Shafer-Landau’s Moral Realism and Supervenience

Here at PEA Soup, Jason Kawall recently raised the possibility that moral realism might be open to a Euthyphro Dilemma kind of objection, which got me thinking about one of those papers that I write and then shelve for some reason or another. In this paper, I take a critical look at Russ Shafer-Landau’s moral realism (as presented in his aptly titled Moral Realism). I figured that perhaps this would be a good venue to discuss its argument.

Shafer-Landau thinks of his view as a kind of non-naturalism, but not of the substance dualism sort, according to which there is some sort of non-natural moral stuff in the world. Rather, he’s a (materialist) substance monist. But he’s also a property dualist: in addition to physical properties, there are irreducibly moral properties as well. This theory is ontologically akin to non-reductive materialism in the philosophy of mind, but Shafer-Landau considers his view epistemologically non-naturalist, since there is no science of ethics.

The worry I have about this view is fairly convoluted, but the basic point is as follows. Shafer-Landau (understandably) wants his view to avoid any “metaphysical profligacy,” a goal that he thinks is accomplished by (1) avoiding any non-physical substances and (2) making moral properties materialistically respectable by having them supervene on physical properties. I have no problem with (1), but I think (2) is suspect, because I think that all non-reductive materialist views fail to confer materialistic respectability upon non-material properties merely by virtue of the supervenience relationship.

The argument for rejecting (2) is a modified version of the argument that Michael Lynch and I put forth in our “The Impossibility of Superdupervenience.” (In addition to considering Shafer-Landau’s specific arguments, the more recent paper has beefed up the conclusion in various ways). Again, the nuts and bolts are as follows. (P1) If the supervenience relation is to transfer materialistic respectability from physical properties to moral properties, then that relation itself needs to be explained, and explained in a materialistically respectable way. (Others, such as Blackburn, Kim, and perhaps most thoroughly Horgan, are responsible for originally levying this explanatory burden; the basic intuition behind it is that even substance dualists have the moral supervene on the physical, so non-reductive materialists need to provide an explanation that presents a more robust supervenience in order to distinguish their views from dualism and make them materialistically respectable.) (P2) Non-reductive materialism does not, in principle, allow for a materialistically respectable supervenience relationship. (C) Thus, non-reductive materialists cannot use supervenience to transfer materialistic respectability from physical properties to moral properties.

Why is (P2) the case? Call the facts about the supervenience relation, whatever the theory in question may say about them, “S-facts.” In short, either the S-facts are themselves some facts of the physical world, which doesn’t do any work in connecting ontologically distinct physical facts to moral facts; or, the S-facts are sui generis, which is not materialistically respectable; or, finally, the S-facts themselves supervene on physical facts, which just generates an infinitely regressing, undischargable, and not materialistically respectable kind of “explanation” of how the moral supervenes on the physical.

As best I can tell, Shafer-Landau offers two rejoinders to the explanatory burden (focusing on Blackburn’s formulation of it). First, he rejects the burden (to, as he puts it, explain fundamental laws of morality) as a question-begging move by his opponents. But I fail to see why this might be question-begging. Each of the different possible ontological positions about moral properties has its own explanatory burdens; just like type-type reduction views must account for multiple realizability, one of the unique burdens of non-reductive views is to account for how ontologically distinct moral and physical properties could be correlated so tightly given the putative fact that they really are ontologically distinct. It would be question-begging to simply stipulate that this burden cannot be discharged; but since there’s an argument that it cannot be discharged (the trilemma in the last paragraph), that claim shouldn’t be question-begging either.

Second, Shafer-Landau does offer something of an explanation for how the moral supervenes on the physical: moral properties just are constituted by (and only by) physical properties. The main problem with this move that Lynch and I stress is that, in order to be a truly non-reductive view, the moral and the physical must be in some sense ontologically distinct. But for Shafer-Landau there is no ontological difference between the two – a fact about the moral world just is a fact about the physical world. So if he’s going to go the constitution-relation route, he can satisfy the burden of explaining the correlations between the moral and the physical, but only at the cost of reducing the moral to the physical (though not, of course, in a type-type way – and Shafer-Landau himself allows that there are other kinds of reductionism).

19 Replies to “Shafer-Landau’s Moral Realism and Supervenience

  1. Very interesting.
    I’m not sure I exactly understand the objection. I think I know what Shafer-Landau would say, though. Try this.
    “There are moral facts that are not identical with any natural facts. There are moral properties that are not identical with any natural properties. Moral facts and properties are, however, constituted by natural facts and properties. I (R S-L) don’t know if constitution is a kind of reduction, and I don’t see why it matters. Constitution is a relation we understand quite well enough — probably better than we understand reduction, which is a bit of philosopher’s jargon.”
    Jeez, once you get Russ started…
    Anyway, I’m certainly not saying that I’m satisfied with this kind of answer; in fact, I am not at all satisfied with it. But I’m wondering whether your worries about it are in the end pretty much the same as mine.
    P.S. Is there a list somewhere of the html tags that are useable in comments for this blog? Very few seem to work. (I tried to blockquote the pretend response from R S-L, and when that didn’t work I tried to indent it, but no luck.)

  2. Jamie, I think that’s a quite reasonable response, actually. But I’ve got three concerns about it. The first is that it’s a bit misleading to characterize such a view as a non-reductive theory, since it’s hard to see how a moral fact is anything beyond a physical fact (which seems as reductive as anything to me), even if the two kinds of fact are not identical. Of course, that’s just about categories, which might not matter much in the end, I suppose. The second worry is more substantive. To my mind, the claim that moral (or mental, or whatever) properties are constituted-by-but-not-identical-with physical properties is substantively problematic. There are reasons, of course, to say that moral and physical properties are not identical (as RS-L goes to great lengths to show). But to then say that moral properties, even though not identical to physical properties, are nonetheless constituted by them, and only by them, strikes me as only a statement of a philosophical conundrum, rather than a solution to one. For now the pressing question is: if X is constituted entirely by Y, how is X not identical to Y (or a subset of Y)? It seems like the proponent of such a view now has a burden to explain this. Finally, in another less-substantive but clarificatory point, Lynch and I point out that if the B (moral) facts are just made up of A (physical) facts, it seems superfluous to bother with supervenience at all, since supervenience is naturally considered and inter-level relation, but this kind of view removes the ontological distinctiveness from the moral and the physical.
    (Incidentally, I did a search on typepad for the html tags, and didn’t come up with much of anything helpful. Maybe another reader could help on this one?)

  3. A question from a lay person. I understand the conceptual utility for nonreductive materialism in the context of philosophy of mind–to account for intuitions about the supposed multiple realizability of mental properties.
    But what is the motivation in this context?

  4. Hi Jamie. Try these. (I hesitate to call them “HTML tags,” but they certainly seem to be the tags Typepad uses.) Just so this comment box does not treat what I’m about to mention as (used) tags, I will use brackets rather than the usual ‘< ' and '>‘ symbols, but you should, of course, use the correct symbols rather than brackets.
    Italics: [em] phrase [/em]
    Bold: [strong] phrase [/strong]
    Underline: [u] phrase [/u]
    Blockquote: [blockquote] quote [/blockquote]
    URL Links: normal HTML tags
    Email links: normal HTML tags
    Hope this helps.
    Dan

  5. Josh,

    To my mind, the claim that moral (or mental, or whatever) properties are constituted-by-but-not-identical-with physical properties is substantively problematic. There are reasons, of course, to say that moral and physical properties are not identical (as RS-L goes to great lengths to show). But to then say that moral properties, even though not identical to physical properties, are nonetheless constituted by them, and only by them, strikes me as only a statement of a philosophical conundrum, rather than a solution to one.

    I agree. (And thanks, Dan, as you see I’m using the “blockquote” tag.)

    For now the pressing question is: if X is constituted entirely by Y, how is X not identical to Y (or a subset of Y)? It seems like the proponent of such a view now has a burden to explain this.

    Here’s what I think.
    In the case of objects, especially material objects, we have a pretty clear intuitive idea of what constitution is. Furthermore, it is at least not obvious that constitution implies identity (as we know from some of the oldest puzzles in metaphysics — there are reasons to deny that a statue is identical to the piece of clay that constitutes it). But it is not at all clear that this intuitive idea carries over to properties or facts.

    Finally, in another less-substantive but clarificatory point, Lynch and I point out that if the B (moral) facts are just made up of A (physical) facts, it seems superfluous to bother with supervenience at all, since supervenience is naturally considered and inter-level relation, but this kind of view removes the ontological distinctiveness from the moral and the physical.

    I don’t really see this. Maybe my problem is that the antecedent of your hypothetical is pretty obscure to me — what does it mean to say that this fact is “just made up of” those facts? I take it that you, too, find this pretty obscure.

  6. Josh, that’s interesting, your reaction to RSL’s view is *exactly* the one I had. To put it Horganly, he doesn’t meet the demands of the material world because he can’t provide for superdupervenience, that is, a supervenience relation whose holding is not only stated but also *explained*.
    Note, however, that his strategy in Ch.4 of MR is to first offer constitution as the explanation of supervenience and then, when presented with the need to explain constitution, claim that he doesn’t have to. (That’s the opposite order from the way you presented his strategy in the post.)
    One problem is that, as Jamie points out, we don’t have a clear notion of what property constitution might come to – and Russ doesn’t help us much in the book. Another problem, which I think is what you have in mind, is that we are still left with a metaphysical relation (now constitution rather than supervenience) that is inexplicable and apparently sui generis.
    We can organize the latter problem as a dilemma. Either (property) constitution entails identity or not. If it does, then RSL’s MR becomes a reductive, naturalist MR. If it doesn’t, then the holding of a constitution relation between natural and moral facts still requires explanation.

  7. Uriah, I don’t really buy the second horn of your dilemma.
    If someone told me that the part-whole relation had to be explained in terms of something else lest it be physicalistically suspect, I would be very unimpressed. That’s not because I think the part-whole relation is physical. I don’t know what to think about that. It’s because it is obviously perfectly respectable on its own, whether physical or not, and probably more explanatorily basic than anything that could be brought to bear on it in any case.
    Russ says some other things about the moral-natural relation besides that it’s constitution. He also says that natural properties realize moral ones, and that it is the relation between determinable properties and the determinates. I really don’t know what to think of this. I guess at this point the worry that Josh (and Michael Lynch) raise seems very salient: how on earth could there be a natural property that was the determinate of a non-natural property? Has Russ sold out to naturalism after all?

  8. I was thinking of the problem very close to how Uriah put it. So let me try to amplify the second horn of Uriah’s dilemma in response to Jamie’s latest comment.
    I agree with Jamie that a constitution relation is very intuitive, and perhaps in some sense explanatorily basic and in some sense “respectable.” However, note that the job isn’t merely to account for(simpliciter) the relationship, but to do so in a way that confers materialistic respectability upon the higher-order (in this case moral) properties. So we would have to ask whether a constitution (or, for that matter, realization) relation is materialistically kosher, even if the relation is intuitive. Consider an analogy with conceptual truths. Some have argued that it’s a conceptual truth that the higher-order properties supervene on the base properties. But why should we think that conceptual truths – even if intuitively obvious – have any place in a materialistically kosher ontology? There are, of course, hard questions about whether conceptual truths, like mathematical and logical truths, can fit into such an ontology.
    Furthermore, let’s grant for a moment that facts about such relations — S-facts — have a place in a materialistic ontology. The trilemma still holds, if the moral and physical facts are ontologically distinct. For we’ll want to know where to locate the S-facts. Either: the S-facts will themselves be physical facts, in which case it’s hard to see how this does any work in providing connective tissue between the moral and the physical, rather than just being a subset of the physical that still needs to be connected to the moral. Or: the S-facts will not be physical, in which case they’ll be either sui generis or somehow tethered to physical facts. Non-physical sui generis facts, contrary to our supposition here in this paragraph, hardly seem materialistically respectable. (In fact, they seem Moorean.) And if the S-facts are somehow tethered to the physical, then we’ll need to account for that further relation, which threatens a regress.
    Incidentally, in my last comment, to reply to Jamie’s question, I was just using “is made up of” as synonymous with “is constituted by.”
    (And, I agree with Jamie that it’s hard to see how it’s non-naturalism to hold that natural properties realize moral properties. But RSL would, I think, hold that while ontologically it’s all natural, it’s still epistemologically non-natural, since there’s no science of ethics. For what it’s worth, I find this to be a somewhat idiosyncratic notion of non-naturalism.)

  9. Jamie, my worry about constitution was not that it required explanation because it was unintuitive or unrespectable and would therefore remain mysterious if unexplained. Let me try to explain better my worry.
    Let’s distinguish two kinds of relation. There are, on the one hand, constitutive or “internal” or otherwise non-contingent relations, such as identity, part-whole, etc. On the other hand, there are “external,” contingent relations, such as causation, spatial proximity, etc. The former are such that their holding or instantiations need not and in fact cannot be *explained*. But the latter are such that their holding or instantiation does call for explanation.
    (Check it: (1) All contingent facts are presumably explicable; (2) The instantiation of a contingent relation is a contingent fact; Therefore, (3) The instantiation of a contingent relation is presumably explicable. So it calls for explanation.)
    Now, constitution seems to be a contingent relation: the clay constitutes the statue in the actual world, but in another possible world where the statue is destroyed through molding of the clay beyond recognition, the clay does *not* constitute the statue. So the clay’s constituting the statue is a contingent fact, and as such calls for explanation. Not in the sense that if unexplained it would remain mysterious, but in the same sense that all contingencies are explicable.
    To just state that some property constitutes another is therefore insufficient to “meeting the demands of the material world.” You must also explain how come the first constitutes the second.

  10. Hmm!
    I’m not sure, you might have something there. But the thing is, the only clear examples we have of ‘fact constitution’ are not examples of a contingent relation at all. I’m thinking of the fact that my socks are red being constituted by the fact that they are scarlet.
    Does that mean the friend of fact constitution is excused from your challenge? Does it mean that the relation in question is not constitution, since constitution is contingent?

  11. Josh,
    I’ve read your articles as well as a few of the other pieces from the reasonably manageable superdupervenience literature and was wondering if you had any thoughts about a strategy I’ve been toying around with for establishing materialistic respectability without superdupervenience.
    Russ grants that moral properties supervene on the non-moral, material properties (assuming the mental supervenes on these as well) so let me likewise assume supervenience. Arguably, being a supervenient property is itself an essential property of properties. With these two assumptions in place, moral properties must be properties that supervene on some property or other.
    The way that I’ve seen the worry about materialistic respectability of merely supervenient properties introduced is that there may be materialistically disreputable bases for properties that happen to have materialistically respectable ones. There might be other ways of bringing out the worry, but in phil mind, this seems to be the important way.
    What the materialist should do now is argue that no materialistically disreputable basis could ground moral properties. They shouldn’t argue for this by appeal to the general truth that if P is a supervenient property that has a material property M as its actual basis it couldn’t have some non-material property N as its basis as I don’t see why we should think this is a general truth. Rather, they could appeal to special features of normative properties themselves to show that nothing but a material property could be the base property on which normative properties supervene. How?
    We can begin by imagining a world that has the same purely materialistic profile as ours that has in addition some non-material properties. We then have to ask whether this difference could introduce a normative difference. Could the second world be more or less valuable than the first? Could there be different duties in the first and second worlds? One way to argue for the ‘no’ answer (this is sketchy) is that if this second world is just like ours over time, then it doesn’t seem that the non-material difference makes a causal difference and without such a difference, we’d have no reason to think that things could be more or less valuable than they actually are or that what we should do could differ. It doesn’t seem that even in ideal conditions could we be in a position to know such things. If you think that it doesn’t make sense to say that there are normative truths that are essentially unknowable, this would suggest that in moving from a purely material world to a mixed world, we haven’t changed the normative profile of the purely material world.
    This is far from complete. There hasn’t been an argument that in a purely non-material world there couldn’t be normativity, but I think it is a start. What is interesting is that this strategy suggests that materialistic respectability might not depend on superdupervenience. It also suggests that Russ could be wrong in thinking that by taking on supervenience he isn’t committed to some form of naturalism.

  12. Clayton,
    Thanks for the comment (and I’m glad to hear that someone’s reading our article!). However, I’m not sure that I buy the issue as you’ve set it up. Now I’m not totally sure that I follow you, but I take it that you want to show that only material properties could be the base properties upon which moral properties supervene. I’m not sure whether that’s true, but, more to the point, I don’t take that to be the challenge. The way the challenge is put out there by, say, Horgan, goes like this, if memory serves. GE Moore included what are usually taken to be materialistically non-respectable facts in his non-naturalistic moral ontology; but, he of course also held that the moral supervenes on the physical – that seems like something of a truism that even a substance dualist would accept. So if a non-naturalist can have the moral supervene on the physical (as seems possible), and if the non-reductive materialist needs to distinguish her view from non-naturalism (as an ontological thesis, rather than an epistemological one, as RSL construes it), then merely stating the supervenience thesis isn’t enough for the non-reductive materialist. She needs a kind of supervenience that confers materialistic respectability upon the higher-order properties, i.e., superdupervenience. This strikes me as a different task than either stating the supervenience thesis or claiming that if the moral supervenes on anything, it must supervene on the physical, because, again, that’s something that Moore would have accepted.

  13. You’re right, Jamie, that even if object constitution is contingent, the paradigmatic cases of fact constitution seem to be necessary. Consider: the fact Jane is a pianist constitutes the fact that Jane is a musician. Here, there is no possible world in which a pianist is not a musician. But there are also other cases of fact constitution. For instance: the fact that I am in Tucson constitutes the fact that I am in Arizona, but there *is* a possible world in which Tucson is not in Arizona (e.g., a world where the Mexicans conquer and annex Tucson).
    This indeed raises some interesting questions. Are there in fact *two* relations of fact constitution, a necessary one and a contingent one? Would appeal to the necessary relation collapse RSL’s view to naturalism? I think much of the difficulty here has to do with the trickiness of the constitution relation when applied to things other than material objects. That’s why I think that unless a thorough account of fact and/or property constitution, appealing to it may have an air of hocus-pocus. That’s also a liability on Russ’s position.

  14. Good examples.
    My gut reaction is this — I reserve the right to disown any or all of it later!
    Sometimes when we say that X constitutes Y, we mean that X completely constitutes Y, and in such cases the relation is necessary. (Just hang on, don’t jump straight to the counterexamples! I’ll open the loophole myself in a minute.)
    Sometimes, we are instead stating only the salient part of the constitution, leaving the rest of the conditions as background. Compare statements about causes, or explanation in general. In those cases, the relation is contingent, for the obvious reason that some of the (jointly sufficient) conditions have been left unstated.
    This seems more satisfactory when applied to your Tucson/Arizona fact-constitution than it does when applied to typical object-constitution examples, though. It seems ok to me to respond to your example by saying, “Well, really your being in Arizona is constituted by your being in Tucson plus certain facts about the political history of the region.” The parallel move for object constitution would be to say that it isn’t just the clay that constitutes the statue, but the clay plus its being in a certain shape… but that doesn’t sound right, it sounds like we are changing the topic (from object constitution to some kind of analysis of a concept).
    Just one more related issue:
    in a typical ‘ethical fact constitution’ case, the relation probably isn’t necessary at all. Suppose I lie to you, thereby acting wrongly. My acting wrongly is constituted by my intentionally telling you a falsehood. But there is no necessary connection between telling a falsehood intentionally and acting wrongly: many intentional falsehoods are morally permissible, even obligatory. This seems to fit the “incomplete constitution” mold, but of course its controversial whether it really does, since Particularists will say that there isn’t any completed counterpart.

  15. I like the complete/incomplete consitution story, Jamie. It sounds pretty plausible on the face of it.
    So: if I am right about the explanatory burden that attaches to merely contingent relations, then Russ must show that the relation between moral and natural facts is that of *complete* constitution (which is non-contingent). (If the relation was indeed determinable-determinate, that would do the trick.) But as your lie/wrong case suggest, the prospects for showing that are not promising.

  16. Josh,
    Thanks, that was helpful. In the seminar where we discussed RSL’s book, we had talked about Moore and there was some general agreement that while there were a lot of writers who thought that moral properties were non-natural but supervenient properties or that there were emergent properties, either their arguments were wanting or these weren’t real possibilities. At some points I vaguely recall RSL saying something like ‘Well Moore was a non-naturalist who accepted supervenience so supervenience isn’t enough for naturalism’, but there has to be more to it than that.
    It was that point in the discussion that people started arguing that since there could (conceivably) be angelic virtue, divine wrath, and ghost pain, it looked like there was nothing ‘in’ a moral property which required it to have a material base and that this caused problems for materialism and naturalism about the normative.
    Anyway, I don’t have my copy of RSL’s book here, but I just remember that few seemed to moved by the observation that Moore accepted supervenience while rejecting materialism, but maybe there was something more to that that we had missed.

  17. Gift Basket

    Sweet Blessings, a new Christian-based online shop featuring cookie bouquets, candy bouquets and gift baskets, opens with a campaign to donate a portion of all profits to Habitat For Humanity. The devastation of hurricanes Katrina and Rita, while not a…

  18. Joshua Glasgow hits the head of the nail when he writes:
    “Shafer-Landau does offer something of an explanation for how the moral supervenes on the physical: moral properties just are constituted by (and only by) physical properties. The main problem with this move that Lynch and I stress is that, in order to be a truly non-reductive view, the moral and the physical must be in some sense ontologically distinct. But for Shafer-Landau there is no ontological difference between the two – a fact about the moral world just is a fact about the physical world.”
    Behind Shafer-Landau’s label “non-naturalist moral realism” hides a view that is very hard to pin down. Arguably it is not an explanatory view at all but rather nothing more than a description of the surface of the moral, a phenomenology.
    Here is why:
    a) Naturalism is defined, not ontologically, in terms of e.g. existence in space and time, but epistemologically, in terms of what is postulated by a scientific theory. In other words a non-natural property is a property not featured in a perfected scientific theory.
    A first result of this definition is that being a stone or a bicycle is also to possess a non-natural property. The second result is that SL’s view is compatible with all views accepting the autonomy of ethics -from non-reductive naturalists to quasi realists.
    -But what about his insistence about the necessity of ADDING moral properties to our ontology?
    b) SL’s supervenience claim is that moral properties are exhaustively constituted by non-moral properties. This amounts to token reduction. So if by reduction and naturalism you mean token reduction, his view is reductive and naturalist.
    ok. But what about those moral properties not type reducible to non-moral properties; what are they? Do they e.g. play any causal explanatory role? Here SL’s answer is, at first waffling, but ultimately: no. Moral properties are not primarily explanatory, etc. properties, if they were then they might figure in a science, but they don’t. Whatever causal power they have are inherited from their natural supervenience base.
    So; moral properties are exhaustively composed by natural properties(“descriptive” SL calls them confusingly; if you are a realist don’t you think that moral language is descriptive?) and they play no ineliminable causal explanatory role.
    This equals regarding moral properties as thin properties, rather than robust properties. Thin properties can be understood as, in Blackburns words, shadows of predicates. This sense of “property” is compatible with any view of moral ontology from Stevenson to Gibbard.
    So, Shafer-Landau’s view is not, behind the surface, a distinct metaphysical view of real moral properties.
    And -what do you thing about this argumet- if he really believed in real moral properties existing out there, why doesn’t he think that moral principles are inductive generalizations from our experience of these facts, rather than apriori?
    Since his moral properties are so thin, the real burden of his realism rests on his account of why certain moral standards can be true objectively. Here, trust me, you will find nothing more than an appeal to the fact that we do find some moral standards to be obvious. -But a) many standards aren’t obvious, and b)why is finding some standards obvious a proof of their objectivity?

Comments are closed.