Philosophical Studies (2005) 126:313–329 DOI 10.1007/s11098-005-2164-1
Springer 2005
RUSS SHAFER-LANDAU
REPLIES TO CRITICS
1.
REPLY TO COPP
I’d like to express my appreciation to Tom Blackson for organizing this symposium, and to David Copp, Mark van Roojen, and Robert Mabrito for giving me such astute and generous comments. I’m pleased to be the beneficiary of their intelligence and expertise. They have raised a number of important issues and objections, and, while I’ll do my best to muster something on my own behalf, I am not intent on avoiding all concessions. I wrote the book more in a spirit of exploration and tentative inquiry than of certitude; some of the criticisms you now have before you amply justify this authorial attitude. Let me begin with David Copp’s concerns. He has quite fairly characterized my claims, and devotes a good deal of his excellent and challenging discussion to showing how his own society-centered view can deal with a criticism I leveled against constructivist theories. I’m going to pass on the opportunity to probe the details and adequacy of his own important and nuanced view, and focus instead on the more general issues he raises. The first of these is whether I have, indeed, managed to raise an insuperable difficulty for the constructivist with my variation on the Euthyphro problem. The basic problem is this. The choices or attitudes of the favored agents whose responses are determinative of moral truth are either based on reasons or they are not. If they are not, then they are arbitrary, and this undermines the theory that would elevate their responses to the general truth-makers of moral claims.
314
RUSS SHAFER-LANDAU
If their responses are based on reasons, then these reasons must be either moral ones, or non-moral ones. If they are moral ones, then moral realism, not constructivism, is true, since there is a set of non-constructed moral truths that serve as constraints on the choices of the favored agents. But if these constraints are non-moral, then there is no reason to think that the responses of the favored agents will respect the platitudes that partly constitute our understanding of the moral notions that we are exploring. I admit that this isn’t anything like a knock-down argument against constructivism, since it rests on a prediction about the failure of constructivisms (such as those of Copp, Gauthier and Korsgaard) that have the agents of construction choose within a set of nonmoral constraints. I didn’t do anything in the book to vindicate this prediction, because I didn’t (and still don’t) know of a way, short of writing another very long book or two, to do it. Here I can only express my pessimism in a nutshell, without providing the very extensive argumentation that would be needed to defend it. I think the clearest way to get at my core worry for constructivism is to start with a concession. Moral constructivisms can avoid the Euthyphro problem. They can do this if their prescriptions are necessarily coextensive with those of the correct version of moral realism. For example, if an ideal observer theory were crafted in such a way as to characterize ideal observers so that, necessarily, their responses captured the independently existing moral truth, then the worry I raised for moral constructivism would not apply. My worry was that a wholly nonmoral characterization of the relevant choosers (and their conditions of choice) is bound to yield a theory whose moral recommendations clash with some of those that fix the boundaries of our moral concepts. This worry does not apply to the sort of ideal observer view I’ve just described. If Copp’s account of the matter is correct, then it turns out that his own view is necessarily coextensive with the correct version of moral realism. And so his version of constructivism would be immune to my Euthyphro worry. As Copp sees
REPLIES TO CRITICS
315
things, the basic society-centered theory, from which his own view derives, is a moral realist theory. Given the truth of this realist theory, which he accepts, we can readily craft a constructivist analogue that characterizes the relevant choosers in such a way as to guarantee that their rational choices yield the correct moral prescriptions. On closer inspection, it turns out that the true moral norms are not necessarily the product of rational choice, but are, in the first instance, best construed realistically. I couldn’t agree more. Though I’m not prepared to sign on to Copp’s specific version of moral realism, I don’t intend here to enter any criticisms of his own theory. Instead, I want to record our apparent agreement on the more general point – namely, that constructivism can be made to work, provided it is derived from, and is necessarily coextensive in its recommendations with, the correct version of moral realism. And now we are in a position to better appreciate, even if not to vindicate, my core worry for constructivism. This concern targeted what I might call non-derivative constructivism; a family of views each of which denies the truth of moral realism. Non-derivative constructivism must make the moral truth contingent on the outcomes of practical rationality. The different versions of non-derivative constructivism will distinguish themselves on the basis of their different conceptions of practical rationality, the different characterization of the favored agents whose attitudes or choices are constitutive of the general criteria of moral truth, and the different conditions of choice faced by these favored agents. They will all agree, however, that there is no moral truth apart from the attitudes or choices expressed by these favored agents. My worry was simply this: without realistic conditions of moral truth to constrain these agents, there is no cause for optimism that the agents will select norms that respect our moral platitudes. I still think that this is true. The conceptions of practical rationality needed to register the desired guarantees are not powerful enough to do the job being asked of them. Kantian constructivisms will likely respect at least most of our platitudes, but this respect is earned by
316
RUSS SHAFER-LANDAU
means of a conception of practical rationality (according to which immorality involves contradiction) that has always seemed to me mysterious and mistaken. Hobbesian instrumentalist views are, while not mysterious, too weak to ensure that those who perfectly exemplify them will select norms that respect our deepest convictions about paradigm cases of morality and immorality. I recognize that this is just autobiography, not argument, but as I say, I don’t see any short way of doing the needed work to vindicate this pessimism. Let me turn then to Copp’s second criticism, according to which my kind of moral realism fails to satisfactorily answer the skeptic. Though Copp does not put it this way, preferring instead to refer to my view as a kind of bald moral realism, I think that what he is objecting to is its nonnaturalistic element. For if a version of naturalistic moral realism were correct, then there would be the sort of descriptive criteria that would provide an answer to some of the skeptic’s concerns. If Copp’s own view is correct, for instance, then we might put the skeptic at ease by letting him know that moral constraints are those that emerge from the requirements of meeting society’s needs. Copp characterizes the skeptic as someone whose position does not rely on argument, but is ‘‘more resilient than any argument that could be given for it’’. If that is so, then it’s no wonder that my view could not dislodge such a person from his skepticism. No argument could. No argument is powerful enough to convince a thoroughgoing moral nihilist of the existence of moral truth (whether realistically construed or not). This is a limitation that philosophers just have to live with. Since we can’t convince a resolute skeptic about the external world, it should be no embarrassment that we are similarly blocked when faced with a moral skeptic. Someone who has a rock-bottom belief that there are no moral truths will not be moved by a presentation of Copp’s view, or Gauthier’s, or Korsgaard’s, or mine. That alone is hardly enough to defeat any of these accounts. Of course there are those who base their skepticism on argument, and in that case we can assess their skepticism,
REPLIES TO CRITICS
317
and perhaps even hope to shake their confidence if our critical analyses are good enough. In the moral realm, the most powerful such skeptics are those who advance a moral error theory. I tried to identify the major error-theoretic arguments in my book, and sought to undermine their force. To repeat, I don’t believe that moral realists (or any other moral success theorist) can provide an argument powerful enough to convince all moral error theorists to switch sides. But I do believe that the realist has the resources to show why error theoretic arguments are not compelling. Perhaps the most powerful error theoretic argument – one advanced by J.L. Mackie, and more recently by Richard Joyce1 – claims that moral requirements, if there are any, must be intrinsically reason-giving. But that is impossible; either there are no intrinsically reason-giving considerations, or those there are (e.g., self-interest) are not the basis of moral norms. So there are no moral requirements. Morality is a fiction. I spend two chapters (7 & 8) of my book trying to show why this line of reasoning is mistaken. If I have succeeded, then moral realism does have something attractive to say even to the skeptic who denies that he has reason to curtail self-interested pursuits for the sake of moral ones. Since Copp doesn’t mention this portion of my book in his essay, I can’t tell whether he finds its arguments persuasive. But he is, in any event, concerned with what seems to be a different form of skepticism. As he says, his skeptic doesn’t demand a showing that any fully rational person will be moral. Rather, he ‘‘wants to be shown, among other things, that there actually is such a thing as a moral requirement’’. Now many, including both Mackie and Joyce, will see this demand as nothing more than a demand to show that there are intrinsically reason-giving norms that are not derived from considerations of self-interest or desire-satisfaction. To show that there are moral requirements is to show that there are such norms. If that is right, then I simply have to refer readers to the aforementioned chapters, where they can judge for themselves whether I have done what’s needed to defend
318
RUSS SHAFER-LANDAU
against this skepticism. But suppose it isn’t right – suppose that we can separate the concept of a moral requirement from that of its reason-giving power. Why believe there are any moral requirements – even ones that are not intrinsically reason-giving? Copp believes that some forms of ethical naturalism, his included, have provided an answer to that question, and that my own reply, which asserts the existence of brute moral facts, is deeply problematic. I am not at all sure that ethical naturalists have done what is needed here. That is not a strike against them, though, because I don’t believe that anyone can do what Copp thinks must be done. I don’t know how one is going to convince an intelligent skeptic of the existence of moral requirements. Someone who doesn’t already believe that it is wrong to torture people for fun is not going to be won over by the presentation of the society-centered view. Offering a set of descriptive criteria for moral requirements, as naturalism does, may aid in the effort to identify which moral requirements there are. But it will do nothing to convince those who don’t already believe in such things. There is an obvious reply to this: what we mean by a ‘‘moral requirement’’ is a norm whose general acceptance would satisfy a basic social need. Since there are such norms, there are moral requirements. Skeptic answered. This will work, of course, only if Copp (or some other naturalist) has succeeded in finally naturalistically pinning down the meaning of ‘‘moral requirement’’. I don’t see that he has. A lot of competent and intelligent speakers don’t seem to mean this when they deploy the notion of a moral requirement, even if many of them do believe that moral requirements will, as a matter of fact, satisfy basic social needs if widely endorsed. Nor are they talking past Copp and his fellow naturalists when they engage in ethical and metaethical disagreements about this subject. My own view is that we must accept the existence of brute moral facts, ones that cannot be explained as having been ratified from some ideal constructivist perspective. I try to
REPLIES TO CRITICS
319
make this palatable by claiming, I think correctly, that any domain that is properly regarded as realistic must admit the existence of brute facts that do not admit of explanation by anything more fundamental. If one is a theist, for instance, nothing will explain God’s existence or perfection. If one is an atheist, nothing will explain the fundamental laws of physics. Copp thinks that this effort to draw a parallel with other realistic domains is misguided, because it ‘‘misses the fact that the skeptical concerns at issue are focused specifically on morality and arise because of the normativity of moral judgment’’. Yet aren’t there other normative domains that are also best viewed realistically? The principles that govern deductive inference, or that determine the conditions under which beliefs are epistemically justified, are both normative and, it seems to me, best construed realistically. They are not true because we, or some set of favored choosers, believe them to be. They just are true. It is an admirable philosophical impulse to want more than this. But not all truths can be explained. If I am right, some basic moral truths are like this. This will not satisfy the skeptic. But I doubt that anything ever will. 2. REPLY TO VAN ROOJEN
Mark van Roojen’s subtle essay is intent on exploring the implications for my view if we accept a version of reasons internalism other than the one I focus on in the book. The version of reasons internalism (henceforth, just ‘‘internalism’’) that I concern myself with is the following: Necessarily, R is a reason for an agent S only if S either is motivated by R, or would, after sound deliberation based on S’s existing motivations, be motivated by R. Let’s call this the official version of internalism. When combined with moral rationalism and the Humean view about the nature of motivation, the official version implies an anti-realist theory that I would like to reject. Now it is true, as van Roojen says, that I can save moral realism even if I accept internalism, provided I can show that Humean views of
320
RUSS SHAFER-LANDAU
motivation are mistaken. I am pleased to see that van Roojen credits me with having achieved that task, but I wasn’t confident of my success, and so felt the need to develop a back-up plan. Thus my critique of the official version in the book’s seventh chapter. Now so far as I can tell, van Roojen agrees with me in rejecting the official version of internalism. But that is not enough, he claims, to insulate my view from worries. For if we follow most philosophers in the area, and accept motivational Humeanism, and also accept moral rationalism, then realists face serious difficulties. That is because there is a version of internalism that is true, and that, together with these two additional premises, generates an antirealist conclusion. Given van Roojen’s avowed allegiance to rationalism, and his stated sympathies with moral realism, I take it that this he is relying on these comments to force us to see that realists need to reject motivational Humeanism. My book contains a brief discussion (173–174) of the version of internalism that van Roojen prefers. Let’s call this Korsgaard’s version of internalism, as, on my reading of her work, this is the view that she endorses: Necessarily, if a rational person S has a good reason to do something at t, then S can be motivated to do it at t. If we accept Korsgaard’s view of what it means for a person to be rational, then it turns out that her version of internalism is true – indeed, a tautology. For, by her lights, a person is rational only if she is motivated by the reasons that genuinely apply to her. If we build this motivational susceptibility into the very concept of a rational person, then, by definition, rational people cannot fail to be motivated by the reasons that apply to them. This explains the necessary connection between reasons and motivations that is alleged in her version of internalism. On this understanding of the matter, Korsgaard’s version can’t be false, and so there’s no point in denying its truth. In the book, I argued that this version of internalism poses no threat to moral realism. Van Roojen disagrees. If we combine moral rationalism with Korsgaard’s version of internalism, and add a motivational Humeanism that is
REPLIES TO CRITICS
321
restricted to rational agents, then we get the conclusion that van Roojen identifies as 4a.
Necessarily, if a rational person S is morally obligated to F at t, then S must, at t, either desire to F, or desire to Y, and believe that by F-ing S will Y.
The worry here is that ‘‘[A]ny substantive moral constraint would be rationally escapable’’. And that would be a real problem for those who endorse moral rationalism – the view that moral obligations entail good reasons for action. Not all moral realists accept moral rationalism. But I do; so does van Roojen. Yet I don’t see why accepting (4a) entails the rational escapability of any substantive moral constraint. Presumably the argument from (4a) to the rational escapability of moral demands runs like this. If (4a) is true, then the content of moral obligations will depend, for rational agents, on what they desire. And since what one desires is a contingent matter, then the content of moral obligations is likewise a contingent matter. What we (rational agents) are obligated to do would depend on what we desire to do; but this makes moral obligations escapable, since all we need to do to free ourselves from any given one is to alter our desires. Realists can’t accept that. I hope that this is a fair representation of the thinking behind van Roojen’s concern. He doesn’t spell out the argument, so I am offering what I take to be the most plausible effort in that direction. Suppose I’ve managed to target the real worry. Then what? Worst-case scenario: concede that (4a) is inimical to rationalist realism, and so (since Korsgaard’s internalism is tautologous) abandon either (i) moral realism, (ii) moral rationalism, or (iii) motivational Humeanism. Since I devote a chapter (the fifth) to argue against motivational Humeanism, a chapter (the eighth) to argue for moral rationalism, and the entire book to defend moral realism, my choice is clear. The worstcase scenario isn’t so bad – provided that my anti-Humean arguments are sound.
322
RUSS SHAFER-LANDAU
Yet even if I have failed to successfully indict the motivational Humeans, I think my view can avoid the difficulty that van Roojen is concerned about. And that’s because (4a) does not threaten rationalist realism. To see this, consider again what must be true if Korsgaard’s version is to be a tautology. It must be that we define a rational agent as (at the least) one who is motivated by the reasons that apply to her. Now if (as Korsgaard, van Roojen and I all agree) moral rationalism is true – if it is true that moral obligations entail good reasons for action – then it must be, on Korsgaard’s version, that all rational agents will be motivated to do their duty. Their duty entails good reason for action, and their rationality entails that they will be motivated by their reasons. If Humeanism about motivation is correct, then these motivations will be at least partly constituted by desires. But if all this is true, then the presence of certain desires, on the part of any rational agent, will not be contingent. Rational agents as such will, of necessity, possess certain desires. Among these will be desires to do whatever it is that they are morally required to do. But it’s not as if we craft the moral obligations by reference to the desires of these agents. That would indeed clash with moral realism. Rather, we assign to rational agents a certain set of desires that can be derived from the moral obligations that apply to them (in conjunction with moral rationalism, Korsgaard’s version of internalism, and motivational Humeanism). Such an assignment is perfectly compatible with moral realism. It doesn’t support moral realism. But it doesn’t undermine it, either. Van Roojen proceeds to make a number of interesting points, especially to do with his distinction between two kinds of internalism. I would very much have liked to discuss these, but owing to space constraints, I want to conclude this portion of my replies by tying it in with a theme that Copp has also developed. And that is the possibility of situating my main points within a constructivist framework. Van Roojen believes that there is a true internalist principle according to which an agent has a reason to do something only if he would be motivated to do it after soundly
REPLIES TO CRITICS
323
deliberating on the basis of relevant evidence. I agree that this principle is true – provided we first grant certain background assumptions. The first assumption is that the Humean theory of motivation is false – evaluative beliefs, and not just desires, can do the relevant motivating. The second assumption is that we are here speaking only of rational agents, where such agents are defined in part as those who are motivated to act on the reasons that apply to them. The third is that what constitutes relevant evidence can include facts about what is morally required. If we grant these assumptions, then, given the truth of moral rationalism, it follows that ‘‘[A]n action would be right only if sound deliberation based on relevant evidence would generate a motive to do it’’. This certainly sounds like a constructivist position. But as I mentioned in reply to Copp’s essay, it may be possible to develop a constructivist position whose moral recommendations are necessarily coextensive with moral realism. Yet it doesn’t follow that the order of moral explanation proceeds as constructivists allege. Van Roojen’s preferred version of internalism may be true only because moral realism is true – only because stance-independent moral facts will comprise some of the relevant evidence that rational agents must take into account in their deliberations. These facts, according to moral rationalism, will entail reasons for action. And an agent’s rationality will be defined at least partly by her being motivated by such reasons, and so, in effect, by the moral demands that apply to her in a realistic way. Thus it is true, on these assumptions, that actions are morally required only if a rational agent would be motivated to do them after soundly deliberating on the basis of relevant evidence. Yet it isn’t as if an agent’s contingent motivations are constraining the content of moral requirements. Rather, the realistic content of such requirements is placing a constraint on what an agent’s motivations can be. This might strike readers as quite odd. If it does, I suggest three diagnoses: (i) a (perhaps tacit) allegiance to motivational Humeanism; (ii) a rejection of moral rationalism, or (iii) a view of rational
324
RUSS SHAFER-LANDAU
agents according to which they might be unmotivated by the reasons that genuinely apply to them. Van Roojen rejects motivational Humeanism, endorses moral rationalism, and accepts the depiction of rational agents given in (iii). I do, too. But given all that, the sort of reasons internalism that he advances is no threat to moral realism – as I wrote in my discussion of Korsgaard’s version (173–174). This kind of internalism is compatible with realism (and will, given certain assumptions, even derive support from it). And though it is also compatible with a form of moral constructivism, this version of internalism does not provide any reason for thinking that our attitudes constrain the content of moral requirements, rather than the other way around. As I wrote in the book, some versions of reasons internalism may turn out to be tautologous. For any such version, it wouldn’t do to deny its truth. But once we grant the assumptions required to make it a tautology, we find that it is in no position to threaten moral realism. 3. REPLY TO MABRITO
Robert Mabrito’s careful, exceptionally clear essay succeeds, I am afraid, in revealing some of the errors that plague my discussion of supervenience. Perhaps the most important of these is my charge that expressivism lacks an adequate account of moral supervenience. I think that this was true at the point at which the book was written. But in the interim, Allan Gibbard has, as Mabrito writes, developed a view that can provide the needed account. I don’t believe that the expressivism within which this account is situated is, in the end, the correct view of things, but for present purposes that is neither here nor there. Mabrito is correct to say that expressivism now has the needed tools to explain why the moral supervenes on the descriptive. Whether this puts the realist behind the eight ball depends on two things: first, whether an explanation of moral supervenience is needed, and second, whether realism can provide one if it is.
REPLIES TO CRITICS
325
Before considering these questions, we must be sure that we’re clear about the relevant explanandum. It does not appear that Blackburn is challenging the realist’s ability explain why the moral supervenes on the natural. The constitution story I favor could do that. So could a number of other realist accounts. This isn’t to say that any of these theories are plausible (though I obviously hope that at least one of them is). It’s just to say that, if they were true, they could do the relevant explaining. If it’s not the supervenience of the moral on the natural that needs explaining, then what is the relevant explanandum here? What Blackburn wants from the realist, but doubts that he can get, is an explanation of (i) why it is a conceptual truth that the moral supervenes on the natural, even though (ii) there are no conceptual truths that link the instantiation of natural properties to moral ones. There are two basic ways to conceive of Blackburn’s challenge. The first is to see it as a request for two distinct explanations. The realist must explain (i), and she must explain (ii). The second is to see it as a request for a single explanation. The realist must explain why, given the truth of one of the theses, the other one should also be true. It’s easy to see that there is no contradiction involved in asserting both (i) and (ii). But we might be puzzled as to why (ii) is true if (i) is, and why (i) is true if (ii) is. I think that Blackburn really has the second challenge in mind. He doesn’t argue that it’s difficult for the realist to explain why moral supervenience is a conceptual truth. And he apparently thinks the realist can readily explain why moral principles linking natural and moral predicates are not conceptual truths. What the realist cannot explain, according to Blackburn, is why, given the truth of one of these claims, the other one should be true as well. Blackburn’s demand is that the realist be able to explain the ban on mixed worlds. A mixed world is one in which the instantiation of a natural property N sometimes does, and at other times does not, subserve a moral property M. The thesis of moral supervenience says that this cannot happen.
326
RUSS SHAFER-LANDAU
But why can’t it happen, if the lack of entailment thesis is also true? If a concatenation of natural properties does not, as a conceptual matter, subserve any particular moral property, why must it do so if it has ever done so? I think that there is a straightforward answer to this question. The answer is one of the two that I provide in the book. The second one – the one that invokes the constitution story – is, as I now see from Mabrito’s criticism, not well-suited to do the needed explaining. But I think that the other one, which Mabrito does not mention, succeeds. My favored reply is that, in effect, there is no mystery that needs solving. Blackburn’s puzzle is no puzzle at all, provided we are justified in thinking that there are metaphysical entailments that link the instantiation of natural properties with the instantiation of moral ones. Moral realists who are ethical naturalists have always believed this. And many nonnaturalists believe this as well. I do; so does G.E. Moore.2 Nonnaturalists will deny that moral properties are natural ones, and they will insist on the multiple realizability of moral properties. So there will not be any conceptual or metaphysical entailments going from moral properties to natural ones. But if I and others are right, there will be entailments (metaphysical, not conceptual) that go the other way, from the natural to the moral. Now Blackburn’s puzzle arises only in cases where both the moral supervenience thesis, and the lack of entailment thesis, are true at the same modality. If Moore et al. are right, these claims will not both be true at the level of metaphysical necessity. The supervenience thesis will be true, but the lack of entailment thesis will be false – there are metaphysically necessary truths of the form, If N, then M. And so there is no puzzle there. Blackburn didn’t see this in his earlier article on the supervenience problem, but he had it clearly in mind in the later article that Mabrito cites. In that piece, he recognizes the possibility of this move, but simply pushes the problem to the level of conceptual truth. The supervenience of the moral on the natural is a conceptual truth. But there are no conceptual
REPLIES TO CRITICS
327
truths of the form, if N, then M. And so the puzzle has been reinstated. Here is the answer to this alleged puzzle that I offered in the book: Assume for now that it is a conceptual truth that moral facts/properties/ relations are supervenient ones. The problem, then, should be that competent speakers of a language can conceive of a world in which the base properties that actually underlie particular moral ones fail to do so. But there is no mystery here, since people can conceive of many things that are not metaphysically possible. If certain base properties metaphysically necessitate the presence of specified moral properties, then the conceptual possibility that they fail to do so reveals only a limitation on our appreciation of the relevant metaphysical relations. There is no deep explanatory puzzle resisting resolution here. (86)
I still think that this is the correct reply to Blackburn’s concern. If it is, then we might ask whether I have really got at Mabrito’s fundamental worry. Maybe he really is asking for an explanation of why it is that the supervenience of the moral on the natural is a conceptual truth, rather than asking for an explanation of the ban on mixed worlds. After all, the expressivist account he offers is really an explanation of supervenience, rather than the relevant ban. Now he may be right that none of the three semantic accounts (invoking causal regulation, general referential intentions, or platitudes) that I mention are capable of doing the trick. That is a very complicated question, and I don’t think that Mabrito has done enough to show that they must fail in this regard. But suppose that they do fail. I introduced these views as the best of the extant semantics for ethical naturalists. I am not an ethical naturalist, and I was discussing these views so as to show that they do not lend support to a reductive ethical naturalism. So even if these semantic views do not explain how the supervenience of the moral is a conceptual truth, I don’t see that this would undermine the ethical nonnaturalism that I endorse. Still, this is all so much defensive work. Do I have anything to offer by way of a positive explanation of why it is
328
RUSS SHAFER-LANDAU
that moral supervenience is a conceptual truth? The answer, I’m afraid, is No. How damaging is this concession? I don’t think that it is very damaging. For I don’t think that all, or even most, conceptual truths require explaining. As I explained in the book (88), I believe that the supervenience of the moral on the descriptive is really a notational variant on Leibniz’s law of the identity of indiscernibles. I don’t think that Leibniz’s law needs explaining. It is one of those basic principles that we rightly use to constrain our thinking in metaphysics. This isn’t to say that we can’t support it at all, but only that any support it does receive will take the form of its being situated in reflective equilibrium, rather than having been derived from yet more fundamental and more clearly justified metaphysical principles. I think that the principle of moral supervenience is exactly like Leibniz’s law in these respects. And so the ability of expressivism to explain moral supervenience, while no doubt a good thing, does not automatically give it a leg up on my kind of realism, since explaining the conceptually necessary nature of moral supervenience is (I claim) not a proper criterion by which to assess the comparative merits of metaethical theories. That most metaphysical realists lack an explanation of why Leibniz’s law is a conceptual truth does not give us good reason to suppose that metaphysical anti-realism is true. The same, I believe, applies for the metaphysics of morals. Now suppose that I am mistaken about that (and most other things I’ve written here). Still, even if expressivism has this particular explanatory advantage, it isn’t clear how heavily to weigh it when assessing the comparative merits of realism and expressivism. Consider: some theists have an explanation of why moral truths are what they are – God commanded them. And some moral realists, such as myself, have no explanation – the basic moral principles are brute metaphysical facts about the way the universe is structured. By Mabrito’s reasoning, this means that, other things equal, we have reason to think that theism is true and that the nontheistic moral realism I develop is false. But until we are in a position to discern whether all other things are equal, the
REPLIES TO CRITICS
329
explanatory advantage had by the divine command theory does not, I think, carry much weight. As I see it, things are exactly parallel in the debates between expressivists and moral realists. And so even on my worst-case scenario, according to which an explanation of the conceptual truth of moral supervenience is an important theoretical desideratum, and one that my brand of moral realism can’t satisfy, this might be only a quite minor failing when weighed against the difficulties that beset expressivism, and the attractions that can be mustered on realism’s behalf. I have done my best to set these out in the book that has received such generous and acute criticism here. I’d like to thank my critics once again for the kindness they have done me in thinking so hard about the merits of these efforts. NOTES 1
Mackie, J.L. (1977): Ethics: Inventing Right and Wrong, Penguin; Joyce, R. (2001): The Myth of Morality, Cambridge University Press. 2 See Moore (1942): ÔReply to my Critics’, in P. Schilpp (ed.), The Philosophy of G.E. Moore (p. 588), LaSalle, IL; Open Court Press. I provide the text of the relevant passage in my book on p. 85.
Department of Philosophy University of Wisconsin-Madison 5185 Helen C. White Hall, 600 North Park Street Madison, WI 53706 USA E-mail:
[email protected]