Vagueness and moral obligation

It sure seems like there is vagueness in moral obligation. For instance, torture of the innocent is always wrong, making an innocent person’s life mildly unpleasant for a good cause is not always wrong, and in between we can run a Sorites sequence.

What view could a moral realist have about this? Here are four standard things that people say about a vague term “ϕ”.

1. Error theory: nothing is or could be ϕ; or maybe “ϕ” is nonsense.

2. Non-classical logic: there are cases where attributions of “ϕ” are neither true nor false.

3. Supervaluationism: there are a lot of decent candidates for the meaning of “ϕ”, and no one of them is the meaning.

4. Standard epistemicism: there are a lot of decent candidates for the meaning of “$”, and one of them is the meaning, but we don’t know which one, because we don’t know the true semantic theory and the details of our linguistic usage.

If “ϕ” is “moral obligation”, and we maintain moral realism, then (1) is out. I think (3) and (4) are only possible options if we have a watered-down moral realism. For on a robust moral realism, moral obligations really central to our lives, and nothing else could play the kind of central role in our lives that they do. On a robust moral realism, moral obligation is not one thing among many that just as well or almost as well fit our linguistic usage. Here is another way to put the point. On both (3) and (4), the question of what exact content “ϕ” has is a merely verbal question, like the question of how much hair someone can have and still be bald: we could decide to use “bald” differently, with no loss. But questions about moral obligation are not merely verbal in this way.

This means that given robust moral realism, of the standard views of vagueness all we have available is non-classical logic. But non-classical logic is just illogical (thumps table, hard)! :-)

So we need something else. If we deny (1)-(3), we have to say that ultimately “moral obligation” is sharp, but of course we can’t help but admit that there are Sorites sequences and we can’t tell where moral obligation begins and ends in them. But we cannot explain our ignorance in the semantic way of standard epistemicism. What we need is something like epistemicism, but where moral obligation facts are uniquely distinguished from other facts—they have this central overriding role in our lives—and yet there are moral facts that are likely beyond human ken. One might want to call this fifth view “non-standard epistemicism about vagueness” or “denial of vagueness”—whether we call it one or the other may just be a verbal question. :-)

In any case, I find it quite interesting that to save robust moral realism, we need either non-classical logic or something that we might call “denial of vagueness”.