Usually we think of memory as an image of a past event surveyed at leisure in the "mind's eye," or as the fragments of the preceding day that are reworked in dream hallucination; but memory is also active when we recognize objects and retrieve the words to describe them, in the bonds and habits that pattern our everyday experience, the unconscious processes that structure behavior, and in movements like dancing or typing, even walking -- indeed, in all actions and automatic mechanisms. Memory is idea or representation, but it is also the process of remembering and the way this process shapes actions, percepts, feelings and thoughts.
With such diversity it is not surprising that the goal of many studies of memory and its disorders has been mainly that of classification and analysis. As a result, memory has been fractionated into separate forms, each localized to a store in which information about preceding events is deposited. The various types of memory assume a location within the network of the store(s) in relation to anatomical region or stage in information processing. Of course, the idea that a memory function is localized to a store does not imply that the store itself is localized. A store could be woven into other components, but the idea does entail some degree of functional encapsulation or the store would be indistinct from the components it innervates.
There are consequences of this point of view for a theory of mind, apart from the problem of memory. For example, a store that updates or matches information received from sites of registration implies that recognition is nonperceptual, or a function that is secondarily recruited to perception, and obligates a theory of perception that is little more than an account of sensory primitives. A similar explanation of the selection and insertion of words into slots in a sentence frame supposes word-equivalent configurations (labels) in a lexical store. Attaching a mental lexicon to a language device, however, obscures the fact that word finding is a constructive process intrinsic to language no less than are the semantic and syntactic "rules" that guide word finding along.
Perhaps because of difficulties with a faculty account of memory, another version has appeared, duplicating the structure of memory for each mental component, a component consisting of a sensorimotor system or some knowledge base. For example, there might be a separate motor memory, a memory for