Abstract
When you stare at a face in a crowd, you can easily recognize that person's face. However, it is very difficult to recognize a face that falls in the peripheral visual field when there are other faces surrounding it. There is a great debate about why this crowding effect occurs. It could simply be a function of the eccentricity and complexity of faces and their many parts (interference between the low-level parts or features). Alternatively, many psychophysical, single unit, and functional imaging studies have demonstrated that upright faces, unlike most other objects, are coded holistically in face selective regions of the brain such as the fusiform face area. We might therefore expect crowding of faces to arise because of interference between higher-level holistic representations of these faces. Here we tested this by presenting upright or inverted target faces in a crowd of additional upright or inverted faces. We found that face recognition was selectively impaired only when target faces were presented upright in a crowd of other upright faces. This selective deficit shows that there is a representation of upright faces that is independent from the representation of inverted faces. More importantly, the results demonstrate that crowding can occur between high level representations of objects (the configural aspects), and not just between the low-level features as is often thought. Crowding - interference between similar stimuli - therefore occurs at multiple stages in the visual system.