ABSTRACT
Eyewitnesses are often interviewed shortly after a crime to produce a facial composite of the perpetrator. We reviewed the relevant literature to identify one promising interview technique to maximise memory retrieval (Cognitive Interview) and one computer system to facilitate constructing facial composites (Mac-A-Mug Pro System). An experimental study is described in which witnesses observed a target person and then, two days later, attempted to generate a facial composite. The resulting composites were of very low quality (not at all similar to a photograph of the target) and of no value in selecting the target from a photo array. An analysis of the psychological components of facial memory is presented to explain why the Mac-A-Mug system performs so poorly in realistic eyewitness tasks, in which the target face is constructed from memory. Suggestions are provided to improve facial composite systems by making them more compatible with the psychological processes mediating face recall. Finally, we offer some suggestions to improve the ecological validity of experimental research to evaluate facial composite systems.
