Abstract
What role does holistic processing play in face recognition? Konar et al. (Psych Sci 2010; VSS2010) showed that the standard composite-face-effect (CFE) was not correlated with face identification. Gauthier et al. (VSS2010) suggested, however, that face identification may be related to other measures of holistic processing. Here we examine this suggestion directly by determining the relationships among several commonly used measures of holistic processing and face recognition. Eighty participants completed 4 tasks: CFE, the whole-part-effect (WPE), the face-inversion-effect (FIE), and the Cambridge Face Memory Test (CFMT; Duchaine & Nakayama, 2006). Participants were tested on versions of the WPE task that used masked stimuli (no hair, ears, or necks) or with stimuli embedded in a generic face outline (including hair, ears, and neck). We found significant CFEs and FIEs. The WPE based on internal features of faces was not significant, whereas the WPE based on whole faces was, suggesting that the WPE may not be driven exclusively by internal (relevant) features. Analyses of correlations among purportedly holistic/configural indices revealed that none of the effects (CFE, WPE, or FIE) were related significantly with one another, with the CFMT, or with upright face identification accuracy. The CFMT and upright face identification were correlated. These results suggest that holistic processing is not a unified concept: different tasks reflect different aspects of processing, with some tasks influenced by irrelevant features outside the face. Finally, although there may be a role for holistic processing in face identification of some special populations (Konar VSS2009), none of these standard holistic measures predicts performance for typical young observers on face identification using standard perception- or memory-based tasks. These results challenge the notion that holistic processing is a unified concept and ubiquitous for face perception.
NSERC and the Canada Research Chair program.