Purposes, intentions and rules

We have the stage and the actor but not the missing alternative to causal explanation. My favoured candidate is the notion of rationality and I shall maintain that the rational man is both a free agent and a proper subject for science. But other claimants will be considered and will contribute to the conclusion. Human action is purposive, intentional and subject to rules. Each of these ideas has been offered as the key to the social sciences and I shall borrow an element from each. Purposive action is explained by reference to its goal, which need not be reached but affects the way in which the agent adjusts to experience. His intentions are bound up with criteria of sameness and difference for possible actions and are crucial for deciding what he chooses to do and what he prefers it to. The rules he follows belong to an external fabric of roles and interpretations, which partly determine the significance of his actions. Not seeing the key in the notions of purpose, intention and rule, however, I shall treat them eclectically, for the sake of a fuller account of rational explanation. This is no mark of disregard for scholars who have taken a stand where I merely pass through and readers who are unconvinced by the line proposed here will have their own ways of reaching an active conception of man.

To set the task, let us recall briefly why we rejected causal explanations of autonomous actions. Although the correct analysis of Cause is fiercely disputed, we insisted on lines involving the idea of law. If lighting the touchpaper is to cause the rocket to fire, then there is at least a law-like connection between events (or facts) of the two sorts. Whether the connection has to do with necessity a priori or de re or with no necessity at all, at least it holds universally in the same conditions and explanation consists, to that extent, in assigning the particulars to their general class. Hence derives authority to infer subjunctive conditionals. Autonomous actions, while doubtless having partial determinants of this causal sort, are not done because they instance a causal law.