One might think that it being the case that X rationally ought to A (exhibit some action or attitude in response to relevant items) is a straight-forward function of the reasons that X has for/against A-ing. And one might think that the sort of normative reason that is relevant here is of the subjective or psychological sort (specifically a broadly cognitive mental state — e.g., a belief or a perceptual experience). One supporting thought for this last thought is that normativity is, at its core, the direct guidance of responses via rational causal tendencies in the mind of the agent (we might call this the “direct rational guidance” view of normativity).
However, quite plausibly, the rational ‘ought’ is, in some respects, a function of appropriate response to facts, not just mental states.
One way to put this is that the ‘ought’ of rationality is a function not only of subjective normative reasons, but also of objective normative reasons. I’m concerned here with the particular type of fact-responsiveness in rational evaluation exhibited in cases like this: Michael is irrationally thinking that his phone is lost and is irrationally frantically searching for it in the utility room, refrigerator, etc., his phone planted firmly in his hand. What makes me irrational in such a case is that I’m culpably ignorant of the fact that the phone is in my hand (because I’m failing to notice this fact, obviously relevant to the rational responses I am exhibiting). What I ought to do is not believe that my phone is lost and not run around frantically looking for it, yet, relative to my present mental states and the inferences that I can and should draw from them, this is not what I ought to do. So it looks like any ‘ought’ of rationality that is purely relative to the present mental states of the agent is a specialized, philosophers’ sense of ‘ought’ — not the one we use in coming to a verdict that I’m being irrational in the “lost” phone case. (Though we might include the absence of the belief that my phone is in my hand among the relevant mental states in attempting to preserve the purely mental-state-relative view of the rational ‘ought’, this will not do as this belief-absence does not hook up with the rules of and psychological tendencies for good inference in the right way.)
If, in these cases, the rational ‘ought’ is a function of both factual (objective) and psychological (subjective) items that do or would function as something like premises in inference broadly construed, we should want to know what that mix of different sorts of reasons is and why. There is an explanatory question to answer. (Extant theorists might also face the following puzzle: though, in some contexts, it seems that the distinction between mental-state-relative and fact-relative normativity or reasons is fundamental, if rational evaluation is a function of both of these things yet is also equally or more fundamental, then we need to rethink how the priority of explanation goes in this respect, perhaps “demoting” the subjective/objective normativity distinction. Relatedly, there might also be some theory-relative worries about whether there is something funny about “adding” or “weighing together” subjective and objective reasons for/against A-ing to get some single, unified evaluation of whether one ought to A.)
Here is my answer to the explanatory question (ignoring the more theory-relative worries mentioned). Start with this slogan: the ‘ought’ of rationality is most fundamentally a concept of appropriate responsibility or accountability, not a concept of immediate rational guidance from present mental states. More specifically, the fact that I have my phone in my hand is something that I, and you and anyone else, would have warrant (and hence, in this specific way, reason) to hold me accountable for knowing (and for drawing appropriate practical and cognitive inferences from this knowledge). This is what renders the fact that my phone is in my hand relevant to the weighing of reasons that constitutes my overall rational evaluation here. Notice that, formally, the “rendering relevant” or “selection” function here allows reasons to hold X to account to A to be relevant to the weighing of reasons for/against A-ing without themselves being part of that weighing/calculation (which would be silly as my reasons for/against doing something other than A-ing clearly are not what directly determines whether I ought to A or not).
(I think a similar sort of accountability-based proposal and “rendering relevant” or “selection” function could help in the debate about whether moral requirements to A are (i) a function of the agent’s reasons for/against A-ing or (ii) a function of reasons (of the agent or of anyone else in the relevant norm-enforcement group) to require of the agent that she A. In particular, it might be that the latter, accountability-related reasons select among potentially relevant reasons that the agent has for/against her A-ing, in this way helping determine whether X is required or obligated to A — where ‘required’ and ‘obligated’ are most directly a function of the agent’s reasons for/against A-ing and hence a type of ‘ought’ or ‘has most reason’. In this way, we would not have to choose between reasons for/against holding accountable to A and the agent’s reasons for/against A-ing in determining which reasons are broadly — directly or indirectly — relevant to determining whether we are morally required to A (and that constitute our being required to A).)
(Years ago, John Gibbons got me interested in these issues and cases in epistemic and practical rationality. Then he wrote a book about them (and some other things). The thoughts above are prompted by Errol Lord’s review of John’s book. Regarding reasons and requirements, see Matt Bedke’s relevant paper and especially Justin Snedegar recent synoptic paper.)