1. Some preliminaries

The concerns pursued in this paper arise at the intersection of social epistemology and moral philosophy: In talking about disagreement, my interest will be in the epistemology and not the politics of disagreement; and in talking about contractualism, I will be interested in its moral-philosophical forms and not the various species of political contractualism. Although I will remain ecumenical as far as possible, in my constructive discussion in the later sections of the paper, I will focus upon thoroughgoing or, as it is sometimes put, “global” forms of contractualism (see Enoch Reference Enoch2009b), views according to which the relevant forms of agreement not only allow us to identify what is right or wrong but also constitute the rightness or wrongness of what is agreed to. “Local” forms of contractualism, by contrast, may help themselves to realist or intuitionist conceptions of the reasons that motivate and justify agreements, and the significance of disagreement for these views will largely be hostage to what realism and intuitionism can say about disagreement. As I have observed, those questions have been explored by others, elsewhere. If contractualism has something distinctive to say about moral disagreement, we should look to the global varieties to say it. Finally, the relevant type of contractualism here will be those for which “rational” agreement is a matter of consensus on what is right – the sort of view found in Scanlon and Habermas, for example – and not those forms of contractualism (or contractarianism) according to which “rationality” is to be understood as a matter of bargaining or the balancing of unreconciled interests, as in the work of Gauthier or, more recently, Ryan Muldoon.

The literature on the epistemology of disagreement focuses – though not exclusively – on the case of epistemic peer disagreement: roughly, cases in which, prior to discovering the disagreement, I would judge that the person with whom I now disagree is as likely to be correct about this matter as I am. This is how I will understand “peerhood,” throughout.Footnote ³ The contestant accounts regarding the significance of such disagreements are, broadly speaking, steadfastness – the view that the disagreement of my epistemic peers gives me no reason, in itself, to alter my beliefs – and conciliationism – the view that, at least in certain circumstances, the disagreement of my peers may constitute evidence that I have erred, and therefore, may rationally require me to alter my view, by suspending judgment or by moderating the credence with which I hold my beliefs. I think it is fair to say that the conciliationist position has considerable intuitive appeal. In the sorts of “toy” cases that predominate in the literature, conciliationism seems to deliver the right judgments: if you and I are equally good at mental arithmetic, and if we arrive at different answers when we each calculate our share of the tab at dinner, we should indeed (and usually do) suspend our judgment as to which of us is correct.Footnote ⁴ The reasons for resisting conciliationism, on the other hand, have largely to do with its purportedly worrisome consequences – skepticism, self-defeat, inauthenticity, among them – consequences which are thought to be especially threatening if conciliationism applies as well to moral disagreement, given how widespread such disagreement appears to be. But apart from avoiding these problems, steadfastness seems to have little in the way of prima facie appeal: at least within strictly epistemological discussions, the view seems to lack the sort of epistemic humility befitting finite and highly fallible knowers. All this to say what I intend to take for granted: that the interesting question is what it would look like for moral contractualism to take moral disagreement seriously – in something like the manner proposed by conciliationism.

2. Disagreement in Rawls and Scanlon

Traditional contractualist theories – like many more recent efforts in moral-political contractarianism – have been developed against the backdrop of acknowledged disagreement, understood as a practical rather than an epistemic problem: they have been constructed as ways of preserving the possibility of agreement within the context of moral pluralism. This may appear obvious in some cases, as with the later Rawlsian idea of “overlapping consensus” among reasonable comprehensive doctrines; but even in the Theory of Justice, Rawls “identifies disagreement in modern democratic society over the proper understanding of freedom and equality as the circumstance that gives rise to the need for a constructivist approach to the principles of justice” (Engstrom Reference Engstrom and Bagnoli2013, 151).Footnote ⁵ The original position describes a point of view from which Rawls anticipates that moral agreement on basic principles will be forthcoming amongst a population with otherwise divergent views of the good life. While Scanlon, too, insists that his account of morality is consistent with value pluralism and “would not exclude moral disagreement” (Reference Scanlon1998, 360), he nevertheless maintains that, at least within the domain of the narrow sense of morality as “what we owe to each other,” “moral principles are universal” and judgments of right and wrong based upon those principles “hold everywhere” (Reference Scanlon1998, 348). In other words, the idea of justifiability to others, or reasonable rejectability, constitutes a point of view – if only we can adopt it – from which the possibility of agreement is secured. This is so, even while we can expect that some principles are reasonably rejectable (or not) according to reasons “that people have only in certain social conditions” (Reference Scanlon1998, 348), a caveat which allows for value diversity although not disagreement, per se. Outside of the narrow sense of morality, on the other hand, Scanlon allows and expects genuine moral disagreement; but contractualism, in his view, does not offer a theory of morality in this broader sense. Like Rawls’, then, the aim of Scanlon’s approach is to secure the moral point of view as one that preserves the possibility of agreement despite what may be empirically widespread disagreement or, in other words, to eliminate the threat posed to moral validity by first-order moral disagreement. Neither Rawls nor Scanlon answer the epistemic question of whether the disagreement of our peers ought to be taken as so-called higher-order evidence of error in our own moral deliberations – that is, neither address how we ought rationally to respond to peer disagreement regarding what ideal deliberative agents will agree to, a question to which I now turn.Footnote ⁶

3. Hypothetical contractualism and hypothetical disagreement

Answering this question requires that we first consider who counts as the relevant sort of peer when it comes to contractualist deliberation. It seems that the consistent hypothetical contractualist should answer: some hypothetical, representative person(s). And that’s because hypothetical contractualists, like Scanlon, insist that, although deliberative engagement with actual other people can sometimes be practically useful in special cases, the question of what is right is answered only by “making a judgment about what others could or could not reasonably reject. This is a judgment that each of us must make for him- or herself. The agreement of others, reached through actual discourse, is not required, and when it occurs does not settle the matter” (Reference Scanlon1998, 393–94; 395 n. 18; and 340–41). As a consequence, Scanlon acknowledges that his view is “monological,” in Habermas’ sense: it involves “a hypothetical process of argumentation occurring in the individual mind,” and does not require “that real discourse be carried out” (Reference Scanlon1998, 393–94; see Habermas Reference Habermas1992, 68).

It seems to me that Scanlon’s position is quite typical in this respect: putting aside its peculiarly contractualist shape, his view is tantamount to an endorsement of moral testimony pessimism – an insistence that, at least in moral matters, we must form our moral beliefs autonomously, that although we may listen to the advice of others, we may not legitimately defer to them (except in special cases). Such pessimism about moral testimony is quite common, and similar commitments may be found in other specifically contractualist moral theories – for example, those of Nicholas Southwood, Stephen Darwall,Footnote ⁷ and, in my view at least, the Rawls of Theory of Justice. Indeed, I believe this strictly hypothetical and monological approach – with its pessimistic consequences for moral testimony – to be the dominant form of moral contractualism. Nevertheless, my aim here is not to defend the attribution of this view to anyone in particular, but to consider the logical possibilities for accommodating conciliationist insights within contractualism, and I begin with a purely hypothetical approach.

Returning to the point, then. To the extent that hypothetical contractualism insists upon a monological deliberative procedure, it appears to be in a difficult spot: ex hypothesi, in deciding what is right, I do not engage in dialogical deliberation with actual other people. Thus, just as the empirical agreement of others is not required in order to constitute some norm or principle as valid, empirical disagreement with others does not invalidate it. I do not ask whether any actual person affected does disagree or reject some norm; instead, I imagine how a deliberatively perfect version of this person would argue (see Southwood Reference Southwood2018), or how a “representative person”, whom I have imaginatively constructed as occupying a particular moral position, would argue (Scanlon Reference Scanlon1998, 171). And yet, as we have seen, it is precisely on this level that familiar forms of hypothetical contractualism have sought to eliminate the possibility of disagreement a priori through the construction of a monological moral point of view.

To be clear, the problem is not that I cannot even imagine disagreement from within this idealized point of view. Sticking with Scanlon’s formulation, in the interests of brevity: if I can imagine that some “representative person” could, all things considered, reasonably reject a norm, even though other representative persons could not, then that sinks the norm – it is not justifiable and is therefore wrong. The disagreement of a hypothetical reasoner with a candidate norm is, in that sense, decisive: it is what forms my actual moral conviction as a deliberator. But the question that concerns us here is whether I can imagine disagreement with my own moral conviction so formed, and that means after I have already canvassed the views of “representative persons” as I imagine them. It is important to bear in mind as well that it is not the mere fact of imagined disagreement that matters – of course, I can imagine that – since it is reasonability of rejection that idealized discourse is supposed to test, in contrast to the messy, potentially unreasonable facts of empirical disagreement. Disagreement that I cannot see as reasonable is irrelevant to the validity of norms and, in the language of the epistemology of disagreement, cannot represent the disagreement of my peer.

But that returns us to the difficulty: I cannot, as solitary reasoner, both believe that norm N is not reasonably rejectable while, at the same time, imagining that it is reasonably rejected by some epistemic peer, understood as a hypothetical, representative person. Thus, hypothetical contractualism excludes the possibility of entertaining peer disagreement from within the moral point of view, as the hypothetical contractualist understands it – that is, as hypothetical and monological. Just as Kantian views of autonomy have seemed to some to preclude a priori any acceptance of (pure) moral testimony, we might conclude that hypothetical contractualism precludes a priori any attitude other than steadfastness toward purely hypothetical disagreement, as a result of the monological structure of its deliberative procedure. But that way of putting things may mislead insofar as “steadfastness” purports to be a position on peer disagreement specifically, whereas the difficulty with monological contractualism is that it seems to exclude the possibility of imagining both that someone disagrees with my considered moral conviction, and that they are nevertheless my peer. Such a view at least shares with steadfastness an unhappy lack of moral-epistemic humility.

Perhaps it will be objected that, despite the complications introduced by the mechanism of a fictionalized inner “dialogue,” the situation here really is no different from any other disagreement: Once I calculate my share of the tab, howsoever I accomplish that, I must believe that my conviction is responsive to the evidence. That’s just what it means to believe what I believe, as the defenders of steadfastness have been keen to observe. The whole point of conciliationism is to insist that the disagreement of an epistemic peer – someone I would have judged beforehand to be equally likely to be right – constitutes new evidence that I may have erred. But this presents a second difficulty for the hypothetical contractualist: What could it even mean for me to ask whether a representative person, whose moral position I have myself imaginatively constructed, is my epistemic peer in the relevant sense?

As Katia Vavova has noted, deciding when the disagreement of others gives us evidence of error, and how strong that evidence is, requires that we settle two preliminary questions: “it will depend on both [1] how close to peers I take us to be and [2] on how confident I am in this judgment” (Reference Vavova2014, 308). Assessing these matters, in turn, requires that I access – and reflect on the adequacy of my access to – your track record as a moral deliberator. Do I believe, all things considered, that you are as likely as I am to be right (or mistaken) in cases of this sort? How confident am I in that judgment? Vavova’s point is that there may not be very many cases, after all, in which I have grounds for a high level of confidence in your epistemic peerhood. But I mention this to observe something different: that as soon as we pose the challenge of peer disagreement, we must give up the favored device of hypothetical, idealized, or representative persons. Questions of track record and credentials and grounds for confidence – the defining questions of peerhood – can be asked and answered only of actual people. They cannot even be coherently posed from within the deliberative point of view as it is characterized by hypothetical contractualism, from which other actual persons have been excluded.Footnote ⁸ Thus, once again, we arrive at the view that, internal to the moral point of view as constructed by hypothetical contractualism, there does not appear to be space even to raise the question of peer disagreement.

The need to address real histories of judgment may be one reason that, in the literature on the epistemology of disagreement, it has largely been accepted that it is actual and not merely hypothetical disagreement that matters. Granted, Thomas Kelly (Reference Kelly2005) has objected that if disagreement is epistemically significant, then that cannot be a fact only about actual disagreement, since whether we encounter such disagreement may be determined by all manner of arbitrary circumstance. Perhaps a dogmatic and violent tyrant has killed or imprisoned all dissenters on some question. Less fantastically, perhaps for purely contingent reasons, I just never happened to have met the people who disagree with me. Thus, in Kelly’s view, if disagreement matters, it must be that hypothetical disagreement matters just as much as actual disagreement: the question is – as Scanlon, too, insists – whether someone who is my epistemic peer could disagree with me, not whether I am aware of anyone specifically doing so.Footnote ⁹ But Kelly’s point is a reductio: If we take this route, we will be forced into skepticism since it is always possible that someone could disagree with me. In other words, Kelly intends to deny the antecedent: he believes that neither actual nor hypothetical disagreement matter, in themselves.Footnote ¹⁰

Conciliationists, on the other hand, have denied that the consequent follows. As David Christensen notes, given my awareness of my own epistemic fallibility, I do have to concede that it is always possible that someone could disagree with me. That’s not an interesting or informative realization. When someone actually disagrees with me, on the other hand, then I have some evidence that this generic, ubiquitous possibility has been realized, that this may be one of those cases where I’ve made a mistake that I’ve not been able to see. Moreover, the sort of disagreement that matters is the disagreement of my peer – that is, reasonable (in my view) disagreement. But, as Christensen adds, I am extremely unlikely to generate convincing proof against my own views through the effort to simply imagine disagreement: “in those cases where we have made a mistake in assessing the evidence about P, we are overwhelmingly likely to make the same mistake in assessing how a rational peer would assess the same evidence about the same proposition” (Reference Christensen2007, 208–09). As we have already seen, neatly parallel things may be said against the idea that it is hypothetical disagreement that ought to matter to moral contractualism: within a monological moral point of view, it seems I will be unable to construct an epistemic peer who disagrees reasonably with my own view of what is not reasonably rejectable.

So, the case of merely possible, hypothetical disagreement appears to be a dead-end for accommodating conciliationist insights within hypothetical contractualism. And we have, in any case, independent, epistemological reasons to think that it is actual moral disagreement that matters. Let us direct our attention thusly: Can the hypothetical contractualist, who has excluded disagreement from the deliberative point of view that is constitutive of the moral facts, take actual peer disagreement seriously as a kind of higher-order evidence that we have failed successfully to occupy that point of view? Though neither Rawls nor Scanlon ever committed such a view to print, I suspect that defenders of either are likely to find this combination of claims attractive. But is the combination plausible?

4. Hypothetical contractualism and actual disagreement

Let us say that you and I are moral-epistemic peers and committed contractualists, each; that, after fully sharing our respective arguments, you believe that, behind the veil of ignorance, rational deliberators would agree to a policy of genetic screening and selective abortion in the case of serious disability, or that no one could reasonably reject such a policy given the foreseeable distribution of consequences and burdens. I believe that they would not so agree. How should we respond?

We might start by reminding ourselves that hypothetical contractualists allow that I can, in exceptional cases, have reason to defer to others about the effects of some proposed principle – because, for instance, I come to discover through actual discourse that my judgments are biased in some way, or because I come to see that others may be better positioned to observe the consequences of some candidate norm (Scanlon Reference Scanlon1998, 395). But in any case where such reasons apply, we are a fortiori no longer peers: I no longer believe we are equally likely to be correct. Moreover, because I ought therefore to defer to you in such a case, I do not conciliate between your view and my own: I regard my initial and contrary view as defeated. So, this response overshoots the target.

A proposal from C. Thi Nguyen (Reference Nguyen2010), offered in a partly related context, might (mutatis mutandis) offer the hypothetical contractualist a way of explaining how actual peer disagreement can matter epistemically, even while actual peer disagreement (and agreement) is seen as metaethically irrelevant – that is, irrelevant to the monological constitution of norms. There is no doubt some spiritual kinship between moral testimony optimism and moral conciliationism, insofar as both views encourage us to give considerable weight to the beliefs of others unless we have strong reasons against doing so; and there is a similar consonance between moral testimony pessimism and moral steadfastness, in the stress that each lays on seeing things for ourselves. Nevertheless, these views do not logically entail one another, and Nguyen has argued that it is possible to combine an anti-testimonial insistence upon autonomous, individual insight with a conciliationist attitude toward disagreement, provided that we treat disagreement as offering us specifically non-moral, higher-order evidence about the reliability of our own completed moral reasoning processes. In other words, moral disagreement with someone we consider a peer gives us a powerful, non-moral reason to conciliate: it gives us higher-order evidence that we may have made an error in our own monological moral reasoning and, as a result, should become uncertain about who indeed is right. Crucially, because such a check occurs only after I have completed my own deliberative process – somewhat as Rawls’ describes the lifting of the veil of ignorance – its invocation might appear compatible with the essentially monological structure of hypothetical contractualism. The resulting view of contractualism would be a sort of hybrid: (I) Only hypothetical agreement matters for the determination of what is right (e.g., what is or is not reasonably rejectable); but (II) actual disagreement matters for our confidence that we have successfully performed the hypothetical, monological procedure.

But such a hybrid view seems intrinsically puzzling and unstable for at least two reasons. First, if actual agreement is deemed such an unreliable guide to moral validity – in ways that motivate the move to an idealized and hypothetical model of deliberation in the first place – then why would actual disagreement be deemed a reliable indicator of the failure to constitute moral validity? It is the same sort of reasoning performed by the same actual people in either case. If their deliberations about what they can and cannot reasonably reject are so imperfect as to justify their exclusion from the moral point of view, then surely the hypothetical contractualist should be perfectly indifferent toward the actual disagreement or agreement of these same people regarding the outcome of idealized deliberation. Inversely, if the actual agreement and disagreement of other people is a reliable guide to whether or not I have successfully occupied the moral point of view, then what is the reason for excluding actual discourse with them from that point of view in the first place? The hybrid view undermines the central rationale for defending an idealized, hypothetical, and therefore, monological form of contractualism.Footnote ¹¹

Second, though perhaps less decisively, it seems to me misleading to characterize this hybrid view as a form of hypothetical contractualism, since actual agreement and disagreement is now playing a central role in the constitution of valid moral norms or, at least, in our belief whether any norm has been constituted as valid. Consider, for a moment, how conciliationists would see a parallel, non-moral case. The central, motivating principle of conciliationism is the recognition of our fallibility as knowers: both that we make mistakes and that we are often unable to see these mistakes for ourselves. That, as Christensen has it, is the “good news” about disagreement: it provides a means of seeing what we otherwise likely could not. Relatedly, conciliationists widely agree that numbers matter, that the distribution of beliefs can be significant with respect to the “news” that disagreement offers. When I do the mental math to calculate my share of the dinner tab, arriving at the belief that it is, say $43, the confidence with which I hold that belief should be tempered until I find out the number at which my equally competent dinner companions have arrived. If that does not seem right to you, consider: When I discover that all of my companions agree with me, I will reasonably feel greater confidence in my result. A fortiori, I ought to have felt less confidence before this discovery. Indeed, Nguyen himself suggests that our commitment to “getting it right” in our moral beliefs may actually give us reason to actively seek out the views of our peers in order to be able to test the reliability of our autonomous moral reasoning; and, although he doesn’t say so explicitly, it certainly appears to follow that my rational confidence in my beliefs will increase as they withstand such efforts at falsification.Footnote ¹² Thus, in advance of such efforts – which may potentially reduce my confidence to effective agnosticism – I should hold my belief with reduced or undetermined confidence, as a kind of hypothesis that awaits testing and confirmation.

Once we concede that actual peer disagreement plays this crucial role in grounding the confidence with which we can assert some norm as valid, we are no longer describing a theory of hypothetical, monological contractualism. It is no longer plausible to say that reasonable rejectability, for instance, involves only “a judgment that each of us must make for him- or herself. The agreement of others, reached through actual discourse, is not required, and when it occurs does not settle the matter.” On the contrary, rather than constituting moral norms that are subject only to some post hoc “double-checking,” the results of our monological deliberation would be better described – if we adopt Nguyen’s approach – as a kind of hypothesis, our starting conviction in the debate with our peers, to be definitively tested in the crucible of actual agreements and disagreements. We cannot know with any confidence what really is and is not a valid norm, in the absence of, or in advance of, such actual discourse.

Thus, the cost to contractualism of granting epistemic significance to actual peer disagreement is that it must give up its insistence upon a purely hypothetical deliberative procedure. What will matter, in the end, for our ability to identify candidate norms as properly constituted, will be the actual agreement (and absence of reasonable disagreement) of actual other people about what we or our representatives would agree to if we were perfectly deliberatively rational. Granted, on the hybrid view, it remains possible to insist that it is the idealized, hypothetical agreement that constitutes a norm as valid; but if we cannot identify any such norms, cannot know that this constitution has indeed occurred, without actual agreement, then it is unclear what is gained by the insistence. Moreover, as I noted in the first place, once we grant the actual views of our actual peers that kind of epistemic weight, it is puzzling why we should insist that the same views are excluded from the process of deliberative constitution. If actual agreement is allowed to be central to the overall process of the production of moral knowledge, then why not opt for a version of contractualism that openly and directly rests on actual agreement?

5. Actual agreement contractualism and actual disagreement

Although certainly not the common view – and despite having been dismissed by some hypothetical contractualists as a “non-starter” (Southwood Reference Southwood2018, 529) – actual agreement versions of moral contractualism have been defended by Jürgen Habermas (Reference Habermas1992), Karl-Otto Apel (Reference Apel1996), Seyla Benhabib (Reference Benhabib2007), and Rainer Forst (Reference Forst2014), among others. The details dividing such accounts need not detain us here. The general, shared picture involves a view of morality as a really existing social institution that arises from the attempt to coordinate action on the basis of principles or norms that can be justified as right to all affected, and which, therefore, embodies a set of intrinsically normative claims about justifiability.Footnote ¹³ In practice, such coordination often happens on the basis of already accepted norms, background agreements regarding behavioral expectations, and so on. Where disputes arise, however, the re-establishment of agreement is pursued through attempts at justification, a form of communicative interaction that is constituted by certain presupposed commitments on the part of participants. For example, we cannot even try to reach an agreement on what is right except insofar as we presuppose that everyone affected must be given the opportunity to participate; that views will be shaped by the force of the better argument (and not by threats), and so on. The constitutive presuppositions are “quasi-” rather than fully transcendental, in Habermas’ view – although this is a point of disagreement with Apel – because, in practice, we are always confronted with approximation: some degree of stubbornness, some practical limits to the length of debate, some limitation to the number of participants. But that these commitments are nevertheless transcendental is clear from our response to serious violations: Competent speakers intuitively recognize that, for example, where some affected person is silenced or threatened or excluded in the course of the conversation, then whatever is “agreed” to under such conditions is not, in fact, an agreement in the morally relevant and binding sense. There are, of course, very many details to be added in any adequately articulated account and many objections to answer. But it is not my aim to offer an independent defense of the actual agreement approach; it is my aim to say something about the place of disagreement in contractualism and, for that, this bare sketch will suffice.

For an actual agreement approach, actual disagreement raises foremost a worry about moral failure: because the validity of norms is a product of actual agreement, actual disagreement signals a failure of some sort to produce a valid norm, or to confirm a norm as valid. As a result, the significance of disagreement will be distinctive for such a view, since it will not be confined solely to disagreement with those in whose peerhood we are confident – although, as I will discuss in conclusion, there are also important differences between peer disagreement and the disagreement of non-peers, whose views may be subject to various defeaters. Nevertheless, the relevant pool of agents for the constitution of a norm is “everyone affected,” and so any actual disagreement within that pool presents a problem. Indeed, insofar as we are considering a “global” version of actual agreement contractualism, according to which there are no moral facts independent of the outcome of the contractualist procedure, such failures are not merely “practical” but “metaphysical” – resulting in real indeterminacy in the moral facts.

At the same time, it is too quick to see in this an argument not just for indeterminacy, but for moral skepticism or nihilism, the complete absence of all moral facts. Steadfast critics who worry that conciliationism will force us into moral skepticism often present the matter in all-or-nothing terms: one is either entitled steadfastly to believe that meat-eating is permissible, irrespective of disagreement, or one must surrender one’s belief in the face of disagreement and remain agnostic. Many conciliationists, on the other hand, recognizing the complexity involved in assigning evidential weight to a specific person’s disagreement on a specific question, opt for a view according to which we will typically be forced to alter the credence with which we hold our beliefs, allowing that only in some cases will our credence be reduced to the point of effective agnosticism (see Vavova Reference Vavova2014). Of course, one wants reasons for such moderation; but these we often have, in the form of defeaters. Indeed, any failure to satisfy a constitutive presupposition of moral discourse will count as a potential defeater; and since approximation to full satisfaction is the most that can be hoped for, even in the best cases, some potential defeaters are always present. For the same reason, however, it would be unreasonable to see this in all-or-nothing terms, as though the mere presence of any uncertainty immediately sank any claim to moral justification, at any level of confidence, tout court. To so insist would, for fallible knowers like ourselves, be tantamount to an absurdity: it would demand perfection in our epistemic circumstances in order to license any degree of belief, a bar we can meet neither in our moral nor in our non-moral convictions. How serious were the failures of inclusion? How far did stubbornness, bias, fallacious reasoning, or ignorance interfere with the force of the better argument? How prematurely did the discussion end? How we answer such questions should have bearing on the confidence with which we endorse the validity – the unfinished validation or constitution – of the relevant norms or principles in the face of ongoing disagreement.

A further point against a prematurely skeptical conclusion: As I’ve already noted, conciliationists largely agree that numbers matter. It is one thing when I encounter the disagreement of a single, usually arithmetically reliable friend, about my share of the tab; it is quite another if I find myself disagreeing with the calculations of dozens of my mathematical peers.Footnote ¹⁴ Similarly, if I find myself in agreement with dozens of the arithmetical peers with whom I am sharing this meal, while only one friend dissents, I have very little reason to conciliate (see Cosker-Rowland Reference Cosker-Rowland2017, 3–4). In the moral cases of disagreement, it matters that we are not, for the most part, faced with a dizzying array of divergent beliefs held by single individuals, but with substantial agreement among large groups, often confronted with a large group of dissenters who themselves share some view. If we are rational, our sense of the relevant reasons will determine where we find ourselves in such disputes; but it will further matter with whom we are agreeing and with whom we are disagreeing, in assessing the strength of the evidence of disagreement as a reason to conciliate since, among other things, this will bear on what sorts of defeaters we have. This does not make conciliationism an apology for ethics by majority rule: contractualism, in particular, is avowedly non-aggregative in its view of moral rightness. But numbers do matter in attempting to assess the epistemic significance of disagreement with our shared considered judgments of what is right. Still, at least for the global-constructivist brand of contractualism, there is no gainsaying the fact that disagreement represents a failure to fully constitute some norm as valid and is thus a source of real moral indeterminacy.

Because of this, numbers can matter in an additional way for the actual agreement approach to contractualism: as an indicator of one type of moral progress. If more and more people come to accept a principle – same-sex marriage equality, or some degree of moral standing for animals – either through argumentation in adequately-conducted discourse, or, at least, if they come to accept a principle that itself has been validated in such discourse, this must count for the actual agreement contractualist as morally progressive – a movement toward the constitution/validation of norms and principles to which all affected can agree. As Scanlon elegantly concludes, “Working out the terms of moral justification is an unending task” (Reference Scanlon1998, 361)Footnote ¹⁵ ; but numbers matter, for the contractualist, because they may signal our progress towards consensus. Full inclusion is, in many cases, not possible; and full deliberative rationality is, in every case, unattainable. That is not a counsel of despair. But the unavoidable incompleteness of the process of normative justification is an important source of moral indeterminacy. This is itself an interesting and important point upon which the actual agreement approach sheds useful light: the literature on moral indeterminacy has often seen indeterminacy as a cause of disagreement; but it has rarely noted that indeterminacy may be a result of disagreement.Footnote ¹⁶

Finally, we can see here another way that an all-or-nothing approach to moral belief misconstrues our situation: in a world of moral indeterminacy and justificatory incompletion, our moral commitments will often involve a measure of moral and epistemic risk, particularly since remaining agnostic will often be practically impossible or will entail unacceptable costs – despite everything, we still must decide and act. An especially clear example may be found in our obligations to future generations, whose non-participation in processes of justification always presents us with a potential defeater for whatever norms we believe apply to the intergenerational consequences of our present decisions. Try as we might to anticipate the point of view from which future generations will morally assess the ways we have impacted them, we must recognize that our judgments involve an ineliminable form of risk – a hypothesis regarding how our descendants will organize conflicting values. The same point holds when we are faced with the actual disagreement of affected parties, here and now: since the relevant sort of disagreement will signal moral failure, understood as the failure (of some magnitude) to fully validate a moral norm, disagreement will also have a distinctive epistemic upshot – it tells us what kind of risk we will be courting in continuing to endorse and act upon a not-fully validated norm, one that is rejected (or, at least not accepted) by some of those affected by it. And it is indeed not just reasonable to moderate the credence with which we hold our moral beliefs – the belief that some norm, N, is justifiable to all affected – when we encounter evidence of the failure to justify N; on the premises of actual agreement contractualism, it is simply true that N cannot be entirely or conclusively justified in the face of ongoing disagreement.