ABSTRACT
Theories of semantic memory tend between two poles. At one extreme is a view often associated with the connectionist enterprise: that the semantic system is a unitary, homogeneous mass, without functional or neuroanatomic specialisation, that capitalises on statistical regularities in the environment in learning about and processing semantic information. On this account, double dissociations of semantic memory are explained in terms of the processing mechanisms characteristic of neural networks, the statistical structure of the environment, and various psycholinguistic factors such as familiarity, frequency, and visual complexity. At the other extreme is a view positing that semantic memory is parcelled functionally and neuroanatomically into a set of discrete processing modules, each tied to a particular modality and/or semantic domain. Under this hypothesis, double dissociations of semantic memory arise from damage to one or another of these modules, or the connections between them.
