The development of “face-space” in infancy

Abstract

Valentine (1991) proposed that faces are encoded in a multidimensional “face-space”. Here, we apply this framework to the development of face processing in infancy. It is thought that faces are encoded as “regions” in the face-space, such that variations in the face input are recognized as the same face identity. In a set of studies with 4-month-olds, 7-month-olds, and adults, we addressed the question of how these “identity regions” develop. The findings indicated that they are initially broad and become narrowed with development. Our results suggest that the face-space model can usefully be applied to face processing in infancy.