To say that I ought to take out the garbage and to say that I must take it out is to say two different things. And, if I ought to take out the garbage, it does not follow that I must. But — apparently — if I must take out the garbage (if I am required to), then it follows that I ought. The ‘must’ seems in some way stronger than the mere ‘ought’ (perhaps ‘must’ is simply ‘decisively ought’ — that is one theory).
There is an assumption here — evident in the Snedegar paper on normative requirement but also in most of the other literature that I have encountered — that I want to question. It is assumed that requirement is (or entails) some kind of overall normative valence (an all relevant things considered, or all relevant things in some domain considered, valence). However, at least in one sense of ‘requirement’ (or ‘must’) this is not true. For, rather famously in the history of Anglo-American theorizing about normative ethics, I might be morally required to PHI and morally required to PSI, but be able to PSI only by failing to PHI (e.g., I can only save a person’s life by breaking my word). If there is, in such situations, some best choice that involves violating one of the requirements, we should say that such requirements are kinds of reasons (to put things in the rather loose but common way) — they are partial or specific not overall normative valences (even if, normally, the weightiness or importance of a requirement usually makes it the case that, when one is required to PHI, it is also the case that one ought to PHI).
Is there a concept of (or might we usefully coin a concept of) normative requirement that is (or entails) a genuinely overall normative valence (i.e. an ought of some sort)? Perhaps in the classic ethical quandary sort of case, one is “all-in required” to abide by the more weighty or important “specific requirement.” But it seems to me that putting things this way does violence to the concept of requirement. For, if abiding by requirement R1 wins out, but only by a little, over abiding by incompatible R2, though it is obvious that one ought to abide by R1 (because this option is just a little bit better), it does not seem that one is required to (because an option being best, or such that one has most reason to take it, does not make an option required).
What makes more sense to me is this: distinguish balance-of-reasons oughts from oughts in which some relevant requirement (that is a reason or partial or specific normative valence) carries the day (makes for the ought). We can keep the result that, in some normal range of circumstances, requirement (plus the circumstances) entails ought. And we have a nice error theory for the (tempting but false) idea that requirement (all on it own) entails ought. I think this does justice to our intuitions. However, the assumptions that I am challenging here (that requirement is or entails on overall valence, that requirement entails ought but not the other way around) seem pretty standard in the literature. So I have to wonder whether I am making a mistake somewhere in my thinking (or if the different framings here, when thought of properly, are merely verbally not substantively different).
I don’t follow this:
More precisely, I don’t follow the argument in the fourth sentence that begins with “For…” The proposal you’re considering is that there is an “overall normative valence,” an “all-in requirement.” In a conflict between two conflicting requirements, R1 and R2, let’s say R1 wins out. So R1 is the “specific requirement” in a given context. So R1 is (in that context) an all-in requirement, or an all-in ought.
There’s no implication that R1 defeats R2 “by a little bit.” It depends what we’re talking about. If R1 is “keep your promises,” and R2 is some duty to save a life in jeopardy, then in a context where R1 is trivial and an occasion for R2 is urgent, R2 will defeat R1 by a wide margin, or just override R2 altogether without remainder. In a case like this, R2 is a requirement even if, ordinarily, R1 usually is.
I guess I don’t see the “violence to the concept of requirement” you bring up.
LikeLiked by 1 person
That’s a helpful challenge. The more important, and more easily defensible, point is that, when abiding by R1 (by PHI-ing) is more important than abiding by R2 (by PSI-ing) — whether or not just by a little bit — it seems perfectly coherent to say that: (a) you ought to PHI, (b) you must PHI and (c) you must PSI as well while (d) making no claims about anything being all-in required (or there being anything that one is all-in required to do). I should not have relied on the “by a little bit” element and should not have claimed that, in virtue of this element, it is odd to say that the agent is required to PHI (abide by R1).
Maybe it matters if there is a requirement (not just most reason) to abide by R1 over R2 when doing both is impossible. But even here I would press the idea that such a second-order requirement is just a special sort of specific valence (with a specific sort of role), not an all-in or resultant valence. The resultant normative valence here seems to me to be the familiar ought. This is partly a flat-footed intuition, but also partly, I think, a function of some inchoate theoretical commitments (that I have not fully articulated). Though I’m here explicitly backing off pushing toward the claim that the no-all-in-requirements way of thinking of these matters is better, I would at this point want to know more about the all-in requirements-framework (what, precisely, it comes to, in substantive, not-merely-verbal terms) and what the virtues of adopting it is. We need to know what, if anything, is at stake. My getting clearer on the background commitments for my intuition here would help for sure.
(One might think that requirements are like aptness conditions for attitudes. Aptness is an all-or-nothing normative property — in this, it is similar to the ought-property — but, in the odd cases, the aptness of a response functions as a reason, being a contributory valence when the focal apt response is not or might not be all-in best. Aptness is not, though, a reason or contributory valence (though it is a specific one, this stimulus calling for that response being something like a simple normative feature). On this model, contrary to what I claimed in the post, requirements are merely associated with reasons or can function as reasons or contributory valences; they are not identical to reasons. So maybe I want to revise this claim as well.)