Abstract

Forensic face matching refers to the comparison of pairs of faces for identification purposes, and is ubiquitous in applied contexts such as passport control. Despite its widespread use, a remarkable number of errors arise in this task even under optimised conditions. In this review, we outline the problem of face matching within the wider context of passport control. We then proceed to review factors that influence accuracy by constraining data quantity within stimuli, through changes in pose, illumination, and image quality, for example. This is followed by a review of factors that influence resource limits within individuals to perform this task, encompassing individual differences and sources of bias.