Some facts are reasons for us to Φ:

We can Φ in the light of, or motivated by, such facts. When we do so, we are in the good case. We Φ because p.

    As we have seen, we have the idea of reasoning from a premise, which may be known, or merely believed… and which may be true or false. But as well as the idea of reasoning from a premise, we also have the idea of being guided by a fact. (Hyman, 2011: 355-356)

Both good cases and bad cases involve an in the light of relation. 

In the good case, this holds between S's Φ-ing and their knowing. In the bad case, between S's Φ-ing and their belief.

Disjunctivism. In the good case, the in the light of relation is distinctive: it requires the agent to know that p.

Standard argument appeals to the Reasons-Knowledge Thesis: S Φs because p only if S knows that p.

A direct inference from the Reasons-Knowledge Thesis to Disjunctivism is a non-sequitur.

Sketch of a better argument: Φ-ing in the light of one's knowledge involves a dependency of one's Φ-ing, relative to the reasoning which results in one's Φ-ing, on one's knowing. The only way to cash this out appeals to the idea that the episode of deliberation requires an in the light of relation that entails that one knows.

My research over the course of the fellowship will be directed at establishing a connection between Φ-ing rationally and having an ability to engage in a certain kind of mental agency: conscious inference.

There is an empirical literature concerning agency and delusion which will be relied upon.

This research will be conducted against the background of the disjunctive approach to rationality outlined.