Techniques for understanding and improving the effectiveness of facial composites (PDF)

This e-book comprises a collection of six research papers that focus on understanding and improving the effectiveness of facial composites. Research in this area is important for practitioners, with the aim of helping the police to identify as many offenders as possible using this forensic method of identification. Firstly, Fodarella et al. provide detailed procedures for correctly assessing the effectiveness of composite systems and techniques in the laboratory. Their contribution is based on a gold standard proposed by Frowd et al. (2005) for face construction and naming [Contemporary Composite Techniques: the impact of a forensically-relevant target delay, Legal and Criminological Psychology, Vol. 10, pp. 63-81]. Fodarella et al. describe these procedures for composite systems included in the current collection: E-FIT and PRO-fit 'feature' systems; EFIT-V and EvoFIT 'holistic' systems; and forensic artists' Sketches. An annex is also provided for inclusion of future systems. Next, Marsh et al. consider the impact of presenting a facial description at the time of encoding which is either congruent or incongruent with a target identity. Their finding is that participants' face recall and feature-based construction are less effective following an incongruent auditory description (cf. a congruent description and silence). They also report an interesting effect based on distinctiveness of the target face. Skelton et al. explore feature selection in whole-face and isolated contexts, based on established research (e.g. Tanaka and Farah, 1993) [Parts and wholes in face recognition, Quarterly Journal of Experimental Psychology: Human Experimental Psychology, Vol. 46A, pp. 225-245]. They confirm the long-held belief that more identifiable composites emerge when selection of facial features (eyes, nose, mouth, etc.) is carried out in the context of a complete face (cf. selection of isolated elements). Ness et al. next consider the importance of the angle of view of a face at encoding and at construction. Their results for a feature system indicate an advantage for construction of a three-quarter (cf. front) view composite when a witness has encoded a three-quarter view of a target, but best performance emerges for construction of both front and three-quarter view (cf. front view) composites when a witness has seen either a three-quarter view or all views of a target. Davis et al. replicate the advantage to naming observed by Bruce et al. (2002) [Four heads are better than one: Combining face composites yields improvements in face likeness, Journal of Applied Psychology, Vol. 87, pp. 894-902] when combining a number of individual composites into a 'morphed' image. Davis et al. find that naming increases as the number of constituent images increase for E-FIT and EFIT-V composites (up to 16 composites were involved in the work). They also replicate the advantage for viewing a composite in a vertically-stretched (cf. unstretched) format. Lastly, Frowd et al. analyse a corpus of over 6,000 naming responses collected over the last decade, to obtain a stable estimate for seven factors of interest. Using regression- and meta-analyses, the main results confirm the benefit of the EvoFIT holistic (cf. feature and sketch) system, the holistic component developed for the cognitive interview, and a short (cf. medium and long) retention interval.