Purchase this article with an account.
Rachel Bennetts, Shobonna Akhter, Kayleigh Ziegler, Charlie Frowd; Individual differences in creation of forensic composite faces: A comparison of multiple systems. Journal of Vision 2016;16(12):73. doi: 10.1167/16.12.73.
Download citation file:
© ARVO (1962-2015); The Authors (2016-present)
Lab-based research has found that people vary widely in their ability to recognise faces. However, few studies have examined how these individual differences relate to performance in more ecologically valid face recognition tasks. Eyewitnesses to a crime are sometimes asked to generate an image of the suspect using specialised software systems – these are known as "composite" face images. To date, no research has examined whether the face recognition ability of the witness affects the quality of the composite, or whether certain systems are better suited to individuals with high or low face recognition ability. This study examined composite face creation in 60 individuals with high, medium, and low face recognition ability, as measured by performance on a standard laboratory test of face memory (the CFMT). Each participant watched a video of a single unfamiliar face (randomly selected from a group of 10 faces), and created a composite image of the face 2-4 hours later. Participants were randomly assigned to one of two commonly used composite systems – ProFit, which asks individuals to select and manipulate individual facial features; and EvoFit, which encourages more holistic processing by asking individuals to select faces based on their overall appearance. A second group of participants (n = 51) rated the composites for their likeness to the original faces. There was no overall difference in likeness ratings for high and low face memory composites. However, there was a significant interaction between face memory ability and composite system: participants with high face memory created better composites with the EvoFit system, whereas participants with low face memory created better composites with the ProFit system. The results indicate that variability in face recognition ability can have significant real-world effects. The findings are also in line with previous work suggesting a link between face recognition ability and holistic processing.
Meeting abstract presented at VSS 2016
This PDF is available to Subscribers Only