AI Models as a Variety of Psychological Explanation

The paper attempts to establish a basis upon which it could plausibly be said that knowledge level models typically used in the development of AT systems such as expert systems could have psychological import. Various modelling methodologies are set out, and it is shown that these methodologies cannot supply psychological explanations of expertise on the basis of ordinary realist assumptions about the mind, since the knowledge level primitives cannot supply the right sort of links between tasks and AI methods. In contrast, an anti-realist, interpretative view of mind is set out, and it is shown how AI modelling methodologies could, in that context, be of psychological value. Finally, a short example gives some concrete expression to these ideas.