Familiarity effects in the construction of facial-composite images using modern software systems

Abstract

We investigate the effect of target familiarity on the construction of facial composites, as used by law enforcement to locate criminal suspects. Two popular software construction methods were investigated. Participants were showna target face that was either familiar or unfamiliar to them and constructed a composite of it from memoryusing a typical ‘feature’ system, involving selection of individual facial features, or one of the newer ‘holistic’ types, involving repeated selection and breeding from arrays of whole faces. This study found that composites constructed of a familiar face were named more successfully than composites of an unfamiliar face; also, naming of composites of internal and external features was equivalent for construction of unfamiliar targets, but internal features were better named than the external features for familiar targets. These findings applied to both systems, althoughbenefit emerged for the holistic type due to more accurate construction of internal features and evidence for a whole-face advantage.

Statement of Relevance: This work is of relevance to practitioners who construct facial composites with witnesses to and victims of crime, as well as for software designers to help them improve the effectiveness of their composite systems.

Keywords:

• facial composite
• eyewitness
• evolutionary algorithms
• facial feature
• memory

Notes

1. In Ellis et al. (1979), photographs of familiar complete faces were correctly named (M) at 80%, internal features at 50% and external features at 30%. The probabilities of not correctly naming the face – calculated as (1 − M/100) – are .2, .5 and .7, respectively. Using the product rule, the probability of not correctly naming both internal and external features seen separately is .5 × .7, or .35. Converting this probability (p) into percentage correct – (1 − p) × 100 – gives 65%, but correct naming was much higher when seeing the complete face (80%). This analysis provides evidence for a whole-face (holistic) recognition advantage.

2. We collected additional naming data from participants who did not inspect composites of repeated identities – as repeating such items may elevate naming levels and potentially promote different results. To do this, an additional 60 football fans (people different to those in Section 3.1.3) named eight composites from one of the eight conditions (familiarity × system × composite-type), selected randomly with equal sampling. Overall naming was 22.3% correct for the additional data without repeats, which is obviously only slightly less than the above, 26.5%, with repeated items. We also found the same pattern of effects by system and familiarity.