Suppose that there is reason for Ronnie to go to the party because there will be dancing there and Ronnie loves to dance. As Daniel Fogal points out in his “Reasons, Reason, and Context,” ‘reason’ here is used as a mass noun, to refer to the action having what I’ll call a specific valence that has quantity to it. (Thus there is some/much reason for Ronnie to go to the party. We might also say that Ronnie has reason — some or much — to go to the party, where having-reason language does not indicate, as it does in some other contexts, that the specific valence is partially explained by Ronnie believing that there will be dancing as against the fact of such). Distinguish specific valence from overall valence or the valence that an option has in virtue of all of the respects of specific valence that bear (e.g., perhaps Ronnie also ought to go to the party because the sum of any competing specific valences here fail to outweigh).

Normative reasons (count noun here, not mass noun) explain there being reason (for some agent). Put another way: the reason-status of an item is that of explaining or helping explain, in a certain way, there being reason (for some agent). Fogal persuasively argues that reasons-talk and causes-talk are the same in that they pick out pragmatically/communicatively salient representatives of clusters of conditions that constitute full explanations or causes. The bits singled out stand in for the whole thing. And because of all of the variation in the sorts of things that it makes good pragmatic/communicative sense to single out as reasons, there is no special metaphysical explanatory role for reasons themselves in explaining reason. Though there are different explanatory roles in the overall explanation of there being reason for X to A (e.g., Ronnie loving dancing and the fact that there will be dancing play different explanatory roles, the former playing what is sometimes called the “source” role), no such roles neatly map onto this as against that item being cited as being a or the reason for Ronnie to go to the party. 

So we might say that the fact that there will be dancing at the party is a reason for Ronnie to go to the party or that the fact that there will be dancing at the party explains why Ronnie has reason to go to the party (maybe part of the assumed pragmatic/communicative context here is that we know that Ronnie loves to dance). But, similarly, in some pragmatic/communicative contexts, the fact that Ronnie loves to dance is the relevant explainer of why Ronnie has reason to go to the party. For these reasons, we should reject Mark Schroeder’s claim that desires are not reasons but rather mere background conditions for reasons. Any distinction between reasons and background conditions (or enabling conditions) essentially involves pragmatic/communicative selection from among the elements of the total explanation of why there is reason for Ronnie to go to the party. And that reflects us, not the metaphysics of it, especially as such categorization crosses over the lines between distinct metaphysical, explanatory roles.

***

There was never much of any reason to think of normative reasons themselves as normatively more fundamental than specific normative valence itself. I never thought this, for years having held a less-sophisticated version of the metaphysics of reasons/reason implied by the mass/count noun distinction in reason/reasons language that Fogal stresses. Perhaps it is because these distinctions have not always been clearly recognized that many philosophers have spoken as if normative reasons (the normative-reason-status of items) were normatively fundamental. We need to stop such sloppiness. And, if we do, then a “reasons-first” program of constitutive normative/evaluative explanation makes little sense. The relevant constitutive-explanatory program in the ballpark would be a “having-reason first” (or “there-being-reason first”) program, according to which specific normative valence is the sole normatively/evaluatively fundamental feature (whether or not there is any constitutive explanation of specific valence by clusters of conditions, any one of which might be pragmatically/communicatively salient and hence a or the reason; after all, specific valence could be explanatorily primitive). But also, if we are concerned with different explanatory elements in the explanation of specific valence, that is a separate issue from what we do or should cite as the or a reason why Ronnie has reason. Fogal’s paper brings this point into particularly sharp relief.

What might a “having-reason-first” program of constitutive normative/evaluative explanation look like? This is something that Fogal’s paper does not address. That is curious because, putting on my somewhat-confused reasons-first theorist hat, I would retreat to claiming that this — or some confused version of this — is what I meant all along. Thanks for helping me get clearer, Fogal!

That constitutive-explanatory program might look something like this: necessarily, specific normative valence is related to overall normative valence via an “additive” or “contributory” explanatory relationship. Hence, the right explanation of ought-features (and other overall valence features) is always via specific having-reason features, specifically via these, pro and con, being weighed or added together in a “contributory” style.

Here are two problems with this view. The first one denies the universal validity of this pattern of explanation for all overall valences (and all ought-features). The at-least-possible cases I have in mind here are those in which a specific valence entails, all on its own, corresponding overall valence. In such cases, necessarily, there is only one specific valence relevant to deciding between a set of options (perhaps a binary set of options, as the plausible cases might suggest). In such a case, there would be no “contributory” explanatory link between specific and overall valence. Certain normative relationships of requirement, correctness, appropriateness, fittingness might work this way (e.g., maybe fitting-attitude standards, if and when they are normative, work this way).

The second problem denies not the reality but the importance of the contributory sort of explanatory link. On this sort of view, what is important, metaphysically, is that a specific valence (again, perhaps a kind of requirement, correctness, appropriateness, etc.), though in principle only one of many factors and subject to being overruled, in conjunction with other factors commonly present, entails the corresponding overall valence. For example, most moral requirements usually entail oughts, but do not if more-important requirements (or other super-important factors) override. On this sort of picture, though some contributory explanatory link is present, it is simply of the right minimal sort for the entailment to hold (given that enabling conditions have been met). What matters is the entailment, not any weighing up of distinct respects of specific valence.

Combining these two objections into one, we might say that specific normative valence includes both merely-having-reason and something like requirement — a specific normative valence that either necessarily or circumstantially (and defeasibly) entails the corresponding ought-feature. (There might also be specific valence that, conjoined with relevant conditions that often or generally obtain, entail that one has sufficient reason to do the thing. But I’ll leave such cases aside for now.)

However, supposing these objections hold, the following not-insubstantial elements of the program remain. 

First, overall valence is always explained, one way or another, in terms of specific valence (and hence in terms of any explanatory basis for this and, by proxy, by what is appropriately cited, on pragmatic/communicative grounds, as a reason). 

Second, the generic quantitative language of having reason (and there being reason) applies to all specific and overall valence (even if using such language is not always conversationally appropriate, given pragmatic/communicative implicature). Whenever there is some normative valence attaching to A-ing for X, it is true that X has reason to A (or there is reason for X to A). 

Third, the explanatory linkage between specific valence (and therefore its explanatory basis and therefore any reason taken as representative of this basis) and overall valence is (a) centrally explanatorily relevant in the evidently “contributory” cases (e.g., additively explained ought-features), (b) centrally explanatorily relevant in certain hard or unusual cases that are not evidently “contributory” (e.g., describing what is going on in the proper balancing of competing requirements) and (c) relevant to good theorizing whenever more than one specific valence is potentially in play in determining the overall valence, providing formal-level description and explanation.