Distinguishing Between Perceptual and Decisional Sources of Holism in Face Processing

Face recognition is thought to rely on holistic processing, where the entire face is processed as a single unitary object. The nature of this holism is often assumed to be perceptual, however recent work applying the General Recognition Theory (GRT) framework suggests that holistic processing of faces may arise from decisional factors. Using Monte Carlo simulations, we examined how known violations of GRT constructs relate to a behavioral measure of holism, the congruency effect, and found that both perceptual and decisional sources of holistic processing can give rise to significant congruency effects. We then explored whether a well-known model of face recognition, one that reproduces the congruency effect, would also reproduce the pattern of GRT results seen in human participants. Surprisingly, like humans, the model showed a consistent decisional locus of holistic processing. This was surprising because there is no explicit decisional component in the model that could give rise to this finding. This suggests that either the model contains some kind of implicit decisional component, or alternatively that there may problems with the techniques used to measure decisional versus perceptual holism from the perspective of GRT.