Abstract
Individual faces are notoriously difficult to discriminate and recognize when they are presented upside-down, an effect that supposedly reflects the disruption of holistic/global processing. Since acquired prosopagnosia has been associated with a deficit in holistic face processing, a reduced/abolished effect of face inversion is expected in such patients. However, previous studies have provided mixed results, showing not only a decreased (e.g. Marotta et al., 2002), or abolished FIE (e.g. Delvenne et al., 2004), but also a better performance for inverted stimuli in some cases (e.g. Farah et al., 1995). These discrepant results may be due to the selection of patients with general object recognition problems, and may also be confounded by associated upper visual field defects.
To clarify this issue, we conducted an extensive behavioral investigation of upright and inverted face processing in two cases of selective acquired prosopagnosia: PS (Rossion et al., 2003) and GG. First, there was no effect of inversion for the two patients at the Benton Face Matching Test, contrary to normal controls. Second, PS and GG both showed a significantly decreased effect of inversion in a simultaneous ABX face matching task across viewpoint changes. Third, pictures of faces and cars were presented at two orientations, in an ABX simultaneous matching task. While both patients showed small effects of inversion for cars in the normal range, they both showed significantly decreased inversion effects for faces. Finally, PS had to identify personally familiar faces and did not show any advantage for the upright orientation, in contrast with two controls familiar with the faces.
Overall, these results indicate that the differential processing of upright and inverted faces is affected following prosopagnosia, reinforcing the view that a key aspect of the impairment in acquired prosopagnosia lies in the inability to form an individual holistic facial representation.