Friday, May 7, 2021

Vagueness about moral obligation

There is a single normative property that is normatively above all others, that overrides all others: moral obligation.

I think the above intuition entails that there cannot be any non-epistemic vagueness about moral obligation.

There are two main non-epistemic approaches to vagueness: deviant logic and supervaluationism. Deviant logic is logically unacceptable. :-) That leaves supervaluationism. But on supervaluationism, there would have to be many acceptable precisifications of our concept of moral obligation. Each such precisification would presumably be a normative property. But only a precisification that was normatively above all others could be an acceptable precisification of our concept of moral obligation. And there can only be one precisification above all others. So there can only be one acceptable precisification of moral obligation.

The above argument is too quick. The supervaluationist can say that in the claim “Moral obligation is above all other normative properties”, we have another candidates for vagueness: “above” (or “overrides”). Then we need to engage in coordinated precisification of “moral obligation”, as well as “above”. For each coordinated precisification, the aboveness claim will be true: “Moral obligationi is abovei all other normative properties.”

I think, however, that once we allow for a variety of precisifications of “above”, we betray the intuition behind the aboveness thesis. That in some sense moral obligation is above personal convenience is not the bold and bracing intuition of the overridingness of morality. Thus, I think that if we are to be faithful to that intuition, we cannot allow for non-epistemic vagueness about moral obligation.

And this, in turn, greatly limits how much non-epistemic vagueness there can be. For instance, if there is no vagueness about permissibility, then it cannot be vague whether something is a person, since vagueness about personhood leads to vagueness about moral obligations of respect. Indeed, it is not clear that there can be any non-epistemic vagueness if there is no non-epistemic vagueness about moral obligation. Suppose I promise to become bald, and I have a small amount of hair. Then I am non-bald if and only if I am obligated to remove some hair.


Brian Cutter said...

I find this idea pretty compelling. Maybe further support comes from the connection between vagueness and semantic plasticity. With a typical vague predicate like "bald," we can change what qualifies as a borderline case by small changes in usage. So, suppose a person with n hairs is a borderline case of "bald" on its current meaning. If people become ever-so-slightly less reserved in their disposition to apply "bald," then the meaning/extension can shift to count an n-haired person as a definite case. But it's hard to believe that our becoming slightly more liberal in our dispositions to use "permissible"/"obligatory" can change an action from a borderline instance to a definite instance of the predicate (or vice versa). (I think Miriam Schoenfield talks about this in an Ethics paper on moral vagueness; I also discuss the issue in a phil perspectives paper from a few years back, though in connection with axiological predicates instead of deontic predicates.)

Actually, for the same reason, I'd be inclined to say that "permissible" isn't even *epistemically* vague. Epistemic conceptions of vagueness need to say how vagueness is different from other kinds of ignorance (not all cases of ignorance, even in-principle ignorance, are cases of vagueness). Williamson seems to think that the distinctive feature of vagueness is that it's ignorance *due to semantic plasticity*. If that's right, then if normative terms aren't semantically plastic, they aren't even epistemically vague.

(though I'm not sure any of this rules on the idea that moral vagueness is a kind of ontic vagueness; maybe there's a single candidate meaning for "obligatory"--the property of being obligatory---and it can be indeterminate whether it's instantiated)

Alexander R Pruss said...

Thanks for the references!

Alexander R Pruss said...

But if obligatoriness is not vague at all, then it seems that baldness is not vague at all. For take the case at the end of my post where I promise to become bald. Then I am obligated to shave more hair off iff I am not bald.

Would it satisfy the "semantic plasticity" requirement if my ignorance of whether I am obligated to shave depended on the semantic plasticity of the terms of my promise to become bald?

If so, here is a possibility: There are cases of vague obligation, but they all depend on the vagueness of language (in promises, commands, etc.). And because of the argument of my post, all the vagueness of language is epistemic.

Brian Cutter said...

Hm.. one possibility is that "bald" is semantically vague (multiple candidate meanings, no one of which it determinately expresses), whereas "obligatory" is ontically vague (only one candidate meaning that it determinately expresses, but it's sometimes indeterminate whether that property is instantiated). In the bald-promise case, then, maybe it's indeterminate (ontically) whether you are obligated to shave due to the fact that it's indeterminate (semantically) whether you are bald.

Alexander R Pruss said...

Doesn't ontic vagueness force one into a deviant logic?

Brian Cutter said...

I'm no expert but I don't see why we couldn't combine classical logic with ontic vagueness. You might think that where there is ontic vagueness, reality is indeterminate, and we can say that a sentence is true iff it comes out true on every "metaphysical precisification of reality", every way of resolving the indeterminacy in reality (instead of every semantic precisification of the language). Then the resulting logic should work out like supervaluationism. So, e.g. we'll get excluded middle. I'm pretty sure people have worked out a theory of ontic vagueness along these lines. (Maybe Robbie Williams or Elizabeth Barnes... but it's been a while since i've looked at this literature)

Alexander R Pruss said...

My preferred way of thinking about supervaluationism is that the sentence expresses more than one proposition, so we can say that it's definitely true when every proposition it expresses is true and definitely false when every proposition it expresses is false. But propositions themselves are sharp.

I was thinking that the different thing with ontic vagueness is that with ontic vagueness the propositions themselves are vague in truth value. After all, if P is a property, then every item either exemplifies P or does not exemplify P: there doesn't seem to be any room for anything else.

Brian Cutter said...

So then bivalence would fail (for propositions), just as bivalence fails (for sentences) on supervaluationism. But that's consistent with saying that every proposition that's a classical logical truth is true. (E.g., you could still say every instance of excluded middle is true, even if some instances have two indeterminate disjuncts.) But maybe you're thinking that the rejection of bivalence (for propositions) is bad enough.

Alexander R Pruss said...

1. I think the rejection of bivalence for propositions is bad enough.

2. T-schema + Excluded Middle implies bivalence (q implies Tq, ~q implies T~q, so (q or ~q) implies (Tq or T~q). So we need to deny the T-schema for sentences that express a unique proposition.

3. The following seems correct: If an atomic proposition p attributes a property P to an object x, then p is true iff x has P, and p is false iff ~(x has P). Thus, excluded middle for property possession implies bivalence for atomic propositions.