| Textbooks cover findings of "face-cells" in monkeys, of a "face area" in humans and of numerous behavioral phenomena that seem to be unique to faces. My thesis offers empirical evidence which cannot be accommodated by the popular account of face recognition as mediated by a domain-specific module. The roles of two factors, the level of categorization at which an object is recognized and the expertise of a perceiver, are argued to underlie many of the dissociations that are obtained at the behavioral, neuropsychological and neurophysiological levels when experiments compare faces to non-face objects.; The first chapter of my thesis introduces the general approach which I took to investigate the hypothesis that faces are "special" at both the behavioral and neurophysiological levels. In Chapters 2 and 3, the role of categorization level in determining the apparent specificity of the neural substrates of face recognition is demonstrated using functional magnetic resonance imaging (fMRI). Chapter 4 highlights the role of categorization level in the neuropsychological domain by assessing the extent to which it can explain the performance of prosopagnosic patients. In Chapter 5, I present a 'developmental' study of the acquisition of perceptual expertise with non-face objects, as demonstrated by an increasing reliance on configural information. Chapter 6 presents a longitudinal fMRI study which demonstrates that the development of expertise can recruit the "face area". Finally, Chapter 7 offers a more exploratory analysis of the changes produced by perceptual expertise throughout the ventral visual object recognition pathway.; These results suggest that the level of object categorization is an important factor which coarsely determines the specialization in the ventral temporal lobe: Subordinate-level categorization engages the "face area" as well as the surrounding cortex, as found in normal subjects using fMRI, and can be impaired by lesions in this area, as suggested by findings with prosopagnosic patients. Furthermore, expertise with subordinate-level recognition of homogeneous object categories can lead to further specialization in this area, and result in the recruitment of an individual's "face area" for expert categorization of non-face objects. Finally, these neural changes may be mediated by qualitative changes in object representations, as suggested by experts' greater reliance on configural information. It is concluded that the cognitive and neuronal aspects of face recognition are better understood by the interaction of task constraints with subject experience than by the activation of a conceptual module. |
