Monday, March 25, 2013

Moral Supervenience

A curious new piece in Analysis from Gerald Harrison argues that 'The moral supervenience thesis is not a conceptual truth' -- or, for that matter, an obvious a priori truth.  I think it is at least the latter, so I was interested to read what Harrison has to say against it.

Moral supervenience (S) is the thesis that the moral truths supervene on the "natural" (read: "non-normative") truths.  This seems self-evident: moral facts cannot "float free" of the underlying natural facts -- there could not be a world just like ours, but where Hitler was actually a virtuous fellow.  Such a radical moral difference would need to be grounded in a corresponding natural difference in the world -- say where Adolf never become a fascist, and instead dedicated himself to the peaceful pursuit of philanthropic ends.  To expand upon this intuitive point:

(1) There seems an important sense in which the descriptive facts are what ground the particular moral facts.  But without moral supervenience, the descriptive facts would be insufficient to determine what moral facts hold in a situation.  There would instead be brute moral truths that aren't fully grounded by the descriptive facts, insofar as the former could change despite no change in the latter.  Not only is this intuitively crazy, but it also violates our theoretical sense of how the particular moral truths are grounded.

(2) The fundamental moral truths are plausibly necessarily true conditional statements, perhaps of the form, "In completely described natural circumstances Ci, the applicable moral truths are Mi."  This then explains the sense in which the particular moral truths Mi are grounded in the descriptive truths Ci.  But it also straightforwardly entails moral supervenience (S).

(3) The falsity of S would make nonsense of the guiding role of moral truths.  Note that, e.g., if someone is virtuous, or morally admirable, then it's fitting for us to admire them, or hold them in a certain positive regard.  Moreover, to avoid moral fetishism, what we should admire in them is not "virtue" in the abstract, but rather the particular psychological characteristics that are virtues.  But if S is false, then our emotions and attitudes cannot be guided in this way.  It makes no sense to admire someone for their psychological characteristics in our world, but then to condemn them for being exactly the same way in exactly the same descriptive circumstances in another world, where the moral truths about them brutely differ.

Harrison addresses something along these lines:

The principle of universalizability (U) states that if an action is right, then every action that is similar in morally relevant respects is right too.

However, he holds that the denial of S is compatible with U, because a brute moral difference would, trivially, be a difference in "morally relevant respects".

Of course, U is generally not meant to be interpreted so trivially.  Rather, when we speak of "morally relevant respects" here, we mean the underlying descriptive properties that have the higher-order property of being morally important.  Harrison goes on to consider:
[U2] if act x is right, then any act having the same properties in virtue of which act x is right, is also right.

Unfortunately, he reads U2 as entailing an implausible absolutism:
Act A might be wrong because it was an act of cruelty. Act B might also have been an act of cruelty in exactly the same way, yet have saved a million lives. According to U2 if A is wrong, B is too.

This, too, is not the best interpretation of Universalizability.  We should not be considering right-makers and wrong-makers in isolation, but rather the totality of natural facts that together entail the rightness or wrongness of an act.  This may include "negative" facts, such as that act A does not save a million lives.  That's a morally relevant natural feature of A, which differs from B, and explains why one may consistently reach different moral verdicts about the two.

When U2 is understood in this (obviously more sensible!) way, it is far harder to deny.  The whole paper seems to rest on neglect of this crucial point. (It re-emerges later when Harrison takes a grounding-style claim to be motivated by U2.)

Admittedly, if one is determined to reject the self-evidence of S, and hence even the better interpretation of Universalizability that I've offered here, then I'm not sure what more could be said to persuade them to reconsider.  But it would seem awfully odd and unmotivated.


  1. I am wondering what should be made of "similar in morally relevant respects." The reason is that I am tempted to a sort of fictionalism about human morality: human morality is the result of the highly social human psychology (which i take to be reducible to natural facts) and not the existence of moral facts or properties.

    On this view:

    (a) Hitler could have been a good fellow if human psychology were different. For example, humans might have had a psychology that valued fascist organizational structure, prejudice, non-humane behavior, and non-evidential propaganda above, say, what we value in the actual world...and...

    (b) The rightness of A and B would hand on what humans are disposed to think about A and B. So A and B could both be wrong or both be right, or one right and the other wrong. It would depend on human psychology.

    So if this kind of fictionalism about morality is true, then moral 'truths' still supervene on natural facts (about human psychology), but the "totality of natural facts" about any given matter will not reach consistent moral verdicts about A and B or about Good Hitler or Bad Hitler.

    But maybe I am misunderstanding S or "totality of (natural) facts." Thoughts?

    1. "human morality is the result of the highly social human psychology (which i take to be reducible to natural facts) and not the existence of moral facts or properties."

      If by "human morality" you mean the anthropological phenomena -- our moral beliefs, behaviour, etc. -- then that seems right, but off-topic.

      If you really think the particular moral facts radically depend on what we happen to approve of... well, such psychological facts can certainly be included in the "totality of natural facts", so is perfectly compatible with moral supervenience. The fundamental and necessary moral conditionals would then be something like "If most folks approve of X, then X is good." (But I do think such moral "mob rule" is pretty abhorrent on substantive grounds. Better to simply deny that there are any moral truths than to bestow such honours upon any sufficiently widely applauded genocide and gratuitous cruelty!)

  2. [Gerald Harrison responds...]

    I have read your reply. I found what you said extremely thought provoking. But I disagreed with a lot - as you might expext. I am sure my disagreement is due largely/entirely to me being a bit of a nincompoop. But here are my initial responses, for what they’re worth!

    First, I would point out that in my paper I did not argue that S is false per se. (I think the natural version of S is false, but I think the non-moral version is true! I’m a divine command theorist). I just pointed out that it isn’t the big deal breaker everyone thinkss. Moral philosophers shouldn't lose any sleep over denying S or spend much time partying if they can respect it: we should all lighten up about it. Denying S (if you have to) won’t cause the kind of havoc many think (or so I think).

    Of course, denying it does mean one has to admit that two situations identical in all of their natural properties can differ in their moral ones: there’s no escaping that. And perhaps some find it obvious that this cannot be so: that it is just self-evident that if two situations are identical in their natural properties they just must be identical in their moral ones too. I admit to not seeing this. If the case for S depends upon seeing this, then I’m a lost cause! Perhaps not seeing it means I lack an important qualification for thinking clearly about moral matters because I've clearly come a bit unstuck, morally speaking. I am open to the possibility. This should be borne in mind when reading the rest of what I say!

    Anyway, you have offered me reasons why I should buy S independent of its supposed self-evidence. 1: Hitler was a bad man and it is silly to think there is a possible world where he’s a good man but has exactly the same character and did the same things because if one does think this one will not be able to explain his actual badness in terms of his natural features. The badness will just be along-for-the-ride, so to speak. 2: The truth of S is entailed by there being necessarily true conditional moral statements. 3: Without S we cannot make sense of the guiding nature of moral truths. 4: S is entailed by a sensible version of U2. I thank you for taking the trouble to provide these, but unfortunately none has worked yet. I will take each in turn and try to explain myself

    When it comes to Hitler I think that pretty much anything you say about the morality of his character and actions I can say as well (hope so, anyway). The only difference is that you can pop the word ‘necessarily’ in here and there and I can’t do that. But I don’t think that makes any real difference. I say that H was a bad man who did many morally awful things. I say that he was a bad man because of his cruelty (among other things) and that his acts were wrong because of the terrible suffering they caused (among other things). Perhaps you think I can’t say these things due to my accepting there is a possible world identical to this one in all of its natural properties but in which cruelty is a virtue and causing others to suffer is right and praiseworthy etc. True, I can’t see any principled way of ruling out such a possibility if I deny S. But I fail to see why it means I can’t say that in the actual world Hitler was a morally bad man because of his cruelty and that his deeds were morally wrong because of the suffering they caused.

    I hate olives because of their bitter taste. But there is a possible world in which I love olives because of their bitter taste. It remains the case that in the actual world the statement ‘I hate olives because of their bitter taste’ is true.

  3. [continued...]

    If I deny S I can’t think of a way in which I can rule out a possible world in which morality loves cruelty because of the suffering it causes. But this doesn’t alter the fact that in this, the actual world, morality hates cruelty because of the suffering it causes. So I think I can say that morality hates Hitler because of his cruelty and the suffering he caused while at the same time acknowledging that there’s a possible world in which morality loves him because of his cruelty and the suffering he caused. I think I can say that Hitler was a bad man because of his cruelty and the suffering he caused just the same way you can.

    You say that there is an ‘important sense’ in which the natural facts ‘ground’ the moral ones. But if I’m right and I can say everything you can say about Hitler’s badness and the wrongness of his deeds it seems to me I can capture it just as well as you and that S didn’t help.

    No. 2. You say that fundamental moral truths are “plausibly necessarily true conditional statements”. Well, I accept that this is ‘plausible’ because I did not argue that S is implausible. But anyway it is the word ‘necessarily’ that is doing all the work of getting you from there to S. What reason, apart from S, is there to think that true conditional moral statements are necessary truths?

    Perhaps the best evidence for what it is that morality commands us to do in the actual world supports some form of utilitarianism. Then the evidence supports the claim that if, in the actual world, Xing would maximise happiness morality bids us maximise happiness. But why should I conclude that this is necessarily what morality bids us do? Why must I think utilitarianism is necessarily true if true at all? I can see why I would conclude this if I already thought S is true and must be respected. I can’t see why I would conclude this otherwise. So this doesn’t offer me an independent route to S and is no good to me.

    No. 3. Your third attempt to get me to see why I must endorse S involves getting me to see that unless I do so I will not be able to make sense of how moral truths guide us. Denying S “makes a nonsense” of the guiding nature of moral truths. But I’m afraid I do not see how. Morality guides, it seems to me, by commanding and commending etc. What’s S got to do with it? In the actual world it seems to me that morality commands me to cultivate kindness and approves of me approving of kindness in others and so on. Why do I have to think that morality would do this in every naturalistically identical possible world before I can think it is doing it here? I just don’t get it, I’m afraid. It seems as odd as thinking that my instruction to the waiter to ‘bring me some coffee’ is an instruction if and only if I would have issued it in all possible worlds in which I am in such circumstances. But the fact there’s a possible world in which I order tea in such circumstances doesn’t somehow mean my actual instruction to ‘get me some coffee’ is not a real instruction incapable of guiding the waiter.

  4. [continued...]

    No. 4. The final route to S that you offer me depends upon a certain principle of universalizabiliy. I mentioned a couple of such principles in my paper. I noted that U2 entails S. But I rejected U2 because it was implausible. You think U2 is implausible as well. But you think my dismissal of U2 depended upon my giving U2 an implausible reading. You offer me a more ‘sensible’ reading. Not as far as I’m concerned. What you offer is a quite different principle. U2 says that if act X is wrong because of its possession of certain naturalproperteies, then any other act that has those natural properties (irrespective of what other natural properties it has) is also wrong. Your ‘sensible’ replacement – U3 – says, if I understand you correctly, that if act X is right, then any act having all the same natural properties as act X is also right. But that’s just S! U3 may well be plausible. I did not argue that S is implausible. But it doesn’t offer me a route to S. It is S! So, at this point I remain a lost cause.

    You say that if someone is determined to hold that S is not true then there isn’t much more you can do. Quite. But I am not quite sure why you think I am ‘determined’ to deny S. I’m not. Again, I did not deny S (as mentioned earlier- I actually think the non-moral version of S is true!). I just pointed out that denying S doesn’t have many of the horrendous consequences many think. If the best metaethical theory turns out to vindicate S, well fine. If it doesn’t, no worries. That’s my position. So I’m hardly ‘determined’ to think S is false.

    But anyway, if all one can do in defence of S is appeal to its supposed self-evidence then that’s quite an important discovery, it seems to me. For despite regular appeals to its self-evidence few actually rest the case for S on that single reed.

    Anyway, I hope what I've said above isn't too exasperating

    [- Gerald Harrison]

  5. Thanks, Gerald, for your detailed response! On the final point, I think it's important that U3 (or, equivalently, S) expresses an independently plausible universalization constraint. There seems something intuitively inconsistent (in a practical, not logical, sense) about adopting different moral attitudes to two non-morally identical situations.

    This relates to the previous point, in that defenders of S's self-evidence are likely to be presuming that the moral truths cannot be arbitrary. But if, as you suggest, the actual wrongness of cruelty is compatible with the possibility of a "world in which morality loves cruelty because of the suffering it causes", then you've turned morality into some arbitrary matter like taste (as you tellingly analogize it to!)

    Of course, this is just the reason many of us find Divine Command Theory a non-starter, so I guess it shouldn't be surprising that people sympathetic to DCT would also be sympathetic to denying the obviousness of S.


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.