There are strong reasons to want actions to figure as causes and effects. We speak of our actions as having effects in the world; and we think that some of our actions, at any rate, are caused by our own wants and beliefs, or perhaps by our acts of will. But the most systematic account of action that has been given recently makes actions turn out to be entities of a sort that could not figure in the causal series. I refer to Alvin Goldman's account in A Theory of Human Action (Englewood Cliffs, New Jersey, 1970). In this esaay, I will support this claim in some detail. The shortcomings of Goldman's act-tokens (the analysantia of actions) suggest that the difficulty could be circumvented by adopting Donald Davidson's analysis of action, 1 but this turns out to be no solution because some actions (pre-theoretically so-called) lack the spatiotemporal qualities necessary, on the one hand, for Davidson's analysis to work and, on the other, for them to be causes or effects. Finally, I propose a way to understand talk which seems to place these problematic actions in the causal series. The proposal leaves much of Davidson's position intact, but it makes a special place for the problematic actions: ones that are not quite events, though they are connected in a perfectly regular way with genuine events.
I can make my objection to Goldman clear only if I summarize the main points of his account of action in A Theory of Human Action. He analyzes action in terms of act-tokens and act-trees. An act-token is at least the exemplification by an agent at a time of an act-property, i.e. a property such that some exemplifications of it are actions. An act-tree is a tree-structure whose nodes are act-tokens, and which displays four kinds of relations (as well as their converses) that may hold between act-tokens: being a temporal part of, being a temporal sequence of, being on the same level as, and level generating. The first two relations are not special to his theory. In a particular case of my exemplifying, from t to t + 1 hour, the property making a pie, for instance, my exemplification, from t + 30 minutes to t + 45 minutes, of