Optimising EvoFIT by Reducing the Number of Faces Shown during Composite Construction

Abstract

The creation of EvoFIT facial composite images enables perpetrators of crime to be identified and subsequently detained. It is therefore important to ensure that the composite construction procedure is optimised to create the most recognisable images possible. During the creation of a facial composite, eyewitnesses view and compare many images of facial shapes and textures to select those which best resemble the perpetrator. However, viewing many images may overwhelm witness working memory, resulting in cognitive overload and resultant memory capacity and decisionmaking deficits. Yet, there is currently no literature exploring the impact of cognitive load during composite construction. This thesis aims to bridge this gap in the literature by investigating the impact of cognitive load during EvoFIT construction and to investigate the importance of face Shape and Texture for composite construction to further optimise the construction procedure.

In five experiments, composite images created using different population sizes to manipulate the cognitive load during the construction procedure were assessed for likeness through composite naming and likeness ratings. The results demonstrated that reducing cognitive load during the construction procedure by decreasing the population size and, therefore, displaying fewer face images to participants, was beneficial for composite likeness. Moreover, reducing the population size for selection of the face shape was particularly important, indicating that face Shape plays a more important role than face Texture in the creation of a recognisable EvoFIT facial composite. Overall, this thesis demonstrates the benefits of reducing cognitive load during EvoFIT composite construction, particularly for selection of the face shape, and develops a theoretically informed construction procedure to increase the number of criminals identified through EvoFIT facial composites.