John Leslie Mackie (1917-1981) was a philosopher, originally from Sydney, Australia. From 1967 until his death, he was a fellow of University College, Oxford. He was in  1974 elected a fellow of the British Academy.

IT may be thought that the topics discussed in the earlier chapters in this book have no bearing at all on ethics. Ethics, it may be said, is autonomous; there is no way of arguing validly from 'is' to 'ought', from any truths or plausible hypotheses about the 'natural' facts to any moral prescriptions or judgements of value. However, this would be a mistake. All that Hume himself claimed is that what he calls 'reason' cannot alone, by itself, establish moral distinctions, not that it has no bearing on them at all. Although, he says, the 'final sentence' which pronounces moral judgements depends on some feeling or sentiment, much reasoning has to pave the way for that sentiment and give a proper discernment of its object; distinctions have to be made, conclusions drawn, comparisons formed, relations examined, and general facts ascertained. Even actions and passions can be irrational if they are based on false beliefs, either about simple matters of fact or about causal relations. Or, to approach the issue from a different side, even if some ultimate or basic ethical principles are autonomous, the derivation from them of practical moral precepts must take facts and causal relations into account.

Even someone who held, therefore, that there are objectively valid ethical principles, discoverable by some sort of intuition, some independent moral awareness or understanding, would still have to pay attention to the topics of those earlier chapters in working out the details of his moral theory. But their significance is much greater for someone who holds, as I do, that morality is a human product, that it is a system of thought and evaluation and control of conduct into which human feelings and desires and instincts and social interactions and reciprocal pressures enter, along with knowledge and beliefs of various sorts. Sociobiology and the theories of games and of collective action are then relevant to morality in two distinct ways. On the one hand they may help to explain already-existing moralities, by outlining the contexts in which they have arisen and identifying some of the forces and mechanisms that have produced them. On the other hand they may indicate constraints on any workable moral systems, constraints which must be taken into account in any intelligent advocacy of moral principles or any worthwhile proposals for the reform of existing moral attitudes and ideas.

THE EXPLANATION OF EXISTING MORALITIES
Sociobiology is the offspring of a marriage between game theory and the Neo-Darwinian theory of evolution. Genetically determined behavioural features, including elements of social behaviour, will be favoured by evolutionary selection in so far as they help the individuals that display them to propagate the genes which determine those features, typically in situations in which those individuals are in relations of mixed competition and co-operation with other members of the same species. It is important that the key process is gene selection, not group selection: there is no general principle that features which promote the well-being or flourishing of a group will be evolutionarily favoured. Tendencies which, if they became widespread, would thus help a group may well be defeated, in competition within the group, by tendencies that are relatively harmful to the group as a whole but are more likely to enable the individuals that display them to propagate their own genes. Thus altruism directed indiscriminately towards all group members will lose out in competition with more selfish behaviour.

Yet it would be almost equally wrong to speak of individual [self-preservation as the motive force in selection] {a line has dropped out of the text here, what is in [] is my attempt to restore it}, since the characteristics that are favoured are not necessarily ones which help the individuals that have them to flourish: an individual's genes may be propagated best by behaviour in which that individual sacrifices itself in helping its offspring or other near relatives to survive and reproduce. In consequence it is not purely egoistic behaviour that we can expect to result from evolutionary selection, but a mixture of egoism with self-referential altruism, that is, altruism directed towards individuals who are somehow related to the agent. But such altruism need not be directed only towards blood-relations: reciprocal altruism can also be favoured by evolutionary selection. If, conditionally upon A's helping B in some way, B will do something that helps A to flourish and have offspring, then there will be an evolutionary pressure in favour of A's helping B and B need not even be a member of the same species as A.

Of particular interest in relation to morality is the way in which retributive tendencies can be selectively favoured. This is most obvious for hostile retribution. Suppose that an animal is injured by another, either of the same or of a different species, where the first is able to do some harm to the second which the second can associate with its own initial aggression. The aggressor will then be discouraged from repeating the attack, so the retaliation which discourages it will tend to benefit the retaliator. More generally, where such situations constantly recur there will be two related selective pressures: among the potential victims of aggression, in favour of retaliation, and among the potential aggressors, against aggression towards retaliators.

Of course there need not be any calculation or deliberate choice by either party; it is rather that the mechanism of natural selection mimics purposiveness, producing instinctive behaviours which resemble those that might well result from intelligent calculation. Thus we can understand how there could come to be instinctive spontaneous retaliatory tendencies, in much the same way as we can understand the development of reciprocal altruism and instinctive spontaneous mutual assistance. Kindly retributive tendencies—a disposition to display 'gratitude' for benefits, which is not quite the same thing as systematic mutual assistance-could in principle be developed in a corresponding way to hostile ones, but it would not be surprising if hostile retribution were in general stronger and more widespread than kindly retribution, simply because occasions in which it is likely to be beneficial will tend to occur more often. This is particularly significant in so far as morality itself can be seen as an outgrowth from retributive (especially hostile retributive) tendencies, as will be shown below. It is hostile retributive behaviour that is initially explained in this way; but in creatures that have a high level of consciousness there will naturally also be retributive sentiments, such as anger and resentment, accompanying and sustaining the behaviour.

A central sociobiological concept is that of an evolutionarily stable strategy. A 'strategy'—that is, a genetically determined behavioural tendency—is evolutionarily stable, relative to a certain context and to certain alternative strategies, if in that context it, rather than any of the alternatives, will be favoured by natural selection in the competition between these rival tendencies within a group of members of the same species. As was suggested above, indiscriminate altruism will not, in most contexts, be evolutionarily stable in relation to a more 'selfish' strategy. But what is thus evolutionarily stable may be a mixed strategy or a mixture of strategies rather than a single uniform strategy. Consider a situation where members of the same species repeatedly come into conflict, perhaps over food or mates or territory, and consider just the two alternative strategies of aggression and conciliation.

When two aggressors meet, they fight and in general both suffer a certain amount of injury, while only one of them eventually gets the item in dispute. When an aggressor meets a conciliator, the conciliator gives way without fighting and without being hurt, while the aggressor gets the goods. When two conciliators meet, they peaceably share the wanted item. Now while aggressors clearly do better against conciliators than other conciliators do, it may well be that aggressors will do less well against other aggressors than conciliators. This will be so if the damage sustained in fighting is considerable in proportion to the value of the items in dispute-both reckoned in terms of the tendency to decrease or increase the propagation of one's genes. If the average value of a fifty per cent chance of getting the item in question is less than the average disvalue of the injuries, then the average payoff to an aggressor in dispute with another aggressor will be negative, whereas the payoff to a conciliator in dispute with an aggressor is merely zero. Now if this is so, if aggressors do better against conciliators than conciliators do, while conciliators do better against aggressors than aggressors do, then neither of these two strategies on its own will be evolutionarily stable.

A population consisting wholly of conciliators could be successfully invaded by aggressor genes, but equally one consisting wholly of aggressors could be successfully invaded by conciliator genes. The evolutionarily stable strategy will be some mixture of the two. The precise ratio will depend on the average values and disvalues of the various possible results: for example, if the average disvalue of the injuries sustained in fighting is equal to the average value of getting the item in dispute, the stable ratio will be equality between aggression and conciliation. But this ratio may be realized in either a mixture of strategies or a mixed strategy. That is, it may be sustained either by there being just as many pure aggressors as pure conciliators, or by each individual being disposed to be aggressive half the time and conciliatory half the time, or, indeed, by some combination of these. Of course, this is only a simple illustration; what is evolutionarily stable may well be a more complicated mixture of strategies that are themselves more complicated (for example, by including various conditionalities) than simple conciliation or aggression.

So far we have been thinking of genetically determined behaviour-instinctive, spontaneous, uncalculating. But the formal character of these developments depends solely on the fact that the genes involved are self-replicating items which cause behaviour which in turn reacts favourably or unfavourably on the frequency of their replication, but whose replication is not always perfect: some mistakes in copying—mutations—occur, and these mistaken copies can in turn produce copies of themselves. Anything with these formal features is a possible subject of evolution by natural selection. Now in human (and in some non-human) populations there can arise self-replicators of another sort, cultural traits or, as Dawkins calls them, 'memes', which include tunes, ideas, fashions, and techniques., Provided that such items can reproduce themselves (fairly faithfully, but with occasional variations) by memory and imitation, and can cause behaviour which then reacts on their own propagation, they too and the behaviour they produce will be subject to evolutionary pressures which work in the same way as those that result in gene selection. Memes, no less than genes, will be selected on account of their tendency to produce behaviour that favours their own propagation—which is not necessarily behaviour beneficial either to the individuals or to the societies in which these memes take root. That some idea or belief or practice has a social function is neither necessary nor sufficient in itself to explain its persistence or its spreading. Its performance of a social function may be part of such an explanation, if it can be shown that its having this function helps to propagate it; but a trait which is socially neutral or harmful may also have the power to propagate itself. Just as infectious diseases can spread among the bodies of a population, harmful ideas can flourish and spread among their minds.

One sort of cultural item which both performs a social function and in so doing contributes to its own reproduction is what we can call a convention. Conventions can arise to solve problems of two sorts, of co-ordination and of partial conflict.

A problem of co-ordination arises where the interests of the parties do not diverge, but where their independent choices of action may or may not maximize the fulfilment of their interests. A typical problem of this sort occurs when each of two or more people wants to meet the other(s), but each has to decide independently where and when to go to do so. For example, in a primitive setting, people may want to meet to barter their respective products. A suitable convention might be that they should meet at the foot of a certain tall tree on the morning after each full moon. It is easy to see how such a convention establishing a regular market place and market day could grow up even without any explicit agreement. If by chance two or three people met, with significant advantage to each, at some striking and therefore memorable time and place, they would be likely to associate that success with that time and place, and so tend to return to the same place at a later corresponding time—say, after the next full moon. Once two or three started meeting regularly, others who happened to find out about it would tend to join in: the advantage of the incipient convention to each joiner would provide a force sufficient to spread and maintain it.

Less obviously, a convention can arise to solve a problem of partial conflict—say, something of the form of a prisoners' dilemma. Suppose that there are two families, of which one catches fish and the other collects edible fruit and roots. They are too much afraid of each other to meet and barter, although each would do better if they exchanged some of their products than if each consumed only its own products, and both families know this. But if each family has to decide independently what to do, each will do better by consuming all of its own products, whether or not the other family gives the first some of what it produces. Since neither family can trust the other, it looks as if no exchange can take place. But suppose that some time when the fishermen have had a good catch they say to one another: 'We might just try leaving two or three fish well in view on that rock that the fruit gatherers often pass, and see what happens.' Of course, it may not work; the fruit gatherers may just take the fish and leave nothing, and then it is likely to be some time before the experiment is repeated. On the other hand, the fruit gatherers may take the fish and leave some fruit in return, and then come back to see what the fishermen do the next day. In favourable circumstances a virtuous spiral could be set up, with each family encouraging the other, rewarding it when it repeats or increases its gifts, but reducing its own offering if the other family reduces its. With such reciprocal sanctioning, a convention of regular exchanges of produce could grow up and be maintained without any explicit bargaining or agreement, and without any prior principles of agreement-keeping or mutual trust between these parties. In fact this sort of convention maintained by reciprocal sanctioning is a more basic social relation than agreement or contract: indeed, even in a society where there is full communication the making and keeping of agreements can itself be seen as a particular example of this sort of convention. For the main motive that each party has for keeping an agreement lies in the fact that if it fails to do so it is less likely to be trusted another time. Nor need this be a consciously calculated motive: rather, the tendency to keep rather than break agreements can be automatically 'reinforced' by the rewards of honesty and the penalties for dishonesty, so that each agent develops a spontaneous inclination to keep agreements.

Situations of partial conflict, where the prior interests of the parties diverge to some extent, and yet where each will do better in terms of those prior interests if they co-operate, each making some concessions, than if each pursues his own interests directly, need not be symmetrical: one party may have an initial advantage over another. Such a situation of unequal partial conflict can still be resolved by the growth of a convention in much the same way as one of equal partial conflict, but the convention that emerges is likely to be differentially advantageous to the party that starts in the stronger bargaining position. Conventions that arise in this way and are sustained by reciprocal sanctioning will thus include 'norms of partiality' as well as 'prisoners' dilemma norms'.

In speaking of 'norms' as well as of 'conventions', we are recognizing that what arise in this way are not only patterns of behaviour but also rules or principles of action which are 'internalized' by the participants. The association of moral sentiments with the practices, in particular disapproval of violations, the feeling that they are wrong or not to be done, and a sense of guilt about one's own transgressions, is a major part of such internalization. Only when this stage is reached can we speak properly of a morality, though as we have seen there can be pre-moral tendencies to behave in ways that coincide with or come close to those that are characteristically supported by moral thinking. But where does the notion of wrongness come from? We have already sketched a possible genetic explanation of hostile retribution and resentment of injuries. We have also seen how co-operation can arise and be maintained either by genetic selection or by the corresponding social evolution of a certain kind of meme, namely conventions. Putting these two together, we can understand how there can come to be co-operation in resentment, where all or most members of a group jointly react against injuries to any of them. Among human beings in particular, certain kinds of behaviour can be recognized and can become the objects of co-operative resentment. This may be the source of the characteristic 'disinterestedness' and 'apparent impartiality' of the moral sentiments—only apparent impartiality, since the conventions which are 'impartially' observed and enforced may themselves be norms of partiality, differentially more advantageous to some kinds of person than to others. The notion of moral wrongness includes three main elements: what is thought wrong is seen as being harmful generally (not just to this or that particular person), as being intrinsically forbidden
(not merely forbidden by this or that authority, but in itself simply not to be done), and as calling for a hostile response (again not merely from a particular person: it is rather that a hostile response from somewhere is needed). All three of these elements can be understood in the light of the above-outlined genetic and social mechanisms, as projections of the sentiments associated with co-operative resentment and hostile retribution. And this notion of wrongness is the central moral notion: other moral ideas fall easily into place around it. A sense of duty, for example, is the feeling that failure to act in such-and-such a way would be wrong in this sense.

Existing moralities vary between societies and even within one society. And even one morality will commonly recognize quite a collection of diverse requirements—for example, such virtues as courage, temperance, perseverance, honesty with regard to property, veracity, promise-keeping, loyalty to friends, patriotism, marital fidelity, parental devotion, chastity, modesty (in two different senses), piety, cheerfulness, compassion, tolerance, fairness, sportsmanship, and so on. And what has to be explained is not only the recognition of these as virtues but also the fact that people maintain a partial, but only partial, conformity to them. The content of moral thought and moral conduct is complex, and an adequate explanation of morality would have to cover much more than a bit of general benevolence and some rules about sexual behaviour. But what I have outlined here is the beginning of such an explanation: we can see how this sort of approach could account for the complex reality that we actually find. We have referred both to genetic (sociobiological) mechanisms and to sociological ones. We have found possible sources of self-referential altruism and reciprocal altruism, and of co-operation within groups guided and sustained by various conventional norms, including co-ordination norms, prisoners' dilemma norms, and norms of partiality. We have traced a possible development of retributive behaviour and retributive sentiments growing into co-operation in resentment and thence into the central moral concept of wrongness.

Is there some competition between the genetic and the cultural accounts? If so, it can be fairly easily resolved. It will be reasonable to ascribe to biological evolution those pre-moral tendencies to care for children and close relatives, to enjoy the company of fellow members of a small group, to display reciprocal altruism and both kindly and hostile retribution, which we share with many non-human animals, but to ascribe to cultural evolution the more specifically moral virtues which presuppose language and other characteristically human capacities and relations, such as honesty, veracity, promise-keeping, fairness, modesty as opposed to arrogance, and so on, as well as those detailed moral principles which vary from one society to another.

I want to guard especially against two possible misunderstandings, both of which would make my account of morality seem more egoistic than it is.

First, I would stress that none of the above-mentioned mechanisms requires calculation, let alone calculation in terms of self-interest, on the part of the agents concerned. With the 'lower' animals there is no question of motives, but simply of genetically determined behaviour, self-referentially altruistic, retributive, and so on. Even among animals, including humans, to whom we can ascribe motives, altruism, resentment, gratitude, norm-following, moral disapproval, a sense of duty, and the like have been explained as direct motives. From the agent's point of view the corresponding actions are not adopted as means to anything else. It is only that both the genetic and the convention-forming mechanisms can mimic purposiveness, can give the agents direct, spontaneous, motives which lead them to act in ways which in fact promote what we can regard as their pre-existing interests.

Secondly, even those pre-existing interests are not to be taken as being purely egoistic. At the genetic level, what we can describe, speaking behaviourally, as the selfishness of genes is not equivalent, even behaviourally, to the selfishness of individual organisms. And at the social level the pre-existing interests that create the problems of co-ordination or partial conflict, prisoners' dilemmas and the like, from which conventions may emerge, are not necessarily, and in fact will not be, exclusively egoistic. The human individuals enter into these situations already with their various direct motives of self-referential altruism (concern for children, relatives, friends, and so on as well as for themselves) and the joint purposes of already co-operating groups, as well as motives of pride, self-respect, honour, and the like. Prisoners' dilemmas, it must be emphasized, can arise whenever the parties have partly divergent purposes of whatever sort—even if, for example, both parties are aiming at the general happiness, but have different views about how it can best be promoted—and are not produced only by a clash of selfish aims.

A wide variety of moral rules and principles could be understood as having developed by the operation of mechanisms of these sorts in different concrete circumstances, and the fact that what is evolutionarily stable—either genetically or, by analogy, socially—may be a mixture of strategies could explain the coexistence within a single social group, even in a stable environment, of different norms with regard to the same sort of choice, and of different degrees of conformity to a norm. Yet despite such diversity and indeterminacy there are some significant common features of the behavioural tendencies and moral principles that might emerge in these ways.

Except where individuals sacrifice themselves for the sake of their own children or close relatives, we should not expect there to be principles of complete sacrifice of the agent; but there could be principles of joining, when others do so too, in enterprises which carry some risk of the agent's being killed or suffering some other serious harm. We should not expect there to be principles of pure altruism or pure benevolence directed unconditionally towards the well-being of a whole community, of loving all your neighbours literally as yourself, the social insects are a striking exception, but their behaviour is covered by the above-mentioned proviso, since with them all members of the hive or nest are indeed close relatives. Equally, we should not expect to find people guided by the pure abstract principle of doing as you would be done by; we can expect them to follow rather the Hobbesian variant of this, to be willing to co-operate and make concessions when others are so too. One is not to do to or for others whatever one would like them to do to or for oneself absolutely or unconditionally, but only in so far as there is or can be established such a practice in which others as well as oneself will join. That is, what we should expect to develop and flourish are norms of co-operation and reciprocation, including asymmetrical ones that reflect the unequal bargaining strengths of different kinds of participants, not norms that enjoin that each agent should do whatever is most likely to promote the general happiness. And in fact we find that the norms that really do the work of controlling people's behaviour are of the sorts that this approach would lead us to expect. People generally feel that it is wrong to kill or assault others, and usually refrain from doing so; provided that those others are not a threat to them; they respect private property and keep mutually beneficial agreements; but they do not feel obliged to decide whether each such act or forbearance is likely to do more than its opposite to promote the general happiness. In an election, though one's individual vote is almost always causally inefficacious, one takes credit for having voted for the right party if it wins the election, and one puts a 'Don't blame me ...' sticker on one's car window if it loses; that is, we feel that all those who participate in an activity deserve a share of the praise or blame for its results, even if each person's action on its own was causally irrelevant.

Or consider people's moral feelings with regard to leaving litter at beauty spots. If the place is free from litter when you arrive, there is a fairly strong moral pressure against leaving the first beer can. One does not think 'Just one can won't make a significant difference', but rather 'It wouldn't be fair for me to make a mess when other visitors have taken the trouble to leave it all so clean'. Even if there is rubbish everywhere already, one may well not say 'One more can will make no difference', but rather 'If I throw my can away, I shall share the responsibility for all this filth'.

Thus the norms that do the work include such ones as these:

Join in enterprises that promote public goods in which you will share, and play your part in them.

Join others in refraining from activities that would produce public harms from which you, along with others, would suffer.

Take a share of the credit for any good results of enterprises in which you have joined, and a share of the blame for any bad results of activities in which you have taken part, without calculating the differential effects of your participation; give credit and blame similarly to others.

Such norms as these are different from the utilitarian one:

Act so as to maximize the expected production of general good.

They are also different from the Kantian one:

Act only on a maxim which you can will to be a universal law of nature.

In short, the moral principles which we find flourishing and effective in controlling people's actions are of the sorts which we should expect to find if we assumed that morality has developed through the biological and social mechanisms outlined above; and they are significantly different from the ones recommended by some influential schools of philosophical moralists. The vital differences are that the operative principles involve concrete relations of reciprocity, co-operation, and joint practices whereas the philosophically advocated ones are more abstract or appeal to merely hypothetical considerations, and therefore build less on each agent's pre-moral motives and offer him less chance of a reward.

CONSTRAINTS ON MORAL DEVELOPMENT AND REFORM If it is agreed that already-existing moral systems can be accounted for in these ways, we may go on to the question whether, and, if so, to what extent, these genetic and social mechanisms generate constraints on any workable and therefore rationally recommendable moral systems.

It is sometimes asked whether we can escape from our own biology. Indeed Dawkins himself, in the final chapter of The Selfish Gene, suggests that 'We are built as gene machines and cultured as meme machines, but we have the power to turn against our creators. We, alone on earth, can rebel against the tyranny of the selfish replicators'. But this is a misleading way of putting it. We are, and will remain for the foreseeable future, what biological evolution over millions of years has made us. But on that practically unalterable basis, cultural evolution has already erected a far more complicated set of superstructures: we obviously do not behave just in ways that are biologically determined. The interesting question is how much freedom of movement is left for the development of still further superstructures, which will start from and yet may modify the existing culturally-determined ways of behaving and thinking, much as cultural evolution has started from and yet modified the biologically-determined patterns.

For example, we may sensibly enquire whether we could develop a morality of pure altruism or pure rational benevolence, which might replace the existing norms which prescribe more specific kinds of conduct and which are characterized, as we have seen, by reciprocity and (sometimes asymmetrical) co-operation. Such a morality would, of course, require each individual to be ready to sacrifice not only himself but also those close to him for any greater advantage to others, however remote. Such a replacement is most unlikely, for three reasons. First, this morality would frequently have to oppose strong genetically ingrained tendencies of egoism, self-referential altruism, reciprocal altruism, and retribution as well as strong culturally developed traits of similar sorts. Secondly, if this kind of morality did begin to flourish among some considerable number or people, it would lay them open to exploitation, to being used by those who diverge not only from this morality but also from the traditional one of reciprocity and conditional co-operation in the direction of pure selfishness. In Dawkins's terms, the more 'suckers' there are, the more 'cheats' will flourish. Or, as Hutcheson said long ago, to do away with reciprocation and a right to the fruits of one's labour 'exposes the industrious as a constant prey to the slothful, and sets self-love against industry'. Thirdly, this sort of morality suffers from a radical indeterminacy in its object. There simply are no natural and obvious ways of measuring the 'general welfare', or of eliciting a 'collective choice' or 'group preference' from the set of divergent preferences of the members of a group. Not even something as apparently neutral and uncontroversial as the Pareto principle is really acceptable in all circumstances. And of course this indeterminacy provides further opportunities for exploitation, for people to represent what are really their own selfish purposes as the promotion of the elusive general welfare.

It is true that we may distinguish between a practical, working, morality which fairly directly guides or controls conduct, and a higher level, critical, moral theory in terms of which philosophers may evaluate and seek to modify the practical, working, morality. All three of our objections tell against the possibility of developing a working morality of pure altruism or general benevolence, but the first two would not tell against the philosophical use of a moral theory based on general benevolence: the points that they make are ones that our critical philosophers would properly take into account in deriving a working morality from their higher level theory. But the third objection tells even against this philosophical use of general benevolence: if there is no satisfactory way of adding up interests or eliciting a collective choice or measuring general welfare then there will be something spurious about any purported derivation of working moral principles from these concepts.

There may, then, be constraints which exclude moralities of the utilitarian type. If there are, I see no reason to regret the fact. As I have argued elsewhere, a morality based on the assigning of rights to each individual would in itself be more attractive, even apart from the fact that it would fit in better with the suggested constraints of reciprocity and conditional co-operation.

Another question is this: do these constraints imply that a workable morality can take account of the interests only of those who are able to compete and to co-operate, and can show no respect for non-human animals, for example, or for seriously and permanently handicapped human beings?

It is true that the mechanisms sketched so far would not in themselves generate moral sentiments in favour of such non-competitors. But they do not positively require that there should not be such sentiments, in the way that they do tell against a morality of pure altruism or benevolence. Hume, in referring to this problem, suggested that considerations of 'justice' would not arise between us and creatures, human or non-human, that were permanently unable to compete with us, but that 'we should be bound, by the laws of humanity, to give gentle usage to them'. Hume's use of the word 'justice' to refer only to moral principles of a quasi-contractual sort may be a bit misleading; but we can understand what he meant. But how are we to explain the sentiments of 'humanity'? Surely as Hume himself did, as a product of 'sympathy' and 'imagination'. Once we have developed, in the ways already outlined, a morality that assigns rights to those other human beings who are participators in the mixture of competition and co-operation that is the ordinary and inevitable human condition, and have acquired dispositions to show some respect for their interests and to feel compassion for their sufferings, it is not difficult for us to extend these attitudes to other creatures, human or non-human, in so far as we see them as being like ourselves and like those towards whom we have already developed respect and compassion. It is true that there is no necessity about this. It is perfectly possible for people to combine the finest moral sensitivity in relation to their fellows with extreme inhumanity towards 'brute beasts' and defective human beings, or indeed to non-defective human beings whom they see as in some way alien to themselves and their associates. All that I am saying is that the contrary is also possible, and is the more likely to come about the more people are imaginatively aware of the similarities between themselves and the non-participators. The constraints we have been considering do not rule out a morality that includes an element of humanity in this sense. On the other hand, they do set some limits to its influence. The kind of humanity which we can expect to be effective will fall far short of the equal concern for all sentient beings, proportionate only to their capacity for feeling pain and pleasure, which seems to be a consequence of utilitarian principles. We cannot expect such equal concern for the interests even of other fully active human beings, even those with whom the agent is acquainted. Putting it bluntly, we can expect people in general to display humanity only when they can do so at not too great a cost to themselves and to those close to them, though as I have said, they may be willing to risk the final sacrifice.

We have seen that where, in the mixture of competition and cooperation, there are parties with unequal bargaining power, the conventions which resolve such problems are likely to include norms of partiality, rules or principles which are differentially advantageous to what was already the stronger party. Any explanation of existing moralities would have to include such a mechanism, since nearly all actual moral systems have been of this biased sort. Duties and rights have been assigned unequally between peasants and noblemen, slaves and freemen, blacks and whites, brahmins and untouchables, citizens and non-citizens, owners and workers, party-members and non-party-members, men and women. But is this widespread and explainable feature an unavoidable constraint on any workable morality? Must any morality that actually controls human conduct merely reflect and perpetuate differences of strength that arise from other forces and relations in society? Of course there have been and are egalitarian moral theories; but the question is whether these can actually operate in practice.

As with our previous problem, we can see how sympathy and imagination can generate egalitarian moral sentiments; but egalitarian, like humanitarian, sentiments are likely to control people's conduct only when the cost to the agents themselves and those close to them is not too great. And, in general, the giving of more equal rights to important classes of those who are fellow participators in the mixture of competition and co-operation would cost more than the displaying of a moderate degree of humanity to non-participators. On the whole, then, we can expect norms of partiality to be replaced by norms of impartiality only in so far as the previously disadvantaged groups find non-moral sources of strength that improve their bargaining position. Otherwise egalitarian moral sentiments will either not arise at all or, if they are generated by sympathy and imagination, will remain largely at the level of theory, and not develop into a genuine working morality. This may seem to be a pessimistic conclusion; but something like it has, I believe, been reached through trial and error by many of those who have been actually engaged in campaigns for racial (etc.) equality. It is also only a generalization of the traditional Marxist thesis that capitalism cannot be reformed by the moral conversion of individual capitalists: as long as the 'relations of production' are unchanged an individual capitalist who attempts to give up exploitation is merely replaced by someone else who plays his social role more efficiently.

In raising this last problem, we have already touched upon the area where there is not only the greatest need for further developments of morality but also the greatest difficulty in devising them, that of the political applications and extensions of morality. Devices for compromise and adjustment of conflicts between individuals have grown up, largely automatically but with some help from deliberate invention, over many thousands of years, and have been widely accepted into moral thinking and into various legal systems. But corresponding devices for compromise and adjustment of conflicts between politically organized groups are relatively rudimentary, and though some principles of international justice are vaguely recognized, they form as yet only a weak system of international morality. Here, I am afraid, I have no specific suggestions to offer, but only a plea that the problems should be considered, and that they should be considered in the light of the understanding of morality in general, and the constraints upon it, that we have been developing. Here, as in all human affairs, the concrete situation is inevitably one of partial conflict, of a mixture of competition and co-operation, of partly coincident and partly divergent interests, but also, and importantly, of unequal and unstable concentrations of power. We cannot reasonably hope for effective principles of political and international morality based merely on abstract ideals. We need principles of adjustment which both allow for and are themselves sustained by the very forces which they try to control.

Hume quotes from Cicero a passage which, as Hume rightly says, anticipates Hobbes's picture of the state of nature, the condition of men in the absence of laws and governments, as a state of war. It concludes:

The difference between this civilized and humane life, and that monstrous one, depends upon nothing so much as upon the difference between law (or justice, ius) and force
(or violence, vis). If we are unwilling to use one of these, then we must use the other. Do we want to eliminate force? Then it is necessary that law should flourish—that is, legal proceedings (iudicia) in which all law is contained. Do we not want legal proceedings, or not have them? Then it is necessary that force should rule. Everyone can see this.

This is, however, only part of the truth, We need not only ius but also mores, morality as well as law, and they must be in reasonable harmony with one another. Besides, we need the right sort of ius and the right sort of mores to contain the forces that would otherwise break out as vis. The problem is to find out what these are. The first step towards solving the problem is to see that this is the problem. Regrettably, not everyone has seen this, and many thinkers divert attention from it by posing questions about moral philosophy in less illuminating ways.



From Persons and Values, Oxford: Clarendon Press, 1985; originally published in Cooperation in Humans and Animals, ed. A. M. Colman (van Nostrand Reinhold, 1982). Footnotes have been omitted.