It is popular, in some perhaps many circles of analytic philosophy, to analyze various normative and evaluative properties in terms of normative reasons. There are, I think, two essential elements to this strategy. First, it seeks to explain the candidate normative or evaluative property (this thing being valuable or good for or valuable to one or that thing being what one ought to do) in terms of this putatively more basic normative valence of relevant actions or responses (reasons, an agent X having reason to A). Second, this valence is pro tanto in that it does not, in itself, render a verdict on whether the action or response that it attaches to is to be done (in this way, it is partial, non-determinative, non-verdictive, non-ought-making). Thus the explanatory work of reasons (X having reason to A) is done via reasons “weighing” (or having “weight”) for and against the action or response in question, one side ultimately having more “weight” (except when the two sides are equally-balanced).
(The normative valence of an action A of agent X maps onto X having reason to perform or exhibit A. There is an essential relationship here to reasons themselves and we can think of these as propositions that, in virtue of either being believed or true or perhaps rational, do essential work in explaining why the pro tanto normative valence attaches to A for X. I am, then, abstracting from any essential explanatory role for reasons in an agent having reason to perform or exhibit A. I’m also abstracting from any other elements in the explanation of the pro tanto action or response valence — e.g., desired or valued outcomes as on subjective or psychological or Humean approaches to reasons, outcomes considered very broadly as the “consequentializing move” in normative description or explanation would have it.)
This second thing worries me. It is plausible that some normative valences that have this sort of role in explaining what we ought to do (or how we ought to respond) are such that, in a broad range of contexts, they are not merely pro tanto (but rather determinative or verdictive or regulative or ought-making on their own). In other words, it might be that X being required to A is sometimes as basic as X having reason to A.
To be sure, it is possible to stipulate that talk of X having reason to A is to cover X being required to A as well (e.g., ‘X has decisive reason to A’). And this is all good and well insofar as it allows us to pick apart the contexts in which normative requirements fail to produce oughts as they normally do (as well as the standard cases in which, after all of the various relevant pro tanto valences bearing on X’s A-ing have been weighed up pro and con, the agent ought to A as against not-A-ing). There might also be special mixed cases in which requirements and reasons work together to determine what the agent ought to do. So this expansive framing of reasons-talk has some potentially good chops for getting at important elements of explanatory detail. However, the downside, it seems to me, is potentially obscuring the difference between what we might call “deontological” as against “weighting” accounts of ought-making. For example, if the more-abstract framing of X having reason to A is in mind, there is nothing obviously wrong with the common proposal that what one ought to do is identical to (and explained by) what one has most reason to do. But there is a tendency, in thinking of things this way, to assume that none of the reasons are requirements — so that the ought-property gets explained by various reasons weighing for and against the response or action and the resultant overall weight coming down on one side or the other.
My conclusion: since it is plausible that (non-pro-tanto or normally verdictive or ought-making) requirements play a basic normative-explanatory role similar to that of (merely pro tanto) reasons, we should be suspicious of — or at least very careful with — the program of explaining the (other) normative and evaluative things in terms of reasons (even on the broad construal of ‘reasons’ and perhaps ‘weighing’ or ‘weighting’ indicated).
(TLDR – One way of avoiding the potential confusion or bias here is distinguishing two kinds of reasons or ways of X having reason to A, one of the “weighting” or “weighing for or against” sort and the other of the “requiring” sort. Gert does this, as does Dancy. I don’t see anything wrong with this kind of approach, as long as one is making the distinction in terms of regulative function — not just two different, mysterious “flavors” of reason — and as long as, in light of these functions, one has a way of accounting for how these two different elements do their ought-making in both pure cases and in cases in which the two sorts of reason compete in the ought-making.)
(TLDR – These thoughts are prompted by reading Justin Snedegar’s “Reasons, Oughts, and Requirements.” Just starting the second, close reading of this paper now. The broader context is that considering the moral requirement to A as a function of either or both of (i) reasons, perhaps requirement-type reasons (Joshua Gert), for one to A and (ii) reasons for everyone to require that one A (Matthew Bedke via a thread in J.S. Mill). What I say above should be read as tentatively rejecting Bedke’s approach in favor of a “mixed” analysis of at least moral requirement in terms of both reasons of the agent for/against the action and reasons of those putatively wronged and observers to be resentful, have outrage, fight back, punish, etc. The even broader context/interest is that of how consent or related (or unrelated) processes might make otherwise adverse actions, such as coercion and including the coercive imposition of law by the state, normatively permissible. In order to answer this question well, I think I need to get a bit more settled on what I think normative, moral permissions — and requirements — are.)
Policy of Truth 
I’m just getting around to commenting on this. Didn’t have time while I was away, and have been too jet-lagged or busy since I got back. We’ve batted these issues around for years, so I’m not sure I’m going to say anything new, but maybe I’ll say it a new way.
In a sentence: I find the idea of a reason for action based on a requirement (say, a teleological one) perfectly intelligible, but find the idea of a pro tanto reason for action that isn’t, or isn’t based on, a requirement puzzling to the point of unintelligibility.
The distinction between a prima facie reason and a pro tanto reason is supposed to be one between an epistemic status and an ontological one. Very roughly, a prima facie reason is prima facie for S because S doesn’t know the rest of the normative story: it’s a reason for S insofar as he knows the normative story about what to do, relativized to his ignorance. A pro tanto reason, by contrast, is pro tanto for S because it has some non-determinative, non-conclusive normative weight for S. It normatively inclines S in the direction of action, but isn’t an all-in consideration to push her into action. But it has this status out of any relation to a requirement that is an all-things-considered reason to act a certain way. It’s as though a pro tanto reason just has semi-conclusive normative weight, full stop, with no further account of why it has the weight it has.
I have no problem with a prima facie reason, but pro tanto reasons completely abstracted from all-in requirements to act strike me as hard to understand. I don’t understand the rationale for the concept, and am not sure I even fully understand the idea itself. The puzzle is how a consideration can have some normative weight without having any ontological connection to a source of normativity that provides all-in reasons to act.
(1) Imagine that Smith has five pro tanto reasons, A, B, C, D, and E. It can’t be the case (can it?), that A-E jointly give Smith all-in reason to take some action, including the action that consists of the sequential performance of A, B, C, D, and E (or of any action consisting of those five elements, regardless of order, e.g., EDCBA, or EABCD, or whatever). If that were the case, the idea of a pro tanto reason wouldn’t do much work. There would instead be some second-order requirement that explained why Smith’s performance of A-E (in whatever order) was required of Smith. And fundamentally, that would be Smith’s reason for action.
The concept of a prima facie reason would make good sense here. Imagine that Smith sees that he has reason to perform A, but doesn’t see that he also has reason to perform B. In that case, he’d have prima facie reason to perform A. But the concept of a pro tanto reason is misleading at best in this context. To say that Smith has pro tanto reason to perform A is true, but it’s only true insofar as Smith has reason to perform the whole set, A-E. A has the normative weight it has in virtue of the further requirement to perform B-E, not on its own.
(2) Different case. Imagine that Jones has five pro tanto reasons, A, B, C, D, and E, none of which is sufficient, and no combination of which is sufficient, for an all-in reason to act. A gives Jones some reason to act, but not an all-in reason. B does the same. So does C, and so on. The question arises: in virtue of what do any of the pro tanto reasons give Jones any reason to act?
It seems to me that this question leads in each case to a regress of reasons to a terminus. (I’m making some foundationalist-type assumptions here about regresses.)
A leads to some terminus T1.
B leads to some terminus T2.
And so on, so that we have five separate termini, T1, T2, T3, T4, and T5.
The question for me is this: do advocates of the approach you’re describing in the post think that there are as many termini–as many fundamental sources of normativity–as there are pro tanto reasons for action? That seems very implausible. If I have a thousand pro tanto reasons for action, am I somehow subject to a thousand separate, autonomous, ontologically unrelated sources of normativity? That’s not quite a reductio, but it’s close.
The idea of a pro tanto reason for action strikes me as theoretically ad hoc. It assigns an unspecified “weight” to a reason without explaining where the weight is supposed to come from. How are the weights set? Why is that considerations have any weight (as opposed to zero weight) at all? It makes sense as an account of the phenomenology of indecision–of a set of reasons that half-incline the agent in a variety of conflicting directions–but otherwise, I don’t get it.
LikeLiked by 1 person
Thanks, Irfan. You give me good occasion to think harder about these matters…
First, I want to clarify (and add just a bit to) my post. The point was that, sometimes, X having reason to A relative to end E (and E might simply be X’s A-ing whenever this is an option) does most of the work in making X’s A-ing the top-ranked option for X. Though other ends relevant to the ranking of A’s options might do some work, they are swamped by the effect of E. In such a case, though the ought-making is perhaps technically an “additive” affair (oughts, after all, are all-relevant-ends-in, within some domain), what seems more essential is saying that E and its importance makes it clear that, barring very strange circumstances, A will come out on top. Relatedly: if one were reasoning things out, one would not need to weight the effects of E on the option-ranking against the effects of the other relevant ends. For example, it is at least plausible that the end of not, in any given case, believing both P and not-P does most or all of the work in ranking this option last (so that one ought not to believe both P and not-P in each case where this is an option). Moreover — and I did not say this in the post — it is plausible that some “reasons” (in the sense of relevant ends) function in precisely this way. Their job is to determine (do the bulk of the work in determining) what one ought to do in some range of decision situations. When relevant ends (“reasons” in this sense) do this kind of work non-accidentally, they seem to constitute requirements (again, think of the rational requirement not to both believe that P and believe that not-P). In this sense, the reasons explanatory framework can explain why some reasons are requirements (have requiring normative force that normally does work sufficient for determining the relevant ought).
Second, I want to try to answer some of your concerns about the ordinary understanding of normative reasons and about explaining requirements (and other normative features) in terms of reasons (relevant ends, their effects on option-ranking or in making for X having reason to A with respect to E). I’ll do this by quoting some bits from your reply to my post and then briefly replying to these myself. The hope is that, while not fully understanding all that you have said or where we agree or disagree, these replies will constitute (and result in) progress.
(A) “I find the idea of a reason for action based on a requirement (say, a teleological one) perfectly intelligible, but find the idea of a pro tanto reason for action that isn’t, or isn’t based on, a requirement puzzling to the point of unintelligibility.” Is the first idea here that X has reason to A, relative to end E (and relative to X’s A-ing realizing or promoting E) only because X is required to realize or promote E? I’m not sure that I have any gripe with this, but neither am I sure how useful such a sense of ‘requirement’ is (such a requirement is hardly very specific; it might not function to make some one option better than others in decision situation). If the second idea is simply that relevant end E has to have some property (being desired, being valuable, being important to X) in order to generate X having reason to A, then I agree. The reasons-first folks generally don’t deny this. But I might have you wrong on either or both of these points.
(B) “A pro tanto reason, by contrast, is pro tanto for S because it has some non-determinative, non-conclusive normative weight for S. It normatively inclines S in the direction of action, but isn’t an all-in consideration to push her into action. But it has this status out of any relation to a requirement that is an all-things-considered reason to act a certain way. It’s as though a pro tanto reason just has semi-conclusive normative weight, full stop, with no further account of why it has the weight it has.” X having reason to A in respect of E is non-determinative only in the technical sense that oughts are always all-in (relative to all relevant ends) in a domain. But X having reason to A in respect of E might do most of the work in generating the option-ranking. (Only if E is the only relevant end does it fully determine the all-relevant-ends-in ranking – and hence that X ought to A, if this is the case.) We can rank X’s options only with respect to E if we like, but this is not the option-ranking that determines best decision (what one ought to do). I think all that the ordinary understanding of normative reasons is committed to is the idea that we can isolate the effects of individual relevant ends on the ordering of options. This can be put, admittedly in a way that might mislead, with respect to option A, in terms of X having reason (some degree or quantity or weight of reason) to A in respect of E.
(C) “…do advocates of the approach you’re describing in the post think that there are as many termini–as many fundamental sources of normativity–as there are pro tanto reasons for action?” I’m not sure I’m getting you here, but I think the (or a common) ordinary understanding of normative reasons has it that each relevant end is a reason and also makes for various reasons, where reasons are simply factors that explain the option valences (or ranking) in terms of the end valences (or ranking). The relevant ends — and the features of them that make them the relevant ends — are the sources of the normative valences (rankings) of the actions (options). So for each end there are tons and tons of related reasons (that go into fully, saliently explaining why X has reason to A with respect to E). If we try to give each of these items “weight” and “add up the weights” to get some quantity of X having reason to A, what we end up with is just a mess. But there are, indeed, often or always, multiple sources of normativity — multiple relevant ends — for each decision situation. I don’t think this is any kind of problem. From the ranking of these ends and how much more or less likely each option would make each end if it were chosen, we just do something like decision-theoretic math to generate the rankings of options. Admittedly, talk of X having (some quantity of) reason to A in respect of E can be deceptive, making the quantitative relationships seem more cardinal than ordinal and ignoring the holistic or relative or systemic nature of rank-ordering. It is a convenient way of speaking, but we do need to watch out for abstracting away relevant truth.
LikeLike
I guess I’m a little lost here. I no longer get the contrast you were originally drawing between a requirement and a pro tanto reason.
If S has just one goal, G, then G functions as a requirement: G is ex hypothesi the only thing that gives S reason for action. Suppose that S has two goals, G1 and G2. Then both do. If both give reason at once, but demand different actions of S, then S has pro tanto reason to do what G1 requires and pro tanto reason to do what G2 requires. In principle, this can leave S at a stalemate about what to do, all in (in the case where G1 and G2 demand incompatible actions). On this scenario, teleological requirements generate pro tanto reasons. But if so, there is no clear contrast to be drawn between requirements and pro tanto reasons. The two things are part of one picture; our pro tanto reasons trace back to teleological requirements.
But suppose we take requirements completely out of the picture. Then we just seem to have free-floating pro tanto reasons. I find this picture a mystery. How does the agent get pro tanto reasons without being required to do anything?
Now that you’re describing pro tanto reasons relative to ends (your E’s), it seems to me that what you’re describing is the first of my scenarios above, not the second. But if that’s so, I don’t understand the problem you’re trying to solve.
LikeLike
[I did two replies – this is the second one, dealing with your scenarios and framings, Irfan. The first one is below and for some reason does not appear as a reply to your reply above. Read that first.]
I think you are right that your first scenario is the most relevant one. So let’s set up some cases like this and see what the reasons framework might say (and what I think you might say or want to say).
Case 1: One relevant goal for X, G1. Three options, A, B, C. Each option promotes G1 to the same degree. No oughts, no requirements, right? And yet: (a) X has reason to A with respect to G1, (b) X has reason to B with respect to G1, (c) X has reason to C with respect to G1. These are the specific valences. And (d) X has sufficient reason, all relevant goals considered, to A and (e) to B and (f) to C. These are the resultant valences for each option. (Having sufficient reason here is technically a resultant valence, not a goal-specific valence, since it takes all relevant goals into account. It just happens that G1 is the only relevant goal.)
Case 2: Same as #1, except that A is the best option (each option still realizes or promotes G1 to some degree, however). So X ought to A. Not clear that X is required to A (unless A is the best option by quite a lot or some other further condition is met such that A wins is some decisive way). I say: X has reason to A (and B and C) with respect to G1, X has most reason (and hence ought) to A with respect to all relevant goals (G1 being the only relevant goal). And: it being the case that X ought to A is a resultant valence explained by X having more reason to A than to B or C with respect to G1 and G1 being the only relevant goal.
Case 3: Same as #1, except that A is the best option by far. So (plausibly) this suffices for X being required to A. I say much the same thing as for #2. Similarly plausible explanatory direction in the opposite direction from reasons to requirements, not the other way around.
Here’s what I conclude from these three cases:
(1) X having reason to A with respect to G1 is something that can occur without X being required to take any option of A, B or C.
(2) At least quite plausibly, X having reason to A with respect to G1 (to some degree or with some weight, relative to X having reason to B and to C with respect to G1) is what explains it being the case that X ought to A (when this is the case) and it being the case that X is required to A (when this is the case). It is true that, if G1 is the only relevant goal, then, if certain conditions are met, X ought to A (or X is required to A). But these conditions are only sometimes met.
(3) One might, and some people do, explain X having reason to A in respect of G1 in terms of the conditions under which it would be the case that X ought to A (or even is required to A) with respect to G1 (e.g, G1 is the only relevant aim, X’s A-ing strongly promotes it, the other options to not promote it or frustrate it significantly). Is this the kind of hypothetical analysis the one that you would endorse?
Finally…
Case 4: Multiple relevant goals, G1, G2, G3. X’s A-ing realizes or promotes G1 to some degree, such that X has reason to A with respect to G1. But suppose that G3 is so much more valuable or important than both of G1 and G2 that the different degrees to which A, B, and C promote G3 easily determines the ordering of A, B and C. So X having reason to A with respect to (the relatively unimportant) G1 does not seem to go along with any particular thing that X actually ought to do (or is required to do). The only hope, again, seems to be to go hypothetical.
I don’t think hypothetical analyses of objective normative reasons (to speak loosely) in terms of (hypothetical) objective normative oughts (or even requirements) is crazy. My intuitions go the other way and, as I have indicated, I think there are strong explanatory reasons that favor the reasons program.
Objective normative reasons, and reasons in general, might depend on standards of good reasoning (and hence rational requirements) — e.g., one idea is that all reasons are essentially things that might appear as premises in relevant bits of good reasoning. I like this idea, but it leaves up in the air how and why rational requirements are normative (if they are – some people implausibly claim that they are not). But it is easy to get way too far away from Kansas with this stuff.
LikeLike
Thanks for that reply. Again, more helpful than you might imagine.
I think there is a clearer way to put my original point. Distinguish between specific and resultant normative valances: ways of (i) X having reason to A relative to goal G1 (in respect of X’s A-ing realizing or promoting G1) vs. things like (ii) X having reason to A (it being open relative to which goal or goals and the goals are not specified), (iii) X having most reason to A, (iv) X having decisive (in the sense of decisively-most) reason to X, (v) X having sufficient reason to A. One might, and I think many do, think of X being required to A as a resultant normative valence — in particular as a species of ought that is particularly decisive (in which the best option is best by a country mile, something like [iv] just mentioned).
But one might also think of X being required to A as a specific normative valence that usually wins the day in a big way. And it might even be the constitutive job of the requirement or the relevant goal to do this work.
As a specific valence, X being required to A with respect to G1 would be “a reason” (in a loose way of speaking) or a way of X having reason to A with respect to G1 (in a less loose way of speaking). If X is required to A with respect to G1, this comes to X having a very strong, usually decisive, reason to A (with respect to G1), whether or not, in the present circumstances or choice situation, this results in X having most (or decisively most) reason to A. So — surrendering to the common, loose way of speaking — I think that many requirements are just reasons that are normally (or that have the job of) decisively settling matters about what one ought to do.
I think we are speaking of requirements in this sense when we speak of our being morally required not to torture people or commit murder. And I think that, at least most of the time, when we speak of requirements, we are speaking of specific, normally-decisive valences (and their goals) — not the corresponding resultant valences (or all-in rankings, within a domain) that they tend to produce. (An important exception would be speaking of what some system of values and norms requires of us. If I have to break a promise to save a life, then morality requires of me that I break my promise. I ought to do this and I’m required to do it, this is what morality requires of me.) If I am right, then one is highly likely to be getting things wrong in speaking of requirements as species of ought (even though they are closely-related to decisive oughts).
So reasons (or X having reason to A in respect of G) are inherently pro tanto in that they contribute to, but need not settle, matters about what one ought to do (in some domain). Specific requirements are just reasons (ways of X having reason to A in respect of G1) that do (generally do, have the job of doing — something like this) settle matters decisively in their favor. We should not think of reasons as things that do not settle matters about what one ought to do, but rather as things that need not settle such matters. Reasons that are requirements do settle such matters (generally do, have the job of doing this) — and decisively so. We don’t want to think of reasons as “merely contributory” (for though they need not they might well settle it as to what one ought to do). And we don’t want to think of requirements — at least not all of them or in all contexts — as especially strong or decisive oughts (as this sort of resultant valence). Even though such decisive oughts (resultant valences) do play an important role in explaining/individuating specific requirements.
Does this help you understand my original point, Irfan? Anyhow, it helps me understand my original point! So, again, thanks for prompting the additional thinking and (hopefully) clarification.
LikeLike