Over the last fifteen years or so, Robert Audi ad I have often discussed the epistemic and practical implications of fundamental moral disagreement.Footnote 1 For me at least, these discussions have been extremely helpful, and I believe they have led to a certain amount of convergence. But this convergence is far from complete, and in this paper I shall outline where I believe our positions currently stand, respond to some of Audi’s more recent arguments, and conclude with some reflections on the practical implications of moral disagreement.

Ornithology and Philosophy

Consider the following caseFootnote 2:

Birds. You and an old and entirely trustworthy friend are wandering together in the countryside. The light is good. Both of you have recently been attending a course for novice ornithologists, and each of you knows that the other has excellent vision and is roughly equally good at bird-identification. At the same moment, each of you sees a bird on a tree-branch about ten metres away. You believe it to be a redwing, while your friend says it is a song thrush. The bird flies away before further investigation can be made.

How should you respond to this disagreement? That is, what is the epistemically rational way to respond? One response was described by Sextus Empiricus as the fifth mode of epochē, or ‘suspension’:

The one based on disagreement is that according to which we find that, both in ordinary life and among philosophers, with regard to a given topic there has been reached an unresolvable impasse on account of which we are unable to reach a verdict one way or the other, and we end up with suspension of judgment. (1996: PH I. 165–6, trans. Mates).

In Birds, this response strikes me as clearly the right one, and the only response in line with Hume’s dictum that one should proportion one’s beliefs to the evidence.Footnote 3 Given what you know about birds (in particular, about the close similarity of redwings and song thrushes) and about your friend’s epistemic abilities and character, you must proportion roughly equal probabilities to the propositions that the bird is a redwing and that it is a thrush.Footnote 4 Henry Sidgwick, when outlining various ‘tests’ for the certainty of apparently self-evident beliefs, wrote as follows:

Since it is implied in the very notion of Truth that it is essentially the same for all minds, the denial by another of a proposition that I have affirmed has a tendency to impair my confidence in its validity…. And it will easily be seen that the absence of … disagreement must remain an indispensable negative condition of the certainty of our beliefs. For if I find any of my judgments … in direct conflict with a judgment of some other mind, there must be error somewhere: and if I have no more reason to suspect error in the other mind than in my own, reflective comparison between the two judgments necessarily reduces me temporarily to a state of neutrality. And though the total result in my mind is not exactly suspense of judgment, but an alternation and conflict between positive affirmation by one act of thought and the neutrality that is the result of another, it is obviously something very different from scientific certitude. (Sidgwick 1907: 341–2; italics added).

Given the context, Sidgwick must here be stating, in his characteristically tentative way, a normative rather than a descriptive principle: reflective comparison should reduce me to a state of neutrality. And when he says that the state he finds himself in in such cases is ‘not exactly suspense of judgement’, I take him to be recognizing that his original judgement still appears to him to be true. But no more credence should be placed in that appearance than in that of a bent stick in water. We might capture Sidgwick’s position, then, as:

Sidgwick’s Principle: A person who judges that p, if she finds that some other person judges that not-p, and if she has no reason overall to believe that other person to be in a worse epistemic situation than her, should suspend judgement on p.Footnote 5

Now consider the following caseFootnote 6:

Rossianism. You are a well-trained and highly-regarded professional moral philosopher, working in an outstanding philosophy department. For many years, you have studied historical and contemporary works of philosophy, considered the various positions available in ethics, and published many books with excellent university presses and articles in top journals. You believe, along with W.D. Ross, that there is a plurality of moral principles. For example, the fact that someone has promised to φ is, you believe, other things equal, an ultimate, non-derivative reason to φ. You have thought hard about each of the components of this view of ethics– the nature of acts and agents, what a principle is, what makes other things unequal, and so on. And you are utterly convinced. With your view now finally worked out to your satisfaction, you decide to discuss it with a colleague in the same department. This colleague has a similar background to you, is equally well trained, and has an equally impressive publication record. She listens patiently while you explain your Rossian view, reflects seriously upon it, and rejects it. Her own position – which she has developed after a period of reflection similar to yours – is that the only ultimate reason for acting in any way is to maximize overall well-being. She explains her view to you, equally patiently, including its implications, and you reject that view.

This case is of course somewhat narrow and artificial. But I suggest that, especially in the light of its analogies with Birds, it captures the essence of a great deal of basic first-order disagreement between moral philosophers.

Audi on Moral Perception and the Ethics of Belief

A good deal of Audi’s Moral Perception, and a paper published slightly earlier (2011b), have much new to say on the issue of moral disagreement.Footnote 7

Audi’s book is of course especially concerned with what subjects can perceive in particular actions. But, as Audi notes (82), his points often carry across to disagreement about moral principles. Audi makes three points in a prolegomenon to the second part of the book, in which most of the discussion of disagreement is found (66). First, just as persons may differ in aesthetic sensitivity, so they may differ in moral sensitivity, and this may lead to disagreement. Second, much moral disagreementconcerns propositions arrived at through inference, and the parties to a disagreement may differ in their standards for a sound inference. Third, those disagreeing are rarely responding to identical evidences.

These points are well taken. How should we understand the Rossianism case in the light of them? Since I already accepted that the case is somewhat artificial, I might just stipulate that you and your colleague are equally morally sensitive, apply the same inferential standards, and are responding to identical evidences. But this would, I think, be something of a desperate strategy, especially in the case of evidence. There is a strong case for saying that the ‘evidence’ (or perhaps the evidential background) for philosophers’ first order judgements extends well beyond philosophical texts, discussions, conferences, and so on, into acquaintance with literature, history, politics, and indeed moral upbringing and experience of human life.Footnote 8 It would, then, be absurd to postulate identity of evidential background in any such case construed realistically. But two important questions now become salient. The example assumes that you and your colleague are epistemic peers. Must the evidential background for epistemic peers be identical, or just relevantly similar? It almost certainly makes a difference to one’s degree of moral sensitivity or acuteness that one was able to play with other children. But perhaps, within a certain range, the number of playmates and playdates one had is insignificant. Second, and more importantly, for Sidgwick’s Principle to have practical bite, must it be the case that someone disagreeing with another can identify that other as an epistemic peer? If so, even if we allow relevant similarity to replace identity in our standard for evidential background, cases of disagreement with epistemic peers may be really quite rare, and those in which actual peerhood can be recognized even rarer.

Audi makes various important points relevant to this issue. First, he distinguishes between disagreement over prima facie moral appraisal and that over final or on balance appraisal (75). The latter, he implies, is more common (e.g. people disagree about the on balance wrongness of assisted suicide, but less so about the prima facie wrongness of killing people). That is almost certainly true, but the case of Rossianism does concern on balance wrongness in the sense most important for philosophers. The Rossian view is that actions are made on balance wrong by their violating one or more of the Rossian principles, even though those principles, as far as their application to individual actions is concerned, are stated as prima facie (or pro tanto).

Audi goes on to characterize epistemic parity as a case in which two individual subjects are (a) equally rational and thoughtful and (b) have considered equally conscientiously the same relevant evidence (76).Footnote 9 He rightly points out that there has to be equally conscientious consideration of the evidence, since parity cannot consist merely in two individuals with the same epistemic virtues being confronted by the same evidence. In any particular case, for example, one party to the disagreement might not have attended properly to the evidence. Rossianism is intended as a case in which there has been (at least roughly) similar attention to the evidence, and Audi goes on to consider just such a case of disagreement with a colleague (77–8). He accepts that when you find your colleague disagrees with you, you should not conclude that you are clearly right – that would be dogmatic. But, he goes on, it is extremely difficult to be justified in believing (a) and (b) above to be true in any real case. To be justified in seeing your colleague as an epistemic peer will require a large amount of complex evidence from a wide range of sources. And deciding what that evidence is (as we have seen) anyway not straightforward.

There are also, Audi notes, non-evidential sources of disagreement (78–80). The parties may hold different background theories. In the Rossianism case, for example, perhaps you accept some form of coherentism based on the idea of reflective equilibrium, while your colleague is a hard-line foundationalist intuitionist, and so readier to discount her common-sense moral beliefs. There are also non-theoretical background beliefs to take into account, some of which may be prejudicial and operate sub-consciously. Perhaps many years ago your colleague’s undergraduate application to Oxford was rejected, which resulted in her forming a view of the place, and the views that emerged from it, as stuffy and anachronistic. Also relevant are what Audi calls the ‘conceptions’ that participants bring to a disagreement, which include what I have suggested might be part of the available ‘evidential background’, if not plausibly counted as evidence. Here Audi actually mentions the possible effects of reading Ross rather Bentham or Mill, as well as literature by authors as different as Austen and Dostoyevsky, and religious convictions. Your colleague, perhaps, is a Christian, and impressed by the theological utilitarian tradition beginning with Cumberland and reaching its apogee of influence in the nineteenth-century works of Paley.

Audi makes, then, a very strong case for the claim that genuine epistemic parity is rare, and awareness of it even rarer. But how big a difficulty is this for Sidgwick’s Principle? Note that the principle is not restricted to cases of genuine epistemic parity, or cases in which the subject is aware of such parity, or even falsely believes it to be the case.Footnote 10 It does apply to such cases, but extends far more widely, to any case of disagreement in which the subject has no reason overall to believe the other person to be in a worse epistemic situation than her. In a real case like Rossianism, for example, it will indeed be extremely unlikely that you and your colleague are epistemic peers, and this is what, if rational, you will believe. The question, however, is whether you have any reason to believe your colleague to be in a worse epistemic position than you. In other words, Sidgwick’s Principle is negative, while Audi’s criticisms are directed at a significantly stronger positive principle.

Audi has further arguments, however, to set against Sidgwick’s Principle (79–80). The first is that, if one rationally believes p, then someone else’s disbelieving p is itself a reason to doubt that the other person is correct in her denial or is an epistemic peer. This suggestion certainly weakens the practical significance of Sidgwick’s Principle, though note first that the suggestion is in effect introducing one more piece of evidence to judge another person’s epistemic qualities. Imagine that in Rossianism your colleague is fifteen years your senior, and so has had a substantially greater amount of time in which to think about first-order ethics. If that gives her an epistemic advantage (note the ‘if’!), then it may counterbalance the evidence of her thinking not-p. But one might wonder anyway whether, independently of the positive justification one has for one’s own belief, the mere fact that another disagrees with one counts as evidence of their epistemic inferiority (see Christensen 2011). The problem is essentially that if we allow it as evidence, its value and indeed status as evidence will immediately be cancelled by the evidence available to the other person about one’s own epistemic position.Footnote 11 When your colleague asserts not-p, you might follow Audi and downgrade your assessment of her epistemic capacity. But you know that she will also be in a position to downgrade her assessment of your epistemic capacity. In other words, if disagreement is evidence of epistemic inferiority, it is evidence for both parties, not just for one or the other. If you are justified in downgrading, so is she, which means that the evidential weight of mere disagreement is the same for both of us.

Audi notes also that each of us is in a better position to assess the evidence available to us and of our response to it than we are to assess the evidence and response of another. (Audi has in mind primarily ‘internal’ evidence, such as that of the senses, or perhaps one’s sense that some proposition is or is not self-evident.) It is easier for you to know that your belief is based on the evidence, for example, than for you to know that this is true of your colleague, and so you are entitled to think that her position may be merely consistent with her evidence. Indeed – though Audi does not himself say this – you are entitled to think that her position may be neither based on nor even consistent with the evidence, and the result, perhaps, of some mental abnormality such as some quasi-psychopathic desire to cause conflict.

Audi’s points here seem to me largely right, though it is worth pointing out that there are limits to the transparency to ourselves of our own mental lives. The question is how weighty they are, and I suggest that, though they may sometimes be highly significant, in many standard cases of disagreement they can be largely discounted.Footnote 12 Your reason for thinking your birding companion is playing games with you, or that your colleague is not playing with a straight bat in her avowal of consequentialism, is sufficiently weak as not to dislodge the conclusion of Sidgwick’s Principle that in cases such as those of Birds and Rossianism suspension of judgement is required.

Audi also draws our attention to a third factor to be taken into account when assessing disagreement. Imagine that in Rossianism, your response to the disagreement with your colleague is to check and recheck your grounds for believing the Rossian view. You ask yourself questions such as: ‘Am I really clear on the content of these principles? Am I accepting the Rossian view partly because of my being so impressed when I heard what an upright and decent person Ross was? Am I taking properly into account the fact that consequentialists like my colleague can make room for Ross-style principles as secondary, practical aids to the maximization of the good? After all, surely one reason I accept those principles is the good that following them tends to do …’. According to Audi, your checking process itself increases your justification for your original belief. As Audi puts it: ‘the belief survives a kind of test’.

This is again an excellent and highly relevant point. And it is not undermined by the kind of cancellation process we saw in the case of the claim that disagreement is itself evidence of epistemic inferiority, nor by any application of Sidgwick’s Principle itself. That principle would be relevant only if you had no reason to believe that your colleague had not engaged in this degree of self-reflection and self-criticism. And in fact of course you do have such a reason, since scrutiny of one’s own beliefs of such depth is rare. But you might ask your colleague to do it, and she might agree.Footnote 13 It might be thought that honest self-scrutiny does provide at least a temporary way out of the implications of disagreement. This is in an important sense an empirical matter, and I suspect the truth may well be that in most real cases analogous to Birds and to Rossianism, practices of self-scrutiny would make little or no difference to the credence rational subjects will have as regards the beliefs in question. Given that this is the case, I suggest that we must again conclude that, though this factor does have some epistemic weight, that weight is not especially great.

As Audi has made clearer in his work than anyone else, and points out in Moral Perception (81), one should not believe that the only justification for a self-evident proposition must be based on a subject’s intuitive grasp of that proposition. It might be thought that, even if basic disagreement in intuitions results in an epistemic impasse, other possible inferential arguments for the propositions might break that impasse. In general, moral philosophers do not just assert their principles at each other, and then move on. Rather, they will engage in detailed and complex argumentation, seeking to support their own principles and to undermine those of their opponents. Unfortunately, however, such argumentation itself is open to the application of Sidgwick’s Principle. First, it rests on background theories and beliefs which are themselves basic, and over which there may be disagreement. Second, people differ in their judgements of the persuasive force of any given inference, and these people are required to suspend judgement unless they have reason to believe their opponents’ judgements of that force to be inferior.

The Limits of Moral Knowledge

Like Audi, I believe that moral knowledge is possible. In particular, there are certain self-evident propositions stating ultimate reasons for action, which if properly grasped under the right conditions can be known and be held with justified certainty. Again, however, we differ on the scope of what can be known, in part because of our difference on the role of lack of disagreement in the conditions for knowledge.

Note first that the Pyrrhonist scepticism I have been advocating has itself been limited to ethics. It may be that some broader application of Sidgwick’s Principle might require suspension of judgement on many more, and perhaps all, issues. If we do bracket this question, then there is at least one area in which it seems highly plausible to claim knowledge: one’s own good, and in particular one’s own hedonic experience. Consider the following principle:

Own well-being. Other things equal, any agent has a reason to advance their own well-being.

The ‘other things equal’ clause here ensures that the principle cannot be undermined by claims about deserved decreases in well-being or undeserved increases, of a kind which undermine, silence, or cancel the self-interested reason. Note also that the claim concerns pro tanto rather than overall reasons. It is not a statement of normative egoism.

This further, more specific claim is also a plausible candidate for moral knowledge:

Own pain and pleasure. Other things equal, any agent has a reason to decrease their own pain and to advance their own pleasure.

Because this claim is limited to pain and pleasure, it cannot plausibly be disputed on the basis of widespread disagreement about the nature of well-being. Any credible desire or objective account of well-being must allow room, in standard cases, for a pro tanto reason of this kind. There is dispute about whether one’s own pleasure and relief from pain must be desired, and whether if they are not desired there can be a reason to promote them. But this dispute can be dealt with reasonably enough through the ‘other things equal’ clause, unlike other more substantive disputes concerning, say, whether accomplishment is an independent component of well-being.

What about the following other-regarding principle?

Another’s well-being. Other things equal, any agent has a reason to advance any other person’s well-being.

It seems plausible both that the territory of plausible moral theories is fully occupied by variations of consequentialist and deontological theories, and that any credible version of each will include something like Another’s well-being. But it is important to remember the third ‘method of ethics’ Sidgwick rightly set alongside deontology and consequentialism: rational or normative egoism. So-called ‘veto egoists’ can accept the principle, noting that, though the reason to promote one’s own good will always trump an other-regarding reason in any conflict, the latter plays a practical role in deciding self-interested ties. Unfortunately, however, standard egoism has been one of the dominant positions in western philosophy, and though currently unpopular cannot justifiably be ignored. In other words, we must suspend judgement even on weak other-regarding principles such as Another’s well-being.

This is not to say that anything goes in the normative domain once we move beyond Own well-being and Own pain and pleasure. Recall Birds. If in a similar case some third person were to venture the opinion that the fairly small bird with a speckled breast was in fact a fully grown golden eagle, both you and your companion could with justification ignore it. Your companion is misremembering what a golden eagle is, joking, or just plain hopeless at identifying birds. In the same way, we can ignore claims that it is wrong to run round trees left-handed or to look at hedgehogs in the light of the moon, at least if these claims concern alleged ultimate reasons.Footnote 14 But within these bounds there is of course a huge range of possibly correct forms of deontology, consequentialism, and egoism, even if we restrict ourselves – as we have little reason to do – to contemporary western ethics. What are the practical implications of this high degree of uncertainty?

Living with Moral Disagreement

Note first that there is a good deal of reasonable disagreement over Sidgwick’s Principle itself, so to that extent it is self-undermining (see e.g. Elga 2010). But we can ask ourselves what the principle requires of us if it is correct, or what it would require of us were it to become a focus of informed, universal convergence.

As we have seen, Sidgwick’s Principle does not require us to suspend judgement entirely in the normative domain. There are certain views we can rule out, there are certain other views still in play, and there is a highly restricted set of principles we can claim to know. Each of us is in a position to know that we each have a reason to promote our own well-being. But what of the other views in the range, in particular deontological and consequentialist positions that may advocate the sacrifice of our well-being for the sake of others or some other moral goal? One view here is that we should act on dominant principles, so as in effect to minimize the expected wrongness of our actions.Footnote 15 Consider, for example, abortion. Some normative theorists claim that killing a fetus is wrong. If it were the case that all theorists believed bringing a child to term was permissible, then a pregnant woman could seek to avoid wrongness by avoiding abortion. One immediate problem, which will arise in many real cases, is that there is a non-trivial number of consequentialist theorists who believe that, given current constraints on resources, it is wrong to have a child. But another more serious problem with the idea of wrongness-minimization is that it itself is a matter of ongoing disagreement.

It might be thought that, given that Own well-being is so well grounded, and nothing competes with it, we should always act so as to further our own well-being. This may sound a disturbing suggestion, but it is important to note the degree of convergence at the level of secondary principles between normative egoism, deontology, and consequentialism. Given the internal and external sanctions of morality in our culture, for many of us there is little or nothing to be gained from breaking certain moral rules, such as those concerning the infliction of harm on others. But we also must accept that we have as yet no weighty argument against the egoist who can benefit by violating everyday moral norms. That is a problem that lay at the heart of ancient philosophy and modern philosophy in earlier centuries, but which to some extent has received less attention in contemporary philosophy at least in part because of the decline in influence of normative egoism. The problem itself, however, has not gone away.

Our main hope here lies in increasing moral convergence, both within and outside philosophy. We are aware of at least pockets of moral progress – for example, the widespread recognition of the irrelevance of race or sex in many contexts in which they have been thought significant for millennia. But what are the implications for philosophy? Complete shutdown is not called for, given the signs of progress I mentioned, which have been mirrored in philosophy. One might strive for complete impartiality. But the fact remains that, as we have seen, acceptance of Sidgwick’s Principle in most cases sits alongside certain moral appearances, and these appearances leave open opportunities for valuable argument and debate – significantly more in cases like Rossianism than in those like Birds. But carried out between those who have suspended judgement as to the correctness or otherwise of the view which they themselves find attractive, such debate would be less adversarial and more constructive than much in philosophy at present. This would have several significant advantages. First, each participant would be more likely to notice the faults in her own position and the advantages in those of others. Second, philosophers would see that there is often greater epistemic benefit in discussing issues with those of radically different views than with some clique of one’s own. Third, the aim of debate would be not the victory of one’s own position but convergence on some truth, which might well be a conglomeration of various elements from several existing ethical theories. Ethical enquiry must be informed by a spirit of open-mindedness, in which those who propose normative principles are prepared both to hold up those principles to the light of rational reflection and the arguments of others, actual or imagined, and to look enthusiastically at the views of others.Footnote 16 Critical argument, of course, would continue to be the mainstay of moral philosophical discussion, but if it were freed of its unjustified dogmatism there would be a greater likelihood of rational and considered convergence (which itself might provide justification, though cf. Kelly 2005) on the truth.Footnote 17 Here, I suspect, I find myself again in agreement with Audi, and in particular on many of the principles of epistemic modesty and humility he sets out in his 2011b.

Contemporary philosophy seems to me analogous to the following.Footnote 18 Imagine a group of equally experienced cavers, lost underground in the darkness, and wanting to get out as soon as possible. They share ideas, and it turns out that they all disagree on the best strategy. Only one of them, at best, can be right about the way to go. Now imagine that each decides to spend the time she has available trying hard to persuade her colleagues to agree with her, focusing only on what she sees as the weaknesses of their views and the strengths of her own. If I were lost, I would much prefer to be in a group whose members, though prepared to state their own views as well as possible, were also ready seriously to look for flaws in their own view and advantages in the views of others. Do we, or do we not, seriously want to find our way out of the cave? If so, then at the very least many of us should change the way we currently do moral philosophy.Footnote 19