AN ARM GOES UP; IT IS A SIGNAL that the assassination is to go forward. The physical process that transpired could equally have been a stretching, a voting, or a taxi-hailing, but this particular process was in fact a signaling. The context in which the arm-rising took place was surely relevant to determining which action it was, for no amount of flailing can constitute a taxi-hailing on a deserted island, or a voting in a monarchy. But let us somewhat improbably suppose that our arm is in a context where its rising could have constituted any of these things. The question is why it was in fact a signaling. And as the physical properties of the arm and its setting do not tell us, the natural place to seek an answer is in the arm-wielding agent's thought. In some way, the agent's own understanding of the event of the arm-rising contributes to determining the further descriptions under which that event constitutes the intentional action that it is.
How is this relation between thought and action to be understood? The broader aim of this paper is to highlight a particular constraint that any theory of action proposing to answer that question must satisfy. There is an architecture present in intentional action that must somehow be reflected in the agent's practical thought, and I will argue that the task of accounting for this structure strongly favors some ways of thinking about moral psychology over others. The motivation for this claim involves a second aspiration that I believe has independent interest for our understanding of mental causation: to identify a phenomenon I will take some poetic license in labeling deviant formal causation. (1) The phenomenon concerns a type of mismatch between practical thought and the resultant action, and merits the label in its resemblance to the problem of deviant causation that plagues the so-called Causal Theory of Action. I will argue that whereas the Causal Theory is thought to be vulnerable in its susceptibility to the problem of deviant causal chains, the "non-causal" neo-Anscombean theory of action that is an increasingly popular alternative faces its own form of deviance--a form of deviance that the Causal Theory is well equipped to handle. The point will be that in accounting for the relation between thought and action, there is more than one kind of way in which the two may deviate. And insofar as the Causal Theory of Action is less vulnerable than the alternatives to the kind of deviant formal causation I will illustrate, this point will amount to a new argument in favor of the Causal Theory.
1. Actions, Causes, Deviance
The Causal Theory of Action (CTA) traces back at least as far as Aristotle, (2) but owes its contemporary heritage to Donald Davidson. (3) It is not the most useful of labels, since there are theories of action opposed to the Davidsonian strand that nonetheless involve appeal to some form of causation. The causal commitment meant to be specific to Causal Theories concerns the pedigree of intentional actions: what distinguishes events that are intentional actions from other kinds of events is the causal antecedent of the event. Intentional actions are those behaviors that are caused to occur by some relevant psychological property of the agent, where the notion of "cause" in use here is efficient causation--"the primary source of the change or rest" of the agent. (4) Particular theories may diverge with respect to precisely what this property is, but we may simply call it "intention." Actions are those things the agent intended to do (or appropriately related in some way to his intention), where intentions are some causally real state of the agent that explains the occurrence of the action.
The CTA holds that being caused by an intention is necessary to transform mere behavior into intentional action, but this condition is not meant to be sufficient. Not just any result with the right cause will count as having been done intentionally; the outcome must also accord in some sense with what the agent had in mind. …