Comparing featural and holistic composite systems with the aid of guided memory techniques

This study compares the effectiveness of two computerised composite construction systems - a holistic, recognition-based system named ID and a featural system that is utilized internationally, namely FACES. The comparison aimed to test whether ID produces better quality composites to FACES, and whether these composites could be improved with the aid of context reinstatement tehcniques, in particular guided memory. Participants (n=64) attended a staged event where they witnessed a female 'numerologist' for 20 minutes. Five weeks later they were asked to return to create a composite of the woman using either FACES or ID. Reconstructions were made in view, from memory after a South African Police interview or from memory after a guided memory interview. In addition, experts for each system constructed composites of each perpetrator. Studies have reported enhanced identification when multiple composites are combined to create a morpho. Hence, the guided memory composites for each perpetrator were morphed to create three ID and three FACES morphs. The complete set of 76 composites was then evaluated by 503 independent judges using matching and rating tasks. The study hypothesised that ID would perform better, but results suggest that the two systems performed equivalently. Results also suggest that the guided memory interview did not have the desired effect of significantly improving participants' memories of the perpetrator, and that contrary to expectations, the morphed composites performed extremely poorly and were rated the worst and identified the least. Related findings and ideas for future research are discussed.

Grantor dc:publisher.institution

Department of Psychology

Year dc:date.issued

2007

Author dc:creator

• Sullivan, Taryn

Advisor dc:contributor.advisor

• Tredoux, Colin

Language dc:language.iso

eng

Handle dc:identifier.uri

http://hdl.handle.net/11427/11642

OAI identifier oai:identifier

oai:open.uct.ac.za:11427/11642