In this seminar I will be outlining the main positive argument of my dissertation, which is that our sense of agency–our implicit sense of being in control of some event in our immediate surroundings by performing a certain action–can be a reliable sense for that the action is a cause of that event. The argument uses the conditions under which it is possible to infer the existence of a causal relation that have been identified in the context of causal modeling theories. (Specifically the conditions under which an event is an intervention on an event c with respect to an event e.) It is argued that the sense of agency is a reliable sense for control–given these conditions–if certain broad but substantial conditions hold of the environments in which agents experience this sense. Finally, a naturalistic epistemology is employed to argue that if the sense of agency is in fact a reliable indicator of control, then the agent knows that its action is a cause of an event when this is true and the source of the agent's belief that this is so is its sense of agency.

This result bears on several issues in the epistemology of causation: one is the possibility of having direct, experiential knowledge of causation; another is the epistemic difference between true experiments in science, that involve control, and natural experiments, which involve only passive observations. A third issue concerns the epistemic regress implied by the idea that we can only infer causal relations (because we cannot directly experience causation), and that to infer any causal facts, we must already have some causal information in our premises. (This is a well-known result in for example Nancy Cartwright's work on causal inferences.)