Abstract

We tend to think every ought statement implies that an actual agent can comply. However, our uses of “ought” suggest that some ought statements fail to have this implication: it is possible that the actual agent ought to do something she has no chance of accomplishing even if she intends to do so. Rather they imply that if the agent and her circumstances were defect-free, she could and would perform the prescribed action. There are two types of evaluation for ought statements. One type of evaluation addresses the question of what to do given the agent’s peculiar capacity and condition. The point of this evaluation is giving personalized action guidance, and so recommends only what the actual agent can do under the actual condition. Another type of evaluation addresses a different question, that of what to do as a type of agents. The point of this evaluation is the coordination of individuals by selecting a shared norm for them: the standard that prescribes all of them to perform the same action and classifies for all of them the same traits as defect. This is why it endorses the ought statements that some actual agents cannot comply with, but that a normal agent could and would do so under normal conditions.