Jump to ContentJump to Main Navigation
Moral RealismA Defence$

Russ Shafer-Landau

Print publication date: 2003

Print ISBN-13: 9780199259755

Published to Oxford Scholarship Online: January 2005

DOI: 10.1093/0199259755.001.0001

Show Summary Details
Page of

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (www.oxfordscholarship.com). (c) Copyright Oxford University Press, 2019. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.  Subscriber: null; date: 17 September 2019

Rationality and Disagreement

Rationality and Disagreement

(p.215) Rationality and Disagreement 9
Moral Realism

Russ Shafer-Landau (Contributor Webpage)

Oxford University Press

Abstract and Keywords

Distinguishes two strands of the classic antirealist argument from moral disagreement. The first has it that the extent of actual disagreement provides the basis of an inference to best explanation to antirealism. The second claims that ideal observers would disagree about moral matters, and this justifies an antirealist verdict. The author Replies to these criticisms in a number of ways, but the one the author likes the most invokes an ethics‐philosophy parallel. Since intractable disagreement among wise, informed and rational people about philosophical matters does not license a verdict of philosophical antirealism, it should not do so when the subject is ethics proper.

Keywords:   ideal observers, moral disagreement, philosophical antirealism

Those who advance antirealist conclusions often rely on considerations to do with disagreement to support their views. The basic idea, developed in the two-part argument offered below, is that the many moral disagreements we actually encounter, as well as the hypothesized moral disagreements we can anticipate among more ideal judges, are best explained by central antirealist assumptions. I think that this line of argument is mistaken at a number of points, and will try to reveal its weaknesses in what follows.

I. The Argument from Disagreement

The argument begins with the introduction of the evidence: the existence of widespread, sometimes intractable interpersonal and intercultural ethical disagreement. The claim is that this evidence is best accounted for on the assumption that moral judgements give voice solely to relatively parochial views, there being no objective truth that they might capture. This view is commonly supported by noting a contrast in the degree of consensus within the natural sciences, on the one hand, and moral enquiry, on the other. There are those who accept the plausibility of realism when it comes to geology or physics or molecular biology, but who take a decidely sceptical attitude towards moral claims. Scientific enquiry, broadly construed, has generated a great deal of consensus within its own communities—consensus on substantive truths within scientific disciplines, as well as consensus on what constitutes appropriate methods for confirming such truths. By contrast, there is much disagreement both about how to make progress in ethical investigations, and about substantive ethical issues.

Most people suppose that the kind and degree of convergence we see in science is attributable to the metaphysical standing of its subject matters. We (p.216) have made great progress in such fields as hydrology, astronomy, and botany because these disciplines study things that are best construed realistically; their stance-independent reality serves as an extremely useful constraint on investigation. In these disciplines, we have some way of determining when theories fail to match up with the reality they are intended to describe.

According to ethical nihilists and subjectivists, pervasive moral disagreement is good evidence for thinking that, by contrast, there isn't any such reality that is serving as a constraint on the development of moral opinions.1 If moral facts were ‘out there’, awaiting discovery, then we would have to charge at least one party to a moral dispute with having made some cognitive error. But given the breadth of such disagreement, surely the most charitable and plausible account of the matter is that no one need be mistaken. Instead, unlike successful scientists, moral interlocutors are simply registering their personal opinions, unfettered by an external moral reality that might check them. Their views reflect the only ethical standards there are, namely, those that have been constructed from personal experience or have emerged as an expression of social understandings. The extent of disagreement in ethics is best explained by the absence of any objective reality that could be captured by our moral judgements.

As antirealists realize, the case as it stands at present requires supplementation. The reliance on a disanalogy between scientific and moral enquiry will go only so far, and is subject to plausible efforts to explain much of the differences away. Antirealists need a further argument for thinking that such explanations (on offer below) fall short. That argument is not far to seek.

Here the relevant thesis is that even idealized enquirers would fail to converge on a set of ethical standards and verdicts, and that this lack of convergence warrants a rejection of realism. As it is usually framed, this argument claims that if everyone were perfectly knowledgeable and rational, and if realism about a domain were true, then there would be unanimity in judgements within that domain. But since we can expect no such unanimity in ethics, even among such ideal agents, we have excellent reason to reject moral realism.

This argument sees realism as committed to the view that at least one party to genuine disagreement must be in error, and that any such error is properly characterized as a cognitive or rational failing. (See, e.g., Wright 1992: 175.) (p.217) Now the powers of rationality, on realistic assumptions, are thought to be sufficiently great as to invariably home in on the truth. Everyone armed with such powers to a perfect degree ought therefore to converge in his or her judgements, in any domain that is best construed realistically. Yet we should expect moral disagreement to persist in even idealized conditions of enquiry. Therefore realism is false. The simpler explanation of moral disagreement—that individuals are reporting or expressing their own attitudes, which don't answer to any more objective standards—is therefore also the best explanation.

Thus the argument from disagreement is really a two-part argument. The first part consists in presenting evidence for an antirealist conclusion—the existence of pervasive actual moral disagreement, and the greater degree of dissensus within ethics, as compared to that found in the natural sciences. Because such evidence is susceptible of competing explanations of more than superficial plausibility, a second element is introduced—the existence of hypothesized moral disagreement under ideal conditions. This is supposed to clinch the case against realism, since (it is claimed) realism entails the unanimity of views among properly ideal moral judges. It is this second element that warrants the antirealist explanation of actual, apparently intractable moral disagreement.

Before assessing this two-part argument, let us consider a preliminary worry. This comes in the form of a charge that realists are forced to mischaracterize the nature of moral disagreement. If realism is correct, then moral error is cognitive error, and this sounds either wrong, or too underdone by way of fully describing the nature of moral mistake. For realists, the natural diagnosis of most actual moral disagreement is that the interlocutors are contradicting one another. It follows that at least one of them is holding a view that is false, and so making some sort of cognitive error. And that is to over-intellectualize what is going on in such cases. Moral error is first and foremost a matter of conative misalignment, and this element is left out of the cognitivist picture of moral mistake.

I hope to have done enough to show that such a view is not the default position, but is rather an expression of a theory that we need to be argued to. According to moral realists, and cognitivists generally, moral mistake does centrally involve false belief, and this is what licenses us in speaking of cognitive error. But this is only the start of a full diagnosis in any given case. One's emotions and moods, one's acculturation and upbringing, one's passions, fantasies, aspirations, and insecurities, all play complex roles in generating the moral views one holds. Moral error, at least of the sort that (p.218) involves attachment to false moral standards, often signals an affective misalignment as well. That moral error always involves cognitive error does not mean that it involves nothing else. The realist can grant such things their place in the aetiology and constitution of moral error, while continuing to regard moral error as mistaken belief.

II. Explaining Actual Disagreement

Let us turn now to the argument proper. The argument has two parts; consider first the evidence supplied by the breadth and depth of actual moral disagreement. At best, such evidence places a burden of proof on the moral realist that will be more or less significant depending on how well the realist can do two things: (i) challenge the evidence as presented, and (ii) challenge the inference to antirealism that is drawn from the evidence.

Regarding (i), it is a commonplace for realists to point to the very great deal of moral consensus within and across societies, and to mention matters on which the scientific community remains deeply divided. Though this is correct as far as it goes, it does not go far enough. At the end of the day, there does seem to be broader agreement in physics and chemistry than there is in ethics, both about method and about substance. This needs explaining.

There are the obvious points: much moral disagreement stems either from disagreement about what the relevant non-moral facts are, or is due to some error of instrumental reasoning. Clearing up these errors and getting consensus on non-moral facts would remove a great deal of moral disagreement. But probably not all.

Some disagreement is attributable to impoverished imaginative and sympathetic capacities. Agents who display such limitations might, at some level, know the non-moral facts and yet be unable to situate themselves in the places of those who would suffer from their moral endorsements. This is a real failing, and one whose elimination would substantially lessen moral disagreement.

In addition, some realists, myself included, will accept the idea that there is no uniquely correct answer to some well-formed moral questions. There are cases in which incommensurable or incomparable values are in play, and others in which various kinds of moral indeterminacy can manifest themselves.2 In such cases, we should expect something other than unanimity even among those who are reasoning well and possessed of all non-moral facts.

(p.219) Ethics is concerned primarily with how people ought to treat one another. Ethical standards provide a basis for action and the cultivation of character; they prescribe what we ought to do and how we ought to be. Scientific or mathematical principles aren't meant to be like this. To the extent that they are allied with normative recommendations, or are deployed for practical ends that threaten the interests or standing of another, there is a correlative increase in the disagreement we can expect within the scientific community. Self-deception, special pleading, and inattention to the nuances of competing views often accompany normative enquiries when self-interest or ideological commitments are at stake; these faults arise as well within scientific communities when scientific matters are crucial to (e.g.) the allocation of resources or the determination of social standing. The fact that so much more is typically at stake in an ethical enquiry partially explains why disagreement is broader there than in the sciences.

These considerations have been advanced by realists for a long while; David Brink (1989: 197–209) has done the best job that I know of in bringing them together and developing their individual and collective strengths in deflating the argument from existing disagreement. The last point he makes, taken from Parfit (1984: 453–4), is that the number of people and resources devoted to the development of secular ethical theories pales in comparison to those spent on scientific investigation. More philosophical effort may yield a greater consensus in ethics, perhaps one day rivalling that found in science. Even if this is too optimistic, we might explain at least some of the present disparity between the degree of consensus in science and ethics as owing to the far greater amount of people, time, and resources devoted to the articulation and testing of scientific theories.

At this stage it is too soon to tell whether this package of explanations for existing disagreement is implausible. Whether the realist's account is satisfactory depends on what else the realist has to say—about ontology, epistemology, moral motivation, and reasons for action. And it depends on what antirealists have to say about these things as well. The ultimate question must be focused on which picture offers the best comprehensive metaethical view, and even if there were some advantage gained by antirealists on matters of disagreement, that advantage might be overridden when surveying the liabilities of realism's competitors and noting realism's strengths. We can't confidently pronounce on the degree to which realism is weakened by considerations of moral disagreement until we have in hand a well-grounded comparative assessment of how realism and its competitors fare on a whole range of additional issues.

(p.220) That said, let me offer a further consideration that strikes me as supportive of the realist's analysis of actual moral disagreement. Disputes within ethics seem structurally to be quite similar to those within philosophy generally. Disagreement about the morality of abortion or euthanasia seems as intractable as that which separates nominalists from Platonists, theists from atheists, foundationalists from coherentists, libertarians from compatibilists. In none of these cases do we defer exclusively or primarily to scientific findings when trying to resolve the problems. Though the deliverances of the special sciences may have some role to play in these different debates, trying to resolve these problems is not primarily a matter of doing mathematics, physics, chemistry, biology, geology, or any combination of these. Philosophy generally, like ethics in particular, is not a science.

If the argument from disagreement is sound, then this should lead us to an antirealism about all philosophical views. For the extent of disagreement within any branch of philosophy is surely as great as—perhaps greater than—that found among substantive ethical views. Yet this sort of highest-order philosophical antirealism isn't very plausible. Even philosophers who are substantively sceptical about some domain (universals, free will, God, or moral facts) write as if their views were correct—not just endorsed by them; not just consistent with the attitudes they antecedently held; not just reflective of the spirit of the times, but correct for everyone, always—correct, simpliciter. It doesn't make much sense to say that moral realism (or nominalism or physicalism) is true for me and false for you, or true in this epoch but false in another. And it is even less plausible to say that endorsing realism is nothing other than a non-cognitive expression of one's practical commitments—that a judgement of realism's truth is not itself truth-evaluable.

Once we note this, the argument from disagreement against moral realism is considerably weakened. For the features that are used to generate a kind of scepticism about morality's status are shared by disagreements in all areas of philosophy. Disagreements in such areas as metaphysics, epistemology, and philosophy of language are pervasive and persistent, largely resistant to empirical resolution, and addressed without the benefit of determinate methods that are agreed by all to yield truth. Things are not relevantly different in ethics. If we are reluctant to see all metaphysical or epistemological opinions as nothing more than expressions of personal commitments, then we ought to be equally reluctant to take persistent ethical disagreement as a good basis for inferring the truth of moral antirealism. The disagreements that arise within moral discussion are not probative evidence for an antirealist diagnosis of morality.

(p.221) III. Explaining Hypothesized Disagreement

This brings us to the second element of the antirealist argument. This is where the real philosophical action is, for two reasons. First, were this second element vindicated—were it true that there would be moral dissensus even among idealized judges, and that such dissensus is incompatible with the implications of realism—then realism would be false even if there were no moral disagreements at all in the real world. Second, the evidence about actual moral disagreement can't be properly weighed without introducing considerations that bear directly on the outcome of idealized moral enquiry. For the realist's best gambit regarding that evidence is to cite a host of failings that beset actual interlocutors who persist in their moral disagreements. It would be natural, then, to suppose that on realist assumptions, once those failings are corrected, the disagreement will disappear. And that is just what the antirealist supposes will not occur. Realists, it is assumed, are wedded to the idea that idealized enquirers will converge on a truth that is guiding their enquiry. The persistence of moral disagreement in such hypothetical circumstances is therefore warrant for rejecting realism about morality.

Realists have two replies to this element of the argument. They might either defend the claim that idealized enquirers will converge in their judgements, or allow that they might not, but then deny that realism is committed to such convergence. I think that the realist is best advised to go this second route.

I don't think that realists, operating from assumptions about idealization and rationality that their opponents will agree with, are able to substantiate the claim that there will be perfect convergence among ideal enquirers or judges. To see why, we should begin with the governing antirealist assumption about rationality. On this assumption, rationality is to be understood procedurally: being rational or exercising one's rationality essentially involves a series of operations over one's existing commitments. These operations involve deduction, induction, abduction, and instrumental reasoning. Being rational is a matter of enacting particular kinds of reasoning processes that have their origin in a set of commitments that are not themselves rationally assessable. On such a view, any belief or end, in principle, can be rational, so long as it is arrived at in the appropriate way. Our basic, non-derivative attitudes will not be subject to rational scrutiny. All that is rationally assessable will be so relative to these basic attitudes, which may, without irrationality, differ from person to person. That is why antirealists say that we should expect disagreement among agents who are perfectly rational.

(p.222) This procedural account is opposed to a substantive one, according to which one is rational only if one possesses certain particular beliefs or ends, and fails to possess others. Antirealists who press the argument from disagreement, claiming that perfectly rational agents may nevertheless disagree about moral matters, will insist that this perfect rationality be understood procedurally. For once we allow for substantive constraints on what can count as rational attitudes, we allow that rational beings, qua rational beings, will share the same commitments (namely, a commitment to the substantive reasons that define the rational perspective). And this, of course, makes it far likelier that if there are genuine moral reasons, then perfectly rational agents will converge on them, contrary to the guiding assumption of the argument from disagreement. Thus antirealists must begin with a procedural account of rationality; in particular, one that takes the relevant procedural input as contingent attitudes that may, without irrationality, differ from person to person.

From this starting point we have two options. First, we might claim that we would continue our moral disagreements with one another, even if all of us were transformed into perfectly rational agents. That is almost certainly true, given our very disparate starting points. But that is hardly sufficient to threaten moral realism. It is also true that set theorists, physicists, and evolutionary biologists would very likely continue to disagree about the foundations of their disciplines, even if each reasoned with perfect propriety from his or her present intellectual commitments. Since that result doesn't cast doubt on the existence of objectively correct results in those disciplines, it shouldn't do so in ethics.

But this isn't the end of the challenge. Suppose we expand our idealization to include features other than perfect procedural rationality. Purge us of our false beliefs; give us all (relevant) non-moral information, vividly presented; add to this the fullest sympathetic and imaginative capacities, and make sure that our attitudes are ordered in a perfectly coherent way. Now ask the question: are those who match this description going to converge on a set of moral verdicts, or a set of moral standards that are to govern those verdicts? If we imagine away all features of ourselves that can serve to distinguish us from one another, as Rawls (1951, 1971) and Firth (1952) do, then we might well expect unanimity among such agents. In that case, the argument from disagreement collapses; there would be no disagreement that the realist is unable to explain.

Antirealists, of course, will object. Their best move is to query the initial conditions—the ‘erasure’ conditions, so to speak, according to which all (p.223) differences among perfectly rational agents are papered over. They might instead follow philosophers such as Brandt (1979) or Railton (1986), who allow that our individuating features can remain with us even under the relevant conditions of idealization. Under those conditions, we might expect some residual disagreement among perfectly rational agents. Even with ample imaginative sympathy and flawless instrumental reasoning, after full and vivid confrontation with all relevant facts, with a belief set in finely tuned reflective equilibrium, it may be that agents will diverge in their moral views, given their incompatible starting points. (Again, if they don't, no problem for realism—the argument from disagreement dissolves.)

For the realist enamoured of the idea that ideal judges will always converge, he must redraft the conditions that define the ideal standpoint, and good luck to him. But other realists can remain agnostic about the contents of the ideal standpoint(s), allow for dissensus among ideal enquirers, and stand fast to the view that even ideal moralists may, on occasion, fail to get things right.

Before I explain this neat trick, let me enter a defensive claim. Since these envisioned scenarios are hypothetical, and in particular involve an idealization of a sort probably never realized, it isn't clear whether we have any evidence sufficient to tip the balance of argument one way rather than another. Predictions about convergence or dissensus under idealized conditions seem more an expression of realist or antirealist assumptions than of conclusions for which we can offer some non-question-begging evidence. Since every one of us falls short of the ideal in so many respects, we may be in a quite poor position to determine the existence or degree of consensus in the counterfactual situation in which we imagine our failings away. In short, we may be in no good epistemic position to determine what the outcome of moral enquiry would be under idealized circumstances. If that were so, then the most powerful element of the argument from disagreement would be severely weakened, and with it, the argument itself.3

This reply has greater plausibility if we are assuming something like Rawls's or Firth's characterization of the agents and their circumstances of choice. But the critic can rightly ask why we should opt for such a characterization, as (p.224) opposed to one, such as Brandt's, that seems antecedently likelier to allow for divergence among perfectly rational agents. Indeed, the critic can expand this challenge into a general one that claims that there is no single specification of the ideal standpoint from which to register moral views. Given the diversity of kinds of moral exemplars, we have good reason to think that there is more than one way to idealize the affective and cognitive nature of ideal moral enquirers and their conditions of choice, response, or enquiry. It is this diversity of idealizations that warrants the expectation of a lack of consensus among ideal observers.

Fair enough. Yet realism is not threatened by this move. Even if there is no uniquely best picture of what an ideal agent looks like, or the conditions under which she chooses, there might still be correct answers to every well-formed moral question. There may be a variety of ways of getting at the truth, and a variety of kinds of conditions under which moral truth is likeliest to be gained. Realists needn't have any quarrel with such claims. As a metaphysical thesis, realism is committed to the existence of stance-independent truths. This thesis is compatible with a kind of epistemic pluralism, of the sort that allows such truths to be grasped in a variety of equally good ways. This is a quite general realist view. For instance, even if there is no such thing as the uniquely best characterization of an ideal physicist, or an ideal standpoint from which to render physical judgements, we may well believe that there are physical facts that are best construed realistically. So too with morality. Realists can endorse the idea of a truth of no one's making, while allowing for a number of different avenues that might reach it.

Yet given the diversity of idealizations, shouldn't we expect dissensus among ideal agents? Suppose we do. That doesn't impugn realism. If divergence amounts to contradiction, then realists must claim that even ideal agents may fail to get things right. This would signal our inability to craft an idealization such that, necessarily, those who exemplify it perfectly track the truth. We would refrain from defining ideal enquirers in this context as those whose judgements invariably reflect the truth. If truth is not constituted by the deliverances of such ideal agents, but rather serves as an independent constraint on their efforts, as realists will insist, then there is no difficulty in entering this conceptual limitation on their best efforts.

Though I don't know how to decisively argue for it, I think that the ‘diversity view’, in which we reject the notion of a single best standpoint from which to discover moral truth, is correct. We certainly have excellent inductive evidence for this claim, in the form of failed (though ingenious) efforts to identify such a viewpoint. Firth's (1952) ideal observer view, Brandt's (1979) (p.225) cognitive psychotherapeutic view, Railton's (1985, 1986) full information approach, and Smith's (1994) advice model are just the most prominent recent attempts to specify a standpoint that necessarily yields moral truth. I won't try to argue against such efforts here,4 but will simply record my impression of their failure, which to me is indicative of the failure of the general project of supplying a set of non-moralized conditions sufficient to entail the identification of moral truth. As I see it, there is no single best perspective from which to secure moral truth, nor any non-question-begging (i.e. non-moralized) method whose perfect exemplification entails a right answer to every well-formed moral question. Such concessions do no damage to realism.

The question is whether ideal agents judging in ideal conditions (whether one model or many) will converge in all of their moral judgements. If they would, then the argument from disagreement entirely loses its force. If they diverge, in a way that entails contradiction, then realists will have to settle for a characterization of ideality that allows for the fallibility of even ideal agents. Because realism does not see truth as constituted by even idealized attitudes taken towards some subject, realism allows for the possibility that moral truth may elude our best epistemic efforts. If dissensus is on the cards, then agents can be ideal, in the sense of being situated in the very best epistemic conditions, perfectly adhering to the very best philosophically constructible methods of enquiry, while still being subject to error. This isn't incoherent, and is what we should expect a realist to say of any given domain, on the assumption that ideal judges will diverge in their opinions.

Compare how things stand in philosophy generally. In most or all philosophical matters, we do, as I argued above, think that there are right answers not of our own making. And yet we have a hard time conceiving of a uniquely best standpoint from which to do metaphysics, or epistemology, or philosophy of science. Whether one ideal standpoint or many, we can say that those who inhabit the standpoint will either converge or diverge in their judgements. Convergence poses no threat to realism. But neither does divergence—realists about right answers for any philosophical domain must simply say that the relevant idealization fails to guard against the fallibility of ideal agents. The only ones who should feel threatened by the prospects of such dissensus are those constructivists who identify truth with the deliverances of ideal agents. Dissensus in that case is very serious indeed—so serious (p.226) as to require either a suspension of the law of non-contradiction, a withdrawal into relativism, or a recharacterization of the ideal standpoint (to ensure that dissensus is avoided).

This yields an important lesson about rationality and rationalism. As we have seen, there are two families of theory about rationality. Substantive views forge necessary connections between rationality and attitudinal correctness. If you like such views, then it will indeed be impossible for ideal agents under ideal conditions to diverge in their views. But those who advance the argument from disagreement adhere to less substantive views of rationality. If, to take some popular examples, to be perfectly rational is just to flawlessly identify best means to adopted ends, to infallibly infer entailed conclusions from valid arguments, or to coherently order one's ends into a compossible set, then we should expect ideally rational agents to disagree about moral issues. Any such conception of rationality, when allied with realism, entails that perfectly rational agents may choose badly or believe falsely. This alliance allows for cases in which one is exercising one's reason impeccably, and yet still landing in error.

This may sound strange, and especially strange for moral rationalists. For many have thought it an identifying feature of any kind of rationalism that, by virtue of the powers of reason, one is thereby able to identify all genuine reasons: the perfect exercise of reason ensures access to truth. But this would be so on just one condition: that every reason exists only in virtue of being a function of our reason. If the content of our reasons was conditioned on their accessibility to reason, then truth would be dependent on our epistemic limitations. But this is constructivism, not realism. Realists who are also rationalists hold that moral reasons are fixed in a stance-independent way. If that is so, then there will be no basis for assuming that our powers of reason must always, if successfully deployed, be able to accurately discover what reasons there are.

Recall that the argument from disagreement gets off the ground only if we assume that some procedural conception of rationality is the correct one. It may not be. If a substantive account is correct, then an agent, in so far as she is rational, will have attitudes that correctly reflect the truth about what reasons there are. Any agent who is perfectly rational will be possessed of all such truths. So all agents who are perfectly substantively rational will be possessed of all such truths, and therefore there will be no basis for disagreement.

I am not arguing here for a substantive conception of rationality, but instead pointing to a vulnerability of the antirealist position. To get the argument from disagreement going, antirealists must first show that no such (p.227) substantive conception of rationality is true. They might do that—certainly, proceduralist accounts are the ones most favoured among philosophers these days. Yet if it comes to that, realists can take over a procedural conception of rationality, and advance fully adequate replies to the argument from disagreement.

Even on proceduralist grounds, we might expect unanimity among ideal agents, as Rawls and Firth did, for instance. Or we might be less sanguine about such prospects, especially if we favour the view that there is a plurality of ideal conditions of choice and judgement. If dissensus is to be expected among ideal enquirers, realists may allow for the possibility of moral error among such agents. Realists would then have to maintain a distinction between truth and epistemic accessibility, even under humanly ideal conditions.5 That, after all, is the keystone of the realist enterprise. It may be corrupt, and the realist structure may therefore be in danger of collapse. But our discussion reveals that moral disagreement fails to supply good reason for suspecting the plausibility of this central realist tenet.

IV. Conclusion

How damaging is the argument from disagreement? Not very. There is the actual evidence about the extent of moral disagreement. There are antirealist assumptions about how this is to be best explained. And then there are the competing realist explanations. The thought was that we could assess the plausibility of the realist explanations by predicting whether ideal enquirers would converge in their moral judgements. But now we see that such predictions are useless by themselves. Perhaps we aren't even in a position to justifiably make these predictions, in which case we would have to suspend judgement about the merits of the argument from disagreement. But even if we could predict these outcomes with certainty, realists are well prepared to defend themselves.

Realists can rely on the analogy between moral enquiry and philosophical investigation generally to substantiate the claim that persistent, intractable (p.228) disagreement among informed and well-intentioned individuals is not enough to warrant antirealist conclusions. If there's no reason to suppose that even perfected enquirers will reach consensus on central moral matters, then there's as little reason to anticipate consensus on any other important philosophical issue. Yet surely there is a fact of the matter, realistically so understood, about whether, for instance, there is such a thing as agent causation, or abstract entities, or an all-perfect divinity. We will have our opinions about such matters, but they are ultimately answerable to a truth not of our own making. That is so despite the presence of intractable actual disagreement about these issues, and even the possibility of such disagreement among idealized enquirers. Therefore such disagreement poses no threat to philosophical realism of any stripe, and so, a fortiori, poses no threat to moral realism in particular.

In addition, realists have satisfying answers to the problem of disagreement on either of the two relevant outcomes that we have discussed. If ideal enquirers converge in their moral judgements, then the argument from disagreement must obviously be scuttled. If the ideal enquirers diverge, then realists will insist that even ideal methods and conditions of choice or enquiry are insufficient to ensure correctness. This is false only if antirealism is true. It may be. But considerations of disagreement fail to show it so.


(1) The antirealist argument from disagreement has been given in similar versions by Blackburn (1985); Mackie (1977: 36–8); Hare (1963); Stevenson (1937, 1948); Williams (1985: ch. 8). Replies have been offered by, among others, Wellman (1975); Hurley (1985); Brink (1989: 197–209); McNaughton (1988: 147–61); Miller (1985); Shafer-Landau (1994a). See also Foot 1958 and 1959.

(2) I defend this claim in Shafer-Landau 1994a, 1995. For a fine collection on the subject, see Chang 1997.

(3) Doesn't this ignorance cut both ways? Only if realists attempt to establish their claims by appeals to convergence under ideal conditions. But realists needn't do that, and few have gone this route in defending their views. The argument from disagreement is a standard antirealist critique; the ‘argument from agreement’, as it might be called, has not been nearly so popular. One good reason for this is that perfect convergence among idealized agents underdetermines a metaethic—such unanimity is compatible with many forms of constructivism, as well as realism.

(4) See, among many possibilities, Velleman 1988; Sobel 1994; Rosati 1995; Loeb 1995; Shafer-Landau 1999.

(5) Perhaps the strongest reasons to oppose such a distinction, at least in the normative realm, are those advocated by reasons internalists. Such internalists claim that there is a necessary connection between the truth (about what reasons there are) and a specific notion of this truth's epistemic availability: this truth must be reachable via a sound deliberative route with origins in an agent's existing beliefs, desires, commitments, etc. I hope to have shown, in Chapter 7, that the internalist arguments advocating an erasure of the distinction between normative truth and this sort of epistemic accessibility are not ultimately compelling.