Suppose that the coarse-grained account, i.e., the Anscombe-Davidson line, of action individuation is correct. On this account, all actions are bodily movements. The description of bodily movements corefer. So, for every action description, each perfect nominal denotes the same action. For example, “Smith’s arm movements,” “Smith’s pump operation,” “Smith’s replenishment of the water supply,” and “Smith’s poisoning the inhabitants” all refer to the same bodily movements. Thus, since all actions are bodily movements, the actions, “Smith is moving his arms,” “Smith is operating the pump,” etc., all refer to the same action.
Goldman and others have raised objections against the coarse-grained account. They have pointed to the fact that if different actions have different properties, then actions are distinct and action descriptions cannot corefer. (We might align fine-grained accounts of action individuation with imperfect nominals for exactly this reason.) Similarly, Goldman and others have argued that the coarse-grained account runs into trouble with regards to temporal order. If indeed all action descriptions corefer, then it seems that Smith has poisoned the inhabitants of the house well before the poison reaches the occupants. This seems awkward; perhaps we could say it sounds counterintuitive.
(Well, that is how it seems to me, Goldman, and a host of others. Someone pointed out to me in conversation recently that this is not so awkward. He mentioned a scene in Shakespeare’s Hamlet where Hamlet announces, “you have killed me” before being stabbed with a poison-tipped sword. In such literary accounts, temporal order is not a problem for coreferential action descriptions. If it’s not a problem for literary accounts, then it might not be considered counterintuitive.)
What if Goldman and others have missed an opportunity to deflate the coarse-grained account because it succumbs to a sort of vagueness problem? Let me explain.
If all actions are bodily movements, and if all descriptions of bodily movements corefer, then all descriptions of action refer to the same action. There may be a way of showing how my different bodily movements, say “Smith reads the newspaper in the morning” and “Smith falls asleep to the radio in the evening,” corefer. If they corefer, then there’s no way of distinguishing between an agent’s actions. All of the action descriptions we can generate about an agent’s actions from birth to death refer to the same action.
Since the agent is the common denominator of all the actions, every action the agent performs becomes virtually indistinguishable. We can just think of the agent slipping from one action to the next without there being much of a difference between the two actions. If we cannot distinguish between the agent’s actions, then his entire life is one big action.
Two problems seem to persist in an account like this. First, it seems to me that it is fallacious. That we cannot distinguish between the agent’s actions in shorter time intervals does not mean we cannot distinguish between the agent’s actions in larger time-folds. We should be able to distinguish between “Mary’s walking as a one year old” and “Mary’s walking as an eighty year old.” If we can distinguish between these two action descriptions, then surely we can distinguish between shorter time slices of Mary’s actions.
Second, time is not irrelevant on Davidson’s coarse-grained account. Since it seems that we would have to amend each of our actions with the locution “at time t1″ (whatever t1 stands for), we could distinguish between action descriptions using the time locution. Actions that occur at such-and-such a time will have a set of descriptions that refer to it. When some action description captures something outside of that time-slice, it is not identical to the action description that designates something that occurs within that time-slice.