Goldman on Level Generation and Act Token

Abstract

Suppose Smith forgets about the pain in his toe. Has he performed an action? One might think that he has, for he can be advised or urged to forget about his pain and he can try to do so and fail or succeed at it. However, neither of these is a sufficient condition for something's being an action. Being brave is not an action, but one can be advised or urged to be brave. Nor is understanding, but one can try to understand. I take it that it is a necessary though not sufficient condition for something's being an action that the verb designating the action take the progressive. It is not a sufficient condition, for sleeping is not an action. But 'sleep' can take the progressive. It is not odd to say 'Jones is sleeping'. In contrast Smith's forgetting the pain in his toe is not an action. Were he asked what he is doing, it would be odd for him to say 'I am forgetting about the pain in my toe'. However, as I shall show, on A. I. Goldman's theory of human action, forgetting turns out to be an action.' The central purpose in Goldman's valuable book is to tell us what a human action is. He begins by arguing that what he designates as the Anscombe-Davidson identity thesis is incorrect (pp. I-2). Suppose S flips the switch, turns on the lights, and illuminates the room. According to the identity thesis, only one action has been performed, but it has been described in different ways. Goldman argues that S has performed not one action, but three. Further he contends that if three different acts have been performed in this case, then there must be some relationship among them. Goldman claims that S illuminates the room by turning on the lights, and turns on the lights by flipping the switch. As an analysis of this relationship, Goldman introduces the technical notion of 'level-generation'. S's flipping the switch generates S's illuminating the room. Level-generation plays a role in Goldman's definition of 'act-token'. If A is level-generated by an act token, then A is an act-token. First, I show that on Goldman's analysis of 'act-token', and 'levelgeneration', S's forgetting about the pain in his toe is level-generated by an act-token and so, contrary to what I take to be facts, Goldman is committed to its being an act token. Second, I revise Goldman's definition of 'level-generation' in a way which circumvents my criticism. It might seem incompatible with forgetting not being an action that one can be advised or urged to forget, one can try to do so, and one can fail or succeed at it. Lastly, then, I shall offer an explanation for these facts which is compatible with my contention that forgetting is not an action. Goldman recursively defines 'act-token' as follows: