The Use of Facial Composites in Person Identification

In criminal investigations, eyewitnesses are often required to construct a facial composite of a previously seen perpetrator to help the police narrow down a pool of suspects. Although research settings have previously indicated that composite identification rates are often low, the development of more recent holistic composite systems has substantially improved naming rates. However, considerably less research has explored other means of improving composite identification.

This thesis investigated three strands of research for this purpose. Firstly, it was necessary to determine whether composite construction is a suitable task for eyewitnesses. In Chapter 2, the recognition accuracy of composite constructors and control participants was tested over forensically relevant delays, to determine whether any adverse effects of construction are likely to extend to applied settings. A null effect of composite construction on subsequent line-up performance was found across both a two-day, and ten-week time delay. Following this, ways of improving composite naming rates were investigated. In Chapter 3, the role of individual differences in composite construction was explored, to establish whether some eyewitnesses are better equipped for the task, based on face recognition abilities. In this instance, an individual's performance on two standardised recognition tests did not correspond to the quality of their facial composite. Finally, the role of context in composite identification was investigated. Chapter 4 suggests that composite naming can be increased substantially when contextual information is provided.

These results are discussed within the context of applied settings and directions for future research are provided.