Abstract
Face identification is possibly the most complex, and likewise efficient task achieved by the human visual system. Studies addressing the effects of face familiarity have documented that the presence of facial representations in memory generally enhances processing efficiency. Nevertheless, it remains fairly unknown how visuo-perceptual processing of personally familiar and unfamiliar faces actually differs. The present study sought to identify sources of facial information that are processed differently due to repeated, real-life experience. The underlying hypothesis was that encountering faces across various viewpoints and distances enhances perception of facial information maintained across such changes. Variations in viewing distances affect the resolution of available information (Sinha et al., 2006; Gilad-Gutnick et al. 2012), while maintaining the overall configuration of facial features. Changes in viewpoint, on the other hand, do not affect the arrangement of facial information along the vertical axis (Dakin & Watt, 2009; Goffaux & Dakin, 2010). Four experiments were conducted testing individuals of the same peer group; i.e. all participants viewed identical stimulus sets containing a minimum of 20 personally familiar identities. The results demonstrate that while processing of local information is unaffected by familiarity, it leads to superior discrimination of the spatial relations maintained either across viewing distances (i.e. the overall configuration of internal features), or viewpoint changes (i.e. inter-feature relations along the vertical axis). These findings are compatible with a multi-dimensional face space (MDFS; Valentine, 1991) account of experience-dependent visuo-perceptual processing differences. Within the MDFS increased real-life experience concurs with enhanced diagnosticity of specific dimensions—those invariant to variations occurring during social interactions. This renders the facial representations of frequently encountered, personally familiar individuals more distinct and robust, as compared to those of unfamiliar ones.
Meeting abstract presented at VSS 2014