ABSTRACT
Computational accounts of the mind are often framed as accounts that treat the mind as an “information processer”; yet, as many writers recognize, it is not always clear just what this means. One way of understanding information processing is rooted in the idea that cognitive activity involves the manipulation and exploitation of inner representational states and structures. Yet again, although there is considerable agreement among cognitive scientists and philosophers of mind about the explanatory need for internal representations in theories of cognitive processes, 1 there has been less clarity and agreement about the nature of the representations. In fact, there have been disagreements about their form, disagreements about the way in which representations acquire their content, disagreements about their explanatory role, and, perhaps most importantly, disagreements about the manner in which they function as representations. The perspective I want to illuminate in this chapter involves a particular type of representation that has a long history in computational accounts and that is increasingly viewed as a promising way to make sense of representation in the brain. Commonly referred to as “S-representation” (see below for more on this label), this type of representation is based upon some form of structural similarity between a representational system on the one hand, and the item, process, condition, or relations represented on the other hand. In large measure, the notion is motivated by our use of models, maps, and simulations in our own problem-solving activities. Just as we use external models and maps to acquire information, make predictions, and navigate through the world, so too, investigators claim that the computational brain uses its own neural models and maps to perform various cognitive tasks.
