VIII FROM PRUDENCE TO MORALITY 1.Morality as something alien SOME of us think that the question, Why should I be moral?, hard though it is to answer, raises no serious doubts. However, others of us, probably a majority, disagree. We feel morality as something to some extent alien, as something external to the workings of our own nature, as something imposed from outside. We have our own aims, commitments, attachments, bents, enthusiasms, often far from selfish or callous, and when they come into conflict with what presents itself, sometimes after rather abstract deliberation, as a moral requirement, we wonder what authority morality could possibly have. There are various well-known, sometimes rather desperate, remedies for these feelings of alienation. They all involve reducing morality to something less external to human nature. One way is to reduce it to self-interest, the narrower the conception of self-interest the better. Another is to reduce it to some rich form of human flourishing. Another, more modestly, is to be content to find morality a place among many persons' own aims. Yet another, more ambitiously, is to make its requirements into requirements of rationality. Moral scepticism certainly has dimensions not present in, say, scepticism about the external world. The problem it presents is not just one of getting one's beliefs well-founded. Morality ends up with decisions that guide action. Any acting-guiding principles must meet the Requirement of Psychological Realism: the source of morality can never stray far from the natural sources of action.1 But what are the natural sources of human action? And more generally, what is the nature of the self? On this, we tend to divide, as much by inclination as by judgment, into two camps. One camp says: the deep sources of human action are sympathy, attachments, commitments---in short, sentiment or feeling. It is the answer 128 MORAL IMPORTANCE of classical British Empiricism, most attractively stated by Hume and accepted by Bentham and Mill. The other camp finds that list of motives acceptable, as far as it goes, but simply too short. They say: there is a further source of human action, namely understanding or reason. This is where the dispute must focus. Morality cannot take its stand far from the sources of human action. But what are they? And can one see morality emerging from one of them? And can its restrictions fit into a social structure that persons of that nature can themselves fit into? The Requirement of Psychological Real- ism, in this form, clearly has got to be met. It gives us a natural starting place for our move from the discussion of prudence to the discussion of morality. 2.(a)Morality and self-interest One way to make morality not at all alien is to reduce it to self-interest. Some persons are self-interested in a very cramped, crabbed, ungenerous way; nothing much matters to them but their own comfort and safety. If morality is to be reduced to self-interest in the sense of what about themselves persons actually care about, then it has to be reducible, in the case of a cramped, crabbed, ungenerous individual, to his own meagre range of concerns. But there are ampler, more enlightened forms of self-interest, and one might mean by the term one of these more idealized forms. For instance, one might see moral obligations as solutions to situations of partial conflict, the best bargain that any one person could strike, whether or not he actually knows it. Or one might mean by the term the flourishing of an individual life, on some rich conception of what human flourishing consists in. If we have the first, narrow sort of self-interest in mind, then the only hope of reducing morality to self-interest is with the help of sanctions. What authority has a judgment of moral obligation? One can answer: they are commands, backed up by sanctions. If the sanctions are severe and inescapable, then morality may well coincide with self-interest. But even most religious believers find it hard to accept that, since the sanctions in this world are not inescapable, life after death is tacked on as a sort of prison sentence. Anyway, religious believers do not often see self- 129 FROM PRUDENCE TO MORALITY interest as the source of morality. Life after death is, if nothing more, a human invention much of the purpose of which was to reduce morality to self-interest, but it is hard to see that it would also be God's purpose in his creation.2 Bentham and Mill rightly stress the importance of social sanctions, but sanctions are important as the causes of the conformity that society needs. Since they are not always inescapable or sometimes even feasible, if the range of moral obligations is not to shrink much beyond our expectations, then social sanctions cannot be the source of obligations. Life cannot present the range of inescapable sanctions that the reduction needs; life after death could but does not.3 The reduction becomes more plausible when the accounts of self-interest become more idealized---not what we think is in our own interest but what actually is. If moral rules represent the best bargain each of us can hope to strike with the rest, and the bargain is better than no bargain at all, then they are reducible to an enlightened self-interest. This notion of moral rules as emerging from a bargain takes us into the territory of modern contractualism. One of the big attractions of contrac- tualism is that it seems to manage many of the benefits of reducing morality to self-interest without its defects. But for now I want to look just at attempts to reduce morality to pure self-interest, and not to self-interest mitigated, as it is in contractualism, by some form of impartiality. It is true, and important, that many norms that are correctly regarded as moral rules are the best bargain that all parties can strike. Indeed, certain norms are so well grounded in that way that they never even need to be the subject of conscious or explicit bargaining. They emerged, and often quite complex social institutions embodying them also emerged, entirely naturally with the growth of social life.4 I shall come back to this important fact later on.5 However, the fact is important to understanding the limits to the forms that society and its norms can take and also the very considerable limits to the efficacy of moral judgment;6 it is not important to understand- ing the authority of moral judgments. Again, unless we are willing to allow the range of moral judgment to shrink far behind our present expectations, morality cannot be reduced to this enlightened form of self-interest. There is more to 130 MORAL IMPORTANCE morality than resolution of partial conflict. There are, for instance, moral obligations to those too weak to be in a position to bargain effectively, such as the present losers in some unfair but securely entrenched distribution of resources or the members of some future generation who will suffer from our high living. The richer the notion of self-interest the more plausible the reduction of morality to it and, up to a point, the more plausible it will be as an account of self-interest. The most plausible of all is the reduction to a rich conception of human flourishing. There is, of course, the danger that one will make the conception of human flourishing sufficiently rich by packing it with ethical assumptions, so making it in effect evaluative. For instance, one might contend that no one can be "truly' or "deeply' happy unless he is to some fairly high degree moral, where the effect of the qualifiers "truly' or "deeply' is to make "happiness' a thoroughly evaluative notion.7 But that is not the real difficulty. We do not need a reduction of morality to something entirely non- evaluative. The question is, Can we make morality less alien?; it is not, Can we make it less evaluative? We shall have solved the alienation problem if we can reduce morality to some normative notion that people see as less alien, that they more easily identify with, than morality seemed at the start. All that we have to do, for example, is convincingly argue that one cannot be deeply happy (so long as the person we seek to pursuade will accept that as a description of something that comes within the compass of his happiness) without being to a large degree moral. If certain moral categories, recognized as moral, could be shown to be also prudential, without doing violence to our original notion of the prudential, then we should have gone a long way to solving the alienation problem. Still, there is another, altogether more familiar, obstacle to reducing morality to human flourishing. A defensible concep- tion of flourishing is bound to be too indeterminate to yield interesting moral results.8 Aristotle's conception, on some interpretations, approached the degree of determinateness needed to yield moral conclusions. But, as I argued in Part One,9 the most that we can hope for from an account of flourishing is 131 FROM PRUDENCE TO MORALITY not a picture of the ideal form of life for all humans to adopt but a relatively short list of prudential values---the values that contribute to making a normal human life go well. Though these important values are valuable for everyone, and though the list of them does even constitute a kind of perfectionist picture of human existence, the conception of human flourish- ing they can be made to yield is far too indeterminate to serve in a reductionist programme. Even if one were willing to accept the list, even if one had no objections to its being so thoroughly evaluative itself, one would not have the materials to which morality could be reduced. The list, I think, includes all the following: enjoyment, deep personal relations, auto- nomy, accomplishment, understanding, and living a life of point and substance.10 Perhaps the list should be a little longer. But these values can be realized in such different ways and combined in a life in such different proportions that there is no one form of life that emerges, let alone a form of life fully enough specified to provide materials for answering all the moral questions that life will present. They do not tell one, in any very definite way, what one ought to aim at in one's own individual life, and certainly do not tell one how one's own aims in life ought to be reconciled with the competing aims of others. They do not, for instance, provide even the rudiments of an account of justice. That obstacle to the reduction is perfectly familiar. Another, and to my mind the greatest, obstacle to the reduction is less familiar. It is what might be called the penetration of the prudential by the moral. What I have in mind is this. One has not got a specification of the prudential at all without a pretty full account of what moral demands there are on us and how they are to be accommodated. Prudential value does not stop at the edges of an individual's own private life. Some persons may see their self-interest in a narrow, crabbed way. But anyone with a defensible idea of prudential values can see what he cares about not just as what as a matter of fact he now cares about, but as what he ought to care about or what he will care about after subjecting his concerns to full deliberation. He will find it hard, therefore, to keep moral and prudential values apart. One of the things he will want is a life of point and substance. What he will see as 132 MORAL IMPORTANCE prudentially valuable, valuable to his own personal life, will to some extent coincide with what he will see as valuable morally. Our understanding of "a good life' cannot be parcelled into "good prudentially' and "good morally'. The very phrase "a good life' may seem ambiguous (good pruden- tially? good morally?), but at any deep level there are not two senses to be distinguished. Part of having a life of point and substance is having a life in which moral reasons take their place, along with other practical reasons, in motivation. This is a complicated matter, and I want to come back to look at it in greater detail shortly.11 But for the moment let us accept that it is a marked and central part of some persons' experience that they cannot live a satisfactory life, seen simply in prudential terms, without some accommodation of morality ---as marked and central a part of their experience as is their self-love. That poses the following problem. To reduce morality to prudence is to reduce it to something that, without morality, has a gaping hole in it. One might think that this is not such a serious difficulty. Why not just say that if morality has been shown to be part of prudence then it has been shown to be reducible to prudence? But that will not do for a reduction, at least in the strong sense that we have so far had in mind. It does not find a place for determinate moral requirements in separate determinate prudential concerns. Instead, what we find is that both moral requirements and prudential concerns are indeterminate. Showing that morality is part of prudence might look like reducing morality to prudence only because the part of prudence to which morality will be reduced is already morality itself. But this is hardly, in any normal sense, a "reduction'. There is a parallel argument about motivation. One cannot treat prudential motives as basic and then establish a motive for morality by showing every moral act to be a necessary condition of reaching some prudential goal. A theory of motivation cannot itself be complete without an account of the force and place of moral reasons. An account of prudential motivation has to come at the same time as an account of moral motivation (and no doubt as accounts of other kinds of pracical reasons as well). There is, in that sense, no starting 133 FROM PRUDENCE TO MORALITY point in a theory of motivation. We have to deny the dualism of prudence and morality: they are not separable either as reasons for action or as motives. It is not that they are not different but that they overlap importantly. I have just mentioned that, until now, we have been considering whether morality can be reduced to human flourishing in a pretty demanding sense of "reduction': namely, that we can generate the content of morality out of the resources of the notion of flourishing. There is no reason why we cannot take a less demanding sense. What we want is to find morality a place inside the domain of prudence. We should still do that if the content of morality could in some way be generated independently of prudence, so that morality was already a set of satisfactorily determinate demands, and we merely showed that flourishing involved acting like that. Indeed, since part of flourishing, when it is properly under- stood, may turn out to be acting morally simply because it is morally, it may also turn out that, to use Robert Nozick's terms, ethical push (the push to our own action arising from our living a prudentially successful life) and ethical pull (the pull exerted on our action arising from the moral demands of others) eventually meet.12 I do not think that they do, but I want to postpone that discussion until we come (very shortly, in Section 5) to how the push and pull of ethics do stand to one another. Let us stop here with the relation between morality and self-interest. I realize that at points the case that I have set out has not been argued enough, but I shall come back to these matters too in Section 5. 3.(b) Morality and personal aims To show that morality is not really alien, it is enough to find a place for it somewhere inside the self. And many persons have goals that are not self-interested, even goals that can plausibly be considered moral. They care about other persons, some- times even more than they care about themselves. Sometimes they care about justice or about living a moral life. Admittedly, one cannot rely on some form of moral concern's being present in everyone, whereas everyone, or nearly everyone, can be 134 MORAL IMPORTANCE relied upon to care about himself. But it is enough, for certain purposes, to find a place for morality inside some persons. If enough persons have enough moral concerns among their own personal aims, then morality will be an important social force. It is undeniable that some persons aim at living moral lives. Now, there are many philosophers who argue that there are no reasons for action except for ones that link with some personal aim.13 We all accept the Requirement of Psycholo- gical Realism. So if these philosophers were right, moral reasons would have to have the same sorts of links with personal aims. Bernard Williams, an advocate of this view, draws a useful distinction between "internal' and "external' reasons for action.14 Internal reasons are ones that stand in a certain sort of relation, shortly to be explained, to some element of an agent's "subjective motivational set'; an external reason is one that has no such relation.15 Williams concludes that, once we see clearly what is involved in the existence of an external reason, we shall have to accept that there are none. The only reasons for action are internal. So if morality is to have any hope of meeting the Requirement of Psychological Realism, moral reasons will have to be reduced to internal reasons. But there is this difficulty. Clearly not just any aim or desire or motive, just by happening to exist, will serve as backing for an internal reason. The difficulty here is very like the difficulty with desire accounts of well-being. A person's well-being cannot be explained in terms of just any desire that he happens to have. Persons sometimes actually want what will not enhance, and sometimes even harm, their well-being. The only remotely plausible desire-account of well-being has to be in terms not of a person's actual desires but of his informed desires---desires in some sense idealized.16 Similarly, if a desire is based on false belief, then it provides no reason for action. As Williams, for instance, well knows,17 internal reasons must also depend upon desires that are in some way idealized. But how ideal? Some candidate criticisms of desires appeal, in effect, to what look very much like external reasons. So the answer requires a full, detailed picture of how deliberation about values ought to go---a picture that those who advocate that all reasons are internal have never really given us. We have just noted the obvious point that if a desire rests 135 FROM PRUDENCE TO MORALITY on false belief it will not support an internal reason. What is more, a person might think that he has an internal reason when he does not (e.g. he thinks he wants such-and-such but does not really), or he might not know that he has an internal reason when he does (e.g. he is unaware that he does really want such-and-such). Also, a desire or an internal reason that one did not know one had can be uncovered in the course of deliberation. For instance, one wants x and comes to see, upon reflection, that y-ing is the most efficient way of getting it. Or one comes to see, upon reflection, that x and y, which seemed to clash with one another, can both be achieved. Or one wants x and comes to see, upon reflection, further or better ways of realizing it. These are all forms of criticism that Williams acknowledges.18 There is nothing wrong with that list, I think, except that it stops too soon; it constitutes a rather meagre description of practical deliberation. If all reasons are internal reasons, then our "subjective motivational set', though it does not have to be static, will change only in response to new facts or to some better entrenched member of the set. But we ask far more searching questions about our aims and resort to more radical forms of deliberation to answer them. Such searching questions arise, it is true, fairly rarely, but they are the especially fateful ones. We can change not only our desires by appeal to some deeply entrenched member of the subjective motivational set, we can also change the deeply entrenched members of the set. I went into such a case when I earlier discussed the relation of value and desire and, again later, when I discussed perfectionism.19 Suppose that I have been quite content to fritter my life away with pleasant pastimes, until I meet someone who has accomplished something important with his life, and then, inspired by his example, I form a new desire to live a life like that. But upon further reflection, I see that not every achievement will serve to give life point and substance. A compulsive achiever may quickly mount the rungs of some conventional ladder of success but not get nearer to accomplishment in the sense I am trying to understand. And an achiever might get great joy from success in the contest, but this sort of joy is not to be confused with the sense of accomplishment that I am after. Nor is the mere exercise of skills or the winning of respect and admiration a 136 MORAL IMPORTANCE sign of accomplishment. So one stage of the deliberation is to separate off what is to count as "accomplishment' in the relevant sense. Another stage is to see whether, when all is said and done, what one has succeeded in isolating looks worthy of being an end in life.20 And one decides this not by appeal to some still more deeply entrenched member of one's subjective motivational set. There is none to appeal to, except for the vacuous desire to have a valuable life, which would not in any case help because what we have to decide is whether accomplishment, defined as we now have managed to define it, is indeed valuable. What is important about this more radical form of deliberation is that subjective, varying, personal desires play very little part, while understanding plays a large part. What is central to the deliberation is seeing fully and independently what the sort of accomplishment in question is, and this is a sort of perception that introduces a new member into one's motivational set with much less aid from the old members than the picture of an internal reason permits. This is not to say that things are desired because valuable, not valuable because desired.21 For one thing, there are very many different sorts of case, my example of accomplishment being only one. In many other cases---simple tastes, for instance---the element of understanding is small and of desire large; rotting bananas are valuable to me because, odd though it may seem to most of you, that is the way I like them. Another thing is that, even in the cases where the element of understanding is large, as it is with accomplishment as a prudential value, desire also plays a role in its status as a value. There is no priority: desire does not precede value; value does not precede desire. In the case of accomplishment we desire it because we come to see what it is. But when we explain what it is we do not describe things neutrally. The language we use in reporting our perceptions selects what we see as important and shows how we view things in a favourable light. But this is where, in the middle of this explanation of what we perceive, desire re-enters at a deeper point. We have to explain what it is to see things favourably. We have to explain what makes certain features of objects desirability features. Some understanding---the sort that in- volves fixing on certain features and seeing them in a 137 FROM PRUDENCE TO MORALITY favourable light---is also a kind of movement. It requires a will to go for what has those features. There is no adequate explanation of their being desirability features without an appeal to this kind of movement. So we cannot, even in a case such as accomplishment, separate understanding and desire. What this means is not so much that the advocates of internal reasons are wrong as that the distinction between internal and external reasons is unreal.22 A prudential value such as accomplishment is valuable for everyone; anyone who fails to recognize it as valuable lacks understanding. It is true that there will be some special sorts of persons for whom accomplishment conflicts with other important values (it raises great anxieties, say) so that they had better pass it up; but these are cases of conflict of values in which accomplishment is still a value. It takes a quite complex story to make clear both the extent to which values are not personal and the way individual differences affect values, a story I told more fully earlier.23 One outcome is that once we get the roles of desire and understanding straight, the distinction between "subjec- tive' and "objective' also looks unreal. Are prudential values subjective or objective? I use "subjective' here to mean "dependent upon an agent's own desires'. Now, prudential values do not rest upon one person's desires. They would not even be intelligible if their status as values came simply from someone's wanting them; they have also to be something that persons generally can see as worth going for. But then it is also true that varying, personal desires matter to prudential values in several ways. Individuals differ in how, or even whether, they can realize some particular value. For instance, desires enter importantly in tastes. The distinction between "subjective' and "objective' has figured importantly in the history of moral philosophy in part because the distinction between the competing views valuable because desired and desired because valuable has been thought to mark a deep division in our options. But the dependence of prudential value on desire and understanding is much less simple, much less a matter of all or nothing, than these dualisms suggest. The distinc- tion between objective and subjective, defined in this common way, does not mark an especially crucial dis- tinction. 138 MORAL IMPORTANCE What this means is that the distinction between internal and external reasons is another untenable dualism. The doubts that some writers feel about external reasons is that it is hard to see how they could ever create a new motive.24 An agent who acknowledges an external reason to x is supposed then to form, on the strength of understanding this reason, a motive to x, whatever motivations he originally had. What is thought problematic about this is that this new motive cannot stand to his old motives in the ways that the non-radical criticisms of desires allow, because they all depend, in the end, upon some member of the agent's personal subjective motiva- tional set, thus making the reason internal, not external. Can one see any further way for a new motive to arise? Could it arise, in particular, simply from a change in understanding? But this scepticism about external reasons depends upon our keeping desire and understanding at a considerable distance from one another.25 Any normal person who is frittering his life away has a reason to try to accomplish something with it. The sort of more radical deliberation I sketched a moment ago is the sort that he would have to go through to recognize this as a reason. It is not that it appeals to some pre-existent member of his subjective motivational set. But it is not free from desire either: the understanding that is needed is a grasp of certain desirability features, and they owe their status as such to normal human aims. It is on the nature of just this sort of deliberation that the important issues about external reasons turn. This deliberation is a matter of coming to see some new kind of thing as valuable, and the question is how it relies upon the present membership of one's motivational set. Certainly a lot is likely to be necessary for us to be able to come to see this new kind of thing as valuable. It is not just a matter of understanding features of the newly valued object that are expressible in value-neutral ways. One has to work one's way into the new vocabulary that gives expression to our fastening upon some features and not others as worth directing our attention to. This vocabulary expresses, so cannot be separated from, a sense of what is important for action. To work one's way into the vocabulary, to come to see things in fresh, subtle, complex ways, is unlikely to be easy. The point is not that one cannot come to see things in this way without also 139 FROM PRUDENCE TO MORALITY falling motivationally under the sway of the perceptions; on the contrary, it is possible, in the detached manner of a good anthropologist, to get inside highly subtle points of view without endorsing them. The point is, rather, that it will often not be at all easy. It may sometimes take guidance, training, and practice. But there is one sort of difficulty that it does not present. We are not faced with two outlooks---our old one and our new one---at a loss for a way of mediating between them. It may look as if this choice, being a choice between two evaluative outlooks, needs an equivalent outlook of the same sort but more basic than each of them. But that is not how such deliberation need go. We have to be able to decide that, say, a life with enjoyment and accomplishment is, other things being equal, better than a life without them. I can make that judgment from within the way of seeing things that sees enjoyment and accomplishment as valuable. What I need, in order to make that judgment without arbitrariness, is to be able to decide, also without arbitrariness though within that perspective, that enjoyment and accomplishment are indeed valuable. This kind of reasoning need neither appeal to a member of the present subjective motivational set nor proceed without touching motives. Understanding and motivation cannot here be very sharply separated. Practical rationality is not bloodless; it itself cannot even be understood independently of motivation. Motivation is not blind; it is shaped, though to different degrees in different cases, by our perception of the nature of its objects. One of the things we have been trying to do in this discussion is to place morality sufficiently inside the self to meet the Requirement of Psychological Realism. But making morality internal is not the only job. We also have to show why morality should have any authority over our actions at all. Making it internal does not, in itself, make it authoritative. We have an alienation problem, but we also have an authority problem. We quite rightly ask, What authority over my actions do highly theoretical moral considerations have? But we should also ask, What authority over my actions do my present desires have? Why should I be moved by any present aim of mine, no matter how it came about, no matter how well founded? The authority of my present aims is hardly 140 MORAL IMPORTANCE more obvious than the authority of my theoretical beliefs. Desires have histories. Some desires, no doubt, have the sort of motivational authority that we are after, but some do not. We need a way to determine which are which.26 So it is a mistake to think that the foundation of agency is the subjective motivational set. We ask, Should I want that? Agency has a role in determining the motivational set. It does not just start with motives; some motives have their origins in the search for the answer to the radical interpretation of the question, Should I want that? There is no starting point, no single foundation upon which all the rest of the building blocks sit, in either practical reason or in theories of motiva- tion. The holism characteristic of theories in the natural sciences applies here too. The sorts of desires that have authority in motivation are shaped by understanding. Under- standing gets its direction from the nature of human desires. Reducing morality to present personal aims is puzzling in a way that reducing it to self-interest is not. Why should I accept some demanding moral requirement when it has a clear cost? It is implausible to answer, Because it is your present desire. It is implausible to answer even, Because it is your desire. What authority does any old present desire have? The concept of a reason is fairly strongly normative. One only gets at a practical reason if one can answer, Because it is worth desiring. Self-interest is a consideration that not only appears in desires but also deserves its place there. Its worth may be exaggerated and may indeed, upon close scrutiny, be much more limited than we tend to think,27 but self-interest is not only internal to the self but also authoritative. On the other hand, the advantage of a reduction to personal aims over a reduction to self-interest is that many of us do care about things other than ourselves, and some of these other things are, unless we are all grossly deceived, well worth caring about. So the question that is important to each of us is, What do I most care about? Not "do I happen to', as if I had to find out some fact about myself, but "should I', given the sorts of radical and non-radical criticism of desire open to us. If I ask myself, Why should I be moral?, I cannot just answer, Because I now want to. The answer begs a very important question. Do desires, regardless of their content or their cause constitute a practical 141 FROM PRUDENCE TO MORALITY reason? Since they do not, I have to find a fuller answer. I have to satisfy myself that morality constitutes a good reason for action. What we can accept as worth desiring gives us a reason and also becomes, since it is part of a desire's being shaped, a motive. So we cannot first supply a theory of motivation and then show how it limits the nature of moral reasons. A theory of motivation cannot itself be complete without an account of moral reasons. If morality constitutes an object worth desiring, then it is part of what life aims at and one of the springs of action. A theory of motivation cannot, any more than a theory of self-interest could, limit morality, because it cannot itself be determined independently of determining moral reasons. It must all come at the same time. The Requirement of Psychological Realism imposes no independent restriction on moral reasons. 4.(|c) Morality and rationality Another important strategy in dealing with the alienation problem is to make morality a requirement of reason. Our reason is part of our nature, so morality can be seen not as imposed upon us from outside but as self-imposed. But, again, solving the alienation problem is only half the battle. There is also the authority problem. Would morality also then be something to care about? If my acting immorally is acting contrary to my rational nature---say, in the worst case, contradicting myself---why should I care? Self-contradiction seems an awfully small price to pay for what, in the moral stakes, can be very great gains. When writers mention rationality, what they have in mind ranges from minimal conceptions that no matter how the notion of rationality might be filled out will have to feature in it (e.g. consistency) to quite rich accounts that embody stringent and contentious normative standards. There cannot be any even fairly rich account of practical reason without decision about how moral reasons fit into the set of good reasons. The only attempts at showing substantive moral principles to be a requirement of reason that have any hope of success are those that use a fairly rich account of rationality. The minimal account, an account that does not incorporate 142 MORAL IMPORTANCE among other things an account of moral reasons, delivers no interesting moral conclusions. For example, at first glance, Kant seems to use only the most minimal notion (indeed only the notion of contradiction). The Categorical Imperative test involves first universalizing one's maxim and then seeing whether, when this new law of human operation is added to the description of how the rest of the world operates, contradiction results. But in cases of even blatantly immoral maxims, it is very hard to uncover anything remotely ap- proaching contradiction in the strict sense. But then Kant explains that he has in mind either of two sorts of contradiction, contradiction in formulation or contradiction in the will. And his explanation of, in particular, contradictions in the will draws upon a very rich account of rationality. There is nothing wrong in this. But both his rich account of rationality and his use of a much-enriched notion of contradiction as a test of right and wrong carry a very heavy freight of teleological views about human nature and, in the end, also moral views. A rich theory of rationality and a substantive moral theory have to be developed together. This means, I think, that it is best to postpone Kantian derivations of morality from rationality until later, when we have substantive moral reasons in front of us. There is no sharp line between minimal and rich conceptions of rationality. One feature of reasons is a kind of impersonality. In the case of moral reasons, impersonality introduces some form of impartia- lity, a commitment to seeing everyone as due some sort of equal standing---what, for short, one might call "equal respect' ---and equal respect, when developed, brings us close to standards of fairness. The movement of thought---rationality, impersonality, impartiality, equal respect, fairness---passes without obvious break from minimal to rich conceptions of rationality. It is not so much that the later notions, once they are given content, add yet more features to the earlier ones as that the later ones make determinate earlier notions that are otherwise largely featureless. It is some such movement of thought that is the best prospect for grounding morality in rationality. It is the really powerful argument of Kant, and it is the argument of some modern contractualists.28 But it is not possible to assess it without assessing the related rich 143 FROM PRUDENCE TO MORALITY conceptions of equal repect and fairness, which I shall come to in the next chapter. It might be possible to derive from a minimal conception of rationality, without intrusions into moral territory, a notion of impersonality determinate enough to enable us at least to answer the question, Why should I be moral? Perhaps, for instance, we can derive a requirement that reasons be impartial, where impartiality has enough content to show that moral reasons trump all other kinds of reasons, while not having so much content that we find ourselves committed to one or other contentious moral stand- ard of fairness. I doubt that even this limited derivation succeeds, but it would be well short of the derivation of determinate moral requirements from rationality, which is our primary interest now. I think that the considerations I come to later go a long way towards suggesting that this more ambitious derivation does not succeed either, and that holism applies here too: the only way to decide whether to accept a certain rich account of rationality, with its considerable moral content, is also to have a way of deciding which moral reasons are good ones, with neither decision taking precedence. It is true that moral language itself has implications for what a moral judgment can be like. To use certain key moral terms is to accept certain commitments, the best known being that a moral judgment must be universalizable, in the relatively weak sense of that term (the interpretations of which also range from the minimal to the rich) that we be able to purge the judgment of all references to particular persons, times, or places. But not only is this not a particularly demanding requirement, it is also an escapable one. It is a requirement at all only so long as one is willing to use those key moral terms as they are used now. The same is true if, to universalizability, one adds further features that one believes to be implied by moral language---for instance, as Professor Hare would propose, prescriptivity.29 Prescriptivity is escap- able in the same way. And prescriptivity is not, in any case, very powerful in its effects. Prescriptivity and universalizability can jointly turn into a test of moral right and wrong of the sort that the Categorical Imperative is only if they are filled out in the context of a fairly rich account of rationality, rich enough to embody judgments about what moral reasons are good 144 MORAL IMPORTANCE reasons. It is true that formal features, such as universaliz- ability and prescriptivity, impose some limits on what can count as a moral judgment. We cannot change the formal features of moral terms too drastically before we find that we have ended up with vocabulary with which we can no longer make what we now understand by a "moral' judgment. But that is not much of a limitation. The questions that we ask now may be slightly but importantly the wrong ones. And the interesting moral features, such as universalizability, though present now, are the sort that could easily be much modified with further understanding. I doubt that they will be, but my belief comes out of a set of views that includes a substantive moral theory. In developing a moral theory all concepts are up for revision. Holism encompasses the network of key theoretical terms too. No firm conclusions about the content of morality can be got out of moral language because the content of morality is needed to help fix moral language.30 I shall come to attempts to make morality a rational requirement later. Still, there is another important kind of dependence of morality upon rationality, which I do want to pursue now. We have a conception of what makes human life in general good, which each of us adapts to his own case. And maximization is the only plausible policy in the prudential sphere: each of us wants, or at any rate should want, to make his own life as good as possible. That is where Part One left us. But there are many writers, especially in the utilitarian tradition, who think that rationality will take us further. The rational policy in the sphere of prudence, namely maximiza- tion, is, they believe, the rational policy in all practical decision. This view crops up again and again in modern moral and political thinking. What generates it, what causes and seemingly justifies it, is well worth trying to pin down. It seems to me to rest on a deep confusion---a mistake about how prudence could possibly be related to morality. In the domain of prudence, maximizing is irresistible. Although we should reject a policy of maximizing one particular value (say, pleasure) at the expense of others, it is hard to resist a policy simply of making one's life as good as one can. One does not aggregate, though, by totting up many small values from various quarters of one's life; instead, one's 145 FROM PRUDENCE TO MORALITY important global desires already incorporate the major magni- tudes that determine one's choice. If one prefers living autonomously to the various comforts of a non-autonomous life, that very decision is a decision that an autonomous life is more valuable. Even when one does tot up, say, many small-scale pleasures to get an overall aggregate value, the value of the life containing these many local pleasures is fixed in comparison with competing forms of life, and so the finally effective magnitudes are fixed by global desires. I have argued the point earlier: desires form a hierarchy, and the whole idea of a hierarchy of desires brings the prudential policy of maximization along with it.31 This justification of maximization in the case of prudence really comes down to the observation that maximization is already built into a person's hierarchy of values. But is it built into it because of the nature of the values or because it is one person's life?32 If it is because of the nature of the values, then behind the justification of the maximizing that we have been talking about there is another that would extend from the one- person into the many-person case. But if it is because it is one person's life, then if maximizing has a justification in the many-person case it must be a fresh one. It seems to me that the especially clear justification for maximization that we have found in the one-person case is dependent upon its being a one person case. It is true that some prudential values, just as the values they are, ought to be maximized. They are such that, in general, the more the better. Pleasure ought, other things being equal, to be maximized; pain ought to be minimized. And this maximizing/ minimizing policy would seem to have to do with the nature of the particular value at stake in this case, and not with its being part of one or many lives. But other values seem different in this respect. Having deep, loving attachments is, for many persons, one of the most valuable things in their lives. But do we think the more the better? Well, we do not seem always to aim at still deeper and more loving attachments. The words "deep' and "loving' describe a kind of relationship, and not values of which more is better. Nor do we want as many deep, loving attachments as possible. It is very implausible that this is the kind of value where nine or ten such relations are, if they 146 MORAL IMPORTANCE were possible, better than three or four. At least there is doubt whether the point of maximizing comes from the nature of these values. And there is little doubt that each of us aims at making his own life as valuable as possible. So there seems to be one strong justification for maximization that arises not from the nature of the values but from their being found in a context of one life.33 So we have not yet found a way of extending our maximizing policy from the one-person to the many-person case. One might think that decision theory has already shown the way. The name "decision theory' is used so widely that it includes any theory, normative or descriptive, concerning the choice, in uncertainty, of the optimal or rational action.34 On this capacious definition, morality and prudence are just two departments of decision theory---along with, for instance, empirical theories of action. What especially concerns us now are the normative departments of decision theory. How might normative decision theory manage to extend the policy of maximization from the one-person case to the many-person? Classical decision theory, in the Bayes or Ramsey tradition, employs the framework of utility maximization: rational choice is choice that maximizes expected utility. We can then see reasoning as fitting means to ends.35 We can assume the existence of a dominant end, not necessarily hedonistic. Such an end can then be seen as what makes rationality possible. Conflict between goals is to be reconciled by calculating how effective they are as means to the dominant end, or how likely they are to lead to it. Conflicting aims can then always be reconciled rationally---that is, by the computations just mentioned---and need not be left to hunch or intuition or fiat, and the operation of reason can be reduced to such computing. These forms of decision theory, therefore, extend maximization from the one-person to the many-person case in the name of rationality. But this way of extending it does not work either. In contrast to this means-end conception of rationality, what we ordinarily call rationality is much less restricted and sharp- edged. It can also include deliberation about ends; it can include the weighing and reconciling of conflicting aims that are regarded as irreducibly plural. What we ordinarily call 147 FROM PRUDENCE TO MORALITY "rational' merges with what we see as "sensible' or "justified'. Means-end rationality is not a superior, or the primary, sort of rationality. In fact, this particular sort of means-end rationa- lity is not even possible. It needs a single substantive super- value, a dominant end in that sense, and there is none.36 "Well-being' or "utility' is best understood as a formal analysis of the concept of prudential value. And prudential values are irreducibly plural. So the use of maximization in this common form of decision theory and in a plausible prudential value theory is not the same. The defects in the first theory do not affect the second theory, because the defects arise from elements that do not reappear in the second. There is an important moral here. So many different theories have some claim to be called "maximiz- ing' or "utilitarian' that they tend to get run together and faults in one seen as faults in another. It is easy to understand why many people are alarmed by this particular version of decision theory. If rationality itself demands a dominant end in prudential action theory, then it will demand it in moral theory too. Once that is allowed, the battle is over. All the competitors of utilitarianism would be proved wrong simply by appeal to rationality. No doubt some people are drawn to utilitarianism because it seems to them uniquely rational, but this appearance of unique rationality comes either from the dubious assumption of a dominant end or from their holding a view about rationality that does not even apply in prudential theory. On the other side, those critics who rightly object that utilitarianism has no claims to unique rationality37 are not really attacking utilitarianism but only one form of decision theory that gets incorporated in some forms of utilitarianism. The hope of an irresistible, uncontroversial extension of maximization from the one- to the many-person case does not die easily. We do, after all, generalize our notion of individual good into the notion of the common good. What is that if not a maximizing notion? In any case, there are standards for the good of many persons that are as obvious and inevitable as the standard for one person. There is this one, for instance: if some in the group become better off and none become worse off, the state of the group is better. It is hard to see how one could resist such a principle. Mill's defence of his proof of utility falls back on a 148 MORAL IMPORTANCE similar principle, which is also hard to resist. He moves from the premises "each person's happiness is a good to that person' to the conclusion that "only the general happiness is a good to the aggregate of all persons'.38 That looks like the plainest non sequitur: how can Mill's sort of ethical hedonism establish anything so specific and so strong as the moral requirement that each person act, at whatever sacrifice to himself, to maximize happiness generally? But Mill explained later in a letter that he did not have anything like so strong a conlusion in mind:39 "As to the sentence you quote from my Utilitarianism, when I said the general happiness is a good to the aggregate of all persons I did not mean that every human being's happiness is a good to every other human being, though I think in a good state of society and education it would be so. I merely meant in this particular sentence to argue that since A's happiness is a good, B's a good, C's a good, etc., the sum of all these goods must be a good.' Perhaps by summing several persons' goods Mill meant no more than keeping them all in our mind together. But perhaps he also meant, a trifle more strongly, that any increase in the good of one of the members also makes the general good greater. However, the type of addition in all these principles is different from the addition in the case of one person. With one person, the aggregation is to a large extent already incorpor- ated in the individual's global desires, and there is nothing comparable to that framework in the many-person case. Also, in the one-person case we have addition in the sense of amalgamation of positive and negative values into an all-in sum; here we have addition only in the sense of the accession of further positive values. That is what makes these principles so hard to resist. But they pay a price in their scope. In most changes, some persons gain and some lose, and these principles can say nothing, therefore, about all of these cases. Mill talks about summing persons' goods as if it were bringing them together in a new, clear, and entirely uncontentious sense. But actually bringing persons together in a group puts them in relationships with one another in which one person's potential gain can be another person's potential loss. Certainly some changes involving both gains and losses will produce better 149 FROM PRUDENCE TO MORALITY states than others. So these principles, though irresistible, are inadequate. What they identify as better is better, but what is better will not always be identified by them, and they cannot identify the best. Also, they are highly conservative: if the status quo contains distributions that are, by any standard, unjust, these principles will not tell us to improve them. There are various ways of trying to give more scope to these principles. With some groups, nations for example, there are goods that benefit everyone, what economists call "public goods'---national security and roads are examples. They meet "common interests' of a strict, uncontentious sort. But these interests are also of the gain-only sort. So not many interests are "common interests' in this strict sense, and even those that are often benefit persons unevenly and so produce results that themselves need ranking. The scope, therefore, is still not great. Another way of trying to extend the scope is the test of Pareto optimality with compensation. Indeed, the compensa- tion test looks, at first sight, like a brilliant device for extending the justification of gain-only situations into gain- and-loss situations. The Pareto test is simply this: the common good is to be considered greater if at least one person is better off and no one is worse off. Changes, of course, seldom leave no one worse off, which is why it is usual to introduce compensation into the test, so that it goes: the common good is greater if, after the gainers fully compensate the losers, at least one gainer is still better off. But the compensation test seems as irresistible as the gain-only prin- ciples, merely because it is not in the end really different from them. If compensation does not actually take place, if the test is operated as welfare economists normally do and we ask only that compensation be possible, then the principle is clearly no longer irresistible. If compensation does take place, then all we have done is to bring more actual changes under the gain-only principle, while doing nothing to remove the damaging limitations of the principle. What it identifies as better is better, but it cannot identify everything that is better. And this makes it conservative: one question we want to ask, central to fairness, is whether or not it would be better actually to compensate, but this test gives no answer. Finally compens- ation is not always possible, a clear case being some charges that bring death. 150 MORAL IMPORTANCE Yet Mill, having advanced the modest, circumscribed gain-only principle, takes it at other points to be a principle of great scope. Perhaps his thought moved like this. The principle of utility, he says, incorporates the view "that equal amounts of happiness are equally desirable, whether felt by the same or different persons'.40 In "happiness' he seems to include "unhappiness' as well; his only assumption, he assures us, is "that the truths of arithmetic are applicable to the valuation of happiness',41 and addition now seems to be the gain-and-loss type. Finally, he takes the sum that results not only as a general assessment of states of affairs, including how fairly gains and losses are distributed, but also as a criterion of action. General happiness is, he says, "the end of action' and "the criterion of morality'.42 However, if this is indeed the movement of Mill's thought, it is blatantly unjustified. It moves from irresistible to alto- gether easily resisted principles. Once we move to situations of gain-and-loss, no one standard seems inevitable; minimal conceptions of rationality here force no principle upon us. Even within the theory of good, most people are somewhat attracted by the standard that the state of affairs is best which maximizes the sum of gains and losses; but many are also inclined to think that a state of affairs in which someone at a terribly low level of welfare is raised is better even if his own personal utility sum is somewhat less. And nothing that made any of the earlier purely maximizing principles seem irresistible supports a purely maximizing approach here. This is even more to the point when we move from prudence to morality. Perhaps there maximizing action should be subject to deonto- logical checks. Perhaps the maximizing distribution is not the fair distribution. Nothing that made the earlier maximization principles irresistible helps pure maximization as a standard of moral action. This shift from uncontentious maximizing principles to contentious ones, from prudential principles to gain-only principles, and then to gain-and-loss principles and finally to moral principles, is obviously eased by confusion. Different standards of maximizing are difficult to keep apart. Just as a successful attack on one is thought to be a successful attack on another, the obviousness of one is easily turned into the 151 FROM PRUDENCE TO MORALITY obviousness of another. But when we face up to gains and losses, and to how our actions should affect them, new problems, centring on distribution, arise. We cannot fall back on how apt maximization is in the simpler parts of prudential theory. Mill, in his proof, seems not to have realized this. Again and again in the utilitarian tradition problems about distribution are solved by working variations in rather simple conceptions of maximization, as if these relatively safe princi- ples of maximization from the simpler parts of the theory have the resources within them to produce, just by some process of iteration, principles of fairness for the much more complicated parts of the theory.43 A very neat illustration of this central confusion is the famous formula "the greatest happiness of the greatest number', and the mistake on which it rests is instructive to lay bare. This famous formulation had its heyday in the eighteenth century.44 Bentham picked it up for a while, although he seems also later to have dropped it.45 Mill, so far as I know, never used it.46 Nowadays, it is often used to state the aim of utilitarianism, but typically by its enemies.47 Its friends have good reason to steer clear of it: the formula requires double maximization of a kind that makes it incoherent.48 Consider this example. Suppose that, to stimulate produc- tion, a manager offers a prize to the worker who assembles "the greatest number of radios in the shortest possible time'. One worker enters for the prize having assembled ten radios in one hour, a second enters having assembled forty in five hours, and a third with a hundred and thirty in twenty hours. The manager is in trouble; his formula does not pick a winner. He runs into trouble, as von Neumann and Morgenstern point out,49 because his formula requires the simultaneous maximi- zation of two non-independent functions.50 The formula "the greatest happiness of the greatest number' runs into the same trouble. The number of persons happy and the amount of happiness are dependent functions, and we are instructed to maximize both at the same time. True, the amount of happiness and the number happy tend to be directly proportional: diminishing marginal utility means that the more persons happy the greater the quantity of happiness. So why does the formula present problems? The trouble comes 152 MORAL IMPORTANCE simply from their being dependent variables. To increase the one merely tends to increase the other. But suppose that in certain circumstances it does not; maximizing happiness in these circumstances, say, is brought about by holding the number slightly under its maximum. And this is not an unrealistic assumption. The law of diminishing marginal utility applies in most but not all cases; some goods can only be acquired when assets reach a certain point, below which increases might be worth relatively little but just at which they might be worth a lot. But then what action is in accordance with the formula? The formula is, after all, offered as a principle by which the members of any set of choices may be ordered, while in the case I have just imagined the principle has no application. Of course, there are other formulae which do not always have an application that, nonetheless, are coherent. But the trouble with this formula is not that it sometimes fails to give an answer but the reason why it does. The simple formula "the greatest good' might not give a result if a sort of incommensurability appeared: if, for instance, there were two kinds of goods where one was not better than the other, nor were they equal. But there a breakdown comes because the elements of the case cannot be brought under its concepts. With the formula "the greatest happiness for the greatest number' the cases where it breaks down can be perfectly well brought under its concepts. It is the formula as a whole that fails; it points in different directions, which suggests that even when it points in one direction only one of the maxima is working as the standard.51 Why has this flawed formula beguiled so many intelligent persons for so long? I suspect that there are two explanations. One is the strong but confused faith that the obvious rationality of maximization in one-person cases must---it is only a matter of finding the right formula to explain it---be transformable into the rationality of maximization in many- person cases. This is von Neumann and Morgenstern's dia- gnosis in A Theory of Games and Economic Behavior. They contrast the economic problem for a solitary person (what they call the Robinson Crusoe case) with that of a person in a social setting.52 Crusoe's aim is to maximize his own satisfaction, which is a function of variables that either are subject to the 153 FROM PRUDENCE TO MORALITY calculus of probabilities or are within his control. But a person in the social exchange setting, though trying to maximize his satisfaction too, finds that the variables of which it is a function include the actions of other persons attempting to maximize their satisfaction. What it is for many persons to achieve maximum satisfaction is not simply the maximization of the number of persons each achieving maximum individual satisfaction. In many-person cases---for instance in Prisoner's Dilemma cases53---each party may individually seek to maxi- mize his utility, while the others are doing the same, and end up with neither his best nor his second-best outcome. It is improper to assume that the rational procedure in the one-person case is transferable to the many-person case. The other explantion is that the phrase "for the greatest number' is a confused but reassuring gesture towards the moral demands of distribution in what seems still to manage to be a maximizing framework. In fact, some writers, aware that the two maxima in the formula point in different directions, suggest that the formula ought to be seen as an amalgamation of two independent principles: a principle of utility maximi- zation and a principle of justice.54 But a moment's reflection shows that it is a pretty feeble gesture towards justice. What does it mean to maximize the number of persons benefitted? Well, if we take it to mean merely that benefitting more persons is to be preferred to benefitting fewer, that is still a long way from satisfying all plausible demands of fairness. If we take it, more subtly, to mean that a distribution benefits a greater number if it yields a net increase in welfare for a larger number, it still leaves us well short of a plausible principle of fairness.55 Is there a still subtler interpretation? Why these two interpretation fail is that they permit very unequal treatment, and treating persons equally occupies a central place in our idea of fairness. Of course, we can go on searching for further interpretations of the phrase "for the greatest number' until we find one that captures the requirement of equal treatment. But the interpretations already briefly canvas- sed suggest that the subtler interpretations necessarily will be ad hoc stipulations of sense for the phrase "for the greatest number' that will make it near to, or the same as, some notion of equal treatment. And if some notion of equal treatment is 154 MORAL IMPORTANCE what we want, it is better to introduce it undisguised, and not to suggest in this spurious way that it really is there all the time just beneath the surface of the formula "for the greatest number'. Why we are tempted by the illusion of an irresistible, uncontroversial extension of maximization into the many- person case is that each step along the way is so short and looks so innocent. We may move from one person's good to uncontentious forms of the common good to contentious forms and finally, with equity and fairness, to issues of morality. But issues of equity and fairness are not settled just with the resources of prudential good or uncontentious forms of the common good. They cannot be settled, either, by appeal only to rationality. To maximize in all many- person situations raises the central issue of morality---the kind of equal respect for all persons' interests that morality represents. This expands somewhat John Rawls' unclear criticism that utilitarianism ignores the separateness of persons.56 If the criticism is that it is a mistake for a moral theory to take maximization as appropriate in many-person cases be- cause it is appropriate in one-person cases, then it cannot be denied. But one can hardly take the difference between the two cases as showing that the maximizing approach is inappropriate, or a non-maximizing approach appropriate, in the many-person case. It is possible to treat the move from one to many as less important than it is; it is also possible to treat it as more important. One should keep all the stages of the move from the one- to the many-person case in mind, because it is easy to mistake both the truth of one claim about maximization for the truth of another, and the falsity of one for the falsity of another. Mill makes the first mistake; Rawls, I shall shortly argue, makes the second. When we come to views about contentious many-person cases, all the competitors need further justification: the mere fact of differences between one- and many-person cases shows nothing. All that we can conclude now is that there is no irresistible, uncontentious conception of rationality that will carry us from one-person to many-person deliberation. Part of 155 FROM PRUDENCE TO MORALITY fixing a rich conception of rationality, which might carry us that distance, is fixing our thoughts about what can, in the end, be regarded as good moral reasons. 5.The nature of the self and the source of morality We have to find space for morality deep inside human motivation. We have also to find morality a position of authority, because there is a lot that is internal to human nature that we rightly choose to ignore. As we have just found, we are not going to make morality both internal and authoritative by making it part of self-interest, or of personal aims, or of some minimal notion of rationality. We have got to find some other way. What I want to propose is that the reason-desire dualism has been as destructive of understanding in theory of action and morality as the mind-body dualism has, until recently, destroyed understanding in epistemology. We treat the word "reason' (or "cognition', "understanding') and "desire' (or "conation', "feeling', "sentiment', "passion') as marking two separate domains. Then we have to explain how they are related, and since the time of Plato we have reached for political metaphors. Those of us of a rational bent give "reason' a commanding, God-like authority over human life. Others of us of an anti-theoretical, empirical bent, rightly suspicious of talk about the domination of reason, have made it the slave of the "passions'. But these political metaphors hardly suit their subject. As we have seen, when it comes to practical deliberation, reason and desire are not independent enough for one to be master and the other slave. The lesson is that we have to scrap this model and all its related dualisms (objective/subjective, internal reasons/external reasons, desired because valuable/valuable because desired) and start an account of practical deliberation afresh. A good place to start is with a fresh, more accurate picture of the self. We want things, but we do not regard a desire, just by its mere presence, as constituting a reason for action; it has also to be able to survive the sorts of criticisms that we have talked about. We acknowledge reasons, both to believe this or that and also to do this or that, but a practical 156 MORAL IMPORTANCE reason does not move us to action unless we care about what the reason points to. Each of us has a conception of what makes his life go well. It is a highly complex conception. With experience and reflection we arrive at some understanding of the ingredients of a good life. A mature conception will include autonomy, understanding of ourselves and our surroundings, enjoyment, deep personal relations, accomplishment, and so on. Many of these prudential values---say, deep personal relations--- incorporate major springs of human action. We are naturally (to a large degree genetically) disposed to deep attachments to certain persons, and the prudential value that I am calling "deep personal relations' arises out of and supports these natural attachments. The influence of desires and values typically runs both ways: for instance, we naturally form bonds to certain persons; these bonds are central to making life good; so our judgments of value support these bonds. And our judgments of value are capable of moderating our action; if upon deliberation we decide that what we now in fact care about is not worth caring about, we can stop caring and act differently. One of our aims is to live a life of substance or value or weight. Another, not far removed, is to understand ourselves and our surroundings. These aims are themselves quite complex. We want to be in touch with a reality outside ourselves. We do not want just to have convincing impressions of having a life of value, of accomplishing something with our lives. We also want to have clear perceptions of the reality about us, including the reality of other persons. They too have interests, which matter as much to them as ours do to us. And since we want to live a life of value or weight, we have to decide what a valuable life is. This is a point, among many others, at which prudence and morality will not stay apart. A valuable life, in the sense that we are now trying to understand, cannot just be a life filled with prudential values conceived in fairly narrow self-interested terms. A valuable life, in the sense we are after, consists importantly in doing things with one's life that are themselves of sufficiently substantial value to turn back on the life itself and make it valuable. And we cannot see what we do in those necessary terms if we have no regard for, 157 FROM PRUDENCE TO MORALITY or if we damage, values generally, including the value of other persons' lives. This is not an argument for morality in the narrow sense. A person might be able to live a life of value and substance by single-mindedly painting his pictures, ignoring all his family responsibilities. But it is a life of value and substance only if his paintings are also of substantial value and, in the end, valuable to the rest of us. Prudence cannot be kept in narrow confines. What is prudentially valuable must, at various points, spread out into areas that do not look like a part of prudence at all. The boundary defining prudence cannot, along this frontier, really be fixed. It cannot be fixed because one kind of value tends to spill over into another kind. Part of living a valuable life is living one with a sense of direction. It is a life open to value of any kind; it is one the direction of which is determined by what is really of value. I have explained prudential value as the fulfilment of informed desire. There are therefore two consti- tuents to prudential values: desire that is formed by appreciation of its object. That it is my, or any one individual's, pain or accomplishment is not a constituent of my, or anyone's, conception of prudential value. How well a person's life goes matters---matters immediately to that person, of course, but also matters generally in this sense: depending upon what happens to him, his life will be more or less valuable. Valuable to him, of course; but I do not have to see things from his point of view or, alternatively, see them sub specie aeternitatis to recognize that. In any case, the subjective point of view and the view sub specie aeternitatis hardly between them exhaust the ways of looking at the world; most important deliberation about prudential value uses neither. That value has the role in a life of point and direction that it does, and that the locus of value is not just oneself, brings one to a certain sort of impartiality. Should I, I wonder, sacrifice some trivial interest of mine (not to get my trousers wet) because an enormous interest of yours is at stake (not to drown)? But my interest, though tiny, is of importance to me, while your (great) interest is of importance of you. We need a move from importance relativized to an individual to a non-relativized notion of importance. And we have one: namely, the values at stake and the reasons for action that 158 MORAL IMPORTANCE they constitute. My interest gives me a reason to do one thing; your interest gives me a reason to do another. In this extreme case the balance of reasons, even if I shall have reason to give more weight to my interests than to yours, is clearly on my wading out. We should lose all conception of strength of practical reason if we did not say so. And the conclusion about what the balance of reasons in this case is does not assume, or need, an independent concern on my part about your survival; one's concerns have to be responsive to how worthy of concern its possible objects are. It underdescribes this case to say merely that, as a matter of fact, some persons care about morality, that they desire to act well. Some philosophers, in effect, stop there;57 they find, at least in some of us, the sort of reason, namely an internal one, that they think necessary in order for us to have a reason to be moral. But it is not just that some of us care about morality in the abstract or about this or that substantive matter that makes it up, as if that fact about our sentiments were at the bedrock of the self. Rather, morality is one of the things worth caring about. That, of course, may be a conclusion that a person will arrive at reluctantly, because moral concerns can conflict with other, especially large and vivid personal concerns of his. One of our important prudential aims is understanding our place in the world, including understanding what are good reasons for action. But, apart from that, we also recognize the force of reasons, quite independently of the desires and concerns we happen to have. If we could not, then we would not recognize the independent authority of reasons in factual matters of truth and falsity. With reasons for belief too, our desires, though they can distort and obscure, do not undermine the independent force of the reasons. We subject both our belief and our action to impersonal demands, demands not associated with our personal desires. There are natural springs of action, including habits and desires. But in each case there are also independent reasons. No one thinks that subjecting our beliefs to those impersonal demands is, though often difficult, psychologically unrealistic. And it is not psychologically unrealistic, though it may usually be more difficult, to subject our actions to them either. How do we resolve conflicts between reasons for action? 159 FROM PRUDENCE TO MORALITY Reasons for action form themselves into a hierarchy. Some are stronger than others; some are global in scope while others are only local. Desires, as we discussed earlier,58 also form a hierarchy. For all of us, our reasons for action shape our desires, though desires are shaped by reasons to different degrees and our hierarchy of desires is never wholly shaped by our hierarchy of reasons. Conflict can be between reasons for action, or it can be between a reason and a desire. If it is a conflict between reasons, we try to settle it by looking to their relative strength as reasons. That, in turn, means looking at the relative magnitudes of the values with which the practical reasons are linked, some of the complications of which we have seen in connection with the measurement of prudential value.59 But only some of the complications, because earlier we were discussing only prudential values and now we are discussing values of all kinds, including moral ones. Since prudential values exhib- ited a complicated structure, with various kinds of incom- mensurabilities appearing, there is no reason to think that the relation between values generally is going to be simple. How are prudential reasons related to moral ones? Are moral reasons lexically prior? How are kinds of moral reasons related to one another? Are considerations of human rights lexically prior to considerations of general well- being? Certainly we are not always going to be able to determine the strength of practical reason by simple inspec- tion. We say, "No individual may be sacrificed indefinitely for the general good', and that seems to many of us as certain as any moral consideration is going to be. And so the notion of certain especially basic rights as trumps over the general good seems well established. But in this case, I believe and shall later argue,60 appearances are deceptive. At any rate, some test is needed; in the face of, say, what seems to be a particularly crass, counter-intuitive consequence of utilitarianism we announce, with firm confidence in its correctness, this or that human right but then, up against some very different problem, find that we have overstated things and have now to back down. We must subject candidate moral reasons---and candidate practical reasons generally---to some test or other. In the Introduction I 160 MORAL IMPORTANCE sketched the two main tests that I think we should try to apply: the test of completeness and the test of correctness. So to resolve conflicts between practical reasons we have to decide just how much worth caring about certain things are. What seems unlikely to emerge from such discussion is what Sidgwick called the Dualism of Practical Reason.61 At least, there is no problem just in the fact that there are prudential reasons and moral reasons and no super-reasons of some transcendent category in terms of which to resolve conflicts between them. We no more need super substantive practical reasons than we need super substantive prudential values. We resolve conflicts between prudential values by using the quantitative notion of value.62 We can resolve conflict between prudential and moral reasons by using the notion of strength of reason. The same is true of "oughts': we are not stuck with oughts (prudential) and oughts (moral), in need of a super, unsubscripted ought to mediate between them; what in the end concerns us are only oughts that connect with things (of any category) worth caring about. Furthermore, the penetra- tion of prudence by morality, in particular that it penetrates it in a way that shifts serious deliberation on to the level of abstraction where the categories "prudential' and "moral' are left behind, makes dualism an unlikely model for their relation. This is not to say that ethical push and ethical pull eventually meet; so far as I can see it points to the opposite conclusion. It is hard to see how, even with a perfectly saintly person whose desires are in complete harmony with the hierarchy of practical reasons, what could be identified as prudential reasons would always support what moral reasons support. Though living a life of point and weight and of conformity to values generally may be seen as prudential values, it is hard to see how their value, purely prudentially, could be greater than a good life itself---if it came to the terribly hard choice between morality and survival. Moral reasons and practical reasons overall might outweigh pruden- tial ones in such a case, but identifiably prudential ones still would be in conflict with them. It is not that death could never be better than dishonour, but rather that it is hard any longer to see the relevant notion of dishonour solely under the heading of prudence---it has to be something less hedged in 161 FROM PRUDENCE TO MORALITY than that. This does, however, call for some qualification. Since morality penetrates prudence, so making the prudence/ morality dualism hard to maintain, my talk about "identifiably prudential reasons' is not entirely satisfactory; prudential reasons run, without boundary, into moral ones. There is, though, another conflict that can look deceptively like a manifestation of the dualism of practical reason, namely conflict between the two hierarchies at work in action. Our hierarchy of desires can conflict with our hierarchy of practical reasons. What we recognize as worth caring about can conflict with what we actually care about. This is not a clash between two kinds of reason, but it is typically a clash between what we see should be cared about and the stubborn fact that we often actually care more about ourselves. But the most important point to make about the putative dualism of practical reason is that deliberation of a sufficiently global scope is not conducted in terms of "prudence', "self- interest', or "flourishing' on the one side and "morality' on the other. It is conducted in terms of strength of practical reasons. This is not to say that reason in the end rules sentiment. That is the indefensible dualism again. It is to say that values, neither expressly prudential nor expressly moral but values taken at a higher level of abstraction, are what we appeal to: the notion of what, all things considered, is worth our concern. Just as when with prudential values we deliberate not by appeal to any one substantive prudential value but to the notion of prudential value itself, here too we step up another rung in the ladder of abstraction. There is nothing mysterious or suspect about this step; it is one more of the same sort that we have continually to take in deliberation. How complete the resolution of conflict between practical reasons will be, and what complexities of structure the hierarchy of reasons will exhibit, we have to wait to see. The same was true with prudential values; we could settle the formal features of prudential values only when we also had a substantive account of what those values were. What resolution of conflict there is will start with one's representing practical reasons to oneself as completely and vividly as one can manage. Why should one be moral? One cannot answer: Because self-interest counsels it, or, Because it is among one's 162 MORAL IMPORTANCE personal aims, or, Because it is a requirement of reason on some uncontroversially minimal conception of rationality. The only answer possible seems to be of the form that, when one represents the full range of practical reasons to oneself, moral reasons find a place in the hierarchy, and a place high in it. The strongest answer, of course, is that they occupy the highest place. But that depends upon the content of each moral reason. There is no dualism of prudence and morality to raise obstacles in principle to moral reasons' being strongest. But if acting justly might make the heavens fall, if that is the sort of requirement that justice turns out to be, then our collective self-interest might well outweigh it. If respecting rights could cause or fail to prevent catastrophe, then again self-interest probably outweighs it. Whether these would be cases of pure self-interest's outweighing morality or of self- interest's taking on moral weight would also depend upon the content of moral reasons. At this level of generality, with no well worked-out account of moral reasons to call upon, there is not much more that can be said in answer to the question, Why should I be moral? So what I have offered is not as decisive an answer as the rejected alternatives. It leaves hostages to fortune: we have to lay out moral reasons and see what place they take in the hierarchy. But until we do that, it is, I think, the best answer we are going to get. IX EQUAL RESPECT 1.Equal respect and psychological realism THERE are two topics from the last chapter that I want now to carry further. The first is what I called equal respect. I said that living in accord with practical reason brought along with it a commitment to a certain vague, not very rigorous form of impartiality---namely, that everyone is to be granted equal respect. By "equal respect' I do not mean anything so demanding as what could be called "equal concern'---for instance, giving as much weight, utilitarian fashion, to the welfare of a stranger as I do to the welfare of my children.1 Yet even such moral theories as do allow some sort of partiality to family and friends also demand some form of equal respect; they accept, say, a weak form of universalizabil- ity that requires treating persons equally unless there are relevant differences between them. In more familiar terms, respecting persons equally is looking at them from the moral point of view. Now this vague, capacious concept of equal respect is too indeterminate as it stands to do much work. So philosophers develop it in various ways. Some develop it into the viewpoint of the Ideal Observer, and others into the viewpoint of the Ideal Contractor. One of the things I want eventually to argue2 is that there is still scope for innovation here, that there are other, unfamiliar developments that are also promising. The second topic is the Requirement of Psychological Realism. In one form or other the Requirement is undeniable. There is no point in announcing moral restrictions unless they fit the human psyche. Restrictions are meant to restrict, so if moral considerations are actually to shape action, they must be able to find a place inside human motivation and, what is more, a position of authority. Now I want to bring these two topics together. How are we to flesh out the notion of equal respect so that it yields 164 MORAL IMPORTANCE plausible moral restrictions but, at the same time, meets the Requirement of Psychological Realism? The very importance of the Requirement makes it tempt- ing to try to derive morality from ground that is not rich enough to yield it. The Requirement would easily be met, if morality could be grounded in self-interest. But self-interest seems pretty clearly not to be rich enough. The arguments for that we have already seen.3 A shrewdly self-interested person is likely, simply for the sake of the quality of his own life, nearly always to respect the interests of others. But there are times when he will not---for instance, if he finds himself a member of an unjustly privileged class in a stable society. Furthermore, the best ways of reconciling morality and self-interest do not help meet the Requirement. There is much to say for the idea that no life can fully flourish, reach its highest peaks, unless it is a moral life. But this is plausible only because our conception of flourishing is so penetrated by morality that the reduction is no longer, in any interesting sense, a reduction of morality to flourishing. We could broaden the ground, and still fairly easily meet the Requirement. We could ground morality in personal aims. Nearly all of us have aims which are not purely self-interested; we are also interested in the welfare of our children or our friends or institutions to which we are committed. Some of the ends of some of us are for the good of some other people. But these natural sympathies are still too limited to be a rich enough ground for morality. There is the same problem as before. The member of a secure privileged class may not feel much sympathy for the exploited. And even if these limited sympathies could be got to spread, the spread would be unlikely to be a purely involuntary movement. Sympathy tends to spread when we see reason for it to spread, and otherwise not. And even if the spread occurred without a reason, as just a brute fact about us, we should be unlikely to grant the restraints that such sympathy generates any author- ity if in reflection we had to referee between warring feelings of sympathy and self-love.4 The natural move at this point is to admit that the grounds are not yet rich enough and once again to make them richer, perhaps by bringing in more of the workings of society. 165 EQUAL RESPECT Morality, it is entirely plausible to suggest,5 has an object, a point, a function---namely, to counteract what it is in the natural course of events that makes things go badly for us. And they are, primarily, limitations of one sort or another: limitations in resources, in our knowledge of facts, in our accord with logic, but especially in our sympathies.6 Limited resources we often cannot do much about. But human limitations we can to some degree combat. Limited resources and limited sympathies together lead to competition, easily escalating into conflict, and we end up losing the benefits of co-operation.7 That is the problem that morality is needed to solve. The object of morality is largely to counteract limited sympathy, and we might think that we can settle the content of morality simply by fixing how best to do it. For instance, when two persons find themselves in a situation that has the form of a Prisoner's Dilemma, or something close to it, they need to save the situation from deteriorating into the third best outcome for each. The best that could be achieved by any device that both parties would willingly go along with is one that leads to the best of the symmetrical results, so we need a device that would ensure the second-best outcome for each.8 Thus, what we need to counteract limited sympathies are some or all of the following: we need more knowledge; we need organization, including legislative procedures; we need rules and coercion to back them up; and we need good dispositions.9 Most of that is true and important. But is it yet rich enough? Can we generate moral restrictions simply by decid- ing what is needed to make things go better? It seems not. Again, we cannot derive any recognizably moral instruction to the members of a stable but unjustly privileged class. We discover nothing about justice between generations, or about our obligations to the Third World. And good dispositions, especially disposition to benevolence, although an important part of the scheme, are not fertile enough to make up these deficiences. It is plausible that what would be needed to make things go better is a very wide benevolence, a sort that spreads far beyond family and friends, and it is furthermore plausible to think that once benevolence spreads that far it will naturally go on spreading quite on its own until it covers distant strangers and future generations too. This may seem to 166 MORAL IMPORTANCE offer hope of generating obligations to the Third World and to future generations. But it is open to an objection similar to one made against rule utilitarianism: extended benevolence will be recognized, by the reflective among us, as being generally useful in making things go better but as having no use in the case of unpowerful distant strangers or totally powerless future genera- tions. One would not have to be very bright to notice that a device of morality that has a point within certain confines has run on into areas where it entirely loses its point. So we should be right, to the extent we can, to fight the promptings of a very extended benevolence, and so we should have in the end no explanation of obligations to the Third World or future generations. Our only resort, it seems, in trying to enrich this sort of ground, is to enrich the notion of making things go "better'. Better for whom?, we might ask. If "better' just means "better for me', then it is no ground for morality. If "better' means "better for some and no one worse off', then it is no ground for morality because it fits so few situations. If "better' means "better for some some though worse for others but better overall', then it may be rich enough to yield morality, but only because we should have to have built most of it into the notion of "better overall', which was meant to be the centrepiece in the device that would generate morality. This same shift from an interpretation too poor to yield morality to an interpreta- tion rich enough but which no longer yields morality because it has had to presuppose it, appears in the report I gave a moment ago of the solution to the Prisoner's Dilemma. The best that could be achieved by any device that both parties to a dilemma would willingly go along with is, it was proposed, the best symmetrical result, namely the second-best outcome for each. Now if this argument amounts to saying that one cannot get a stable solution unless one settles for the outcome that everyone will freely accept, then no doubt it fits some situations. Sometimes, no doubt, that will be the only deal that can, as a matter of fact, be struck. But it does not fit all situations; sometimes one can have, for a while (perhaps even a century or two), a stable solution based on ignorance, indoctrination, apathy, or bullying, and then one has no reason not to opt for institutions that make one privileged. 167 EQUAL RESPECT The only way to give one a reason is to regard each person's freely accepting a solution as required, not by the facts of a situation, but by, say, the respect due to each person as an autonomous agent. But, again, what we are then doing is appealing to a conception of the moral person, which already carries much of the moral content that we were meant to generate. So it seems to me that although the Requirement of Psychological Realism makes it tempting to ground morality in self-interest, or in personal aims, or in the social function of morality, none of these grounds is rich enough to give us what we think of as morality. We seem to need a notion of moral reasons for action that breaks away from self-interest or personal aims or universal social pay-offs. What I now want to argue is that familiar forms of utilitarianism break too sharply, and that familiar forms of contractualism do not break sharply enough. So the first have trouble meeting the Requirement of Psychological Realism, and the second have trouble generating moral requirements. 2.The utilitarian view of equal respect I started by explaining that equal respect, the way I am speaking of it, constitutes the moral point of view. There are different ways of trying to capture it. One well-known way is the device of an Ideal Observer: the moral point of view, it is proposed, is a benevolent view from a position above the fray, granting everyone equal consideration. This can easily turn into the utilitarian conception of merging interests by maxim- izing utilities, counting everybody for one. John Rawls, of course, thinks that it merely turns into an undesirable impersonality and not the true impartiality of the moral point of view, which he sees as captured by a different device, the device of the Ideal Contractor:10 the moral point of view, it is proposed, is the view of equal contractors from behind the Veil of Ignorance. There are still other accounts of the moral point of view, which I shall come on to later, but let me for the moment concentrate on these two. The perspective of the Ideal Observer easily leads to utilitarianism, and that of the Ideal Contractor leads, Rawls 168 MORAL IMPORTANCE thinks, to his two principles of justice. It is hard to see exactly how, in each case, the perspective stands to the substantive moral view derived from it. It is clear, at least, that they are not the same. Each perspective, though meant to make the moral point of view more determinate, is itself described so vaguely that it can be further specified in different ways. There is a strong argument that the rationality of the Ideal Contractor would lead him, not to Rawls' two principles, but to Average Utilitarianism.11 And there are strong arguments that the benevolence of the Ideal Observer would not let him accept some of the sacrifices of one person to others that utilitarian aggregation permits.12 Still, since notions like rationality and benevolence are very vague, one way to give these rather empty conceptions of equal respect more content is to fix what follows from them. So I want to take the view of the Ideal Observer as fixing on utilitarian aggregation, and that of the Ideal Contractor as fixing on Rawls' principles or some- thing close to them. Let me start with the utilitarian conception of equal respect. One element is the Benthamic formula "Everybody to count for one, nobody for more than one', which is a principle of equal weight or lack of bias. It is not a separate principle over and above the principle of utility. Mill's account of its status, it seems to me, is right. It is, he says, simply part of what is involved in applying the principle of utility.13 If one's aim is to maximize utility, whose utility it is is irrelevant. One is barred from being a respecter of persons by being a respecter of utilities alone. I think that the second point to note about the utilitarian conception of equal respect is that merging persons' interests into a single moral judgment by maximizing them is a distributive principle. It is a view, right or wrong, about when sacrificing one person for another is justified. It is just a modern muddle to contrast sharply distributive and aggregat- ive principles, as if an aggregative principle could not also be fully deliberately distributive.14 It crops up commonly in regarding, as economists often do, an aggregative principle as a principle of "efficiency' and other principles as ones of fairness. Similarly, no plausible principle of distribution ---think, for instance, of Rawls' Difference Principle---could be 169 EQUAL RESPECT purely distributive, without some maximizing tendency, as if reducing everyone to the same level of misery could satisfy it. Every plausible principle of equality is based on the thought that everyone matters and matters equally, and to stress only formal features of distribution is to recall the equally but to forget the matters. Even a principle of a minimum acceptable level of welfare has, if not a maximizing, at least a quantitative element. And the principle of utility, too, represents another conception of the distribution that equal respect for persons requires. But that still leaves untouched what may seem to be the major fault in the utilitarian account of equal respect. Merging persons' interests is maximizing their utilities. The utilitarian conception of equal respect simply transfers the structure of intrapersonal trades to interpersonal ones. Does my justified willingness to suffer a cost now for a greater benefit for myself later also justify my having to suffer it for your greater benefit? And if it did, I might lose out not only in this trade to you but, for similar reasons, in another to him and in another to her. Should I accept the cumulative effect of those trades even if, were things to fall out awkwardly, they built up into something disastrous for me? The looseness of utilitarian restrictions on trade-offs creates problems with the Require- ment of Psychological Realism. Would a person who believed that morality obliged him to go to the wall merely to increase the aggregate social utility be able to find a place for the obligation in his motivation? Would he if, in the name of morality, he had to sacrifice the persons he most cared about to benefit persons for whom he cared nothing? It is at this point that many people also protest in the name of morality: "One person may not be sacrificed, without limit, for the good of others', or "The well-being of one person cannot simply be replaced by that of another', or "This would not take seriously the distinction between persons'.15 The first two protests announce what are clearly moral principles; the last seems to gesture towards some fact, and this appearance has led some writers to regard "the separateness of persons' as a reason to drop the utilitarian conception of equal respect.16 But since there is no fact of "separateness' that anyone has overlooked, no delusion that a group of persons is one 170 MORAL IMPORTANCE super-person, the protest that utilitarians overlook separate- ness amounts to no more than the claim that one ought not to transfer the model of intrapersonal trades to interpersonal trades. It is an expression of one view about equal respect, and so not a reason for choosing it. Still, for good reason or bad, it is at this point that many persons turn to the contractualist view of equal respect. 3.The contractualist view of equal respect It is quite clear that Rawls regards the perspective of the self-interested contractor behind the Veil of Ignorance as one, though perhaps not the sole, moral point of view.17 The contractor's reasoning is an acceptable way of reaching many different kinds of moral conclusions, not only conclusions about justice. Roughly speaking, ignorance supplies impartial- ity, while self-interest supplies the moral significance of the separateness of persons. As for justice, this productive tension yields, Rawls thinks, two principles:18 1. Each person is to have an equal right to the most extensive basic liberty compatible with a similar liberty for others. 2. Social and economic inequalities are to be arranged so that they are both (a) to everyone's advantage and (b) open to all. His two principles are certainly, at first sight, very attractive. And they echo other most attractive-sounding intuitive convictions such as one that Rawls employs, "Each person possesses an inviolability founded on justice that even the welfare of society as a whole cannot override'.19 But it is strange that principles, a major motive for whose support comes from the counter-intuitive consequences of the principle of utility, should turn out to be every bit as counter-intuitive themselves. Rawls remarks at one point that, no doubt, some of these attractive intuitive convictions are overstated a bit,20 but the trouble goes far deeper than that. The perspective of the Ideal Contractor yields a cut-off for all utilitarian trade- offs: the worst-off group has to be as well off as possible. But that restriction---at least our intuitions tell us---is far too strict. Rich societies, where the worst off are well off, do, on the face of it quite reasonably, allocate some resources---to art, for instance---in a way that mainly benefits the better off. And all 171 EQUAL RESPECT societies allow some persons' life prospects to sink to disastrous- ly low levels in order to protect the relatively minor benefits of the many. There are over twenty thousand avoidable deaths each year in England because of poor help or the victim's ignorance or hesitation, and to remedy the situation money could be, but is not, taken from art or education.21 And the French Government knows that some people, through no fault of their own, die each year in automobile accidents because of the beautiful roadside avenues of trees, but it does not cut them down. So we do allow trade-offs between the common good and disastrous drops, even when the drop is as disastrous as death and the good merely more aesthetic pleasure. And even if the avenues of trees caused no deaths, the French maintain them at a huge cost that could be used to improve, say, the housing of the worst off. How might Rawls reply? He might remind us that his two principles of justice are not meant to apply at every point in society; they are meant to apply to what he calls "the basic structure', that is to say, to "the way in which the major social institutions fit together into one system, and how they assign fundamental rights and duties and shape the division of advantages that arises through social cooperation'.22 So they apply to the political constitution, to forms of property, to general economic organization, and suchlike. Once these general organizational features are in proper order, what happens thereafter, Rawls says, satisfies pure procedural justice; that is, it is the outcome of a fair procedure where there exists no other test for fairness of the outcome than that it is what in fact comes out.23 Furthermore, the two principles are not meant to be applied to the worst-off individual, but to the worst-off "representative' individual---that is, to a whole social group.24 But these exclusions clearly do not meet some, and probably do not meet any, of the examples. Since the contractors are self-interested, they see the society whose basic structure they are fixing as "a cooperative venture for mutual advantage'.25 They will not agree with one another at all, unless they secure as much advantage, and certainly as little disadvantage, as they can hope for. They will, Rawls thinks, adopt a maximin strategy, and this play-safe approach will 172 MORAL IMPORTANCE appear in each of the two principles of justice. It will appear in the first one as a strong right to self-preservation,26 so strong that at one point Rawls gives saving life priority over even avoidance of slavery.27 It will appear in the second principle as the requirement that the worst off be as well off as possible. This Difference Principle, Rawls says, would require at least a minimum acceptable level of welfare,28 so it would, even more, require protection of life. Both principles would have to give an especially high priority to these protections, simply because life is a necessary condition for the use of any liberty or the enjoyment of any good. That the two principles apply to the basic structure does not matter; they will also have to apply, as Rawls openly acknowledges,29 to "life prospects'. Just as principles that govern choice behind the Veil of Ignorance (e.g. maximin) reappear in some form in the principles of justice chosen there (e.g. the Difference Principle, with its requirement of the best worst), so one would expect the actual arrangements of society sanctioned by the two principles to reflect certain of their most important features. Rawls conceives of a sequence of stages: the original choice, the constitutional convention, legislation, and finally judicial decision.30 The Veil of Ignorance does not totally lift as soon as the first stage is through; it draws back slowly as we progress through the sequence.31 So play-safe will still have some point at each stage, and the institutions, laws, and priorities that appear further along in the sequence will embody the important protections guaranteed by the prin- ciples further back.32 Otherwise, Rawls' criticism of utilitarian- ism would entirely fail; he stresses the extreme demands of utilitarianism ("the sacrifice of the agent's private interests, when this is necessary for the greater happiness of all') and argues against them that "A rational person, in framing his plan would hesitate to give precedence' to a principle that allowed these sacrifices.33 Whatever play is left in the basic structure, it must not be play that can lead to such sacrifices. So the same failing that appears in self-interest as a ground of morality reappears in Rawls' contractualism, because self-interest is not strongly enough checked there.34 True, Rawls places self-interest (as well as the demands of social co-operation and narrow sorts of rationality) in an entirely 173 EQUAL RESPECT new setting, namely inside a conception of the moral point of view, where each is meant to be checked by the impartiality imposed by ignorance. But they are left too robust. Everyone has to benefit from the bargains struck by the contractors;35 their decision has to be unanimous.36 Everyone must think that he has struck the best deal he could.37 Each must find the terms reasonable.38 Now, the elements of self-interest, of best possible bargain, and of narrow rationality add, without doubt, to the attractiveness of Rawls' view of equal respect. They make it seem that contractualism manages at the same time to be a non-reductionist moral theory and yet comfort- ably to meet the Requirement of Psychological Realism. No heroic self-sacrifice is asked for; the burdens of moral commit- ment are kept well within human capacity.39 Contractualism needs only, as Rawls puts it, "the psychological law that persons tend to love, cherish, and support whatever affirms their own good',40 whereas utilitarianism expects our sympa- thies to be widely extended. But morality seems uncomfortably more demanding on the human psyche; at least, it needs more than Rawls provides, and a large element of extended benevolence is an obvious candidate. In recent papers, Rawls has, while not at all retreating from the ground of checked self-interest, made use of another ground. The principles of justice, he says, "must issue from a conception of the person', the free and equal moral person.41 But this does not really help. As Rawls well knows, the notion of a free and equal moral person is too empty, on its own, to generate many moral conclusions; it needs principles like the two principles of justice as its further specification. But only principles like them; it cannot be those two principles, because they are too strict. And the idea of the moral person does not show where else to go. Of course, another place to go is to some other form of contractualism. Ignorance, as we have just seen, does not check self-interest toughly enough. Most recent forms of contractualism drop both self-interest as the contractor's motivation and the large element of ignorance that is con- sequently needed. Instead, they enhance the role of rationality and of social co-operation. For instance, we could take contractors to be motivated to find reasons for principles of 174 MORAL IMPORTANCE conduct that others cannot rationally reject as a basis for informed, unforced general agreement.42 Impartiality would then be represented largely by treating people as informed and unforced. Rationality would exclude any rejection of prin- ciples unreasonable given the aim of finding social agreement between informed and unforced people.43 T. M. Scanlon argues that we should say "No one could reasonably reject' rather than "Everyone could reasonably accept', because the latter would not be strong enough.44 Suppose there were a principle requiring quite severe sacrifices from a group, the members of which were, through the happy chance of their being public-spirited, willing to accept them. Their willingness might be reasonable; for example they might have some entirely worthy supererogatory reason. But, because of the great sacrifice, it would not be unreasonable for them to reject it. This version of contractualism is therefore just as strong as Rawls'. What is supposed to make it reasonable for them to reject the principle is that it is contrary to their self-interest. All this again gives too much scope to self-interest to be able to generate morality. The earlier examples that counted against Rawls count here too: for instance, a person who will die young unless a cure is found for his congenital disease has a reason to reject rules that allow diversion of funds, say, to the arts, which will satisfy what are clearly lesser interests of the rest of us. Moreover, its notion of rationality seems so narrow that it is doubtful that it can explain the authority of moral rules. It does not require us to be moved by more than the need for agreement and by a desire to justify our actions on grounds that no one can reasonably reject; beyond that our motives may remain intact. What all of us together can agree to is likely, therefore, to have large elements of compromise to it.45 If someone sees an important interest of his threatened and is by nature not given to much self-denial---say he is an adult male who will have none of the "women and children first' approach to life---then he has a reason to reject any set of rules with "women and children first'. But then why should I, who am a little more altruistic and who believe that there is a lot to be said at least for a policy of "children first', think that my obligations are determined by rules that emerge from a compromise with him? The trouble is that the device is meant 175 EQUAL RESPECT to build only fairness into rules but it ends up building in more; it builds in self-interest of a form that raises doubt about their standing as moral rules. The obvious change now would be to shift from "No one could reasonably reject' to the looser "Everyone could reason- ably accept'. But we should need to make the notion of rationality compensatingly richer. Then, depending upon the degree of richness we add to it, either of two things is likely to happen. First, if we make rationality not very rich, we shall find that the ground of morality has not become rich enough yet either. Suppose, like David Gauthier, we see rationality on the model of rational bargaining.46 The moral point of view, we could then say, is the view of an arbitrator trying to reach a fair compromise between persons with conflicting interests. He would not try simply to maximize the satisfaction of the interests, because that could sometimes require virtually sacrificing one of the parties; he would try to strike a bargain that is acceptable, always given the need to reach agreement, from the standpoint of every party.47 But this would merely be the overly strong requirement yet again. But then, second, if we make rationality a very rich notion, we are likely to find that the ground is rich enough to generate morality, but only in the way we have already seen. We shall have already derived morality in some other way and planted it in the account of rationality. Suppose that, instead of making the moral point of view that of an arbitrator, we make it, as B. J. Diggs does, the view of a moral judge, who hears each party's advocate and then, in a spirit of reconciliation, renders a "reasonable decision'.48 But if the requirement that the decision be "reasonable' is no stricter than that the judge cite good reasons, including moral reasons, then there is no doubt that the conception is rich enough to give satisfactory moral answers. But then what is reasonable is no longer worked out from the needs of mediating between conflicting interests, and the conception of a contract will be doing no work. So the fault that appears in Rawls' contractualism reappears in these later versions, because their checks on self-interest, although different from Rawls', do not do the job any better. Where can the contractualist go? It is hard to see how we ought to tinker with the contractualist machinery, because it is 176 MORAL IMPORTANCE hard to see what else we want it to come out with. What cut-off to utilitarian trade-offs are we after? The thing to do, of course, is to look for a weaker restriction. A few years ago Amartya Sen looked for a much weaker one that would still have the form of an absolute cut-off for utilitarian trade-offs. If there were one, so weak as to be irresistible yet still eliminating utilitarianism, that would be a result of no small importance. He suggested the Weak Equity Axiom, which goes: if one person is a less good utility convertor than another, say, because he is handicapped, so that at any level of goods he is less well off, then he must be given something more than the other.49 One might even make the requirement weaker still: the unfortunate one must not be given less, if he is below and the more fortunate one already above a minimum acceptable level of welfare. But these "weak' requirements are still too strong. Surgeons, in choosing between patients for a kidney transplant, are often guided by a patient's prospects. If one patient, Adam, has been chronically ill, below the minimum acceptable level for most of his life, with poor prospects of surviving long after the operation, and another patient, Eve, never as ill, has good prospects, Eve may be the one to be operated on.50 That this very weak requirement turns out too strong is also a result of no small importance. It suggests that we ought to give up trying to make moral theory accommodate an absolute cut-off to utilitarian trade-offs. It suggests that instead the structure of moral theory should perhaps be a plurality of same-level principles. Instead of a cut-off to trade-offs, there would be a counterforce to them: say, a principle of welfare maximization sanctioning trade-offs and a principle of equal welfare restricting them, with sometimes the one and sometimes the other winning out in conflict. In the earlier case, the surgeon could help Eve, the better-off one, a lot and Adam, the worse-off, only negligibly, and it is obviously this great disparity that points towards helping the one who is better off. So let us change the story, now making the disparity less great. Adam, we shall say, is an otherwise healthy man of seventy who with a transplant will live about another ten years; Eve is a chronically ill woman of thirty-five who with a transplant would live about another five years. 177 EQUAL RESPECT Adam, we are strongly inclined to think, has had a fairly good go at life, and although his prognosis is better than Eve's, the pull towards levelling the balance is obvious. The benefit to Eve is less but not negligible, and she has had, and will at the end of her life still have had, so much less than Adam has had already. But if intuitions are to guide us, we ought not to stop there. We ought to exercise our intuitions on more cases with just the features that in the last case seemed decisive against a purely utilitarian trade-off. Suppose Eve were a new-born baby with prospects of five years, and Adam six years old with prospects of ten. Now intuition strongly favours Adam. Or suppose Eve were six with prospects of five more years, and Adam twelve with prospects of ten. Intuition still favours Adam. What if Eve were fifteen with five more years, and Adam thirty with ten? Now intuition is probably just baffled. So, if intuitions are to guide us and, contrary to common practice, we consult a wide enough selection of them, they do not clearly support the model of same-level principles either. Faced with a puzzling set of intuitions, what we should do is to probe more deeply and try to make sense of them. Why do our intuitions---not, of course, that they are unanimous--- favour Eve, thirty-five years old, with prospects of five more years, over Adam, seventy, with prospects of ten? We tend to assume, in concocting examples like this, that the number of years may be taken as a measure of well-being. A prospect of ten more years, we seem to be assuming, is roughly twice as good as the prospect of five. But, on reflection, that is far from clear. Many central prudential values have to do with what gives a life point or substance. One's projects, what one accomplishes in life, are clearly central among them. Certainly not all years of one's life matter equally to the overall quality of one's life. Adam, at seventy, has probably largely made what he can of his life while Eve, at thirty-five, is probably just in the critical period. It is not that a seventy-year-old cannot still be in, or indeed just arriving at, this critical period, or that a thirty-five-year-old cannot but be in it. Individuals vary in this way, as in most others. But, as a general thing, years at the end of a normal span of life and, even more, years at the start, are not as crucial to the success of a life as years in the 178 MORAL IMPORTANCE middle. And intuitions conform largely to the general thing. So in this case it is doubtful that Adam's ten years weigh as heavily, even in terms of prudential value, as Eve's five. This variation in the value of years of life also explains the other cases. If Eve is a new-born baby or a child of five and Adam is not only older but with twice the prospect of survival, then Eve's years, even though so much less than Adam has already had, do not (at least as a general thing) have the prudential weight that Adam's years have. If Eve is fifteen with five more years and Adam thirty with ten, the prudential calculations are far harder to do, and so it is not surprising that intuition is far more pushed to come up with a verdict. I think that we are put off the mark in thinking about all of these cases by the incorrect totting-up conception of well- being. Under the sway of that conception, we think of overall well-being as a sum of several small values (in this case the number of years of life) each representing the same quantity of well-being. But this ignores how central global judgments are to the conception of well-being and to its measurement. The number of years in these examples do not represent amounts of well-being. Indeed, what we need to do is to take a case where they are more likely to do so. Suppose Eve were thirty with the prospect of five more years, and Adam thirty-five with the prospect of ten. Now perhaps intuition favours, as does almost universal medical practice, Adam. The result is scarcely conclusive, even within the severe limits of appeals to intui- tions. But once the amounts of well-being at stake are identified accurately, their maximization seems to be playing a very large role. It is easy---altogether too easy---to juggle numbers and produce examples that seem to support the model of same- level principles. What is far harder is to produce examples in which the numbers can be taken seriously as representations of well-being. Suppose we could bring about this outcome: (1)Adam Eve 5050 To isolate issues, let us say that we, who can influence the outcomes, are not Adam's and Eve's parents, are not even 179 EQUAL RESPECT known to them, and that neither of them is aware of what happens to the other. If we can juggle numbers ad lib, then we could get outcomes: (2Adam Eve 0100 and: (3)Adam Eve 0101 On the maximizing model outcome (2) is not better than outcome (1), though (1) has the distinct merit of treating Adam and Eve equally. But then, if equality makes (1) better than (2), why should we ever accept the maximizing model? It is not at all clear what that model would say about the choice between (1) and (2); it might be interpreted as making them morally equivalent or merely as silent on the matter. If the latter, there might even be a case for adding to the maximiz- ing model the tie-breaking rule that when the total is the same the (more) equal outcome is to be preferred. Perhaps we should give equality at least that much weight. This possible development of the notion of equal respect does not seem to me unimportant; indeed, it takes on, I think, great importance in another context, which I shall come to in the next chapter.51 But it is not what we are interested in now. Our interest now is whether, if we give equality enough weight, as most of us would, so that we rank (1) over (2), we could give it no weight when the tiniest increase in the total appears on the scene, as in (3). Most persons find it deeply counter-intuitive to rank (3) over (1). And that seems to be as clear a demonstration as we need of the superiority of the model of same-level principles over the simple maximizing model. But can the figures in (1) and (3) be taken seriously? They ignore at least two features of reality, one well-known and the other not. The well-known one is diminishing marginal utility. If the zeros in outcomes (2) and (3) represent a life of no value at all (and this suggestion gives the example a lot of its rhetorical force), then it is most unlikely that Adam and Eve, 180 MORAL IMPORTANCE both capable of a good life in outcome (1), would, by a redistribution of the same resources, be capable of the outcomes in (2) and (3). The less well-known one is incommen- surability. There are increases in well-being---clear, undeniable ones---that cannot be got on to the same scale with certain other values; no amount of one kind of positive value may equal a certain amount of another. That Eve might go from 100 in outcome (2) to 101 in outcome (3)---some tiny but undeniable improvement---does not mean that the increase belongs on the same scale of well-being as the measurement of the values of Adam's and Eve's whole lives.52 The fact, say, that Eve finds a way in outcome (3) to get one more good meal out of the distribution than she could in outcome (2) gives her something of positive value. But one good meal, more or less, does not affect the quality of a life as a whole. There are losses in certain values (Adam's?) that cannot be made up by certain gains in other values (Eve's?). The judgments that most matter in all of these cases are global judgments about the value of whole lives. That fact also provides one answer to the charge that the maximizing model ignores the separateness of persons. On the totting-up conception of well-being, overall judgments of value are reached by adding many small values. And so we think that there is no real difference between a case in which I endure a hard period of struggle and deprivation now to accomplish something later and a case in which you endure it so that I can accomplish it. But they clearly are morally different. In the first, intrapersonal, comparison my present loss is compensated by my gaining later; the compensation goes on within the boundaries of a single life. In the second, interper- sonal, case, that sort of compensation is impossible. However, that is not the only morally acceptable sort of compensation; as we have seen, to think so is to accept the over-strict constraint of contractualism. Still, these two cases are not morally equivalent. We have to take your life and mine as wholes. The totting-up conception is not the right one. My life is better for enduring the hardship in order to bring off the accomplishment. That is a global assessment of one life. But that I can accomplish the same if you suffer the hardship tells us nothing about your life as a whole. We need another global assessment for that. And to make it we should have to look at 181 EQUAL RESPECT what the deprivation does to your life, considering your life in all its other respects too. Just by knowing that a certain gain and loss would be justified in my life one does not know that the gain and loss divided between our two lives would also be justified. The notion of total well-being does not work like that, and a plausible prudential value theory would show why: to make the interpersonal judgment we should have to judge how good each life is, and we clearly lack the informa- tion for that. That is not to say that one person's deprivation could never be justified by another person's accomplishment. No one thinks that. It is just that justification has to be constructed out of materials that we have not yet got. What we must do is to find cases in which the numbers can be taken seriously as indicators of well-being. Suppose that we could bring about either outcome (1) or: (4)Adam Eve 4556 It is possible to see how we might decide on such global values. We would use the measure at work in strategic prudential deliberation. We would estimate how much we should value Adam's or Eve's life in outcome (1). We would do the same for outcome (4). That is, we would rank the lives; we would decide that Eve's life in (4) was better than either life in (1), which in turn was better than Adam's life in (4). And we could, at least sometimes, introduce some cardinality; we could say that the gap between either in (1) and Eve in (4) was greater than the gap between Adam in (4) and either in (1). Since all of these judgments can coherently be made just by appeal to strategic prudential deliberation, the model of same-level principles can be coherently stated. It is the view that the moral turning-point is not the prudential turning- point. That is, it says that for a deviation from equality in (1) to be morally justified we should need more than just the simple excess of gain over loss that we have in (4). We should need much greater gain---say, the one in: (5)Adam Eve 4580 182 MORAL IMPORTANCE But where is the moral turning-point? The model of same-level principles cannot stop at saying only that we need "a great excess'. It has to go on to say when the excess is great enough. If the moral turning-point is not the prudential turning-point, how do we find it? Suppose that this is the story. With our resources we could educate both Adam and Eve so that each would have an interesting, worthwhile career---outcome (1). Or we could, because Eve has excep- tional musical talent, concentrate resources more on her; Adam would stop his education earlier and so be in the running for less interesting jobs, but Eve would be (let us make the excess really great) a new Mozart---outcome (5). I think that, if outcome (5) were a possibility, nearly all of us would opt for it---even just for Eve's sake, let alone for the benefit to mankind of having a new Mozart. But surely it does not take as great a bonus as a new Mozart to justify a deviation from equality. A new Brahms would do. Or a Richard Strauss. Or perhaps even a Franz Leha$0r. But then where does the turning-point come? It would be unreasonable to expect great precision in the answer, but still we need an answer. Even parents, who have many more reasons to keep to equal treatment of their children than we detached observers have with Adam and Eve, often think that it is justified to concentrate resources on a talented child if the child has prospects of ending up, say, as violin in a good quartet. No doubt, we should not think that it would be enough merely for Eve to get occasional fun from playing her violin later on in life. Occasional fun, like the occasional good meal, is probably not the right kind of value to change the global judgment about the prudential value of a whole life. To reach the moral turning-point, we should need something more. We should need, for instance, a change in the order of magnitude of what Eve could accomplish with her life. Becoming a member of a good string quartet seems the right kind of change. But this is also the kind of change we should need to reach the prudential turning-point, too. The obscurity of both the prudential and moral turning- points makes conclusions difficult. But two conclusions seem to me plausible. First, the moral turning-point is at least not far from the prudential turning-point. Second, there are no 183 EQUAL RESPECT obvious independent criteria for where the moral turning- point is. These two conclusions suggest that the model of same-level principles is also too strict. It puts forward the requirement of a "great' margin of gain over loss, provides no clear criterion for "great', and once we have a plausible conception of well-being we find it hard to distinguish the margin required from a simple margin of gain over loss. 4.The two views compared Let me sum up what I have to say about utilitarian and contractualist conceptions of equal respect, turning now from particular versions of contractualism to contractualism in general. In sharp contrast to the uninvolved, God-like gaze of the Ideal Observer, the Ideal Contractor is meant to be involved, but the constraints on him, in any version, strongly incline him to see the world through the eyes of the worst off. But this gives us the dictatorship of, or at least some strong dominance by, the worst off. Why think that this is the way to see things morally? It seems to suffer from distortion by not being uninvolved enough. It distorts moral judgment in areas outside justice, and it is too strict even as a criterion of justice. It is because the principles that come out of contractualism give such strong protections that they so clearly meet the Requirement of Psychological Realism. But the attraction they get from this may not go deep. That every contractor has to be got to agree means that contractualism can advertise itself as "a genuine reconciliation of interests'.53 It would be a great comfort to know that the demands of morality will command universal assent, constrained only by ignorance or some fairly narrow conception of rationality. But the argu- ments so far suggest that it is unwarranted optimism to think that they will. Contractualism may look as if it has its roots in firm psychological ground, but no theory is well-grounded if, in the end, it fails to produce the goods. The failing in Rawls' version of contractualism is not corrected in later versions and, what is more important, seems not correctable inside a contractualist framework at all. It is not easy to pin down the essence of contractualism, but I think 184 MORAL IMPORTANCE it is this: an agreement on what are to be moral rules coupled with certain weak constraints on the parties to it. The constraints may vary from version to version but may not be too strong. They are meant to have minimal moral content, just enough to account for equal respect. So they must include such things as the parties' having to be motivated finally to reach agreement, or their being ignorant in certain ways, or their employing fairly narrow forms of rationality such as prudence or principles of bargaining. But they may not include specific moral constraints or a rich account of moral reasons for action, because this device is meant to generate them. But they must include enough to capture what we now mean by a contractualist account of equal respect: that is, they have to include something to bear the moral weight of the separateness of persons, because that has now become defini- tionally imbedded in the term "contractualism'. Since "separate- ness of persons' has no content except for the negative point that the utilitarian standard for trade-offs is wrong, contract- ualism cannot yield utilitarianism. Therefore, it cannot, even though this could be called "contractualism' on some looser definition, require merely that the parties agree, ignoring their own eventual positions yet knowing enough to reckon reliably that they have an equal chance of being in each position,54 because that yields average utilitarianism. So "contractualism', as it is now used, needs self-interest in a form that prevents the sacrifices allowed by utilitarian trade-offs. And even if, to overcome the vagueness of this account of equal respect, we were not too strict about keeping the moral content to an absolute minimum but helped ourselves to some of the principles that supposedly "follow' from it as fixing it further (so treating the account of impartiality and some of its main "derivations' as a single package), we should still have to have this strong form of self-interest present; it is what gives moral weight to the separateness of persons. But then contractualism never adequately faces up to the threat to morality that self-interest poses; it never makes the decisive break with self-interest that morality seems, in the end, to be. Meeting the Requirement of Psychological Realism is only a necessary condition. Still, it is necessary, and the utilitarian conception of equal respect seems to fail it. There is a distant 185 EQUAL RESPECT but not unreasonable hope that if we build a fairly prosperous society, ordered by just laws, peopled by citizens with a secure sense of their own worth, then envy and rancour will disappear, and tendencies to opt out of or to cheat the system will wane.55 But utilitarianism, since it demands great sacri- fices, needs saints, and no happy growth of society is going to work that miracle. That is why John Mackie calls utilitarianism "the ethics of fantasy'.56 But I wonder whether that is not largely a misunderstanding. Of course, not many of us will behave like Captain Oates;57 few people are going to agree to their own sacrifice. But this fixation on agreement is the contractualist's, not the utilitarian's, and it is just what produces the great distortion in the contractualist's outlook. What it is psychologically impossible for me to agree to do may still be morally justified for you to do to me. And the world does not generally ask for sacrifices; it exacts them. Surgeons decide who gets the transplant. Governments decide how much will go on health services or road safety. One might not cheerfully give to charities, yet still might fairly happily vote for political programmes that force one to do more or less the same through taxation. Anyway, if morality is demanding, it is demanding. The Requirement of Psychological Realism does not require that morality be comfortable, merely that we be willing to work at it. So I do not think that the Requirement of Psychological Realism ranks the utilitarian and contractualist conceptions of equal respect. They both pass. What would rank them is the adequacy of their rules governing trade-offs. We have rather stark versions of the principle of utility generating restrictions on trade-offs that are too loose. We therefore devise rather stark anti-utilitarian restrictions, which are attractive at first sight (as indeed they should be, since they are largely shaped to look good up against utilitarian looseness), but then at second sight turn out to be too strict. That, I believe, is not too bad a summary of the present state of moral and political philosophy: there are two main competitors, one too loose, the other too strict. We have either to tighten the former or to loosen the latter. 186 MORAL IMPORTANCE 5.A view of equal respect that allows some partiality How much out of my own slant on things must I lift myself in order to rise to the moral point of view? Utilitarians say sternly that I must not look through the eyes of any interested party; contractualists let me look at things interestedly, while using constraints that mean that I must look through the eyes of the worst off. But each of us, in looking out at the world through his own eyes, finds natural and proper objects of regard, respect, and commitment; our attachments---not just to persons but also to causes, standards, and institutions--- express much of what we value. Is this entirely to be left out of the moral point of view? And if it is, what mutilated rump of a moral agent is going to be left? Anyway, do we not deep down, despite the cant we all find it easy to fall into in theorizing, think that the partiality of some of these particular commitments is at the heart of the moral point of view? Do we not think that it is all right to care more for family than for just anybody, and not because some fully impersonal moral perspective finds itself able to grant us the dispensation but because the personal perspective is there, right at the base of morality, in the moral point of view itself? These are worries about any rigorously impartial form of equal respect, including both utilitarian and contractualist. However, the rationale of this more personal view of equal respect cannot be that morality reduces to personal aims. As we have already seen, such a reduction would merely solve the alienation problem by ignoring the authority problem; per- sonal aims are not always aimed at what is valuable. Any rationale will have to show both how to generate these more personal forms of obligation and how to link them to values. A more attractive way is to say that there are not only personal and impersonal obligations but also, in parallel, personal and impersonal values.58 An impersonal obligation is one that applies to all agents---such as certain duties of help and of avoiding harm, or the utilitarians' single super-duty of maximizing well-being. A personal obligation is one that arises out of an agent's standing in a special role relation to others---such as parent or friend or promisor or citizen. And one can see this distinction underpinned by a distinction 187 EQUAL RESPECT between personal and impersonal values. From our personal point of view, we might think, we fix on many proper objects of attachment, which express much of what we value. Indeed, some values, we might say, can be seen only from a personal perspective. It is not that there are no impersonal values; on the contrary, some things are valuable regardless of personal perspective. For example, pleasure and relief from pain look valuable to any properly functioning human eyes. But other things are valuable only as seen from the point of view of individual tastes and aims and commitments---for instance, rock-climbing or playing the piano well. So, on the face of it, there are two kinds of value, personal and impersonal, and an account of the moral point of view cannot forget either one. There is clearly some sort of useful distinction to be drawn between "personal' and "impersonal' values. The proposal a moment ago was that "personal' values are tied to aims that vary from person to person. But it is not clear in what sense there are such things. For me to see something as a value, from any perspective at all, requires my being able to see it against a certain background. The background must include general human aims; my own individual aims are not enough. This is not at all to opt for an objective approach to values; it is to state what is required for any talk about values to make sense. Let me take prudential values as an example. Certainly some values can be called "personal' with no problem. They are not widespread; they are tied to special, even rare, sorts of character. For instance, a few people find rock-climbing an achievement, and so they value it highly; more people find it terrifying and do not value it at all. But these two groups are not disagreeing in their values in any interesting way. Nearly all of us value challenge, excitement, and accomplishment, and none of us values simply being terrified. It is just that we differ in what we find challenging or terrifying. But we all have to be able to connect what we value to some characteriza- tion that makes it generally intelligible as life-enhancing. The mere fact that the object I aim at is aimed at by me is not enough. A child might aim at not stepping on any crack in the pavement, but that is not enough, either for us or for him, to decide that carrying it off is valuable to him. Of course, he might be unhappy if he does not carry it off; he might have 188 MORAL IMPORTANCE fun bringing it off. But happiness and fun are impersonal values. We quickly start reaching for impersonal values in cases like this, because simply achieving the object aimed at does not make his life any better. Indeed, some aims, if brought off, make one's life worse. The relation between desire and prudential value is complicated, but, whatever the full story, nothing is made valuable just by being aimed at. And what more is needed---the characterization that makes it generally intelligible that the object aimed at makes life better---makes all prudential values in one way impersonal. That it would make his life worse is, like that it would cause him pain, an impersonal reason not to do it. What I see simply through my own eyes and cannot see through other eyes cannot count as a value. Nor, of course, can it be a reason for admitting partiality into the moral point of view. What we very much need, therefore, is an account of the complex ways in which values are not personal (do not depend upon the varying desires of individuals) and the ways in which individual differences affect values. I said something about this earlier.59 The analogy with secondary qualities does not seem to me really to help here. Where it fits best (but how well I shall leave open) is in an account of values generally: values depend in one way or other on the human way of seeing the world.60 But what we are after now is an explanation of something else: a difference in kinds of value, personal and impersonal.61 Personal values, in the way we are now using the term, do not arise out of the human way of seeing things. All values do that. Personal values arise out of varying individual ways of seeing things. The analogy is not with, say, colour perception but more with (but still very little with) the difference between normal and deviant colour perception. None of this, though, helps explain the ways in which values are, and are not, personal. Enjoyment is a prudential value. It is valuable to anyone. But the particular ways in which persons do enjoy themselves, even can enjoy themselves, of course vary a lot. And it can be reasonable for a person to forgo enjoyments to devote himself to some all-absorbing project; there is no ideal balance of prudential values that means that enjoyment must figure to this or that degree, or to any degree, in a good life. Prudential values, as described at 189 EQUAL RESPECT some fairly general level of classification, are valuable to anyone; individual differences matter, not to what will appear in a profile of general prudential values, but to how, or how much, or whether a particular person can realize one or other of these values. For instance, accomplishment is one general value. I may want to become an accomplished pianist, and that specific accomplishment may become very valuable to my life overall, while not being of the slightest interest to the rest of you. But this is a matter not of our having different ways of experiencing the same determinate object (on the model of variations in colour vision) but of our having different determinate forms of the determinable value of accomplish- ment. In addition, the general value of accomplishment goes on imposing criteria on candidate determinations. I cannot, just by having it as my aim, turn anything I do on the piano (say, setting a world record for the fastest playing of the black keys from top to bottom) into something valuable---a "per- sonal' value. Rapid playing of the black keys, like navigating around the cracks in the pavement, takes more than being aimed at to be a value. In any case, we may not move from the existence of personal values, in the sense above, to the existence of personal moral obligations. It is true that deep personal relations are an important value, and that for different ones of us the value will be realized in relationships to different particular persons. But from the fact that it is those, and only those, particular relationships that each values, it does not follow that each has a moral obligation only to, or especially to, those persons. That may indeed be so, but not simply because the value is, in that sense, personal. The chain of reasoning, if there is one, will have to be longer than that. There may, none the less, be one. And we can see the rough makings of at least a link or two of the chain. It is still plausible to think that moral obligations will rest, in a way that has not yet been spelt out, on what an agent does, or should, find valuable. Since we all admit that there are values that are in some sense personal, some moral obligations may be too. For instance, since the value of deep personal relations can be realized only in particular ones, if any obligation is to grow out of this value it may turn into an obligation not to 190 MORAL IMPORTANCE persons generally but to certain particular persons. That there is no short-cut does not mean that there is no longer route to this conclusion. Obligations may then turn out to fit the human psyche more comfortably than they promised to do with, say, utilitarianism. And they hold out hopes of eventually, in aggregate, constituting a set of demands rather like the law: they may demand specific limited forms of positive action to identifiable and usually not too distant persons, and prohibit limited forms of negative behaviour, often to the whole of humanity. The whole set, one can reasonably hope, will turn out to be something that, like the law, we can live within. The familiar yield of this line of thought is, roughly speaking, common-sense morality. There will be such a thing as being morally quits, free to relax and go about one's own business. The moral point of view becomes in this way, the view of an individual agent, who forms intentions, acts, and can be held responsible for what he does. Then we might find, in the idea of an agent-centred point of view, scope for making an agent's intentional action the centre of moral assessment, thus demot- ing outcomes (even forseen outcomes) to which the action is causally connected to a peripheral place. And we might then also be able to give more weight to distinctions between acting and omitting, and between obligation and supererogation, than either utilitarians or contractualists can.62 If all of this can be carried off, it will deliver a deontological conception of equal respect. It would not be the only form the deontological view could take, but since it is so promising in coming to terms with the Requirement of Psychological Realism, it is an important one. But the number of missing links in this chain of reasoning is the great worry about it. By tailoring personal obligations to fit typical personal aims, it too solves the alienation problem without offering any clear solution to the authority problem. Where are these personal obligations coming from? Not from the Requirement of Psychological Realism, because that can be met in many ways. Nor from the mere existence of a distinction between personal and impersonal values either.63 Impersonal values such as the well-being that both utilitarians and contractualists appeal to are clearly valuable, but the obligations connected to them seem to spread too widely and become too demanding. This 191 EQUAL RESPECT sort of deontology generates obligations that do not spread too widely or become too demanding but are not founded in anything comparably obvious. By trimming obligations to fit personal aims one can get them inside human motivation. But it is likely that everyone will face moments when even this personal sort of deontology makes a moral demand on him that conflicts with his actual motivation. Then he will want to know what authority these obligations have, why they should move him. That is the trouble: since it is so hard to see where these personal obligations are coming from, it is very hard to accept that they constitute reasons for action, let alone reasons strong enough to outweigh one's actual present motivation.64 The answer cannot be: These are reasons because they fit pretty well what anyway persons generally desire. It cannot even be: Because they coincide with what is prudentially valuable to you. Whatever the Requirement of Psychological Realism amounts to it cannot be anything as crude as that. That is not fatal to this sort of deontological view. It just shows that it too has to solve the authority problem. I shall be coming back to deontological views later.65 All of these views about equal respect---utilitarian, contrac- tualist, and deontological---are views about how much one person's interests may be sacrificed to the interests of others. They are all views about fairness and justice. So what is needed is an account of equal respect that leads to satisfactory answers to that whole range of questions. There are no short cuts. General considerations about value do not supply one; moral psychology does not either. To think that they do is grossly to underestimate the scale of the problem. The only way to establish anything of interest is to produce a satisfactory substantive account of fairness and, in the end, of morality itself. X FAIRNESS 1.Two problems: fairness and the breadth of the moral outlook EACH of us has desires that he can satisfy only along with other persons. I cannot rendezvous with you, unless we both go to the same place. We cannot free the car from the snow, unless we both push at the same time. This sort of case does not often raise problems. Since, typically, each party has the same aim, agreements or conventions arise easily, often spontaneously, to bring off the benefit of the co-ordination. But there are also cases where each of us has an aim that he can indeed bring off on his own, but where the results of his doing so are less good than if he brings it off together with the rest. I want to be prosperous, but I would prosper more by working with the rest of you. But what each puts in and each takes out can vary enormously, and agreements and conventions do not typically arise so easily. These cases raise some of the hardest problems about fairness---problems of co-operation. The most famous example of a problem of co-operation is the Prisoner's Dilemma.1 But there is another overlapping sort. Sometimes what matters are cumulative effects of action, effects which singly make only a small, often imperceptible difference, and occasionally no difference at all, but together make a great difference. If one person living in the centre of my town burns logs on his fire, the deposit that the smoke makes on the buildings will be imperceptible, and the rain will wash it away. But if enough people do, the buildings will be permanently damaged and everyone (even the ones who have enjoyed a fire) will, on balance, be worse off. This category only overlaps that of Prisoner's Dilemmas, because cases of this type can involve only one person. If I once cut across your lawn, no damage is done; if I do it often enough, I wear a path. The trouble in these cases, whether involving one or many persons, arises simply from the fact that each consequence 193 FAIRNESS on its own is slight or, with time, entirely reversed. Whatever other persons living in the centre of my town do, my burning a log fire will make no perceptible difference. Whatever I do at other times, my now cutting across your lawn will make no perceptible difference. Co-operation raises many problems, including some of the most urgent political problems that we face. I want to talk about two central moral problems. Co-operation raises problems of fairness. What should the terms of co-operation be? If I voluntarily benefit from co-operation, should I not myself co-operate? And should I not even if nothing bad would happen if I did not? And co-operation raises problems of moral outlook. How wide does our perspective have to be to take in everything that matters morally? The problem of outlook is distinct from the problem of fairness, because although it arises in cases of fairness, it also crops up in cases where fairness does not. If a moral theory tells me to look only at the results of the act that I am contemplating and to compare them to the results of the alternatives open to me here and now, I shall miss the fact that acts can be parts of sets, of practices, of activities. What matters is the result not just of my now cutting across your lawn but of my making a habit of it. These are difficult problems, one way or another, for all moral theories. In the case of utilitarianism, for instance, the problem of moral outlook arises like this. If no one in my town burns logs, then I do no harm by doing so myself and, what is more, I get the benefit of a log fire. If everyone does, the same is true. Whatever number between all and none burn logs, the same is true. So utility, it seems, will be maximized if I do.2 And the problem of fairness arises like this. Suppose I know that virtually no one else will burn logs, although they would like to. Then it will produce no harm and some benefit if I do. But then I am just free-riding, exploiting other persons' restraint. Recent reaction to these well-known difficulties for utilitar- ianism is itself curious. Contractualists complain, with some cause, about utilitarians' reducing all morality to benevolence. But then contractualists turn round and reduce all morality, with no more plausibility, to fairness. It is reasonable, in light of the trouble utilitarianism runs into over fairness, to think 194 MORAL IMPORTANCE that a solution to problems of co-operation will have to occupy a central place in morality. Yet it distorts morality to make fairness the whole, or even the heart, of morality. For one thing, since contractualists concentrate on cases of co- operation, they naturally look for moral principles in an agreement between the parties who have to co-operate, but, as we have already seen, it distorts principles governing trade-offs to say that they have to emerge from agreement; it makes them much too strict. For another thing, fairness has trouble explaining benevolence. It is natural to look for the principles that should govern co-operation in what free and rational persons would agree to. We must expect to have to agree to some constraint, if we are to have a hope of benefiting from the constraint of others. But it is an unnecessarily roundabout explanation of why, say, cruelty to children or to animals is wrong. The constraint against cruelty has moral force not because it would emerge from an agreement; it should emerge from an agreement because it already has moral force.3 What is more, fairness has trouble with sorts of obligations it ought to find easier. Should very long-term promises be binding, if entered into in good faith and if circumstances have not changed? Suppose a young bride promises her husband that, should he die first, she will never remarry.4 Then suppose that, years later, with her husband dead, she deeply regrets that promise of her distant, youthful self, and wants to remarry. Is she obliged to keep her promise? How would it help to appeal to the rules that parties to co-operation would agree to? They could shape their institution of promising so that such very long-term promises, without legal standing, were binding or that they were not. But the benefits of co-operation would not be threatened by adopting either institution. So, to decide between them, they would have to look outside the contracting situation. They would have to appeal to some richer, wider conception of morality than merely rightness as fairness. Fairness, as important as it is, is neither the whole of morality, nor the device that generates the whole of morality. Why have contractualists been tempted to make it the whole, or the heart, of morality? They seem to have confused ---using the terms in the somewhat special sense I am attaching 195 FAIRNESS to them---equal respect with fairness, the rules governing all moral action with the rules governing co-operation. They seem to have regarded co-operation as an especially important case, as the place where the transition from the State of Nature to Society, from prudence to morality, takes place. But although there is a sense in which justice is the whole of morality and another sense in which it is only a part, it is equal respect that is plausibly thought the whole of morality; fairness is indeed just a part. 2.The consequentialist's problem of finding a broad enough outlook A moment ago, I mentioned two problems that would be my subject, first a problem about the breadth of the moral outlook, and second the problem of free-riders. The first is a problem for consequentialists. And most of us are consequentialists in the relevant sense; for most of us believe that some consequences matter morally in some way. But, then, how are we to take in all the consequences that matter? One source of trouble is a plausible, and sometimes it seems even inescapable, notion of agency---the notion of each of us as an isolated agent, cut off from other agents whose actions we are driven to treat as part of what is given, cut off even from our own long-term, life-structuring aims. Morality has dif- ferent branches; there is, most importantly, act-morality (assessment of the rightness and wrongness of acts), but there is also agent-morality (assessment of the goodness and badness of character and of responsibility for acts). But then what is the central concern of act-morality except evaluation of acts? And what are acts except what each of us can do here and now each time an opportunity to act arises? So an isolated agent, to the extent that consequences matter to him, reasons by listing his options occasion by occasion and calculating which provides the best outcome. One example of an isolated agent is an act utilitarian, at least on a tight interpretation common now;5 let me call him a "strict act utilitarian'. Such a utilitarian reviews his options case by case and plumps for the one that has, impartially considered, the best outcome. But to reason in that way does 196 MORAL IMPORTANCE not, in the end, produce the best outcome.6 For most of us, one of the things that gives most value to our lives is deep attachments to particular persons. And these attachments are incompatible with strict act utilitarian reasoning; they do not leave one able, occasion by occasion, having reviewed all actions within human capacity, to choose the best. Of course, a strict act utilitarian can include among his options becoming a person of deep attachments. But that is not a mere action of the sort that the isolated agent passes in review. It changes the whole source of his action. We are naturally inclined to form deep attachments, and if we know what makes life valuable we will refine and reinforce this already strong inclination. But such an inclination forms part of the long-term life-structuring features of one's character; and it undercuts the strict act utilitarian's whole mode of deliberation. The outlook adopted by the isolated agent could not be the right one, for any form of consequentialism. There seem to me to be three reasons for this. For one thing, it ignores too much of our real situation of choice. Typically, we are short of time, facts, and fellow- feeling. Therefore, what guides our choice must take these shortcomings into account. And what guides choice are not just principles and beliefs, but also feelings, emotions, habits, dispositions, and policies. Modern moral philosophy has largely ignored the towering importance of moral education: we are set on a moral course early in life, changing which, except in small ways, is not often or easily done. It is always tempting to think oneself or one's situation an exception, so moral dispositions, to do their job, must go fairly deep. And dispositions have to be shaped to what usually happens, not to the exceptional case. So a moral person will be moved by inclinations that have to be, for good reason, to some degree insensitive to exceptions. Another important feature of our real situation of choice is our lack of agree- ment on matters on which, for the survival of our life together, we have to agree. Sometimes we need agreed practices. Sometimes a deal must be struck, even at the cost of one's own moral beliefs. I may think that I have the better moral principle to apply in a certain case; others may think that they have and we cannot here agree to 197 FAIRNESS differ;7 so perhaps the best we can do is let the toss of a coin or a majority vote decide. These thoughts prompt some distinctions. We ought to distinguish the criteria of right and wrong from a moral decision procedure.8 The criteria are the features that make acts right or wrong, while a moral decision procedure is the thinking one should go through to decide what morally to do. Obviously we should want a person's moral thinking as much determined by consideration of right-making features as possible. But there are good reasons, some of which we have just seen, why moral thought cannot be so pure. Then it is worth distinguishing different sorts of moral decision proce- dures. There is a practical decision procedure: what should guide one in everyday life when one is short of time, facts, and fellow-feeling. Then there is a reflective decision procedure: what should guide one on the rare occasions when those shortcomings are absent. When they are absent, then, clearly, since the procedures designed for the typical case no longer fit, we should be dim not to look for new ones. I shall come back to this distinction in a moment, but it is clear that part of morality at least is a practical decision procedure. And as far as that part goes, the deliberation appropriate to it is nothing like the deliberation of the isolated agent. There is a second reason why the isolated agent's outlook cannot be right. The value theory at the base of consequential- ism rules it out. I discussed earlier9 ways in which prudence is thought to, but does not, deliver important results for morality. Now I want to discuss one way it is not thought to, but does, deliver one. We naturally expect prudential value theory to be neutral as between different versions---say between act and rule utilitarianism---but I think that it is not so. This is because with the shift from a simple hedonistic account of prudential value to an informed-desire account, the focus of evaluation shifts from short to long periods of time, and in many cases to whole lives. For instance, a person may find that what would give his life value would be rich, loving relationships, or some measure of accomplishment, or living autonomously. Even valuable short-term pleasures have to be fitted into a fairly long-term context; a person would have to decide how much place to give to living for day-to-day 198 MORAL IMPORTANCE pleasures compared to other ways of living. Therefore the isolated agent does not properly identify the primary object of evaluation; it is the quality of lives, not what we do occasion by occasion, which a hedonism of short-term experiences fits better. On a plausible theory of prudential values, the important values are what give life point, structure, or weight. But then this must influence how moral thinking can be conducted. Anyone who centres his life on deep attachments to certain persons adopts a whole way of living. He makes commitments that themselves become a powerful source of action; they reduce psychological freedom. The same is true of a person who tries to lead a life that accomplishes something. He makes a commitment that changes his feelings and attitudes, that takes up his time and narrows his attention. These large-scale decisions are the ones that, for the most part, determine action, and they demote the sort of occasion- by-occasion decisions that the isolated agent goes in for to a modest corner of deliberation. There is a third reason why the isolated agent's outlook could not be right for any sort of consequentialism. It fails to see the outcome of acting in concert with others. If the isolated agent is a utilitarian living in my town, he will, following the pattern of deliberation we have already traced, conclude that he ought to burn a log fire. And if this is the model for moral deliberation, then ideally everyone would think the same, and the buildings will be badly damaged and everyone will be worse off. This tunnel vision clearly misses too much. Before it missed the importance of whole lives; now it misses the importance of whole groups. I have mentioned three reasons why a consequentialist would not adopt the isolated agent's outlook. Let me return for a moment to the distinction between a practical decision procedure and a reflective one. The first of the three reasons is that the isolated agent's outlook is not right for the practical decision procedure. That leaves it open whether it is right for the reflective procedure. However, the second and third arguments show that it is not right there either. When, very rarely, we are not short of time, facts, and fellow-feeling, we clearly should not use decision procedures made for normal times. Should we then use strict act utilitarian thinking? Well, 199 FAIRNESS "act utilitarian' is a loose term, too loose to make an answer obvious, but there is at least some reason to answer, No. With time, facts, and fellow-feeling, we can sit back and ask whether a certain rule of the practical decision procedure applies in a certain case, and if we decide that it does not, we can sometimes free ourselves from its grip. The rules of good citizenship, for instance, are like that; it does not take much effort to miss a vote, if one is convinced that it is all right to do so. One thing that makes it easy is that what many duties of citizenship require has value only as a means to other values, which sometimes are not at stake. Admittedly, the practices that form our practical decision procedure differ a lot in the strength of their grip on us. Still, some of the things we value are very different indeed from the rules of good citizenship. The central values of our lives always retain their value. Furthermore, they get a firm grip on our lives. For example, our deepest commitments to particular persons go so deep that we never recover our original freedom of action, if ever we had much. Some are genetically based; many that are learned still go deep. If the training starts early, we can to some degree overcome our nature. But if we see no reason to overcome it, if indeed we see every reason not to, then we allow it perma- nently to shape, and to limit, our action. We are no longer able to enter into, and exit from, the behaviour occasion by occasion. Then there is another kind of central prudential value. If making some sort of contribution to knowledge is a large part of what gives weight to my life, this goal would then take a lot of my time. I might, if I had become a different sort of person, have been caught up, say, in politics. Indeed, as it is, I still might, even as I actually turned out, promote well-being more, all things considered, by taking time off now and then to serve some good political cause. But I should first have to find out how good the cause is, and spend even more time discovering how to be effective in helping it, and much more time in doing the job. It is not that I have not, strictly speaking, got the time. And, of course, I ought not to sit quietly at my desk, if I could help avert some political disaster. But there comes a point where I draw a line, even knowing that I shall be missing chances to do better. But once the line is 200 MORAL IMPORTANCE drawn, my policy to get on with my work is not shaped by the normal shortcomings we have mentioned. The policy gets its point not because time is short to reach a decision, but because life is short. These points about the reflective decision procedure raise again the question of exactly how ideal it is thought of as being. My points all rest on the assumption that it is not so ideal that it imagines human nature away; a decision procedure is, I assume, a procedure that people would sometimes use. This means that we may imagine people with ample time to reflect, in possession of the relevant facts, and finding their self-interest quiet; because all of this sometimes happens. But it also means that we cannot just imagine people emotionally tied to others at one moment and detached at the next, or the human life-span other than it is, or a person in society untouched by whether co-operation occurs. Any decision procedure ideal in such an extreme sense would lose interest for us. So the practical decision procedure is this: I ignore case-by-case consequences and, in deciding what to do, consult the principles that express the habits, dispositions, and practices that have the best long-term consequences in the light of normal human shortcomings. The reflective procedure is this: when the shortcomings are absent, I assess the consequences occasion by occasion to determine whether a particular case is an exception to my general rules; I also assess consequences occasion by occasion at those early times when I choose the general features of my way of life; but then, having chosen them, I give up act-by-act deliberation in large parts of my life and settle what to do by appeal to my goals and commitments. As any reader familiar with the recent work of R. M. Hare will have noticed, this contrast between the practical and the reflective decision procedure is in some respects like his contrast between two levels of moral thinking, the intuitive and the critical.10 It is a mistake to regard these two procedures, or these two levels, as merely something to smarten up dowdy old utilitarianism. They are, as Hare claims, part of the structure of any moral thought. But my proposal is not identical to his. For one thing, I think that 201 FAIRNESS there are reasons for distinguishing more than two levels, as I shall explain shortly. Also, I should want to draw the lines between the levels somewhat differently. Indeed, there are many different forms that a multi-level structure can take. Clearly, much more must be said about, and in defence of, a multi-level structure; some of it I shall come to in this chapter, and some more of it in the next. One large difference between Hare's proposal and mine has to do with where act consequen- tialism fits into deliberation, and I should say something more in a general way about that.11 It seems that, although some thought that goes on in the reflective decision procedure is act consequentialist, some is not. Our commitments and life-structuring aims largely take away our freedom to deliberate occasion by occasion; this loss of freedom is not made appropriate by the normal shortcomings that shape the practical decision procedure, so it does not disappear when they do. I spoke earlier of strict act utilitarian- ism but it is hard to see how the reflective decision procedure could be thoroughly act utilitarian even in broader senses. Just as act-by-act deliberation is barred from the practical decision procedure, it is at times barred from the reflective procedure. If that is enough to say that the practical procedure is not act utilitarian, it is enough to say the same about parts of the reflective procedure. This conclusion does not mean that consequentialist deliberation is not, ultimately, at some particu- larly deep level, act by act. In a sense it has to be. One department of moral thought is act-morality. Its business is assessing behaviour, not exactly what people do but what they choose to do. What they do often comes not from choice but from habits, emotion, and so on. But we want to get at moments of choice, and in the case of emotions and habits the relevant choices lie back in such acts as those that formed the habits or did not curb the emotions.12 Act consequentialism is true simply in virtue of the proper object of assessment in this department of moral thought. But the choices that matter are often not only in the background, but far in the background. Ultimately a consequentialist has to be an act consequentialist, but it is often very ultimate indeed. Where act consequentialism has its unqualified place is not in the decision procedures we have spoken about but among the criteria of right and wrong. 202 MORAL IMPORTANCE To the extent that consequences determine right and wrong, what makes an act right is there being no other possible act with better consequences. But even this role for act consequen- tialism has to be taken in a way compatible with prudential value theory. And there the primary object of evaluation is, as we have seen, lives looked at in a long-term way. So, to the extent that consequentialism has a hand in assessing what people do, it has to look at how what they do affects the quality of life. And what a person does can affect the quality of lives in very many different ways: by the dispositions he encourages or discourages, by the goals he sets himself, by co-operating with others, by adopting conventions, and so on. These are all things that we choose to do. We do not just look from an act outwards, towards its further and further conse- quences; the prudential value theory shows that we should also look from the whole of life inwards, to find the acts that give it its general contours. This means abandoning the narrow perspective that sees acts as making ripples, some of which admittedly last a long while, but also as having to be repeated again and again, with the deliberation appropriate initially appropriate repeatedly. Some of our most important acts are not like that; they are important because they alter the character of deliberation; they relegate act-by-act deliberation to a relatively small segment of life. So, no consequentialist would want to adopt the local outlook of the isolated agent. It would certainly be wrong for his practical decision procedure; it would also be wrong for his reflective decision procedure. He would want to adopt a much more global outlook that took in not only the effects of individual acts but also ways of life, commitments, aims of life, and the cumulative effects of what groups do. This is still a consequentialist outlook, only not the isolated agent's. My talk of local and global outlooks touches on many issues in Derek Parfit's discussion of whether certain theories are self-defeating.13 But it seems to me that the framework of local and global outlooks is in some respects to be preferred to Parfit's own framework. It is not that it leads in the end to different conclusions from his but that it reaches them more directly. Parfit argues persuasively that someone adopting what he calls the Self-interest Theory, who thereby develops a 203 FAIRNESS disposition "never to be self-denying', will often make things worse for himself simply in self-interested terms.14 And he argues that a group adopting Consequentialism as a moral theory, the members of which thereby develop dispositions "always to do good', would also make things worse even in their own consequentialist terms.15 Both these theories are, he concludes, "indirectly self-defeating'. They are because "It would make the outcome worse if we were always disposed to make the outcome best'.16 But what is it to be disposed to make the outcome best? It could be rather different things. It could be to be disposed always---on each occasion, taken occasion by occasion---to do what would make the outcome best. That is, it could be to apply one's standard of value, whatever it is, locally. Or it could be to be disposed always---consistently through one's life---to do what would make the outcome best, leaving it still to be decided whether one's standard is to be applied locally or globally. According to Parfit, belief in the Self-interest Theory or in Consequential- ism will be likely to produce certain dispositions (never to be self-denying, always to do good), which in turn may produce bad effects. But all that belief in either theory will produce is a disposition of the second sort, while to get the result that the disposition would make the outcome worse, and so the result that the theories are self-defeating, we need the first sort. More importantly, it would not be mere belief in these theories that produced bad effects in their own terms; it would be belief in them plus a decision to apply them locally.17 Yet whether we should approach self-interested or consequentialist deliberation locally or globally depends upon the content of prudential value theory, and the best prudential value theory is not neutral between local and global outlooks. The Self- interest Theory does not lead us to be "never self-denying', and Consequentialism does not lead us to be "do-gooders'. Neither is capable of leading anywhere until it incorporates a value theory and so makes the closely connected choice between local and global outlooks. Once the theories thus become capable of motion, they will in fact lead us somewhere rather different. Parfit sees the Self-interest Theory and Conse- quentialism coming to advise their own abandonment; they become to some large extent "self-effacing'.18 But the best 204 MORAL IMPORTANCE response to the real enough cases where applying the theories turns out to be self-defeating is not to stop believing the theory but to stop applying it tactically rather than strategically. Can one apply them strategically without schizophrenia? Would desiring the best yet having to turn one's face, occasion after occasion, against doing the best require a divided self: one self rational grand strategist, the other unthinkingly obedient lower ranks? Is it better simply to suppress one's original beliefs? But this exaggerates the psychological tensions. I do indeed have to turn my face against occasion-by-occasion maximization. I have to live with knowing that my policy, centred on maximizing well-being, is not maximizing it in one way: I am passing up chances locally to maximize. But I have the consolation that, looking at things overall, and accepting the limitations that reality imposes upon me, I am doing the best. I know that I am psychologically incapable of entering and exiting, at will, from policies, commitments, and attach- ments. I have to live with that, and much similar, knowledge. But it does not take a divided self to be able to.19 Once a consequentialist shifts to a global outlook, Prisoner's Dilemmas look different. Merely by thinking strategically, even without thinking morally, he might sometimes agree to accept a political solution to a Prisoner's Dilemma. For example, we might all agree on an inescapable system of taxation, or one with swingeing penalties for cheaters. But, as is well known, such political solutions do not work in many real cases in which defection cannot be made impossible or sanctions effective. Then we need one or other psychological solution. And we have one: we are imagining an agent who is impartially benevolent. But acceptance of a principle of impartial benevolence, like most moral solutions, gets us only so far. It is true that if both parties are impartially benevolent, in one way we have a complete solution. Each party will co-operate whether the other party co-operates or defects; in either event co-operation would maximize benefit impartially. Still, that does not get us far enough. Anyone who is impartially benevolent will want to bring about mutual co-operation; that is the most beneficial outcome of all. And if he does not know that the other party is impartially benevol- ent, then he will want to look for a way to ensure that the 205 FAIRNESS other party will still co-operate. If the other party is indeed impartially benevolent, then he too will look for a way to ensure co-operation. But there are often different ways to do this. And then their problem becomes one of co-ordination.20 In co-operation generally, even if everyone were impartially benevolent, and moreover thinking strategically, we should need both to co-ordinate action (for instance, if there were several equally plausible "best' options) and to be assured both that enough others are going to co-operate for co-operation in fact to have the best results, and also that not so many will co-operate that it would be better if some struck out on beneficial action of their own.21 There are many important, sometimes life-and-death questions about the emergence of co-operation, not all of which is it reasonable to expect moral philosophy to be able to answer. For instance, there is the question, How can an individual with certain endowments (for instance, with a global outlook and a belief in impartial benevolence) find reassurance about the motives of other persons, or manage to co-ordinate his actions with theirs? That is a complex question about how co-operation in fact gets off the ground; it is not a question for moral philosophy. Then there is the question, Does a person with only these endowments have reason to be assured and to co-ordinate his actions with others? But that question is ill-conceived. There is much more to human nature than strategic outlooks and moral principles, and a good deal of it comes into play in making co-operation likely and so rational. There is genetically based behaviour; there is a wide range of emotions; there are natural dispositions---for instance, a willingness to gamble. It is most unlikely that a key social institution such as punishment, say, came into existence through an agreement that required trustworthy character and rational assurance of each other's trustworthiness. It is much more likely to have grown from feelings of guilt and anger, and from dispositions of self-protection and retaliation. And there is no sharp divide between "political' and "psycho- logical' solutions to Prisoner's Dilemmas. Once key institutions have arisen, their rules, conventions, and sanctions produce practices and eventually habits and dispositions. Habits give more assurance about what others will do. The actual genesis 206 MORAL IMPORTANCE of the social institutions and individual attitudes that promote co-operation is bound to be highly complex. Many institutions and attitudes existed well before humans rose to the level of self-consciousness that allows moral reflection. Moral philo- sophers, solely with rationality and moral principles as their resources, could not be the architects of our social environment; in any case, our social environment is already in place and moral philosophers are at best handymen capable of piecemeal alteration. Still, there are questions that a moral philosopher must face. It would be proper to ask whether, taking for granted all the various ways in which co-operation actually gets going, morality can give a reason to join in beneficial co-operation. And it would be proper to ask if, when beneficial co-operation is already under way, a particular morality will recommend undermining it or not doing one's bit. To these questions, global consequentialism can often provide satisfactory answers. As we have seen, mere growth of altruism is sometimes not enough. But the global outlook would also tell us to do whatever is necessary---agreements, conven- tions, education, persuasion, punishments, forfeits---to hit on and sustain beneficial co-operation.22 Or, rather, since this sounds too much like the implausible picture of the moral philosopher as architect, it would give us reason to sustain and refine those forms of co-operation that we find already in place. A narrow consequentialist outlook gets one into troubles; shifting to a broader one gets one out of many. But how many? In particular, how much does it help with free-riders? 3.The free-rider problem and a minimal solution Broadening the consequentialist outlook, on its own, is no help with free-riders. If enough people follow the rule not to burn log fires, the buildings in town escape harm, and I should increase welfare, impartially considered, by making myself an exception. Perhaps, therefore, the fully moral outlook needs one more step back. Perhaps we should look at hypothetical as well as actual outcomes. "What if everyone burnt a log fire?' The reply, "Not everyone will', seems to miss the point. Of course, if they are decent sorts, they will not. But the point is that a few of us 207 FAIRNESS can indulge ourselves without damage only because the rest of you are denying yourselves. This line of argument obviously has some force. However, the move to hypothetical outcomes seems to drain consequentialism of its intuitive plausibility. What matters morally are actual changes to people's lives. If the Town Council came to believe that not everyone's constraint was necessary to avoid the bad outcome and hatched a workable scheme to let us take turns burning log fires, there would be no moral objection to it. "Not everyone will' seems not to miss the point. And this second line of argument has force too. These two arguments seem to me to make up a deeper dilemma than the Prisoner's. The pull of each is a cause---not the only one, but nowadays an important one---of the deep split between Kantians and utilitarians. What seems right in the first argument is that it is objectionable that my life can be made better with a log fire only because your life will be worse. I am being a parasite and so violating equal respect on some conception that we should not want to lose. And, in a way, what the rest of you actually do is morally irrelevant. It is beside the point that you will actually restrain yourselves. And that is because the point is my behaving like a parasite. But what is morally pertinent about my behaving like a parasite is not a hypothetical outcome either, what would happen if you all behaved differently from the way you will.23 What seems right in the second argument is that the outcomes here are all real enough: I actually get the cream of life because the rest of you are actually making do with the skimmed-milk. How much do we have to do to get fairness back into the picture? I cannot see how adopting a principle of maximizing well-being, even with a finely articulated multi- level structure that its global application could give, can get it back. And therefore I cannot see how utilitarianism, at least with the resources we normally associate with it, can get it back either. What I think we ought to do is to look for the very minimum resources needed to reintroduce fair- ness. A minimalist approach has not only elegance to recommend it, but also prudence. The greater the resources, 208 MORAL IMPORTANCE the more likely they are to produce trouble elsewhere in the system. One approach would be to take a hint from the fact that it is real, not hypothetical, outcomes that matter in free-rider cases and see how far we might get simply by developing or supplementing the principle of impartial maximization. It would be a relatively modest project. Our aim would be to find not answers to all moral questions, but merely a solution to the free-rider problem. The place to start, I think, is with that vague but fateful notion of equal respect. Every moral theory has the notion of equal respect at its heart: regarding each person as, in some sense, on an equal footing with every other one. Different moral theories parlay this vague notion into different concep- tions. Ideas such as the Ideal Observer or the Ideal Contractor specify the notion a little further, but then they too are very vague and allow quite different moral theories to be got out of them. And the moral theories are not simply derivations from these vague notions, because the notions are too vague to allow anything as tight as a derivation. Too vague, but not totally empty; although the moral theories that we end up with put content into all these notions, the notions themselves also do something towards shaping the theories. We move too quickly across this ground; there are positions that have not yet been properly explored. So let us set off from this vague basic idea of equal respect, which impartial maximizers parlay into a principle of trade- offs: maximizing well-being, counting everybody for one and nobody for more than one. What would such maximizers say, animated as they are by a certain conception of equal respect, about cases where maximization gives no result? Imagine a case where the welfare total stays the same and only chances of having it vary.24 Suppose, for instance, that a government finds itself able to act in either of two ways. If it acts in one way, then, depending upon matters outside its control, either one group will be helped or another group will be equally helped. If it acts in the other way, then the first group is bound to be helped as before, but the second group will definitely get nothing. What should the government do? Whatever it does, 209 FAIRNESS the total help it gives will be the same; so the welfare sum provides no ground for choice. Yet there does seem to be an obvious ground: the first way of acting gives each group a chance. What does an impartial maximizer say about this case? His principle of trade-offs (maximization) has no implications for it. The fault in that principle is not that it gives the wrong answer but that it gives none. What answer, then, should he give? That question comes down to another question, What deep, animating conception of equal respect do we think his principle holds? We have to look behind the familiar principle of trade-offs (maximization), and its constituent principle of equal weight (everybody to count for one), to the conception of equal respect behind them both. Suppose that you care equally about the welfare of your two children. You count each for one and neither for more than one; other things being equal you would have them at the same level of welfare; you would accept a loss of welfare for one of them as long as the gain to the other were greater. But then one day you find yourself confronted with a choice about which your view about trade-offs tells you nothing: the benefit will be the same whatever you do, the only choice being between directing it to only one of them or giving each a fifty per cent chance at it. From your moral views, as spelt out so far, it is unclear what you would do. There is no account of the principle of impartial maximization that provides reasons for doing either. You would have to think out your principles further. And here you might well reach different conclusions. You might conclude that your deep conception of equal respect can give no further guidance, that either some entirely new moral notion is needed or that the question is unanswer- able. Or you might conclude that your deep conception, when developed further, gives the answer that the options are morally equal. Or you might conclude that your deep conception, when developed, points towards giving each child an equal chance. The last seems to me a perfectly natural, perhaps the most natural, development. And there is this further explanation (I do not pretend that it is anything as conclusive as an argument). You as a parent would not be choosing between 210 MORAL IMPORTANCE acts that determine the substantive well-being of your children. You would be choosing simply between acts that affect their chances. But chances may affect one's well-being. If one child's welfare matters just as much as the other's, then it seems in the same spirit to say that when it comes to chances of welfare, the one child's matter as much as the other's too. Where the notion of "greatest sum' provides no criterion, it does not follow that an impartial maximizer has to be without one; there is no reason why his position cannot be taken to be rich enough to cover this case. So if an impartial maximizer parlays equal respect in this way, he will have, first, a principle of maximization and next, when maximization does not rank options, a principle of equal well-being and finally, when equalization does not rank options, a principle of equal chances at well-being. It is, as principles of equality go, a modest addition to the principle of maximization. Still, no doubt some persons will want to deny that impartial maximization can be "spelt out' to have even this modest egalitarian implication. What I want to say is that there is no good reason for their doing so. They might argue that if one adopts maximization as the principle governing trade-offs, then one accepts that, in a certain sense, individuals do not matter: the sum matters; which particular individuals get which part of it does not. But that individuals do not matter in this sense does not imply that they cannot matter in the sense that these further, tie-breaking egalitarian principles require. The sort of concern for individuals that is displayed in maximization is not inconsistent with the sort of concern displayed in this modest egalitarian addition. Indeed, the same deep consideration that generates impartial maximization can be seen as also generating these ways of filling its gaps.25 Yet if an impartial maximizer spells out his notion of equal respect in this way, what is it that he has added to his views? It might, of course, really be an entirely independent principle. The most powerful objection to impartial maximization is that it cannot account for fairness. So it is not surprising that its principle of trade-offs is silent on the sort of case that we have been looking at. But what I charitably call "spelling out' the deep maximizing conception of equal respect may look to others suspiciously like recognizing its inadequacy and adding 211 FAIRNESS to it an independent principle of fairness that will turn around and start qualifying the maximizing principle itself in cases of other sorts. But I do not think that this is so. The addition is made within the constraint of the principle of trade-offs: loss is justified whenever gain is greater. And although the deriva- tion is not especially tight, it is natural to think that each person's chance at welfare matters and matters equally, because of that deeper consideration that makes us think that each person's welfare matters and matters equally, and because of the causal connection between chance at welfare and welfare itself. The principle of impartial maximization plus the supplementary principle of equal chances constitute a combin- ation of views that is no less natural, and no more or less puzzling, than any other important view about equality. It is little assurance, for instance, to be able to get such a view out of a contractualist framework, because there are many such frameworks, yielding principles on trade-offs that range from the quite strict to the quite permissive. Contractualists have to choose one particular framework, and that choice is itself the choice of a particular conception of equal respect. Nor is my suggesting that maximizers develop their own principle in this way any more puzzling or in need of justification than making equal chance an independent principle in some Ross-like pluralist kit. This sort of pluralism is incomplete; it has still to explain how we are to trade-off, say, a principle of impartial maximization against a separate principle of equality. To explain that would simply be to spell out this deep notion of equal respect. The spelling out that a pluralist has to do is the same sort of project as the spelling out that I have just done---neither one more natural or theoretically deeper than the other, and neither one different in kind from the spelling out that a contractualist has to do. All such spellings-out are conducted in much the same way and uncomfortably close to present theoretical bedrock.26 However, I hope that my disclaimers about the tightness of the derivation do not suggest that my argument is identical in form to certain arguments for indirect utilitarianism: namely, start with (usually conventional) results that you want the principle to yield and then tinker with the principle till it does yield them. On the contrary, my argument moves from the 212 MORAL IMPORTANCE notion of equal weight in calculating trade-offs to a notion of equal chance; it tries to spell out in a natural way the deep notion of equal respect that underlies both.27 If the principle of impartial maximization can be spelt out in this way, then there is one implication to be faced. Utilitarianism, which is just the principle as the sole criterion of right and wrong, looks as if it might be similarly spelt out. But would we seriously contemplate giving the package of the principle of utility plus the two tie-breaking egalitarian sub-principles the name "utilitarianism'? What I want to propose is that the answer does not much matter, that there is nothing deep at issue, that we might say what we like. We might prefer to say that a utilitarian is one whose sole criterion is utility maximization and that where it did not rank options there is nothing to choose between them. The principle of utility is silent on what to do about equal maxima. We tend to say that there is nothing to choose between them because usually there is not---on any version of utilitarianism, the spelt-out version included. But there are rare cases where the equal maxima differ because in one of them persons are benefited or get equal chance at the benefit whereas in the other they do not; and these cases pose a fresh problem. What should a utilitarian say about them? He might well indeed say that there is nothing to choose between them. Or he might say, I think, that he believes that the notion of equal respect comes back into play again here and gives us something to choose between them. Would the second position be "utilitar- ianism'? If, as I think, there is no real case for saying that a principle of equal chances is contrary to the deep spirit of the principle of maximization, then it is hard to answer. It does now seem the sort of verbal question that needs legislation rather than argument. This spelling-out may seem unimportant.28 After all, it can help only where total welfare is the same whatever we do and it is merely chances that differ, and cases like this may seem of great rarity and little weight. But that is not really so. The spelling-out has implications for procedures. For instance, in all cases where benefit or burden is about the same for each member of a group but only some can or need to have it, it suggests that we ought to select randomly. Conscription is a 213 FAIRNESS good example. When the national, or other, interest points in no particular direction, then the chance of being selected should be made equal. And it should be made equal not because, say, any other procedure would cause resentment and unrest, although they matter too, but for a much more fundamental reason: each person's fate matters and matters equally and random selection is demanded simply by that. And a lot of cases of cumulative effects are also examples. If I were to burn a log fire, I would get about the same benefit as you would. So long as there is only one exception, whoever the exception is, the welfare sum will be the same; all that changes is the chance of having the benefit. But does the spelling-out help with fairness? To return to our point of departure, free-riding: How do we get fairness back into the picture? If I burn a log fire because I need it and the rest of you do not, then my benefit does not rest objectionably upon your burden. That is because the introduc- tion of needs also introduces equal respect: people's actual interests matter, and matter equally. But if I benefit because it is my turn, or because it is my name that came out of the hat, then my benefit also ceases to rest, in any objectionable sense, on your burden.29 Chance at an actual benefit also matters. If I simply award myself the benefit, then not everyone has an equal chance at it. If, on the other hand, it is simply my turn, then everyone will in time have his chance. If my name came out of the hat, then everyone has had a chance. These modal matters---neither actual consequences nor hypo- thetical ones---also count.30 I think that the principle of equal chances is also important in explaining what is objec- tionable in free-riding. To check free-riding, to get fairness back into the picture, we do not have to stop me from having the benefit while you suffer the burden. That still happens, but now unobjectionably because of the availability of one of a set of reasons for my getting the benefit: I needed it; I won the toss; it was my turn; etc. In other words, we need only show, by citing one of these reasons, that equal respect has not been violated. Equal weight in calculating different persons' interests would show it. Equal chance at a benefit would also show it. Do we need any- thing else? It seems not. We do not, for instance, need that 214 MORAL IMPORTANCE extra step backwards to hypothetical consequences: "What if everyone...?' This has consequences for both those who frame and those who follow rules. To allow no one to burn log fires, when some could without harm and there are feasible ways of letting them, is senselessly stern. Those who frame laws could, as we have seen, give everyone an equal chance to have a fire; or they could amend the rules so that fires could be burnt in public buildings on special occasions. Sometimes they will have to play safe, because they do not know how often the rule will be broken or because what is at stake is so important. Sometimes they will have come to the point where finer calculation costs more than it is worth. But they should get as close to the edge as possible. If they do, then those who follow the rule will not be left much room for man{ouvre. If the rule embodies the best estimate of compliance and of the importance of what is at stake, then generally I am in no position to conclude otherwise. That is why the practical decision procedure is so important in discussing fairness. Most of the rules, the breach of which is unfair, are rules of the practical decision procedure. They are framed for the usual context of choice. But when, unusually, the context changes, those rules clearly suit less. There cannot be any easy move from the rules of a practical to those of a reflective decision procedure. Do I have a duty to vote? Well, suppose I ask as a citizen of a large marginal state in the United States, where the chance of my vote's making a difference is about one in one hundred million and where the population is two hundred million.31 Suppose, also, that whether the better person wins matters; if he does, there will be some average net benefit to each citizen's life over a certain period. Then, I will have to reason roughly like this. There is only a very slight chance of my vote's making a difference. On the other hand, the expected benefit of my voting will be twice the average net benefit,32 less the costs of my voting. So unless my case is most unusual, this sum will turn out positive. In general, in situations of that sort, I ought to vote. One does not have to look beyond outcomes to find a reason to vote, or to justify a rule about voting. But if I ask as a member of a much smaller body, say my college's Governing 215 FAIRNESS Body, and it is a case where I know perfectly well how the others will vote and that my vote clearly will make no difference, then outcomes may give me no reason to vote. But there seems still to be a consideration of fairness: may I excuse myself from voting? May I, even if there are no other outcomes that matter? Why should I be the one to get the afternoon off? Perhaps I should toss a coin over lunch with you. It is easy to make what Derek Parfit calls mistakes in moral mathematics, such as ignoring small chances or small benefits.33 But the free-rider problem is not solved just by avoiding these mistakes.34 This takes us to the reflective decision procedure, where deliberation gets complicated. There is the consideration of fairness, which can go beyond actual consequences: for instance, who should get the chance? On the other hand, only actual outcomes matter; maybe it is not worth the bother of tossing a coin with you, and since my vote will make no difference I need not go. We have also to keep in mind that although actually we have only partial compliance to our rule the ideal would be full compliance: burdens, let us say, would be fairest and outcomes best if we all faithfully voted, and my voting, joined with other persons', now and later, might be a break for higher moral ground. And we have to recognize that our attachments and commitments and life-structuring aims limit the freedom of our occasion-by-occasion deliberation. So deliberation on the reflective level has many sides. I know that many persons will find this solution of the free-rider problem simply too weak.35 They will think that it still allows too many exceptions, and they will insist that the free-rider problem shows that we need other, tougher prin- ciples. I want now to explain why the tougher principles seem to me to be too tough. 4.The possibility of tougher, Kantian solutions Kant himself is not a consequentialist of the familiar modern variety. Consequences do enter his theory, but his test is not to determine first the consequences of everyone's doing such and such ("What if everyone ...?') and then their value ("It would be disastrous').36 The test looks elsewhere: in one formulation, 216 MORAL IMPORTANCE it asks each of us to see whether he can act on his proposed policy and simultaneously accept everyone's acting like that.37 The question is not Would it be disastrous? but Would a contradiction appear? But, clearly, the appearance of the sorts of contradiction that Kant has in mind will turn largely on how we describe a person's policy. What I propose doing may be "skipping votes', or it may be "skipping a vote when there is one spare', etc. How specific can my description get? If it cannot get as specific as we want (providing, of course, that it avoids any essential mention of particular persons, place, or times), and especially if it has to be kept at a fairly high level of generality, then perhaps one of the sorts of contradiction Kant has in mind will appear.38 Even so, however, the test would not account for fairness. If it is kept at a fairly high level of generality, then the test would not allow me, or anyone else, to miss the vote, no matter how the lucky person was chosen. It would not allow me, or anyone else, to burn a log fire, even if the City Council drew names out of a hat. It is not only senselessly stern; it does not get to grips with what fairness is. Perhaps, then, the account of a person's policy can be as specific as needed. It can mention anything morally relevant, and the fact that I do no harm by skipping the vote seems at last that.39 You and I know perfectly well how the vote will go: the motion that we support will pass by two votes, when a one vote margin is just as good. But I am quicker off the mark in the car park. "There is a vote spare, so I'm leaving', I say, and beetle off. You would not have minded a free afternoon either, but now you are stuck. My departure means that you cannot reason, If I skip the vote, no harm will be done. Yet, my behaviour, on a fully specific account of it, does not seem to lead to any contradiction. Still, what I do may be unfair. And not only does the test seem to allow the unfairness but also, worse, it does not locate its source. The intuitive thought behind this test is this: if it is morally all right for me to do it, then it is all right for everyone, similarly placed. Nearly all of us accept that; we disagree only over how to fix what it is to be "similarly placed'. But your situation might be like mine in all relevant respects---your capacity to enjoy an afternoon off just the same; your need for extra time just as great; your absence 217 FAIRNESS no more harmful than mine. What is important to fairness need not lie in any feature of you and me, nor in any of our circumstances. It lies in the availability of one of a set of reasons for one of us getting the benefit. If you and I toss for the free afternoon, or if I do indeed need it more, etc., my missing the vote is fair. If not, it is not. Indeed, that is why the case Kantians sometimes put up for banning consequences from mention in one's policy (not allowing one to state it as, for instance, "skipping a vote, when no harm will be done') is weak.40 They rightly say that each of us would be able to argue that his skipping the vote would not be harmful and so each of us would be an exception. Nor does it help, they correctly add, that in fact not everyone will, because the rest may be decent sorts paying their moral fare while I ride free. However, the most that this shows is that the mere fact that I would do no harm in skipping the vote does not of itself justify me in skipping it. Justification needs more. What more, however, need not be, as they conclude, that my policy, formulated without mention of consequences, passes the universality test. It could instead be the availability of one of the reasons that I just began listing. Suppose, then, we amend the test to accommodate this point. We now include how the benefit is distributed as part of the relevant circumstances. We might even find grounds in Kant's text for doing so. We could appeal to another and, according to Kant, equivalent formulation of the categorical imperative, the "respect for persons' formulation: persons should always be treated as ends, and never merely as means.41 And when I speed out of the car park, then whether or not I am conscious of using you as a means, I am none the less using your attendance in order to justify my freedom. I certainly give you no chance to exercise your rational will; I do not ask you, as a free and rational person, to consent. The "respect for persons' formulation brings us very close to modern contractualism; the test is less immediately Would a contradiction appear? than Would free and rational persons agree?42 And it seems at first sight as if this test is just what we need to generate the list of reasons that would justify exceptions. If you could do with a free afternoon just as much as I, then you are not likely to agree to my unilaterally 218 MORAL IMPORTANCE awarding it to myself. What we both could agree to, however, would be some procedure, such as taking turns or tossing coins. We are likely also to agree, on maximin grounds, to give place to greater need. What more would we give a place to? The answer depends upon how rich a notion of rationality we are using. If we use the spare notion of self-interested rationality from modern contractualism, the answer is: nothing more than I have already proposed for the list. If we use Kant's potentially richer notion of a rational will governed by the objective ends of life, the answer is: it is hard to say. What makes my skipping the vote fair is my having the right sort of reason for being an exception, and it is hard to see how the Kantian test, made richer by his teleological account of life, identifies which reasons are of the right sort. It would be able to if its key terms "categorical imperative', "rational will', and "respect for persons' had more content. It is true that it is a perfectly good thing to say of my giving you an equal chance, or of my weighing your needs impartially against mine, that I thereby show respect for your person. It is a good way to sum it all up. Still we have to know how to fix the content of this respect; we have to know which substantive reasons for my getting the benefit will make it fair and which will not. It is determining what constitutes fairness that puts content into the vague formula "respect for persons', not the formula "respect for persons' that discloses to us what constitutes fairness. The potentially richer, Kantian notion of rationality is so indeterminate that it does virtually no work. These are, in outline, three ways of understanding Kant's test. If one of them is indeed correct, then in a sense it does not matter which it is: none of them helps enough with the problem of fairness. What seems most promising is taking the "respect for persons' formulation in the direction that contract- ualists do. This might, depending upon whether the particular form of contractualism uses a narrower index of well-being, lead to slight differences from my proposal, but only slight ones. With something like health, say, it is hard to see how they could differ. For many diseases the optimum rate of inoculation is ninety per cent; inoculation beyond that brings no benefit to public health, and any inoculation imposes a real (small) risk on the child of serious illness. What rational agents 219 FAIRNESS would be likely to agree to, either directly or through the political structure, is inoculation to the safe threshold and fair procedure for excepting people. The contractualists' way of developing Kant's test produces a solution, but not the tough one that we set out looking for. And, as we have seen,43 it has the disadvantage of being in other ways too strict. 5.The solution of an agent-centred deontology The most interesting deontological solutions, I think, are anyway not the tough ones. They are the ones that have about the same practical results as my minimal solution but, many would say, back them up with better theory. Indeed, my minimal solution may seem not minimal enough. It uses a rather wide-ranging principle, the principle of equal chances, which applies to anyone whenever a tie between lexically prior principles, which also apply to anyone, needs breaking. This breadth creates trouble. Kurt Baier gives the following nice counter-example.44 Suppose I have two lottery tickets, each of which would win me a million. I believe that two charities, Oxfam and Care, would make equally good use of the million, though of course the good they do would go to different persons. Maximizing does not decide between them. But would it be wrong for me to give both tickets to one charity rather than one ticket to each? That is what an undiscriminat- ing principle of equal chances seems to say, but it is most implausible. What seems more plausbile is that I am free morally to keep or give away the tickets, or to give them both to one charity, or to give one to each, just exactly as I see fit. The examples that I used earlier---a parent taking a decision affecting his children and a government taking one affecting its citizens---were indeed intuitively plausible, but that may be precisely because each is a case where an agent-centred obligation is present, an obligation arising from special role-relations (parent, government) involving responsibility, including a responsibility to deal equally. My undiscriminat- ing principle of equal chances, therefore, looks too broad, and all that seems justifiable is one tied to a specific role. This then suggests a rather different approach to free-rider problems. Baier's own suggestion is this.45 In free-rider 220 MORAL IMPORTANCE problems there is a network of special claims linking the parties, a set of role rights to equal treatment. All of us in the group benefit. So each of us, in return, is subject to a special claim, arising from our role as willing participants, to help produce the benefit. Not all of us are needed, so as many as possible ought to be exempted. But no one has a special claim to be exempted. So every one has an equal claim. The reasoning all takes place within this network of claims. The vocabulary of "roles' and the "claims' they give rise to suffers from a lot of obscurity. Look at the decisive move from a tough to a gentle solution---"When not all are needed, as many as can ought to be exempted'. How are we meant to get from "No one has a special claim to be exempted' to "Every one has an equal claim'? The argument is supposed to work within a network of moral claims but it starts, not with a moral claim, but with the non-existence of any such claims. How do we get from that to the existence of equal claims? Well, some notion of equality is coming into play, perfectly legitimately. But what notion? Is it not my general one? No, one might think, it could not be such a general principle; the Lottery Example shows at least that much. But nothing is shown by one intuition on its own. We ought to look at other cases of the same form (that is, cases without a special obligation to treat persons equally, including granting equal chances, of the sort that arises from role-relations). Suppose we have a doctor treating victims of some disaster. The victims have been sorted on reasonable criteria (for instance, hopeless cases, non-urgent cases, etc.). In the savable- and-urgent class you and I are left: each savable but time is left to save only one. The benefit from saving each is identical---and all the rest. Since you are injured in the legs and I in the arms, I walk up to the doctor when he is free and say, "Here I am'. Would you not be inclined to shout, "Not fair---just because he can walk and I can't. Give me at least a chance. Toss a coin.' Ought the one to be saved be chosen randomly? Probably. But is this just another case of a role-relation, where our equal claim is built into the role of "doctor'? I doubt it. But, anyway, take next not a doctor but a layman with first-aid skills that would save one of us. And then take an ordinary person with food that will save one of us 221 FAIRNESS from starvation. And then take a person not with food but with money with which one of us can buy food. And then take a person with a lottery ticket.46 A fairly wide principle of equal chances does not seem to me to be ruled out by our intuitions. To make his case, a role-deotonologist has to be able to specify roles in a way that makes it clear that the related claims do indeed arise from them and not, say, from non- role-related principles hovering in the background. For in- stance, the role "father' carries certain rights and is subject to certain claims, at least conventionally and often legally. But do these rights and claims also carry weight morally? The rights and claims that convention and law have granted to fathers have changed through history. There are two problems about reading claims off role-descriptions, which in the end reduce to one. If the claim is already built into the role- description, it is the result of historical contingencies and so has no obvious moral authority. If it is not, then we are choosing to build it in but have then morally to justify our choice. Of course, a claim's being the result of historical chance does not mean that it could not, after the moral assessment of the related institution, rise to moral status. It just means that it would have moral status only in virtue of the further judgement. In any case, the role-description in free- rider cases is something like "co-operator' or "co-producer', and it is very implausible that either description has built into it, as a piece of English or as a matter of convention, an equal claim on exemption from producing the benefit. So, since the moral claim is not already built into the description, we must be choosing to build it in ourselves. Our choice, of course, may be defensible. So the problem in these two alternatives comes down to the same. Either way we have to justify choosing to regard certain claims as part of certain roles. How are we to do it? One answer is this. The role in question is "co-producer of a benefit'. In general we each enter co-operation on a roughly equal footing. We agree, tacitly at least, to co-operate, and our initial equality transfers to equal claims to the benefit. Now, one benefit of the co-operation is ending up in the lucky position of being able to enjoy the benefit without having to 222 MORAL IMPORTANCE put in the work, and the initial equality, it is reasonable to think, also gets transferred to an equal claim to this lucky position. Without such terms, free and rational persons would not agree to co-operate. But, as we have seen,47 this contrac- tualist argument seems weak at just the point we should have to rely upon it here. It can yield equal claims of too strong a sort, so it leaves us with the job of having to distinguish equal claims that have moral status from equal claims that do not, without supplying the equipment to do the job. So let us translate the argument out of the contractualist mode, retaining most of the early moves: we co-producers enter the co-operation on roughly equal footing, so the benefits should be divided equally between us, including the chance at exemption, in virtue simply of our status as co-producers. But that cannot be the whole story. It is characteristic of role-deontology that it stops too early. What would it say of this familiar sort of case? Suppose two tribes, the Northerners and the Southerners, are driven by climatic changes to abandon their homes for the temperate middle region. They happen to settle on adjacent ground and start tilling and planting. Soon the Northerners find that their land is barren and that they cannot survive on it. They move in on the Southerners, whose early crops are just being harvested, and ask to share with them. They echo Locke: "We did not get as much and as good.' Do the Southerners have equal claims on the benefits simply in virtue of being co-operators, that persons generally---the Northerners, for instance---do not? Most persons would say that Locke's proviso, or something like it, would have to come into play. The claim to any benefits, individual or co-operative, are limited by the claims of persons generally to an equal chance at a good life. But a role-deontologist might now reply that, although a principle of equal chances strictly for co-operators has to be sanctioned by a broader principle of equal chances, nonetheless the narrower principle is generated simply out of the role- relations. It is because we co-operate in producing benefit that we co-operators have equal claims, in the absence of competing claims, on the benefit. That does not matter. The crucial point is that to get the narrower principle in place as a moral principle appeal has to 223 FAIRNESS be made to the broader principle. So to solve the free-rider problem, we do seem to need the broader principle. In any case, it is by no means clear what the background for the co-operators' equal claim on the benefits of their cooperation is. It is initially attractive to propose that co-operators, as co-operators, ought to have special claims on the benefit simply because without their co-operation there would be no benefit. But the proposal is exaggerated. Sometimes they will co-operate without equal claim on the benefit, providing they have a reasonable hope of coming off better than they would without co-operating. Still, there is a point there. Sometimes, no doubt, we should co-operate only because we should have equal control over the benefit. The equal claim is sometimes an incentive to, indeed a necessary condition of, our producing it. In those cases it is odd to treat every one as having an equal claim; awarding them the claim ensures the non-existence of the necessary condition of what they have the claim to. But then a multi-level moral structure could accommodate the co-operators' equal claim on an upper storey, supported lower down by some more general principle of equality. What we all need to do, including the role-deontologist, is to find the best overall account of fairness. We also need to find a justification not only for claims arising from the role of "co-producer' but also from the role of plain "producer'; that is, we need a good overall account of property rights and how a person's labour comes into his claim to its products. And we need a much better general understanding of what it takes to turn the rights and obligations generated by social institutions into moral rights and obligations. These are the subjects of the next three chapters. I think that it is in these larger considerations that role-claims have a much harder time finding a place. XI RIGHTS 1.The need for a substantive theory I want to take, as my point of departure, a dissatisfaction with the present state of philosophical discussion of rights that all of us, regardless of our views, feel. It is obviously unsatisfactory that the term "rights', at least as it figures in the philosopher's lexicon, comes so close to being criterionless. It is still less satisfactory that the compound term "human rights'--- and it is on these rights that I want to concentrate in this chapter ---comes even closer. As a result, philosophers often give the impression of plucking human rights out of the thin air. For instance, in the middle of a justly well-known discussion of abortion, we find the author settling most difficult problems by introducing a right to determine what happens in and to our bodies,1 and if anyone should doubt, as many of us do, that there is exactly that right, we are unsure how to settle the matter. Also, we spend a lot of time these days arguing over the structural features of rights: whether they can be traded off against one another and against utility, whether they are, say, "trumps'2 or "side constraints.3 Yet we all know that there is no real hope of settling these arguments until we can say what rights there are, what their extent is, and what makes them rights. We cannot, for instance, sketch out merely the rough outlines of a theory of rights---for example, rights as side constraints---and let the filling be added later. The filling is what settles whether one right can have different grounds on different occasions, and whether a right can differ in impor- tance from one occasion to another, and whether a right with one kind of ground may be traded off against other rights or against utility, while a right with another kind of ground may not be. In short, structural features and substantive features have to arrive in a theory together. It would be perfectly safe to draw conclusions about the structure of rights in the 225 RIGHTS absence of a developed substantive account if the answers to the substantive issues were obvious, but they are scarcely that. But I think that virtually everyone agrees with these complaints, including those who make claims about rights that lack the foundations that the complaints desiderate. I doubt, therefore, that I have really to argue the need for a substantive theory of rights, but no doubt I should explain further what I mean by such a theory. I think of a substantive theory in contrast to a formal theory---that is, one primarily concerned with the sort of structural features I have mentioned; and also in contrast to a conceptual theory---one primarily concerned to explore the relation of the concept of a "right' to such concepts as "duty' or "permission' or "entitlement';4 and also in contrast to a taxonomic account---one concerned, as Hohfeld's was, with cataloguing the different types of legal or moral relations that rights consist in.5 A substantive theory overlaps these theories, but it is unlike them in being mainly concerned with the content of the concept, with its criteria. And it does not just tell us, as some philosophers have usefully done,6 the characteristics in virtue of which individuals are bearers of rights. It would also tell us why those characteristics justify the ascription of rights to them, and what else, if anything, does, and how, if at all, our present criteria might be improved. 2.First ground: personhood If there are such things as human rights, then they are rights we have independently of actual laws, conventions, or special moral relations. Therefore, it is likely that their substantive theory will draw on, although also modify, elements from the natural rights tradition,7 and will go something like this. Taking one's own course through life is what makes one's existence human. We value our humanity, so we value what makes life human, over and above what makes it happy. Now, as we saw in Part One,8 the freedom that makes life human has many sides. The distinction between positive and negative freedom is by now familiar to us, but perhaps the systematic way to study all the complexities of freedom is to study the complexities of "agency'. One component of agency is deciding for oneself. Even if I constantly made a mess of my life, even if 226 MORAL IMPORTANCE you could convince me that if you managed my utility portfolio (on the usual understanding of "utility') you would do a much better job than I am doing, I would not let you do it. Autonomy has a value of its own. But autonomy, on its own, is not enough. It is not enough to be able to choose one's path through life if one cannot move. One needs limbs and senses that work, or something to take their place. But that is not enough either; it is no good being able to choose and having the capacity to act, if one is so racked by pain or by the need to keep body and soul together that one cannot spare a thought for anything else. We surely also need some minimum health and leisure. Nor is this enough if others then stop us; we need liberty.9 We need other persons not to interfere in those areas of our life which are the essential manifestations of our humanity---namely, our speech and associations and worship. Nor is this quite enough; it is not enough not to be fenced in, if we are unable to form any conception of where to go. We also need a mind capable of assessing things, which means that we need some minimum education and access to persons worth hearing or reading. This personhood consideration, to give it a not unfamiliar name, goes some way towards making the notion of a human right more determinate. It generates most of the conventional list of civil rights: a right to life, to bodily integrity, to some voice in political decision, to free speech, to assembly, to a free press, to worship. It also lends support to a form of positive freedom, namely to a right to a minimum provision.10 And it says something about just how wide the concept of liberty is: as far as the personhood ground goes, only a narrow concept finds support; the right to liberty is not to do whatever fancy prompts (so that any restriction on satisfaction of desires is some restriction on liberty) but only, more narrowly, to do what is essential to living a human life.11 It also provides a right not to be tortured, because torture aims at destroying one's capacity to decide and to stick to the decision. We say that personhood concerns what is needed for human status, but it is tempting to be more generous and say that it concerns what is needed for human flourishing. Yet which should we say? What makes it so easy to shift between the two notions, human life and good human life, is the difficulty 227 RIGHTS of knowing where to separate them. But it seems that the minimalist notion is the one that the philosophical tradition, with reason, supports. If we had rights to all that is necessary for the good life, rights would be too extensive. We should then have a right not just to minimum material provision, but to any unsubstitutable component of a better life. If this were the way that the tradition regarded rights, then we should obscure the distinction, with no obvious compensation, be- tween, on the one hand, what enhances well-being and, on the other, what rights demand. However, the personhood ground, despite its importance, may easily lead us to underplay the dynamic side of life. For instance, we know that a person can be oppressed, can even be shipped off to the Gulag Archipelago, and yet, because oppressors are not perfectly efficient, not only retain his autonomy and creativity but find them enhanced. But one who is oppressed can still ask of his oppressor, "Who is he to have such control over me, whether it makes me or breaks me?' Rights, it would seem, must also secure the distribution of control over the central features of one's fate. Every responsible person must be granted control over certain matters affecting him. What is crucial is not just (human) status, but also control. The right to control the centre is a strong form of negative liberty. Liberty is the absence of barriers to living out one's life plan. In its weak form, one is at liberty if, for whatever reason, one can live by one's life plan, as someone still can who has a form of life imposed upon him which, by luck, he anyway wanted. But in its strong form, one is at liberty only if one could have lived by one's life plan, even if it had been different. The ground for the strong form of liberty is not obvious, but personhood must in any case be some of it. A person's values are not static; and fences that do not block now may block later. So personhood requires a strong enough form of liberty to guarantee the movement characteristic of human development. That still does not guarantee complete control of the centre; it is not so strong a form of liberty as to stop a paternalist from arguing that his short-term violation of someone's control of his own life will make him freer in the long run. But the plausibility of a paternalist's claim has to be assessed. Given human limitations, 228 MORAL IMPORTANCE how likely is it that a paternalist will deliver the goods? But now the justification for strong liberty is beginning to shift from personhood to practicalities, which probably deserve to be regarded as a ground on their own. 3.Second ground: practicalities The personhood consideration leaves a lot unsettled. For one thing, it says that there is a right to some political voice. Yet how much? And what, too, is the minimum material provision? Where life expectancy reaches 40? Or would it have to be 60? And what is minimum education? Literacy? Or would it have to be the ability to ponder the meaning of life? And we have, it says, some sort of right to bodily integrity, because without security of body we have no security of action. But, to raise Robert Nozick's question, does this right bar a state's forcibly taking one of my kidneys for transplant? Does it then bar a very accommodating state from demanding a pint of my rare blood which, it says, it will take in my own house while I sleep and leave me to wake in the morning as fit as ever? Well, certainly it is no good expecting the personhood consideration to protect me against such an accommodating state, which threatens nothing essential to my human status. But there are practical considerations that are obviously relevant here. A line has to be drawn somewhere; the personhood consideration shows that much. And men and governments are not scrupulous and are prone to domination and self-serving. Moreover, the line has to be clear and, for safety's sake, at some remove from the vitals. I can, it is true, still lead a human existence if the state takes one of my kidneys, but one might well fear that the chances of doing so in such a pervasive and interventionist state would be slim. Whatever the result of this line of thought, there is no doubting the relevance of this kind of consideration. On any account of rights, I should think, practicalities play a large part in determining their final shape. 4.Third ground: the private sphere Personhood supplies only the weightiest sort of reason for rights, the survival of one's human status. However, Mill 229 RIGHTS thought that whether something, weighty or light, falls within one's private sphere also counts. How I dress or part my hair is hardly central to my living a human existence. Still, it is my business: even if what I wear upsets you, surely I am within my rights dressing as I please. Is the private sphere, then, a further ground for human rights? Personhood yields a right not to a general liberty but only to certain specific liberties. Mill, however, proposed that liberty should extend to whatever we wish to do, so long as it does not harm others. His harm test allowed him to define a self-regarding, or private, class of actions. But there are problems with Mill's broad conception of liberty. First, it lacks any clear value supporting it. Our status as persons is clearly valuable to us, but control over what is merely private is far less obviously so. Second, the justifications of the broad conception of liberty that carry weight, Mill's own, for instance, seem really to be an elaboration of the old personhood ground rather than the provision of a new ground. Stopping me from wearing the clothes I want, as inessential as any particular set of clothes may be to my human standing, certainly touches my self-respect. The idea of a person that we have already made appeal to is of a self-determiner. But to deny me freedom to express my own tastes does indeed threaten my status as self-determiner. Exactly which clothes I choose may be trivial, but my status as an independent centre of taste and choice is not. In the interest of keeping the concept of a right as sharp as possible, one might want to keep the personhood ground as narrow as possible---the absolutely minimum conditions for carrying out a life plan. If so, a right to dress as I please would not be supported by it. On the other hand, one might want to make it a bit more capacious---what is necessary for human dignity. One would lose some sharpness with the extension, but one ought to keep in mind that, in any case, the narrow notion would not itself be especially sharp anyway. It turns on what is central to human status and what is necessary for carrying out a life plan, and what is central or necessary is essentially moot, and can itself be seen narrowly or broadly. So, in any case, there will have to be a lot of not very sharp-edged debate about these moot matters, largely in terms 230 MORAL IMPORTANCE of practicalities, in order to fix the boundaries of rights. The loss, therefore, from adopting the slightly more capacious conception is not great. And the gain is a notion of liberty which is more in accord with the philosopical tradition, yet which still has a ground of some substance. This ground, although not as circumscribed as it might be, is not so extensive as to give us, as Mill thought we had, a right to general liberty. Mill focused on the private domain when it seems better to focus on one's status as a self-determiner. And some things that are "self-regarding' are also too remote from one's being self-determining to be protected by this slightly more capacious conception. 5.Fourth ground: equal respect There is another consideration, equal respect, that seems different from any of the preceding ones. Utilitarian thought proceeds by the simple device of trading off goods and bads by their magnitude, paying no attention to whether the trade-off occurs within one life or between different lives. It is just this particular lack of attention that invites the rejoinders: one person may not be sacrificed, without limit, for the good of others; the well-being of one person cannot simply be replaced by that of another. Yet to say that it cannot simply be replaced implies that sometimes it can and sometimes it cannot. And to say that one person may not be sacrificed without limit implies that sacrifices are allowed but only up to a point. Each suggests a line but does not supply it. And it is not that we can expect rights to supply it, because this consideration is meant to be a more fundamental one from which rights are derived. These lines represent what has been called the moral significance of the separateness of persons.12 It takes a lot of work to see what this separateness consideration amounts to. It might be hoped that the separateness consideration could be made more determinate through the idea of contracting. If persons have first to agree with one another, then each can exercise a veto in his own interest, and a veto seems the ideal instrument for expressing the separateness of persons. But we must give the contracting situation concrete shape, for instance to decide upon what degree of ignorance to impose upon the 231 RIGHTS contractors, in order to be sure that it will indeed express the separateness of persons. There are several different contracting situations. Which captures the right conception of separate- ness? It would help to have some rough independent idea of what that is. Besides, as we have seen,13 contractualism gets the restrictions on trade-offs too strict. I think that the way forward lies in this direction. One cannot sacrifice a person without limit because each of us has only one life, and each person's fate matters and matters equally. So we may take the separateness consideration as being, in effect, some form of a principle of equality, not a principle requiring equal treatment (say, meting out equal portions of resources) but that different and altogether deeper matter of treating persons as equals, of showing them equal respect. Thus, the limit on trade-offs that the separateness consideration desiderates will be the limit imposed by equal respect. But what is that? Equal respect is a little less indeterminate than the separateness of persons, but is still notoriously indeterminate itself. Does equal respect require (i) merely equal weight ("everybody to count for one, nobody for more than one' in the way that utilitarians mean), or (ii) a minimum level of well-being, above which obligations cease, or (iii) an equal start with equal prospects, after which inequalities resulting from just transactions are themselves to be regarded as just, or (iv) equal goods, except where inequalities work to the advantage of the worst off, i.e. Rawls' maximin, or (v) equal goods with equal prospects, or (vi) equal well-being, or (vii) equal opportunity? I know that many persons believe that the use of notions such as "the separateness of persons' and "treating persons as ends' is precisely to choose between those competing principles of equal respect. But this is to try to chose between these principles by appeal to notions so vague that they can be given content only by the choice between these principles. Nor can we remedy the vagueness by trying further to plumb the depths of our intuitions about either equality or separateness or respect for persons, because intuitions are superficial things. There will be no depths until we choose them and put them there. What we need is a well-worked-out theory of equality; such a theory would, at the same time, give content to all these notions. 232 MORAL IMPORTANCE We made a start on this theory in the last two chapters and shall not really finish it until we get to the final chapter.14 Anyway, a theory of rights is an important part of it, so at this point we could only be in the middle of building it. Suppose, however, that what will eventually emerge will be certain rather demanding principles of equality---say, for the sake of argument, maximin or equal well-being or equal resources, all of which, I think, prove to have a place in a full substantive theory. If so, then equal respect constitutes a new ground for rights. Personhood requires, among other things, minimum provision. Practicalities may lead us to define the minimum fairly generously. But equal respect requires something stronger than either: some qualified form or other of equal share. The only reason to doubt that equal respect is a new ground for rights is that it may turn out to be reducible to the personhood ground, or vice versa. There is a long philosophical tradition which has it that, a few difficult marginal cases aside, we are all possessors of human standing and hence of human worth. Clearly we differ in the degree that we possess the various features that constitute human standing---for example, in our capacity for autonomy, rationality, evaluation, and action. But, according to this tradition, the notions of person- hood and moral worth do not admit of degrees; anyone inside the boundary, no matter how far inside, is equally inside. If, in the end, it is this equal possession of human standing that is the ground of all rights, then it is not clear whether equal respect reduces to personhood, or personhood to equal respect, or each is only part of another, deeper ground that combines both. But the notion of equal respect is still too obscure to allow us to settle the matter now. So, until we have a theory of equality that makes the notion more determinate, it is best for the time being to treat it as a separate ground. 6.What rights does the substantive theory yield These then seem to me the three elements of a substantive theory of human rights: personhood, practicalities, and equal respect. (The privacy ground, the fourth possible ground I considered, is best thought of merely as a development of the notion of personhood.) 233 RIGHTS Do these three grounds yield the human rights that moral theory needs? I have suggested that, in some form or other, they yield rights to life, to bodily integrity, not to be tortured, to autonomy, to the central civil liberties, to minimum provision, to a strong form of liberty (that is, to sovereignty at the centre of one's life, so, for instance, I think that it could plausibly be argued, to sexual freedom and possibly, although less easily argued, to freedom to drink and take drugs), finally to some form of equal share (that is, share in what makes a good life possible: although whether the equality is to be, for instance, only at some starting-point or as far as possible continuously, and whether it is equality in resources or need-satisfaction or desire-satisfaction are all questions that the well-worked-out theory of equality will have to settle). And the equal share is not only a share in material goods but also in the powers and opportunities that matter just as much to a good life, and so it requires that advantages be open to all and that no advantages be denied without due process. We have also seen, however, how much these three grounds leave indeterminate. Take, for example, the right to privacy. Is it a human right?15 What are its boundaries? The personhood consideration would yield a right to whatever privacy is necessary to human standing: whatever privacy is needed for the thought and communication that go into forming one's life plan, whatever privacy is needed in develop- ing the personal commitments central to one's life plan. But that, it must be admitted, is not a very extensive right. What, for instance, does it tell us about the invasions of privacy that worry us now: phone-tapping, electronic eavesdropping, access to one's medical or financial records? The only resources that the three grounds supply at this point are various tangled practical considerations. Why do governments want to tap phones? Which is now the greater danger---public intrusion in private life or private subversion of public life? What is at stake in debates about modern intrusions on privacy is a certain sort of power, and if nowadays the greater threat is from government power, and if the threat is great enough to undermine one's life plan, protection should go to individuals. But this sort of argument is not timelessly valid; it allows the possibility that at other times, in other circumstances, the 234 MORAL IMPORTANCE powers should go elsewhere. In general, it would mean that, to the extent that practical considerations determine the bound- aries of rights, they are subject to periodical redrawing. But I suspect that practical considerations are, at this point, all that we have to go on. And since practical considerations enter into the determination of virtually every human right, human rights have neither sharp nor fixed edges.16 Another important consequence of this account is that rights have more narrow boundaries than convention says. There is no general right to liberty, but only rights to specific liberties. There is no right to say or publish what one wants, not even a prima-facie right subject to limitation by conflict with other rights, but merely a right to express what matters to the centre of one's life. There is no broad right to determine what happens in and to one's body, only a limited right to the bodily integrity necessary to carry through one's central aims, where the boundary of the "centre' is essentially moot. 7.The need for a second level to the substantive theory Rights are linked to such values as autonomy, liberty, and equal respect. The substantive theory suggests that they are to be seen as protections of these values and so seen within a generally instrumental or teleological framework. That consti- tutes the first level in a substantive theory: showing in what rights are grounded. But, unfortunately, that does not yet get us very far towards understanding the structure of rights, for the teleological framework need not be utilitarian, nor even consequentialist. Autonomy and liberty are, on the face of it, not utilitarian values at all, and equal respect is less a consequence of respecting the corresponding right than a value the content of which is itself best expressed in terms of the right. And we need answers to many questions. Do rights conflict with one another? How can we settle conflicts? May rights be traded off against each other? Against well-being? Do rights differ in relative importance? How is importance estimated? Are any rights absolute? Should we aim at maximizing the observance of rights? Can rights be forfeited, or diminished, by wrongdoing? These questions raise parallel questions about the values to 235 RIGHTS which the rights are linked. Do these values conflict? Can they be reduced to one value, or to one metric? Are they all the sort of values that we should seek to maximize? All of these questions, whether about rights or values, are unavoidable and require a second and deeper level of explanation. The first level in a substantive theory is concerned with how rights are grounded in values. But the second has to establish the character of the values themselves.17 8.A sampler of values that rights protect So I turn now to the second level of a substantive theory, the investigation of the nature and structure of the values that rights protect. I want to consider just three of them: autonomy, liberty, and equal respect. I consider just these three not because I believe that all rights can be derived from them alone; on the contrary, there is much more to the personhood ground than just autonomy and liberty, and I omit the practicalities ground altogether. I take these three simply as examples for closer study. But to them one must add well- being, with which I shall start. Well-being is best explained, I argued in Part One, not in terms of states of mind, but in terms of fulfilment of desire, where "fulfilment' is meant without psychological overtones and implies merely that what is desired comes about. The desires that are relevant on this account are not only persons' actual ones but also those they would have, if they understood the nature of possible objects of desire; a person's own conception of what is in his interest, therefore, is not definitive. This interpretation has an important consequence. "Well- being' is not to be seen as the single overarching value, in fact not as a substantive value at all, but instead as a formal analysis of what it is for something to be prudentially valuable. Therefore, well-being will be related to substantive values, such as autonomy or liberty, not by being the dominant value that subsumes them, but by being an analysis of, and the related suggestion of a metric for, any prudential value. It should be seen as providing a way of understanding the notions "(prudentially) valuable' and hence "more valuable' and "less valuable'. Well-being, therefore, is not what it is 236 MORAL IMPORTANCE about objects that makes them desirable. What makes us desire the things we desire is something about them---their features or qualities. Now, on the face of it, the values that back rights---for example, the three values that I have singled out, autonomy, liberty, and equal respect---seem clearly to fall outside the ambit of well-being. However, sometimes that is because the notion of well-being in use is too narrow and would have seemed too narrow quite apart from this issue. Consider autonomy first. Would you, for greater serenity, surrender your autonomy? If you understand what is at stake, then unless the pain of autonomy is in your case very great, no doubt you will prefer autonomy. But that preference can, I think, be brought within the ambit of well-being. For what reasons might one have for thinking that the value of autonomy is not the sort of value that well-being encompasses? One reason would be that autonomy is of absolute value, value greater than any elements of well-being. But most of us think that, on the contrary, there could come a point, say with a psychiatric patient, where autonomy was so painful as to justify reducing it. Another reason would be that the value of autonomy, though not absolute, is not given by its place in informed desires either. One way to test that would be to see where trade-offs sanctioned by these different views of the value of autonomy---its values as its place in informed desires and its putative true value---diverged. It is hard to find examples, and that suggests that autonomy falls largely within the ambit of utility. Liberty too, especially the strong sort of freedom to do not only what one in fact chooses but also what one might have chosen, seems clearly to fall outside the bounds of well-being. It concerns distribution of power, not distribution of well- being, even on its formal interpretation, because what is distributed here is not what constitutes the value or signific- ance or quality of life. It goes beyond quality of life to considerations of control: every responsible person must be granted control over matters that crucially affect him. This is the point that John Mackie makes in arguing that, if we understand goals properly, we see that they are less theoretic- ally deep than rights:18 237 RIGHTS A plausible goal, or good for man, would have to be something like Aristotle's eudaimonia: it would be in the category of activity. It could not be just an end, a possession, a termination of pursuit... But Aristotle went wrong in thinking that moral philosophy could determine that a particular sort of activity constitutes the good for man in general... People differ radically about the kinds of life that they choose to pursue. Even this way of putting it is misleading: in general people do not and cannot make an overall choice of a total plan of life... I suggest that if we set out to formulate a goal-based moral theory, but in identifying the goal try to take adequate account of these three factors, namely that the "goal' must belong to the category of activity, that there is not one goal but indefinitely many diverse goals, and that they are the objects of progressive (not once for all or conclusive) choices, then our theory will change insensibly into a right-based one. We shall have to take as central the right of persons progressively to choose how they shall live. All that Mackie says here about goals seems to me true, but no stopping point. It is not likely that control would be valued for itself; it is valuable because of the value of what can be controlled. If others were to intrude but only at the very periphery of one's life, the power to repel the intrusions would not matter much. If the intrusion were at the centre but were motivated, for instance, as with some surgical intrusions, by an uncontentious and beneficial aim and carried out by sure techniques, again the power would not matter much. However, in the real world, intentions of intruders rarely are honourable, or values agreed, or techniques sure. So, in the real world, power at the centre of one's life matters immensely. But its value derives from the value of the whole way of life at stake. Mackie's argument is that since what is valuable in life is a kind of activity, a way of life, and since conceptions of the good life are diverse and developing, liberty must be "central'. But liberty's being "central' in this sense is compatible with its being derivative. Mackie's argument is meant also to establish that liberty is not part of a more fundamental teleological structure, but that is what it does not do. Certainly the fact that what is valuable in life falls under the category of activity would establish this, only if an activity could never be a goal. But it can. It can even, on a proper conception, be a 238 MORAL IMPORTANCE component of well-being. "Socrates dissatisfied' describes not a state at the end of an activity, but a way of life. The fact that a person's conception of a valuable life changes and matures does not suggest that a right to liberty is in some important way more basic than goals. It suggests only that no one goal may be permanent or authoritative. But it is that we have goals, that we have the chance of making our life valuable, which gives value to the right to liberty. So, contrary to what Mackie suggests, it is not the right to liberty that is basic but the valuable life, on some conception or other. And the valuable life is basic in just the sense that Mackie tells us he has in mind: it captures what it is that gives point to the rest of the moral structure. There is another reason that might lead one to think that liberty falls outside the bounds of well-being. Utilitarians, Mill for example, have claimed that we ought to respect liberty because, if we are at all wise to the ways of the world, we shall realize that paternalism is largely counter-productive and that in the long run general welfare is best served by strict non-interference.19 All that is true, but many find it, with reason, a lame defence of liberty. Liberty, they would plausibly insist, is itself valuable, valuable apart from this link with well-being. But when one spells this out sufficiently, one merely discovers a new link with well-being. We value our status as persons and want to live recognizably human lives, and liberty is a central component of that. But then liberty, like autonomy, can be fitted into the scheme of our preferences, and its value explained by its place there. So there is more than one kind of link with well-being. It is a mistake to move from liberty's being valuable in itself to the conclusion that well-being does not encompass it. Using the formal notion of well-being does not commit one to monism in values. One can value many different things, and value them not for any state of mind that they result in, but for themselves. And since well-being encompasses two central features of personhood, autonomy and liberty, that common disjunction between either grounding rights in well-being or grounding them in personhood is unreal.20 Liberty seems, then, best understood as belonging to a teleological structure. Whether the ends of that structure, the 239 RIGHTS valued ways of life, can be brought within the ambit of well-being is contentious. But the contention is between conceptions of prudential value. Nothing that has so far entered the story about rights rules it out. So let us move on to the last and hardest case. Equality is a very different kind of value from the others. Unlike autonomy and liberty, which focus in a way on the value of one life, equality focuses on the comparison of lives. It is not a prudential value at all; it is a moral value, in a way the moral value. So equality presents the best case for saying that rights are grounded in more than just prudential values. Yet when we survey principles of equality, we are faced with an embarras de choix. I mentioned earlier seven perfectly familiar principles, and there are still more. So our first job is to make the notion of equal respect more determinate. Now, when one reflects on these various principles of equality, one sees that some belong on different levels; they are not, after all, competitors but principles about different things. For instance, there is the most fundamental level where equality is to be seen as an interpretation of impartiality---that sort of impartial- ity that constitutes the moral point of view. We all agree that to look at things morally is to look at them, in some sense or other, impartially, granting every person some sort of equal status. Of course, we should have to make this notion of equal status more determinate---say through one interpretation or other of the Ideal Observer or Ideal Contractor. In any case, principles of equality can be principles of impartiality in this sense: they can express the spirit with which one will, if one is moral, consider the facts of the matter. Then there is a second level, the level of moral principles resulting from applying first level principles to facts. For instance, Rawls derives, from his Original Position, his two principles of justice.21 And on this second level there are different kinds of principles generated which can reasonably claim to be principles of equality. There are, for instance, theories of rights which see people as possessors of equal basic protections and entitlements. There are also theories of distribution of resources which might, for instance, say that resources themselves should be equal, or that their pay-off in well-being should be equal. Then there is a further wrinkle: I 240 MORAL IMPORTANCE made a case in the last chapter for distinguishing a reflective from a practical decision procedure, so one might see this level itself splitting in two. But we can leave that aside for now. Then there is a third level of principles simplified for action on the large social scale, where knowledge is short and justice rough. For example, a utilitarian might adopt a principle of equal resources, because on the level of social policy one cannot consult individual utility functions or hope effectively to control utility levels by manipulating shares of resources. Hence, equality is a multi-level and, within a level, multi-dimensional notion. This is not surprising: equality is merely a formal notion (sameness of some feature), and sameness of different features matters morally. Our moral notion of equality fragments. There is not one principle of equality, but many. Now how do principles on these different levels support rights? Third level principles, being rough rules of thumb, would support rights which were themselves only rough rules of thumb and so rights of a kind less strong then we are interested in now. First level principles, on the other hand, principles which in effect express the moral point of view, would support perhaps too strong a kind of right. John Mackie and others think that they provide the one absolute human right, namely, a right to equal respect in the proced- ures that determine the compromises and adjustments between all the other, non-absolute rights.22 But the doubt here is whether this is too much the whole of morality to be anything as specific as a right. It is absolute because it is moral standing itself, and morality can never recommend suspending the moral point of view. But whether or not it is best to regard this as a right, it is the principles on the second level that are the promising candidates for what we normally regard as human rights, and it is on them that I want to concentrate. The competing principles on the second level provide rival ways of fixing the moral point of view. Some, as we have seen, seek to express the moral significance of the separateness of persons. Others give expression to the demands of everybody's counting for one in utility calculations. What they are all concerned with is what is permitted in the way of trade-offs between prudential values. They are all views about rights 241 RIGHTS and distribution. What is common to all conceptions of equal respect is the belief that some sacrifices of one person for another are permissible, but that a limit is imposed by everyone's equal status as moral persons. Well, what limit? To explain that is to explain our conception of moral status. And to explain it demands going beyond such edifying but empty formulas as "no person may be sacrificed without limit' or "respect for persons', which leave so much undetermined that nothing can be got out of them until more is put into them. And the most promising way to do this is to use second level principles as their further determination. The way to fix the second level is to fix the principles restricting trade-offs.23 We have already looked at various models for the restriction, and the argument was roughly this.24 One model is that of a line beyond which trade-offs are not allowed, a line which defines one's unbargainable person- hood. Suppose one proposed the generous line supplied by maximin: the worst off to be as well off as possible. But this is not a line we should long defend. Maximin, intuition tells us, has exceptions; rich societies where the worst off are well off do, on the face of it quite reasonably, allocate resources---to art, for instance---in a way that further benefits the well off. Suppose then, to accommodate this, one redrew the line so that it prevented sacrifices of one person for the already better off, disastrous drops in welfare or, at least, drops to dismally low levels. But recall the example of the beautiful roadside avenues of trees in France. We do allow trade-offs between the common good and disastrous drops, even when the drop is as disastrous as death and the good is merely more aesthetic pleasure. Suppose, therefore, that one accepted that neither of these lines would do and that one looked for something weaker, something like Amartya Sen's Weak Equity Axiom, which goes: if one person is a less good utility producer than another, say because he is handicapped, so that at any level of goods he is less well off, then he must be given something more than the other. But this "weak' requirement is still too strong. Surgeons, in choosing between two patients for a kidney transplant, often rightly give it to the one with better prospects of survival, even if the other has been chronically ill and has always been less well off in the past. Intuitions, therefore, go 242 MORAL IMPORTANCE against the model of an uncrossable line. A natural move at this point is to fall back on the familiar model of same-level principles: to abandon the idea of a cut-off for utilitarian trade-offs and opt instead for a counterforce. There are certainly examples that seem to support this model, especially against the model of maximizing well-being. But if we look at enough examples, we can also find plenty that go the other way. We can generate both sorts of examples just by juggling the figures in the transplant case for the age of the patient and the years of survival. So on the testimony of our intu- itions none of these models is satisfactory: not the model of an uncrossable line, nor that of same-level principles, nor that of maximizing well-being. It is easy to announce attractive-sounding principles. But if one collected the trade- offs that fairly widespread intuitions support, and if one took the task seriously enough to collect a large number of them over a wide range of cases, this "undisputed set of trade-offs', so to speak, would undermine every proposal that moral philosophy has yet produced of principles govern- ing trade-offs. This is not fatal to them, but it certainly shows how far we are from having any satisfactory arguments on the subject. In a situation like this, one has to get behind the intuitions to see how they work. I tried to do this earlier.25 What emerges, I suggested then, is that the model of maximizing well-being has a role to play in certain fundamental delibera- tions about trade-offs. It is not that my earlier argument was decisive in favour of any single model. There are many other tests that a model of equal respect has to pass, one important one being the plausibility and explanatory power of the substantive theory of rights that it is part of. 9.The second level The second level of a substantive theory has to explain the structure of rights---how we can settle conflict between rights themselves and between rights and well-being, whether rights differ in importance not only among themselves but also a single right from occasion to occasion, how such importance is estimated, and so on. The first level stated the existence 243 RIGHTS conditions for rights, the second level must state the rules for their operation. The trouble with bringing values such as autonomy and liberty within the ambit of well-being and, in addition, accepting a maximizing model of equal respect is that the restrictions on trade-offs that look likely to emerge are unpromisingly weak. We should look for stronger restrictions on the maximizing model---not arbitrary ones tacked on to turn it into what we should like it to be but ones that it itself implies. There are, I think, at least two. First, there are restrictions in prudential value theory that, because of underdevelopment of the theory, have not begun to be properly appreciated. We have still to understand, and to work out, how radical the consequences are of the shift to a formal conception of well-being. Since well-being is not itself a substantive prudential value, it is the nature of those values that are that determines the structure of informed desires. Thus, there is nothing in the formal conception of well-being that rules out one value's being incommensurable with another. We have to look at the substantive values to see. And what we find, I argued earlier,26 is that they do exhibit one form of incommensurability---discontinuity. We find values such that no amount of one can outweigh a certain amount of the other. So there are certain inflexibilities in prudential value theory that therefore get transferred to a metric of well-being. If I had a friend prodigiously shrewd and overflow- ing with savoir faire who could save me a dozen minor false turns if he took over the management of my life, I should still value my autonomy so highly that I would not contemplate surrendering it. But what if he could save me, not a dozen, but a hundred minor false turns? But the trouble is this: minor false turns do not seem to be weighty enough, even in large aggregates, at least the aggregates that life presents us with, to balance the value of living life autonomously. Some things might indeed be weighty enough---great pain or anxiety, for instance---but minor false turns are not. Not even a thousand ---someone might press---or ten thousand? But now the problem is to get one's mind around the question. If one's life contained a huge number of false turns, one right after the other, they would scarcely leave space for anything to go right; they 244 MORAL IMPORTANCE would be a pretty good indication that something else, far more serious, was wrong. Or consider another case, liberty. Can the value of a person's being free to live what he regards as the only life of substance and significance open to him be equalled by the upset or distress that his doing so might cause others? Well, Mill has plausibly argued that upset and distress are simply not in the same league as a person's making something out of his life. The stakes are so different, given the world as we know it, that not even large aggregates of persons upset and distressed, hundreds of them or thousands, come into the same league. If it were a matter of some minor liberty, say one's liberty to bathe nude, one would expect upset and distress, if there were enough of it, to match the value of the liberty. But if it is one's most major liberty at stake---to live out one's life plan---why believe that life presents us with aggregates of upset and distress that will, judged simply on the values accounted for in a prudential value theory, match it? This suggests a certain structure for the second level of the substantive theory. We need rights to be strong protections of the individual, but not too strong. That remark is a bromide, of course, but it does also state the greatest problem facing any substantive theory. There are no absolute rights except for the dubious "right' to moral standing itself, which it is probably less confusing not to regard as a right at all. Otherwise, the existence conditions for rights give no reason to think that rights cannot be traded off against one another or against well-being. And rights differ in importance: both one right with another and (as we have just seen in this last mention of liberty) one right on different occasions. So we need a theory that not only allows trade-offs but also explains how they work. They work because behind rights is the deeper notion of the differing importance of rights. So the theory must supply the criteria of importance and a metric derived from them. This is the chief job of the second level of a substantive theory. Our notion of importance must come from the weight of the values to which rights are linked. It is hard to see what else it could possibly come from. This would provide a basis both for trade-offs between rights and between a right and well- being. We can see this in the case of liberty: two liberties will differ in importance depending upon how close they get to the 245 RIGHTS centre of one's life plan (say, freedom of worship for a religious person compared with the freedom to dress as he pleases), and a minor liberty (say, the freedom to go on wearing imported clothes) can be less important than a substantial economic improvement (say, by import controls). And the theory could also explain why rights have some kind of priority over well-being. Rights are grounded for the most part in the values of personhood, and though these values come within the ambit of well-being there is a tradition in political theory that contrasts them with "well-being' on a narrower but not uncommon understanding of the term. To bring out this contrast, we might distinguish personhood from prosperity, where "personhood' means what I have been using it to mean and "prosperity' means all the rest of the values that make up well-being. (So "prosperity' has still to be understood quite broadly; it would include, for instance, certain forms of intellectual and artistic as well as purely economic flourishing.) Then, using this narrower conception of well-being, we could say that rights characteristically outrank well-being, that often rights should be respected even at the cost of maximum well-being, meaning that the values that make up personhood characteristically outweigh those that make up prosperity and that protecting them is often more important than maximizing prosperity. And we can say that certain key rights, such as the liberty to live out the centre of one's life plan, trump certain increases in prosperity, because here we start encountering incommensurabilities: for example, no amount of upset and distress can match the value of such a major liberty. But all of these judgments take place within a framework that also shows when trade-offs are allowed and how they work. So rights have priority over well-being but not a strict priority. The full account of this priority is a large part of the job for the second level of the substantive theory, and talk about rights' being lexically prior to, or trumping, well-being does not do justice to the complexity of what goes on. That is the first restriction, which comes out of the prudential value theory. A second restriction comes from the notion of equal respect. Equal respect, whatever else it might lead to, leads to maximizing well-being and so to a threat of over-weak restrictions on trade-offs. But, as we have seen,27 246 MORAL IMPORTANCE the home of the maximizing model is in the most general characterization of the criterion of right and wrong; maximiza- tion has a smaller role in the reflective decision procedure, still smaller in the practical decision procedure, and smaller still when we get away from individual decision procedures to a political one. So there is another restriction on maximization; it comes into play only occasionally. There are, as what is already a large literature on indirect applications of maximiza- tion makes clear, several different forms that this restriction may take.28 I began describing a multi-level structure in the last chapter, and should say what more I can about it now. We are led to some form of multi-level structure, I believe, by reflection on the differences in kind between moral considerations. One example is this. It is clear that some moral principles are indeterminate; they have exceptions that need stating, and they need constant interpretation to adapt their key terms to fresh circumstances. Often, though perhaps not always, behind those principles there are other, more funda- mental ones, that help shape these determinations and interpret- ations.29 Another example is this. A principle that fits one setting (say, small-scale interpersonal dealings) does not fit another (say, large-scale social arrangements), for which we need either a new principle or a new version of the original principle. At various earlier points,30 we came upon reasons to think that the principles of beneficence needed on the large social scale would be different from the ones needed on the small interpersonal scale. So we end up with rather different principles, but close cousins of one another, and we need to understand more clearly what the relation between them is. A multi-level structure seems to me the best explanation of these, and many similar, matters. But I do not want to overstate its powers of explanation. The multi-level structure that I want to describe may not exhaust morality. But it is, at any rate, the structure of that large part of morality that has to do with our concern for well-being. Also, it seems to me somewhat arbitrary exactly how many levels of moral thought one distinguishes and precisely where one draws the lines. The types of moral considerations that I have in mind shade into one another; there are no sharp divisions between them. Still, I want to suggest that it is useful to distinguish four levels. I 247 RIGHTS mentioned earlier three levels, one of which might be seen as splitting in two, and that is the idea I want now to develop. On the deepest level of a moral theory, in what earlier I called the general characterization of the criterion of right and wrong, on a level where one defines how one person's fate weighs against another's, the maximizing principle applies.31 It may not be the only principle that applies there, but that anyway is where it applies. It is a principle that applies to single acts. That is so simply in virtue of the department of morality of which it is a general characterization, namely act morality. It is the job of act morality to provide principles of intentional action. But the maximizing principle is indetermin- ate. It gives us the aim or object of moral action. But when we add to it the most plausible prudential value theory and when, further, we acknowledge the need for co-operative action among moral agents on fair terms, then the focus of moral principles moves away from the single acts of an isolated agent to life-structuring commitments, attachments, agreements, and institutions. This brings us to the next two levels, what earlier I called the reflective and the practical decision procedures.32 One important influence on the shape of principles is our actual situation of choice. We are characteristically short of time, facts, and fellow-feeling, and we need standards for these circumstances. These standards constitute the practical level. But we also need criteria for going behind those practical, everyday standards in order to decide on their exceptions, to amend them, and in general to deliberate when we find ourselves free of the restrictions characteristic of normal life. The considerations we then use constitute the reflective level. For example, since it is rare for us to know reliably how much different persons get out of some good, we need a policy for usual cases, and so on the practical level the principle of equality that we shall apply to them is: treat persons equally. But on the reflective level, when our knowledge is not restricted in that way, the principle of equality will take the form of equal respect, which allows trade-offs---as in, for instance, the medical cases that we discussed at some length earlier.33 Still, we have already seen too that many principles change little between these two levels. Any decision procedure, 248 MORAL IMPORTANCE including the reflective one, has both to deploy a prudential value theory and to accept the importance of group action. Our commitments, attachments, aims, and co-operative enter- prises incorporate central and important values. The prin- ciples on the practical level governing those parts of our lives are not replaced by other, radically different principles on the reflective level. The only difference between the two levels in this respect are those further considerations available on the reflective level that come into play when we amend or interpret our principles.34 These two decision procedures are for an individual; they answer the question, How should I decide what to do? But we also need an answer to the question, How should we as a group, especially as a large society, decide what to do? This shift to a political decision procedure brings with it systematic changes in the principles we use. A change in scale imposes further limitations in knowledge; now we typically lack knowledge about not just how much different individuals get out of some good, but also, since most of them are total strangers to us, what their conception of the good is. It also brings with it limitations in trust; we cannot depend upon an office-holder or an ordinary citizen in a large impersonal setting always to be perfectly scrupulous, so we must take the wide and long view and find rules that can cope with these variations in moral reliability. Beside limitations, change in scale also brings changes in function; our government has as its job promoting not your and my happiness but the setting in which we can pursue it. So the social principle of equality should take the form: equalize all-purpose means. For one thing, it does not matter how admirable one individual's conception of the good life may be; what is to be protected is our living a human existence, and that means protecting one person's capacity to live out his life plan as much as another's. And a society needs procedural principles. Each of us has a right to be a self-determiner. But the right, formulated no more fully than that, is indeterminate; it needs spelling out both for the individual decision procedure and especially, and rather differently, for the political decision procedure. Each of us should have a voice in political decision. But precisely what sort? Each of us also wants good social decisions, and so we 249 RIGHTS want complicated questions to be answered by knowledgeable persons. The determinate form of the principle is likely, therefore, to embody some compromise between values---say, a right to a vote in elections in a representative democracy. That is a proposal of one way---to my mind a useful way---to divide up the levels. But is the effect of the division to restrict maximization? Does it not, instead, just usher maximiza- tion from the moral scene altogether? My proposal that maximization should be thought of as part of the general characterization of the criterion of right and wrong looks rather like the proposal that Geoffrey Warnock and John Mackie both make that, though concern for well-being does indeed enter morality, it enters as its general aim or object and not as part of the content of its principles or rules.35 And Warnock and Mackie see their proposal as the abandonment of the maximizing test, at least in any recognisably utilitarian form. But I think that they want the levels more insulated from one another than they can be. If making life go well is the aim of the rules, it cannot be kept totally out of their content. It would be extremely odd if it were not allowed to play at least some role in making the rules determinate, in interpreting and amending them, or in reconciling their conflicts. For instance, as we have just seen, we have to get behind rights to the values they protect, and when these values conflict we have to know how to weigh them against one another, and the maximizing test enters there. It plays a role on the reflective level and also, though to a lesser extent, on the practical and political levels. But the greatest worry about a multi-level structure comes from moral psychology. Can the various psychological states that it posits all co-exist? The worry arises at many points, but let us consider just one. Bernard Williams poses the question, Where, either in society or in the human psyche, are we going to find a place for a division between a practical and a reflective decision procedure?36 In a society we might just manage to locate it in a division between social classes---a moral e$0lite, say, who will manipulate the beliefs and disposi- tions of a moral proletariat, who, for their own or the common good, will be denied much access to the workings of the reflective decision procedure. But where can we possibly locate 250 MORAL IMPORTANCE it in the human psyche? We might try locating it in a division between more and less reflective or reliable parts of human nature, with the first manipulating the dispositions, perhaps even the beliefs, of the second. But both suggestions are suspect. The behaviour of the e$0lite class looks very much like plain deceit, which, though it may in the end have a justification, certainly calls for one. But the sort of manipula- tion required inside an individual life looks simply psycholo- gically impossible. No one individual, at least no tolerably healthy, integrated individual of the sort that we should want both to be ourselves and also to have around us, could maintain on the practical level full-blooded commitments to co-operation, to individual persons, to his own life-structuring aims and, at the same time, on the reflective level regard all of these values as merely instrumental. But these worries largely rest on a misconception. It is wrong to assume, as Williams does, that a multi-level structure with the maximization of well-being at its base has to treat well-being as a substantive super-value, all other values being merely instrumental. For instance, a person might value commitments to certain individuals. It might be central to what makes his life valuable, and its value would partly depend upon his sense of the value of the fate of the other person. Of the values that come into this story, none is instrumental. In general, any prudential value, and its related principle of action, will be intrinsic, not instrumental. The case of the principles of fair co-operation is more complicated, but they too in an important sense are of intrinsic value. They have an instrumental side to them: they define necessary conditions for our reaping the benefits of co-operation. But they also in part spell out the consequences that our notion of equal respect has for co-operation; for instance, the principle of equal chances states a requirement that does not have merely instrumental force. There are, it is true, some rules and attitudes on the practical level (for instance, respect for promises) that will have only instrumental force. But virtually everyone accepts that some rules and attitudes are merely good policy, their justification coming from outside themselves. That fact, of itself, hardly raises problems of moral psychology. So whether there are problems depends upon precisely which 251 RIGHTS rules and attitudes are regarded as instrumental. Since thought about well-being will, as we have seen,37 focus on the shape of whole lives and on the importance of the action of whole groups, we need not in general worry that our important commitments at the practical level will be unnerved by thought on the reflective level; most of them reappear there. And the ones that do not are, I think, psychologically managable without intrinsic status. It is a mistake to regard the practical and reflective levels as consisting of entirely distinct kinds of deliberation. Many of the principles of the reflective level reappear on the practical level. There is no level, including the reflective, on which one can reason in the occasion-by-occasion manner of the isolated agent. There remains the considerable problem of telling when to shift from one level to another, and the related danger that we should always want to, or perhaps think that we ought to, shift to the one most authoritative level and stay there. But there is, I think, no one "most authoritative' level. The idea that there were would get some support if there were a single substantive super-value and one level of deliberation to which it was confined. But the general characterization of the criterion of right and wrong, for instance, has no more authority than the personal decision procedures. The reflective and practical decision procedures do not deal with what has only derivative value; on the contrary, it is only on those levels that we get proper representation of substantive prudential values and the importance of groups, both of which make it impossible for us to assess acts occasion by occasion. And it is only on the political level---with, say, the right of each citizen to a vote---that we can accommodate certain values---say, autonomy---given the need for a decision procedure for that large scale, with all the constraints that it brings. Our choice is not between thinking about political matters on a more or less authoritative level; with certain issues our choice is between thinking about them on a political level or not at all. As to when to shift levels, any fairly full account of the multi-level structure, in defining the levels, will give some guidance. I doubt that any multi-level theory, no matter how fully developed, will make it entirely clear when the shift should come. That will always to some extent present problems. But many problems about when to 252 MORAL IMPORTANCE shift level seem to me to be an ineradicable feature of moral life, and though a good account of the multi-level structure should ease them, it will not entirely remove them. To return now to rights. Where do they fit into this multi-level structure? Clearly rights will appear on the reflect- ive and practical levels and also on the political level, and any particular right is likely to need somewhat different formula- tions for its different appearances. There is unlikely to be one fully determinate set of rights appropriate to all levels. But generally the most important appearance of rights will be on the political level: they will define how society ought to be arranged to protect certain central values. Even when the threat comes not from the government but from other individuals, social arrangements are usually the most effective protection. On the social level it is likely that we should want much tougher principles than we should want for an intelligent, scrupulous individual in his private life. And since political principles have to deal with what is generally the case, the restrictions arising from prudential value theory---that person- hood is characteristically more important than prosperity ---have even more effect. Consider, for example, the right to minimum material provision. Both a government's knowledge and its aims are limited. The first, rough definition of the right is to "minimum material provision', provision needed to live "a recognizably human existence'. But this definition leaves the right highly indeterminate. Each particular society has to make it determinate in the light of its own circumstances. How high can a particular society afford to set its minimum? Practicalities will have to enter. At this stage the maximization of well-being doubtless will have to play some role. As a society gets richer, at any rate if it is not already very rich, we should want to fix the minimum at a higher point, not because it strictly follows from the concept of the "minimum' that we should, or from the notion of what humans "need' in contrast to what they merely "desire'. As we have seen,38 those notions are too indeterminate to have such logical powers. But by an appeal to practicalities a minimum is finally fixed. A public policy is adopted. What one can then claim as a right is determined by the policy, and not by considerations of well-being. It would not matter that in a particular case 253 RIGHTS well-being, on the broad conception, might be increased by holding some individual under the minimum material provi- sion. The social structures have been fixed where they have for good reason, and this fact does not constitute a reason to change them. And denying someone minimum provision is wrong in itself, and not because it produces, say, pain or unhappiness. I started out by suggesting that the first level of a substantive theory will ground human rights in personhood, practicalities, and equal respect. Still, a substantive theory cannot stop there; it has to answer questions about the structure of rights, in order to do which it is forced on to a second level of theory about how these values and well-being are related to each other. I finished by suggesting a framework for their trade-offs and two sources of restrictions on them. To object to this substantive theory, contrary asseveration is not enough---for instance, to insist that it is of the essence of rights but they are "side constraints' or that they are more potent "trumps' than this theory makes them.39 Nor is it enough to back up the asseverations with an intuition or two. Those who object will have to come up with a better substantive theory, if they can. XII DESERT 1. The moral interest IF we gathered persons equal in strength, intelligence, wealth, and political power and gave them an equal start in some job, inequalities would soon appear. Some would do more or better than others. Our responses to persons are dictated not only by their sameness---by the equal respect we owe them--- but also by their differences---by their desert. However, the word "desert' is not used quite as narrowly as I have just suggested. Some uses are of no interest to moral theory. Others are of interest but have already entered the theory under another heading (e.g. "equal respect'). So what moral theory needs is not so much "analysis' of the word "desert' as isolation of the uses that matter to it.1 Let me quickly explain the uses that I think do not matter. A woman certainly deserves the same wage as a man for doing the same work. But this is not the desert we are interested in. That is because it appeals to equal respect: everyone is to be treated equally unless there are relevant differences between them, and sex is irrelevant to pay. So this case matters to moral theory but is already catered for under a different heading from "desert'. Someone in a dangerous or hard job deserves extra pay. But this is not the desert we are interested in either. The extra pay is not reward but compen- sation; we want to make the person suffering risk or extra burden in some way equal to the rest of us who do not. The same is true of an injured person's deserving compensation; compensation is an attempt to equal things out. Also, not all differences between persons matter. In deciding that Miss Grenada deserves to win the Miss World Contest, I may show each contestant equal respect, but in this case the relevant differences have nothing to do with moral merit. Miss Grenada deserves to win but does not deserve any 255 DESERT credit for having been born beautiful. You may deserve more pay than I because you pick more grapes, but if you pick more simply because you have more strength and stamina, then what you do is the natural result of these endowments, and they, like Miss Grenada's endowments, are not enough to earn you moral credit. And the sort of desert that matters to us should not be confused with entitlement either.2 If I am silly enough to offer #1 to any child in a nursery group who says the alphabet standing on one foot and ten children up and do it, then all ten deserve #1. Since they met my conditions, they are entitled to it. Some cases of desert mix entitlement and genetic endowment and so are doubly not the sort of desert that matters. If Jessica writes a good examination paper, she deserves a good mark. She is entitled to it even if she is effortlessly bright and has not done a stroke of work. But an effortlessly good mind is like a pretty face, no moral credit to its owner. What then is the sort of desert that matters? It obviously does not come just from success---from picking more grapes or writing a good examination paper. That might tempt us to think that it comes from effort---from being able to say "I try harder'. But it cannot be simply that either. Effort can be aimed at the wrong things, or it can be bungled. Also a person who manages self-control from the start of his life, bit by bit, may in the end be both more creditable and spend less effort overall than a person who did not manage it from the start and now always struggles greatly but in vain. The sort of desert that matters is, I think, part of a simple, even rather crude, and by no means unchallengeable concep- tual framework. At its centre is a notion of par: what people of a certain type---with certain physical and psychological assets ---would find it natural to do. I speak of a type of person, but the assets that matter morally are often so widely distributed that the type then becomes any normal human. Where there is a par, people can do better or worse. When a person meets an obstacle that is not the sort that he naturally scales, he has to try to overcome it. And if he comes to a pit that is not the sort that he inevitably falls into, he just has to take care that he does not. He becomes deserving, of praise or blame, by scaling 256 MORAL IMPORTANCE the obstacle or falling into the pit. In the case of deserving praise, for instance, a person must do something (an element of success) better than type (an element of effort) and do better than type when he need not have (an element of autonomy). That is the not entirely trouble-free conceptual framework into which the notion of desert that I want to examine fits. It is not easy to make precise what persons with certain physical and psychological assets find it "natural' to do. What is "par' for someone cannot be defined in terms of his own past performance, because then anyone who already does well could earn credit only by doing better and better. Effort, it seems, has to have some place. Anyone who regularly does well earns credit if he could easily not do well. And anyone whose doing well has become habit earns credit now for his having made himself into this sort of person, which is no easy job. And what is "natural' is not to be taken to exclude what is "learned'. It is natural to be tempted; it is all too natural to succumb. Only stiff training changes that. But this training should probably be thought of as among the psychological assets of some types of per- sons, making a new, more self-controlled kind of behaviour natural. This account of desert sounds commonsensical, but it is also, by the standards of contemporary political thought, revisionary. A highly responsible job, society thinks, deserves more pay. But why? It cannot be on grounds of compensation, because greater responsibility generally makes jobs better, not worse. Nor can it be on grounds of earned credit, because the attractions of a responsible job are so great that what is hard is to resist them. And, according to this account, other common political sources of desert---brains, brawn, value or contribu- tion to the community---fail to qualify for moral desert too. Is anything left? My scaling the obstacle and your not may just show that we were not equal to start with anyway. It may all be the result either of the genetic lottery or of the environmental lottery. This is the determinist attack on the conceptual framework, the strongest attack of them all. I believe, and shall just assume, that the framework survives the determinist attack, that people can deserve the fairly full- 257 DESERT blooded moral praise and blame that I shall go on shortly to talk about. There is another, though less serious, attack on the framework. With time, people change, and sometimes in ways that make us wonder whether praise or blame is still appro- priate. Is it right to blame the seventy-year-old for his misdemeanour when he was ten? Is it right to go on blaming the criminal who repents and changes? These worries are connected with, but not exhausted by, worries about personal identity. Perhaps the only conception of personal identity that would make praise and blame appropriate is itself indefensible. Again, I believe, but shall just have to assume, that the most plausible view about personal identity leaves intact all of the problems about desert that I shall now turn to.3 2.Is desert a moral reason for action? Let me, to focus our minds, start with something obvious. It is our nature to respond to what other persons do. We character- istically have certain feelings about what they do, and the feelings lead naturally to action. Each responsive feeling has its own appropriate object and leads on to its own appropriate action. At the sight of courage we feel admiration, which naturally leads to praise. If we are helped, we feel grateful and offer thanks or return the favour. If we are harmed, we resent it and retaliate. Sometimes our feelings are inappropriate: only envy, and no admiration, at another's achievement. Or they are the wrong degree: small grudging admiration for a large achievement. And we know that some objects are entirely inappropriate to certain responses. Indeed, the intro- duction of an inappropriate object has a devastating effect on the response; it destroys it. I can admire what a person achieves. But if I am told to admire him because that will make him happy, I just cannot. His happiness, whatever it might be a reason for me to do, could not possibly be a reason for me to admire him. And if, solely to make him happy, I gabble words of praise, the last thing I am doing is praising him. The introduction of his happiness into the situation is like finding a painting beautiful and being told by its owner, who want to help one's response along, how much it cost. 258 MORAL IMPORTANCE But now, granted that certain responses are appropriate to their object, why think that it is a moral matter whether or not I have them? Granted that they often give me a reason for action, why regard it as a moral reason? Another person's desert surely enters into my moral assessment of his character. But why think it enters the assessment of how I ought morally to act towards him? When someone harms me, it is natural to retaliate but ought I to turn the other cheek? What is striking about most standards of right and wrong is that they leave no place for desert. Of course, a pluralist's standard, such as W. D. Ross', does,4 but that is because it can accommodate anything that intuition prompts one to add. But other tests, tests which look to a unified standard based on goodness perhaps, or on a fount of morality such as rational autonomy, have no place for desert. This is true even of Kant's Categorical Imperative test. Though Kant has the strongest possible views about the obligation to punish wrongdoers,5 it is hard to see how he gets them, or indeed any view that makes it an obligation to respond in certain ways to desert, out of the Categorical Imperative. I shall come back to this shortly. And modern contractualists, whose test is close to Kant's, have no place for desert either.6 Contractualists look at the world through the self-interested eyes of the worst-off representative person, and our obligations derive, though sometimes only indirectly, from what is in his interest. So when it comes to rewards and punishments, a contractualist will look to incent- ive, deterrents, and protection, perhaps checked by rights to liberty. But these are all inappropriate to desert; they are the sorts of extraneous considerations that destroy response to desert as desert. Of course, the same is true of the utilitarian test, according to which incentive, deterrents, and protection matter but desert does not. Either these tests are all too narrow or desert does not belong in a test of right and wrong. That is what I want now to try to decide. 3.Is merit? Suppose someone achieves something quite out of the ordinary; I admire it and praise it. Suppose it benefits mankind; I am 259 DESERT grateful and would happily join with the rest of you in honouring and rewarding him. It is the most natural thing in the world for these private responses to grow into public institutions, giving voice to responses that, if left private, might never find adequate expression. Persons are morally equal but are not equal in the way we regard, honour, or reward them. All of this is natural, reasonable, and not to be inhibited. A good society will, for many reasons, be rich in these diversities. There need not be anything misdirected, out of proportion, confused, or irrational in any of it. Indeed, what would be wrong is not to respond. But what sort of wrong would it be? Sometimes it would be unjust; we might honour a minor benefactor while ignoring a much greater one. But that answer does not go far enough; we might ignore all benefactors equally, which would be wrong simply because they deserve our thanks. Desert is itself a justification, not needing to be bolstered up by another. It is right to thank benefactors simply because that is the response that is fitting. It is wrong to ignore them because that response does not fit their acts. Talk about "fittingness' is, I know, sometimes dismissed as obscurantist. Reasons for actions, some say, have to be on the order of utilitarian reasons. But this is just fiat. Desert is a reason on its own, governed by criteria that are far from obscure. What would be entirely wrong would be to try to introduce utilitarian reasons into desert.7 As we have seen, it destroys a response to inject extraneous considerations into it, and utilitarian reasons are extraneous. Authenticity is not merely the best or the purest form of responses such as admiration, gratitude, or appreciation; it is the only form. A person who utters words, or does acts, of admiration, gratitude, or appreciation only on utilitarian grounds becomes a person without admiration, gratitude, or appreciation. If utilitarian- ism has no place for desert, desert has no place for utilitarianism either. However, that these responses to merit are natural, even fitting, does not yet mean that merit is a moral reason for action. Is it? Ross thought it clear that gratitude was sometimes a moral duty. When I read my colleagues' manu- scripts for them, I hope they are grateful. But if, when I send mine to them, they read them only because they now see 260 MORAL IMPORTANCE themselves as obliged to, they may be helping me, but it is far from clear that they can also be expressing their gratitude. Perhaps, when it comes to desert, even moral duty is extraneous matter the injection of which destroys the response. Certainly, if you, as depressingly many people do, turn the help I give you into a debt which I can collect, then you destroy gratitude. You transform it into exchange. You destroy my act as a favour, and you destroy your response as thanks. Repayment is certainly not gratitude. And duty does not look much like it either.8 Perhaps gratitude is not quite the thing for us to be focusing on. Gratitude is a feeling, a motive, a source of action, while actions, not feelings, are the primary subject of moral duties. Still, if "gratitude' is not quite the word for Ross to have used, it is clear that he had an act in mind---the act of returning good for good. It may destroy your motive as one of gratitude to inject considerations of duty into it, but none the less perhaps you morally ought, considering that I read your manuscript, to read mine. But if we now shift our focus to acts, there are several easy confusions that we need to avoid. It is easy, in the way I mentioned a moment ago, to transmute, without being quite aware what one is doing, a good turn into a process of exchange and so a grateful act into a debt. But the fact that a great many persons make returning good for good into a duty gives the rest of us no reason to think that it really is. It is also easy to confuse the moral assessment of agents with the moral assessment of acts---agent-morality with act- morality. We are now interested in the second, and in the first only to the extent that it bears on the second. And it is easy to slip unawares between deontic categories---for instance, what is morally admirable but not a matter of right and wrong (say, an act of heroism), what one ought to do but is not strictly an obligation (say, a simple act of charity), and what one must do (say, keeping a solemn promise). Unfortunately moral philo- sophy has not yet come up with any clear definitions of deontic categories, nor indeed any satisfactory argument that we need a distinction between, for instance, what we merely ought to do and what we strictly must do. But for our present purposes we should concentrate on a fairly weak deontic category: ought one morally to return good for good? Does the 261 DESERT fact that I read your manuscript enter into the set of considerations that determines what would now be morally right or wrong for you to do? It is hard to enlist intuition on either side of this question. Suppose that you feel genuinely grateful to me and yet, at the same time, follow a policy of reading only the manuscripts on which, in the time you have, you can help most and that I do not qualify. Would you be doing wrong? You would not, it is true, be ruled by feelings of gratitude; you might also be inhibiting the action towards which these feelings naturally prompt you. But that hardly settles the matter. Even in cases in which good turns pile up mountainously, as they often do in parents' help to a child, it might not be wrong for a child to ignore them in deciding what to do. If the child's policy is to help where he can help most so he leaves home for field-work with Oxfam, does he do wrong? Certainly if you did not want to do me a good turn for reading your manuscript, something would be wrong with you. And if a child were not deeply moved by the needs of parents who have been good to him, something would be much more seriously wrong with him. And there is a good case for thinking that the wrong in each case would be moral. But some of the morality that enters here looks very much like agent-morality. Since feelings connect with emotions, what feelings we have can, of course, be of great moral importance. To be deficient in certain feelings is to be insensitive to, in certain ways unsympathetic towards, other persons. How one relates to persons emotionally is part and parcel of having the dispositions that enable one to act morally towards them. And failing in feelings of gratitude, while not dead centre of this set of moral emotions, is still a member of it. But this is not the sort of link with morality that we are after. That there is a link between one's feelings and generally desirable behaviour does not show that one's having received a good turn is to be placed among the criteria of what it would be morally right or wrong for one now to do. That morality needs certain emotions does not mean that it would be morally wrong ever not to follow them. Admittedly, any child whose feelings are in proper order will have deep emotional ties to parents who have done him great good. Such ties grow naturally, and they, along with other deep personal 262 MORAL IMPORTANCE relations, constitute one of the central prudential values of life. A child would need a strong reason to free himself from the ties and the actions to which they lead. It would take something like the chance of saving others' lives in a famine, and not just a chance of conferring some marginally greater benefit, to justify a child's going against the prompting of these deep ties to his parents. They are too large a part of what makes the lives of all of them valuable to be put aside lightly. The wrongness of ignoring those ties does probably qualify as a moral matter, even as a matter of act-morality. But the reason for this has nothing to do with a duty to return good for good; this link with morality is still not the kind we are looking for. Indeed, it is because feelings of gratitude are so important that a duty to return good for good is suspect. It would make for a dreary life to have the duty replace the feelings, and following the duty would have the effect, in the way we saw, of undermining the feelings. It is probably better to live a life in which the feelings, and not this duty, are at work. Justice, I think, is a better case to press than gratitude. I said before that desert is a reason on its own, not needing to be bolstered by other reasons such as justice. But the cases of desert that most have an aura of moral right and wrong seem also to be instances of justice. We have to remember that sometimes desert rests on the obligation of equal respect, and then, of course, desert is a case both of justice and of moral right and wrong, but that is not the sort of desert that we are interested in. We are interested in merit. Is merit a matter of justice? Having an Order of Merit does not make Britain juster than the United States. Awards are not a moral matter (though giving them unfairly is). Rewards, however, look more like one. Is it not unjust not to give more---say more pay---to one who meritoriously contributes more? Are not American universities juster than Oxford, because their salaries recognize meritorious production while Oxford's recognize only the passage of time? Some people display determination, dedication, or industry, while others around them, who are like them, do not. A person expects, at the end of the day, even if he worked freely and unselfishly, to be treated with some regard for what he did. He wants to be acknowledged for 263 DESERT what he is, and what he does is much of what he is.9 But what should we conclude from this? No one would deny that there are powerful arguments for differences in pay: incentive, compensation, efficient distribution of labour. But merit lacks their obvious power. Oxford salaries are not obviously less just than American. Where merit has overwhelming power is in an argument for differences in response, even in reward. But rewards need not be money. At the end of the day, a person does indeed want to be acknowledged for what he is and what he does. And we, his partners in the enterprise, would be wrong not to admire his achievement and to be grateful for the benefit it confers on us. But the wrong need not be a moral wrong. All meritorious benefactors---good workers, social reformers, poets, musicians---deserve these forms of recogni- tion. But admiration and gratitude have, in the end, to be in people's minds. What societies may choose to do can only be tokens. Nor can admiration or gratitude be commanded, even on moral grounds, without injecting destructively extraneous matter. And they lead naturally to actions, to forms of respect and recognition. Now some societies are content to leave it at that. Other societies believe that it is better to give these spontaneous workings an institutional form. But few people who prefer institutional expression think that leaving it to spontaneous expression is morally wrong. Admittedly, if we leave it to spontaneous expression, and if we are all brought up extremely undemonstrative, some meritorious benefactor might go all the way through life not knowing how much he is appreciated. That would be the greatest of pities; it would show a great fault in us. We ought, in some way, to have made him understand. But it seems best, all things considered, to say that we ought because only that is fitting and its omission unfitting, and let it go at that, and not make it into a moral wrong.10 Sometimes, however, the case for institutionalizing a response is irresistible, and the arrival of an institution brings along with it a new set of moral issues.11 The appropriate response to a student who does well is admiration and praise, and to a student who does better it is greater admiration and praise. By institutionalizing these responses in a system of marks, we do more than just praise. Marks become statements: 264 MORAL IMPORTANCE they inform whomever it may concern that Sally is a little better than Paul, who is a lot better than Ann. The core of the institution, however, remains a response to achievement. What is extraneous to the response remains extraneous to, and potentially destructive of, the institution. Utilitarianism is still destructively extraneous. And for any institution of great value it matters greatly that we separate clearly what is essential to it from what is extraneous. Of course, nearly all of us accept that an extraneous consideration may come along weighty enough to outweigh the demands of the institution, but it will still remain extraneous. Yet, once there is an institution, giving a student a mark that he does not deserve is a plain injustice: it deceives and it can injure. And the institution consists in criteria of success that contain a promise: if you do this, you get that; so entitlement also enters. Clearly, then, moral obligation enters with the institution. But it does so, because honesty, injury, and entitlement enter, and none of these is the sort of desert that concerns us now. Is merit then a moral reason for action? It seems not. Even allowing that people are responsible for what they do, even giving them all credit for what they do, it is hard to find a case for turning that credit into a criterion of moral right and wrong. This might look puzzling: what sort of reason for action could it possibly be, if not moral? It is not a prudential or aesthetic reason; it is not a reason arising from personal aims. Still, there are, I think, other sorts of reasons. Another person's merit is a reason for me to respond in a certain way simply in virtue of its being the appropriate response: it is not just that my feelings would be functioning badly if they did not prompt me in this way, but also, beyond that, that my response, my action, would not otherwise fit the situation. The element of appropriateness itself constitutes a reason. 4.Is my own demerit? How do I respond to my own wrongdoing? I find that I am a pretty willing retributivist in my own case. Only occasionally do I see myself as a suitable subject for manip- ulation. I do, I suppose, when the trouble is a relatively simple bad habit that hypnosis or aversion-therapy would 265 DESERT break. Still, some wrongdoing, even if it could be stopped that way too, is more complex. I do not fully understand its origin, which is a lot of the trouble. It comes from my not seeing things straight: from coarseness of perception, lack of imagina- tion, atrophy of feeling, bad values, self-deception, insecurity. In those cases I do not know how I should want the manipulation to work: will it cure one fault and make others worse? There is no alternative but to see things straighter. To manipulate myself, or to get others to do it, with sufficient fineness of control would require that I already have so grown morally as to make the manipulation largely unnecessary. Anyway, only with such growth would I stay autonomous; I am not prepared to turn my moral fate over to someone else's idea of how I ought to be functioning. I do not suppose that I am at all unusual in this. For most of us, therefore, there is no alternative to moral growth. That is what makes certain responses to wrongdoing not only natural but fitting. I feel guilty: I know that what I did was wrong, that it matters that it was, that I am responsible; and I therefore feel "depleted of energy, unable to stand upright and partake of life'.12 And if I see that I have acted not only wrongly but shabbily, I also feel shame. Sometimes I might be able to decide that I was so frightened or threatened or provoked that I could not have done otherwise. But sometimes I can find no excuses. Is there anything inappropriate in my then feeling guilty? If what I did was wrong and matters and is inexcusable, it is proper that I should sense its wrongness, fatefulness, and its source in me. What would be inappropriate is not to react like this; a person without these responses would fail in perception and understanding. And the fact that the perception is charged with feeling does not make it any less a matter of understanding. The understanding would fail on its own terms if it were not highly charged. But is it right that I should also be unable to stand upright and partake of life? Well, it is hard for my understanding to be adequate without these effects following naturally. And the period of suffering and detachment has value. I cannot easily take part in life when I am absorbed in coming to terms with what I have done. The human frame supplies only so much psychic energy. The sense of crisis, of the impossibility of 266 MORAL IMPORTANCE carrying on without first coming to terms with what one did, are important in bringing about deep change. None of this is to dismiss these natural responses as merely serving some practical purpose, as utilitarians sometimes do. What they often have in mind is that guilt has no value in itself but can be a useful handle for steering behaviour. However, guilt is not just an instrument of manipulation; it is a response the appropriateness of which must be come to terms with. Yet any case for the appropriateness of guilt will also show that it is often inappropriate. It is notorious that feeling guilty and being guilty often disastrously part company.13 But this ought not to raise any general mistrust of guilt feelings. For instance, it would be a superficial psychological theory that claimed that all guilt feelings are pathological. Sometimes I see my wrongdoing all too plainly, and I hold myself responsible, and guilt feelings follow. If my beliefs change in these cases, my feelings change accordingly; there is nothing pathological. If one wants to argue that all guilt feelings are pathological, one has to attack at a very deep level. One has to argue that our using the concepts "right', "wrong', and "responsible' to organize our understanding of ourselves is itself mistaken, a symptom of the same pathology. However, for our present purposes the greatest flaw in guilt is this. The most appropriate outcome of the perception of guilt is moral growth. But what if further growth is not possible? Suppose that I see that my wrongdoing was genuinely just a slip that I shall be able to guard against in future. Or suppose that I have gone through the painful moral growth and am really now changed. It would seem in each case that then I should shake off the guilt and go back to my normal life. That is what we might call the Repentance View, according to which the appropriate pattern is: perception of wrongdoing, guilt, and finally repentance and correction. According to it, once I have gone through that process, I do not have to do more; that---neither more nor less---is the appropriate response. Yet if I go through the process quickly, though I have repented, have I yet atoned? When I confess to a priest, he may accept that I am already a changed person, but he still gives me a penance. Atonement is the purest form of the wages 267 DESERT of sin; the wages do not depend on moral growth and, if one believes that Jesus atoned for our sins, they can even be paid by someone else. On what we might call the Atonement View, the appropriate pattern has one last stage: perception of wrongdoing, guilt, repentance, and (the last stage) payment of the moral debt. Let us think for a moment about the relative merits of the Repentance and Atonement Views. Dostoevsky, the most powerful advocate of the belief that criminals themselves feel the need for punishment,14 might be enlisted on the side of the Atonement View. But he is really, I think, in the opposing camp. There is a strong widespread feeling, a feeling which moves Raskolnikov in Crime and Punishment to turn himself in, that one must suffer to atone for wrongdoing---not suffer to grow, but suffer simply to atone. But that is not where Dostoevsky leaves things. Raskolnikov goes to prison ready to atone but not prepared to repent. On the contrary, at first in prison he sees his crime merely as a blunder; he simply lacked the psychological strength to carry off his vision of a masterful Napoleonic life. But eventually, through the example of Sonia, who follows him to Siberia and devotedly attends him, he begins to see the hollowness of the vision. He sees how empty his life would have been even if he had managed to rise to truly Napoleonic force. The displays of power he valued were not the opposite, but the obverse, of his weakness. His problem was not impotence but emptiness, and the murder he commit- ted was an exercise of power that left him even emptier ---"agonizing, everlasting solitude'. And he finally finds in the possibility of his feeling toward Sonia as she does towards him a way in which his life might take on the substance that it has lacked. It is only then that Raskolnikov, not having before seen any point in his punishment's coming to an end, looks forward eagerly to release. It is spring, just after Easter; Sonia had recently read him the story of Lazarus. Raskolnikov, Dostoevsky says, "had risen again and knew it'. So what Dostoevsky has to say about guilt and punishment is, as one should expect, far subtler than the Atonement View. He says, with Nietzsche,15 that guilt can be an illness like pregnancy---its outcome a new life. He says that the criminal's need for punishment, if it takes the form of a desire for 268 MORAL IMPORTANCE atonement, can be superficial, incomplete, and self- interested.16 He sees that it can be completed by becoming repentance, and that without some such completion a crim- inal's need for punishment will be propelled by the same destructive forces inside him, only now turned against himself, that led originally to his crime. Once Raskolnikov has changed, atonement is of no interest to him; it is merely the social penalty to be got through.17 One way to test the Atonement View is to consider the question, If Raskolnikov had been condemned to die, would he, after his moral change, have been able to see his execution as a further appropriate, required, event? Of course, in some moods, especially with certain examples at the front of our minds, we are ready to say yes. If all the Nazis on trial at Nuremberg had in the end truly repented, not all their judges would have been willing to let them off. Still, despite the power of such examples, the Atonement View threatens our whole grasp on the concept of the appropriateness of responses to wrongdoing. The Atonement View focuses on wrongdoing. The most appropriate object for our response, however, is something more complex: what a person did---both wrongdoer and wrongdoing. But then, when I repent and change, I create a problem. Now that I and my act have in a way come apart, to which should one respond? According to the Atonement View, the appropriate object is still the act (or, perhaps, the evil of the act). But this sets up a strain. It divorces the appropriate object from the enduring moral person with whom one now can deal. I would not want such a divorce in my own case. I should want people to respond to my wrongdoing by regarding me as accountable and as able to come to terms with my own wrongdoing. I should want that respect, and also concern and help. But if I change, I should want them to respond now to the person I am now. What leads people to the Atonement View is, I suspect, an attempt to match up my response to my own wrongdoing with other persons' response to it. The two responses would certainly not be identical. My response would be: guilt, repentance, and (according to the Atonement View) paying the penalty. Your response would be: resentment, anger, sometimes also fear, avoidance, self-defence, retaliation. Both 269 DESERT of these chains of responses can be appropriate. At a certain point they converge: your retaliation can be my penalty. If both responses are appropriate, then it seems necessary for me to pay a penalty; otherwise your appropriate response, retaliation, would be left without the partner it needs in my appropriate response. But when I do repent and change, a lot in your response loses appropriateness. Cer- tainly fear and self-defence do. So, I think, would anger and retaliation. Your anger is aimed primarily at the com- plex of my person and my act. But now I have changed. Why not now accept, as the aphorism puts it, that "He that repents is angry with himself; I need not be angry with him'?18 The slippery moral notion of what is "fitting' gives no help here. To be used correctly it has to be tied to our notion of an appropriate response. Admittedly, some- one might insist that only Raskolnikov's death would fit the intensity of people's reactions to the murders he com- mitted. But the intensity of people's reactions has no inde- pendent moral status. It has itself to be appropriate, and will be appropriate only if it fits Raskolnikov's wrongdoing. That has to be the primary judgement of "fittingness'. Nor will the elusive, near-empty notion of "respect for persons' help much. Kant relies on this notion to justify his allegi- ance to the Atonement View. But it does not do it. The alternative to the Atonement View is not to deny a person the dignity of being morally accountable. To think so is to succumb to the false dichotomy that bedevils this subject. There is nothing mechanistic, manipulative, degrading, or lacking in respect in the Repentance View. The Atonement View is not the only way to show respect for persons. And once the Atonement View is separated from the notion of "respect for persons' and stands on its own, it looks both inappropriate and empty. On its own, there is nothing characteristic to it but insistence on the collection of moral debts. If anything, it is the Repentance View, in which a person is given more weight than an act, that shows respect for persons. Well, then, is my demerit a moral reason for action? In the case of merit, there is no move from a response's being appropriate to its being a moral reason. But demerit is 270 MORAL IMPORTANCE different. Though both reward and punishment are responses with standards of appropriateness, punishment is a response to something moral and gets its moral status from that. You respond to my wrongdoing with self-defence and retaliation. You do it to protect rights that I infringe. It is morally wrong of me to infringe them, and this gives you a moral reason to stop me. And I respond to my own wrongdoing with repentance. If I ought not to have done it, I ought to do what is necessary not to do it again. So there is a moral reason to punish. But is it based on my demerit? The analogue of the claim that one must reward for the good done is the claim that one must punish for the wrong done. But the ground of the moral reason to punish turns out to be something different from pure desert. It is protection of other persons' rights and my repentance and change. The effect of choosing the Repentance View over the Atonement View is both to give desert a large role, but, so far as it is to have moral force, to restrict it to its workings in repentance and change. So if the claim that one must punish because of the wrong done is taken to mean that we must take the further step represented by the Atonement View, then there is no reason to agree. My demerit is a moral reason only within the limits imposed by the Repentance View. 5.Is the demerit of others? Whatever is right for me is also right for anyone like me in the relevant ways. If, say, you punch me on the nose, it is all right for me to punch you back---with certain provisos. So long as you are in fact like me---capable of repentance and reform ---then my response has to be appropriate to you as a possessor of these capacities; it may not go beyond what is appropriate to your exercising them (e.g. no pure atonement); and it must be in proportion to your wrongdoing. Turning the other cheek would be right only if it spoke to, challenged, helped transform, the wrong in the wrongdoer; otherwise it would be not noble but narcissistic.19 Whether I may punish others all depends, therefore, on what the wrongdoer is like. If the wrongdoer is an exceptional moral agent who spontaneously repents and reforms, there is 271 DESERT virtually no place for punishment. If instead he is a normal sort, whose response is confused, ambivalent, and weak, punishment may provide him with the urgency to find the clarity, the unity of purpose, and the strength. If he is an adult who still needs to learn but resists the lesson, then maybe punishment would teach the seriousness of wrongdoing where words alone would fail. If he is someone who cannot learn no matter what the lesson, then punishment has no place as a response to the person that we are dealing with. The hardest claim to accept, I think, is the last---the response to someone who cannot come to terms with his own wrongdoing. Punishment takes the form of a dramatic acting out of a spiritual movement that ought to be taking place in the wrongdoer: a wrongdoer naturally suffers guilt and cannot partake of life. Punishment gets moral force from being an external form of, a symbol for, this internal process.20 But then confusion is almost irresistible: symbol gets mistaken for symbolized. We start to think that to be punished is actually to go through what one ought to go through. But if the wrongdoer is incapable of going through the internal process, then, putting the interests of other persons aside for the moment, we have no reason to put him through the external. Why use symbols when there is no one to communicate with? The symbolic status of punishment no doubt also partly explains our intuition that, besides deterrents, protec- tion, and reform, a further justification of punishment is its being fitting or appropriate. We think it fitting not only that a wrongdoer be unhappy but that everyone recognize what he did as wrong, and the strongest form of recognizing this is by making him suffer. But this is pure confusion. It is fitting that he suffer and that what he did be recognized as wrong. But making him suffer does not always, and often will not, make him suffer in the way he should. It is an entirely different form of suffering, which needs an entirely new justification. One does not transform a monster by imposing upon him for a while the outward trappings of the inner life of a beautiful soul. All of that concerned our response to the person of the wrongdoer. Part of our response to any wrongdoer, however, looks to ourselves. Of course, it is all right for me to protect 272 MORAL IMPORTANCE myself and others. That needs no arguing. So for now I shall just acknowledge it and say more about it later. Before moving on to the large social scale, let me sum up what lessons I think we should carry over into it from small scale interpersonal dealings. Punishment has a rich life of its own, quite independent of social institutions. It starts with the appropriateness of a person's response to his own wrongdoing. This, since it is purely a one-person affair, is not yet what can be regarded literally as "punishment'. But it influences the appropriateness of interpersonal responses to wrongdoing, which certainly can be called "punishment'. The appropriate response to wrongdoing is: perception of wrongdoing, guilt, and repentance. That is true in the one-person case; it is also true in the interpersonal case. My response to your wrongdoing is appropriate only when, and to the extent that, it contributes to your going through the same process: perception, guilt, and repentance. Considering only the person of the wrongdoer, punishment has no place at the two extremes: with a good person who spontaneously reforms or with a bad person not capable of reforming. It has a place only with the majority of us, who fall between the extremes, but then only as part of our coming to terms with our own wrongdoing, which severely limits it. But considering the victims as well, we may all defend ourselves. This means that demerit on its own---pure payment of moral debt---has no moral force; it gets it only through its connection with values such as good or right. It connects with good through protection of would-be victims. It connects with right through reform of wrongdoers; reform gets its importance not from making the quality of a wrongdoer's life higher but from making his acts better. Prudential values are not the only ones with a place here; there is also the moral value of doing right and doing what is necessary to do right. 6.The social response As soon as punishment becomes a social institution, it becomes in an important sense less pure. It is used not just to respond to wrongdoing but also to solve a lot of social problems. Let me just tick off a few of the extra jobs that punishment takes on. First, punishment is put to the rather different job of 273 DESERT teaching, a job that it does well. It is an emphatic denunciation of crime; it is a symbol of society's reprobation.21 However, its message is not always or primarily directed to the individual actually being punished, but to society at large, and these two interests may well diverge. Second, punishment is put to the job of deterrence. Having the rule, "Anyone who does this gets that as punishment', deters only because it is meant seriously, which requires punishing offenders. Third, punishment is pressed into the service of reform, because it is a good way of putting new pressure on the psychic mechanism and because it will serve as an occasion for autonomous moral growth. But reform, especially if it is moral growth, is a reasonable aim only in some cases. Fourth, punishment is put to the job of public relations. People have not only to be taught and deterred and reformed, but also to be reconciled to living in peace with one another. Justice has to be seen to be done. Since most persons do not notice fine details of particular criminal cases, and since judgements about the amount of fault are subtle and contentious, there is strong pressure to punish kinds of acts rather than particular ones. But all of these additional jobs assigned to punishment make it less and less pure. That is to say, they do not sit at all comfortably with punishment seen merely as an appropriate response to wrongdoing. The large-scale social perspective is hard to reconcile and harmonize with the small-scale interper- sonal perspective. Punishment, from the small-scale perspect- ive, can accommodate the job of reform, at least the sort in which the wrongdoer comes to terms with his own wrongdoing by repenting and changing. But further jobs such as protection, deterrence, and education, if they become too central to the institution, will destroy it as punishment. Yet protection, education, and especially deterrence, are important---some would say the important, even the justifying---function of a social institution of punishment. Once punishment becomes a social institution we seem to have a conflict on our hands between the small-scale perspective in which guilt appears at the centre of the picture, and the social perspective, in which solving co-operation problems is central. It would certainly be a mistake to make what dominates the small-scale perspective---wrongdoing, guilt, repentance 274 MORAL IMPORTANCE ---dominant in a social institution of punishment. Parking on a double yellow line brings a fine, but unless it is also dangerous guilt and repentance are inappropriate. And society has as one of its chief functions fostering and protecting beneficial co- operation, and that requires sanctions for welshing on agree- ments or for not doing one's bit. The most successful strategy for an individual in iterated, few-person Prisoner's Dilemma games is what game theorists call Tit for Tat: to co-operate initially, to be quickly provoked into punishing the other player's defection by defecting oneself next time, and then to repay co-operation with co-operation and defection with defection.22 Co-operation on a social scale would be even more efficiently promoted if society as a whole were provoc- able by deviations from its rules. Though rules do not get their whole justification from providing solutions to co-operation problems, they get much of it there. The social institution of punishment has a rationale independent of guilt and repent- ance. It comes from the causal chain: certain great social benefits require conformity; conformity requires sanctions; sanctions require punishment. This clash between large-scale and small-scale perspectives is connected with a clash between retributive and utilitarian justifications of punishment. The widespread belief that any entirely satisfactory justification will have to combine elements of both retributivism and utilitarianism could then be re- phrased: it will have to manage to see its subject from both of these perspectives. And this is not going to be easy. What is dominant in one perspective destroys what is dominant in the other. The social institution of punishment serves very many interests, which are not easily made compatible. The various elements of the structure can, it seems, be kept in balance, but it is an extremely precarious balance. The social institution of punishment, it seems to me, has some such structure as this. At its core is one person's response to another's wrongdoing: the deliberate imposition of an unwelcome state in return for wrongdoing, where the notion of a "return' carries with it a rough notion of proportionality. Next, outside the core but still close to it, is the job of bringing a wrongdoer to his senses. Next, shifting attention from the wrongdoer to his future action, comes the job of protecting 275 DESERT potential victims. Next is the job of deterrence, first of the wrongdoer himself and then, at one further remove from the centre, of others. Next, at some distance from the centre, comes the job of education, teaching others or raising the general moral tone. The core and what surrounds it are in unstable relation. If what surrounds it displaces the core, the institution collapses. Even something as close to the core as self-defence, if it becomes too much the centre of society's activity, will destroy the activity as punishment. If I sit on someone to stop him from attacking me, and if I go on sitting on him because each time I go to get up he starts again, I am not punishing him. It is a perfectly appropriate response, but since the core is missing, it is not punishment. And a social institution centred on self-defence would be concerned with anticipating, prevent- ing, warding off wrongdoing so that it never becomes actual, not with responding to it once it does. The same is true of every other surrounding function of punishment: reform, deterrence, teaching, public relations; if any of them gets too close to the centre, it destroys the whole institution. However, these surrounding functions can hardly be dispensed with. They are what justifies punishment. Desert, which appears in the core in the notion of a "return', is not a justification. It played no role in the justification worked out earlier, so can play none here; the introduction of a social institution opens up no new place for it. So if there were no surrounding functions, there would be destruction of another kind. Justified punishment needs both core and surrounding functions. If no core, then no punishment; if no surrounding functions, then no justification. Once we give up the Atonement View, as many Kantians and all utilitarians do, then we face a problem. All of the justifications of punishment appeal to features that are not only inessential to but also, if they loom too large, destructive of its status as punishment. This has been recognized as a problem for utilitarians, but it is far more general. The misimpression that it is particularly utilit- arians who are in trouble may be helped along by the fact that the important modern Kantians, the contractualists, generally avoid the whole messy subject of punishment.23 How far can one go? How far can one increase the 276 MORAL IMPORTANCE importance of the surrounding functions, which are after all what justify the whole business, without destroying the core? We should not have gone too far if, say, we were to do the following. If we had a wrongdoer who could be brought to his moral senses whom we punished with that hope, then we should be acting in return for his wrongdoing and also in hopes of his changing. The core survives alongside its extra function. But exemplary punishments usually go too far. To survive as punishment they have to stay within the bounds of an appropriate response to the wrongdoing. There is a rough proportionality built into the essence of punishment---very rough indeed, hard to make precise, but undeniably there.24 A state could hang shoplifters. The authorities could even call it "punishment', but the rest of us would not go along with it for long. It is no longer an institution whose purpose is to make return for the shoplifting but becomes one whose aim is controlling behaviour by fear. The same applies (although it is a good deal harder to argue) to capital punishment. It is certainly never an appropriate response to the person of the wrongdoer---not if he can repent and certainly not if he has already repented. Its justification has to be the general deterrent effect of the rule, "if you do this, you will be executed.' For the most part, attention focuses on whether such a rule does really deter, but it should also focus on whether what would successfully deter would also succeed in falling within the range of responses appropriate to the particular act. For if it does not, then the desert at the core is missing. Still, desert, I want to say, is only the core and not a justification. But then all the worst fears seem to rise; all sorts of abuses seem to be allowed. Punishments need no longer fit the crime. Worse, we may punish the innocent. But these fears, which probably account for a lot of allegiance to desert, are caused by misunderstanding the different roles desert can play---as core and as justification. Does the punishment no longer need to fit the crime? May we impose a penalty of whatever size proves to have best effects? No, as we have just seen; and no, because of desert.25 Punishment, in its core, is a return for wrongdoing; a rough proportionality is part of it. It destroys the institution to let the 277 276 MORAL IMPORTANCE importance of the surrounding functions, which are after all what justify the whole business, without destroying the core? We should not have gone too far if, say, we were to do the following. If we had a wrongdoer who could be brought to his moral senses whom we punished with that hope, then we should be acting in return for his wrongdoing and also in hopes of his changing. The core survives alongside its extra function. But exemplary punishments usually go too far. To survive as punishment they have to stay within the bounds of an appropriate response to the wrongdoing. There is a rough proportionality built into the essence of punishment---very rough indeed, hard to make precise, but undeniably there.24 A state could hang shoplifters. The authorities could even call it "punishment', but the rest of us would not go along with it for long. It is no longer an institution whose purpose is to make return for the shoplifting but becomes one whose aim is controlling behaviour by fear. The same applies (although it is a good deal harder to argue) to capital punishment. It is certainly never an appropriate response to the person of the wrongdoer---not if he can repent and certainly not if he has already repented. Its justification has to be the general deterrent effect of the rule, "if you do this, you will be executed.' For the most part, attention focuses on whether such a rule does really deter, but it should also focus on whether what would successfully deter would also succeed in falling within the range of responses appropriate to the particular act. For if it does not, then the desert at the core is missing. Still, desert, I want to say, is only the core and not a justification. But then all the worst fears seem to rise; all sorts of abuses seem to be allowed. Punishments need no longer fit the crime. Worse, we may punish the innocent. But these fears, which probably account for a lot of allegiance to desert, are caused by misunderstanding the different roles desert can play---as core and as justification. Does the punishment no longer need to fit the crime? May we impose a penalty of whatever size proves to have best effects? No, as we have just seen; and no, because of desert.25 Punishment, in its core, is a return for wrongdoing; a rough proportionality is part of it. It destroys the institution to let the 277 DESERT degree be fixed by anything extra. That is what the institution is; that is the institution we are interested in. Considerations of utility maximization have no more place here than they do in marking a student's answer (though, as before, this is not to say that they have no moral place at all; nearly all of us accept that they can override the internal considerations). May we punish the innocent? Suppose a government took to trumping up charges of currency irregularities against political dissidents. Well, even if it did, it would not be a case where what justified punishment was also justifying punish- ment of the innocent. For reasons we have already seen, it would no longer be punishment. This is not a semantic point. A prison guard can well say to the dissident he locks up, "I am punishing you for something you didn't do'. For the word "punishment' to be used correctly, it is enough that there be some deemed wrongdoing. There is enough of the normal semantic background there for the guard not to have made a linguistic error, though also enough missing for him to feel a bit uneasy about going on talking in that way. And if this case is not isolated, if the state regularly uses the trappings of punishment for political suppression, people are likely eventually to stop calling it "punishment' or to shift to a new or obviously ironic way of using the word. Desert is necessary, in general, for the institution to be punishment and to be regarded as such. Political suppression is plainly not punishment. This argument is like one that, some years ago, Anthony Quinton used. He proposed that certain forms of retributivism be seen as a claim about the meaning of "punishment', and Herbert Hart later dismissed this as the ploy of putting a "definitional stop' to serious objections to utilitarianism.26 But my point is not quite Quinton's. He wanted to rest the weight of the argument upon the semantics of the word "punishment', and it would not carry the burden. I rest it on the character of the institution of punishment, the rich connections it has in our life, with its accommodation of our natural and proper responses to our own and other persons' wrongdoing. The important point is about the institution the justification of which is in question. We need not follow what a prison guard or a man in the street would say. Our purpose is rather different. We are building a moral theory, one of the aims of 278 MORAL IMPORTANCE which is precision and the avoidance of confusion. Other institutions---e.g. institutions of political suppression---are not punishment. To talk about them is not to talk about punish- ment. What justifies punishment may not (often does not) justify them. And if it is replied that political suppression is justified by considerations out of the same stable as the ones now being used to justify punishment, what would this show? All that it would show is that, although we may have a good justification of punishment, we had better also have resources to prevent those considerations' going on to justify those other institutions---when, that is, they are not justified. But what if the abuse was not on a large social scale, but was a lone individual twisting the institution, unobserved, to some good end? Cases would, no doubt, be very rare, but life might sometimes take on the flavour of an Agatha Christie story. We can imagine a detective, in what looks like an open-and-shut case, getting the hunch that not all is as it seems. Green, let us say, had a flaming row with his partner Grey, who was heard to threaten him, and an hour later Mrs Green found Grey, smoking revolver in hand, standing over her husband's dead body. Everyone is satisfied of Grey's guilt, except our detective, who begins to find a trail leading to Mrs Green. But she has five young children who need her; Grey is a bachelor; it was probably a crime of passion. No one will know if the lone detective probes no further, and even if the truth eventually comes out, no one will lose faith in the police for not being as clever as Hercule Poirot. Still, strictly speaking, what will happen to Grey will not be punishment. It will not be the workings of an institution whose justification we are trying to determine. This may not seem to matter much; we could just call it pseudo-punishment instead. But it matters because, as calling it pseudo-punishment shows, it brings a change in subject. The real complaint is not that the Repentance View would justify punishing the innocent, but that it does not prevent pseudo-punishment. Why is that a complaint? Why should its job be to do that? It is only one part of moral theory and should not be expected to have all the answers. All that can be reasonably asked of it is that it is able to fit into a larger moral theory that does supply the rest of the answers. 279 DESERT There is a better way to resist the Repentance View. It is to take up a position somewhere between the Repentance and Atonement Views. Desert, one would say, allows punish- ing a wrongdoer (it does not require it; the permission may be subject to the restrictions in the Repentance View), but lack of desert prohibits punishing or pseudo-punishing an innocent person (it prohibits it more strongly than the Repentance View does, in which the prohibition comes from what is necessary for the institution's survival, while now it comes from lack of desert). We can call this the "Account- ability View'.27 There is an ambiguity in the Accountability View that needs clearing up right at the start. Should we interpret its prohibition as being absolute? The moral prohibition, Do not intentionally kill the innocent, has great moral weight, but there is the rare exception. It was right for the sailors to eat the cabin boy since otherwise most of them, including the cabin boy, would have died.28 Similarly, the prohibition, Do not execute the innocent, must have exceptions, and if it does, the vastly more general prohibition, Do not pseudo-punish the innocent, which covers actions ranging all the way from execution down to a #1 fine, is likely to have many more. The tendency in modern discussions of punishment is to set the demands far too high: a justification of punishment is often regarded as ruled out if it ever allows pseudo-punishing the innocent. But few of us are absolutists in general moral theory, and it is hard to see why we should be in this department of it. But if the prohibition is not absolute, the divisions become altogether more difficult to locate. It is easy to say how the Accountability and the Repentance Views differ. The basic idea of a non-absolute version of the Accountability View is that desert has moral weight and that it is wrong, other things being equal, to punish the undeserving. The Repentance View contains nothing comparable to this prohibition. So, according to the Accountability View, we need extra counterweight to balance the moral weight of desert, and therefore more weight to justify pseudo-punishing the innocent than the Repentance View requires. That is the theory. But when we get down to particular cases, it is not at all clear that we do require this extra 280 MORAL IMPORTANCE counterweight. Of course, there are examples that spark our intuitions. But is it desert that has weight in those cases? The Accountability View consists of two claims, a permis- sion and a prohibition: one may punish the guilty because they deserve it; one may not punish the innocent because they do not. But if, as seems desirable, the permission is interpreted in a way compatible with the Repentance View, the only point of difference between them will be the prohibition. But a theory of punishment is not the whole of moral theory. So although the Repentance View restricts pseudo-punishment of the innocent only so far as the institution's survival is at stake, the whole moral theory of which it is a part is likely to provide other restrictions. An obvious one is rights. I spoke of the status and moral weight of basic rights in the last chapter. Liberty is such an important right that it takes a very great moral counterweight to override it; and it protects an innocent person threatened with pseudo-punishment. Does the Account- ability View give any greater protection? If not, then the Repentance and Accountability Views merely offer different explanations of the same degree of protection, and we should not need "desert' to explain that degree. Think again of Mrs Green and Mr Grey. And suppose it was Mrs Green, after all, who did it. Then we have Mrs Green on one side, a woman who had something in her that led her to murder, who will be or should be torn by remorse and unable to carry on with life until she has come to terms with her own guilt, who needs punishment and whose punishment should end when she no longer needs it. Then, on the other side, there is Grey, who has no guilt to come to terms with but has a life to make as much of as he can. It is hard to see how, even without giving desert a moral weight, it could be right to let Grey be convicted. Of course, the case need not be so straightforward. Perhaps the legal system is corrupt; perhaps Mrs Green would get a much longer sentence than she deserved. That would then make the moral reasoning much more difficult; now it becomes, among other things, a choice between injustices. But the complication that interests us is different. To test the two views properly, we should need to take a series of cases, in which we added bit-by-bit to what rides on Mrs Green's 281 DESERT escaping punishment until we reach the point where they outweigh Grey's right to liberty, and then we need to see whether they are not yet enough to balance Grey's desert. Give Mrs Green not five children but ten. Add a child with a fragile psyche who would kill himself for the shame of it all. Have we got there yet? And does desert need still more? Since rights have instrumental value (protections of the greatest prudential values), we have some rough notion of how to weigh them. Desert is quite different. It is not just that our intuitions about its weight will differ greatly. That would be all right if, with thought, we could bring them into some order. But that is what is so hard to see how to do. And that is the result of its being so hard to attach a sense to the notion of "desert'. When it is what persons deserve in return for the good and ill that they do, we understand the rough proportionality involved. But negative desert, a person's not deserving punish- ment, when the prohibition is not absolute, introduces a different sort of proportionality. And, to fit into the structure of the Accountability View, the protection conferred by negative desert has to be different from the protection conferred by basic rights. That is the notion that is so difficult to attach a sense to. Let me sum up what I have to say about the social institution of punishment. It is a natural growth that owes little to conscious human decision. It has grown from two very different grounds. It grew up as a solution to certain co- operation problems. It grew up simultaneously as one of a large set of natural responses to the behaviour of others, each with its own criteria of appropriateness. We tend now to see conscious human decision as more influential than it is. An institution like punishment has not been, and could not be, created by a rule utilitarian or a contractualist decision about which laws and institutions to install in a society. One does not choose one's deep responses to other persons, and without them one would have only a pale imitation of the institution of punishment. What is more, although we might, we seldom do consciously choose a solution to a co-operation problem. The size and complication of the set of co-operation problems that a large society presents have meant that our solutions to them have come from many different sources, conscious decision 282 MORAL IMPORTANCE being only a minor one. It also means that one has to be cautious about grand-scale social engineering. It is not that we should not try to introduce much more conscious choice into our social institutions than is there already. It is, rather, that we are not the architects of our social institutions. We inhabit them rather as we would some natural shelter, and both our freedom of movement and our understanding of the structure are limited enough so that we can, at best, only alter them a bit here and there. These limitations cast more light on the nature of the multi-level structure of moral thinking. We are tempted to think that the practical, everyday level (what I earlier called the practical decision procedure) contains rules and institutions that we choose on the more detached, critical level (what I called the reflective decision procedure). But there is a sense in which institutions cannot be chosen. With the institution of punishment, a part of it is an appropriate response to wrongdoing, and an appropriate response is not the sort of thing one chooses. One could, it is true, choose not to have an institution of punishment at all, but that is the kind of decision that means knocking down props of that natural structure the architecture of which we scarcely fully understand. 7.Retributivism and utilitarianism There is a lot to be said for avoiding talk of "retributivism' and "utilitarianism' for as long as we can. Each term covers a confusion of views, many of which have attractions; and neither term adds much to our conceptual kit for exploring the middle ground. Utilitarians, rightly it seems to me, stress that punishment is justified only by being something more than a return for wrongdoing. But they tend then to concentrate on general prudential values (in some versions, simply happiness) and to ignore other values that are especially important to punishment ---namely, moral values. If what I have done is wrong, then it is right for me to do what is necessary to avoid doing it again, even apart from the effects on the quality of my life. True, "reform' is one of their justifications, but they tend to see it as change caused by social pressure, rather than as an agent's 283 DESERT autonomously coming to terms with his own wrongdoing. And they generally underestimate the role that desert plays in punishment. By doing that, they turn punishment into a social institution without depths, as if it were entirely the creation of social planners without life before they came on the scene, whereas it is its rich prior life that largely determines what they can do with it. Retributivists are right, it seems to me, to stress that desert has a central place. Still, it is not the large place assigned it in the Atonement View, nor even the smaller one in the Accountability View. Retributivists misplace desert. It belongs in the core. It is what makes sense of guilt, remorse, and repentance, without which punishment becomes a bloodless imitation of the real thing. One of the strongest arguments for retributivism has always been that it dignifies the person as someone capable of accepting responsibility, whereas utilitar- ianism demeans the person as one suitable for manipulation. But this argument is no longer available. There are alternatives to retributivism that do not deny moral responsibility. The trouble is that famous classical utilitarians have sounded rather toughly determinist,29 but this is only another example of the baleful legacy of psychological hedonism with its simplistic language of the push-pull of pleasure and pain pay-offs. However, one does not have to give desert moral status to make room for responsibility. One does that just by giving desert its place in the core. And that role for desert is compatible with utilitarianism. On its central claim about desert, indeed, retributivism is wrong. Desert does not require, permit, or prohibit punishment. More generally, although desert is often a perfectly good reason for action, it is never a good moral reason. Moral reasons have a special kind of urgency, which desert has not got. XIII DISTRIBUTION 1."Distributive' justice WE punish, reward, exempt, promote, and elect some persons and not others. But that is not what we usually regard as "distributive' justice. A transfusion can shift blood from someone who can spare it to someone who needs it. But that is not it either. What we usually have in mind, instead, is distributing material goods and services: food, clothing, shelter, tools, land, raw materials, education, medical treatment, labour. And in a mature industrial society in which land, factories, raw materials---tangible goods generally---become less important in generating wealth and government-granted licences, franchises, and benefits become more important, we also have in mind certain legal powers, titles and claims. And we have in mind, of course, the general wherewithal, money. What about happiness or pleasure? Happiness and pleasure partly turn on how one responds to goods, services, and powers, and one's responses, though affected by what is distributed, cannot themselves be distributed. All of this is a bit arbitrary. There is no best way to divide off the potential objects of distribution; the division will be in some measure stipulative. I prefer to keep the class fairly narrow: what is up for "distribution' are holdings: material things, services, certain legal powers, money. One reason for concentrating on holdings is that they specially lend them- selves to distribution. They can be easily parted from their holders; they can be amassed in excess of their holder's need; they are easy to shift from person to person. Holdings are, therefore, tempting to redistribute. Another reason for focusing on them is that someone usually claims them as soon as they come on the scene; they often owe their existence, after all, to some person's work. Though holdings, 285 DISTRIBUTION because of their transferability, attract redistribution, they also, because of ownership, especially repel it. No setting for distribution seems really paradigmatic; settings vary along dimensions all of which seem to matter to what we say about distribution. So each position on one of these dimensions can be seen as a value of a possible variable in the ultimately authoritative distribution function that we must now try to find. First of all, the nature of the group of persons varies. The group might be a family, or a commune, or a city-state, or nation-state, or the entire population of the world at a given time, or the world population over several generations.1 Will the moral principles governing distribution not vary as the group varies? Second, the nature of the holdings varies. Some, such as natural resources, are in effect manna from heaven. Others exist only because certain individuals have put a lot of work into making them. Can principles of distribution apply regardless of how holdings appeared on the scene? Third, the nature of the economy varies. There are autarkic economies, where people are able, perhaps even forced, to make things for themselves. There are exchange economies with specialization of labour and complex forms of co-operation. It is easy to identify the producer in the first setting but not the second. Will an individual claim to holdings not also depend upon what the methods of produc- tion are? Is even the notion of "property' stable across kinds of economy? Then the affluence of groups varies. With superabundance, Hume remarks, questions of distributive justice do not even arise.2 We might find that each of us may take what resources he wishes, leaving as much and as good for the rest. Or we might find that we would leave enough but not as much. Or perhaps, whatever we do, not everyone can survive. Hume also thought that when we get to the other extreme, when persons have to fight for survival, distributive justice is "suspended'.3 But the two extremes---superabundance and struggle for survival---are not symmetrical. Plausible prin- ciples of distribution just do not apply to superabundance; they do, however, apply to desperate scarcity---it is only our 286 MORAL IMPORTANCE psychological capacity to live up to them that characteristically runs out.4 So once below the level of superabundance will not persons' claims to holdings depend upon how much there is to go around? And are there any principles of distribution that apply to all levels of wealth, from comfortable abundance to desperate scarcity? The intersection of two or more of these dimensions defines a possible setting for distribution. One setting would be a modestly well-off family. It is tempting to treat the family as a paradigm for all distribution. What should we strive for if not the brotherhood of man? Why not then use in society a principle of distribution close to the one at the heart of the family: from each according to his ability, to each according to his need? Then the bulk of society's assets could be regarded as forming a common stock, and the notion of "private property' would have only a peripheral part to play. The principle of distribution that mattered would then be what has been called "patterned' or "end state', not "unpatterned' or "historical'.5 But at the same time it is most odd to take the family as a general paradigm. There is the obvious problem: why take the brotherhood of man as a political ideal if it is a psychological impossibility? Another setting for distribution, one that fascinated seventeenth- and eighteenth-century philosophers, is settlers landed on a new continent each of whom can take what resources he can use and leave as much and as good for the rest. In this setting a new principle appears: property. It is not that it is totally missing from families but here for the first time it becomes central. If I take and till a plot of land when the rest of you have perfectly good ones yourselves, most of us would be willing to say that I develop claims that the rest of you do not. While it is easy to see why a family would be taken as the paradigm for distribution, it is harder to see why an autarkic collection of New World settlers would be. Labour seems central to our conception of ownership, and autarkic gatherings provide relatively pure cases of one person's labour, which help to fix our thoughts about ownership. There is that much justification. But it has often been pointed out that there is nothing like seventeenth-century abundance any more. There is very little in the way of autarky either; what is 287 DISTRIBUTION produced now is usually the result of such complex interaction that it is hard to isolate one person's contribution. And what does it help to be clear about just acquisition, when most important original appropriations happened so long ago that we cannot now begin to recover all the information about them that we should need to decide about their justice?6 Autarky, because of its inefficiencies, will, where possible, give way to an exchange economy, which presents a new setting for distribution. Exchange usually brings with it the institution of money, which allows storing value and concentrat- ing power in a few hands. All the principles that apply to autarky apply in an exchange economy too. But there are, in addition, the principles that govern agreements. And there are also the standards of fair exchange. Equal respect would require that the parties to the bargain start from equal bargaining positions, and large concentrations of power threaten that. And once there is fair exchange, procedural justice enters.7 Any group with fair exchange will be prepared to accept that, up to a point, there are fair outcomes that cannot be described in advance---unpatterned ones, ones that are fair merely through being the outcome of a fair procedure. So although "end-state' principles still apply, "historical' ones seem to apply now too. Whatever case there is for making the abundant autarkic economy our paradigm applies even more to an exchange economy; the introduction of exchange is a small move towards present reality. But since the move is so small, the case against applies just as strongly too. The inefficiences of solitary production are likely to lead to a co-operative economy. Now, let us say, you organize, I prepare the raw material, and he assembles it. As products become more complex, techniques become more important and are borrowed freely from others. Life is better; population gets larger. Soon abundance of resources disappears. What I have inherited gives me a holding, but you arrive on the scene to find nothing left for you to appropriate, let alone as much and as good. What principles apply now? One important addition is fairness in participation. If I voluntarily benefit from co-operation, it would seem that I should not free-ride. But what is striking is the principles that no longer apply at all comfortably. The deepest requirements of equal respect, the 288 MORAL IMPORTANCE ones that seem to justify ownership, are now unmet: you do not find as much and as good for you. That hardly reflects badly on my distant ancestors, who probably satisfied the Lockean Proviso. Nor does it reflect badly on my nearer ancestors, who may also have satisfied it. The problem is ours and in the present: what may well be at fault is my now going on holding the resources. And products are no longer just my doing, or even yours, mine, his, and hers in identifiable proportions.8 Society now makes its own important contribu- tion; so does tradition. Not even the products of thought retain much purity. A medical researcher might make a discovery of great commercial value. He might have worked terribly hard to bring it off. But even so, who trained him? Who moved the subject to the point where the discovery became possible? Who built the lab in which he worked? Who runs it? Who pays for it? Who is responsible for the enduring social institutions that present the commercial opportunities? One who cleverly exploits the social framework has both his cleverness and the framework to thank. In short, life has got not only better but also so much more complicated that justifications for property, imported from simpler times, fit awkwardly. Anyway, the concentrations of wealth and power that a co-operative economy generates are likely to lead to central government's exercising, in the name of liberty, more and more control. And in a mature industrial society services begin to take on as much importance as manufacture. Land, machines, and raw materials become rivalled by government-granted licences (for example, to a doctor to practise), franchises (for example, to run a television network) and benefits (for example, to the support of the welfare state) as means of livelihood.9 Then our concept of property, developed in the days of the supreme importance of tangible goods, fits even more tenuously. Why should we have property rights defending our tangible goods against invasion and have no comparable rights defending what are for many of us even more important to our lives? It cannot be because governments granted the licences, franch- ises, and benefits; so did they in many cases originally grant the lands. There is an obvious reason for our taking a co-operative economy as our paradigm: it is where we are now. But that is no reason for us to treat it as paradigmatic in 289 DISTRIBUTION other settings in which distribution also takes place: our families, to take the very small scale, or the world divided between rich and poor nations, to take the very large. I shall stop there with my sampler of paradigms. A lot of confusion arises from our not appreciating how many and how heterogeneous the paradigms are. Some persons see most goods as ultimately part of society's common stock and so they look for moral principles to use in distributing them fairly.10 Others see goods as by and large coming into existence with name tags already on them and so they deny that society has any moral right to distribute them at all.11 Both attitudes are appropriate, but in different settings. And the setting that we are most interested in---a modern, fairly affluent industrialized nation-state---is just the setting in which it is hardest to tell how far either set of attitudes fits. So let us now concentrate just on that setting and ask, Which principles really apply to it? and, Are they moral principles? 2.Role-based principles There must be something that a person can do to earn himself some moral claim on holdings. If so, his claim will rest on what he did; it will be tied to a status that he has and that persons generally do not. But we need, as always, to find what precisely the status is that has these powerful effects. It cannot be merely "appropriator'.12 Even appropriation that satisfies the Lockean Proviso is not enough. Suppose I appropriate a bit of land, leaving as much and as good for the rest of you,13 but that, because of my constitution, I consume practically nothing. Everyone's equal share, let us say, even when worked, still gives a pretty meagre life, and if the rest of you could use my plot productively you would be an important bit better off. It was reasonable of Locke to add to his Proviso about as much and as good a second Proviso against waste.14 The two Provisos together, however, reduce the bare act of appropriation to insignificance. What is at work in generating the claim is some such general view as that everyone's well-being matters and matters equally. The act of appropriation on its own seems to be playing no role---not less than a full role but no role at all. Appropriation leaving as 290 MORAL IMPORTANCE much and as good seems more an application of a general rule of assigment of holdings---namely, everyone should have equal holdings for productive use---than an act that generates its own moral claims on them. Is the relevant status then "producer'? There is a long tradition in political theory that sees the act of mixing one's labour with resources as the key claim-generating fact. Now, there are ways of interpreting this view that make it quite ridiculous (for example, the one that Nozick guys: "If I own a can of tomato juice and spill it in the sea . . . do I thereby come to own the sea, or have I foolishly dissipated my tomato juice?').15 But there are much more sympathetic interpreta- tions. If I collect seed that would otherwise rot and plant it in earth that would otherwise go unused, and do not thereby deplete the soil, then the crop is almost entirely my doing. And it is hard to resist the thought that the more purely a product comes to being all my own work the stronger my claim to it is. What I have a claim to, then, is not the land I have used but the crop, the value that would not have existed but for what I did. It is true that some contributors to the happy outcome ---the seed and the use of the land during the growing season---were not my doing, but it would be the meanest dog-in-the-manger attitude to deny me the use of what no one else wanted in the production of what I definitely needed. My labour is in the crop, so in a way I am in it too. We move easily along the path of thought: "It's my doing; I am in a way in it; it's mine.' This sentiment is natural and deep and worthy of respect. It is a sentiment that, as has been pointed out,16 has even become well embedded in many languages in an etymological link between the words for "proper' and "prop- erty'. The sentiment is very likely also to have some evolutionary source; some form of property may be a human correlate of animal territoriality.17 The labour theory of property can be seen as what much good philosophy is: an attempt to put before us in an accessible, articulate form parts of common sense that are themselves shaped by powerful, natural sentiments. The weakness of the labour theory is not its ancestry but its scope. It does not get us at all far. If my claim is limited to what does not involve appropriation, I have claim to very 291 DISTRIBUTION little indeed. Even my using land for one growing season constitutes short-term appropriation. No product is purely my own work, devoid of trace of resources that others might have used. What comes closest is the product of thought, say a new technique that I think up on my own, but even then I use food and shelter while I do the thinking. Clearly, I ought to have some security of claim on what I need to labour on, but the labour test is largely silent on what my claims would be. If I till and plant land, may I claim the crop or the land? If the land, for how long? Does my power extend to transferring it to someone else? Anyway, we are talking about the modern world and, as we have seen, it is often dauntingly difficult to identify what a final product---even a lone scientist's discovery in his lab---owes to one person all on his own. The pure labour test is too indeterminate in any distributive setting and too tenuously connected with our actual setting. It is not that it has no weight: it gives a producer some claim on some objects in some circumstances. It is just that the labour test scarcely advances us beyond the vague intuition from which it starts. Perhaps there is something behind the labour test that is more determinate. Does the labour test really have a desert test at the back of it? Is it not the mere producing that matters but the desert it creates? The heart of the pure labour theory is the movement of thought, "It's my doing, so it's mine.' But I doubt that it could be desert at work here,18 at least in the strict sense of desert centred on actions sufficiently out of the ordinary to attract praise or blame. We should not want claims to holdings to rest on that, or else only the specially meritorious would have them. New World settlers, each harmlessly tilling his plot, may show no special merit to speak of, but they ought to have some claims on holdings. And, in any case, as we saw earlier, merit does not itself have moral force so it cannot confer it.19 I think that it is only another, broader use of desert that comes into play here. We use "desert' not only when we think there is special merit, but also when we think merely that something ought to happen. The movement of thought, "It's my doing, so it's mine', can be construed, "I did it, so I deserve to have it' (i.e. "I ought to have it'). This, I think, is the way to understand the argument at the heart of the pure labour theory. It rests not on merit but 292 MORAL IMPORTANCE on pure doing. Desert-qua-merit can be the ground of a claim about desert-qua-ought. But a claim about desert-qua-ought is just the assertion of what ought to be done, not a reason why it ought. Desert-qua-merit is not the right kind of backing for the labour test. Desert-qua-ought is not backing of any kind. Since each of these two tests, appropriation and labour, is inadequate on its own, the obvious thing to do is to merge them. Locke, for instance, used not a pure labour test, but one incorporating large elements of appropriation. However, this mixture, though better than either ingredient on its own, does little to cure either the indeterminateness of the pure labour theory or its tenuous relevance to the present. We are still left with the problem of getting beyond the truism that productive users have some claim on some holdings in some circumstances. How do we make these claims more determinate? What is going to make them relevant to the modern world? Role-based considerations are at best the shortest of first steps towards distributive principles. They leave us to take all the rest of the steps in some other way. 3.Rights-based principles We all have a right to minimum material provision---the right to what is necessary to carry out any life plan.20 What that comes down to, however, is a claim on secure use of goods, not a claim that approaches ownership of them. It could be met without property at all---say, by a sugar-daddy who attaches no strings to his gifts. Also, as we have seen,21 the notion of minimum provision suffers from no small indeterminateness itself, and the resources of a theory of human rights on its own cannot remedy it. Still, minimum provision is clearly a claim on the distribution of holdings. Does the right to liberty also generate distributive claims? Say that you and I want to strike a bargain with each other. When the bargain benefits us both, it seems gross interference for others to stop us, unless we are harming them.22 So if we have a claim to secure use of goods, then we have some power over their transfer, and we are on our way to the usual rights of ownership. But whether that is so all turns on the extent of liberty. I argued earlier23 that our human right is not to a 293 DISTRIBUTION general liberty, so long as no harm is done to others, but only to certain specific liberties, namely to the freedoms necessary to live out one's central life plan. In any case, my freedom to act is not cancelled just by my act's causing harm.24 For example, the fact that my deciding not to marry Sadie will distress Sadie, even blight her life, does not threaten my right to marry whom I wish. Sadie's liberty does not include the freedom to violate my equally important liberty. But once the necessary conditions for everyone's living out his life plan are met, we move out of the realm of major harms into the realm of relatively minor harms. Now what I do will not affect that crucial matter of whether or not you can make something valuable of your life; that, we are supposing, is already guaranteed. It can only determine just where in the positively valuable range it will fall. Suppose that you want to buy and build on some land, and that you, the seller, and the builder all happily reach agreement. The building, however, will be an intrusion into an area of great natural beauty. Zoning regulations and planning permission do constitute an interfer- ence in your deal, but not the sort that infringes your liberty. Once in the realm of relatively minor harms, we have to balance gains and losses. The public loss of amenity may well outweigh further enhancement of your personal resources for living a good life. So, rights give us claims on property no stronger than what can be shown to be essential to living out our central life plan. This takes us back to the minimum material provision. But that requirement seems satisfiable without property rights of any strength. If property gets moral backing, it cannot be at as deep a level as human rights. Human rights have to do with the necessary conditions of living a recognizably human existence, and property, what- ever importance it may have, does not have that kind of importance.25 Why not then be content with claims to minimum provision and not look for anything stronger? The Lockean Proviso of as much and as good imports further distributive claims set up by equality. But why not think that, once everyone is up to some minimum acceptable level of well- being, we have satisfied all patterned demands on distribution and where holdings go thereafter may be determined by the 294 MORAL IMPORTANCE unpatterned demands of fair transaction? But then if I were to chance upon some minerals on my vast estates that would greatly improve the quality of life, and in amounts far more than I and my heirs could ever use, and the rest of you are just at the minimum acceptable level, then I would do nothing wrong in hoarding them. And if there were no demands on distribution except minimum provision, then if, by chance, I were the first settler to scuttle off the boat, I could do a few quick calculations and claim vast tracts of land leaving just enough for the rest of you to toil your way up to the minimum level. There must, at least in some settings, be distributive claims arising from equality. 4.Equality-based principles In those rare cases in which social life starts from scratch ---take again a boatload of New World settlers---equal respect requires equal distribution. In start-from-scratch cases, a principle of equal distribution might be subject to qualification, but nonetheless it would dominate the scene. At least, so virtually all philosophers say. However, many of them go on quickly to add some such point as this. Other considerations do indeed soon appear on the scene. The settlers mix their labour with the land, and this creates special claims. Some work harder than others, and this creates personal desert. One freely exchanges with another, and liberty protects the inequal- ities that inevitably emerge. And labour, desert, and liberty all push equal distribution well into the background. No doubt, equality carries over into this more complicated setting in at least some attenuated form (say, in the idea of a minimum provision), but other principles now dominate the scene. However, we have seen that the other principles meant to push equal distribution into the background do not themselves have the muscle to do so. Mixing one's labour yields only the slightest and most indeterminate of claims.26 Desert has no moral weight on its own, and if it figures in moral and political thought it does not do so at any very deep level. And liberty supports a claim to no more holdings than are necessary for carrying out a life plan.27 So equal distribution remains an important principle even in our actual, complex setting. 295 DISTRIBUTION What form does this egalitarianism take? I must now draw together what I have said about equality here and there earlier in the book. There is no one moral or political "principle of equality'. Equality of different things matters at different points in moral and political theory, and we have to keep them straight.28 At the deepest level, there is the requirement of equal respect---to see things morally is to grant everyone some sort of equal standing. But the notion of equal respect is extremely vague and needs to be made more determinate, for instance through the device of the Ideal Observer or the Ideal Contractor. What is common to all conceptions of equal respect is the belief that some sacrifices of one person for another are allowed but only up to a limit. Different determinations of equal respect place the limit at different points. This gives us a new level of moral theory, the level of determinations of the root notion of equal respect---for example, a contractualist's principles of justice or a utilitarian's principle of maximizing. Also on this second level there are likely to be different kinds of principles that can reasonably be claimed to be principles of equality. There is a principle of equal rights: that everyone possesses equal basic protections and entitlements. There is some sort of principle of equal distribution: for example, that everyone should have equal holdings, unless inequalities would have desirable incentive effects. Then, there would be another level, on which principles are adjusted to the needs of decision on a social scale, where knowledge is short and a government's role is limited by the workings of whatever other social institutions there happen to be. For instance, a society would want to define both rights that go beyond strictly human rights and also principles of distribution responsive to the conditions of the particular society. And it is likely that disagreement about principles at lower levels will give way to a greater measure of agreement at the social level.29 In short, equality is a multi-level, and within a level a multi- dimensional, notion. So principles of equality are prone to level-ambiguity. They are also prone to object-ambiguity. We are now interested in principles of equal distribution, but very different views are possible as to what a distribution ought to make equal. 296 MORAL IMPORTANCE Equality of exactly what? It might be equality of holdings, or of the well-being that the holdings produce, or of only a certain part of well-being such as meeting basic needs, or of only (or perhaps especially) a certain part of basic needs such as needs for basic capabilities of limbs and senses that work, and so on. As we have seen,30 we have to get behind talk about basic needs and basic capabilities to the importance they have in our lives; it is prudential values that constitute the ultimate consideration. It is true that needs look as if they connect more securely with obligation than mere desires do; basic needs look like the "bread' of life and mere desires like the "jam'. But, as we have seen,31 this appearance is deceptive. No matter what appears on the list of basic needs, there are likely to be persons who want things off the list more than they want things on it. For instance, a group of scholars might well want books for study more than they want exercise equipment for health, and if they see clearly what they are about, the books will not seem like mere "jam' to them. Once one sees how substantial the values may be that connect with mere desires, it is odd to insist that needs create obligations where desires do not (or create greater ones). That would be to insist that we have an obligation (or a stronger one) to give the scholars what might be of considerably less value to them. If we want to claim that needs have more importance than desires, it is unlikely to be that they have more moral importance. It must be a different claim---for example, that needs have more political importance. A government might decide to provide wheelchairs and ramps for cripples but not allow any individual cripple to substitute what he might care for more---books, say---to the same value. But a father might well accede to his crippled son's request for books in place of ramps, if he were satisfied that, all things considered, that would make his son's life considerably more valuable. Health is always and mobility is usually a necessary condition of living a good life. But what moral weight has a necessary condition to a good life when achieving it might prevent one from having a good life? It is an indefensible position that gives moral weight to trivial effects on what is only a necessary condition, and no, or less, weight to what can be central to the state for which it is necessary.32 Disagreement with this 297 DISTRIBUTION conclusion usually rests on a confusion of levels. The standards of distribution for a parent are not likely to be identical to the standards for a government. Needs generally trump desires; what is at stake with basic needs generally matters much more than what is at stake with mere desires. Governments will therefore naturally concern themselves with basic needs, which is a demanding enough job on its own, and with mere desires less or not at all.33 This assignment to different levels may look a bit too easy. There is a case, of no little strength, for saying that moral judgement---at any level---ought to deal only with the neutral all-purpose means to a good life, with basic needs in that sense. The case for saying so rests, indeed, upon equality. We think that equality of opportunity is not nearly strong enough to satisfy moral demands; in the world as it is, with persons of different genetic and social endowments, it merely lets competi- tion decide who gets what and some persons go to the wall. Equality of opportunity offends our deep sense of equal respect. The fear is that so does equality of well-being.34 It seems to let in too much---for example, the expensive demands of those dedicated to learning and the arts. You with your modest aims in life should not be pushed aside by me with my resource-hungry aims, as worthy as my aims may be. Our deep sense of equal respect seems to require our staying neutral between conceptions of the good and our equalizing all-purpose means rather than well-being. But, in the end, morality cannot stay neutral. No moral theory could get by entirely on a thin, neutral conception of the good.35 Perhaps a government could, and should, often stay neutral. But a father, in distributing resources to his children, ought some- times to take account of the fact that one of them will just squander them. And even a government cannot do its job without taking a view on the relative merits of at least some competing conceptions of the good. It could not decide when its obligations to health care were sufficiently met, unless it made up its mind how valuable other claims on its resources were---for instance, the cultural values of society, its learning and art. This again suggests that we must avoid level-confusion. Some principles fit one level far better than another. We can avoid object-ambiguity partly 298 MORAL IMPORTANCE by avoiding level-ambiguity. Equality of what? It depends on the level. It is no good arguing over principles of equality unless we are clear what level we are talking about, sometimes even what place within the level. Equality is merely a formal notion: sameness in some respect. Sameness is relevant at many different points in moral theory. There is, in a way, no philosophical topic of "equality',36 and the move from egalitar- ianism or anti-egalitarianism in one level or setting to the same in another is always dubious and often illicit. The debate between egalitarians and anti-egalitarians, everyone would agree, has tended to take place too much at the level of asseveration. Egalitarians have the upper hand, because equality, in the form of a deep principle of equal respect, is clearly central to all morality, and it is easy to slip from that relatively uncontroversial basic-level egalitarianism to more contentious forms of egalitarianism at higher levels. In contrast, anti-egalitarians often argue as if all that they need to do is to define their non-egalitarianism, as if egalitarianism wins the day only for lack of alternative. But the most intuitively plausible position on this matter is egalitarianism in some levels and settings, and inegalitarianism elsewhere. But to know where to be egalitarian, and where not, requires a well-worked-out account of equality, starting with the deepest notion of equal respect. What then does equal respect require? As we have seen,37 it does not require an uncrossable line, a line protecting one's unbargainable personhood. It does not require, as contrac- tualists think it does, maximin. It does not even require that, if possible, everyone be brought up to, or not be pushed below, some minimum acceptable level. These restrictions are all too strong. The natural thing to do, in response to their excessive strength, is to look for some weaker restriction which is still an absolute cut-off to maximizing trade-offs---this one, for instance: in distributions between two persons one of whom is worse off than the other at all levels of goods the worse off should get at least not less. Weak as this is, it is still too strong. Surgeons deciding who will get the transplant often rightly ignore it. There is no absolute cut-off on maximizing trade- offs. In view of that it is natural to change tack and to suggest 299 DISTRIBUTION that equality of distribution is one among several same-level principles, a principle of maximizing well-being perhaps being another. Equality then imposes not a cut-off on maximizing trade-offs but a counterweight to them. But this model fails too.38 In small scale interpersonal dealings---say, a surgeon deciding whom to save---the trade-offs that are morally allowable are largely, perhaps entirely, determined by maxi- mum well-being. All of that is just drawing together what I have argued earlier. Our current question is, Are there equality-based principles of distribution? We have to avoid level-ambiguity. We are now interested in distribution on the social scale, in a modern, fairly affluent industrial economy. And that, I think, is the level that most of us have in mind when we talk about "distribution'. The state is now the big force in distribution; the important disputes these days are about what governments should do. If, as nearly everyone agrees, the principle of equal distribution applies in start-from-scratch settings, then it applies in our present-day setting. None of the moral considera- tions meant to supplant it---mixing labour, desert, liberty ---manage to do so. One obvious feature of the large social scale is plurality of agencies of distribution, so in a society with a tradition of family responsibility, churches, and private char- ities, the government will tend to operate on the large scale. Another obvious feature is the government's lack of knowledge of individual utility functions, and for that reason governments will often have to treat persons as if they all had about the same needs and capacities. Yet another feature, if the society is unregimented, is that persons will have very different aims in life. In these necessarily rough deliberations, nearly all of us would be willing to use the same principle: that we should equalize holdings. And it would be a particular sort of holdings, namely all-purpose means, that we should distribute, not out of moral respect for neutrality between differing conceptions of the good life but out of ignorance of individual differences and a belief that there will be plenty of them. Even a utilitarian would accept the principle; if everyone has about the same needs and capacities and it is all-purpose means that we distribute and Diminishing Marginal Utility applies, then 300 MORAL IMPORTANCE to maximize is to equalize. We would start off, then, with a principle of equal distribution of all-purpose means. Next we would start qualifying. There would be an incentive qualification. Every plausible principle of equality grows out of the root thought that everyone matters and matters equally. So to stress only the formal features of distribution is to recall the equally but to forget the matters. It would obviously not do to reduce everyone to the same level of misery. There is a quantitative concern in the deep notion of equal respect, which both tells us how to take "equal distribu- tion' and prompts us to qualify it. If an inequality would work as an incentive to make the worst off better off, it should be allowed. Then there would be a non-envy qualification. If an inequality would make the better off still better off without depriving the worse off it too should be allowed. There would be a perfectionist qualification. Governments should not be entirely neutral between conceptions of the good; some goods constitute the central cultural values of society. Although subsidies to art and learning benefit some persons, often the already better off, more than others, they should be allowed. And finally there would have to be some sort of limited utilitarian qualification. The package of equal holdings is unlikely to give the crippled and ill as good a life as it gives the average citizen. Certain kinds of person should get bonus holdings. And even normal persons could sometimes come under this qualification. In those special circumstances in which a government does know something about individual utility functions, then it would be a waste to deny extra to those who have worthy but expensive aspirations when others can achieve their aspirations with less. No doubt these qualifications would then attract qualifica- tions. All of them would attract what could be called, after a prominent member of a certain set of values, the fraternity qualification. Equality of holdings measures outcomes in terms of material goods and services. But some prudential values are little affected by our holdings---for instance, the quality of our personal relations. If inequalities in wealth are of any size, they will, by dividing society into the powerful and the weak, destroy its sense of fraternity. What one means here by "fraternity' is not easy to pin down. It is sometimes used of the 301 DISTRIBUTION relation of granting another person equal respect. This is the moral relation between persons and is already taken accout of in the moral point of view. But "fraternity' is also used of a kind of free, easy, naturally concerned relation between persons. It is a relation that not only makes moral relations more natural but also makes life better. It is yet another but also an important prudential value, and as such needs to be balanced against those prudential values that are enhanced by holdings.39 What is worrying about a society guided by a principle of equal holdings along with an incentive qualifica- tion is that it seems to regard holdings and the values they promote as all the values there are, or at least all that a state need be concerned with. But they are not. And who better than the state to concern itself with such large scale values as fraternity? These are, then, the equality-based principles of distribu- tion that apply to our present-day setting---the principle of equal holdings of all-purpose means with its nimbus of qualifications. 5.One setting and its principles How should we apply these principles---role-based, rights- based, and equality-based---to a modern, fairly affluent indus- trial society? Role-based principles are particularly elusive. When one tries to pin them down, they slip away into considerations of either rights or equality. What precisely is the obligation-generating role? Is it "mixer of labour' or "productive user' or something not yet described? In a way, the precise description does not matter. When resources are plentiful, we need a way to parcel them out. And when persons are more or less alike, an obvious way is to give control to whoever can productively use the resources, leaving roughly the same for others. Mixing labour with them is generally both a sign of productive use and a strong emotional link which makes their assignment obvious and natural. But this is largely an appeal to equal distribution; mixing labour and productive use are doing little work at any deep level of justification. And the idea of mixing one's labour owes much of its power to our feeling that, through our labour, we 302 MORAL IMPORTANCE become in some sense part of what we make, and that therefore the inviolability of persons gets transferred in some measure to things. It does not matter that I am part of what I make in only a metaphorical, perhaps even quite strained, sense. The inviolability of myself is connected with my right to make something out of my life, and in virtue of that right I should be granted some degree of control over resources. This line of argument gives a basis for control of resources and a limit to it. It explains the need for a Locke-like Proviso that I must leave what the rights of others demand for them. And it explains why some of the things I mix an enormous amount of labour with, say my children, I can never own. I cannot control, in the name of the inviolability of my self, what would violate another's self. But here an appeal to "mixing labour' turns into an appeal to rights. This reduces the list to two: rights-based and equality- based principles. It is not that a philosopher treats this as a list of the material out of which he then constructs distributive rules for his society. That would again be the mistake of seeing the philosopher as the architect of social institutions and the institutions themselves as constructed out of purely philosoph- ical materials.40 The distributive rules of any actual society are made from far more varied materials. One important material is deep human sentiment. We are people of limited altruism; we will work for ourselves and for those close to us but not for just anybody. We are people who naturally enough see ourselves in what we make. Even though the moral significance of mixing one's labour is obscure, its emotional force is clear, and that is enough to make certain possible distributive rules salient. And the sentiment is entirely appro- priate; in some sense, obscure though it may be, I am in what I make, and to ignore that is to ignore me. Mixing one's labour matters more in the genesis of distributive rules than in their justification. Another important kind of material is the social and economic relations that we find ourselves in. We are not a tribe; we are not linked hierarchically by personal ties. In the early industrial period there was no powerful central government and few of the political institutions that would make a welfare state even a possibility. Another material is the solutions that we come up with to Prisoner's Dilemmas and 303 DISTRIBUTION related cases. We desperately need institutions that encourage co-operation. Another material is what we think will prove productive of well-being. And to some extent, though probably relatively small, our moral beliefs no doubt also go to shape our actual distributive rules. It is not that solutions to Prisoner's Dilemmas and other cases get us, just by themselves, to a determinate set of distributive rules. Many different sets would do equally well as solutions. Nor for the same reason, do human sentiments move us much along the way towards determinate rules. What take us to a determinate outcome are probably the influence of economic and social relations, considerations of efficiency, and no doubt also historical contingencies that I have not even listed. The full causal story does not matter to us now. Some such forces have shaped the distributive rules that we in modern Western industrial societies have: our mix of govern- ment economic regulation, taxation, welfare state, and a large measure of private property. Our social institutions have already been built, largely out of non-philosophical material. A philosopher arrives late on the scene. His job is to assess what has arisen perfectly well without him. When we turn to that job, two facts strike us. Economic relations have become so complicated and resources so unplen- tiful that considerations of mixing one's labour scarcely any longer fit. And governments now have the power and the institutions to minister to needs that previously, for lack of alternative, could only be met privately; we could now, if we wished, adopt a principle for social distribution from the family paradigm, perhaps such as "From each according to his ability, to each according to his needs.' What generated many of our present distributive rules can no longer justify them. Many of them are glaringly obsolete. Our rights-principle will do some of the job of assessment. Institutions must ensure the minimum material provision. But that gives us only a moral floor and leaves unassessed everything above it. For that further job of assessment we have the principle of equal holdings with its array of qualifications. However, those principles in effect leave the field largely to considerations of well-being. We should have to decide about the value of incentives and their damage to fraternal relations. 304 MORAL IMPORTANCE We should have to decide what degree of private property most enhanced the quality of life. What sorts of property rights are most economically efficient? Does diffusion of economic power prevent concentration of political power? How plastic is human motivation? Would a change from private property to communal property bring the psychological changes that would make the new institutions in the end more beneficial? I shall not try to answer any of those questions, which are both difficult and familiar. It is their nature that now most concerns us, and it is clear that they raise issues that can be brought within the ambit of well-being. There would also always be a place on a social scale for a principle of equal holdings, no matter how much qualified. And the assessment of distributive rules cannot be carried on independently of their consequences for other rights, such as political liberties. So our assessment should have to balance rights against one another and especially against considerations of general well-being. On the social scale central rights such as minimum provision, auto- nomy, liberty, would have an entrenched, protected place. But on the deepest level of moral assessment we should have to go behind rights to the values that they protect and be prepared to trade off these values against one another and against well-being, in the way that we have already seen in the earlier discussion of rights.41 6.A glance at another setting What principles apply to distributions between nations?42 The people of a neighbouring nation, hit by famine, appeal to us for help. Let us make the story as simple as possible. If we help them, let us suppose, we are not just promoting worse famine later.43 And this is not a nation we have exploited, nor have we profited from another nation's exploitation; its trouble, let us say, was that it was too poor a nation to be worth anyone's exploiting.44 And their famine was not the result of idleness or incompetence on their part; their ancestors simply did not get as much and as good.45 So with the failure of the Lockean Proviso, the justifications of property seem to get no grip at all. The other principles that are prominent in autarkic, exchange, and co-operative economies---fidelity to 305 DISTRIBUTION agreements, fair bargaining, fair participation, and co- operation---also lose relevance. One principle that still applies, of course, is the right to minimum material provision.46 And despite the failure of the Lockean Proviso, property in a way still applies. Someone like Robert Nozick, who makes the Lockean Proviso a necessary condition of property, simply has to stop talking about "property' in the present case.47 But our actual concept of property, which is a legal notion largely fixed by considera- tions of efficiency, is more complex and more flexible than that. We allow some weight to the argument "It's our doing, so it's ours', and also to the protection of successful co-operation. We have the surplus that our neighbours are now eyeing, partly because we are successful co-operators. But you and I co-operate for our own benefit. It is too facile to argue that we have goods that we can give to others, if the distribution destroys the conditions necessary for the production of the goods. And we cannot simply absorb our neighbours into our co-operative community, if co-operation requires that there be a fair measure of values, expectations, and standards of behaviour in common between co-operators.48 Then there is the connected point that co-operation is greatly enhanced if the parties are likely to have repeated encounters with one another. Defection in iterated, few-person Prisoner's Dilemmas is discouraged by a healthy risk of retaliation the next time round. So obligations between persons in different communities have much less backing than obligations in small communities---less backing not only by sentiments of affection but also, perhaps even more important, by self-interest. But though this fact may go a long way towards explaining why we make obligations to those distant from us less stringent than obligations to those close, it does little to justify it. It would justify it only if moral reasons depended for their force on their reducibility to internal reasons or to self-interest. But they do not. This brings us to another important principle: the claim of special relations. A government has obligations to its own citizens that it does not have to its neighbours. As it stands, that point, though undeniable, does not have much strength. A government is an institution consisting of rights and 306 MORAL IMPORTANCE obligations, but it, like any institution, stands in need of moral sanction, and it may receive it only in a modified form. But the claims of special relations can be put without reference to governments. Our government is an agent for us, for co- operators. We have a claim on the benefits of our co-operation that not everyone has, a claim arising from our having been the ones who produced it. But labour supports no claim unless as much and as good is left. Our neighbours can invoke the fatal objection that they never asked us to labour, and that if they had had the opportunity they would have been just as willing to labour as we have been. So the point about special claims is likely to change into a point about the scope of morality. It is likely to come down to the claim that it is no good demanding of the human frame what it cannot deliver. I am motivated to co-operate by the prospects of benefits for myself and those I care about. I am willing to make sacrifices for you, a fellow co-operator, to get you to co-operate. But does our motivation just run out if we are asked to make sacrifices for starving neighbours well outside the co-operative circle? How great a sacrifice can morality ask of us? No one denies that it can ask us to help, where the help is great and the sacrifice small. But then this might be the more testing case where what is needed to save our neighbours' lives is a sacrifice that would blight our own life prospects. Can the human frame deliver so much? Many say that in general it cannot, and that therefore, though such sacrifices would be admirable, they are not morally obligatory. But there is an evasiveness about calling persons who make such large sacrifices "saints' or "heroes'; putting people on pedestals not only honours them but also keeps them at a morally comfortable distance. The only defensible form of the requirement that moral demands fit the human frame is the relatively weak one that it must make sense for us to work at them, to see them as relevant ideals.49 In any case, many moral principles already incorpor- ate concessions to human capacities. Moral theory cannot ignore the importance of extra-moral human institutions. It has to recognize their natural genesis from deep features of human nature. We can use moral grounds to amend and even to reject an institution. But if the institution serves an 307 DISTRIBUTION important purpose, connects with something deep in human nature, and has no easy substitute, then we shall probably be limited in how much we can change it. Property is one such institution. When a person considers producing something, he weighs the cost and benefit. If there is no net benefit why should he sweat to produce it? And he usually measures his benefit, for good moral reasons we shall come to in a moment, in terms of benefit to himself or to certain persons near to him. So the existence of the product depends upon his control of its destination. The natural transition, "It's my doing, so it's mine', is natural in part because it acknowledges the place of control in the existence of the product. Another such institu- tion is parenthood. One of the most important prudential values is deep personal attachment, and one of the most important attachments is a parent to his child. Once one has made such a commitment, it changes the whole direction of one's deliberation. One will no longer reason occasion by occasion and ask, What would maximize well-being? And institutions interlock. One cannot assess each institution in isolation from the rest. Parenthood links with property. A commitment to particular persons alters one's motivation. It gives one a reason to produce things, if one can control their destination. We probably could not successfully change these institutions in any thoroughgoing, revolutionary way. And we probably have no moral reason to do so. 7.How are the principles structured? What are the deepest level principles, the principles underlying distribution in any setting? The principles that apply in a modern industrial state are minimum provision, promotion of well-being, and equality of all-purpose means (with all of its qualifications). So, if the many-sided practical arguments about, for instance, incentives and protections of political rights come out in a certain way, then property rights are likely also to be prominent in this setting. But now take a different setting. The principles that apply between rich nations and poor are somewhat different; minimum provision and promotion of well-being carry over, but property fits less well. With the failure of the Lockean 308 MORAL IMPORTANCE Proviso in this international setting, a lot of the backing for property rights disappears, and control over holdings is relevant only indirectly through various considerations of incentive. What this suggests is that certain principles---for instance, property rights and equality of all-purpose means ---are not ground-floor. Property loses grip in the international setting and never had much in the small-scale family setting. Furthermore, equality of all-purpose means gets its point from the largeness of the social scale. As we have seen, it loses relevance in most small-scale settings. So property and equality of all-purpose means are, I would propose, political principles, principles for the large, social scale; they are not, at any deep level, moral principles. That is no demotion for them; depth is not the same as importance. Take the case of property. What apparently moral rationale there is for property---mixing labour, productive use---allows us to take only the smallest first step towards determinate property rights: they support some control over some holdings in some circumstances. All the major work to make property rights determinate is left to practical considera- tions: the effects of property rights on well-being and on human rights.50 So many criticisms of property rights, Marx's for instance, can be easily accommodated within that practical debate. Do the property rights of capitalism dehumanize both work and enjoyment? Well, appeals to liberty, autonomy, and human dignity can all be accommodated in a good prudential value theory. So there can be no use of property rights as if they were moral rights at some deep level, to block genuinely fundamental moral influences on distribution. This would be to misunderstand what kind of moral work goes into producing determinate property rights. The work is done by the sort of deliberation about human rights (and the values behind them) and well-being that we saw earlier in the chapter on rights.51 Obviously if the work is done by rights and well-being, then considerations of rights and well-being will be able to modify any details of the property rights that owe their existence to them. There are three important principles that put in an appearance in all distributive settings: minimum provision, promotion of well-being, and some form of equality. I have 309 DISTRIBUTION just argued that certain principles on the large social scale do not hold on the small interpersonal scale. But that is not to suggest that small is deep and large superficial. What we are after are the principles that hold on all scales. Of course, that search may just be misconceived. Why think that there are any? Or if there are some, why think that they are particularly deep? Clearly one is entitled to neither presumption. One has to look. One has to work out a substantive theory and then see whether there are indeed ubiquitous principles and whether the ubiquitous support the non-ubiquitous. And this requires having a substantive theory of prudential values as well. In other words, one has to see what structures the values and principles themselves display. Think of determinate property rights once again and ask what supports them. In the last two centuries two traditions have dominated the assessment of property.52 The British utilitarian tradition assesses it in instrumental terms; forms of property are good or bad depending upon their social effects, such as security and incentive. The Continental tradition assesses it largely in terms of the growth and flourishing of the self; Hegel thought that self-realization was enhanced by fairly traditional forms of owning and exchanging, while Marx, on the contrary, thought that capitalist ownership degraded the self, made it thing-like. But these two traditions are really only one. They both approach the assessment of property with the critical apparatus of prudential values. They look like separate traditions because classical utilitarians confined prudential values to happiness without making quite clear what the boundaries of happiness were, and most of us value things besides, even more than, what in any ordinary sense we should call our happiness.53 In particular, we value our humanity, our dignity as self-determining agents; we greatly value our autonomy to choose and liberty to carry out our own life plans. These are all, I have already argued,54 properly regarded as prudential values. So the terms of assessment of both these traditions should be brought within the single framework of well-being. What more should we bring in to assess property? We might appeal to equality of holdings or to the right to minimum provision. But they both can be met by very 310 MORAL IMPORTANCE different sets of property rights, and so give us little to go on in assessing any particular set. We might appeal to certain political rights---for instance, the right to have an equal voice in social decisions. We might find a link between certain fairly strong property rights and certain protections of political voice. If so, then we should have to weigh these political rights against the prudential values that also figure in the assessment of property. That is a sort of weighing that I discussed earlier55 and that I shall come back to in a moment. Think now of equality of all-purpose means, and ask the question that we just asked about property. What supports the principle? The principle is plausible only along with its array of qualifications. The qualifications are motivated by strong quantitative considerations. For instance, one qualification allows deviations from equality if they make everyone better off, or even if they make some better off at no cost to the rest. Another qualification allows deviations if they improve the general welfare, even if, for instance, the gain is in non-vital interests (say, subsidy for the arts) and the loss is in literally vital interests (say, road safety). There is the apparently non-quantitative qualification that restricts deviations from equality in the name of fraternity, but even fraternity, in this context, is best understood as a prudential value and so is to be weighed in a quantitative way against more material prudential values such as wealth---we have, as always, to decide where greater value lies. So there is a strong element of maximizing well-being in all of the qualifications. The import- ant question is, Do considerations of well-being in effect take over? Is the principle of equality of all-purpose means allowed a place on the political level only because it too finds a justification in maximum well-being? Well, the case that we have found for equality of all-purpose means works only on the social scale. However, we cannot confine deliberation, even on the social scale, to all-purpose means. We have to get behind them to a measure of the importance that they, and other things, have in our lives. The ultimately authoritative deliberation goes on in terms of prudential values.56 And we have seen that equality of all-purpose means (or of well-being, for that matter) imposes no cut-off on trade-offs except for this: that persons are to be 311 DISTRIBUTION treated equally except when deviations produce greater gain than loss.57 The most plausible model in which equality is independent of maximum well-being is the model of same- level principles---for example, a principle of equality of well-being and a principle of maximum well-being that have to be weighed against each other. On this model we should expect equality to show up on the political level justified not by maximum well-being but by the deep level principle of equality. But, as we have seen, there is no such deep level principle. The circumstances of the social scale make equality a dominant principle and its qualifications limited in power. But what determines its dominance and the scope of its qualifications is maximum well-being. So on the political level equality is indeed a consideration to be weighed against, and to qualify, the promotion of wealth and other of the more material elements of well-being. The model of same-level principles does fit there. But on the moral level, below not only the political level but also the level of an individual's practical and reflective decision procedures,58 equality is not a consideration that competes with well-being. On this level equality is equal respect, the moral point of view. Equal respect has weight; well-being has weight; but they do not have weight on the same scale. Equal respect is not the competitor of well-being, but its regulator. Equal respect is the principle governing trade-offs between different persons' well-being. It is the principle: treat equally except when gain is greater than loss. Think now of the last of the three principles. What supports the right to minimum material provision? As we have seen, human rights protect a certain kind of value. In the case of this right, what is protected is the value attached to our carrying out an autonomously chosen life-plan.59 The mini- mum provision is defined as the necessary material provision for our being able to do so. But the value that we attach to our doing so is prudential; it gets its value from what it contributes to making a life, seen simply from the point of view of the person who lives it, a good one. This value is to be weighed, in the way we saw earlier,60 against other prudential values attaching to rights, such as autonomy and liberty, and all remaining prudential values. And the ultimately authoritative 312 MORAL IMPORTANCE weighing is the moral one: equality except when gain is greater than loss. Two general points emerge from thinking about distribu- tion. One is the multi-level structure that its principles display. Property rights, for instance, are high storey and cannot be used, as some conservative political theorists use them,61 to restrict what is on the ground floor. If, say, a freak storm wipes out your crop and not mine, then even though you had as much and as good as I, since the point of the Lockean Proviso is to recognize your equal claim to a good life, I must help you. Property rights have to be limited by such considerations, because no rights are so powerful as to override these central values. The only way to miss this point is to have no sense of what a substantive theory of rights is like. The second point is a specific instance of the first. There is no one principle of equality; principles of equality crop up on many levels. What we need in moral and political philosophy more than anything else is to identify them and to plot their relations to one another.