The Argument from Moral Experience

Abstract

It is often said that our moral experience, broadly construed to include our ways of thinking and talking about morality, has a certain objective-seeming character to it, and that this supports a presumption in favor of objectivist theories (according to which morality is a realm of facts or truths) and against anti-objectivist theories like Mackie’s error theory (according to which it is not). In this paper, I argue that our experience of morality does not support objectivist moral theories in this way. I begin by arguing that our moral experience does not have the uniformly objective-seeming character it is typically claimed to have. I go on to argue that even if moral experience were to presuppose or display morality as a realm of fact, we would still need a reason for taking that to support theories according to which it is such a realm. I consider what I take to be the four most promising ways of attempting to supply such a reason: (A) inference to the best explanation, (B) epistemic conservatism, (C) the Principle of Credulity, and (D) the method of wide reflective equilibrium. In each case, I argue, the strategy in question does not support a presumption in favor of objectivist moral theories.

1 Introduction: The Pervasiveness of the Argument

It is often said that our moral experience, broadly construed to include our ways of thinking and talking about morality, has a certain objective-seeming character to it. There are two ways in which experience of this sort has been thought relevant to the central questions of metaethics. The first involves the traditional idea that moral experience – especially our dispositions to use the moral vocabulary in various ways – is among the best evidence we have about what it is we are (thinking and) talking about when we talk about morality: We are, it is claimed, talking about a realm of (putative) fact.¹ The second way is not inconsistent with the first. It involves an inference from morality seeming a certain way (or our practices somehow presupposing it to be that way) to the reasonableness of a presumption that it is that way. It is widely thought that the objective-seeming nature of our moral experience supports a presumption in favor of objectivist theories (according to which morality is a realm of non-relative facts or truths) and against anti-objectivist theories such as Mackie’s error theory (according to which it is not). The presumption can be defeated, it is claimed, only if arguments against objectivist theories prove successful. I will call the argument that our moral experience supports objectivist theories, the Argument from Moral Experience or the AME, for short.

In some form or other, the AME (or at least the presumption it is said to ground) has played an important role in the debate over moral realism.² Many of those accepting it consider themselves to be moral realists and take the AME to support their realism.³ Thus, Jonathan Dancy claims that a version of the AME is “the main argument for moral realism” and even “perhaps the only argument for realism, remaining thoughts being used for defence/offence.” (1986, pp. 172, 175). He describes “the simple form of the argument” as follows:

[W]e take moral value to be part of the fabric of the world; taking our experience at face value, we judge it to be the experience of the moral properties of actions and agents in the world. And...we should take it in the absence of contrary considerations that actions and agents do have the sorts of moral properties we experience in them. This is an argument about the nature of moral experience, which moves from that nature to the probable nature of the world. (Dancy 1986, p. 172).

According to David Brink, the “presumptive case in favor of moral realism...shift[s] the burden of proof to the moral antirealist.” (Brink 1989, p. 36). David McNaughton puts it even more forcefully, saying, “The realist’s contention is that he has only to rebut the arguments designed to persuade us that moral realism is philosophically untenable in order to have made out his case.” (McNaughton 1988, pp. 40–41).⁴ This assignment of the burden of proof helps to explain why so much of the debate over moral realism has centered on arguments against it, and far less has focused on arguments in its favor. If the negative arguments can be defeated, it is widely thought, then moral realism wins by default.

But those who view themselves as moral realists cannot appropriate the AME as readily as they have often claimed, for it can be argued that our moral experience does not presuppose that moral realism is correct, but at most that morality is objective in the sense that there are non-relative moral facts, truths, properties, or correct answers to our moral questions.⁵ All moral realists are objectivists, but many objectivists do not consider themselves to be moral realists. For example, most Kantians and Kantian constructivists would agree that a claim of objectivity is implicit in our moral thinking, and that this supports objectivism about ethics. But they would not agree that ordinary moral experience thereby supports moral realism. On their views, certain moral judgments are practically correct even though moral properties are not real and the correctness of moral judgments does not consist in correspondence between our statements and the world.⁶

Even those in the broadly non-cognitivist⁷ camp, who hold that moral utterances are (primarily) something other than straightforward assertions of putative fact, often claim it an advantage that their theories can accommodate (or so they maintain) the most important features of our objective-seeming moral experience. For example, Simon Blackburn says that “the most forceful attack” faced by the moral projectivist, “is that he cannot accommodate the rich phenomena of moral life,” but goes on to argue that his “quasi-realism” can indeed accommodate these phenomena. (Blackburn 1993, p. 158)⁸

The near universal acceptance of the AME helps to explain why the error theory has had so few adherents in recent years. The error theory, made famous by Mackie (1977), combines cognitivism, the view that the central function of moral utterances is to make factual assertions, with the view that all (positive) assertions of this sort are false.⁹ Thus it is a prime example of an anti-objectivist theory. A weak, burden-shifting, presumption in favor of objectivist theories would not be sufficient to explain why the error theory is held in such low regard, however. Instead, a much more ambitious version of the AME seems to be in play, one which holds that there is an overwhelming presumption in favor of objectivist theories. If the error theory is true then it is not a fact that torturing infants for fun is morally wrong. Anti-objectivist implications like this are not merely taken to place the burden of proof on error theorists, but to make that burden extremely strong.

Indeed, Mackie himself was in the grip of an ambitious version of the AME. Mackie was convinced that “ordinary moral judgments include a claim to objectivity, an assumption that there are objective values.” (Mackie 1977, p. 35). This was the basis for the cognitivist component of his error theory. But he also believed that in virtue of this claim views like his faced an uphill climb:

But since this is an error theory, since it goes against assumptions ingrained in our thought and built into some of the ways in which language is used, since it conflicts with what is sometimes called common sense, it needs very solid support. It is not something we can accept lightly or casually and then quietly pass on. (Mackie 1977, p. 35).

I agree with Mackie that objectivism should not be rejected lightly or casually. If objectivism is needed in order to validate common sense, then it might be very disturbing to find out that it is wrong. But that in itself gives us no reason to believe that objectivism is correct.

In what follows, I ask, first, in what sense is it true that we experience morality as a realm of fact? Although this is largely an empirical question, I argue that evidence we already have suggests that things are much messier than objectivists have supposed. Second, what support does our moral experience give to the claim that morality is such a realm? I argue that even if moral experience were to display morality as a realm of fact (or presuppose that it is), that would not in itself support objectivism.

The stakes are high. If I am right, then, at a minimum, the playing field should be leveled. One might even have thought that those defending an existence claim should themselves accept the burden of proof. If so, and if, as Dancy claims, the AME is the only argument for objectivism, then its failure would mean that anti-objectivist views like the error theory have the upper hand.

2 Do We Experience Morality as a Realm of Fact?

In what sense is it true that we experience morality as a realm of fact? To begin, “our moral experience” is not as uniform as that expression suggests.¹⁰ Undoubtedly there are both cross-cultural and intra-cultural differences. Still, there are some obvious commonalities for many of us, and perhaps we can base a reasonable gloss on “our moral experience” on these.

Brink (1989, pp. 25–29) presents one of the fullest cases for the claim that our moral experience has an objective-seeming character. Moral utterances, he points out, are often in the declarative mood, and thus appear to be statements of fact. Our beliefs that, “one should not be held responsible for actions one could not have known were wrong,” that “goodness deserves reward,” and so on, contain implicit references to moral properties, facts, or knowledge, he argues. Furthermore we often wonder what morality requires, and when we do, “we often deliberate as if there is a correct answer to the question before us.” We disagree with those whose moral views seem mistaken. We recognize that we have been mistaken in the past (and might be now), but we also believe that we can be right.¹¹

David McNaughton (1988, pp. 19, 48, 56) adds other elements, centered on the phenomenology of our moral experience. Morality seems to be in the world apart from our happening to encounter it, not something that depends for its existence on our subjective inputs. We seem to have moral perceptions, which, although not strictly analogous to visual perceptions, nevertheless are like them in that they seem to be of something outside of us. We seem to see conduct as right or wrong. And when we are moved to act morally, it seems to be in virtue of our recognizing morality’s authority over us.¹²

This appears to be an impressive case for the claim that we experience morality as a realm of fact. Of course, few non-philosophers have thought about the issue in terms that make explicit reference to moral facts, and it seems illegitimate to attribute much metaethical theory to most people. But most people do seem to recognize a distinction between realms of fact and others (such as fashion) in which normative facts are not thought to play a central role. Perhaps they would be disposed on reflection to place morality in the former category. Or perhaps our moral practice and thought in some more inchoate way presuppose objectivism. For example, the claim that we can reason about moral questions can be true, it might be claimed, only if morality is objective. Our moral beliefs and practices might be appropriate if objectivism is correct, but not if it is incorrect.

But is it clear that even this more limited presupposition is present? It is surprising how often philosophers are willing to generalize about complex, subtle, and largely empirical matters like this based merely on their own experience and intuitions. The evidence may well take us in more than one direction. People’s beliefs and presuppositions, especially about philosophical matters, are often intrapersonally inconsistent, and interpersonal and cross-cultural differences surely complicate things. Thus, even if we find substantial evidence of a commitment to objectivity, we may also find evidence against such a commitment.

Indeed, certain features of common moral experience do suggest that we experience morality as something that is not objective.¹³ For example, just as we talk about moral beliefs, we often talk about moral feelings and attitudes as well, and in other contexts these words typically signify something other than beliefs. In fact, people often say things that seem quite incompatible with objectivism, such as that in ethics “it’s all relative,” or that what it is right for a person to do depends on that person’s own decisions. We cannot dismiss such statements as the products of confusion merely because they appear to conflict with views we think widely held.¹⁴

There are other problems for the claim that we experience morality as a realm of fact. In particular, anti-objectivists claim that their theories can accommodate many of the features often cited as pointing towards objectivism. Unless ordinary people have metaethical views inconsistent with that claim, we need a reason for thinking that their experience presupposes objectivity, and such a reason has not been supplied. Mackie, for example, thought it entirely appropriate to reason about questions of value, even though he denied that values are objective. Barring the implausible assumption that ordinary people think reasoning is only appropriate when it comes to objective matters, ordinary moral reasoning could be of just the sort Mackie endorses. Or it might fail to reflect either an objectivist or an anti-objectivist presupposition.

Similarly, because Mackie thought of morality as something to be made and not discovered, it is not at all surprising to find him writing about values in the declarative mood. Similar things can be said about disagreement on questions of value, and even, perhaps, about the doubting and wondering mentioned by Brink. Once again, unless we have reason to believe that ordinary people are in the grip of a metaethic incompatible with Mackie’s, proponents of the AME need to say more if they are to make persuasive the claim that ordinary moral practices presuppose objectivity.

In fact, much of the evidence cited in favor of the objectivist reading of moral experience – for example, that we tend to express our moral thoughts in the declarative mood – seems weak. We talk about matters of taste in the declarative mood (saying, for example, that chocolate is better than vanilla), but ordinary people would find “gastronomic objectivism” about such matters utterly implausible. Similarly, the fact that we seem to have perceptions of morality does not strongly support the claim that we experience it as objective, since we seem to have perceptions of the (comparative) goodness or badness of some foods’ tastes but many would find odd the claim that we experience the taste of food as in fact good or bad or that we typically presuppose objectivism about gastronomic value.¹⁵

Indeed, even if Mackie and other anti-objectivists are wrong about the reasonableness of ordinary people’s practices and beliefs (on the assumption that there are no moral facts), it may still be inappropriate to treat such people as presupposing moral objectivity, at least if by “presuppose” we mean to imply a sensitivity to the rightness or wrongness of these metaethical claims. The rightness or wrongness of those claims is unlikely to have had an impact on ordinary people’s thought. Still, if anti-objectivists are wrong about these matters, perhaps there is another sense in which ordinary people’s views and practices could be said to presuppose some form of objectivism: Their views would not be right and their practices would not make sense unless objectivism were correct.¹⁶ Presupposition in this second sense might be relevant to some (but not all) versions of the AME. But whether ordinary people’s views and practices in this second sense presuppose some form of objectivism is an important issue in the debate over moral objectivity. It is inappropriate to make a burden-setting presumption in that very debate rest on a particular resolution of a matter at issue between the two sides.

Undoubtedly, much more could be said about these questions.¹⁷ But it should already be clear that our moral experience is much more varied and complex than proponents of the AME have assumed. Rather than continue to investigate these matters here, I shall set them aside and ask whether experience of the sort gestured at above, even interpreted in a way reasonably favorable to objectivists, would by itself support objectivism. The answer, I will argue, is that it does not. To supplement it objectivists would need a favorable resolution of certain issues already central to the debate over moral objectivity. That being so, a presumption in favor of objectivist theories is not warranted.

3 Is the Presumption Justified?

Does a presumption even require a justification? It is by now a philosophical commonplace that we have to start somewhere. But we are not being asked to view the objectivist presumption as itself a philosophical starting point. We argue for the presumption on the basis of our moral experience, experience that is thought to be evidence for objectivism. That being the case, it seems fair to ask how the evidentiary support is supposed to work. Of course, we could make a further presumption here, that our moral experience is evidence for objectivism (even if we can’t explain how this evidence is supposed to work). However, that would seem rather ad hoc, to say the least. And, although those who appeal to moral experience often seem to be taking the evidentiary link for granted, there are hints of various attempts to support it. In what follows, I consider four related strategies for supporting the presumption. Although these appear to be the most promising, I argue that none of them is in fact successful.

3.1 The Best Explanation

The most prominent approach suggests that objectivism allows us to best explain the nature of our moral experience. For example, Brink argues that we can most readily explain our moral thought and behavior on the hypothesis that moral inquiry is directed at discovering independently-holding moral facts, and that this supports moral realism:

...general considerations about the nature of inquiry and considerations about moral inquiry in particular are most easily explained on the assumption that moral inquiry is directed at discovering moral facts that obtain independently of our moral beliefs and at arriving at evidence-independent true moral beliefs. I take this to establish a presumptive case in favor of moral realism and to shift the burden of proof to the moral antirealist (Brink 1989, p. 36).

But the claim that moral inquiry is directed at discovering moral facts, even if true, does not by itself support the claim that such facts exist, any more than the fact that prayer is directed at God supports theism. The most obvious way to supplement it would be with a second explanatory claim—that the best explanation for the direction of moral inquiry involves the hypothesis that realism (or objectivism) is true. Brink may have something of this sort in mind when he suggests that, “features of actual and possible moral inquiry are hard to understand on antirealist assumptions and much easier to understand on realist assumptions. Realism and realism alone,” he says, “provides a natural explanation or justification of the way we do and can conduct ourselves in moral thought and inquiry.” (Brink 1989, p. 24).

This argument, however, seems to run explanation and justification together. The fact that we need to understand our moral experience does not support the claim that we must show it to make sense. But conflating explanation and justification is fairly common, and this may misleadingly boost the appeal of the AME. It is often facilitated by speaking of the need to “make sense of” our moral experience.¹⁸ We do need to understand that experience. But the claim that our moral practices are most naturally justified on the hypothesis that moral realism (or objectivism) is true is controversial, and even if correct, could not support moral realism (objectivism)—not, that is, without the added premise that our practices are typically justified, or something of that sort. Such a premise would be question-begging in this context since anti-objectivists believe that insofar as these practices presuppose that morality is a realm of fact they are not justified.

What then of the claim that our moral practices can be explained more naturally or easily on the assumption that moral realism is true? Brink seems to be assuming that what he calls a more natural or easier explanation is a better explanation, and thus that we can best explain our moral practices, experiences, and beliefs on the assumption that moral realism is true. But he does not spell out how such an argument would proceed. One possibility (expressed in terms of objectivism) is as follows: If we experience morality as a realm of fact, there must be some explanation for that. The easier and more natural (and thus best) explanation is that morality is such a realm and that we are familiar with it as such. We think moral beliefs can be mistaken or correct because they can be mistaken or correct; we deliberate because there are answers. And so on.

But if we are to accept the objectivist’s explanation as best, we need more than the mere claim that we experience morality as objective because it is objective. We need to know how its being objective can explain our having the experiences we do.¹⁹ Or at least we need some good reason for thinking that its being objective would explain why our experience turns out as it does. But such a reason has not been given in the context of this argument, and we cannot simply assume that one is available.

More importantly, whether moral facts help to explain our moral experience is a special case of one of the central controversies in metaethics: whether moral facts figure in the best explanation of anything.²⁰ It is inappropriate to make a presumption in favor of moral objectivity depend on the outcome of a major controversy in the debate over whether morality is objective. If the explanatory claim were to succeed then we would be beyond the point where we needed the presumption, and if it were to fail then the presumption could not be grounded on its success.

It might be tempting to think that a demand that moral facts be shown to figure in best explanations sets the bar too high for a mere presumption. The presumption is only meant to set the burden of proof, not to establish that objectivism is correct. Shouldn’t a showing that a particular explanation is the most initially appealing be sufficient? Perhaps that is what Brink has in mind when he says that realism provides an easier and more natural explanation for our moral experience. The easier and more natural explanation, he could argue, is good enough to support a presumption in realism’s favor. But it could be overridden if a better explanation were ultimately to be discovered.

The problem with this line of thinking is that whether or not realism provides the most initially appealing explanation is itself in issue between objectivists and anti-objectivists. First, it is not agreed that realism (or objectivism) does provide an easier and more natural explanation of our moral experience. Second, even assuming that easiness and naturalness are explanatory virtues, they are certainly not the only such virtues. Anti-objectivists are likely to point to others – such as parsimony or fit with the best current psychological or evolutionary theory – as supporting their own explanations.²¹ Here again, the presumption cannot be grounded in a claim that reflects an unresolved issue in the debate.

3.2 Epistemic Conservatism

Another possible ground for the presumption focuses mainly on one particular aspect of our moral experience—our beliefs. Many philosophers have felt that, other things being equal, we should favor the theory that requires us to give up as few of our current beliefs as possible.²² It is not clear, however, that the conservative strategy could work here, even in its own terms. If it is admitted that ordinary people are not explicit objectivists, then for them at least, a belief in objectivism itself is not available for conserving. Still, ordinary people do have beliefs objectivists might think cannot be true unless objectivism is true. And, it could be argued, objectivism allows us to conserve more of these beliefs than do anti-objectivist approaches such as the error theory.

For example, most people believe that our moral thinking is fallible. Objectivists often allege that there is no room for this to be true within an anti-objectivist framework. If this allegation were correct, then a principle of conservatism might have us grant some credibility to objectivism, since objectivism would follow in some sense from things we already believe—things the conservative approach tells us to treat as having some credibility. Likewise, the objectivist could argue, our moral beliefs themselves cannot be correct unless objectivism is correct. If we are to conserve these beliefs, they could say, we must accept some form of objectivism.

But the principle of conservatism is quite controversial, and for good reason. It is implausible that the mere fact that we already believe something should by itself give that belief any warrant, quite apart from any other beliefs we might have about the reliability of the processes that led to the formation of that initial belief. Suppose I believe that the next Lotto number drawn will be even, but suppose further that I have no actual evidence for that hypothesis. I should believe that the odds are 50/50, in spite of the fact that I find myself with the ungrounded belief that they are not. That I do believe the number will be even gives me no warrant whatsoever for believing it. Indeed, even if I knew that most ordinary people had the same belief, this too would not by itself give me any evidence that it is true.²³

Conservatives might try to restrict the range of beliefs to be conserved, so as to avoid counterexamples of this sort, for example by pointing out that we cannot see any way that my belief about the Lotto could be explanatorily connected to its own truth. In other cases, we can see such a connection. For example, I believe that the Battle of Hastings was fought in 1066, though I cannot remember my evidence for this belief. Still, I have reason to maintain the belief since it is likely that it was produced by a reliable process (such as my having read a history book). But such a reason is lacking in the case of my belief about the Lotto. Perhaps, then, we should conserve only those beliefs for which a plausible credibility-imparting explanatory story regarding their provenance can be told.

Anti-conservatives will no doubt argue that to adopt an approach like this is to abandon conservatism entirely. But we need not enter into that debate here, for, returning to our previous examples, the restricted conservative strategy would require us to find a credibility-imparting explanation for the widespread belief in moral fallibility, or for our moral beliefs themselves—a showing that these beliefs are plausibly explained by the states of affairs that make them true. And that is just another instance of the question whether moral facts figure in the best explanation of anything. As we have just seen, resolving this question cannot be a prerequisite for grounding an independent presumption in favor of objectivism.

There may be other ways of restricting the range of beliefs to be conserved, in order to avoid the implausible implications of wholesale conservatism. But if so, and if proponents of the AME wish to rely on them, they need to produce them. And they must show us that these approaches are not themselves disguised versions of the inference to the best explanation strategy.

3.3 The Principle of Credulity

A third suggestion relies on another principle it is sometimes thought we must accept if we are to avoid falling into skepticism. This principle, often referred to as the Principle of Credulity, holds that we are entitled to presume that things are pretty much as they seem to be, unless there is substantial evidence to the contrary.²⁴ Just as the appearance of an external world is sometimes thought to entitle us to presume that there is such a world, the appearance that there are moral facts is thought to entitle us to believe that there are such facts.²⁵

Even if we needed such a presumption in order to deflect external-world skepticism, however, that would not by itself give us reason to adopt a similar, but distinct, presumption to get us out of moral skepticism. Perhaps external-world skepticism is so implausible that we would never be satisfied unless we were to deny it. But that would not establish that moral skepticism involves a price high enough to make its avoidance imperative as well. (Indeed, most anti-objectivists think that the price is nowhere near as high, if there is any price to be paid at all.) Instead, the suggestion seems to be that a very general presumption that things are as they appear (barring substantial evidence to the contrary) is necessary for avoiding external-world skepticism, and that the very same presumption also has implications for the moral case.

Now, given the implausibility of attributing a belief in objectivism to most non-philosophers, it would be inappropriate to claim that morality seems objective to the average person. So here again we may feel compelled to rely on other pre-philosophical beliefs thought by objectivists to be best accommodated within an objectivist framework. For example, since it seems as though we can reason about morality, we might be entitled to believe that we can reason about it, in the absence of substantial evidence to the contrary. If objectivism in fact does best accommodate this belief and others like it, a reasonable extension of the strategy might hold that we are entitled to accept it, at least provisionally. If morality is not a realm of fact, the argument would go, then things are not as they appear.

Many people would find the kind of anti-skeptical strategy described above very unappealing, however, even in the external world case. If there are arguments against skepticism that are not based on the Principle, they would say, then we can examine them. But if they are unsuccessful, then we should let the chips fall where they may. Unless the Principle has some significant appeal beyond the avoidance of skepticism, they would argue, we are not entitled to adopt it for that purpose alone. By itself, the Principle seems question-begging as a response to skepticism.

In fact, the Principle of Credulity is implausible for precisely the reason conservatism (likewise commonly defended as necessary for avoiding skepticism) is implausible. Seemings in and of themselves give us no more warrant than mere believings. And just as the scope of conservatism’s reach would have to be restricted in order to avoid counterexamples such as my belief about the next Lotto number, so the Principle of Credulity would have to be restricted as well. Suppose it seems to me that the next Lotto number drawn will be even. Unless there is a plausible credibility-imparting explanation for things seeming this way – a showing that this seeming is explanatorily related to the truth of the beliefs it generates – there is no reason to treat it as providing any epistemic warrant whatsoever. In the moral case it would have to be shown that moral facts figure in the best explanations of these seemings, a matter, as before, at issue between objectivists and anti-objectivists. Such a showing cannot be a prerequisite for an independent presumption in favor of objectivism.

The anti-objectivist about morality can take no comfort in the defeat of a more general anti-skeptical strategy, however. For, as objectivists often point out, moral skepticism is only interesting if it is a distinctive form of skepticism. It would be no victory at all for the anti-objectivist to show that there is no more reason for believing in moral facts than there is for believing in an external world! What is wanted is some justification for treating moral properties differently. This suggests a familiar constraint for anti-objectivists: They should not rely on arguments that provide as much support for external-world skepticism as they do for moral skepticism. But there is an asymmetry here. For it would not be unreasonable to hold that even if external-world skepticism turns out to be false there is still no good reason for believing in moral facts. For example, even if we believe that the best explanation for the appearance of an external world is that there is an external world, nothing about the moral case follows. Thus anti-objectivists can reject the unrestricted Principle of Credulity as a way out of either sort of skepticism, remaining agnostic (at worst) about external-world skepticism.

It could be claimed that any route out of external-world skepticism would apply equally well to moral skepticism. But that claim would need to be defended, and such a defense has not been produced, nor does it seem obvious that it could be. In fact, the unrestricted Principle of Credulity is not the only anti-skeptical strategy worth considering. For example, a more modest presumption that our senses and memory are reasonably reliable would still allow us to avoid external-world skepticism but it would not have the implication that morality is as it appears to be, since the appearances relied upon in the moral case do not result (primarily) from the operation of these faculties.²⁶

3.4 Wide Reflective Equilibrium

A final strategy, also closely related to conservatism, relies on a more general epistemic principle having to do with considerations of theoretical coherence. It is hard to see how we could avoid appealing to such considerations in evaluating theories. Just as it is often thought that good moral thinking involves a search for “wide reflective equilibrium” (WRE), good reasoning in general seems to involve such a search as well. We appear to have no choice but to start off with the beliefs we have, not all of which cohere with one another. Our aim, it is often thought, should be to bring our beliefs into a coherent balance after careful consideration. Arguably, such a balance provides the best sort of justification available to us.

How might this approach to justification favor a presumption of objectivism, given the nature of our moral experience? Here again the focus is primarily on our beliefs. It might seem as though the approach commits us to the view that even our prereflective beliefs about the nature of morality (such as our belief in the possibility of moral reasoning) have some initial warrant, warrant they carry with them into the balancing process. After all, it could be asked, how could beliefs that have no warrant, even when made coherent upon reflection, produce beliefs that are warranted? Such an argument would not rest on the false assumption that nothing can have properties different than the properties of its parts. Instead it would be an application of what is sometimes called the garbage in, garbage out (GIGO) principle. Unless the beliefs we start out with have some warrant, the argument would go, they have nothing to pass on to the beliefs that emerge in WRE. Their initial warrant, on this view, might be outweighed. But all of our initial beliefs – including those thought by objectivists to be most compatible with objectivism – have some warrant, since the beliefs that emerge in WRE have warrant.

This argument seems highly suspect, however, for it opens a back door to epistemic conservatism by once again treating mere belief as sufficient for warrant. According to this approach, I should again treat my belief that the next Lotto number picked will be even as having some warrant, because the WRE approach requires me to treat all of my initial beliefs as having some warrant. Yet the claim that belief is sufficient for some degree of warrant is just as implausible when put in terms of WRE as it was when put in terms of conservatism. Once again, the mere fact that many of us have beliefs about the possibility of moral reasoning (and the like) does not in itself show those beliefs to have any warrant.

Is there any other way in which WRE might be brought in to support the presumption? A hint may be found in Ronald Dworkin’s response to Crispin Wright’s claim that realism about a particular domain has the burden of proof:

Unlike an argument in a court of law, however, the course of a philosophical investigation is fixed not by any free-standing methodological postulates, like Occam’s razor which Wright cites, but by how opinion stands when the investigation begins. No skeptical argument can succeed, for anyone, unless it brings him skeptical conviction, and that means that none of us can accept such an argument unless we find its premises convincing even when we grasp their skeptical import. We must find these premises more plausible than what they require us to abandon (Dworkin 1996, p. 117).²⁷

Expanding on Dworkin, we might argue as follows: Either our current moral beliefs (some of them, at least) are justified (enough to be presumptively true) or they are not. If they are justified, then objectivism must be justified as well, since the beliefs can’t be right unless objectivism is right. If they are not justified now, then once we have gotten appropriately close to WRE, our web of belief will either contain those moral beliefs or not. If it doesn’t, that will be because there appeared to us to be some reason to take them out. But that is just to say that our moral beliefs (and hence objectivism) are presumptively true. We are entitled to accept them, the argument would go, unless there is some reason for rejecting them. Such a reason would indeed defeat the presumption in favor of objectivist theories, but in its absence, the presumption holds.

Such an argument, however, would be in one sense too strong, and in another too weak. If it were meant to show our initial moral beliefs to have some (defeasible) epistemic warrant, it would fail. For it is true of all of our current beliefs that we would not abandon them, upon reflection, absent a (putative) reason for doing so. But that does not mean that each of our beliefs is presumptively true. To think otherwise is once again to allow conservatism to sneak in, since it is to hold that any belief is to some degree warranted merely in virtue of being held.

Alternatively, if the argument is not meant to show that these beliefs have warrant, then it cannot show that the burden of proof is on the anti-objectivist, as proponents of the AME claim. On this reading, what looked like an argument is merely a description of what many people happen to believe (and perhaps a prediction that they are unlikely to give most of those beliefs up in response to philosophical arguments). It would be an overstatement to call such beliefs presumptively true. At best, we could say that they have been presumed to be true.

Indeed, Dworkin himself denies that the considerations he points to show where the burden of proof lies or, “that our convictions are right just because we find them irresistible, or that our inability to think anything else is a reason or ground or argument supporting our judgment” (Dworkin 1996, p. 118). No doubt most people do have moral beliefs, and it would take a reason (or at least a putative reason) for them to give these beliefs up. But if we reject conservatism, we will hold that this alone does not give those beliefs any positive epistemic status. Whatever else is needed in order for them to earn that status is needed by both our current beliefs and any competitors. There is no asymmetry.

Still, while it is implausible to think that individual beliefs have warrant merely because they are held, perhaps we have no alternative but to treat as warranted certain theories that unify and accommodate those beliefs. Where else could we look for justification? If objectivism does indeed unify and accommodate the various theoretical and substantive beliefs about morality that in part constitute our moral experience, then perhaps we should provisionally accept it.

Conservatism, however, is no more plausible at the level of theories than it is at the level of individual beliefs. Suppose, to take a silly example, the theory that this bus has been commandeered by Martians nicely unifies and accommodates my beliefs about each of the various passengers, that they are non-human and that they are from a nearby planet, for example. If the beliefs about the passengers are unwarranted (as they would be if their only warrant came from my holding them), then a theory that unifies and accommodates them is (without more) unwarranted as well.

But how, without lapsing into complete skepticism, can those employing the method of WRE avoid holding that sooner or later coherence with mere beliefs is warrant conferring? Is there a way to look at the WRE strategy that does not have conservatism’s implausible implications? Perhaps. Our initial beliefs represent the beginning of our search for WRE, not the end point. Often, these initial beliefs will have to be modified or rejected if the entire corpus of our beliefs is to be made coherent. Thus arguably we should treat only those beliefs that are able to survive significant reflection as having some degree of warrant, and not those that merely appeared plausible before we began to think things through more carefully. If this is correct, then acceptance of the WRE framework would not automatically allow us to treat an initially appealing theoretical gloss on our moral beliefs as warranted. The claim that it is warranted would depend, on this approach, on whether it could survive the minimally necessary process of reflection.

How then should we envision this process of reflection? One answer, similar to a proposal gestured at above, begins with the claim that we have justified beliefs about the reliability of the processes that led to our beliefs. These beliefs about reliability might be thought to have a bearing on the question of how much credibility the initial beliefs should be given. For example, if a person’s commitment to the existence of angels stems entirely from dogmatically accepted religious beliefs he now sees reason to abandon, he may also have reason to avoid giving his belief in angels any weight in WRE. In contrast, arguably, we should see our perceptual beliefs as warranted when they are made under normal circumstances, since we are justified in believing that perception is normally a reliable belief producer. If this is correct, then the WRE strategy cannot yet support objectivism, since it would be question-begging to assume that our beliefs about morality (and our moral beliefs themselves) are the products of similarly reliable mechanisms.

In any case, this approach to WRE seems problematic since it is not immediately clear how our beliefs about reliability themselves get justified. An alternative might be to look for justification in the way our beliefs fit into our larger explanatory picture. Thus, we might treat as warranted only those beliefs whose existence can plausibly be explained in a way that supports the claim that they are true. Finding a plausible explanatory theory is often thought to be the point of WRE. On such a view, the coherence of a set of beliefs is constituted in part by the beliefs’ fit into such an explanatory picture.²⁸

How would the beliefs we have been discussing fare under this proposal? My belief that the next Lotto number drawn will be even cannot plausibly be thought to be best explained on the hypothesis that the number will in fact be even, nor can it be supported by any other aspect of a plausible explanation for my coming to have it. The same is true of my beliefs about my fellow bus passengers, and the theory that unifies and accommodates these beliefs. What about the everyday beliefs pointed to by those employing the AME? These would give objectivism some measure of warrant only if the best explanation for our holding them makes it more probable that objectivism is true. But once again whether such an explanation is available is merely a special case of the question of whether moral facts figure in the best explanation of anything, a question not appropriately considered in this context.

Presumably, other ways of looking at the WRE strategy are available, and perhaps some of them can also avoid the implausible implications of conservatism, either at the level of beliefs or at the level of theories that purport to accommodate those beliefs. But however we interpret it, the WRE strategy should not treat our beliefs – and a fortiori should not treat controversial theories that attempt to explain those beliefs – as having warrant simply because the beliefs are held. Instead the strategy should hold that the reflective process itself (or at least, the availability of such a process) is what confers warrant on some, but only some, of those initial beliefs. Whether it does so in the case under consideration here is a question that has not yet been answered.

4 Conclusion

The AME has exerted an enormous influence on the debate between objectivists and anti-objectivists. Yet its underpinnings have never been carefully examined. If I am right, objectivists have unfairly attempted to shift the burden of proof to anti-objectivists, and anti-objectivists like Mackie have been all too quick to accept this burden. Unless someone does a better job of defending the AME, we should treat objectivism and anti-objectivism as, at a minimum, starting on an even footing.