Humeans hold that actions are movements of an agent's body that are suitably caused by a desire that things be a certain way and a belief on the agent's behalf that something she can just do, namely perform a movement of her body of the kind to be explained, has some suitable chance of making things that way (Davidson 1963). Movements of the body that are caused in some other way are not actions, but are rather things that merely happen to agents.

Actions can, of course, be explained in other ways. Perhaps every action can be explained by neural activity, or by goings on at the sub-atomic level, and presumably many actions can be explained by the states of the world that make the beliefs that figure in Humean explanations true: That is, the states that make those beliefs knowledge. But Humeans insist that belief-desire explanations are distinctive because their availability is what makes our bodily movements into actions (Davidson 1971a). A belief-desire explanation of a bodily movement is thus, as we might put it, a constitutive explanation of an action (Smith 1998). Other explanations of actions may be available, but they are all non-constitutive: their availability is not what makes our bodily movements into actions.

We can represent the Humean's view as in figure 1.

Humeans may seem to hold that the constitutive explanation of an action has four basic elements: two psychological (a desire for an end and a means-end belief), one non-psychological (a bodily movement), and a relation that holds between them (a causal relation of the right kind).