Wednesday, March 30, 2005

Supervenience vs. Hume's Law

Hume's Law: you cannot derive 'ought' from 'is'; in other words, factual premises cannot entail an evaluative conclusion.

The supervenience constraint: the moral facts are, in a sense, 'fixed' by the natural facts. There is no possible world that is exactly like our own one in all factual respects, but where it was 'morally right' of Hitler to carry out the Holocaust.

It seems to me that there is some tension between these two principles. If the natural facts 'fix' the moral facts, then wouldn't that set up an entailment relation between them? If you had an accurate theoretical account of just how this 'fixing' was determined, couldn't you reason straight from the natural facts to a moral conclusion? (And don't we attempt to do this all the time - albeit with imperfect natural and theoretical knowledge?)

So why does anyone believe Hume's law?


  1. It seems to me that, if you concede Hume's law, you're not going to be able to derive many (if any) moral principles at all.

  2. I think you mis-understand supervenience. It's little more than a particular sense of the phrase "depends upon".

    Just because moral facts are affected by physical facts, doesn't mean than physical facts define or fix moral ones.


  3. Richard, an important aside: Hume derived ought from is. His law refers only to objective facts: facts other than facts about our attitudes toward those facts. Once you add in the facts about sentiments, Hume thinks you can derive ought from is.

    Actually, it's not only an aside. I think it gives a Humean answer to your question. This means that your question is good evidence against any position that isn't of the Humean sentimentalist, reductive variety.

    MP has given the reply to such an argument, however. Yet, Hume would surely rebut this with the argument from queerness.

  4. Jim, I do like that point - James Rachels suggests much the same in an article I'm reading at the moment. (Though I think he intends it as an improvement on Hume rather than an accurate interpretation of what the man actually said. But that's fine by me; I'm more interested in philosophy than history!)

    MP, my understanding of supervenience is that if B supervenes on A, you cannot have a change in B without a change in A. That sure looks to me like A 'fixes' B. Anyway, it's the concept that matters, not the word, so call it whatever you like. The point is, it does seem extremely plausible that moral facts are fixed in just this way by the natural facts. And this would seem to disprove Hume's Law.

  5. I think it's consistent to deny the supervenience constraint while affirming Hume's Law.

    Suppose the supervenience constraint holds. Then, if

    1. We know ALL the natural facts in case A, and

    2. We know ALL the natural facts in case B, and

    3. We know that the natural facts in case A are the same ones as those which obtain in case B, and

    4. We know that moral fact X obtains in case A,

    then we could infer

    5. Moral fact X obtains in case B.

    It is consistent with the supervenience thesis to say that if any of 1 through 4 fail to hold, then we could not infer 5.

    Item 4 includes knowledge of a moral fact. So one could affirm the supervenience thesis and still claim that in order to make an inference about the moral facts in case B, one would first need to know something about the moral facts in case A.

  6. Yeah, good point, though I was hoping to pre-empt that with my remark about how we also need "an accurate theoretical account of just how this 'fixing' was determined".

    Presumably the moral fact X obtains precisely because of the various natural facts (about suffering and so forth). So if we knew just how they were related, we wouldn't need to appeal to any moral facts (e.g. from other 'cases') at all. We could use the theory to "reason straight from the natural facts to a moral conclusion".

  7. Historical Aside:

    Richard, I stand by my reading of Hume, though Rachels is certainly smarter than me. I can't see how he thinks its an "improvement" on Hume, when I've obviously just interpreted Hume himself as saying it. Perhaps I'm reading in. Let me know if you find any data resolving the matter.

  8. Richard--

    But the theoretical account would, presumably, not itself be a natural fact. Presumably it would be a moral principle of the form

    "Whenever natural facts a, b, c obtain, moral fact X obtains"


    "Whenever an act is an act of killing for pleasure, that act is wrong"

    This principle would need to be part of your inference. Your inference would look something like

    1. Bob killed Sue for fun.
    2. [moral principle above]
    3. Thus, Bob did something wrong.

    So even if supervenience commits you to the possibility of a theoretical account of the supervenience relation (and I am not sure that it does), I still don't think you would need to say that factual premises can, all by themselves, entail an evaluative conclusion.

  9. But the same holds for any cross-disciplinary comparisons (e.g. within science), so if that's all that Hume's law amounts to, it's not a very interesting claim.

    For example, you can't reason from "the shaker contains NaCl" to "the shaker contains salt" without the theoretical premise: salt = NaCl. Does that show there's a fact/chemistry gap? If so, these gaps are fairly trivial.

    Besides, entailment merely requires that whenever the premises are true, so is the conclusion. So long as some moral theory is correct, it need not actually be included in the entailment. The theory correctly states that the natural facts determine a particular moral conclusion. That is, there is no possible world where the natural facts hold but the moral conclusion does not. So the natural facts entail the moral conclusion.

  10. Richard, all that seems right to me. A few additional points:

    There's a difference between derivation and entailment. Consider this line of inference:

    1. the shaker contains salt
    2. Salt=NaCl
    3. Thus, the shaker contains NaCl

    1 entails 3, but (I claim) 3 cannot be derived from 1 unless I include 2 in my derivation. This is because 2 is not analytically true, so 3&1 says something "more" than just 1, even though 3&1 is necessary given 1.

    I think something similar might be true in moral reasoning. Pasting from my last comment:

    1. Bob killed Sue for fun.
    2. [moral principle above]
    3. Thus, Bob did something wrong.

    Given 1, 3 has to be true. Nevertheless, 3&1 is "more" than 1. If I knew 1 but didn't know 2, I would not be entitled to believe 3. This might not seem important in the example just given, since the moral principle at issue is fairly obvious. But there are probably some true moral principle which are not at all obvious, despite being necessary. (E.g. there are probably true, necessary, non-obvious principles concerning abortion, euthanasia, etc.)

    The relevance of all this is that both "derive" and "entail" appear in your statement of Hume's Law. So it's not completely obvious (to me) whether the possibility of a theoretical account explaining the supervenience relation would make Hume's Law true or false.

    Finally, I'd just like to reiterate my previous point: Affirmation of the supervenience constraint does not commit one to the possibility of a theoretical account of the supervenience constraint. For instance, some (most?) particularists hold that the supervenience constraint is true, though they typically deny that any (non-trivial) true moral principles are available.

  11. One last thing: I think that it is an open question whether there's a "fact/chemistry gap." If chemical facts supervene on (say) microphysical facts, it is still possible that no theoretical account of that relation is possible. I don't think it is trivial to ask whether this will turn out to be the case. But that's possibly not important for present purposes.

  12. Yeah, your comments about derivation vs. entailment are spot on. It's the latter I'm interested in, since - as we've noticed - the former 'problem' is nothing unique to ethics.

    Now, you also write:
    "Affirmation of the supervenience constraint does not commit one to the possibility of a theoretical account of the supervenience constraint."

    I'm not sure that this impacts my point. It's helpful to have a theory, but it doesn't change the brute facts. And if the facts in (3) supervene on those in (1), then it's impossible for (1) to be true and yet (3) false, right?

    Put another way: even without a moral theory, factual premises would still entail a moral conclusion. We just wouldn't know what it was!

    But I don't think we need to be so pessimistic. We can (correctly) infer Hitler's immorality from the facts of the holocaust. So that seems to show that the fact/value gap can be overcome.

  13. Well, here's the thing. I'll agree that natural facts do entail moral facts, given supervenience. For instance, given supervenience, the facts about the holocaust entail that the holocaust was wrong. If these same facts are repeated in a new instance -- another holocaust -- then the new holocaust will be wrong. The problem is that the same facts cannot be repeated; there is only one holocaust which is factually just like Hitler's holocaust. So to say "All holocausts factually the same as Hitler's holocaust are morally wrong" is not much more than saying "Hitler's holocaust is morally wrong."

    So: In a trivial sense, supervenience does imply that an "is" can entail an "ought." The problem is that supervenience only implies that this entailment occurs one time.

    To get the sort of entailment we'd like (i.e. an entailment relation which can hold in more than one case) we need something more than supervenience -- we need something like

    "There is no possible world that is exactly like our own one in all _morally relevant_ factual respects, but where it was 'morally right' of Hitler to carry out the Holocaust."

    This is not supervenience. I guess it's universalizability.

  14. I think that Frank Jackson has made something very much like the point you make here.

  15. Heh, that's often the way of it. Do you recall where he talks about this stuff?

  16. In Section 4 of an article called "Unprincipled Ethics" (Midwest Studies in Philosophy XX (1995)), Gerald Dworkin addresses an argument which (I think) is closely related to the one you make here.

  17. Check out chapter 5 of Frank Jackson's From Metaphysics to Ethics.


Visitors: check my comments policy first.
Non-Blogger users: If the comment form isn't working for you, email me your comment and I can post it on your behalf. (If your comment is too long, first try breaking it into two parts.)

Note: only a member of this blog may post a comment.