Accuracy of relational and featural information in facial-composite images

Abstract

Facial-composite images are crucial in helping to bring perpetrators to justice and thereby maintain national and international security. Psychology can assist in identifying the accuracy of such facial information as well as ways to improve
their identification rates. The current experiment demonstrates that composites tend to differ in terms of their accuracy of featural information (individual facial features) and relational information (spacing between those facial features) depending on the type of composite software used. Composites constructed with feature systems tend to have more-accurate feature information, whilst those constructed with holistic systems tend to have more-accurate relational information. Further, it was found that not only the addition of sunglasses on poor-quality composites, but also viewing composites from the side, can increase their identifiability. This could be applied by the police to potentially increase identification and arrest rates of perpetrators.