Abstract
It is well established that faces are processed by configural mechanisms. This configural processing could be the product of a face-specific mechanism or a content-general configural processing mechanism. If they are content-general, prosopagnosics (individuals with impaired face recognition) should show impairments with a wide range of tasks involving configural processing. To determine whether our sample of prosopagnosics have problems with face configural processing, they were tested with a task requiring discrimination between faces that differed in either part configuration or the face parts themselves. The prosopagnosics showed deficits with face configuration discriminations but also showed comparable deficits with face part discriminations. We then assessed non-face configural processing using three independent measures. On a task with houses that was parallel to that used with faces, the prosopagnosics performed normally with house configuration and house part discriminations. When tested with a Navon global-local task, the prosopagnosics showed normal levels of global interference on local response times and local interference on global response times, and they showed a normal sized advantage for global response times compared to local response times. Finally the prosopagnosics scored normally on three tests of visual closure, which have been argued to require configural processing. Taken together, these results suggest that the prosopagnosics have impairments with domain-specific face processing mechanisms. These mechanisms appear to process faces holistically rather than processing facial configuration and facial parts separately.