Abstract
Face space is a powerful framework that can explain various face processing phenomena in humans, including poor recognition memory for unfamiliar faces and other-race effects, however, it has yet to be applied to other species. We examined whether the face space framework can aid our understanding of face processing in the chimpanzee, a species that shares many of the same cognitive specializations for face processing as humans. Five chimpanzees discriminated all combinations of 20 female faces (380 dyads repeated 10 times), including a 20-face population average, in a matching-to-sample task. Multidimensional scaling was then used to generate a two-dimensional plot representing discrimination performance (% correct). Typicality ratings for the 20 faces were also obtained from human primate experts. From the MDS plot, we measured the length of each identity vector with regard to the origin, the vector angle between each face in the 190 dyads, and calculated the mean typicality ratings. As predicted by the face space model, the average face was positioned centrally in the MDS plot, having the second shortest vector. Human experts' typicality ratings were significantly correlated with vector lengths, where the average face was rated most typical. Chimpanzees' performance was significantly correlated with both the typicality ratings and vector lengths, such that distinctive faces were discriminated better than typical faces. Notably, the worst performance was for the average face. Finally, performance also correlated with vector angle. Faces separated by the shorter angles, suggesting similar diagnostic features, were discriminated more poorly than faces separated by large angles. An analysis of the faces suggested that the dimensions could be described as lower face width and overall head size. These data suggest that face processing in chimpanzees fits within a face space framework, and supports evolutionary continuity in face processing in Hominoids.