Abstract
The human face transmits a wealth of nonverbal signals that readily provide crucial information for social interactions. The brain, as a decoder, flexibly filters the information arriving at the sensory inputs to rapidly achieve complex perceptual categorizations, such as people identification and expression. Response classification techniques have revealed that the face system uses differential effective information to achieve face identification (Bubbles - Gosselin & Schyns, 2001) and expression (Smith et al., 2005). Yet, a fundamental question remains unresolved (Calder and Young, 2005): does facial information used for face identity and expression tap into a unique facial representation system, or does this processing occur in dedicated systems? To address this question we tested PS, a pure case of acquired prosopagnosia with lesions sparing the neural substrates dedicated to expression (STS, amygdala). In marked contrast to normal observers, PS does not use the eyes to identify faces but the mouth (Caldara et al., 2005), highlighting defective representations for identity. However, PS' expression categorization is extremely effective compared to face identification. Consequently, the PS case represents an exceptional opportunity to test whether the face system relies on a unique representation system or can instead flexibly adapt to face expression categorization by using dedicated representations. Patient PS and normal observers categorized neutral, happy and fearful faces by expressions while using Bubbles. Strikingly, patient PS consistently used only the mouth to categorize these expressions, even for fear in which the eyes are highly diagnostic for normal observers. Importantly, eye movement patterns showed that PS spontaneously attempted to extract information from the eye region, indicating a general, selective deficit impinging on the representations for the eye region. Critically, these results clearly demonstrate that a unique representation system is used to code face features, and support a single, flexible (in normal observers) coding system for face representations.