown that unfamiliar face matching from both high- and low-quality closed circuit television video images to photographs is highly prone to error, even when viewpoint and expression are matched as closely as possible. The current experiments made use of a filmed, staged reconstruction of a bank raid that was captured on CCTV and on high-quality broadcasting video. Experiment 1 tested the ability of members of the public to match actors captured on CCTV to photo-spreads containing similar-looking distractors. Further experiments, each testing different groups of subjects, investigated matching ability using both high-quality photographs (Experiment 2) and broadcast-quality video material (Experiment 3). Experiment 3 also investigated the effect of disguising hairstyle, and varied whether or not the target was present in the photo line-up. The results of these experiments confirm those of previous work, that matching the identity of unfamiliar faces is highly fallible, even when high-quality footage is used. Experiments 4 and 5 tested matching ability using two-alternative forced-choice and single-item verification tasks. Performance remained highly error-prone even with the simplest question asked. The legal implications of the results are discussed.