Javier Gonzalez de Prado Salas

Agents sometimes have rational grounds to doubt their own reasoning abilities. If the reasons for self-doubt are misleading, it may be that such self-doubt makes it rationally required for the agent to refrain from endorsing the conclusion of an actually correct piece of reasoning. Imagine, for instance, that Jane performs a correct mathematical deductive inference, but learns that she has taken a pill that is likely to make her mathematical abilities unreliable, without her being in a position to notice whether this is so. The intuition is that, in cases like this, it may be rational for the agent to refrain from endorsing the conclusion of her reasoning, even if the pill has not been effective and Jane’s abilities were not impaired (Christensen 2010; Lasonen‐Aarnio 2014). This would be so despite the premises of Jane’s reasoning remaining perfectly good, deductive reasons for the conclusion.

Some authors have claimed that these cases involve a conflict among rational epistemic requirements, more specifically a conflict between the requirement to respond to one’s reasons and the requirement to have coherent attitudes (Worsnip 2018; Pryor forthcoming; Coates 2012). My aim is to argue against these views. I propose that rational self-doubt undermines the agent’s ability to competently rely on her premises as reasons for the relevant conclusion. In this way, the agent’s possession of such reasons becomes precarious, since in order for a fact to be available to an agent as a reasons for a conclusion, she needs to be able to treat it competently as such a reason (Lord forthcoming, Sylvan 2015). Thus, an agent with misleading higher-order evidence can remain coherent while properly responding to the set of reasons that she actually possesses – a set that is reduced due to the acquisition of the higher-order dispossessing evidence.