Abstract
The theoretical notions of holistic/configural processing, and of face-space, have each been very successful. Authors typically refer to holistic/configural processing when explaining why upright faces are discriminated better than inverted faces, or faces better than objects, and when explaining findings from associated paradigms (e.g., composite effect, part-whole effect). Authors typically refer to face-space when explaining why distinctive faces are discriminated better than typical faces, own-race faces better than other-race faces, or why adaptation aftereffects occur. In this theoretical presentation, I argue that (a) although each theory has been independently successful, both in fact purport to explain exactly the same thing - the coding of facial identity - and so, as a field, we must consider the relationship between them, (b) our current approach of simply picking the most convenient theory in a given paper is not sustainable in the long term, and (c) the problem of the relationship between holistic processing and face-space is not, as many of us might have assumed, intractable. To illustrate how progress might be achieved I propose three theories, and sketch potential or actual empirical studies relevant to testing them. Theory 1 is that either holistic processing or face-space is not, in fact, related to face identification (e.g., holistic processing subserves face ‘detection’). Theory 2 is that both are related to identification, and make independent contributions (e.g., timecourses following stimulus onset are different; they derive from different cortical regions in fMRI; multiple regression for individual differences in face recognition shows independent contributions of holistic processing strength and face-space coding ability). Theory 3 is that holistic processing and face-space coding are the same thing, predicting tighly interlinked empirical findings (e.g., strength of holistic processing differs for typical and distinctive faces; adaptation aftereffects are in some way “special” for upright faces compared to inverted faces and objects).
Supported by Australian Research Council grants DP0450636 and DP0984558.