Abstract
Prosopagnosia is an impairment at recognizing faces following brain damage whose study offers invaluable information to understand the neural and functional aspects of normal face processing. It has been claimed that a deficit in integrating facial features, i.e. holistic/configural processing, underlies prosopagnosia (e.g. Levine & Calvanio, 1989; Sergent & Signoret, 1992; Barton et al., 2002), but the exact nature of this impairment remains unclear. To clarify this issue, we tested the patient PS, a pure case of acquired prosopagnosia following a dominant lesion in the right inferior occipital cortex sparing the right fusiform face area (‘rFFA’, Rossion et al., 2003). First, we show that PS can categorize a stimulus as a face even when it requires holistic processing. Specifically, her detection of faces in 2-tones Mooney figures or in Arcimboldo's paintings (faces defined only by the organization of non-face features) is accurate and fast. Second, a whole brain analysis of fMRI data during a face detection task with Mooney face stimuli indicates that holistic face detection is subtended primarily by the patient's ‘rFFA’, [q(false discovery Rate)[[lt]]0.05; 34, −50, −14] as in normal controls, rather than by low-level visual areas. These results indicate that the prosopagnosic patient PS is able to integrate facial and non-facial features into a global/holistic generic face representation. This preserved ability contrasts with the patient's impaired ability to integrate facial features into a global individual face representation (Ramon & Rossion, 2007), suggesting a functional dissociation between two different forms of holistic face processing. The first one allows the basic-level categorization of the stimulus and is common for faces and objects, being deficient in patients suffering from both prosopagnosia and integrative visual agnosia. The second process is necessary for individual encoding of faces only and can be selectively impaired following brain damage.