I'm really enjoying Scanlon's new book "Being Realistic About Reasons" (all citations below to this book), but I'm stumbling on the part about pure normative truths and the explanation of supervenience. Any help would be much appreciated.
Scanlon takes the reason relation to hold between tuples, (p, x, c, a), where p is a fact (a reason), x is an agent, c comprises some conditions (circumstances), and a is an act type. In a case where p = the metal is sharp, x = you, c = normal circumstances where you would avoid the pain by not pressing the metal, and a = not pressing the metal (3), Scanlon says the reason relation R(p, x, c, a) obtains (30).
He calls this a mixed normative fact because it contains normative and non-normative elements; this reason relation only obtains when the metal is indeed sharp. Aside from these, there are pure normative claims: "the essentially normative content of a statement that R(p, x, c, a) is independent of whether p holds. This normative content lies in the claim that, whether p obtains or not, should p hold then it is a reason for someone in c to do a. So I will take what I will call a pure normative claim to be a claim that R(p, x, c, a) holds (or does not hold), understood in this way" (36-37). Later he says the contingent non-normative elements are "subjunctivized away" in pure normative claims (40). So it looks like the pure normative claim for the sharp metal case is something like this:
(Pure?) It is normatively necessary that, for all agents, x, if it were the case that p and x were in c, it would be the case that x has a reason to a (40-41).
One of the advertised payoffs for separating the pure from the mixed is that it allows us to explain the supervenience of the mixed normative facts on the non-normative facts by appealing to the pure normative facts. But I am puzzled.
First the pure facts. As Schroeder's review in the AJP also points out, (Pure?) is an odd candidate for pure normative content. It looks like is is a claim about when pure normative content would hold (i.e., a condition under which there would be a reason). It not only subjunctives away p, but the reason relation itself. And it looks like a claim in need of a truth-maker.
I get my head around this by thinking of the pure normative content in terms of Ewing's moral laws (or Enoch's norms with modally maximal jurisdiction). True claims about them are necessarily true; they tell us, for any possible non-normative facts, whether they have normative significance (42); and they do not contain any natural elements. Whether they are purely normative themselves is questionable, though. They seem to be normative status conferring without themselves being normative (or natural). But I would love suggestions about how to understand pure normative content in other terms.
Second, explaining supervenience. Why can there be no change in the reason-relatedness status of the tuple (p, x, c, a) without some change in the non-normative features exhibited by (p, x, c, a)? The classical realist explanation is to posit real necessitation relations between non-normative properties and normative properties/relations (I think of a really strong towline). Ewing definitely wanted to avoid such "necessary synthetic connections", so he gave a different explanation: there are necessary moral laws (here, we can say normative laws) that somehow confer normative status on items with certain non-normative properties. Scanlon seems to opt for this second explanation, and finds supervenience less puzzling in light of it.
I am not so sure. First, I see dimly what it would be to have an explanatorily basic necessary synthetic connection between distinct existences (though I try to avoid them). But I do not see what an explanatorily basic necessary normative law would be. For a familiar law L (say, a traffic law), there are some non-legal facts in virtue of which L. But Ewing-Enoch-Scanlon necessary normative laws are not laws in virtue of anything. (Well, for Ewing, it might be in virtue of God's attitudes). For me, sui generis laws seem more puzzling than sui generis necessary synthetic connections.
Second, I see dimly what sui generis normative properties would be on the classic realist explanation (though I try to avoid them) - they are just intrinsically authoritative properties necessitated by certain non-normative properties. But I do not see what the normative statuses (properties?) conferred by necessary normative laws amounts to. Is the reason relatedness for a tuple *reducible* to facts about the necessary law? Is it that, for a tuple to be reason related just is for the pure normative law to decree that tuples like that are reason related? That makes the normative authority of the status (property?) somewhat elusive. But the alternative is to have non-normative properties, irreducible normative properties/relations that are intrinsically authoritative, and Scanlon's explanatorily basic necessary normative laws ensuring (describing?) the modally systematic connections between the two. I do not see how that is a better explanation than positing explanatorily basic necessary synthetic connections between the two. It can be hard to see how it is a different explanation.
Last, Blackburn's explanatory burden still lingers. Though there is no agreement about what that is, I think it comes into focus when we attend to the full statement of general strong supervenience.
(SS) Necessarily, if any tuple (p, x, c, a) exhibits the reason relation, R(p, x, c, a), then there are some non-normative features that the tuple exhibits, NN, such that, necessarily, any tuple (q, y, d, b) exhibiting NN bears the reason relation, R(q, y, d, b).
I take it that the wide scope necessity is conceptual and the narrow scope necessity is normative.
I think the burden is to explain how it is that we can know *as a conceptual truth* that there are normative necessities (synthetic relations or laws) dictating which non-normative properties are of normative significance when the normative necessities are not conceptual truths. If a normative necessity is to be discovered (perhaps a priori but not via conceptual analysis), how is it that we know *as a conceptual truth* that there is some necessity to be found prior to discovering it? Very odd. (BTW, my friend Cole Mitchell has a great paper on Blackburn's challenge.) This probably isn't on Scanlon's plate, so we can set it aside if you like.
This is too much, isn't it? OK. Thoughts?