Abstract
While recognition of familiar faces is extremely good, the recognition of unfamiliar faces is prone to many errors. These differences could be accounted for by either perceptual or conceptual knowledge that we have about familiar, but not unfamiliar faces. Since both factors are confounded in familiar faces, the sole contribution of perceptual or conceptual information can only be examined by using either of these factors during face familiarization. In a set of experiments, participants were asked to learn unfamiliar faces while rating them based on either their perceptual, physical features or based on their inferred personality traits. Findings show that learning faces using conceptual judgments yielded much better recognition than learning faces based on perceptual judgments. This effect was found both when perceptual questions included holistic facial aspects (e.g. is the face symmetric?) or when they asked about specific facial features (e.g., how thick are the eyebrows?). Interestingly, conceptually learned faces were more resilient to changes in viewpoint and illumination than perceptually learned faces. Furthermore, recognition of low-resolution, degraded images of the learned faces was better following face learning with conceptual than perceptual judgments. Finally, we tested whether conceptual judgments are mediated via previous experience, using the same design with other-race faces. Interestingly, no advantage for conceptually than perceptually learned faces was found for other-race faces, suggesting the important role of previous experience in associating semantic information to faces. These findings suggest that associating semantic meaning to newly learned faces, rather then focusing on their perceptual aspects, enhances their perceptual representation and improves face recognition. These findings suggest that conceptual information is likely to play an important role in our superior recognition of familiar faces.
Meeting abstract presented at VSS 2016