All Souls Night


Although it is Parfit’s views about motivation, reasons, and ethics that concern me, I am going to start with a comparison at a little distance from his discussion, and indeed from moral philosophy altogether, in the theory of probability. We return to moral philosophy soon enough.

            Single case probabilities are those of particular events, such as Eclipse winning the 2.30pm race this afternoon, or my dying within the next twelve months. The contrast is with probabilities whose topic is general, and more naturally introduced with the indefinite article: ‘the probability of a child being autistic’ or ‘the probability of a philosophy book weighing over two kilos’, for example. Here frequencies provide natural truth-conditions, or truth-makers, but in the single case it is not so easy, and this motivates the pessimistic but plausible view that no satisfactory theory of such probabilities is to be had. Frequencies require sets of events, and putting singular events into sets does not help unless we have a principle for selecting the right sets. Any such principle is in danger of leaving all single case probabilities at either 0 or 1, although we will usually not know which, or, if the world is indeterministic, leaving them equally unknowable. Hence, probabilists such as Richard von Mises had banned them altogether, and scepticism about the notion—either counseling elimination, or at best an error theory— seemed to be the only scientific course.

            And yet outside the philosophy classroom the notion refused to die. Horse races still took place, and bookmakers thrived as before. Discussions about a particular horse’s chances filled sporting columns and bars. Insurance companies and actuaries still worked. People made up their minds, or changed them, oblivious to the pained cries of philosophers. So Ramsey, more charitable than von Mises, offered an explanation and vindication of the practices connected with voicing and discussing single-case probabilities. The manifestation, he said, of a sincere judgment of this kind is a distribution of confidence, and that in turn can most easily be regarded as a disposition to buy or sell bets at appropriate prices, under certain idealized conditions.[1]  Such dispositions, of course, are not true or false in themselves. But they have what I called a propositional reflection, in the single case judgment. Thus, if I am inclined to risk $1 on Eclipse winning the 2.30, in the hope of getting $2 back if he does, I can voice this disposition by saying that Eclipse’s chances are at least 50 – 50. The disposition is discussable, for it may be that this is a very foolish bet. If you know that Eclipse is off-form, or entirely outclassed by the field, or has shown symptoms of equine flu, you may helpfully seek to dissuade me. It is these discussions that fill racecourse bars and the sporting columns and the single-case probability proposition is the focus for them. This is only a sketch of Ramsey’s story, but it represented, in my view, a considerable improvement on the error theory of von Mises, or indeed any other proposal on the table. [2]  It is naturalistic, explanatory, and justificatory, all in one shot.

            Consider now a bookmaker Bill who is worth listening to on his subject. He offers odds more shrewdly than his competitors—that is, while offering better prices than they do, thereby attracting punters, he still makes a nice profit. Bill listens to gossip, looks at gallops, studies form more carefully than, say, a worse bookmaker, Kevin. Bill can offer 100 to 1 against horses like Eclipse, say, and still rake in a profit, although Kevin only dares offer 10 to 1 against it: a worse price, with fewer buyers, netting less profit.

            Nevertheless, Bill will not regard himself as infallible. There may be occasions when he kicks himself, not because an outsider won, since the improbable sometimes happens, but because he should have known better. He neglected a possible source of evidence, relied on a stable boy of whose plausibility he should have been more wary, and so on. ‘I was wrong’, he might say, ‘Eclipse’s chances were nothing like as bad as I made out’. In other words, Bill has, in thought or talk, a discursive practice of improving and refining his sensitivity to evidence, and thence his dispositions to buy and sell bets, and to make the judgments that are the propositional reflection of those dispositions. Kevin does the same, but is not as good at it as Bill.

            In all such thought and discourse, the single-case probability is quite naturally, and quite faultlessly, treated as a topic. But now the single-minded, or indeed myopic, philosopher comes along, and says that this is all very well, but what on Ramsey’s theory does the truth that Eclipse’s chances have one value or another consist in? This is not a particularly useful question, for as I discuss below, it is shoehorning a distinctively pragmatic theory about the nature of a practice, into one working in terms of truth-conditions. If it is insisted upon, it can get either of two answers. One is the unhelpful homophonic answer: it is true that Eclipse’s chances are high if and only if Eclipse’s chances are high. Alternatively, the response might be to point to the evidence—the standards in virtue of which Eclipse is to be regarded as a good bet, or the grounds for his being a good bet: such things as him being fit, trained, healthy, ridden by a champion jockey and having a better record than the rest of the field. These are not the things that identify the semantics of the term—‘being ridden by a champion jockey’ is no part of the meaning of ‘likely to win’. The semantic anchor of the judgment lies in the dispositions that make up the degree of confidence in the event. But these grounds are the kinds of things to which Bill is exquisitely sensitive. Kevin is not. [3]



We now turn to ethics, where, over many years, I have tried to articulate and defend a  parallel position, standing, for instance, to error theorists exactly as Ramsey stood to von Mises.[4] Moral and evaluative propositions are foci for the arguments and thoughts with which men and women discuss, reject, accept, ways of conducting their lives. We urge them on each other in order to change peoples’ practical inclinations: their motivations and concerns, their sense of honour, guilt and shame, or of what will do and what will not. So now I turn to examples of the things Parfit says, and see how they might sound if we applied them to Ramsey’s theory. Since I do not want to put words into Ramsey’s mouth, I shall invent a persona, Bramsey, to act as his spokesman.                  

Blackburn suggests that such attitudes (moral and conative attitudes) might be mistaken in the sense that we would not have these attitudes if our standpoint were improved in certain ways. But to explain the sense in which this standpoint would be improved, Blackburn would have to claim that, if we had this standpoint, our attitudes would be less likely to be mistaken. This explanation would fail because it would have to use the word ‘mistaken’ in the sense that Blackburn is trying to explain. (OWM, II, 16)

The parallel will be:

Bramsey suggests that betting dispositions might be mistaken in the sense that we would not have these dispositions if our standpoint were improved in certain ways. But to explain the sense in which this standpoint would be improved, Bramsey would have to claim that, if we had this standpoint, our betting dispositions would be less likely to be mistaken. This explanation would fail because it would have to use the word ‘mistaken’ in the sense that Bramsey is trying to explain.


But Bramsey already explained why Bill is a better bookmaker than Kevin, and why Bill might wish to be better still, and avoid the mistakes he regrets and that trouble his midnight hours. Similarly with ethics. The only asymmetry is that in actual betting the desirability of a particular upshot, winning the bet, is a given, whereas in ethics what counts as desirable is often itself contestable. But just as some people pretty much know what to look for in horses, so too in ethics we have our standards, and scramble about for grounds as best we can. Often enough one fairly uncontroversial improvement for most of us would be becoming better informed. Others might include becoming more sympathetic, or imaginative, or better able to prioritize or select information, or even coherent. So say I, and I hope you do as well.

            Parfit, however, believes that expressivists are not entitled to a conception of improvement.[5] They are not to be allowed to use the words ‘better than’. When I say that some evaluative opinion is defective, he asks rhetorically ‘in what sense are these opinions bad rather than merely different from Blackburn’s opinions’? (OWM, II, 394) Parfit does admit that we have to ‘give priority’ to our own present point of view (OWM, II, 395). But he thinks that he alone can explain ‘what it would be’ for our present beliefs to be mistaken: ‘they would be mistaken if they were false’. Since he also admits that expressivists can say that evaluations are false (see below), brandishing this explanation scarcely helps. But in any case anyone should be able to distinguish between saying ‘X is bad’ and ‘X diverges from my view’: in my midnight hours I can fear that there are divergences from my view that are not bad, and vice versa, since I have no doubt that I am imperfect, like everyone else.

            We all have standards and we all have views about who is worth listening to and what kinds of information they will be deploying to make them worth listening to. Anyone can be worried that he might be falling short by those standards or even by standards of which he is as yet unaware but which he should be grateful to learn, but that is not the same as worrying that he might be falling short in his opinion. One of his worries might be that his opinion of himself is unduly complacent.

In defending quasi-realism Blackburn also claims that some apparently external meta-ethical questions are really internal moral questions. That may be so. If we ask expressivists whether it is really true that acts of a certain kind are wrong, they can consistently answer Yes. But we are asking what it would be for conative attitudes and moral judgments to be true or false, correct or mistaken. This is not an internal moral question. Though Blackburn suggests he need not answer this question, that is not so. (OWM, II, 16)


In defending quasi-realism Bramsey also claims that some apparently external, meta-probabilistic questions are really internal probability questions. That may be so. If we ask expressivists whether it is really true that certain probabilities obtain, they can consistently answer Yes. But we are asking what it would be for betting dispositions and probability judgments to be true or false, correct or mistaken. This is not an internal probabilistic question. Though Bramsey suggests he need not answer this question, that is not so.


First of all Bramsey never says that betting dispositions are true or false (nor have I ever said that desires or other conative states are true or false).[6]  Betting dispositions are not true or false, or in that sense correct or mistaken, but in choosing one a person can certainly be foolish and in need of advice. Their propositional reflections, in the judgments that express them, are what may be said to be true or false, correct or mistaken. As with probability judgments Bramsey has the two kinds of answer already offered as to what this means: he can give the homophonic answer, which ought to be enough by itself, and he can go on to say in virtue of what evidence he makes any particular judgment. I don’t believe I have ever ‘suggested’ that I need not answer the question: I have only suggested that the answer that can easily be given is unlikely to be found interesting. [7]

            Parfit is himself in no position to insist that the question ‘what would it be’ for our judgments to be true or false needs a rich robust answer. When he is not just repeating that they are true or false, the answers he himself gives are no richer or more robust than mine. His own best explanations of ‘what it would be’ for moral judgments to be true or false consist in restating them in closely equivalent terms. ‘What it would be’ for something to be wrong, for instance, is that there are decisive reasons against doing it, and so forth. Like talk of things ‘meriting disapproval’ or ‘deserving censure’ these are moral claims, voiced from within a particular sensibility. It takes a moral vocabulary to voice them. Hence, as contributions to metaethics, they are of no value by themselves.

            Similar charges are repeated throughout OWM. Rather than go through them one-by-one, which, although there are further things to correct, would involve a good deal of repetition, I shall stand back a little and offer four mutually supporting diagnoses of what has gone wrong.


(1) The first is Parfit’s bizarre, bipolar, division of the realm of reasons into those that are ‘object-given’ and those that are ‘subject-given’ or ‘state-given’. This immediately blinds him to the actual structure of Humean theories. The Humean holds, in St. Augustine’s lovely phrase, that  ‘in the pull of the will and of love appears the worth of everything to be sought of avoided, to be thought of greater or less value’. In other words our reaction to what our reason determines to be the facts of the case is also a function of our concerns (passions, desires, inclinations). It is not that we typically attend to our passions or desires, for our beliefs about what we desire do not serve as premises in any kind of inference. But the passions themselves are nevertheless determinants of what we adopt as reasons for choice or action. They have to be, if reasons are to engage the will. Kantians differ from Hume on this, certainly, but that is no reason for denying Hume his theory, nor is there any serious reason for thinking that Hume is wrong about it.

            To the Humean, ‘object-given’ reasons exist, of course. That there is a bull in a field gives me a reason to detour around it. That is because I am rather afraid of bulls. It might give some young Spanish blood an opportunity to go and pirouette at it. The difference lies in our feelings about bulls, or about danger, or the sources of pride or shame in each of us.


(2) Second, Parfit is ungenerous, to say the least, in the resources he allows to the Humean tradition. He supposes that Hume can only talk, as he constantly does, of policies, precautions, demands, rules, and other elements of practical life being reasonable or not, by forgetting his own views. But that mistakes the entire point of Treatise Book II, part III, sect. 3, and equally of Book III, part I, sect. 1. Hume is not concerned to circumscribe the use of the terms ‘reasonable’ and ‘unreasonable’ as terms of approbation or the reverse, but to discover the influencing motives of the will and to locate them partly in our passionate natures.[8]  A Humean can perfectly well say that if I become scarlet with rage because someone overtakes me on the motorway, I am unreasonably angry, but he will go on to add that it is my passionate nature that is at fault, not my perception of the road nor my perception of the ‘relations of ideas’, or in other words my capacity for such things as logic or mathematics, or even my ability to put my maxim through the formula of universal law. This last is especially likely to be impotent, because such a person probably already thinks that other people are just like himself, when they have been provoked as he has just been.


(3) Thirdly, I find it curious that, in a work so concerned, rightly, to distinguish normative from descriptive issues, Parfit disallows Hume or myself, to do the same. Thus, talking of Williams’s confirmed wife-beater, he imagines me saying ‘you have decisive reason to treat your wife better, whether you want to or not’, and believes that I can only say this if I withdraw some other claims (OWM, II, 458). I cannot imagine what he has in mind. I would prefer to put my distaste at this character by saying ‘there is decisive reason to treat your wife better, whether you want to or not’, which makes it slightly clearer that I am moralizing rather than just describing this person’s profile of concerns. The decisive reason—the one that should motivate any decent person to desist—is his wife’s distress. I do not particularly commend the other locution, according to which ‘he has decisive reason’ since it runs more risk of being taken simply as a false description of the agent’s psychology, instead of a true evaluation of his situation.

            Parfit then describes my distinction, in Ruling Passions, between reasons as described by Humeans and Reasons, with an upper-case ‘R’, as described by Platonists, Kantians, and intuitionists. He says that I ought to revise my view to claiming that ‘there are object-given Reasons. There are some reasons that everyone must acknowledge, whatever their sympathies and inclinations’. (OWM, II, 459)  This, allegedly, would put me ‘closer to Hume’s real view’. It’s particularly hard to follow the train of thought here, since it seems to suggest that really Hume endorsed something approximating to Parfit, or perhaps Plato or Iris Murdoch, celebrating goodness or norms or reasons as primitive realities that ‘would give us reasons in the way the sun gives light, ‘because it is out there, shining down’’ (OWM, I, 46). I do not think this interpretation will gain many adherents among Hume scholars. My own immediate response is that ‘there are some reasons that everyone must acknowledge, whatever their sympathies and inclinations’ is not normally taken as a remark about transcendental sunshine. It is naturally taken as a moral remark, and one that I naturally applaud, just as I might say, for example, that you must pay your debts if you can, whether you want to or not. I do not advocate letting you off the hook because you don’t want to pay. But if it were a descriptive remark, entailing that people must of physical or metaphysical necessity acknowledge some particular things as reasons for some particular behaviours or concerns, from which it follows that they do so acknowledge them, whatever their sympathies and inclinations, then it is on the face of it false, and saying so puts me entirely at one with Hume (and if ‘acknowledging’ a reason includes motivation by it, with Kant). The ambiguity or slide I am describing is not particularly difficult to spot: ‘you cannot be a sensible knave’ might be an injunction or plea to someone tempted in that direction, or, much less likely, a crashingly false assertion entailing that a particular option is one that human beings never take. Once more, then, we have a simple confusion between assessment of someone’s moral situation, and a description of the psychological elements that actually play a role in motivating them.


(4) The fourth point cuts deeper philosophically. The pragmatist tradition, of which Ramsey is one example, and expressivism another, takes as its starting point not the ‘truth condition’ or the ‘analysis’ of a proposition, but the activities in which it occurs. It is the activity of giving and accepting odds that is Ramsey’s focus, and those of voicing admiration and other attitudes that are Hume’s focus. If we think, as perhaps ‘analytical’ philosophers have occasionally thought, that providing analyses and ‘accounts’ of facts and truths exhausts the business of philosophy, then we shall be blind to the merits of many philosophers, from Berkeley to Wittgenstein. Hume is especially made invisible by this prejudice, both in his writings on moral philosophy, and those on causation. Hume is very little concerned with ‘analysis’ in the twentieth-century sense, but centrally concerned with the mechanisms of the mind that eventually issue in our making the kinds of judgment we do. He is often much more nearly a naturalized epistemologist than anything else, and of course his pyromaniac attitude to metaphysics, and that would include ‘analytic metaphysics’, is always in the background.

            An example of insensitivity to this tradition comes when Parfit endorses Andy Egan’s  attempt to land me with a priori certainty that my own stable moral beliefs are true. [9] He writes:

Blackburn might instead reply that, on his view, each of us could still claim to know that our own judgments were true. We can talk of ‘knowledge’, Blackburn writes ‘if we rule out any possibility that an improvement might occur’. But we cannot turn our judgments into knowledge merely by claiming that we could not possibly be mistaken. And people with conflicting judgments might all make such claims. Blackburn’s claim would then mistakenly imply that contradictory beliefs could be true, and count as knowledge. (OWM, II, 396)


But of course I never dreamed of saying that whenever we talk of knowledge it follows that we are right to do so. It is the human practice that is the explanandum, and our practice with epistemic vocabulary is, I hold, insightfully described by seeing a claim to knowledge or the allowing of such a claim, as an evaluation of the remoteness of the possibility that improved positions should undermine confidence—with the consequent removal of motivation to further investigation and inquiry.  I think this approach, seeing a claim to knowledge as a kind of stop-valve, is much better than that of post-Gettier ‘analyses’ of ‘truth conditions’ for knowledge claims which have spectacularly led nowhere over nearly fifty years. [10] Obviously claims to know things can be false. Inquiries can be closed too early, when further investigation would indeed have reversed a judgment. Seeing someone claiming to know something when in my view it’s a very open question whether what they claim to know is true, I refuse to allow their claim, precisely because in my view the issue is not closed, and further thought or investigation might reverse their verdict. Seeing someone claiming to know that p while someone else claims to know that ~p I can allow that at most one of them is right, or, very often, if they are each sensible enough fellows, that the confidence of each of them is unwarranted, and they are both premature.



I have offered these four suggestions in a friendly spirit, as perhaps explaining things that might otherwise remain bewildering in OWM. For I confess that on first reading in it my understanding was quite baffled. I could not construe much of Parfit’s discussion unless he genuinely believes that ‘these people’ cannot appeal to reasons at all.[11] Reasons, he appears to think, are his own private property. The rest of us flounder in darkness, cut off from the sunshine beaming down on Parfit. But he himself has written that ‘reasons for acting are facts that count in favour of some act’ [12]—so does he really believe that Humeans are somehow debarred from looking for, finding, selecting and prioritising such facts and features? Can Augustine not take the fact that it is raining as a reason for carrying an umbrella? Or take the fact that its food is inedible as a reason for avoiding a restaurant, or take the fact that it’s not a bull but a cow as a reason for relaxing? And so on and so on? Does Parfit really think that Hume and Hare, Williams, Gibbard and myself, are debarred from the most elementary forms of practical life? [13]

            Perhaps he does. He repeats many things along these lines: Blackburn (or Gibbard) cannot appeal to value-based, object-given facts about the objects of desire since ‘we have no reason’ to want certain things for their own sake. The phrase is a bit odd, since if I want something for its own sake, I typically need not have some other reason for wanting it.  Waiving that for a moment, I am certainly happy wanting some things for their own sake. I want the welfare of my children for its own sake, not for the sake of anything else, such as their ability to support me in old age or to perpetuate my works in the future. And the question of whether my children are happy or flourishing is ‘object-given’ in any sense I can imagine to matter—it is a question of how things stand with my children—while the fact that it matters to me as much as it does is a fact about my values, sentiments, and concerns. The fact that the welfare of my children matters little to Parfit is a fact about his.[14] Do I have a reason for wanting the welfare of my children for its own sake? As I say, the question is more than a bit off-colour—if I want it for its own sake why should I need to scramble around for a reason? Wouldn’t another reason mistake the phenomenon, distracting my thoughts from where they kindly rest, which is on my childrens’ welfare? But I certainly think I am reasonable in wanting it. It is not as if I am likely to admit criticism of any movement of mind that led me to wanting it.[15] It is not as if I feel ashamed of my care and concern. I wouldn’t want a mad Kantian or Stoic or Buddhist therapist to take it away. [16]


I shall end by drawing attention to two other issues. The first is quite short, and the second more interesting. The short one is that Parfit quotes me as saying that inconsistent desires (that is, desires that cannot both be realized) ‘can be faultlessly inconsistent’ (OWM, II, 391) and says that this stands in shameful contrast with the use I make of ‘desire or other conative states’ when I develop quasi-realism, which, of course, requires me to see people as disagreeing with each other in moral judgments. Alas, I am far from ashamed. I only said that desires can be faultlessly inconsistent. And I had in mind only impractical desires—ones at the same end of the spectrum as idle wishes. Here is the footnote from which Parfit extracts the saying:

One reason why I do not think ‘desire’ is a good response to work with is that evaluation is an activity that imposes norms of consistency whereas, in their less practical manifestations, desires can be faultlessly inconsistent. It is only when we come to do something about them that we have to tidy them up. [17]

I find it almost incomprehensible that Parfit could miss the clear point of this footnote, much expanded throughout the book.[18] The point was that the attitudes that gain expression in moral and evaluative discussion are typically more than ‘mere’ desires, although of course sharing their practical direction of fit. This is why I preferred the word ‘concern’ and emphasized that in expressing a concern or attitude one is putting something into public space for acceptance or rejection. Morality is centrally focused on the social world. It is about what we are to choose, condemn, encourage, avoid, forbid, insist upon (Hume, of course, realized that this need to coordinate on a ‘common point of view’ was the essential core of morality, explaining why we had impersonal terms of approbation or disapproval in the first place [19]). And secondly, the whole point of having the moral proposition, the reflection of our practical stances, as a focus for discussion, is to enable us to make public these things. It is in intention a device to further coordination. And then disagreement comes in.

Suppose, to take a simple example, that my wife prefers hill-walking to sailing, and I prefer sailing to hill-walking. Our desires are different but there is no disagreement. However, if we are both to take a holiday together, then plans reflecting those preferences are inconsistent; we have to do something about it, and we may well disagree what it is. Gibbard’s gerundive judgments ‘hill-walking is the thing to do’ or ‘sailing is the thing to do’ are the focus for the attempt to coordinate and reconcile our preferences. That is what the propositional judgment is for. When it is a moral issue rather than one of whim or taste, we enter the territory of shame and guilt, anger and resentment. We are further up the scale of importance, and coordination matters more. One of our concerns might be never to do one thing or another, and then we conceive of ourselves as being under a necessity. But it is we who put on the yoke of necessity, as the Chorus so perceptively said of Agamemnon.[20] Ajax, having slaughtered only sheep, felt he had no alternative but to kill himself, whereas anyone of lesser mettle might have lived putting up with the other heroes’ sniggering, and a sufficiently barefaced anti-hero might have gone on insisting that they weren’t sheep but men.

Parfit is fond of saying that headaches cannot conflict, or be mistaken or unsoundly based. (OWM, II, 393 ff.) But neither are they subject to discursive pressure, from our own thoughts or those of others, and nor are they matters on which coordination is an issue. Plans often are. Parfit pillories Gibbard for allegedly implying that they all are, which he can then refute with Buridan’s ass cases, in which planning to go for one bale of straw does not imply disagreeing with, or rejecting as bad, a previous plan to go to the other. But all Gibbard needs to do is section those off, and talk of the same kinds of publicly expressed, gerundively-girded plans that require coordination and perhaps unanimity. When we start to reach for normative vocabulary, we signal that this is what we are about.

The last thing on which I want to comment is one where I am in the dock with unusually large and distinguished company, including Hare, Williams, Korsgaard, Nowell-Smith, and Gibbard.[21] These writers have each charged, as Hume did in the sections I mentioned above, that a pure rational intuitionism fails to engage with the practical nature of ethics. But Parfit cannot see that there is a problem. Suppose there is a moral judgment which we accept, to the effect that we ought to do something. These writers fear that on a purely rationalist or cognitivist account a gap opens up: we might accept the judgment, but what then? Parfit construes us as asking whether we should take account of the judgment. But, he complains, this is asking whether we should do what we should do­ —and that is hardly an open question. Hence rationalism’s critics cannot frame a question that is not answered by the same cognitive judgment that is already premised.

This is not at all a correct reading of the dialectic. If it were, we would indeed have a kind of standoff. The critics fear a gap between cognition and practice. Parfit replies with another cognition. He sees this as answering any question they intend to ask; they see it as just more of the same, leaving as big a gap as before.

To undo this logjam we should hear the critics differently. They are asking how, on the cognitivist view, motivation and cognition are related. And to make the question pointed, let us rephrase it in this form: could there be a society with impeccable cognitions, ones that pick out just the right ‘normative reasons’ or ‘object-given reasons’, but whose awarenesses of these (including their awareness that they are reasons) do not engage with any motivations: any concerns, desires, attitudes, endorsements, or feelings? [22] The answer, such critics feel, ought to be that there could not. In the absence of the motivations the ‘cognitions’ could not be construed as making the morality of the society. They require other interpretation. But Parfit’s iterated ‘should’s do not get him so far. Their ‘direction of fit’ remains, in his eyes, representational, and he can provide no necessary connection with any states of mind with the motivational direction. All he will be able to say, in tackling this question, remains in the sphere of cognitions. Our thoughts, so pure and elevated, turn to the sun, revolve in the light of the sun, bask in goodness which, like mathematics, lies far above our contingent concerns and desires, passions and motivations. Of course, Parfit and other realists of his stamp may plead and cajole that we make these transcendental celestial cognitions practical. But it remains logically possible (or on the face of it, quite likely) that nobody listens, in spite of their having the most perfect repertoire of moral judgments that they sincerely accept. And if nobody listens, then their motivations may be totally and blithely orthogonal to their cognitions. They have moral understandings, but across the board these are completely inert.  It is the fact that his intutionism opens this possibility that the critics with whom I am numbered rightly condemn.

            I have, perhaps selfishly, concentrated upon elements in OWM which particularly grated on my ears. If this means that I have sounded too self-centred, I apologise. My excuse is only that I think I speak for a significant number of scholars, philosophers and psychologists, who look with some bemusement at the direction of contemporary metaethics. Why it has taken this turn must be a question for sociologists rather than philosophers, for we deal only with the arguments, and these, unfortunately, seem to offer no explanation. [23]


[1] More generally, to incur costs or benefits in acting on the supposition that an event will occur. Thus confidence that a large asteroid will hit the earth shortly would not be manifested in betting on it, since there will never be a chance of collecting on the bet, but it could be manifested in what would otherwise be behaviour to be avoided as too costly, such as spending all one’s retirement savings on wine, women, and song.

[2] Of course in ‘Truth and Probability’ Ramsey does much more. (A) He imposes a logic, showing how the need to avoid being in a state in which you could lose whatever happens implies conforming to the classical probability calculus; (B) He develops a joint theory of value and probability: a way of solving both for a scale of values and a scale of probabilities jointly. (C) He frees up the dispositions in question from even idealized betting, to more general measures of value and belief jointly.

[3] I am indebted to Jamin Asay for pointing out that it is quite consistent for the expressivist to describe these grounds as ‘truthmakers’ for the normative or probabilistic judgment, but with a distinctive non-metaphysical account of the truthmaking relation. See Asay, JaminTruthmaking, Metaethics and Creeping Minimalism’, Philosophical Studies, forthcoming.

[4] We could expand the discussion to cover not only practical but theoretical reason, and the norms in play governing our conduct of it. But for brevity I shall talk only of practical reason.

[5] It is not only expressivists, however, whom he might be fighting. Moral sense theorists presumably stand in the dock as well; most of his objections apply as much to writers such as John McDowell and David Wiggins as they do to me. Hume interpreted as a moral sense theorist would not escape Parfit’s wrath. In fact expressivism as such might not be itself a sensible target for him, given his relationship to Kant, whose morality of imperatives and law is naturally interpreted as expressivist but who thought that obedience to some imperatives or laws is demanded by pure reason.

[6] Other expressivists may say this. Michael Smith, for instance, favours the idea that moral commitments may belong to the subset of beliefs that are also desires. I prefer to say that they are practical commitments that are voiced in sentences that naturally take on the clothing of expressions of belief. But nobody says that they are both capable of truth or falsity, and not so capable. Perhaps the combination of seeing a commitment in terms of mental states with a practical direction of fit, and seeing it in terms of a propositional reflection or expression of such a state, is not easy to comprehend at first glance. It preoccupied Kant, as he discussed the antinomy of taste. But if a student writes that silly old Kant thought that disinterested pleasure in an object is both incapable of truth (falls under no rule, etc.) and capable of truth, (the resolution of the antinomy of taste) is he or she hailed as a first-class commentator on the Critique of Judgment, instead of someone who is just at the starting line of that great work?   

[7] Parfit may have been misled by my allowing that if asked what it is for a value judgment to be true, I am ‘not very forthcoming’. But he had, after all, read the entire passage for he later quotes it (OWM, II p. 398). It says: ‘Just as the quasi-realist avoids naturalistic reductions, so he avoids saying what it is for a moral claim to be true, except in boring homophonic or deflationary terms. The only answer we should recognize to the question ‘what is it for happiness to be good?’ is happiness being good’. This, far from suggesting that I need not answer the question, answers it. Hearing it in a different register perhaps I should also have gone on on to cast around for other features of happy states, which help to ground the judgment that they are good, just as Bill can cite the features of Eclipse that make him a good bet. But in this case it would be easier to draw a blank.


[8] I believe there is but one sentence in the entire discussion that Hume needed to have amended: ‘actions may be laudable or blamable; but they cannot be reasonable or unreasonable…’ from the beginning of Bk III (Treatise, p. 458). But in the context it is clear that he is still talking as an anatomist, not as a moralist. In other words he is still talking of which faculties are in play when we make moral distinctions (more than those of cognition), not about which words we can select to abuse people who do silly things, or commend those who avoid them. For similar reasons a Humean need not be afraid of calling some of Parfit’s queer specimens ‘irrational’, signaling an inability to understand or get on all fours with them, and a conviction that this is not due to them having a superior understanding of anything. I certainly couldn’t get on all fours with people who don’t care about agony on Tuesdays, and so forth. Hume himself never uses ‘rational’ or ‘irrational’ in his writings on morals, except in the first sense that the OED gives, to designate creatures capable of using reason, or not.

[9] Egan, Andy. ‘Quasi-Realism and Fundamental Moral Error’, Australasian Journal of Philosophy v. 85 (2): 205—219. My rebuttal, which was actually made available to Parfit, is ‘Truth and A Priori Possibility: Egan's Charge Against Quasi-Realism’, Australasian Journal of Philosophy 2009, v. 87 (2): 201-213. Also on the web at

[10] The work which best explains this is Edward Craig’s neglected masterpiece, Knowledge and the State of Nature (Oxford: Oxford University Press, 1990)

[11] Perhaps he has been lulled into this view by works such as Christine KorsgaardSkepticism about Practical Reason The Journal of Philosophy Vol. 83, No. 1 (Jan., 1986), pp. 5–25, or Elijah Millgram,  ‘Was Hume a HumeanHume Studies, vol 21, 1995, pp. 75–94. But neither Hume nor Humeans are skeptical about practical reason; they give highly plausible accounts of it. They are at most skeptical about pure practical reason, which is a different thing altogether. Even in Korsgaard’s view what she calls ‘content skepticism’ about pure practical reason is a formidably plausible position.


[12] Parfit, D. ‘Reason and Motivation’ Proceedings of the Aristotelian Society, Supplementary Volume 71, 1997, p. 121.

[13] One of the commentators in OWM, Allen Wood, evidently does, holding that Humeans are ‘either radically defective specimens of humanity who are incapable of feeling respect for anyone or anything, or else every time they do feel it they commit themselves to contradicting their own metaethical theories’ (OWM, II, p. 62). I find it sad that this radically defective specimen of writing should have been passed to appear in a book with a reputable publisher. If we descend to this, what sermon might I not preach on those who think their own attitudes but reflect the instructions of the universe, and what circle of hell should be reserved for philosophers who weave fantasies encouraging them to think that this may indeed be so?

[14] Perhaps using Parfit’s terminology I might say that my reason for cherishing them is object-given sure enough—but only given to a certain kind of subject, of which I am one and he is not.

[15] The importance of this concept is developed in my paper ‘The Majesty of Reason’ first published in Philosophy, 85, 2010, pp. 5–27 and reprinted in Practical Tortoise Raising (Oxford: Oxford University Press, 2010, pp. 283–307.

[16] I am well aware that most current interpretations of Kant try to ignore the bits about sympathy and warm-hearted fellow feeling being burdensome, and things it would be blissful to shed. However see Critique of Practical Reason, trans. L. W. Beck, Macmillan 1956, pp. 118–9.

[17] Ruling Passions (Oxford; Oxford University Press, 1998) p. 118, footnote.

[18] In the Introduction to Spreading the Word I wrote that in my opinion too few philosophers framed the golden words of Quinitilian, the Roman rhetorician, above their desks: ‘Do not write so that you can be understood, but so that you cannot be misunderstood’. On a subsequent occasion Bernard Williams protested to me that this was an unrealistic aim: one can always be misunderstood, and there may often be deep psychological forces making it almost impossible for some people to understand you. This is indeed so, and this particular part of OWM sensitized me, not to the impropriety of Quintilian’s ideal, but to the just pessimism of Williams’s reservation.

[19] David Hume, Enquiry Concerning the Principles of Morals, 9.1. 

[20] Aeschylus, Agamemnon, l. 217.

[21] The discussion of ‘the normative argument’ occupies OWM, II, 413, ff.

[22] Kwame Anthony Appiah has recently explored, in his superb book The Honor Code, ways in which whole societies can be locked into practices which, at some level, they know to be immoral, and which can only be combated by mobilizing a sense of the way they dishonor them in the eyes of others. I certainly accept the possibility, although it can only exist against a general background of motivation by judgment of right and wrong. Compare money: a person may provide goods and services without wanting either goods and services or money in return, nor, if he is offered money, need he care how much. There might even be a society-wide area in which money is not to be thought of, such as close family transactions. But unless there were a general habit of caring about money and its quantity, it wouldn’t be money.

[23] I owe thanks to Jamin Asay, Jonathan Bennett, Michael Gill, Michael Smith, Peter Godfrey-Smith, Valerie Tiberius, Stan Husi, Kate Manne and David Wiggins for valuable encouragement and commentary.