Abstract

Rule-based explanations are accounts of phenomena that cite or appeal to rules . Such explanations of human action are ubiquitous in philosophy and in everyday life. The endeavor I call "computational" cognitive science presupposes a type of rule-based explanation of human behavior that appeals to rules like those that comprise computer programs. This dissertation furnishes an analysis of the concepts of rule and rule following assumed in the two varieties of explanation in order to determine whether they are conceptually compatible. I argue that they are not, unless the rules that govern changes in computer states can be shown to give rise to the full normative character of everyday rule following behavior on the part of a suitably programmed machine. ;According to the analysis in the first part of the dissertation following an everyday rule involves an agent that interprets some linguistic token as a rule, i.e. as a normative standard for his behavior, and decides whether to act in accordance with it. If he succeeds in acting in accordance with it, he has followed the rule. The second part of the dissertation examines the mode of action of rules in computers, where at base there is no agent and no decision whether to conform to rules as normative standards. I argue that such machine behavior does not amount to rule following, only to acting in accordance with rules. The prospects for simulating everyday rule following behavior on a programmable machine are negatively assessed, as is the possibility of producing such a machine with mental states like those that characterize human beings. Doubt is thus cast on the usefulness of rule-based explanations that appeal to the kinds of rules that govern present-day programmable computers in accounting for essential aspects of human mentality