The butcher, the baker, the candlestick maker: investigating facial recognition for multiple-perpetrator crimes

Doctoral Thesis


Permanent link to this Item
Journal Title
Link to Journal
Journal ISSN
Volume Title

University of Cape Town

In the United States, 20% of all violent crime is committed by multiple perpetrators. Despite the prevalence of multiple-perpetrator crimes, most published eyewitness research uses a single-perpetrator paradigm: that is, witnesses view a crime committed by a single perpetrator whom they must recognise later. Multiple-perpetrator crimes, however, present with several problems. Police procedure for administering multiple-suspect parades is poorly defined. Furthermore, eyewitnesses must make multiple identifications, and are tasked with a unique memory problem of perpetrator-role assignment. I studied these problems in the following ways: (a) a survey among South African detectives (N = 75) to investigate how multiple suspect parades are administered in practice; (b) two face recognition experiments where the number of face-attribute pairs was manipulated at encoding to investigate the effect of set size on both item recognition (for attributes and faces), and associative memory performance (i.e., matching identity to role; N = 70, and N = 67); (c) an eyewitness experiment where participants studied a simulated crime committed by up to 10 perpetrators whom they had to recognise later (N = 200); and (d) a set of simulations testing a revised version of the Interactive Activation and Competition network proposed by Burton et al. (1990) as a computational account of the memory difficulties experienced by eyewitnesses to multiple-perpetrator crimes. Overall, the results suggest that associative memory is particularly vulnerable to the negative effects of set size, and that role-players in law and psychology should consider the implications of these difficulties in court and the laboratory.