ABSTRACT
In recent years, support has grown for the idea that the human memory system consists of separate, functionally specialised subsystems. Fractionation of this sort is indicated by selective impairments of memory function in neuropsychological patients and complementary dissociations in normal individuals (see e.g. Shallice, 1988). Theoretical accounts of these findings typically use an information processing framework and are expressed in the form of verbal descriptions or relatively crude flow diagrams. We refer to these accounts as “conceptual models”. This type of theorising usually applies at a general level, and is particularly useful where it would be premature to attempt a more detailed account. However, it is much less useful where an understanding of underlying mechanisms is required. The lack of commitment to unnecessary detail in typical conceptual models allows them to integrate large bodies of different types of data, but can make them difficult to apply in specific circumstances. For example, it can be difficult to generate precise predictions with confidence. One way round this problem, when circumstances permit, is to devise and explore explicit computational models.
