Gorr and Horgan ~ attempt to analyze the locution 'P's A-ing at t is intentional (unintentional) under the description "A-ing"' in order to understand Davidson's contention that actions are never intentional or unintentional simpliciter but only under a description. Their analysis of intentional action under a description requires that the agent know that the description applies to the action. I believe that their analysis fails for reasons which I hope will show that no analysis can forge such a tight connection between knowledge and intentional action.
Gorr and Horgan offer the folowing analysis:
(I) /~ A-ing at t is intentional under the description 'A-ing' if and only if (i) this event is an act, and (ii) P knows, at t, of this act, that it is a A-ing by him.
Is the knowledge condition necessary? If we adopt a JTB+ analysis of knowledge (justified true belief plus whatever condition handles Gettier-type examples), we can divide this question into questions concerning the necessity of each knowledge condition. Trivially, the truth condition is necessary, for P-s A-ing at t is intentional only if it is an A-ing by P at t. Is the belief condition necessary? Must one believe of one's actions that they are describable in a certain way in order for the action to be intentional under that description? Gorr and Horgan consider actions done by rote such as mechanically shifting into third gear. My intuition is that such an act is intentional even though one does not at the time believe (and thus does not know) that one has performed it. Gorr and Horgan attempt to explain away this intuition by noting that when I 'come to' I will be surprised to find out that I have shifted into third gear. They aver that it is counterintuitive to say that one can be surprised by one's intentional action. Although such cases are relatively rare, there is nothing counterintuitive about them. Suppose a man encounters his ex-wife at a party and begins to make rude remarks to her,