Against overgeneralisation objections to the argument from moral disagreement

According to the argument from moral disagreement, the existence of widespread or persistent moral disagreement is best explained by, and thus supports, the view that there are no objective moral truths. One of the most common charges against this argument is that it “overgeneralises”: it implausibly forces its proponents to also deny the existence of objective truths about certain matters of physics, history, philosophy, etc. (“companions in guilt” objections) or even about the argument’s own conclusion or its own soundness (“self-defeat” objections). My aim in this article is to provide a detailed clarification and assessment of this overgeneralisation charge. Various (mostly empirical) issues relevant to assessing the above objections have so far not been sufficiently investigated. Nevertheless, there are good reasons to believe that realists have exaggerated the significance of their overgeneralisation charge. Both its companions in guilt and its self-defeat versions likely fail.


Introduction
Many philosophers have been impressed by the apparent prevalence or persistence of moral disagreement. For example, while the Inuit are said to have left old and sick people for dead, most contemporary Westerners shudder to think about this practice. While some of us believe that abortion is permissible, others judge it to be impermissible. Often, conflicting moral judgements do not even seem to converge over time, thus constituting persistent disagreement. 1 From antiquity onwards, it has been speculated that the existence of widespread or persistent moral disagreement supports moral anti-realism, i.e. the view that there are no objective moral truths. 2 The most prominent argument to this effect has the form of an inference to the best explanation. 3 Why do some societies tolerate parricide, and others do not? Why do some individuals judge abortion to be permissible, while others judge it to be impermissible? Realism entails that in all such cases of (persistent) moral disagreement one of the disagreeing individuals fails to grasp the objective moral truth. 4 However, a much better explanation has sometimes claimed to be that there are no objective moral truths at all, and that moral disagreements consequently only arise because people differ in their upbringings, cultures, psychological dispositions or the like (e.g. Doris and Plakias 2008;Leiter 2009;2010a;Fraser and Hauser 2010;Mackie [1977] 2011). 1 As will be explained at the end of this article, the above examples of parricide and abortion may actually fail to qualify as genuine moral disagreements. 2 Note that moral realism and anti-realism are sometimes defined in ways other than the above metaphysical one (which has been proposed by, for example , Brink 1989;Joyce 2007;Miller 2009); for example, in semantic terms (e.g. Dummett 1978;Sayre-McCord 1988), or as entailing an epistemic condition (e.g. Boyd 1988). Many of my arguments below against overgeneralisation objections do not (fully) apply on such alternative understandings of realism and anti-realism. 3 For a helpful overview of different kinds of anti-realist arguments from moral disagreement, see Enoch (2009). For the history of such arguments, see Gowans (2000). Finally, it should not go unnoticed that some meta-ethicists (e.g. Strandberg 2004;Huemer 2005) have put forward disagreement-based arguments in favour of realism (rather than against it) as well. 4 To be precise, realism has this implication in cases in which the individuals' judgements logically contradict each other. Judgements such as "bailing out bank X is impermissible" and "bailing out bank X is obligatory" could both be wrong (if bailing out bank X is optional).
In what follows, I will abstract from this complication.

Pölzler 262
Here is how John Mackie, the most famous proponent of this anti-realist argument from moral disagreement, puts it: …radical differences between first order moral judgements make it difficult to treat those judgements as apprehensions of objective truths. Such variation…may indirectly support second order subjectivism: radical differences between first order moral judgements make it difficult to treat those judgements as apprehensions of objective truths…actual variations in the moral codes are more readily explained by the hypothesis that they reflect ways of life than by the hypothesis that they express perceptions, most of them seriously inadequate and badly distorted, of objective values (Mackie [1977(Mackie [ ] 2011. One of the main objections against this argument from moral disagreement has traditionally been that it "overgeneralises" (e.g . Brink 1989;Huemer 2005;Shafer-Landau 2006;Enoch 2009).
Suppose widespread or persistent moral disagreement really forced us to deny the existence of objective moral truths. In this case it seems that we would have to deny the existence of objective truths about various non-moral matters that exhibit widespread or persistent disagreement too, such as about (particular debates in) history, physics, philosophy, etc. (call this the "companions in guilt objection") and about the argument from moral disagreement's own conclusion or its own soundness (call this the "self-defeat objection"). But these implications seem absurd. Surely there are objective truths about physics, history and philosophy; and surely proponents of the argument from moral disagreement must accept the existence of objective truths about the existence of objective moral truths and the soundness of the argument from disagreement. Thus, proponents of the objection conclude that the argument from moral disagreement is flawed.
Their prominence notwithstanding, overgeneralisation objections have so far not been developed in much detail. Often, realists have only provided rather cursory formulations that were still taken to suffice to refute or weaken the argument from moral disagreement (for exemplary formulations, see Brink 1989;Enoch 2009). Anti-realists do not seem to have been able to come up with fully convincing replies to these objections either (for brief discussions, see Doris and Plakias 2008;Leiter 2009;2010a;Mackie [1977Mackie [ ] 2011Plakias 2011). My aim in this article thus is to provide one of the first thorough clarifications and assessments of overgeneralisation objections against the argument from moral disagreement.
It is important to note right at the outset that overgeneralisation objections have a conditional form. They claim that if the argument from moral disagreement were sound, then it would implausibly force us to adopt anti-realism about certain non-moral matters. My considerations about these objections therefore do not depend on the argument from moral disagreement actually being sound. They may also be accepted and deemed interesting by anyone who rejects this argument (that is, who rejects them for reasons other than overgeneralisation objections, e.g. because s/he believes that there is not much or much persistent moral disagreement at all).
How are overgeneralisation objections supposed to work then? And how compelling are they? My attempt to answer these questions involves three steps. In the first section, I will introduce what I take to be the most promising version of the argument from moral disagreement. In the second and third sections, I will then address companions in guilt and self-defeat objections against this argument. It will turn out that various (mostly empirical) issues relevant to assessing these objections have so far not been sufficiently investigated. Nevertheless, there are good reasons to believe that realists have exaggerated the significance of their overgeneralisation charge. Both its companions in guilt and its self-defeat versions likely fail.

The argument from disagreement
Whether the argument from moral disagreement overgeneralises strongly depends on how one understands this argument. On some versions (e.g. Doris and Plakias 2008) 5 the argument is indeed   5 Among others, Doris and Plakias (2008) reason as follows: science is a paradigmatically objective endeavour; morality involves much more fundamental disagreement than science; thus, morality is unlikely to be an objective endeavour. In order to refute this argument companions-in-guilt-style one would only have to come up with non-moral matters which also involve much more fundamental quite likely to overgeneralise. But of course, the success of overgeneralisation objections should not be measured against just any, but rather against the most promising construal of their target. In what follows, I will therefore introduce what I take to be the argument from moral disagreement's most plausible version. One helpful and common way of stating the argument from moral disagreement is as claiming that (P1) there is widespread or persistent moral disagreement, that (P2) this fact is best explained by moral anti-realism, and that (C) moral anti-realism is therefore likely true (e.g . Enoch 2009;Meyers 2013). 6 This formulation raises the following two questions: How widespread or persistent must moral disagreement be in order for the argument from moral disagreement to work? And why is anti-realism supposed to provide the best explanation of this disagreement?
Elsewhere (Pölzler 2018), I suggested that the correct answer to the first of these questions is contingent on one's answer to the second question. Moral disagreement must be widespread or persistent enough that anti-realism turns out to be the best explanation of it. This is because even very widespread or persistent moral disagreement could turn out to be compatible with moral realism (e.g. many people could fail to see the objective moral truths because they are partial or irrational, such as argued, for example, by Huemer 2005; Enoch 2009); and even very little or non-persistent moral disagreement could suffice (e.g. moral disagreement among people who are unlikely to be partial or irrational, such as philosophers, as argued by Leiter 2009;2010a).
If these considerations are correct, then specifying the required level of prevalence and persistence requires a conception of what makes an explanation of moral disagreement a good explanation, and an investigation of the extent to which moral realism and anti-realism meet these criteria. Philosophers widely agree that the quality of explanations importantly depends (at least inter alia) on the degree to which they exhibit the following three theoretical virtues: (1) ontological parsimony, (2) consilience, and (3) coherence (e.g. Harman 1968;Thagard 1978). 7 Proponents of the argument from moral disagreement have accordingly typically attempted to support the superiority of anti-realist explanations of widespread or persistent moral disagreement by showing that these explanations are more ontologically parsimonious, consilient and coherent than realist ones.
Consider, first, ontological parsimony and consilience (see Leiter 2009;2010a;Plakias 2011). Anti-realist explanations of moral disagreement exclusively involve non-moral hypotheses, such as Mackies' hypothesis that individuals' differing moral judgements arise from their differing "ways of lives" ([1977] 2011, 36-37). Realists, in contrast, also appeal to objective moral facts. The party that is right, they argue, is often right because s/he grasped the objective moral truth about what is at issue; and the party that is wrong is wrong because s/he failed to grasp that truth. Now suppose that these objective moral truths are taken to be non-natural (e.g. Shafer-Landau 2003;Huemer 2005) or irreducibly natural (e.g. Boyd 1988;Brink 1989). Then, realist explanations of moral disagreement are less ontologically parsimonious than anti-realist explanations (they are committed to an additional kind of facts) and they are less consilient (non-natural or irreducibly natural moral facts do not plausibly explain [many] phenomena apart from people's moral judgements, while facts about differing ways of lives explain many other phenomena too). disagreement than science, but which cannot plausibly be interpreted in anti-realist ways. And depending on precisely how much more disagreement is required this may be a rather easy task (just think, for example, of various areas of philosophy). 6 In recent years many proponents of the argument from moral disagreement have claimed anti-realism to provide the best explanation of a particular kind of widespread or persistent moral disagreement, namely disagreement under certain epistemic conditions (such as full information or rationality) (e.g. Doris and Plakias 2008; for related construals in more general debates about disagreement, especially in terms of "faultless disagreement", see e.g. Kölbel 2004). Discussions have accordingly often proceeded under the assumption that the only plausible way for realists to explain instances of (persistent) moral disagreement consists in showing that these epistemic conditions are not met. In fact, however, realism may have several other explanatory resources at its disposal as well (e.g. Sneddon 2009;Fitzpatrick 2014). I therefore prefer my above formulation of the argument from moral disagreement, which does not prejudge how realists can explain instances of (persistent) moral disagreement. 7 Roughly, ontological parsimony entails that an explanation is better than another if, other things being equal, it commits one to fewer (kinds of) entities; consilience entails that an explanation is better than another if, other things being equal, it explains more (kinds of) facts, and coherence entails that an explanation is better than another if, other things being equal, it is more consistent and its elements support each other more strongly.
Worries about ontological parsimony and consilience may be raised about realist explanations of literally any phenomenon. 8 The most promising argument for the superiority of anti-realist explanations of moral disagreement that is specific to the issue of moral disagreement focuses on coherence. According to realism, a moral judgement is true if and only if it correctly represents an objective moral fact (e.g. Brink 1989;Shafer-Landau 2003). For a moral fact to be objective means for it to be "fixed", in the sense that it does not vary with the mental states of subjects. Realism therefore entails that, contrary to judgements about taste, social conventions, etc., the moral judgements of disagreeing individuals cannot both represent the facts correctly. One of the parties' judgements must be false.
Of course, moral errors should occasionally be expected. If moral disagreement were really widespread or persistent, however, realism would imply that people err very often about moral matters or even if they have carefully reflected on them. This would raise the worry that the objective moral truths may be impossible for us to epistemically access, i.e. that we are unable to know these truths. Meta-ethicists widely agree that a scepticism of this kind provides strong evidence against realism (Brink 1989;Nagel 1989;Shafer-Landau 2012). Most plausibly, Rowland (2017) has defended this connection by arguing that unknowable moral truths could not be practical in the sense that they make a difference to what we ought to do; and that being practical in this sense is a necessary condition for the existence of moral truths (morality essentially is about what to do). Now that we have some understanding of the argument from moral disagreement, let us come back to the worry that this argument implausibly generalises to certain non-moral matters. There are three main ways in which moral anti-realists may attempt to address this worry. 9 First, they may try to show that the argument from moral disagreement does not commit them to anti-realism about the relevant non-moral matters because disagreement about these matters is not or may not be sufficiently widespread or persistent (call this the "empirical strategy"). Second, they may argue that the argument from moral disagreement does not commit them to anti-realism about the relevant non-moral matters because realism about these matters provides the best explanation of disagreement about these matters (call this the "defusing strategy"). And third, they may concede that the argument from moral disagreement forces them to adopt anti-realism about certain non-moral matters but argue that anti-realism about these matters is actually plausible (call this the "bullet biting strategy").
Realists have traditionally attacked the argument from moral disagreement on several fronts (e.g. Brink 1984;1989;Huemer 2005;Enoch 2009). There is no reason why varying combinations of responses should not also be employed in its defence. In order for any version of the overgeneralisation objection to succeed, it must therefore be the case that it resists any combination of the empirical, the defusing and the bullet biting strategy, as explained above. But do overgeneralisation objections meet this condition? In the next two sections, I will discuss this question first with regard to companions in guilt objections and then with regard to self-defeat objections.

Companions in guilt objections
Companions in guilt objections assume that much or persistent disagreement cannot only be found with regard to morality, but with regard to certain non-moral matters as well. 10 Common examples include: (1) history or particular debates within history, such as whether the Reformation in 16 th -century England was a top-down or bottom-up process (Huemer 2005;Wedgwood 2010); (2) physics or particular debates within physics, such as whether string theory is true (Brink 1989;8 As considerations about ontological parsimony and consilience apply so widely, one may hesitate to include them into the argument from moral disagreement. But proponents of this argument typically do not seem to understand it in a provisional sense. They believe that antirealism provides the best explanation of widespread or persistent moral disagreement all things considered. And the overall quality of an explanation depends on its ontological parsimony and consilience just as well as on its coherence (see Enoch 2009). 9 The first and second way simply mirror realist strategies against the argument from moral disagreement. I will come back to this fact in the conclusion of this article. 10 For general investigations of companion in guilt arguments in metaethics, see Cowie (2018) Non-moral matters differ in terms of the prevalence and persistence of disagreement about them. They also differ in terms of the plausibility of realism providing the best explanation of this disagreement and in terms of the plausibility of realism about them. 11 Thus, the strength of companions in guilt objections strongly depends on the particular non-moral matters that realists appeal to. My aim in what follows is not to decisively refute any particular version of these objections. Rather, I attempt to develop and support anti-realists' empirical, defusing and bullet biting strategies in general ways: by making it plausible that, once supplemented by more specific arguments and combined with each other, these strategies have (considerable) force against any companions in guilt objection that might be put forward against the argument from moral disagreement. I will do so by first discussing the empirical strategy, then the defusing strategy, and finally the bullet biting strategy.

The empirical strategy
Just as the argument from moral disagreement requires evidence for the prevalence or persistence of moral disagreement, proponents of companions in guilt objections owe us evidence for their empirical hypotheses about non-moral disagreements as well. For example, do scholars really widely or persistently disagree about the causes of the Reformation in 16 th -century England, about string theory or about humans having free will? With regard to many supposed companions in guilt, the evidence actually seems too sparse and controversial to allow for reliable conclusions about the prevalence and persistence of disagreement about them. First, just as in the case of morality, it is often unclear what would even count as (persistent) disagreement about the relevant non-moral matters. Some of this lack of clarity is due to the fact that it is difficult to tell what makes a disagreement persistent and what distinguishes the judgements of a particular non-moral debate or discipline from those of other debates or disciplines. 12 Another challenging conceptual task confronted by proponents of companions in guilt objections involves specifying what it means for discussants of a non-moral matter to disagree about this matter. Take, for example, philosophy. Understanding and progress in this discipline has traditionally resulted from a "culture of adversarial debate" (Wedgwood 2014, 37) which exaggerates disagreements. But if a determinist only assigns a probability of 45% to the hypothesis that humans have free will and an indeterminist only a probability of 55%, then is this already sufficient for regarding them as disagreeing with each other? 13 11 The labels "realism" and "anti-realism" have acquired significantly different meanings across different subject areas. For the sake of simplicity, I will nevertheless consistently assume our above understanding with regard to morality. By realism and anti-realism about any particular non-moral matter, I thus mean the views that there are or are not objective truths about this particular matter. 12 An obvious example for the latter of these complexities is philosophy. Many claims (e.g. about the causes of moral judgements or the mechanisms of political ideologies) are regarded as philosophical by some scholars and as non-philosophical (e.g. psychological or sociological) by others. 13 Again, as acknowledged above, the obvious point to make here (as well as in response to some of the following passages) is that the same Second, the extent to which any conception of (persistent) disagreement in some non-moral debate or discipline applies to the actual world is an empirical question. Answering this question is exacerbated by various factors. As suggested by the determinism/indeterminism case above, reading off (persistent) disagreement from textual (or also verbal or behavioural) data is bound to be controversial. Disagreements about a particular non-moral matter might also actually be grounded in disagreements about another matter. Thus, in order to determine the prevalence or persistence of disagreement about some non-moral matter detailed investigations are inevitable.
The problem is that for many of the non-moral matters that proponents of companions in guilt objections have appealed to such investigations have not yet been conducted. For example, nobody as far as I found has yet systematically studied historians' views about the causes of the Reformation in 16 th -century England or physicists' views about string theory in ways that address complications such as those mentioned above (e.g. reducibility of these disagreements to disagreements about other questions). Moreover, with regard to philosophy, although there have been some relevant studies (e.g. Philpapers 2012), they have so far been rather coarse-grained too, and were not fully representative in terms of their samples.

The defusing strategy
Suppose we do find much or persistent disagreement about some non-moral matter. Following the logic of the argument from moral disagreement, this finding does not by itself force proponents of the argument to become anti-realists about that non-moral matter. In order for this (potentially problematic) commitment to be entailed it must also be the case that anti-realism about this non-moral matter provides the best explanation of the disagreement about this matter. Unfortunately for proponents of companions in guilt objections, (persistent) disagreement about many of the non-moral matters that they have appealed to is better explained by realism than anti-realism about these matters -even if this disagreement is as widespread or persistent as disagreement about morality.
First, in contrast to some versions of moral realism, realist explanations of non-moral disagreements tend to be as ontologically parsimonious and consilient as anti-realist ones. Consider, for example, string theory. To be a realist about string theory entails explaining disagreements about the existence of strings in terms of an objective fact: some people are right and others are wrong about the existence of strings, because some people see and some fail to see the objective fact about their existence. But the fact that strings do or do not exist is an ordinary natural fact which does not inflate one's ontology, and it also likely explains various other kinds of facts apart from physicists' judgements about strings (for example, the fact that certain kinds of elementary particles do or do not stand in a relation of "supersymmetry" to each other). 14 Second, many realist explanations of non-moral disagreements are also perfectly coherent. Disagreement about these matters can typically be explained without invoking the epistemic inaccessibility of their truths. For example, historians disagree about the causes of the Reformation in 16 th -century England because the available historical sources do not allow any confident conclusion about these causes (and not because these causes cannot be known); physicists disagree about the truth of string theory because the technical equipment to test the predictions of this theory has not yet been developed (and not because the truth of this theory cannot be known); and so on (see Leiter 2010b; Mackie [1977] 2011). 15 But suppose disagreement about an objective non-moral matter did force us to regard this matter as epistemically inaccessible. Another reason why non-moral versions of realism typically more coherently explain disagreement, compared to moral realism, is that this implication often would not (strongly) undermine these non-moral versions of realism in the first place. In contrast to morality, holds true with regard to morality as well. I will address this worry in the conclusion of this article. 14 The claim that the existence or non-existence of strings explains facts such as the one mentioned above is controversial. Not much hinges on it (and similar claims in this section), though. For as pointed out above, my aim is not to refute any particular version of companion in guilt objections, but rather only to lay the theoretical foundation for responses to these objections. 15 Again, one might take issue with this claim about string theory. Some physicists have argued that string theory may be untestable (Wolchover 2016). many objective non-moral matters (such as history or physics) are not practical in the sense that they need to make a difference to what we ought to do. That these matters are epistemically inaccessible hence does not entail that there are no objective moral facts about them (given Rowland's abovementioned argument for the impossibility of sceptical moral realism). Finally, we also often have strong independent evidence for the existence of the objective non-moral truths at issue. For example, while proponents of the argument from moral disagreement are probably right that moral facts cannot plausibly be said to explain anything apart from our moral judgements (and maybe not even these judgements; see Harman 1977;, facts about the existence of strings figure in plausible explanations of other phenomena as well; while there is no presumption in favour of realism about morality (see Loeb 2007;Pölzler 2018;Pölzler and Wright 2020), we have a strong prima facie reason to be realists about historic events; and so on.

The bullet biting strategy
Our above considerations suggest that widespread or persistent disagreement about many non-moral matters is better explained by realism than anti-realism about these matters. But suppose proponents of the argument from moral disagreement do turn out to be committed to anti-realism about some non-moral matter. In order for any companion in guilt objection to succeed, there is still one further requirement that must be met. It must also be the case that this implication is problematic, i.e. that anti-realism about the relevant non-moral matter really is implausible.
At first sight this requirement may seem negligible. How could anybody deny, for example, that there are objective truths about matters such as the causes of the Reformation in 16 th -century England or the truth of string theory? However, not all supposed companions in guilt are as innocent as history and physics (see Lillehammer 2004). The metaphysical status of some philosophical debates, for example, is highly contested. Take claims about the nature of beauty. It does not seem implausible (and in fact quite likely) that these claims purport to represent subjective rather than objective facts. To the extent that interpretations such as these are plausible, and anti-realism about some non-moral matter is thus a serious contender, the fact that the argument from moral disagreement generalises to this matter cannot be (strongly) held against this argument at all.

Self-defeat objections
Let us now proceed to the second version of the overgeneralisation worry, i.e. to self-defeat objections. Self-defeat objections are based on the hypothesis that there is widespread or persistent disagreement about either or both of two particular non-moral matters: (1) about whether there are objective moral truths; or (2) about whether the argument from moral disagreement is sound (e.g. Huemer 2005;Shafer-Landau 2006;Enoch 2009). Suppose proponents of the argument from moral disagreement were right that widespread or persistent moral disagreement grounded anti-realism about morality. Then, some realists have claimed, in combination with the above hypothesis, this argument would entail that there also likely is no objective truth about whether there are objective moral truths ("anti-realism about the realism/anti-realism debate") and about whether the argument from moral disagreement is sound ("anti-realism about the argument from moral disagreement"). But these implications render the argument from moral disagreement self-defeating. Thus, the argument must be rejected.
Michael Huemer summarises this self-defeat version of overgeneralisation objections as follows: …if the argument from disagreement is sound, then it refutes itself, since many people do not agree with the argument from disagreement. The argument would likewise refute any metaethical position, due to the nature of disagreement in metaethics (Huemer 2005: 146).
Self-defeat objections are more promising than companions in guilt objections. Compared to many versions of the latter, anti-realists have a much harder time weakening them by biting the bullet or by criticising their empirical hypotheses about the prevalence and persistence of relevant disagreements. However, I will show that all things considered, self-defeat objections likely fail to seriously undermine the argument from moral disagreement as well. This is because disagreement about whether there are objective moral truths and about whether the argument from moral Pölzler 268 disagreement is sound is likely best explained by realism about these matters.

The bullet biting strategy
Some versions of companions in guilt objections are best met by showing how anti-realism about the supposed companion is actually plausible. The point and defining feature of self-defeat objections is their (purportedly) being immune against any such response. Proponents of the argument from moral disagreement, it is assumed, cannot adopt anti-realism about the realism/anti-realism debate and about the argument from moral disagreement. In my view this assumption of self-defeat objections is correct.
First, contrary to some recent suggestions (see e.g. Plakias 2011), the argument from moral disagreement cannot possibly be reconciled with anti-realism about the moral realism/anti-realism debate. In claiming that objective moral truths do or do not exist, both moral realists and anti-realists do not intend to express non-cognitive mental states, falsities or merely subjective truths (as anti-realism would have it). 16 They rather state that it is objectively true that there are or are no such truths. So sure, if there were no objective truths about the existence of objective moral truths, then moral realism would be defeated. But so would anti-realism be, and thus the argument from moral disagreement's own conclusion. 17 Second, proponents of the argument from moral disagreement also cannot concede that their endorsement of the soundness of this argument is only an expression of non-cognitive mental states, is false, or is only subjectively true. Doing so would undermine the argument from moral disagreement's dialectical function. Interpreted in such an anti-realist sense, the argument could not provide any of those meta-ethicists with a (strong) reason to deny the existence of objective moral truths to whom it is mainly addressed, namely meta-ethicists who are not yet convinced by the argument from moral disagreement (i.e. realists, meta-ethical agnostics, proponents of intermediate positions, etc.;Shafer-Landau 2006). 18 The empirical strategy The empirical response to the self-defeat objection attempts to show that disagreement about whether there are objective moral truths or about whether the argument from moral disagreement is sound is actually not (very) widespread or persistent. Above, I raised awareness for several complexities in establishing empirical hypotheses about disagreements. All of these complexities must be addressed by proponents of self-defeat objections as well. That said, compared to issues such as the causes of the Reformation in 16 th -century England or the truth of string theory, there may be stronger reason to believe that disagreement about the existence of objective moral truths and (to a somewhat lesser extent) disagreement about the soundness of the argument from moral disagreement are quite widespread and persistent.
Consider, first, the debate between moral realists and anti-realists. Beginning at least with Plato, a great many of the smartest minds of their day have reflected on the existence of objective moral truths. Today hundreds of academic philosophers all around the world devote large parts of their professional lives to the study of meta-ethics. Nevertheless, we still find numerous variants and sub-variants of realism (e.g. Railton 1986;Brink 1989;Shafer-Landau 2003) confronting numerous variants and sub-variants of anti-realism (e.g. Harman 1996;Blackburn 2000;Mackie [1977] 2011). Even if inferences from textual and verbal evidence can be shaky and some of these meta-ethical disagreements may be reducible to non-meta-ethical disagreements, it hence does seem plausible that there is widespread and persistent disagreement about the existence of objective moral truths.
Moreover (contra Leiter 2009;2010a), even disagreement about the soundness of the argument from moral disagreement seems widespread and persistent. Sextus Empiricus (2000) took moral disagreements to support his purportedly anti-realist view that nothing is good or bad "by nature" almost two millennia ago. Leiter himself (2009;2010a) draws on Friedrich Nietzsche's (e.g. 1966) contested argument from moral disagreement among philosophers. And, at least since the publication of John Mackie's Ethics in 1977, debates about the anti-realist implications of moral disagreement have become firmly established in meta-ethics, with many philosophers affirming and many others denying such implications (compare e.g. Brink 1984;1989;Boyd 1988;Shafer-Landau 2006;Enoch 2009, to Loeb 1998Huemer 2005;Tersman 2006;Doris and Plakias 2008;Leiter 2009Leiter , 2010aFraser and Hauser 2010).

The defusing strategy
The above considerations suggest that proponents of self-defeat objections may be right that those advocating the argument from moral disagreement cannot adopt anti-realism about the existence of objective moral truths and the soundness of this argument, and that there is widespread and persistent disagreement about these matters. However, just like in the case of companions in guilt objections, conceding this does not yet mean that the self-defeat objections succeed. In order for these objections to be convincing, anti-realism about the realism/anti-realism debate and about the argument from moral disagreement would also have to provide the best explanation of disagreement about these matters. But is this really the case? Do self-defeat objections withstand the defusing strategy as well? There are good reasons to answer in the negative, and thus to reject these objections after all.
Let us begin with ontological parsimony and consilience. Both the existence of objective moral truths and the soundness of the argument from moral disagreement are matters of philosophy. So called methodological naturalists (e.g. Kornblith 2002;Papineau 2011) argue that just like science, philosophy ultimately studies synthetic propositions about the natural world, i.e. propositions that do not merely hold in virtue of the meaning of their constituent concepts, and that we come to know through experience. 19 This metaphilosophical view is controversial. But if it is right -and I think there are good reasons for believing that it is right -then by appealing to objective philosophical truths, realist explanations of disagreement about the existence of objective moral truths and about the soundness of the argument from moral disagreement are not committed to any extra kinds of facts. They are as ontologically parsimonious as anti-realist explanations of these matters. Moreover, realist and anti-realist explanations of these matters need not differ in terms of consilience either (as the natural facts studied by philosophy might also explain certain kinds of non-philosophical facts).
Realism about the existence of objective moral truths and about the soundness of the argument from moral disagreement does not only perform well with regard to ontological parsimony and consilience; it is also highly coherent. One reason for believing so is that in explaining disagreements about these matters one often does not need to invoke the epistemic inaccessibility of the objective truth about them. For example, studies about cognitive dissonance (e.g. Festinger 1957) and philosophers' proneness to cognitive biases (e.g. Schwitzgebel and Cushman 2015) and rationalisation (e.g. Schwitzgebel and Ellis 2017) can make it seem plausible that at least some realists may fail to see the objective truth of anti-realism and of the soundness of the argument from moral disagreement because of a desire to ground their first-order moral views in an objective reality (see Leiter 2009;2010a). 20 Disagreement about the soundness of the argument from moral disagreement also often traces back to the simple fact that we do not (yet) have much evidence about how widely and persistently people really disagree about first-order moral questions (such as, for example, about abortion; see also the conclusion).
Finally, according to the argument from moral disagreement, the epistemic inaccessibility of the objective truths about a certain matter only undermines realism about this matter if it is incompatible with realism. But the potential epistemic inaccessibility of the realism/anti-realism debate and the argument from moral disagreement does not seem to be incompatible with realism about these matters (e.g. Shafer-Landau 2006;Leiter 2009;2010a). Both the question of the existence of objective moral truths and the question of the soundness of the argument from moral disagreement are not practical in the sense that their answers need to make a difference to what we ought to do. They are questions in second-order (meta-) rather than first-order ethics. These questions' epistemic inaccessibility hence would not entail that there are no objective moral facts about them (given Rowland's above argument, according to which we ought not to believe in unknowable moral facts because these facts could not make a difference to what we ought to do).
What is more, we also have independent reasons for believing that whether there are objective moral truths and whether the argument from moral disagreement is sound is not just constituted by the mental states of subjects or is only a matter of falsities or expressing non-cognitive mental states (as anti-realism about these matters has it). For example, both of these matters depend on empirical claims that appear to be uncontroversially objective, such as the question of the prevalence and persistence of moral disagreement (It does not seem plausible at all that claims such as these allow for non-cognitivist, error theoretic or subjectivist interpretations.).
Thus, even if proponents of the argument from moral disagreement had to concede that the objective truth about the existence of objective moral truths and the soundness of the argument from moral disagreement are difficult or impossible for us to access, their argument would not self-defeatingly extend to these matters.

Conclusion
Does the argument from moral disagreement implausibly generalise to (particular matters of) physics, history, philosophy, etc., or to its own conclusion or its own soundness? My above considerations do not provide any definitive answer to this question. Nevertheless, we found reasons to believe that moral realists have exaggerated the significance of their overgeneralisation charge. For any supposed companion in guilt, it is likely true that either disagreement about it is insufficiently widespread or persistent, this disagreement is best explained by realism about this matter, or realism about this matter is plausible. As disagreement about the existence of objective moral truths and the soundness of the argument from moral disagreement is likely best explained by realism about these matters, moral anti-realists need not worry too much about their argument being self-defeating either.
This result takes the sting out of one of the most prominent objections against the argument from moral disagreement. Taken by itself, it thus significantly strengthens this argument. But anti-realists should not crack their bottle of champagne too soon. For as has surely already been recognised by many readers familiar with the debate, both the empirical and the defusing strategy against overgeneralisation objections simply mirror (and develop) common realist strategies against the argument from moral disagreement: the objection that moral disagreement is insufficiently widespread or persistent (e.g. Brink 1984;1989;Boyd 1988), and the objection that this moral disagreement is best explained by moral realism (e.g. Huemer 2005;Enoch 2009). Just like in many non-moral cases, these realist objections (in particular the first one) may have some force with regard to morality as well.
To provide but one example, consider the cases of apparent moral disagreement mentioned at the beginning of this article. Do the Inuit and Westerners really morally disagree about the permissibility of parricide? And do pro-life and pro-choice advocates really morally disagree about the permissibility of abortion? Both of these cases admit of alternative interpretations as well. The Inuit and Westerners may actually agree that parricide is impermissible except if it is the only way to prevent much greater harm to one's community. Their differing moral judgements may only result from their applying this shared principle in different circumstances: circumstances in which parricide is never (Westerners) vs. is sometimes (Inuit) the only way to prevent much greater harm to one's community (see, for example , Brink 1989;Sinnott-Armstrong 2006). Moreover, disagreements about the permissibility of abortion may sometimes boil down to disagreements about non-moral questions. For example, it seems plausible that many pro-life and pro-choice advocates' moral judgements would converge once they reached agreement about the questions of whether foetuses can feel pain or whether God forbids having an abortion (see, for example , Brink 1984;1989;Boyd 1988).
Let me thus close by emphasising again that this article neither attempted to establish the soundness of the argument from moral disagreement nor depends on this soundness for its considerations to be valid. I am not sure whether the argument from moral disagreement will succeed. Much more evidence would be needed in order for anybody to be warranted to proclaim so (in particular, much more empirical evidence about the prevalence and persistence of moral disagreement; see Pölzler 2018). All that I purport to have shown in this article is that the argument from moral disagreement should not be easily rejected on grounds of its overgeneralising.