An application of caricature: How to improve the recognition of facial composites

Abstract

Facial caricatures exaggerate the distinctive features of a face and may elevate the recognition of a familiar face. We investigate whether the recognition of facial composites, or pictures of criminal faces, could be similarly enhanced. In this study, participants first estimated the degree of caricature necessary to make composites most identifiable. Contrary to expectation, an anticaricature was found to be best, presumably as this tended to reduce the appearance of errors. In support of this explanation, more positive caricature estimates were assigned to morphed composites: representations that tend to contain less overall error. In addition, anticaricaturing reduced identification for morphed composites but enhanced identification for individual composites. Although such improvements were too small to be of value to law enforcement, a sizeable naming benefit was observed when presenting a range of caricature states, which appeared to capitalize on individual differences in the internal representation of familiar faces.

Acknowledgements

The work was funded by a grant from the Engineering and Physical Sciences Research Council, EP/C522893/1. Partial funding was also received from the EPSRC as part of a public understanding of science grant, Future Face, GR/R86034/01. The authors would like to thank the London Science Museum for allowing us to collect data, the journal reviewers who made many insightful comments and suggestions, as well as Yvonne Plenderleith, Clare Thomson-Bogner, Anna Law, and Steve Fields for their assistance with data collection.

Notes

¹There is an unfortunate inconsistency in the literature over the term “facial composite”, which was used by Young, Hellawell, and Hay (1987) to describe a novel face produced from the top and bottom halves of two different photographs. In the forensic literature, and that used here, it refers to a face image of a suspect, typically made up of many component parts but also including sketches (e.g., ACPO(S), 2000; Davies et al., 2000; Frowd et al., 2005a; Frowd et al., 2005b; McQuiston-Surrett, Topp & Malpass, 2006; Shepherd & Ellis, 1996); note also that the term “composite” was in forensic use (e.g., by Davies, Milne, & Shepherd, 1983) prior to Young et al. (1987).

²“Morph” is another ambiguous term in the literature, meaning a change in form, but it has also come to refer to an average of two or more images, typically faces, since each is morphed to the average shape. We have adopted the term “morphed composite” (e.g., Bruce et al., 2002).

³PRO-morph is a software component of the PRO-fit facial composite system marketed by ABM in the UK. PRO-morph was used here for both caricaturing (all experiments) and averaging the individual composites to produce morphed composites (Experiment 1B).

⁴An analysis of incorrect names generated by participants was considered, since these data may provide an indication of guessing; also, composites with lower incorrect names can limit a waste of police time. However, in all the experiments presented here, the data suggested that caricaturing exerted only a weak influence on incorrect name production. For simplicity, these data are omitted.

⁵An analysis by target naming was conducted to check for the presence of a sampling bias. In spite of randomly assigning participants to testing booklets, chance differences in the naming of the target photographs may still occur, and this in itself may lead to differences in the naming of the composites. Overall, target naming was very high, at over 98% correct at each caricature level, and did not differ significantly, F(2, 24) = 0.04, p>.1, thus underscoring the absence of a sampling bias. This analysis was also conducted elsewhere in this paper; target naming was found to be similarly high and did not differ significantly by composite level, p>.1.

⁶Note that these results should not be taken to indicate that EvoFIT is in general better than sketch, since the composites are not directly comparable—good sketch artists typically do rather well (e.g., Frowd et al., 2005a).