Parts and Wholes in Expression Recognition

Abstract

Holistic face identification has been inferred from composite face effects (Young, Hellawell, & Hay, 1987) and effects of configuration on part recognition (Tanaka & Sengco, 1997). Dissociations between face identification and expression recognition suggest that the latter process is more part-based than holistic. If it is, composite and configuration effects should not be found for expression recognition. Three reaction time (RT) experiments are reported. Composite effects disconfirmed neither holistic nor part-based recognition. A null configuration effect disconfirmed holistic recognition, and indicated that expressions are recognised as descriptions of parts and part relations. But a face inversion effect and a face context effect also showed that these descriptions are recognised holistically, in the special sense that they are encoded with a description of the face as a basic level object, together, as an undecomposed whole.