Terence Horgan and Mark Timmons

J. L. Mackie argued that if there were objective moral properties or facts, then the supervenience relation linking the nonmoral to the moral would be metaphysically queer. Moral realists reply that objective supervenience relations are ubiquitous according to contemporary versions of metaphysical naturalism and, hence, that there is nothing especially queer about moral supervenience. In this paper we revive Mackie's challenge to moral realism. We argue: (i) that objective supervenience relations of any kind, moral or otherwise, should be explainable rather than sui generis; (ii) that this explanatory burden can be successfully met vis-a-vis the supervenience of the mental upon the physical, and in other related cases; and (iii) that the burden cannot be met for (putative) objective moral supervenience relations. What is the connection between the natural fact that an action is a piece of deliberate cruelty say, causing pain just for fun and the moral fact that it is wrong? It cannot be an entailment, a logical or semantic necessity. Yet it is not merely that the two features occur together. The wrongness must somehow be "consequential' or "supervenient'; it is wrong because it is a piece of deliberate cruelty. But just what in the world is signified by this 'because'? (J. L. Mackie, 1977, p. 44) M o r a l r ea l i sm is the doc t r ine tha t t he re a re m o r a l fac t s , and tha t these facts a re objec t ive r a the r than be ing s o m e h o w cons t i tu t ed by h u m a n bel iefs , a t t i tudes , o r convent ions . ~ This view is increas ing ly p o p u l a r in r ecen t ph i losophy . M a n y ' new-wave ' m o r a l real is ts also con tend , in o p p o s i t i o n to the r ece ived ph i losoph ica l op in ion t h r o u g h o u t much of this cen tu ry , tha t m o r a l r ea l i sm can be m a d e to square with a b r o a d l y na tura l i s t ic , or mater ia l i s t i c , s tance in me taphys ic s and ep i s t emology . 2 J. L. M a c k i e c l a imed o the rwise , on the basis of wha t he ca l led the " a r g u m e n t f rom quee rnes s " . H e m a i n t a i n e d tha t ob jec t ive mora l p r o p e r t i e s o r facts, and ob jec t ive supe rven ience re la t ions b e t w e e n nonm o r a l and m o r a l p r o p e r t i e s o r facts, wou ld be dec ided ly quee r , f rom the pe r spec t i ve o f any sensibly na tura l i s t i c ph i losoph ica l pos i t ion tha t gives p r o p e r p r i d e o f p lace to science. H e set for th the quee rnes s a r g u m e n t in two forms: one focusing u p o n mora l p r o p e r t i e s p e r se, and the o t h e r focusing u p o n the supe rven i ence re l a t ion b e t w e e n n o n m o r a l and m o r a l p rope r t i e s . 3 But the new-wave m o r a l real is ts have m e a n w h i l e Synthese 92: 221-260, 1992. © 1992 Kluwer Academic Publishers. Printed in the Netherlands. 222 T E R E N C E H O R G A N A N D M A R K T I M M O N S produced plausible-looking replies to both versions, thereby shifting the burden of proof to those who sympathize with Mackie. Our project here is to revive and rejuvenate the form of the queerness argument concerning supervenience. Briefly, our argument will go as follows. First, Mackie's allegation that there would be something queer about a genuine, objective, supervenience relation of the moral upon the nonmoral reveals an important, and insufficiently appreciated, burden which must be shouldered by the metaphysical naturalist whenever he invokes supervenience. The burden is to argue that there are naturalistically acceptable explanations for all specific supervenience facts. Second, it is plausible that this burden can be successfully met vis-h-vis the supervenience of the mental upon the physical, and in other related cases of supervenience frequently discussed in recent metaphysics and philosophy of mind. But, third, there are strong reasons to believe that the burden cannot be met for (putative) supervenience relations between objective nonmoral facts or properties and (putative) moral facts or properties. Hence, it appears unlikely that moral realism can be accommodated within a broadly naturalistic philosophical Weltanschauung after all. Before proceeding, let us make some preliminary remarks about metaphysical naturalism. First, the term 'naturalism', as we will use it here (and as new-wave moral realists use it and related terms), should not be understood as implying a commitment to any specific 'reductionist' philosophical program. The doctrine traditionally called 'metaethical naturalism', which asserts that there are meaning equivalences between descriptive terms or statements and moral terms or statements, is only one species of metaphysical naturalism in the broad sense. (Newwave moral realists explicitly disavow this specific doctrine, while also claiming to embrace naturalism in metaphysics and epistemology.) Second, in the course of the paper we will say something about how we think metaphysical naturalism, in the broad sense, should be understood. In particular, we will argue that even a broad and nonreductive naturalism carries explanatory demands, involving supervenience, that often have been overlooked in recent philosophy. Third, we recognize that some philosophers are disinclined to accept metaphysical naturalism not even the 'laid back', California style, naturalism we will characterize below. Some of these same philosophers, we realize, are also inclined to accept moral realism. But we emphasize that even anti-naturalist moral realists might well find cornTROUBLES ON MORAL TWIN EARTH 223 pelling our argument that moral realism does not comport well with metaphysical naturalism. 1. Q U E E R N E S S A R G U M E N T S A N D I N N O C E N C E BY A S S O C I A T I O N Mackie's first argument focuses directly upon putative moral properties, moral facts, and moral relations. He writes: If there were objective values, they would be entities or qualities or relations of a very strange sort, utterly different from anything else in the universe. (Mackie, 1977, p. 37) 4 Here the alleged queerness stems from what such a fact, property, or relation would have to be like. Mackie claims that morality is essentially practical, and that one aspect of its pragmatic nature is this: if there were any objective moral facts or properties of the kind posited by ordinary moral thought, then they would have to be intrinsically motivating or reason-providing, in the sense that mere recognition of the presence of such a property would motivate or provide a reason for acting appropriately. He maintains that the idea of a property which "has to-be-pursuedness somehow built into it" (ibid., p. 40) is, from the perspective of a thoroughgoing metaphysical naturalist, ontologically queer. But the moral realist can reply by denying Mackie's contention that moral properties must be intrinsically prescriptive. This claim is a version of what is called 'internalism', a doctrine which many philosophers consider implausibly strong and not a part of common moral thought at all. 5 David Brink, for example, remarks that not only is internalism dubious, but "in fact, it seems unlikely that any belief so recherche could be part of common sense moral thinking" (Brink, 1984, pp. 11415). Even if the moral realist can dodge Mackie's first queerness argument in this way, however, there still remains a question about the supervenience relation that allegedly holds between certain natural facts or properties, and certain moral facts or properties. Mackie's second queerness argument, quoted as our epigraph, articulates this question. Whereas the previous argument focuses upon the putative supervening fact or property, here the focus is on the putative relation that allegedly holds between the natural and the moral. But the moral realist can reply by asking just what exactly is supposed to be the problem, concerning the supervenience relation. Granted, the 224 T E R E N C E H O R G A N A N D M A R K T I M M O N S connection between the natural fact and the supervenient moral fact cannot be any sort of logical or semantic entailment; old-fashioned 'ethical naturalism', which attempted to analyze moral terms or statements as equivalent in meaning to naturalistic terms or statements, foundered upon G. E. Moore's "open question" argument; and Moore's objection likewise undercuts any putative one-way entailment relations between natural and moral facts or properties. Granted, too, that the connection is not just co-occurrence of the corresponding natural and moral facts or properties, involving mere coextensionality of naturalistic terms or statements and the corresponding moral terms or statements; rather, supervenience is a relation with strong modal force; it is a necessary connection. 6 But in the context of recent developments in metaphysics, philosophy of language, and philosophy of mind, the fact that supervenience is a necessary connection without being an entailment relation is not enough to show that it is metaphysically queer. For one thing, there is wide acceptance of Saul Kripke's contentions: (i) that a name is a "rigid designator", denoting the same entity at each possible world in which it denotes anything at all; (ii) that therefore an identity statement consisting of two names flanking the identity predicate is necessarily true if true at all; and (iii) that therefore some necessary truths are empirical, rather than being knowable a priori (Kripke, 1972). These days, few philosophers believe that all necessary truths are analytic, or reflect entailment relations. In addition, since the early 1970s there has emerged a decided trend toward positions in metaphysics and philosophy of mind which, although still broadly naturalistic ('materialistic', 'physicalistic'), nevertheless repudiate earlier naturalistic programs for 'reducing' higherlevel terms, statements, concepts, or theories to those of physics or the other natural sciences. And, ironically, the concept of supervenience has been invoked with increasing frequency, in an effort to articulate theses that are naturalistic without being excessively reductionistic. 7 So in a philosophical milieu where empirical necessary truths are widely accepted, and where supervenience relations are widely invoked for the purpose of articulating nonreductive naturalist views, it is not surprising that moral realists should be unimpressed by Mackie's second queerness argument. A perfectly sensible-looking realist reply to Mackie is given by Brink: T R O U B L E S ON M O R A L TWIN E A R T H 225 There is nothing strange or queer about the supervenience of moral properties upon physical properties. Assuming materialism is true, mental states supervene on physical states, yet few think that mental states are metaphysically queer (and those who do do not think that supervenience makes them queer). Social facts such as unemployment, inflation, and exploitation supervene upon physical facts, yet no one supposes that social facts are metaphysically queer. Biological facts supervene on physical states, yet no one supposes that organisms are queer entities. Macro-scopic material objects such as tables supervene on micro-scopic physical particles, yet no one supposes that tables are queer entities. In short, it is difficult to see how the realist's use of supervenience in explaining the relationship between moral and physical properties makes his position queer. (Brink, 1984, p. 120; cf. 1989, pp. 177-80) 8 This plea of innocence by association, drawing upon a comparison between the moral case and relatively uncontroversial cases in which one sort of property or fact supervenes upon another, is intended to shift the burden of proof back onto the shoulders of the anti-realist. And, of course, if one assumes that moral facts are otherwise as unproblematic as the ones Brink mentions in the above passage, then it is difficult to see what the problem is supposed to be with supervenience in the moral case. But we detect a deeper problem with moral supervenience, not detected by Brink and not explicitly formulated by Mackie although it perhaps underlies Mackie's provocative question about what "in the world" is signified by the 'because' of moral supervenience. The problem is connected to a little-discussed concern about objective supervenience relations generally, a concern about how to square appeals to supervenience with a broadly naturalistic stance in metaphysics. Once this general concern is made clear and is appropriately addressed, it will become clear that in the case of moral supervenience, a special problem arises a problem that cannot be side-stepped by a plea of innocence by association. 2. S U P E R V E N I E N C E , M E T A P H Y S I C A L N A T U R A L I S M , AND E X P L A N A T I O N Two interrelated theses seem central to a broadly naturalistic, or materialistic, world-picture: (1) that the ontologically primary or fundamental entities in the world are all part of the subject matter of physics; and (2) that the strictly physical facts (i.e., the facts describable in a canonical 226 T E R E N C E H O R G A N A N D M A R K T I M M O N S language of physics) determine, or fix, all the facts. We will call these theses physical ontic primacy and physical determination, respectively. Both are rather vague. Initially it seems fairly plausible to try explicating them via a generalized supervenience thesis like the following: (S) The world could not have been different in any respect without having been different in some strictly physical respect. More generally, there are no two physically possible worlds which are exactly alike in all strictly physical respects but different in some other respect. 9 At any rate, (S) seems a plausible explication of (1) and (2), provided we suppose, or stipulate, that no physically possible world contains entities like Cartesian souls that happen always to inhabit some living body and whose mental properties happen to be completely determined by the body's strictly physical properties and provided we suppose that our actual world is 'physically possible' in this anti-Cartesian, antivitalist, sense. (Since Cartesian souls and vital spirits are supposed to be incorporeal substances, capable of existing independently of any material substrate, surely no version of Cartesian dualism or of vitalism not even a version that countenances possible worlds of the sort just described, and counts our actual world among them should qualify as a species of metaphysical naturalism.) I° Upon closer consideration, however, it turns out that (S) really does not do full justice to the pre-theoretic ideas of physical ontic primacy and physical ontic determination. True, (S) expresses a modal kind of determination, viz., that across the physically possible worlds, there is no variation in higher-level facts without some underlying variation in strictly physical facts. And (S) thereby expresses a corresponding kind of ontic primacy for strictly physical entities and properties, viz., the facts about these entities and properties determine, in this trans-worldly way, all other facts. But the trouble is that this global kind of determination is itself based upon specific supervenience facts that is, facts about supervenience relations between certain specific physical facts or properties, on the one hand, and certain specific higher-level facts or properties, on the other. Yet such supervenience facts are not strictly physical. Why, then, do they obtain, rather than different supervenience facts, or none at all? Suppose this question simply has no answer i.e., that certain specific supervenience facts are utterly unexplainable on the basis of other facts. T R O U B L E S O N M O R A L T W I N E A R T H 227 Then surely these unexplainable inter-level facts, and the properties that figure in them, would thereby qualify as 'ontologically primary' in the naturalist's vague, pre-theoretic, sense. And, since such facts are the explanatory basis for the trans-worldly determination expressed by (S), surely the world's higher-level facts would thereby be 'determined' (in the relevant naturalistic sense) not by strictly physical facts alone but, rather, by the conjunction of the latter and the explanatorily sui generis supervenience facts. These considerations reveal that there is more to the theses of physical ontic primacy and physical determination than is captured by (S). Although the truth of (S) is plausible a necessary condition for the truth of these two interrelated theses, it is not a sufficient condition. In order to obtain an adequate explication, we must conjoin (S) with the following chain: (E) All facts about supervenience relations are explainable on the basis of other facts, rather than being explanatorily sui generis. A central feature of a naturalistic world-picture is that even if there are facts about the world that are explanatorily basic and hence not susceptible to explanation themselves, supervenience facts are not among them. Although explanation may have to end somewhere (perhaps with certain facts of physics, such as fundamental physical laws), facts about synchronic inter-level dependence relations should not be among those that are explanatorily sui generis. An appreciation for this dimension of naturalism is, we suggest, the crucial insight to be gleaned from Mackie's query about what "in the world" the supervenience relation might be. In the contemporary philosophical milieu, this question is best construed as expressing the demand that specific inter-level necessary connections be explainable. This demand is entirely legitimate, from the naturalistic perspective; moreover, it has nothing specifically to do with moral supervenience per se but, rather, is equally appropriate vis-?a-vis putative supervenience relations of any kind. And the alleged 'queerness' problem can be construed as the contention that if moral realism were correct, then moral supervenience relations would be explanatorily sui generis and, hence, would not be integratable into a naturalistic world-picture. 1~ In order to bring home the point that metaphysical naturalism demands that any objective supervenience facts be explainable rather than 228 T E R E N C E H O R G A N A N D M A R K T I M M O N S sui generis, it is useful to consider G. E. M o o r e ' s meta -e th ica l posi t ion. M o o r e denied that there are mean ing equiva lences , or one -way entailm e n t re la t ions, l inking na tura l t e rms or express ions to mora l ones. But he did not deny that mora l p roper t i e s are superven ien t upon na tura l p roper t ies . On the cont ra ry , he explicit ly aff irmed this posi t ion, although he did not use the t e rm ' supe rven ien t ' in ar t iculat ing it; he held that cer ta in propos i t ions l inking na tura l p roper t i e s to mora l ones in par t icular , p ropos i t ions of the fo rm, " A n y t h i n g that has na tura l p roper ty P also possesses the p r o p e r t y of intrinsic g o o d n e s s " are (synthetic) necessary truths, t2 H e also held tha t these propos i t ions do not d e p e n d for their t ru th (or thei r synthet ic necessi ty) upon anyth ing else; ra ther , the necessary connec t ions they express are metaphys ica l ly rockb o t t o m , and thus are not expla inable by any o the r facts. 13 But this la t ter v iew is surely unaccep tab le f rom a b road ly natural is t ic perspec t ive , no twi ths tanding M o o r e ' s al legiance to mora l superven ience ; unexplainable superven ience re la t ions would be just in to lerably queer within a natural is t wor ld-p ic ture , and so l ikewise would be the mora l p roper t i es that supposed ly figure in such relat ions. R. M. H a r e , who in t roduced the t e rm ' supe rven ience ' into the philosophical lexicon, recent ly has m a d e ra ther similar observa t ions abou t M o o r e , superven ience , and Mack ie ' s queerness a rgument . H a r e wri tes as follows. ( H e lets ' p ' r ep re sen t a s t a t emen t of the fo rm 'Fo r all x, if G x then Fx ' , where ' G ' expresses a na tura l p r o p e r t y and 'F ' expresses a mora l p rope r ty . ) Moore opted for . . , claiming necessity for "p ' . . . but at the same time rejecting the view that the necessity involved is purely conceptual . . . . I myself at one t ime. . , thought that the supervenience of moral properties could be appealed to in order to refute such a view . . . . I had given up this hope by the time I wrote The Language of Morals . . . . My reason was that it does not seem to me impossible that someone should maintain, as Moore did, that 'Necessarily p' states some kind of synthetic a priori truth. This leaves the relation between descriptive and moral properties looking queer. It is, as Blackburn says, 'an opaque, isolated, logical fact, for which no explanation can be proffered' (p. 111). This is a sitting target for Mackie's 'argument from queerness'. (Hare, 1984, p. 7) H a r e here acknowledges , in effect , that mora l superven ience is compa t ible with indeed , is a c o m p o n e n t of M o o r e ' s posi t ion. But he alleges (with S imon Blackburn , 1971) that the posi t ion is quee r anyway , because it posi ts specific supe rven ience facts which cannot be fur ther explained. H e thus cons t rues Mack ie ' s second queerness a r g u m e n t essentially in the way we p r o p o s e to cons t rue it viz., as d i rec ted not at T R O U B L E S ON M O R A L TWIN E A R T H 229 objective supervenience per se but, rather, at putatively unexplainable supervenience facts. Concerning the queerness argument, Hare continues: [T]hat argument does not rely on supervenience only on that robust sense of reality which the [moral] realist lacks. He [the moral realist] can, if we are prepared to swallow it, go on saying, without offence to supervenience, that there just is the sui generis nonnatural property which all things of a certain kind necessarily (but not analytically) have . . . . Like Blackburn, I find this hard to swallow, but others may have stronger stomachs (stomachs these days are getting stronger). (Hare, 1984, p. 7) What is hard to swallow, of course what offends a naturalistically minded philosopher's robust sense of reality is the idea that certain facts about specific nonmoral/moral connections are metaphysically rock-bottom, are utterly incapable of further explanation. Moreover, the queerness of such putative sui generis facts is only heightened by the claim that the relevant connections are necessary without being analytic. ~4 So the moral of Mackie's second queerness argument is that objective supervenience facts must themselves be explainable, at least in principle, in order to fit into a naturalistic world-picture. This explainability requirement holds quite generally, and not merely for moral supervenience. 15 Moore's position vis-a-vis moral properties clearly fails this test; and both Mackie and Hare intimate that moral realism is bound to fail it. We agree, and will so argue below. Although we will assume henceforth that any position worthy of the label 'metaphysical naturalism' will assert or entail that supervenience facts are explainable, we recognize that some philosophers, including some moral realists, might wish to dispute this claim (see, for instance, Post, 1987, chap. 6). But, as we will argue, the explainability demand is actually satisfiable for psychophysical supervenience relations and for other related cases, and is not satisfiable for putatively objective moral supervenience relations. This difference alone constitutes a form of moral queerness i.e., an important respect in which moral supervenience differs from these other kinds of supervenience. Thus, this difference alone will suffice to undercut the plea of innocence by association, and to shift the burden of proof back to the new-wave moral realists. A moral realist who chooses to claim that unexplainable moral supervenience relations comport with metaphysical naturalism will face the very demanding task of making this claim plausible despite the explainability of other familiar kinds of supervenience.16 230 T E R E N C E H O R G A N A N D M A R K T I M M O N S What about a version of moral realism that construes the relevant relation between nonmoral and moral facts or properties as identity, rather than supervenience? On this view, there is apparently nothing to explain concerning moral supervenience. Of course, identity versions based upon claims of meaning equivalence are hard to take seriously, in light of Moore's open question objection. However, someone might advocate a general view about the individuation of facts (and/or properties) according to which two sentences (and/or predicates) can express the same fact (property) even without being equivalent in meaning; and might conjoin this general position with the contention that moral sentences (predicates) express the same facts (properties) as do certain sentences (predicates) whose content is nonnormative. But identity versions of moral realism still face an explanatory burden, similar to that faced by the moral realist who espouses supervenience over identity. Now, though, the burden arises at the metalinguistic level. Even if goodness, for instance, is identical to some specific natural property, there remains the task of explaining why this natural property, rather than any other one(s), counts as the correct referent of the term 'goodness'. So although our subsequent discussion will be explicitly directed at supervenience versions of moral realism, it will be applicable, rnutatis rnutandis, to identity versions as well. 3. O B J E C T I V E V S . N O N O B J E C T I V E S U P E R V E N I E N C E Moral supervenience is common ground in meta-ethics; it is espoused by nonnaturalist moral realists like Moore, by new-wave moral realists like Brink and Sturgeon, and by anti-realists like Hare, Mackie and Blackburn. But supervenience in ethics is viewed rather differently by the anti-realist than by the realist, and the differences have important consequences concerning what needs explaining. We will consider this matter in the present section. The upshot will be that the explanatory demands faced by new-wave moral realists are substantially stronger, and hence substantially harder to satisfy, than the corresponding demands faced by many anti-realists. And this point is generalizable: for any domain of discourse (e.g., mental, aesthetic, semantic), if one holds a supervenience thesis about the given domain relative to some lowerlevel domain (e.g., microphysical), then a realist position concerning the higher-level domain will impose a stronger explanatory burden visT R O U B L E S O N M O R A L T W I N E A R T H 231 5-vis the relevant type of supervenience than will most anti-realist