Derek Parfit's Tanner lectures, published with comments and replies in two enormous volumes, must be considered, by any reckoning, as a landmark in moral philosophy.Footnote ¹ Parfit spells out his striking synthesis of contractualist and consequentialist conceptions of moral reasoning, with a view to giving the unique and final answer to the question implied in his title: On What Matters. The result is wide-ranging, lucid, and endlessly engaging, the product of a mind that is relentless in its pursuit of valid argument and beautifully attuned to hidden fallacies.

The two volumes contain an introduction by Samuel Scheffler, extensive commentary from Susan Wolf, Allen Wood, Barbara Herman and Tim Scanlon, elaborate replies to some of their arguments, and a book-length quantity of appendices devoted to ancillary aspects of the argument. These additions weave long paper chains around the central concepts and account for much that is fascinating in Parfit's enterprise. Notwithstanding the additional forces that Parfit marshals, however, he failed to convince me that rule consequentialism is any more plausible at the end of his argument than it was at the beginning.

In Reasons and Persons, which appeared in 1984, Parfit defended a teleological ethic against Kantian deontology, arguing that moral reasoning aims to make the outcome of our acts as good as possible. He supported a form of Act Consequentialism, while arguing that the emphasis placed by ordinary morality on the distinctness of individual persons relies on an unfounded view of personal identity. Without a robust concept of identity we cannot block calculations that weigh one person's cost against another person's benefit. And such a view of identity, Parfit argued, is not available to us. All the considerations that had made utilitarianism seem counter-intuitive to other moral philosophers, were swept aside by an argument that was boldly metaphysical as well as relentlessly dismissive of the values by which less cerebral mortals live. The book was a tour de force, advocating an esoteric moral doctrine which, because nobody could possibly live by it, nobody had any motive to refute.

On What Matters contains no arguments about personal identity, and allows the reader to assume that, when Parfit refers to persons, he means enduring individuals who identify themselves in the first person, and as the same at different times. Nevertheless, the aspect of the moral life that most obviously depends on this – the practice of praising, blaming, punishing, rewarding and in general holding self and other to account for what we are and do – is more or less unmentioned. There is, in Parfit, a distinct aversion to the aspects of the moral life that suggest the entanglements that make us what we are. Every now and then, it is true, he acknowledges that others do not matter to us equally, and that their claims on us may be more or less demanding, more or less rewarding, more or less strong. But when it comes to considering what matters in itself this fact sinks into the background, to reappear only as a qualification to other and more abstractly grounded features of our condition.

Consider, for example, the love for our children which, among normal people, fuses all the circuits in the utilitarian calculator. For Parfit this is just another input into a trolley problem – in this case a problem concerning whom we should rescue from drowning. He writes (Vol. 1 page 385) that ‘the optimific principles would not… require you to save the strangers rather than your child. If everyone accepted and many people followed such a requirement, things would go in one way better, since more people's lives would be saved. But these good effects would be massively outweighed by the ways in which it would be worse if we all had the motive that such acts would need. For it to be true that we would save several strangers rather than one of our own children, our love for our children would have to be much weaker.’ And that, Parfit goes on to argue, would have many bad effects in the long run.

What is remarkable about this line of reasoning is that, even if it upholds common sense, it does so on grounds that entirely undermine the obligations on which common sense is founded. It ignores the fact that our children have a claim on us that others do not have, and that this claim is already a reason to rescue them in their hour of need, and needs no further argument. It ignores, one might say, the human reality of the situation that Parfit claims to be imagining, in favour of the spectral mathematics that provides the measure for all his comparative judgments.

In the second volume, in a long and careful discussion of meta-ethics, Parfit defends what he calls ‘cognitive non-naturalism’ or ‘non-naturalist cognitivism’, which is the view that there really are normative facts, but that they are not identical with any natural facts. In the first volume he argues that normative reasons are object-given, rather than subject- or state-given. It is not the state-given fact that I desire X that gives me a reason to pursue it, but the object-given fact that X has value or is good.

The thesis that reasons are object-given is defended under the name of Objectivism, and Parfit amasses a variety of arguments against the opposing Subjectivism, which is the view that reasons for action are given by our states of mind, notably by our desires. He also distinguishes partial from impartial reasons, the first directing us to what is best for us, the second to what is ‘impartially best’, or best as recognized by any rational being who gives no special weight to concerns of his own.

There is a question as to what ‘best’ means in this context. ‘Best’ is the superlative of ‘good’, but goodness is not, in Parfit's eyes, the only value: there is also, for example, dignity. To put it simply: the good justifies desire; dignity invites respect. (Vol. 1, Chapter 10.) Maybe there are aesthetic values too, but their absence from the narrative is one of several significant lacunae. In any case, the need to say more about the meaning of ‘best’ does not dwindle, but grows as the argument proceeds. This is because Parfit defines Consequentialism as the claim that ‘whether our acts are right or wrong depends only on facts about how it would be best for things to go’ (Vol. 1, page 373). ‘Best’ here means ‘impartially best’, and it is in terms of this definition that Parfit makes out his case for Rule Consequentialism, which he presents as part of what he calls the Triple Theory. This claims that ‘an Act is wrong just when acts of that kind are forbidden by a principle that is:

1. (1) one of the principles whose being universal laws would make things go best (Rule Consequentialism),

2. (2) one of the only principles whose being universal laws everyone could rationally will (Kantian Contractualism),

3. (3) a principle that no one could rationally reject’ (Vol. 1 page 413).

Parfit calls this last part of the theory Scanlonian Contractualism, since he derives it from the argument of T.M. Scanlon's What We Owe to Each Other,Footnote ² in which the reasoned consent of each relevant individual is made fundamental to moral decisions.

Philosophers have thought that Kantian deontology, the liberal theory of the individual veto and the consequentialist search for the best outcome are separate and incompatible approaches to morality. Not so, argues Parfit. Properly understood they are climbing the same mountain from different sides. The three conditions above identify precisely the same set of principles, namely those that, universally adopted, would make things go best. Those are the principles (the ‘optimific principles’) which rational beings would will to be universal laws, and which none of them could rationally reject.Footnote ³

In the course of arguing for the Triple Theory Parfit offers a careful analysis of Kant's formula of humanity, according to which we must act so as to treat humanity always as an end and never as a means only. He points out that there is a great intuitive difference between treating people as a means without harming them, and treating them as a means regardless of the harm. In the course of his discussion, however, the fundamental motive for Kant's appealing and suggestive formula seems to evaporate. Kant aimed to show that the universal principles of morality come to earth in the encounter between individuals, and that the moral law causes all calculation to stop at the threshold of the other person, beyond which we cannot trespass without his consent, and in certain circumstances not with it either. However good the consequences of some course of action, they cannot be pursued by coercing or destroying the innocent person who stands in its way. That is the obstacle presented by Kantian deontology to the Consequentialist, and Parfit reconstructs Kant's formula of humanity in order to smooth the obstacle away.

I doubt that he succeeds in this enterprise, however. Kant's formula of humanity reflects the intuition that, in the I-You encounter, the other has an equal standing with the self. When my actions impinge on your freedom then I must solicit your consent to them, by offering reasons that justify me not in my eyes only, but also in yours. I make your will equal to mine. My reasons aim to be your reasons, and your reasons in turn to be mine. It is this mutuality of practical reasoning that is the root of moral judgment.

That view has been defended in other terms by Stephen Darwall who, in a lucid review, argues for a position that Parfit would describe as subjectivism. For Darwall ‘reasons are, in their nature, considerations that agents can see as reasons from the perspective they take up in deliberating about what to desire, believe, and do’.Footnote ⁴ It is true that reasons are based in object-given facts, but facts seen from the perspective given to agents by their beliefs and desires. Modern Humeans would go further, and argue that it is only against the background of our emotions, attitudes and desires that facts can provide us with reasons, and that without that background the facts are simply inert. Parfit does not accept this. His object-given reasons are reasons regardless of our motives, since they are addressed to everyone in general, or no one in particular, depending how you look at it. They tell us what we should do, and if they fail to motivate us that is simply proof of our irrationality.

In this matter Parfit is, of course, at variance with a long tradition in moral philosophy, from Hume and Hutcheson to Williams, Blackburn, Korsgaard and others today. Philosophers in that tradition worry about the connection between reasons and motives. If reasons state features of the object that we can recognise regardless of our motives, then there is a gap between accepting them and acting on them. Unless we are motivated to act on them, reasons have no weight for us. The ability to recognize them, to make accurate extrapolations to the ‘optimific principles’ from which they flow, to use them in determining whether it is right to throw the fat man from the bridge or to rescue five 60-year-olds rather than three 30-year-olds from the rising tide – these purely cognitive abilities surely do not, yet, amount to a morality, if they do not move us to act on them. Parfit does not recognise the force of that objection. Humeans of that kind, he believes, are asking whether, having recognised the presence of objective reasons, we should act on them. And this is merely asking whether we should do what we should do. (See Vol. 2, pages 423ff.)

Clearly, someone here is going round in circles – either the one who says that reasons are reasons when they state the right kinds of fact, or the one who says that reasons are reasons when they invoke the right kind of motive. But whoever wins the argument, the fundamental difficulty won't go away, the difficulty noticed by Hutcheson when he distinguished ‘exciting’ from justifying reasons, and by Kant when he argued that the objectivity of moral judgment requires that reason alone be a motive to action. Kant argued that reason could be a motive to action, and would be so when detached from all ‘empirical conditions’: as when I ask what ought I to do, regardless of my interests and desires. However, Parfit's non-naturalist cognitivism does not raise the question of motivation. It tells us merely that reasons exist, and that they are not ‘natural’ properties of the objects that exhibit them. Hence there could be experts in discerning these reasons, who sit at their desks calculating answers to the endless dotty problems that fascinate Parfit, but who have never inclined to lift a finger to help, or a fist to reproach, the humans whose antics they view from their college window. If that is what morality is like, then it seems odd that we should pay so much attention to it or think that it tells us what matters.

Parfit is surely right that we think of moral reasons as given to us by objects and not merely projected into the world by our attitudes. Those whom Parfit describes as subjectivists need not deny that. They wish rather to affirm that whether a fact gives me a reason to act depends in part on my desires and attitudes. This does not mean that our reasons are ‘merely subjective’, in the sense of giving no independent foundation to what we decide to do. Moral attitudes have intentionality: they are cognitively rich, open to rational criticism and imaginative enhancement – so much Simon Blackburn has shown in an admirable defence of what he calls ‘quasi-realism’, a defence that Parfit carelessly dismisses in a short section of Vol. 2.Footnote ⁵ Moreover, as I suggest below, whether we are subjectivists or objectivists at the meta-ethical level, the problem of motivation will remain. Before returning to the point, however, we need to be a little clearer about the idea of the ‘best’.

Rule Consequentialism is attractive in part because it makes sense of an important feature of moral judgments, namely that they are often comparative. Kant's moral philosophy seems, at times, to take little account of this and also to have great difficulty in explaining it. In our most urgent moral dilemmas we ask ourselves which of two courses of action would be better, or which among a number of actions would be best. This fact is easily dealt with on a consequentialist view – too easily, some would say. Consequentialists treat moral reasoning like economic reasoning, and sometimes spell out their thoughts in terms of preference orderings and their aggregation.Footnote ⁶ The temptation then is to graft as much mathematics as we can on to our moral discourse and to rewrite morality as ‘moral arithmetic’, to use an expression put to a related use by Buffon. The trolley problems do this for Parfit. As the examples unfold, and the mathematics takes over, the relation to ordinary moral thought becomes more and more strained.Footnote ⁷

Some sense of this can be gained from Parfit's discussion of the non-identity problem in vol. 2. This problem, inherited from Reasons and Persons, arises from the fact that some of our actions produce benefits and harms to particular future people, while others merely lead to a state of affairs in which people in general exist, while harming or benefiting no one in particular. Here is one of the cases that Parfit invokes (Vol. 2, page 223):

With relentless determination Parfit conducts the reader through case after case of this kind, arguing that Scanlon's view, that reasons are inherently personal, will not account for all the many instances in which we might be called upon to make a moral choice. But the importation of precision does not hide the fact that the examples considered are entirely unlike real moral dilemmas, and entirely shaped by the arithmetical obsession of their author. Real dilemmas come about in the way that Scanlon says they do, from what we owe to each other or – as Darwall prefers to put it – from the ways in which we hold ourselves and others to account. A spectral version of moral reasoning can survive in the world of the trolley problems; but it exists there detached from its roots in the person-to-person encounter, lending itself to mathematical treatment partly because the philosopher at his desk has thought the normal sources of moral sentiment away.

That is not to deny the fundamental truth in consequentialism, which is that moral reasoning makes comparisons. When Anna Karenina asks herself whether it is right to leave Karenin and to set up house with Vronsky, she is asking herself which of two courses of action would be better. But although she is making a comparative judgment, it is not one that can be resolved by a calculation. She is torn between her obligations to her husband and her child, and her love for Vronsky. Her dilemma is not detachable from its peculiar circumstances – her husband's vindictiveness and coldness of heart, her son's sweet devotion, Vronsky's Leichtsinn and Anna's knowledge of his faults. Dilemmas of this kind exist because we are bound to each other by obligations and attachments, and one way of being a bad person is to think they can be resolved by moral arithmetic. Suppose Anna were to reason that it is better to satisfy two healthy young people and frustrate one old one, than to satisfy one old person and frustrate two young ones, by a factor of 2.5 to 1, ergo I am leaving. What would we think, then, of her moral seriousness?

This is but one reason for thinking that the idea of an ‘optimific principle’ is both obscure in itself and unable to do the work that Parfit requires of it. Take away the trolleys and the lifeboats and we rarely know how to calculate ‘the best’, either in the particular case or when considering the application of principles. The consequences of our actions stretch infinitely outwards in both space and time. The best of intentions can lead to the worst of results. And values are many and in tension with each other. What place should we accord to beauty, grace, and dignity – or do these all creep into our deliberations as parts of human happiness? There is no knowing how Parfit would answer such a question, for his book is entirely devoid of moral psychology, and has nothing to say about what happiness consists in, by what scale it should be measured or about what human beings gain from their aesthetic and spiritual values.

More importantly, Parfit overlooks the actual record of consequentialist reasoning. Modern history presents case after case of inspired people led by visions of ‘the best’, believing that all rational beings would adopt those visions if only they would think about them clearly. The Communist Manifesto is one such vision. It gives a picture of ‘the best’, and argues that all would work for it, the bourgeoisie included, if only they understood the impeccable arguments for its implementation. Those who stand in the way of revolution are self-interested; but they are also irrational, and would change sides if they thought seriously about principles that everyone could will to be laws. Since their interests prevent them from thinking in that way, violent revolution is both necessary and inevitable.

Lenin and Mao, who put this document into practice, were adept at trolley problems. The moral arithmetic always came out in their favour, as they switched the trolley of history from one set of possible victims to another. And when the fat man had to be pushed from the bridge there was always someone ready to do the job for them, who could be quietly pushed from the bridge in his turn. The result was the total destruction of two great societies, and irreversible damage to the rest of us. Why suppose that we, applying our minds to the question what might be best in the long run, would make a better job of it? Moreover, is not this possibility – indeed probability – of error at the root of what is so objectionable in consequentialism, which turns wrongdoing into an intellectual mistake, thereby excusing it? When the Kaiser, looking back on the calamity of World War I, said ‘Ich hab’ es nicht gewollt' he spoke as a consequentialist, as did all those apologists who regretted the ‘mistakes’ of Lenin and Mao.

Which brings me back to the question of motives. The fundamental intuition behind contractualist arguments – Scanlon's, Gauthier's and Darwall's included – is that morality exists in part because it enables us to live on negotiated terms with others. We can do this because we act for reasons and respond to reasons too. When we incur the displeasure of those around us we attempt to justify our actions, and it is part of our accountability that we should reach for principles that others too can accept, and which are perforce impartial, universal and law-like. When the fault is ours we blame ourselves, and the good person blames himself more severely than others would. We recognize obligations to those special people who depend on us and on whom we depend, and we exist at the centre of a sphere of accountability, which stretches out from us with dwindling force across the world of other people. Our moral principles are the precipitate of personal relations, in which we are face to face with those who have a claim on us and who are more interested in our virtues and vices than in our ability to derive output from input on our pocket moral calculators. Hence what Strawson calls ‘reactive attitudes’ – including guilt, admiration and shame – form the core of our moral sentiments, bearing the indelible mark of the I-You relations in which they are ultimately rooted.Footnote ⁸

Some of this has been carefully spelled out by Darwall in The Second-Person Standpoint.Footnote ⁹ But to give a full account of what it involves we must go beyond the contractualist emphasis on advocacy and the resolution of conflict. Morality governs each personal encounter, and its force radiates from the other, when I address him I to You. In seeking the motive of our moral behaviour, therefore, it is not enough to rewrite it as the upshot of a contract. To understand the moral motive we must explore the deep metaphysical questions surrounding those two crucial words – ‘I’ and ‘You’. The radical scepticism about personal identity put forward in Reasons and Persons enabled Parfit to brush the metaphysical questions aside, and with them the whole realm of moral psychology. With the argument of On What Matters it is no longer possible to be so dismissive.

Kant was perhaps the first moral philosopher to understand that moral thinking presupposes a metaphysic of the self. We are distinguished from all other beings in the universe, he argued, by the fact that we refer to ourselves in the first person. (See the opening pages of Anthropology from a Pragmatic Standpoint.) The ‘I’ concept brings with it the privileges of first-person knowledge, including the knowledge of our freedom. And it is for this reason that we are accountable to each other and compelled by practical reason to take responsibility for what we are and what we do. You can doubt that Kant succeeded in shoring up his metaphysical vision, to the point where sceptical Humeans too would be compelled to accept it. But the vision was deepened by the suggestive arguments of Hegel's Phenomenology of Spirit and Philosophy of Right, which outline the ways in which self-knowledge and freedom depend upon the mutual recognition that binds rational choosers together in communities of their kind. The I-You encounter is implanted in the very centre of each of us, and with it comes the knowledge that we are judged.

Parfit is aware that Kant's moral philosophy is bound up with a metaphysical theory of freedom, and in a short and dismissive chapter he rejects Kant's theory of ‘noumenal causality’, supposing it to denote a relation that is outside time, and therefore outside change, and therefore not part of our world. He then goes on to argue that, since our acts are, pace Kant, merely ‘events in the spatio-temporal world’, and therefore subject to the same causal determination as other events in time, we cannot deserve to suffer on account of them. (Vol. 1, pages 263–272.) Instead of pausing to consider what this means for attitudes like praise, blame, remorse and shame, Parfit rushes on to discuss Kant's impossibility formula – which tells us that ‘it is wrong to act on any maxim that could not be a universal law’. (Vol. 1, page 275.) The whole question of responsibility, and the moral life that depends on it, is thereupon left lying in the gutter, never, so far as I could tell, to be picked up and set on its feet.

Here and there Parfit acknowledges that there are appropriate occasions of blame. He notices, in his discussion of the concept of wrong, that there are acts that ‘would give us reasons for remorse and give others reasons for indignation’ (Vol. 1, page 174). But he says nothing about what remorse and indignation consist in. Wrapped deep into indignation is the aim to inflict suffering – the suffering that comes when affection is withdrawn and the wrongdoer is pushed to the edge of the community. It is precisely when my indignation at your wickedness causes no suffering in you that I retreat to what Strawson calls ‘objective’, rather than ‘reactive’ attitudes. I then cease to view you as a responsible person and see you instead as a case for treatment or, worse, as a zombie or a thing. If, however, Parfit is right in thinking that no one deserves to suffer, then no one deserves to be the object of indignation, and indignation, which is founded on the belief that the suffering inflicted by it is deserved, becomes irrational.

Similar things can be said about remorse, guilt and shame. These states of mind are also forms of suffering – sometimes very intense suffering. People can die of shame, and take refuge from guilt in suicide. Even if we condemn such excesses, one thing is certain, which is that the one who suffers through remorse, guilt or shame believes that his suffering is deserved, and if no suffering is ever deserved these emotions too become irrational.

Parfit is led into this paradoxical position by his idea of wrong, which he treats as an intransitive property of actions, an action being wrong (roughly) if there is decisive moral reason not to do it. (Volume One, chapter 7.) But there is another sense of wrong in common usage, expressed by a transitive verb. People wrong each other, and when wronged the victim may demand retribution or apology. The common law of tort is based entirely on this idea, and since the common law is an enduring compendium of moral reflection it is worth attending to what it says. The wrongdoer (tortfeasor) has the legal obligation to compensate the victim, by putting him as nearly as possible in the position in which he would have been, had the tort not occurred. The wrongdoer can escape this obligation if he can show that he was not responsible for the apparent consequences of his act, and J.L. Austin has given us some nice reflections on what that means.Footnote ¹⁰ The legal process has its parallel in religious usage. This is because, in Judaism, Christianity and Islam, the idea of wrongness is derived from the transitive sense of the term. The wrongdoer wrongs God, and stands in need of God's forgiveness.

Turn attention now to the primary moral experiences of all of us. As children we do things that our parents have forbidden, or which cause pain and anxiety to others. We are taught that we must then apologize. If the wrong inflicted is severe, the culprit must be contrite, show a due recognition of the offence and the extent of his or her blameworthiness, trying meanwhile to compensate the victim for the harm that has been done. If we are lucky a process of atonement begins, and the wrongdoer, at first deprived of the affection that he craves, is by degrees forgiven, and reincorporated into the affections of his victim.

That process in the life of persons, which begins in childhood, is re-enacted again and again in the lives of decent and caring people, who strive not to hurt each other, and who work to be forgiven when they are tempted and fall. It is set before us in the Jewish ritual of Yom Kippur, and in the Roman Catholic theology of the confessional. It is known to all of us since it lies at the heart of the moral life, and reminds us that we are called to account for ourselves, and to bear the weight of our faults. But in Parfit's strange, bloodless philosophy there is hardly a glimpse of it.

If it is true that our moral thinking is rooted in relationships of accountability, then these must provide the raw material from which the motive of morality is constructed. We strive in all our doings to be reconciled with those who matter to us – to answer to the claims that are rightly made on us, and to give to our dependents what they need. We seek to make amends for the wrongs that we inflict, and to respect the dignity and needs of those whom we encounter. Growing out of this is a process of moral reasoning, and Kant is surely right that reasons must be valid for each and all of us, if they are to have a role in moral argument. For morality compels us to see ourselves as others see us, and to justify our actions in ways that ignore the fact that those actions are ours. Hence moral reasoning will often approximate to two of the paradigms spelled out by Parfit in his Triple Theory – the Kantian paradigm of universality, and the Scanlonian paradigm of the individual veto, which between them establish what I call a ‘calculus of rights and duties’.Footnote ¹¹ But to think that this reasoning will come gently to rest on the bed of the ‘best’, is to ignore the whole motivational context from which morality arises. We do not and cannot know what principles are conducive to the best; but we can, through dialogue and reflection, reach agreement and consensus with those who have a claim on us. Why is that not enough?

This brings me to a further point about reasons. In his sole excursus into aesthetics Parfit dismisses the idea that there can be reasons for aesthetic judgments. (Vol. 1, pages 53–4.) Like gustatory sensations, the experiences we derive from works of art and other objects of beauty have causes, but not reasons, for their occurrence. ‘If we ask what makes some musical passage so marvellous, the answer might be “Three modulations to distant keys”. This answer describes a cause of our response to this music, not a reason.’ (Vol. 1, page 53.) Parfit goes on to conclude on these grounds that music is ‘the lost battlefield and graveyard of most general aesthetic theories’ (Vol. 1, page 54).

It is true that the considerations adduced by a critic are not, straightforwardly, reasons for action, and maybe not reasons for belief. But they are addressed by one rational being to another, with a view to obtaining agreement over something that matters. If you think of the scherzo of Schubert's B-flat trio as light, trivial, and merely pretty I respond by drawing attention to things you need to hear in it – for instance, the sustained canonical writing which is so utterly unforced that you can easily fail to notice it. I describe the things you might have missed in the modulations and the overall structure. And having got your attention to coincide with mine I work on your response: the poignancy of the repeated notes – have you noticed it, felt it, reacted to it? The hint – as in so much of Schubert – that the gift of life is the greatest gift we have, and as easily withdrawn as offered. The melody itself makes you think of this, as the repeated notes enter areas of harmonic light and shade, as though the shadows of hurrying clouds run across the musical landscape. Sure, my use of metaphors puts this kind of reasoning in a different category from the practical or the theoretical. But that is because I am giving reasons for a response, reasons that are accepted in an act of rational attention in which emotion plays a constitutive part.

The example ought to remind us that there are many ways in which we engage in reason-governed dialogue with others, only some of which aim at decisions, and many of which aim simply at a consensus of appreciation and a shared network of values. We reason towards consensus, in a thousand different ways, and because life is so rich and varied, and because our values can be arithmetised only in the mad world of the trolley-problem, and maybe not even there, we know, or ought to know, that our comparative judgments will seldom have the kind of exactitude that consequentialists desire. But they tell us what matters all the same.