ABSTRACT
The patterns of differential impairment/sparing of cognitive function subsequent to brain damage can provide strong constraints on theories of the organisation and representation of cognitive systems in the brain. As the articles in this Special Issue of Cognitive Neuropsychology attest, the phenomenon of categoryspecific semantic1 deficit is a compelling case in point. One well-iterated demonstration of the types of constraints imposed on theories of cognitive organisation is provided by recent evaluations of a widely received explanation of category-specific semantic deficits: the sensory/ functional theory (e.g., Warrington & McCarthy, 1983, 1987, 1994; Warrington & Shallice, 1984). This theory assumes that category-specific semantic deficits emerge as a result of damage to a type or modality of information (e.g., visual/perceptual vs. functional/associative) upon which successful recognition/naming of objects from the impaired category differentially depends (i.e., living things vs. nonliving things, respectively). However, a nearly exhaustive review of the literature on category-specific deficits (Capitani, Laiacona, Mahon, & Caramazza, 2003-this issue) establishes the fact that the majority of patients with such impairments do not present with a disproportionate deficit for a type or modality of knowledge. From this fact it can be concluded that the cause of category-specific semantic deficits cannot be damage to a type or modality of information. This conclusion
implies the rejection of the sensory/functional theory as a viable theoretical framework with which toexplain the existence of category-specific semanticdeficits (see also Caramazza & Shelton, 1998). How, then, does one account for the facts of category-specific semantic deficits? Perhaps the most straightforward proposal is that the organisation of conceptual knowledge in the brain is subject to domain-
specific principles (Caramazza & Shelton, 1998). In this article, we assume a domain-specific organisation of conceptual knowledge, and ask two questions: If we assume a domain-specific architecture, (1) what constraints are placed on further theoretical assumptions that might be made regarding the organisation of conceptual knowledge in the brain, and (2) what constraints are placed on assumptions about how conceptual knowledge is represented? Relevant neuropsychological and functional neuroimaging data, as well as alternative theoretical proposals, are recruited in an attempt to clarify these two questions.
