The evidentialist view is that relations between your non-factive mental states and your beliefs determine what your evidence is and thereby which propositions you justifiably believe. Your beliefs are justified you have sufficient evidence for your beliefs (and you believe for the right reasons). Some evidentialists adhere to a strict ethics of belief, one according to which for any body of evidence there is one and only one attitude (believe, disbelieve, or suspend judgment) you can justifiably take towards the propositions you consider.
What evidence is there for evidentialism? Some evidentialists argue for evidentialism on the grounds that relations between your beliefs and your evidence determine whether your beliefs are rational or reasonable from the epistemic point of view. If justified beliefs are reasonably held beliefs and reasonably held beliefs are justified, it follows that your beliefs are justified if they are supported by sufficient evidence and unjustified otherwise. Others argue for evidentialism on the grounds that alternative views are faced with a version of Moore’s Paradox (Smithies 2011).
Suppose (if only for reductio) that the accessibility thesis (the conjunction of the positive and negative accessibility thesis) is false:
PAT: You have justification to believe p iff you have justification to believe that you have justification to believe p.
NAT: You lack justification to believe p iff you have justification to believe that you lack justification to believe p.
If the accessibility thesis is false, you can have simultaneous justification for certain doxastic attitudes that are not rationally co-tenable. (They are not rationally co-tenable because they give rise to an epistemic version of Moore’s Paradox.) If they are not rationally co-tenable, they cannot be attitudes that you have simultaneous justification for holding. So, the accessibility thesis is not false.
Now, consider the pragmatist view. The pragmatist thinks that you can have sufficient justification to believe p even if you have equally good evidence for p and for ~p—provided that forming the belief that p will further some significant practical end. (In cases where there is sufficient evidence for believing p or believing ~p, the evidentialist and pragmatist views agree that your beliefs are justified only if they conform to the available evidence.) I ask you to consider the view not because it is true (I have nothing but contempt for the view), but because I think it is a contingent matter whether anyone has sufficient evidence to believe it. Raised on a steady diet of pragmatist propaganda, it seems someone just might have sufficient evidence to believe that she has sufficient justification to believe p where she does not also have sufficient evidence that p is true. PAT implies that if our subject permissibly believes that she permissibly believes (say) God exists (where she has no more evidence that God exists than she does for the hypothesis that God does not exist), she permissibly believes that God exists even though she violates the evidentialist’s standards.
Here, now, is my anti-evientialist argument. William has sufficient justification to believe that he permissibly believes that he permissibly believes God exists. William, however, does not have sufficient evidence to believe that God exists. So, according to PAT, it is permissible to believe without sufficient evidence. According to the evidentialist, it is never permissible to believe without sufficient evidence. Thus, the evidentialist view is mistaken.
The evidentialist can respond by rejecting my claim that it is possible for someone to have sufficient evidence to believe the pragmatist view or by rejecting PAT. The first response is completely ad hoc since the evidentialist view is that the evidence you have depends upon psychological facts about you (e.g., what you seem to remember, what you find intuitive, how things look to you, etc.). Surely if by virtue of some contingent psychological facts someone can have sufficient evidence to believe evidentialism, someone can have sufficient evidence to believe pragmatism. Suppose instead that the evidentialist rejects PAT. If they do so, they have to dispense with one of the arguments for evidentialism. Is that all that damaging to the view? Not necessarily. There are other arguments for evidentialism. For example, there is the view that the justified belief just is the rational belief. The thesis S’s belief is rational iff it is justified (R=J) seems to support evidentialism because it seems natural to assume that following the evidence is what the rational person does. It is often what the rational person does, but perhaps the rational person follows the evidence and comes to believe that pragmatism is true. If the subject discovers that she holds beliefs that conform to the pragmatist’s standards, it might seem that her beliefs are rationally retained even if the subject’s beliefs do not conform to the evidentialist’s standards. (Perhaps the subject is non-culpably ignorant of the fact that she does not live up to the evidentialist’s standards.) If so, R=J does not support the evidentialist view. Indeed, R=J would seem to suggest that the evidentialist’s view is mistaken.
Richard (in the comments) asked what I think is _the_ question to ask. I've offered a sketch of an answer. Adequate? Well, if not yet, hopefully soon. Thanks, RC.