Dreier on the supervenience argument against robust realism

Blackburn has put forward a very influential argument against moral realism, which turns around the supervenience relation. Dreier’s version of the supervenience argument has a narrower target. It should be effective against non-reductive, robust moral realism, by revealing an explanatory cost that non-robust, naturalistic forms of moral realism do not have. The present paper argues that naturalist realism can explain the necessity involved in the supervenience relation only by assuming an unrestricted application of Boolean closure principles. Pending an independent argument to the effect that even an indefinite and perhaps infinite combination of natural properties is itself a natural property, robust realists are entitled to reject Dreier’s attempt to build the Boolean closure clause into the very formulation of strong supervenience. The conclusion is that moral naturalists cannot claim a clear advantage over robust realists with regard to the challenge posed by the supervenience argument. A final section makes some remarks on how moral expressivism could answer the supervenience argument.

1 Universidade Federal do Rio de Janeiro.Largo de São Francisco de Paula 1, 20051-070, Rio de Janeiro, RJ, Brasil.E-mail: wilsonpessoamendonca@gmail.comDreier on the supervenience argument against robust realism Wilson Mendonça 1 I Notoriously, Simon Blackburn (1984Blackburn ( , 1985) ) has put forward a very influential ar ument against moral realism, which turns around the supervenience relation.James Dreier's version of the supervenience ar ument (Dreier, 1992(Dreier, , 2015, n.d.) , n.d.) draws on Blackburn's original version, but has a nar ower target.While Blackburn aims to show that the supervenience of the moral on the natural poses a pro lem for a l brands of moral realism, thereby sup orting moral antirealism, Dreier claims that the supervenience ar ument is effective only against non-reductive moral realism.In Dreier's view, the supervenience ar ument reveals an explanatory cost of what he ca ls "robust realism" , a cost that naturalistic forms of moral realism do not have.(They may have other costs and at the end of the day be in a worse position than non-reductive realism, but this is beyond the grasp of the supervenience ar ument.) At least in part, this difference is due to the fact that Blackburn makes use of a relatively va ue notion of the supervenience of the moral realm on the natural realm, whereas Dreier has in mind a much more precise formulation of what the supervenience relation is.Dreier's prefer ed formulation is based on Jaegwon Kim's formula of Strong Supervenience of mental properties on physical properties (cf.Kim, 1984).It is the fo lowing: α is the set of moral properties and β is the set of natural, "descriptive" properties.SS then says that whenever a moral property F is instantiated, then necessarily a natural property G is instantiated, which in turn necessarily determines the instantiation of F. Although this is strictly no part of SS, Dreier takes each of the sets α and β as closed under Boolean operations.That is to say: if F and F´ are both moral properties, so are the combined properties of being both F and F´, being either F or F´, not being F, and so forth.This ap lies also to any natural properties G and G´: arbitrary combinations of these properties with the connectives of propositional logic are eo ipso natural properties.
It is not clear why Dreier assumes Boolean closure in the general explanation of the strong supervenience of the moral on the natural.Dreier (1992, p. 23) asserts, without going into much detail, that the analysis of supervenience claims requires Boolean closure of the relevant sets of properties.This would be no pro lem if we had only non-naturalist moral realism in view.For non-naturalist moral realism is the claim that the set of moral properties, on the one hand, and the set of natural properties, on the other hand, are entirely disjoint.Dreier recognizes that the Boolean closure condition may be pro lematic in the context of naturalism, "for many naturalists maintain that a class of properties closed under Boolean combination and containing no moral properties would have to be very sparse indeed" (Dreier, 2015, p. 5).Anyway, he imposes the closure condition and tries to evade the pro lem by setting the focus on non-naturalism.
However, as we wi l see, he cannot now get the intended result that robust realism is not a le to explain SS, hile naturalist realism is.Even if it is true that robust realism has a cost regarding the explanation of SS, naturalist realism can ar ua ly explain SS only by assuming unrestricted ap lication of Boolean closure principles.Pending an independent ar ument to the effect that even an indefinite and perhaps infinite combination of natural properties is itself a natural property, robust realists are entitled to reject Dreier's attempt to build the Boolean closure clause into the very formulation of strong supervenience.If it is reformulated along these lines, the supervenience ar ument leads to an impasse: neither robust realists nor moral naturalists can (as of yet) discharge the burden of explaining the strong supervenience of the moral on the natural.The next sections wi l get into the details of why this is so.

II
If true at a l, SS is an analytical, conceptual truth.The outer box expresses analytical necessity.If it stands in need of an explanation, SS, as any other conceptual truth, requires an analytical, conceptual explanation.This cannot be the notion of explanation invoked by the supervenience ar ument.After a l, adopting the metaphysical attitude of non-naturalism does not make it impossi le for a philosopher to find analytical explanations for conceptual truths-any more than being a moral naturalist makes it easier for a philosopher to explain conceptual facts.Thus, the ability of finding explanations for SS, in this sense of explanation, namely, conceptual explanations, cannot be the test that should reveal the superiority of forms of naturalist realism is-à-is non-naturalism.What then is the sense of explanation in which, according to the supervenience ar ument, the non-naturalist cannot, while the naturalist can, explain the relation of supervenience?
From the point of view of the supervenience ar ument, what stands in need of explanation is the necessary connection represented in SS by the inner box.This is metaphysical necessity.Putting the pieces together: SS is a conceptual truth about a metaphysical necessity.And the cha lenge is to say whence comes the metaphysical link between the supervenient moral property F and the subvenient natural property G.If it is true that the natural property G neces itates the moral property F, what grounds this necessitation?What accounts for the strong metaphysical determination of the moral world by the natural world?This is the operative question in the supervenience ar ument.Now, it seems that Dreier makes two assumptions here, both of which can be questioned.The first assumption is that it makes sense to look for a genuine explanation of the metaphysical fact of necessitation expressed inside SS.The second assumption says that there is no reason to doubt that the naturalist is a le to deliver such an explanation.
Let us ask: What is so crazy in the "fundamentalist" reply to the effect that the necessitation of Fs by Gs is a brute metaphysical fact, a very basic feature of the universe, which in the end can only be registered, but not explained?Explanations must come to an end somewhere.And why not here?Dreier does not consider this fundamentalist move.Admitte ly, in the context of the present debate it may look a bit unsatisfactory, perhaps arbitrary in a sense.But it is certainly not obviously a hoc, as it is a general fact that many metaphysical necessities have no explanations in any substantial sense.Think of He erus being Phosphorus or heat being molecular motion.
However, even if the fundamentalist move proposed on behalf of non-naturalists proves in the end to be arbitrary, it remains to be seen whether the putative explanations proposed by moral naturalists avoid a similar arbitrariness.Take, for instance, the explanation of the necessitation fact offered by the analytic naturalist who claims that the moral chara erization of an object fo lows strictly, pace Moore's open question ar ument, from its natural description along with the meaning of moral ter s (cf.Jackson, 1998, chapter 5).Then it is absolutely impossi le that a difference in the moral facts goes unaccompanied by a difference in the natural facts in a l possi le worlds, exactly as SS claims.This means that the "metaphysical glue" between the natural and the moral is ultimately accounted for by an analytical glue, which is expressed by meaning identities eventua ly involving Ramseyification and other very complex devices.The explanation would end here, in facts about the meaning of moral words.Isn't it "arbitrary", in a certain sense, that we must stop here, that we cannot ask for the grounds of the facts about meaning?I take it that Dreier thinks this is not necessarily so.He even admits, towards the end of Dreier (n.d.), that "there may be no true explanation" of the metaphysical necessitation of moral properties.Most of the time, however, he assumes (i) that such an explanation is needed.(And he ar ues that the non-naturalist cannot deliver the needed explanation.)Further, he assumes (ii) that there is no unsurmounta le obstacle for the naturalist who is trying to give a genuine, non-arbitrary explanation of the facts of supervenience.
Before considering what Dreier has to say e ecia ly about the second assumption, I wi l comment on his grounds for believing that SS is true.

III
There are, of course, alternative formulations of the supervenience relation.Both Blackburn and Richard Hare (1952), for instance, thought that the form of supervenience that is relevant for the metaphysics of morality is Weak Supervenience.This is what you get from SS by simply deleting the second box, thus making the link between the moral and the descriptive properties contingent or, in other words, by substituting mere determination for necessitation: Dreier (n.d.) discusses the adequacy of WS under the heading "Moral Contingency." He fina ly rejects the idea that WS captures the most distinctive claim of moral realism.If WS were a l that can be said about the relation between moral facts and natural facts, moral realism would entail a very counterintuitive form of the moral luck thesis: there would be possi le worlds in which torturing innocent people just for fun is mora ly right.Of course, SS entails WS, but not vice versa.Since Dreier takes SS to be true, he also takes WS to be true.What he rejects is that we can forma ly capture moral realism's fu l range of claims without an inner necessity operator.Dreier is right here.Even radical non-naturalistic versions of moral realism (for instance, Moore's version) require more than the contingent determination of the moral by the natural.
There is at least another formulation of strong supervenience, which Dreier quickly considers and also rightly dismisses, namely, Analytical Supervenience: This interprets both modalities involved in the relation of supervenience as cases of conceptual or analytical necessity.
Plausi ly, the fact that the instantiation of a moral property must be accompanied by the instantiation of a natural property is explained by the constitutive features of our concept of a moral property and its relations to the concept of natural properties.This means that the outer modality is analytical or conceptual.But it is in this sense even more plausi le that the necessary determination of the moral by the natural, as encoded by the inner necessity operator, concerns the very nature of the cor esponding properties themselves, not the concepts refer ing to them.If this is so, the inner box cannot express analytical necessity and AS must be rejected.
The ar ument dealing with WS shows that the supervenience relation between moral properties and natural properties is dou ly modal, whereas the ar ument whose target is AS reveals that the necessity encoded inside the supervenience formula is metaphysical.The conclusion, as Dreier repeate ly asserts, can only be: "SS is true." From the viewpoint of robust realists, these ar uments are convincing.Thus, robust realists can be rightly viewed as committed to the truth of SS.It should be noticed that these ar uments are completely silent on the issue of the Boolean closure condition which Dreier closely associates with the notion of strong supervenience.So far at least the robust realists' commitment to the obtaining of a strong supervenience relation between the moral and the natural cannot be extended to the thesis that each of the sets of moral properties and natural properties is closed under Boolean operations.The significance of this point wi l emerge presently.

IV
Turn now to the reason why robust realism fa ls prey to the supervenience ar ument.The decisive passage in Dreier (n.d.) reads thus: But according to Robust Realism, the moral properties are fully distinct, and distinct in kind, from the non-moral ones.(That is why we confidently speak of the 'non-moral properties'.)So, Robust Realists cannot explain the necessary connection by the usual expedients of analysis, reduction, or identity.How, then, are they supposed to explain the necessary connection?They cannot.They have no explanation.
Dreier seems here to infer from the principled unavailability of explanations by way of analysis, reduction, or identity outside the context of naturalism that robust realists have no explanation for the relevant necessary connection.Presuma ly, he wants to say that analysis, reduction or identity are the only possi le modes of explanation here.These are indeed incompati le with the robust realists' assumption of disjointness of the natural (the non-moral!) and the moral domains.But who is to say that there may not be an explanation that does not proceed by way of analysis, reduction or identity?It is anyway simply wrong to infer from the fact that robust realists cannot take recourse to "the usual expedients of analysis, reduction, or identity" that they cannot explain the necessary connection of the natural to the moral.This is admitte ly a minor point in the critical evaluation of Dreier's position.A more important point is as fo lows.When ap lica le, the usual expedients of analysis, reduction, or identity do show that the relevant sets of properties are not distinct in kind.They vindicate reductionism in ethics by entailing the thesis that the set of moral properties is in fact a subset of the set of natural properties.Moreover, if moral properties are (identical to) natural properties, then the explanation of the necessary link between the moral and the natural fo lows automatica ly: the moral strongly supervenes on the natural because the moral is included in the natural.It could then seem that SS (together with the usual expedients of analysis, reduction, or identity) is sufficient for the falsity of robust realism.But this is not so.Beyond SS (and the usual expedients of analysis, reduction, or identity), the naturalistic explanation of supervenience also requires at least the assumption of the Boolean closure of natural properties, which, so far as we could see, is independent from SS.
What fo lows from SS alone is a claim to the effect that any moral property F is coextensive to a huge (as Jackson, for instance, recognizes: infinite) disjunction of natural properties G i , each of which is sufficient, in each possi le situation, for the instantiation of the multiply realiza le property F. To get a naturalistic reduction out of this, one needs to assume (i) that the huge disjunctive combination of natural properties G i is itself a natural property and (ii) that necessarily coextensive properties are identical.The first assumption is a case of the general Boolean closure clause which Dreier is wi ling to build into his chara erization of strong supervenience.
Both assumptions have been vehemently discussed in the relevant literature (for instance, in Van Cleve, 1990 andParfit, 2011, p. 296f.).Whether or not they can be ultimately justified is not important for the present purposes.The crucial point is that the commitment to the truth of SS without the Boolean closure is not sufficient to show that naturalism is better placed than robust realism when it comes to explaining the facts of supervenience.
The outcome is that we can fo low Dreier in construing the supervenience ar ument as cha lenging the moral realist to explain the necessary connection that is part of SS.We can also agree with Dreier, at least for the sake of the ar ument, that the robust realist fares ba ly at trying to meet the challenge.However, we have seen that naturalistic explanations by the usual expedients of analysis, reduction, or identity are committed to claims that are not entailed by, and go far beyond the content of, SS.By relying on these pro lematic claims, naturalism's ap roach to the necessitation of the moral by the natural-maybe better: usual naturalistic explanations of the necessary connection-cannot claim a clear advantage over robust realism with regard to the cha lenge posed by the supervenience ar ument.

V
How could a defender of moral ir ealism react to the supervenience ar ument?To which form of supervenience are moral ir ealists committed?None of the above ecified forms (SS, WS, AS), as a l of them quantify over (and are therefore committed to) moral properties, which are inexistent to the eyes of the ir ealist.We can sti l take SS as a starting point and remedy this by replacing a l normal second-order quantifiers with substitutional quantifiers.α is now the set of moral predicates and β is the set of natural, uncontroversia ly descriptive predicates.
This means that the form of supervenience that the irrealist can be cha lenged to explain turns out to be a dou ly modal, asymmetric dependence relation between the vocabularies.What remains to be chara erized (and explained) by the ir ealist is the sort of necessary connection represented in IS (the substitutional modification of SS) by the inner box: (A l second-order quantifiers are substitutional.) Here we can profita ly draw on a distinction proposed by James Kla ge (1988), namely the distinction between "ascriptive" and "ontological" supervenience.The latter is a real relation between worl ly (sets of) properties.The former is some sort of order e impose on moral discourse with a certain purpose.According to the ascriptive view of supervenience, F-judgments are necessarily tied to, and supervene on, G-judgments, but this is not bac ed by underlying relations in the objecti e world.Rather, the "metaphysical glue" so cherished by the realist makes place here for a pragmatic link.If moral discourse should uide us effectively, then we cannot a low that mora ly different ways of chara erizing things go with identity of the underlying natural descriptions.In other words, we are pragmatica ly coerced to re ect strong supervenience, on pain of nu lifying the pra ical point of moralizing ta k.This is, in extremely compressed form, the distinctive claim of moral expressivism paradigmatica ly associated with Blackburn.It may have a l sorts of pro lems, which cannot be a dressed here.My contention is only that it passes the test of the supervenience ar ument, provided IS is substituted for SS.