Kosher Status: Eleazar is an Orthodox Jew considering whether or not two foods have the same status in kashrut (Jewish dietary law). Although he could figure it out for himself with some contemplation and study, he decides instead to ask his friend, Esther, who he deems reliable about all things having to do with kashrut. Esther says that the two foods are both kosher, and Eleazar comes to believe and act accordingly, on the basis of her testimony.

Moral Status: Eleazar is a conscientious person considering whether or not men and women have equal moral status. Although he could figure it out for himself with some contemplation and study, he decides instead to ask his friend, Esther, who he deems reliable about all things having to do with morality. Esther says that men and women have equal moral status, and Eleazar comes to believe and act accordingly, on the basis of her testimony.

One of the driving intuitions behind the literature on deference to moral testimony is that there seems to be something inappropriate about moral deference per se.Footnote 1 Several philosophers note that this intuition can function in powerful arguments against moral realism and its constitutive theories. According to these arguments, realism and its constitutive theories seem to falsely predict that there is nothing inappropriate about moral deference per se.Footnote 2 In this essay I show that these arguments all face the same serious problem: their use of the evaluative predicate “inappropriate” is ambiguous, and this matters because the most plausible disambiguation yields an invalid argument. It turns out that the inappropriateness of moral deference likely cannot even in principle count against moral realism and its constitutive theories.

1 The metaethical challenge of moral deference

It is unremarkable that we can appropriately defer to reliable testifiers about various matters of fact, both mundane and arcane—for example, where to find the closest coffee shop, how the periodic table is organized, who has the longest tenure on the United States Supreme Court, whether Gödel’s Incompleteness Theorem is proven, and so on. After all, these are questions about objective, mind-independent fact to which different people may have different degrees of reliable access, and nothing in principle blocks reliable testifiers from making intelligible assertions that count as answers to these questions, which can then be received by others. In domains like these, we have reason to expect that, in everyday life, we can appropriately defer to testimony.

Put simply, the metaethical challenge of moral deference is that moral realism and its constitutive theses, especially cognitivism and objectivism, falsely predict that there is nothing inappropriate about moral deference per se. These metaethical theories describe the moral domain in such a way that we would expect deference to moral testimony to be just as appropriate, and in parallel circumstances, as deference to non-moral testimony about matters of objective fact. Yet, contrasts between cases like kosher status and moral status generate the widespread intuition that there is in fact something inappropriate about moral deference per se.Footnote 3 And so realism and its constitutive theories face disconfirmation.

As I use the term, ‘cognitivism’ in metaethics is the thesis that moral judgments, unlike judgments of gustatory preference, are beliefs; they are mental states that aim to get the world right. ‘Objectivism’ is the thesis that moral norms are the same from every point of view; the fundamental moral facts do not change relative to subjective standpoints. Both cognitivism and objectivism are components of moral realism, which I take to be the thesis that there are objective moral facts, many of which we know, and that we can believe and express moral propositions that are true or false in virtue of these facts.Footnote 4 These facts are also mind- or stance-independent in the sense that they do not exist in virtue of, and are not produced or constructed from, individual minds or collections of minds. Any defense of moral realism will in turn serve as a defense of objectivism and cognitivism.

According to moral realism, then, the moral domain has exactly those features that underwrite appropriate deference to testimony in realistic, non-moral domains. This fact is over-determined: several of the properties that realism predicates of morality are independently sufficient to predict with some degree of confidence that we can appropriately defer to moral testimony. It seems that, for example, cognitive mental states, objective truths, and mind-independent facts are each apt for the practice of deference to testimony.

So, moral realism is faced with the following simple challenge, here adapted from Sarah McGrath (2011):

  1. 1.

    If moral realism is true, then there is nothing inappropriate about moral deference per se.

  2. 2.

    There is something inappropriate about moral deference per se

    Therefore,

  3. 3.

    Moral realism is false.

Call this the “Realist Challenge.” The Realist Challenge can be made less ambitious by weakening the premises and conclusion. For example, one can argue that, given moral realism, deference to moral testimony is probably not inappropriate, and so the fact that it seems inappropriate presents some evidence against realism.

One might likewise target a constitutive thesis of moral realism, for example, cognitivism or objectivism. Consider a problem for cognitivism posed by Allison Hills (2015: 2–6). Here is a reconstruction of Hills’ argument:

  1. 1.

    If moral cognitivism is true, then there is nothing inappropriate about moral deference per se.

  2. 2.

    There is something inappropriate about moral deference per se.

    Therefore,

  3. 3.

    Moral cognitivism is false.

Call this the “Cognitivist Challenge.” The cognitivist challenge relies on the thought that “Anyone who accepts [cognitivism] about moral judgment thinks that … testimony can be the right kind of reason for a belief, and at the same time ‘the wrong kind of reason’ for forming a moral judgment” (Hills 2015, 6). The latter thought calls the first into question; in other words, the inappropriateness of moral deference provides evidence that moral judgments are not beliefs after all.

One might, finally, defend the challenge posed by Julia Diver (2006) that objectivism about moral norms predicts that deference to moral testimony is more appropriate than deference to testimony in a particular domain that is less objective than morality, namely, aesthetics. Here is an adaptation of Driver’s argument, which exhibits a somewhat different structure from the previous two arguments:

  1. 1.

    If moral objectivism is true, then moral deference is just as appropriate as aesthetic deference.

  2. 2.

    Moral deference is less appropriate than aesthetic deference

    Therefore,

  3. 3.

    Moral objectivism is false.

Call this the “Objectivist Challenge.” Driver’s Objectivist Challenge is comparative in nature; it has to do with how the objectivity of the moral domain compares to other domains. However, notice that the upshot for my purpose here is the same: a thesis constitutive of moral realism seems to be incorrectly optimistic vis-à-vis the appropriateness of moral deference.

Note that each of McGrath, Hills, and Driver develop these challenges in order to respond to them on behalf of the respective metaethical targets. But the point of this essay is that each of these arguments faces a problem that militates against even the need for such responses. For our purposes, we can focus on a maximally general and relatively weak form of argument, such that, if it fails, then so do its more particular and stronger cousins. Letting ‘M’ stand for some realist-friendly metaethical thesis, here is the form of argument that the rest of this essay will address:

  1. 1.

    M predicts that there is nothing inappropriate about moral deference per se.

  2. 2.

    There is something inappropriate about moral deference per se.

    Therefore,

  3. 3.

    M is disconfirmed by the inappropriateness of moral deference per se.

Call this the “General Challenge.” If the General Challenge fails in the way that I claim below, then so do any of its particular instances, as well as any of its stronger iterations. After all, if no metaethical thesis of this sort is even so much as disconfirmed by the inappropriateness of moral deference, then stronger, deductive arguments, including the Realist, Cognitivist, and Objectivist Challenges, must fail.

2 Two kinds of inappropriateness

We can distinguish between two kinds of inappropriateness. On the one hand, something might be epistemically inappropriate; on the other hand, it might be non-epistemically inappropriate—for example, it might be morally or otherwise practically inappropriate. These two kinds of inappropriateness can come apart.

Some simple examples will serve to illustrate both points, which I intend to be relatively banal. Consider the claim that, if you want to learn everything that some person does, then it is not inappropriate to place them under constant surveillance. The sense of “inappropriate” that makes this claim plausible is the epistemic sense. Constant surveillance is an epistemically appropriate means to learning about what a person is doing. However, under normal conditions, constant surveillance is also a morally inappropriate violation of privacy. So, something can fail to be epistemically inappropriate yet still be morally inappropriate.

Now consider the claim that it is not inappropriate to acquire a false belief in order to save someone’s life. The sense of “inappropriate” that makes this claim plausible is the moral sense. The claim is true when saving someone’s life is a compelling but distinctly moral reason to acquire a false belief. However, under normal conditions, intentionally acquiring a false belief constitutes an epistemically inappropriate way of exercising one’s rational faculties. So something can fail to be morally inappropriate yet still be epistemically inappropriate.

This distinction is essential for understanding why certain kinds of deceptively simple arguments do not falsify certain otherwise plausible doctrines. For example, consider the following facile argument against evidentialism:

  1. 1.

    If evidentialism is true, then it is inappropriate to believe something solely for pragmatic reasons.

  2. 2.

    It is sometimes not inappropriate to believe something solely for pragmatic reasons.

    Therefore,

  3. 3.

    Evidentialism is false.

Call this argument the “Ambiguous Anti-Evidentialist Argument.” The most plausible disambiguation of “inappropriate” in premise 1 of the Ambiguous Anti-Evidentialist Argument is epistemic, whereas the most plausible reading of “inappropriate” in premise 2 is not epistemic, but moral or practical. Hence, the argument, at least in its most plausible disambiguated form, turns out to be invalid. Disambiguating the terms explicitly makes the invalidity of the argument obvious:

  1. 1.

    If evidentialism is true, then it is epistemically inappropriate to believe something solely for pragmatic reasons.

  2. 2.

    It is sometimes not morally inappropriate to believe something solely for pragmatic reasons.

    Therefore,

  3. 3.

    Evidentialism is false.

Of course, it is open to a defender of the Ambiguous Anti-Evidentialist Argument to claim either that the sense of “inappropriate” in premise 1 is moral, or that it is epistemic in premise 2, but these are both very implausible claims. So the defender of the Ambiguous Anti-Evidentialist Argument faces a kind of dilemma: either they must offer an invalid argument with very plausible premises or a valid argument with very implausible premises.

3 Epistemic inappropriateness and moral realism

The metaethical theses targeted by the General Challenge predict only that there is nothing epistemically inappropriate about moral deference. After all, metaethical theses like moral realism are general theses having to do with the metaphysics, epistemology, and semantics of the moral domain. They are properly silent on whether some particular first-order practice like moral deference is inappropriate in some other—for example, moral or practical—sense.Footnote 5

Consider some of the metaethical commitments that ground the rational prediction that deference to moral testimony is not inappropriate per se:

  1. (1)

    Objectivity Moral facts are objective.

  2. (2)

    Propositional truth There are true propositions with moral content that can be expressed via ordinary, descriptive assertions.

  3. (3)

    Reliability People can differ with respect to how reliable they are in their moral beliefs.

  4. (4)

    Independence Moral truths obtain independently of human minds, stances, and attitudes.

  5. (5)

    Progress We are increasingly on track in appreciating the moral domain over time.

Each of (1)–(5) are epistemically relevant claims about the moral domain, but each says very little, if anything, about what is non-epistemically good or bad, right or wrong, and appropriate or inappropriate. In particular, each is silent about whether deference to moral testimony is non-epistemically inappropriate. This is unsurprising. After all, whether a practice is epistemically inappropriate is determined by distinctly epistemic norms and aims: norms and aims having to do with epistemic phenomena and goods like belief, truth, justification, reliability, correct assertion, tracking reality, and knowledge. And these are just the phenomena and goods about which metaethical theories, such as moral realism, have something to say.

If this is right, then we can disambiguate premise 1 of the General Challenge in the following way:

  1. 1.

    M predicts that there is nothing epistemically inappropriate about moral deference per se.

  2. 2.

    There is something inappropriate about moral deference per se.

    Therefore,

  3. 3.

    M is disconfirmed by the inappropriateness of moral deference per se.

If the General Challenge is to be valid, then “inappropriate” in premise 2 must be given an epistemic reading.Footnote 6

First, a point about the literature: those who have offered instances of the General Challenge simply have not independently motivated the claim that the inappropriateness in premise 2 is epistemic rather than moral. Yet, any argument against moral realism or its constitutive theories that appeals to a failure to predict the inappropriateness of moral deference itself fails if it does not clarify the sense of “inappropriate” at issue. After all, if the inappropriateness of moral deference is not epistemic, then it cannot even in principle provide disconfirmation of moral realism. This is the first horn of the dilemma that faces anyone who offers a version of the General Challenge: if the inappropriateness predicated of moral deference is not epistemic, then it is of the wrong kind for posing a challenge to moral realism. Because of this, the extant metaethical challenges from moral deference do not establish their conclusions.

Second, and more substantively, it is very implausible that the inappropriateness in premise 2 is epistemic. Three main factors militate against an epistemic reading of premise 2. The first factor is purely theoretical, the second has to do with the intuitive data, and the third has to do with the best explanations on offer for the truth of premise 2.

Consider again the metaethical commitments that ground the rational prediction that there is nothing inappropriate about moral deference per se, listed above. Suppose that an agent is simply interested in moral truth—what prevents her from relying on reliable testimony for it? There is nothing in the standard cases of moral deference, for example, moral status, that serves to block agents from finding out about the reliability of others, and nothing in the transmission of mere moral information via the expression of propositions that could somehow undermine the truth of what is expressed. Without begging the question against the realist (for example, by saying that moral truths are not propositional truths), there is no reason to think that the truth or falsity of a non-moral proposition (for example, matzo is unleavened bread) survives testimonial passage, but that the truth or falsity of a moral proposition (for example, stealing is wrong) somehow vanishes or transforms via testimony. The fact that Smith asserts that stealing is wrong to Jones does not by itself plausibly threaten the truth of the proposition just because Jones happens to believe it on a testimonial basis. If mere truth is the goal, then the fact that a person’s testimony is especially likely to be true is an excellent reason to believe its content. We may criticize the recipient of such testimony in non-epistemic ways, but her behavior seems epistemically impeccable.

What goes for truth goes also for justification. Insofar as testimony from a justified believer confers justification in general, there is nothing about moral content by itself that somehow gets in the way of that conferral. Unless we presuppose the truth of some metaethical conception that says so, reliability does not somehow cease to be good evidence for truth in the moral domain. It is for reasons such as these that Julia Driver can write, of the chief epistemic good of knowledge, “A consensus has been reached that moral knowledge can be transmitted via testimony. That is, one’s belief, say, that ‘x is wrong’ can be true, and can be justified solely on the basis of the testimony of a reliable and trustworthy authority” (Driver 2015, 27).Footnote 7 Justification and truth do not exhaust the desiderata likely relevant to knowledge, but Driver’s point applies to whatever other epistemic component one thinks is important—for example, proper function or reliability.

The second factor that militates against an epistemic reading of premise 2 is phenomenological and is also part of the explanation for the consensus identified by Driver. If there were something epistemically wrong with moral deference, then we should expect our intuitive reactions to the central cases to be different from what they are. The overwhelming sense one gets from the cases is that they reveal something questionable about the agent or the agent’s behavior. The cases do not resemble, say, Gettier cases, in which there is a widespread intuition that some specific epistemic good, for example, knowledge, is lacking. In those cases, the presence of epistemic luck directly drives the intuition that there is something distinctly epistemically amiss, with (as far as I know) no one thinking that there is anything non-epistemically (say, morally) amiss in these cases. But in cases of moral deference, there is no such distinctly epistemic feature. And this leads us to look not to metaethics for an explanation of the data, but somewhere else—plausibly, normative moral theory.

This leads us to the third and final factor that militates against an epistemic reading of premise 2: the best candidate explanations for the truth of premise 2 on offer in the extant literature are not epistemic. Of the explanations for the inappropriateness of moral deference that are apparently epistemic, perhaps the most popular and plausible is the alleged requirement that moral agents achieve “moral understanding” beyond mere moral knowledge.Footnote 8 According to defenders of this requirement, the moral domain is distinctive in that, for any p with moral content, those who judge that p ought not only know that p is the case, but know why p is the case.Footnote 9 And moral understanding of p, so the claim goes, is not transferrable via merely testifying that p.

While the requirement for moral understanding is an epistemic requirement in the sense that it demands something distinctly epistemic of moral agents (viz., knowing a reason), what matters for purposes here is whether the violation of this requirement constitutes an epistemic or moral failing. Given the justifications for the understanding requirement in extant literature, it seems that failures of moral understanding are always moral or practical failures. This derives from the fact that the most plausible justifications for the requirement are themselves distinctly moral or practical, not epistemic.Footnote 10 Morality is extremely important, and it matters, practically and morally, whether we fully understand it strictures. Moreover, understanding the underlying moral reasons for moral truths may help us to dynamically apply our moral knowledge in a way that we would otherwise be too narrowly informed to do.Footnote 11 Unlike the ordinary person’s knowledge of chemistry or history, it is important that we are able to apply our moral knowledge to diverse and new situations, day-by-day, moment-by-moment. Something beyond mere knowledge, resembling expert knowledge, is apt for this kind of dynamic application.

Alternatively, perhaps mere moral knowledge is not sufficient for worthy or virtuous action. Actions done on the basis of knowing why they are good might be more morally valuable than actions done merely on the basis of knowing that there is such a reason, and this is facilitated by moral understanding.Footnote 12 Some version of the claim that moral agents ought not only do the right thing, but ought to do it for the right reason or in the right way, is one of the more widely embraced theses in the history of moral philosophy. It seems that moral understanding is apt for the satisfaction of such a requirement. But again, these explanations for a requirement of understanding are decidedly non-epistemic—they refer to concerns squarely in the domain of normative moral theory. When a moral agent fails to have moral understanding, she is morally or practically criticizable.

In order to be an epistemic requirement in the relevant sense (and hence to pose a problem for moral realism), the demand for moral understanding would have to arise internally to the metaphysics, epistemology, and semantics of moral belief in a way unexplained by realism. Instead, the extant accounts of moral understanding just surveyed locate its origin in concerns external to the metaethical features of morality—especially, the distinctly moral or instrumental value of understanding moral matters for oneself.Footnote 13

Several of the other extant explanations for the intuitive data are straightforwardly non-epistemic, with no need for argument that they are so. For example, perhaps in the moral domain, figuring out moral matters for ourselves constitutes a morally valuable kind of autonomy.Footnote 14 When we believe and act on someone else’s say-so in moral matters, we fail to be fully morally autonomous. Even Karen Jones, who ultimately defends moral deference, provides multiple reasons for thinking that some resistance to moral deference is morally good. Here is what she says about her central case of Peter, whose female colleagues in a cooperative can, unlike him, see that a prospective new member is sexist:

Peter’s [refusal to morally defer] shows us the prima facie plausibility of the three considerations against accepting moral testimony: he shows a keen and admirable awareness of the importance of moral issues. Moreover, his desire to have, and to understand, morally adequate reasons for acting makes us fairly sure, that, should Peter fall into morally bad circumstances, he will not be following the leader into error. This sort of independence—the independence of those who want to keep their moral beliefs in step with what they take to be cogent reasons—gives us a sense of the notion of autonomy which shows why autonomy is thought to be something valuable. Finally, had Peter accepted the women’s testimony, he would still not have been any the wiser about how to go on to apply ‘sexism’ to new, nonegregious, cases (Jones 1999, 62).

All three of the prima facie considerations Jones offers against the appropriateness of moral deference are distinctly instrumental or moral rather than epistemic. Alternatively, perhaps Andreas Mogensen is right that moral deference interferes with a valuable kind of authenticity (Mogensen 2015, 15–20). When we acquire and act on the moral beliefs of others, we (in some sense that Mogensen leaves undefined) fail to figure matters out for ourselves in a domain that, unlike many non-moral domains, is “central to our identity” (Mogensen 2015, 18). If this proposal is right, then what’s inappropriate about moral deference has something do with practical norms deriving from our personally important identities. See Skarsaune (2016) for a different explanation that appeals to the value of authenticity. Yet another non-epistemic explanation is Alan Hazlett’s appeal to the “social goods” achieved by a resistance to deference in moral (among other) matters: diversity and dialogue (Hazlett 2017, 59–64). The point here is not to delve into which, if any, of the extant non-epistemic explanations are correct, but to show more generally that as long as some non-epistemic explanation is correct, then the General Challenge will not go through. Put simply, realism and its constitutive theses are properly silent on these kinds of value questions.

So there are three powerful lines of evidence that the epistemic reading of premise 2 is very implausible: theoretical, intuitive, and by implication of the best explanations of the data. We have now seen the full dilemma that faces the defender of the General Challenge. Given the epistemic disambiguation of premise 1, the defender of the General Challenge is guilty either of an equivocation or severe implausibility. If premise 2 is given a non-epistemic reading, then the argument is equivocal and hence invalid. On the other hand, if premise 2 is given an epistemic reading, then the argument is very implausible and likely unsound.

Finally, it is worth saying a word about an interesting, more general implication of the framework on offer here. If I am right that moral deference might be morally but not epistemically inappropriate, then it seems that our moral and epistemic obligations can conflict, resulting in certain kinds of practical or deliberative dilemmas. While this might cause trouble for normative moral theories that inextricably link morality to non-moral rationality (most notably, Kantian approaches), it should not worry the moral realist per se, any more than the conflicting epistemic and moral demands highlighted in the equivocal arguments of Sect. 2, above. For example, the fact that intentionally acquiring a false belief in order to save a life is epistemically but not morally inappropriate poses no problems for moral realism, though it may potentially pose problems for any first-order normative theory that interrelates epistemic and moral obligations such that they cannot come apart. Notice, too, that a mere linkage between epistemic rationality and morality is not necessarily inconsistent with the existence of conflicting duties. For example, Hills (2010) argues that moral and non-moral beliefs are both governed by epistemic standards determined by our purposes in those domains (performing morally worthy actions and achieving our ends, respectively). Adapted to Hills’ framework, the implication of the argument above is that following the norms that help us to achieve our ends may not always be maximally conducive to performing morally worthy actions, and vice versa.Footnote 15

4 Conclusion

I have not given an explanation for the widely accepted datum that there is something inappropriate about deferring to moral testimony. What I have done is show that extant arguments from the inappropriateness of moral deference to the falsity of moral realism cannot establish their conclusions without disambiguating the key evaluative term in the first two premises—“inappropriate.” I have further shown that anyone who disambiguates “inappropriate” faces a dilemma. On a non-epistemic reading, the argument is invalid, because moral realism only predicts that there is nothing epistemically inappropriate about moral deference. So, on this more plausible reading, the inappropriateness of moral deference cannot even in principle constitute a problem for realism. But on an epistemic reading, the second premise is almost certainly false, and hence the argument is unsound.