Abstract
Two experiments exploring the differential processing of distinctive and typical faces by adults and children are reported. Experiment 1 employed a recognition memory task. On three out of four dimensions of measurement, children of 5 years of age did not show an advantage for distinctive faces, whereas older children and adults did. In Experiment 2, however, subjects of all ages classified typical faces faster than distinctive ones in a face/non-face decision task: the 5-year-olds performed exactly as did adults and older children. The different patterns in performance between these two tasks are discussed in relation to possible cognitive architectures for the way young children represent faces in memory. Specifically, we examine two alternative architectures proposed by Ellis (1992) as precursors for Valentine's (1991a) multidimensional adult face-space and discuss whether implementations of these spaces should be based on a norm-based or an exemplar-based framework.
