Abstract

Holistic facial-composite systems assist eyewitnesses in constructing accurate facial images of unfamiliar offenders through repeated selection from a population of face exemplars. While a larger population of faces offer a potentially broader choice, they may impose excessive cognitive demands when a witness is evolving a visual likeness, thereby impairing the effectiveness of the resulting composite. Over the course of three experiments, the population size was varied using the EvoFIT holistic system during the evolution of facial shape and facial texture. Guided by Cognitive Load Theory, lowering extraneous load by reducing the population size was expected to enhance composite accuracy, although evolutionary theory would suggest that reducing population size too much is likely to limit the emerging likeness. Experiment 1 demonstrated that successively reducing population size (from 72 to 54 to 36 faces) successively enhanced accuracy, as measured by correct naming and likeness ratings of the emerging composites. Experiment 2 demonstrated that this advantage applied independently to both facial shape and texture. Experiment 3 revealed that an additional reduction in population size (from 36 to 18 faces) improved composite naming accuracy further. These findings highlight the potential importance of minimising cognitive load during face construction and its positive impact on forensic practice.