(NT) An agent ought to believe that p only if p is true.
Here's a passage from Andrei Buleandra's paper in Dialectica:
One problem with the above formulation of the norm of truth is that it's not easy to see how it can regulate one's thinking. An agent does not have a direct access to the truth, so, most of the time, he is unable to follow the norm. Shah attempts to solve this problem by using the distinction between objective and subjective norms. He argues that the norm of truth is an objective norm the acceptance of which activates in us truth-sensitive dispositions which constitute our acceptance of the subjective norm of evidence (Shah 2003, 481, fn. 41). The subjective norm is: one ought to believe that for which one has evidence. Shah writes:
The role of the objective norm, whose acceptance is expressed in the phenomenon of transparency, is to provide a standard of success for subjective norms of good evidence that an agent can directly apply to his deliberation. It thus is not a brute fact that subjective norms for rational belief must be evidential in character. Rather, this constraint falls out of transparency, which is just an agent's recognition of the authority of the objective norm of truth (Shah 2003, 471).
Thus, according to Shah, the fact that an agent applies the norm of truth to his doxastic deliberation explains the requirement of evidence for belief. The agent ought to believe that proposition for which he has evidence. Evidence points to truth and therefore, in searching for the truth, one has to be sensitive only to evidential factors. Only evidential factors can be reasons for belief.
First comment. There's something strange right at the start, "An agent does not have a direct access to the truth, so, most of the time, he is unable to follow the norm." Hmmm... Can't really "follow" a norm that doesn't tell you what should or may be believed, but you can easily avoid violating it by believing nothing. Even if you're just bent on getting in on the believing game, what justifies the move from "can't access the conditions that determine whether you manage to avoid violating TN directly" to "most of the time unable to follow". We can supply the missing premise, which is that any norm you cannot determine directly whether you've violated is one you most of the time won't follow, but that premise doesn't look very good.
Second comment. I'm not sure what objective and subjective norms are. Suppose the idea is that "ought" is ambiguous and what determines a norm's status as objective or subjective depends upon how "ought" is interpreted. Then the problem is that when the subject asks, "What should I believe?", only one of these norms (at most) will be relevant to that question. Suppose instead that "ought" is not ambiguous, rather, there are many obligations that you're under and some obligations have to do with objective matters (e.g., not believing the false) and some obligations have to do with subjective matters (e.g., bringing belief in line with the evidence). Now the objective and subjective norms can both address the question the subject has in mind when she asks, "What should I believe?", but it seems we can generate epistemic dilemmas:
(i) Necessarily, one ought to believe that for which one has (sufficient) evidence.
(ii) Possibly, one has (sufficient) evidence for believing p even if ~p.
(iii) Necessarily, one ought to believe p only if p is true.
There's no dilemma if you deny that you can have sufficient evidence for believing p even if ~p, but then it seems that we avoid dilemmas by means of infallibilism and skepticism.
There's no dilemma if you deny that "ought" is understood as anything but a prima facie obligation. But how are we supposed to work that out? Not in a Rossian way, I take it. Won't (iii) always trump (i)? Won't the view really amount to the claim that you ought to believe on sufficient evidence--when you're right to do so?
I've been looking through Shah's 2003 Phil Review paper and I think he avoids the mistake by not making it. He doesn't assert (i).
My concerns thus far are picky little concerns about trying to generate an obligation to believe falsehoods from a prohibition against believing falsehoods. There's this interesting passage that I wanted to comment on.
Steglich-Petersen (2006) argues that Shah's explanation of doxastic transparency in terms of following the norm of truth for belief is flawed because it is impossible for an agent to break the norm of truth while engaged in doxastic deliberation. Since a norm or a prescription is by definition something that can be broken it follows that the norm of truth is not actually a norm; doxastic deliberation is not regulated by any norm. Steglich-Petersen writes:
According to Shah and Velleman's explanation it is a necessary condition for an instance of deliberation to count as deliberation about belief that the deliberation exhibits transparency. This means that if transparency is produced by the norm of belief, this norm motivates one necessarily and inescapably to act in accordance with it. The transparency is immediate and does not involve an intermediary question about whether to conform to the norm of belief; the norm is thus unlike the norms governing promising. It is thus doubtful whether a consideration which necessitates motivation should be considered a normative consideration at all. This argument does not undermine the claim that deliberation about belief necessarily exhibits transparency, but only challenges the thesis that a norm or normative considerations can explain the necessity (Steglich-Petersen 2006, 507).
An argument, "it is impossible for an agent to break the norm of truth while engaged in doxastic deliberation. Since a norm or a prescription is by definition something that can be broken it follows that the norm of truth is not actually a norm"
So, I think it's possible to break the norm of truth while engaged in doxastic deliberation--you do so whenever doxastic deliberation results in a false belief.
Objection: breaking requires awareness.
Reply: I don't think that's right about the verb, but suppose it is. Consider the norm, "I oughtn't act without sufficient reason". Can that be broken? Maybe someone deeply irrational can break it. Same with the truth norm. Can someone rational break it in awareness? No. Same with the truth norm. So, while I agree that the truth norm is unlike the promising norm, it seems very similar to the "You should never act without sufficient reason" norm. Which is a norm.