Witness interviews: does recall of relational information improve identifiability of a facial composite?

Abstract

Facial composites are used by Police to generate lines of enquiry; unfortunately composites made by traditional ‘feature’ systems are not often accurately named. One reason could be that these systems tend to rely on descriptions of the criminal’s facial features, when it has been shown that relationships and distances between facial features—the relational information—is of importance for face recognition. Here, we present two experiments to investigate the usefulness of probing for relational information within witness interviews. Participant-witnesses underwent a typical cognitive interview (CI), an interview in which featural information was probed for before relational information (FR), or an interview in which probing for relational information preceded probing for featural detail (RF). Composites constructed of familiar targets with no delay were recognised better in the former two interviews than the latter, suggesting that relational information interferes with subsequent recall of featural information. However, after a 24 hour delay composites constructed of unfamiliar targets did not differ significantly for naming rates by interview type. This indicates no naming benefit for recalling relational information. However, RF composites were rated as significantly better likenesses to target images after a 24-hour delay, and so future work could explore this further to assess what aspect of the image is improved by recall of relational information.