J. L. MACKIE

* From "Retributivism: A Test Case for Ethical Objectiv­ity" by J. L. Mackie. Published for the first time by permis­sion of Mrs. Joan Mackie. All rights to further publication reserved.

I

In my Ethics: Inventing Right and Wrong I have argued that there is a real issue about the objectivity or subjectivity of moral principles and values, and I have tried to clarify this issue, which is often treated in a confused way, and to say what I take the central problem to be. I have also argued that neither objectivism nor subjec­tivism can be dismissed as incoherent, and nei­ther view can be established merely by concep­tual or linguistic analysis, or indeed by any other sort of a priori argument. To decide whether moral principles and values are objec­tive or subjective in the important sense, we have to consider which view can be developed so as to provide the best overall explanation of all the relevant phenomena or data, all the moral (or related) "appearances" with which we are confronted. I would relate this question to what Hume, for example, says in Book III of the Treatise. I do not think that he has shown in any conclusive way by the end of Book III Part i that moral distinctions are not derived from "reason" and that they are derived from a "moral sense" or from "sentiment." Hume's case for these theses is still incomplete at this point. It is completed only when he has shown, by the end of Book III Part iii, how on a "senti­mentalist" foundation he can develop a socio­logical and psychological explanation of both the "artificial" and the "natural" virtues, that is, can explain why these various dispositions oc­cur and why they are cultivated and approved of in human society, fairly uniformly though with some variations.

An examination of retributivism will, I sug­gest, contribute usefully to this discussion. Many people seem to have moral intuitions of a retributivist sort. But are these genuine intu­itions which reveal objectively valid moral prin­ciples or requirements, or are they rather ex­pressions of sentiments for which we should seek, and can find, psychological or sociological or even biological explanations? I shall argue that purportedly objective retributivist princi­ples cannot be defended, but that retributivist sentiments-deeply ingrained, highly developed and organized-can be readily understood.

It is becoming more common for philos­ophers to declare their adherence to the retribu­tive theory of punishment. This may be under­stood as an offshoot of the widespread tendency to reject utilitarianism in favour of some view which takes as central such concepts as those of justice and rights. However, this adherence may go along with a very broad or loose account of what the retributive theory is. Thus Jeffrie G. Murphy says that "A retributive theory of pun­ishment is one which characterizes punishment primarily in terms of the concepts of justice, rights, and desert-i.e. is concerned with the just punishment, the punishment the criminal deserves, the punishment society has the right to inflict (and the criminal has the right to ex­pect)."' J. G. Cottingham, in an admirable re­cent article, finds among views which have been called retributivist at least nine distinct ap­proaches, and almost as many distinct strategies for justifying punishment, not all of which would fall under Murphy's description, broad though it is.2

Cottingham's headings are as follows: Repay­ment Theory (that by being punished the crimi­nal repays a debt to society); Desert Theory (which seems to reduce to the bald assertion that crime deserves punishment, that it simply is just that the offender should be punished); Pen­alty Theory and Minimalism (I shall say more about these two later); Satisfaction Theory (that punishment is justified by the satisfaction that it gives to the victim of the crime or to others who have been injured or outraged by it); Fair Play Theory (that failure to punish would be unfair to those who, unlike the criminal, have obeyed the rules, and so have forgone satisfactions or advantages that the criminal has gained by breaking them); Placation Theory (that a god who has been angered by, say, a murder must be placated by the execution of the murderer, or else he will go on being angry with the whole nation); Annulment Theory (that the punish­ment somehow cancels or annuls the crime); and Denunciation Theory (that punishment is justified as an emphatic denunciation of a crime).

These are certainly a mixed bag. Cottingham takes the notion of repayment of a debt as what is central in retributivism, and rightly complains that several of the approaches he has listed have little to do with this. But let me suggest a way of bringing some sort of order into this mass of ideas.

It is true that retribuo in Latin means "I pay back." But surely the central notion is not that the criminal repays a debt, pays something back to society, but that someone else pays the crimi­nal back for what he has done. Punitive retribu­tion is the repaying of harm with harm, as re­ward is the repaying of benefit with benefit. We can class as an essentially retributivist approach any that sees at least some prima facie rightness in the repaying of evil with evil, especially a proportionate evil, but returning no evil where no evil has been done. This entails that justifica­tions of punishment are retributive in so far as they are retrospective. On the other hand, sug­gestions (which Cottingham mentions under Penalty Theory and Minimalism) that a previ­ous wrong act is a merely logical or definitional requirement for something to count as punish­ment, not in any way a moral justification, would be better not counted as retributivist.

Approaches that count as retributivist in this sense of involving retrospective justification must, however, be further subdivided. One im­portant division is between what I would call positive and negative retributivism. Positive re­tributivism sees the previous wrong act as in it­self a reason for inflicting a penalty. The crime at least tends positively to justify a penalty; but it may or may not be held that the wrong act is in itself morally sufficient for the penalty, or that it absolutely requires the penalty, irrespective of all other considerations. Negative retributivism (for which "minimalism" may be another name) holds that those who are not guilty must not (that is, morally must not, not logically cannot) be punished, that the absence of a crime mor­ally requires the non-infliction of a penalty. There is also a quantitative variant of negative retributivism, that even if someone is guilty of a crime it is wrong to punish him more severely than is proportional to the crime. And of course there is an analogous quantitative variant of positive retributivism, that a crime of a certain degree of wrongness positively calls for a pro­portionate penalty.

This distinction is important for at least two reasons. One is that once they are distinguished negative retributivism will be found to have a greater measure of general acceptance than positive retributivism. Many thinking people who hold firmly that it is morally wrong in itself to punish the innocent-not merely definition­ally impossible-and also wrong in itself to punish the guilty beyond the degree of their guilt, are far less confident that a previous wrong act in itself, if considerations of deter­rence and the like are clearly set aside, even tends to provide a positive justification of a pen­alty. People are often negative but not positive retributivists. The other reason is that if we try to develop and explain these various retributiv­ist theses, the positive and negative ones seem to call for or to fit in with quite different lines of thought. For example, very general and obvi­ously attractive principles about the rights of individuals or human rights provide immediate backing for the thesis that it is wrong to punish the innocent; but these take us no distance at all towards a positive reason for punishing the guilty just because they are guilty.

Even negative retributivism, however, is not without its problems, particularly if it is taken to include permissive clauses: you must not punish the innocent, but you may punish the guilty, you must not punish excessively, but you may punish up to a proportionate degree. These permissive clauses still do not assert positive retributivism, they do not say that wrong acts are positively a reason for imposing penalties; but they do say that wrong acts somehow cancel the basic reason for not imposing penalties; the guilty person loses his immunity in proportion to his guilt. Strictly, then, we should recognize permissive retributivism as a third variety, inter­mediate between negative and positive. The three principles can be stated briefly as follows:

• Negative retributivism: One who is not guilty must not be punished.
• Permissive retributivism: One who is guilty may be punished.
• Positive retributivism: One who is guilty ought to be punished. (This "ought" may vary in strength between indicating only a weak prima facie reason and asserting an ab­solute requirement.)

The "must not," "may," and "ought" in each of these is moral. Also, these are to be consid­ered as principles, morally valid in their own right. Obviously, the same statements could oc­cur as derived rules in, for example, a utilitarian theory, where they would be explained and jus­tified by their beneficial consequences, but they would not then constitute any sort of retributiv­ism. For completeness, we should add quantita­tive variants of each of these three principles. Once these complications have been introduced, some further explanation is certainly needed both for the quantitative version of negative re­tributivism and for both versions of the permis­sive view. However, I leave these problems aside; they are as nothing compared with the difficulties in positive retributivism.

So far I have been drawing mainly formal dis­tinctions, and have thus isolated the general form of a positive retributivist thesis, that a wrong act in itself positively calls for the impo­sition of a (proportionate) penalty. But this is a very dark saying. Why should it be so? Is there any way in which we can make sense of this curious principle, at least by relating it to some network of other moral ideas?

It is here, as developments or explanations or further justifications of the positive retributivist thesis, that several of the approaches listed by Cottingham come in: repayment, satisfaction, fair play, placation, annulment, denunciation. "Desert" is not such a further explanation, but is just the general, as yet unexplained, notion of positive retributivism itself. But these (and one or two more that have been suggested, which Cottingham does not list) are a remarkably un­impressive collection. They are, indeed, the philosophical analogues of the sort of oppo­nents that a chess player would like to have if he were going to play, blindfold, a dozen or so op­ponents at once. Each of them can be dispatched, with ease, in a very few moves.

Satisfaction: What this might justify, if it jus­tified anything, would seldom be proportional to the wrongness of the crime. It would entail severer penalties where the surviving victims of the crime were of a vindictive nature, but lighter ones where the victims were forgiving or "Chris­tian" in spirit. It would entail the punishment of any behaviour in proportion to its unpopularity, whether it was criminal or not. In any case, this is obviously a consequentialist justification: the penalty is being justified by the satisfaction it is likely to produce. Its connection with retributiv­ism is merely that widespread satisfaction, felt by others besides the surviving victims of the crime, will be produced by the punishment of criminals only if many members of the society are already retributivists. But that does not make this itself a retributivist justification, and it has no power to explain any strictly retribu­tivist principle.

Placation: This, too, is obviously consequen­tialist. The reason it offers for punishing rests upon the supposed bad consequences of failing to punish-the god will still be angry. But these consequences themselves depend upon the sup­posed retributivist opinions of the god in ques­tion. Far from explaining retributivism, this ac­count requires that there should be some other morally adequate explanation of retributivism, if it is to be ascribed to a morally respectable god.

Denunciation: This also seems to be consequentialist. It has sometimes been called the "educative" theory of punishment. It tries to justify punishment by way of its tendency to produce in the public an attitude of greater hos­tility to the crime. This also, therefore, has no tendency to explain the retributivist principle, though, like satisfaction and placation, it may presuppose retributivist views. A public which already sees punishment as being deserved by the wrongness of the previous act is more likely to interpret a punishment as a dramatic asser­tion that the act punished was wrong.

These three approaches, then, are not really retributivist, though in one way or another they ride on the back of retributivism. Three others of the approaches listed, repayment, annulment, and fair play, do a little better in that they are genuinely retributivist in our sense. They are re­trospective, non-consequentialist, because in each case the alleged justification is complete once the penalty has been inflicted; no further results of this are relied upon. But otherwise these approaches are equally unsatisfactory.

Repayment: How does the criminal's suffer­ing or deprivation pay anything to society? No doubt repaying a debt often hurts the person who pays it, but it does not follow that anything that hurts someone amounts to his repaying a debt. If Jones owes Smith $10, it will hurt Jones just as much if he has to throw $10 away as if he has to give it to Smith, but throwing it away will not repay the debt. If the criminal's suffering or deprivation or incarceration doesn't do any good to society, it cannot be the payment of a debt to society. So this account is simply inco­herent unless it is transformed into a theory of reparations, which are not punishment, or into the satisfaction theory which, as we saw, is consequentialist and will not explain retributiv­ism.

Annulment: This notion, as Cottingham says, goes back at least to Hegel. But the remark he quotes (" . . . das Aufheben des Verbrechens, das sonst gelten würde, and die Wiederherstel­lung des Rechts") would be better translated as referring to "the annulment of the crime, which would otherwise remain in force, and the resto­ration of right." Hegel's idea seems to be that as long as a criminal goes scot-free, the crime itself still exists, still flourishes, but when the criminal is adequately punished the crime itself is some­how wiped out. It is not, therefore, by any sort of repayment or restitution that "right is re­stored," but just by trampling on the previously flourishing crime. But this notion is simply inco­herent: there is no comprehensible way in which a penalty wipes out an otherwise still-existing crime. What makes this suggestion plausible is a confusion with deterrence and prevention. A penalty may help, by deterrence, to wipe out, not the past crime, but other similar crimes which, but for the deterrence, would be com­mitted either by this same criminal or by others; that is, it may help to wipe out a criminal tradi­tion or practice. Similarly, the locking up or killing of a criminal will put a temporary or a permanent stop to his criminal career. But in neither case is it the past crime that is annulled or wiped out, and deterrence and prevention are plainly consequentialist justifications, not re­tributivist ones.

Kant's remark that if the "last murderer" is not executed blood-guilt will adhere to the peo­ple, which Cottingham takes as an appeal to the placation theory, might be better understood as something like an anticipation of Hegel's notion of annulment. The guilt, the badness, of the crime is still floating around as long as the criminal is unpunished, and by letting him go free the people would share in this guilt as ac­cessories after the act. But to suppose that this guilt is extinguished by the carrying out of the penalty requires the incomprehensible Hegelian notion of annulment, and to say that the people would be accessories if they did not punish the criminal presupposes, and therefore cannot ex­plain, the principle that the crime in itself mor­ally requires the penalty.

Fair play: Here the suggestion is that the var­ious members of society are all in competition with one another, a competition governed by rules. The criminal has gained an unfair advan­tage by breaking the rules; to restore fairness this advantage must be taken away from him. Unlike our last two approaches, this one is not incoherent; it makes perfectly good sense in cer­tain contexts. It has its clearest exemplification in the award of a penalty in football (soccer) when there has been a foul in the penalty area, that is, when a player or his shot has been ille­gally stopped in the neighbourhood of the other side's goal. It is also retributivist in our basic sense of being retrospective; the justification is complete when the penalty has been imposed; fairness has then (roughly at least) been re­stored, irrespective of whether there are or are not any further desirable consequences. The trouble with this approach, however, is that it has little relation to most cases of punishment. Any serious attempt to apply it would lead to bizarre results, not at all well correlated with what is thought of as desert or degree of wrong­ness or guilt. It implies that penalties should as far as possible be proportional to the advantage that the criminal has gained, in the social com­petition, by breaking the rules, just as a foul near the opponents' goal is penalized much more than one elsewhere. Thus if a businessman has secured a contract worth $100,000, but has exceeded the speed limit in order to get to the relevant appointment on time, he should pre­sumably be fined $100,000, whereas a fine of $1 would be enough for someone who murders a blind cripple to rob him of $1. And so on. Un­successful attempts at murder (or anything else) should not be punished at all. There should be an advantage rule, as there is in football, where the referee will not penalize an infringement if the innocent side still has the advantage any­way. So while there are things that this fair play principle can plausibly be used to justify, it has little bearing on most legal penalties and will not serve as an expansion or explanation of the basic priciple of retribution or desert, of the idea that a wrong act in itself calls for a penalty proportionate to its degree of wrongness.

But perhaps it will be said that what matters is not the loot but the self-indulgence. Someone who commits a crime, even if he gets very little out of it, has still let himself go in a way that law-abiding people do not, and his punishment only brings him back into balance with their self-restraint. But when we switch to this topic of self-indulgence the analogy of a competitive game becomes even less apt. The ideal of equality in self-indulgence is so remote that we hardly even think of it. If it were unusual self­indulgence that we objected to and wanted to counterbalance, it would be the rich, the owners of yachts and racehorses, and businessmen with elastic expense accounts, that we should be get­ting after, not thieves and murderers. Again many murders arise out of domestic conflicts, and most of these only from long-standing problems. When people eventually kill their husbands or wives or lovers, they have probably already shown far more self-restraint, in not having so earlier, than those who are more hap­pily or more placidly situated, and on the score of equality in restraint and indulgence would be entitled to let their hair down and kill someone once in a while.

What is basically wrong with the fair play ap­proach, as these rather fantastic examples bring out, is that it focuses on the advantage that may have been gained by the criminal in some sort of social competition, whereas the point of pun­ishment surely lies not in this but in wrongness of his act and the harm that he has done or tried to do. We can agree that he has acted unfairly, and that this implies that he cannot say that it is unfair if he suffers some proportionate penalty, if someone acts against him in a way that would otherwise have been similarly wrong. But this takes us only to a defence of permissive re­tributivism. It is still quite unclear how fairness to the law abiding is secured by the punishing of the law-breaker in itself, that is, unless this pun­ishment secures some benefit for the law-abid­ing. So the fair play theory still fails to yield any defence of the principle of positive retributiv­ism.

A suggestion not on Cottingham's list, which is often used to support retributivism, is that it peculiarly respects the dignity of persons on whom penalties may or may not be imposed. We can see the force of this as a support for negative retributivism, though it is an under­statement to say only that someone's dignity has been invaded if, while innocent, he is made to suffer for the sake of some benefit to others. Again, it may help to explain permissive retrib­utivism: someone's dignity is perhaps being re­spected if it is his own choice that cancels his immunity. Penalties which accord with negative and permissive retributivism do not invade any­one's basic area of freedom. But these consider­ations give no support whatever to positive re­tributivism once it has been clearly distinguished from the other two views. It is quite incomprehensible how punishing a person because he has done wrong respects or enhances his dignity. Even as an argument against retri­butivism's traditional rivals as positive justifica­tions of punishment, deterrence and utilitarian­ism generally, this has force. At most it shows that there should be restrictions on the deter­rent (etc.) use of punishment, and this is com­monly agreed by utilitarians although they hold that deterrence and the like are the only positive reasons for punishing. The utilitarian view can be and usually is developed in a way that intro­duces the negative and permissive theses as de­rived rules. Positive retributivism cannot gain even a comparative advantage by this appeal to dignity.

An even stranger suggestion on the retributiv­ist side is that the criminal has a right to be punished. In general a right is something that the right-holder can either exercise or waive as he chooses, and most criminals would gladly waive this alleged right. If one of them would not, it could only be either because he actually wanted the penalty for itself, and then it would not be a punishment, or because he was himself a very keen retributivist-but then this, as a moral explanation of retributivism, would be circular, like the theories of satisfaction and pla­cation we considered earlier. An associated view is that the criminal wills his own punishment. This may be just the point already noted, that it is by his own choice that he has lost his immu­nity; but this supports only the negative and permissive theses, not the positive one. Alterna­tively, what is invoked here may be Rousseau's general will: the criminal, qua citizen, sub­scribes to and helps to make the universal law by an application of which he is punished. Where this is true, it will help to justify punish­ment, and may indeed justify positive re­tributivist theses as derived rules about the im­position of penalties. Once the validity of the universal law attaching a penalty to a crime is established, the only further justification needed to provide some positive reason for imposing the penalty in a particular case is the retrospec­tive one that a crime of the relevant sort has been committed. But this-which is, perhaps, the core of what Cottingham calls the penalty theory-explains positive retributivism only as a working rule, not as a justifying principle in its own right. Everything depends on the prior validity of the universal law, and this, on the present approach, depends on the two ques­tions, whether the general will has actually made the universal law that is being applied and whether it was reasonable for it to do so. Since all the arguments for the positive retributivist principle are so weak, an affirmative answer to the second of these questions would have either to invoke non-retributivist considerations, or to move to some non-objectivist point of view, or both.

I I

And that, it seems, is that. Every one of these approaches fails completely to supply any co­herent expansion or explanation of the re­tributivist principle, and, as far as I know, there are no others on offer. On the other hand it is plain that a considerable number of people have what would be called an intuition that a wrong action in itself calls for the infliction of suffering or deprivation on the agent. In fact we might go further than this. Many people, perhaps most people, find unsatisfactory what I might call the standard compromise position about punish­ment.

What I mean by the compromise position is this. Negative retributivism is explained as a consequence of the rights of innocent persons, and permissive retributivism on the ground that a non-innocent person has to some extent lost those rights: rights make sense within a system of reciprocal rights, so that someone who has violated the rights of others cannot fairly claim the corresponding rights for himself. This sys­tem of reciprocal rights may in turn be ex­plained within, say, some kind of utilitarian the­ory, or again it may be put forward on its own. In either case, this line of thought brings us to the point that someone who has done wrong may be punished, though someone who has done no wrong must not. But-the compromise runs-the positive decision to punish is justi­fied, if at all, by the benefits that are likely to result from so doing, for example, by way of the deterrence of crime. So we combine negative and permissive retributivism (themselves ex­plained immediately, in terms of rights, though perhaps also, more indirectly, in terms of utility) with a positive justification of penalties that ap­peals directly to utility.

This compromise looks neat, and has there­ fore commended itself to some philosophers. But I think that most people will, on reflection, find it unsatisfactory. They will not feel happy about imposing a penalty on someone for the sake of benefits to others, even if he has lost his right to immunity, unless they can say further that he positively deserves this penalty. Strange though this may be, we want to be able to ap­peal to a positive retributivist principle as well as to whatever utilitarian or in general consequentialist justifications there may be. Also, positive retributivism is not a completely isolated moral principle: it has a counterpart in the view that gratitude and reward are morally appropriate. We feel that someone who has helped another person deserves gratitude, and that someone who has done good generally, especially something outstanding good, deserves some reward. And he deserves a reward simply because of what he has done, not because re­warding him will bring about this or that good further consequence.

Indeed, we could go further still: our basic moral concepts themselves of good and bad, right and wrong, cannot be adequately analysed unless we include a retributive element, the no­tion of returning benefit for benefit and harm for harm. What is bad or wrong, for example, is not simply what is harmful or what is forbid­den: it is thought of also as what deserves con­demnation, where condemnation is some kind of hostile and unpleasant reaction. Equally, what is good or right is not simply what is bene­ficial or what is recommended: it is thought of also as what deserves praise and commendation, what calls for a friendly and pleasant response.

Have we, then, been engaged in a vain and unnecessary task of trying to explain and justify positive retributivism? It is rather true merely by definition that what is wrong or bad deserves punishment and that what is right or good de­serves reward? Is is part of the very concepts of wrong, bad, right, and good that they carry such deserts with them?

It would be a mistake thus to dismiss the problem. For even if it is part of these concepts that good and bad deserve reward and punish­ment, there is still the question whether and how those concepts themselves are to be justi­fied or explained. For example, if the concept of something's being bad combines the notions of its being harmful and forbidden and deserving a hostile response, we still have the question why these three notions should go together, in particular why what is harmful and forbidden de­serves a hostile response. That these elements go together is a synthetic judgment, even though the combination of them emerges from the analysis of bad.

Now as allegedly objective moral principles, this combination, and the corresponding combi­nation of being beneficial, recommended, and de­serving of a favorable response, are still quite ob­scure, though they are undoubtedly deeply entrenched in our thought. But as soon as we make the Humean move of saying that moral distinctions are founded on sentiment, not on reason, the obscurity disappears. If we ask, not, "Why do wrong actions deserve penalties and good actions deserve reward?" but rather "Why do we have an ingrained tendency to see wrong actions as calling for penalties and good actions as calling for reward?" then it is not difficult to outline an answer.

First let us consider the tendency to feel and show gratitude for benefits, and its hostile coun­terpart, the tendency to feel and express resent­ment of injuries. There is no doubt that these are very widespread human dispositions, and what look like the same or similar tendencies are not difficult to detect in some non-human animals. Nor is there the slightest difficulty in understanding how such instinctive patterns of behaviour, and feelings that harmonize with that behaviour, could have developed by an evolutionary process of biological natural selec­tion. Since gratitude will reward the benefactor, and since tendencies that are rewarded are thereby psychologically reinforced, gratitude will tend to encourage further similar benefac­tions, and so will benefit the creature that feels and displays gratitude. We are not, of course, saying (as Hobbes did) that what looks like gratitude is really egoistically calculated behav­iour aimed at encouraging future benefits. The feeling of gratitude may be and normally is straightforward, sincere, and non-calculating; it is, so to speak, the natural selection process that does the calculating, not the grateful agent him­self. Thus uncalculating gratitude will be pro­duced because of the benefits it is likely to bring, but the agent himself does not act gratefully for the sake of or in anticipation of those benefits. Similarly resentment of injuries is likely to dis­courage similar injuries, and so again will bene­fit the creature that feels and displays resent­ment, and thus there will be natural selection in favour of resentment, though again the resent­ment in the agent himself will normally be spontaneous, not calculated.

Gratitude, however, is a favourable reaction by a particular agent to actions that have bene­fited him, and resentment similarly is a hostile reaction by a particular agent to actions that have harmed him. Reward and punishment are more generalized forms of retribution: they are responses to actions that are seen as good or bad in general, not as good or bad to some par­ticular agent, and the retributive attitudes which we have to explain are those of seeing the good or bad actions as calling for reward or punish­ment as such, not as calling for favourable or hostile reactions from particular agents. The at­titudes to be explained, therefore, are general­izations of gratitude and resentment in these two respects: specific reference to those to whom good or harm has been done is elimi­nated, and the attitudes arise in persons gener­ally who know about the good or harm, not only in those who are themselves benefited or harmed.

However, once we have explained gratitude and resentment, it is easy to understand these generalizations of them. Hume would say that such generalization arises from sympathy; but a sociological explanation would be better than this purely psychological one. Within a social group, people tend to feel and show gratitude and resentment on behalf of one another, in some respects at least, because they are recipro­cally benefited by doing so. Cooperation in re­sentment and gratitude grows up for the same reason that gratitude itself does, that it tends to be beneficial to each of the cooperators. This may be either an instinctive tendency, to be ex­plained by biological natural selection, or a cul­tural trait or "meme," to be explained by a pro­cess of social evolution that is an analogue of biological selection. In saying this, we must guard against three errors. First, we are not say­ing that this cooperation is calculating from the point of view of the cooperators themselves: they feel genuine sympathy, righteous indigna­tion, and the like. But they have come to feel in these ways as a result of some biological or so­cial selective process in which the benefits of such cooperation has played a key role. But, secondly, we are not saying that this coopera­tion in gratitude and resentment has arisen be­cause of its general utility, because the social group as a whole has been benefited by it. No doubt the group is benefited, but that would not in itself be a good explanation. A practice which would be socially useful may fail to arise; it is unlikely to arise unless it produces results which favour its own spreading and surviving, unless it works so as to propagate the genes or "memes" on which it depends; and in general this re­quires that it should benefit the individuals who engage in that practice. But cooperation in gratitude and resentment meets this require­ment: it can grow up as a convention which in fact benefits all parties, even without being de­liberately adopted for that purpose. However, thirdly, it is just not the case that all sorts of gratitude and resentment are generalized equally. Only some kinds of harm are socially, cooperatively, resented, and cooperation in gratitude is even more restricted. Again we must seek and can find sociological reasons for these differences: only with particular kinds of harm are the conditions favourable for the growth of a convention of cooperative hostility to them, so only some kinds of harm are seen as wrong and as calling for general resentment and punish­ment. For example, theft is cooperatively re­sented, but commercial competition is not.

In conclusion, therefore, I maintain that it is easy to understand why we have a deeply in­grained tendency to see wrong actions as calling for penalties and some sorts of good actions as calling for reward. Though retributive principles cannot be defended, with any plausibility, as al­legedly objective moral truths, retributive atti­tudes can be readily understood and explained as sentiments that have grown up and are sus­tained partly through biological processes, and partly through analogous sociological ones.

That is why I have, in the title of this paper, called retributivism a test case for ethical objec­tivity. On objectivist assumptions, its force and its persistence are incomprehensible, but on a subjectivist or (in Hume's sense) sentimentalist approach they are easily understood.

NOTES

1.      Jeffrie G. Murphy, in "Cruel and Unusual Punishments," to appear in the proceedings of the 1979 Conference of the Royal Insti­tute of Philosophy on Law and Morality.

2. J. G. Cottingham, "Varieties of Retribution," Philosophical Quarterly, July 1979.