ABSTRACT
The occurrence of semantic deficits in which a patient’s knowledge of objects from two broadly defined semantic categories (typically, biological kinds and man-made artefacts) is subject to contrasting degrees of impairment has been subject to a wide range of theoretical interpretations. One of the most influential of these accounts has proposed that such patterns of neuropsychological performance arise because of differences in the type of information that is most salient for the distributed cognitive representations of concepts in different semantic domains (Warrington & Shallice, 1984). In this formulation, visual attributes are held to be important to the task of distinguishing among exemplars of natural categories, while representations of artefacts are thought to be more heavily dependent on functional knowledge. For instance, according to this hypothesis, different types of fruit or breeds of dog are told apart mainly by virtue of their distinctive colours, shapes, sizes, surface markings, etc., whereas tools (e.g. different types of knives or saws) are distinguished in terms of their usage, and can differ markedly in appearance. That it is possible to produce an apparently category-specific effect, given these representational assumptions, by preferentially damaging a type of knowledge rather than a conceptual category, has been demonstrated convincingly in a connectionist framework (Farah & McClelland, 1991). The account has come to be known as the
sensory-functional theory (SFT) (Caramazza, 1998; Caramazza & Shelton, 1998).
