MERCY, UTILITARIANISM AND RETRIBUTIVISM ANDREW BRIEN

Even though early utilitarians specifically allowed for the remission of legislated punishments in certain specified circumstances - what are generally today called pardons or 'acts of mercy' - they argued that in a perfect utilitarian system, mercy would cease to exist. So, we find Bentham saying at one point, 'Remission of punishment, yes; for that, there may be good reason on various occasions; but they are all of them capable of being, and all of them ought to be, specified'. A t another point we find him setting out the case against mercy: What...in every case, the greatest happiness of the greatest number requires, is - that if, on the occasion in question, the application of the punishment in question would be conducive to that happiness, the punishment should be applied; if not, not .... Under a government which has, for its actual end, the greatest happiness of the greatest number, thus it is that mercy is unknown. That mercy and utilitarianism are incompatible is also a claim found in some of the most influential contemporary philosophical discussions of mercy. In her famous article 'Mercy' in which she resurrected the contemporary debate about mercy, Alwynne Smart, expanding on Bentham's point, wrote: Real mercy is never a possibility for [the utilitarian] because he must always impose what is according to his ethic, the fully justifiable penalty . . . . [T]here is no significant sense in which the utilitarian can say, 'I ought to do such and such, but special considerations persuade me to act differently on this occasion'. For

493

AFa)REWBRIEN

him, the statement, 'I shall act mercifully' can only mean 'I shall impose a penalty less than the one which will produce most good,' which in turn can only mean 'I shall impose a penalty less than the one which w i l l produce most good because this action is the one which will produce most good'. Many who agree with this view also deploy another argument for the incompatibility of mercy and utilitarianism that relies upon a particular analysis of the concept of mercy. This argument seems to show that mercy is logically possible and morally justifiable only within a moral outlook based around notions of justice, justice-based desert and retribution. Since utilitarianism accords these notions no independent moral weight, the utilitarian is precluded from acting mercifully. These conclusions are disturbing. Commonsense morality accords mercy a high moral value, yet also often gives great weight to utilitarian considerations, especially in matters of public policy. Moreover, if it is true that mercy and utilitarianism are incompatible then the ideal utilitarian legislator cannot act mercifully. God may be merciful but He cannot be a utilitarian; and if He is a utilitarian, He cannot act mercifully. The purpose of this paper is to argue that commonsense morality is correct and that the views found in the contemporary and classical literature are mistaken; in other words, contrary to conventional philosophical wisdom, acts of mercy are logically possible and justifiable within a utilitarian ethic and any version of it that grounds public policy. Importantly, I shall also argue that unless the retributivist modifies her view about the nature of mercy, she is morally unable to find a place for mercy within her outlook. A consideration of this problem is important for a number of reasons. Mercy, while highly valued, has not been studied as thoroughly as some other moral ideas. This is surprising since mercy and justice are seemingly incompatible and there is a prima facie difficulty in accommodating mercy within a utilitarian outlook. Much still remains to be discovered about mercy and its place in moral life. An examination of this problem provides an opportunity to analyse the concept of mercy - with some surprising results.

494

MERCY,I3IK/I'ARIANISMANDREIRIBUI'IVEM

Further, any discussion of punishment must explain the status of mercy; and discussions of utilitarian punishment have been remarkably silent of the topic of mercy. More importantly, however, the solution I advance will involve adopting a version of utilitarianism that is particularly useful, not only within the context of punishment, but the broader context of social and political philosophy and public policy. This version of utilitarianism does not suffer any of the traditional problems (victimisation, high decision costs, irrational rule-worship) faced by act and rule utilitarianism even though it retains the positive features of those doctrines. Throughout this paper the discussion centres on the relationship between utilitarianism (as a foundation for public policy) and acts of mercy, and the discussion will take place largely within the context of punishment. A discussion of the capacity of utilitarianism to accommodate the virtue of mercy is not considered in detail. As well, it should be taken as given that utilitarianism is a tenable outlook; whether it is, however, is not part of the present discussion. L The Classic Problem

Utilitarianism may be loosely characterised as the view that an option - action, or rule, or policy - is right if that option on balance produces or tends to produce when impartially calculated, at least as much good for the greatest number of people affected by that option being realised as any other option available. It is an agent's duty (and within the context of this paper that agent will be a public official) to realise only those options that do this. All other options that are open to an agent but which realise less value are morally wrong. Now, if mercy involves doing less than is justified, as is the case when a punishment is remitted, then such a remission is morally unjustifiable from the utilitarian's point of view. Choosing to act mercifully seemingly involves producing less value than is possible given the circumstances, whereas an agent is required to produce as much as possible. For this reason, mercy is morally wrong for the utilitarian. Suppose, in contrast, that some act does maximise value. Such an act is justified for the utilitarian and she is bound to perform that action. In this case, however, doing less than is justified, as mercy is

495

~

BRIEN

thought to involve in the case of punishment, is not logically possible. There is no sense in which the utilitarian may act mercifully by remitting a penalty that is itself justified because it maximises utility. The utilitarian seems to be caught in a dilemma: either mercy is unjustified because it involves doing less than is justified and so lacks any utilitarian justification. Or it is justified because it maximises utility; but then the act cannot count as an act of mercy, since there is no justification for any heavier penalty at all. In this case the act ceases to be a departure from anything and so it fails to be an act o f mercy. Mercy is logically and morally impossible within a utilitarian ethic. The utilitarian can act rightly, but never mercifully; or mercifully only by acting wrongly. This is the classic problem. Two other conundrums not widely discussed in the literature arise from the fact that mercy often is thought, pre-theoretically, to be supererogatory. If mercy is supererogatory one must then explain how utilitarianism can accommodate such actions. More problematic is a clear contradiction in our beliefs about mercy: if mercy can be justified within a utilitarian-motivated system then whenever mercy is justified it is obligatory. Mercy then seems to be both obligatory and supererogatory. A resolution of these conundrums will have to wait until we analyse the concept of mercy and discuss utilitarianism in more detail.

H. Mercy 'Mercy', 'merciful', the phrases 'act of mercy' and 'is merciful' are used to refer to many different sorts of things. For example, people, actions, the weather, the sea and divinities may all be described as 'merciful'. T h e present discussion is confined to actions. When an action, f, is described with moral approval as 'an act of mercy' or as 'merciful', typically it has a number of properties. It is, for example, an intentional, rational action, rather than capricious or accidental. One cannot act mercifully by accident. Acts of mercy aim to, or in fact do, prevent or at least interrupt, possible, likely or actual courses of events, which are threatening or burdensome to the putative beneficiary. In this way these events are

496

MI~CY, ~ A R I A N I S M

AND REIRIBUIIVISM

prevented from occurring or continuing either at all or to some significant extent. It can be seen from this that all acts of mercy are not merely aimed at particular identifiable things, usually people, but that such actions have specific targets, namely, the suffering, dire need or situation, threat or other burdensome state of affairs under which an actor labours o r with which she is threatened, and which she is powerless to alleviate herself. Thus, a person in a coma, who is mercifully killed has her dire situation alleviated; the criminal whose punishment is remitted or the beggar who receives alms, both have (to some significant extent) their dire needs met. The interesting point here is that the agent who is the putative object of an act of mercy (the beneficiary) may be unaware of her dire need or situation, or the threat facing her. For example, a person in a coma may be quite unaware of her comatose state. Yet if she is killed, such a release may be described as a 'mercy killing'. Her coma, it may be known counterfactually, would be regarded as a burden were she aware of it; and certainly it is not inappropriate to describe being in such a state as a burden not only to the comatose person's relations but to the person herself. Nor is it wrong to say that she is suffering a dire need. We possess all sorts of needs when we are asleep, f o r example the need for air, and simply being unconscious does not make them evaporate. In addition, to be an act of mercy f must be the sort of action which, from the mercy-giver's point of view, would be thought to confer a benefit upon the putative beneficiary by releasing her f r o m some dire situation or by meeting some dire need. All actions that are candidates for the epithet 'mercy' are helping actions. They must be thought to do or be likely to do the beneficiary some good. It is this property of the action that prompts the putative mercy-giver to select this action rather than some other action that may not assist the beneficiary. We can test this against our ordinary language intuitions: it just does not make sense for an agent with power over another to knowingly heap further burdens upon that person and then call it 'an act of mercy'. It can seen from this that in any mercy context there is a beneficiary and a benefactor. There is, importantly, a disparity in power and this power structure is found in all cases where 'mercy' is

497

ANDRE-WBRIEN

predicated o f some action, person, thing or event. So, on the one hand, there is a person who has the capacity to help another in some dire situation - the putative mercy-giver - and on the other hand there is a person who is in that dire situation and who is powerless to help themselves, and who must rely upon the putative mercy-giver to alleviate that person's dire situation. The beneficiary, is therefore, within the power - legitimate, illegitimate, or circumstantial - of the benefactor. Further, whoever is describing some action as an act of mercy, must believe that the beneficiary is vulnerable to the benefactor's acts and/or omissions and must be aware that the benefactor is well placed, if not uniquely so, to alleviate at least to some extent the beneficiary's need. We can see this in clear cases, such as the duellist who is merciful to his adversary, or in a more mundane case, such as the aircraft passenger who has it within his power to make a f e l l o w passenger, who is terrified of flying, suffer still further or to reassure him and in this way alleviate his fear. A few more comments need to be made about this account o f mercy. This account is meant to capture the idea that acts of mercy involve some sort of intervention in, and departure from, an actual, possible or likely course of events. This course of events may be uncaused by the putative mercy-giver, such as poverty or natural disaster; or it may be something that the putative mercy-giver has some reason in that context to bring about. For example, in a duel the vanquished being made to suffer by the victor. In the case o f punishment, mercy involves a forbearance or refraining f r o m punishing as harshly as one has the capacity to punish. It involves a departure from some standard, and it results in a wrongdoer not being treated as harshly as he might otherwise merit. On this conception, mercy is essentially a certain sort of actoption, identified by certain characteristic properties, and the terms ' m e r c y ' and 'merciful' when used to refer to actions, are names. Acting mercifully, on this view, involves one agent (the benefactor) choosing a particular act-option; namely, that act-option bearing the properties set out above. This account of mercy captures an idea reflected in ordinary language and which is a fairly standard part o f the recent discussion of the concept of mercy, that there is a difference

498

MERCY,LrrlL1TAR/ANISMANDREIRIBUIaVISM

between acting mercifully and being merciful. An action may be identified and specified as an act of mercy without any reference to the motives of the actor or other springs for the action. Consequently, observers are licensed to call f an act of mercy in virtue of the properties it manifests, even if nothing is known about the motives o f the actor, or more generally, the springs for the action. Another point to note is that the need or burden that is the target of an act of mercy may not itself prompt the act of mercy. An agent may act mercifully, for example, out of self-interest. Suppose some agent, call him Draco, who is a wealthy landowner, is owed a large sum of money by an impoverished tenant farmer. The time comes to repay the debt, but the tenant farmer is unable to do so. His options are clear: repayment or be sold into slavery. Socrates who happens upon this scene urges mercy upon Draco. But all Draco is concemed with is his right to the money or the life. Now suppose that Draco entertains hopes of a public career. Socrates knows this and points out to Draco that should he compel repayment of the debt then the public will surely revile him. Draco remits the debt in order to win public office. A spectator, who may know nothing of Draco's motives, is licensed to call Draco's action 'an act of mercy'; and even if Draco's motives were known, a spectator would say something like: 'Draco acted mercifully so that he would be elected.' This highlights the point that in order for f to be an act of mercy, f need not spring from, or be an expression of, compassion for another's plight, or care or concern for her welfare, or be an expression of goodwill for the beneficiary, as some writers have claimed it must and as is often associated with 'mercy'. Draco's release of the tenant farmer from his pledge may count as an act of mercy, even though Draco fails to possess and display any goodwill and compassionate concern for the tenant farmer's welfare; or in other words, he lacks the very dispositions associated with displays of the virtue of mercy. This highlights the point that 'being' a merciful person, that is, possessing the virtue of mercy, is to possess that trait of character that disposes a person to perform merciful actions and select merciful act-options based upon certain characteristic springs. Such springs include that of rendering assistance to another who is in dire need and

499

ANDREWBRII~

who is powerless to help herself, and who is, because of this helplessness, within the beneficiary's power. Moreover, f may be an act of mercy even though it may not be selected because it is the gentler option or because the putative mercygiver accords some especial moral value to gentleness, but because it is an option that better promotes some goal that may have nothing to do with the welfare of the beneficiary. Draco, when he finally forgoes his claim over the the tenant farmer's life, does so, not because he has been swayed by Socrates' advocacy of mercy and so suddenly values gentleness, but because his selection of the less harsh option will save him from the wrath of the public. This account of mercy may seem to generate the following difficulty. On the account of mercy advanced here a putative mercygiver will often not only be faced with the option of not acting mercifully to person P (option A) or acting mercifully to person P (option B) but acting in some way that is, from the same point of view, even gentler than option B. Call this option C. Choosing option B rather than the even more lenient C may be described as an act that is merciless. Hence option B is at the one time both an act of mercy (since it is gentler than option A) and merciless (since it is harsher than C). The solution is simple. Option B confers a benefit upon P but not as great a benefit as option C. It is in virtue of this that B appears merciless. This, however, is the wrong way to look at the situation. The original burdensome course of events which selecting option A would have maintained, has been departed from by both B and C. For B to be merciless it would still need to impose upon P a great burden. Yet if B is merciful, as ex hypothesi it is, then it must significantly reduce or remove altogether the burden that A would have imposed. It does not seem that B could warrant the epithet of 'merciful' (in view of being less harsh than A) and still be sufficiently harsh enough (in view of option C) to impose a great enough burden upon P to warrant the epithet 'merciless', Not conferring as much of a benefit as one can does not count as a merciless act, if it is also true that the not-so-beneficient act confers, nevertheless, a significant reduction in the original threatened burden, as ex hypothesi B does. For this reason, we may say that option B was

500

MERCY,~ A R I A N I S M ANDRETRIBUTIVISM

not as merciful as C. But the tag 'merciless' has the connotation of 'lacking in mercy'. Clearly option B is not lacking in mercy, when compared to option A. This answer holds while the same point of view is maintained; that is, while we can see options A, B, and C. If, however, we view options B and C only then, of course B may appear merciless when compared to C. But then we have changed our point of view. We would not have any reason to think that B was merciful, since this description of B depends upon the observer knowing about option A. For B in this sort of case to be thought to be both merciful and merciless is nothing more than having one sort of property in one context, and another incompatible property in another. Against this it might be argued that motives do matter. Suppose that A murders B in order to collect an inheritance. Unknown to the murderer is the fact that B is fatally ill, in great pain and and longs to die. B has received a benefit from A, and the action was rational, intentional, rather than accidental. Has A acted mercifully? I think not, and the account of mercy developed here can show why. Part of what it is to perform an act of mercy is to be aware that your action is going to benefit the object of that action. And in this example A was unaware that her action was going to benefit B. Moreover, it must be directed, in some way, at alleviating the dire situation of the beneficiary, even if the motivation for the action was self-interest and the ultimate beneficiary is someone other than the person in the dire situation, as was the case when Draco acted. In the murder-for-gain example, the target of the action was not the suffering of B, even for selfish reasons, since B's suffering was unknown to A. So, this cannot count as an act of mercy, since the benefit conferred was conferred accidentally. What this highlights is that while motives do n o t matter, intentions do: the putative mercy-giver must be thought to be selecting this particular act-option because it will alleviate the dire situation of the beneficiary. The r e a s o n why she does this is irrelevant to correctly describing some action as an act of mercy. This can be seen in another way too: as a matter of the way language operates, people are licensed to call an action 'an act of mercy' even if nothing is known about the actor's motivational state. A familiar example will suffice. Suppose we are walking in the

501

~

BRIEN

woods and happen upon two men, one holding a sword at the other's throat. They are the obvious participants in a duel. They do not see us. Suddenly, the man who is obviously the victor throws down the sword and walks off. It is clear that he intended to throw down the sword, though we do not why, i.e. his motives are unknown. It would be natural to describe his action as an act of mercy, even though we know nothing about the springs of his action. Motives then do not matter when it comes to identifying an action as an act of mercy. If motives were always important to identifying an action as being of a particular type, rather than being important to determining the moral quality of an action, as motives are, then agents would be prevented from describing any action as being of this or that type unless they had complete information about the springs and aims of that action. In this respect 'mercy' is like many other names for actions, such as 'kindness', 'gentleness' and so on, which may be predicated of an action without any knowledge of an actor's springs. This is a very broad account of mercy. It captures actions that are not only regarded as paradigm examples of mercy, such as the mercy of the victor to the vanquished, but also the mercy shown by the mercy-killer to her 'victim', the angel of mercy to the beneficiary of her action, as well as the mercy that characterises the work of Mother Theresa and the Judaeo-Christian God. This is possible since mercy is conceived as a particular sort of act-option that is chosen because it alleviates the dire situation of the beneficiary, and which involves, essentially, a departure from some ongoing or threatened course of events. An important point to note is that some kind, beneficent, just or other actions that benefit others will be merciful too. But not all of them, since such actions can occur in contexts that lack power structures, dire need, threatening burdens, or imposing situations. One action may have two descriptions, but all acts of mercy must have the sorts of properties discussed above. Moreover, the reason for choosing a merciful act-option may not be clear, though some cases of putative mercy carry their reasons with them: mercy-killers act necessarily out of compassion for the beneficiary, for example. But typically, when it is claimed that 'Jones acted mercifully' only his action is being described and no claim at all may be made about the springs for his action. Given this analysis

502

MERCY,LU'K.IrAR/ANISMANDREIRIBUTIVISM

of mercy, the problem for utilitarianism is that of finding a way of justifying those act-options which it is appropriate to describe as 'an act of mercy'.

IlL Mercy and Retributivism It is important to distinguish the classic problem from the other conceptual problem advanced by some contemporary writers. For example, Igor Primoratz argues that 'acts of mercy are logically possible only against a background of a conflict of two ideals: the principle of just, deserved punishment and the ideal of mercy.' Consequently, utilitarianism cannot logically accommodate mercy because utilitarianism does not give any independent moral weight to the notions of desert and justice, and it makes sense to say that we have shown mercy to an offender only if we have remitted completely or partially a deserved punishment. Now, Primoratz is correct in thinking that there is a conflict between mercy and retributive justice. Retributive justice involves (speaking generally) people getting what they, as a matter of justice, deserve; mercy involves them receiving something less; so, prima facie, there is a tension between mercy and retributive justice. In fact, much of the contemporary debate has been concerned to resolve this problem. Primoratz is mistaken, however, in thinking that it is logically wrong to say that mercy has been shown to an offender only if a retributively deserved punishment has been reduced or remitted altogether. An offender is any actor who fails to comply with a law, or more generally, a rule. In virtue of failing to comply, a particular sanction or penalty is permitted. There is no necessary connection between being an offender or being liable to punishment, on the one hand, and desert and retribution on the other. It is natural to describe cases where an agent declines to inflict a penalty that is licensed, and instead inflicts a penalty less severe, as 'an act of mercy', even if we are unaware of the justification for the penalty and the authorisation to inflict it. For example, systems of sanctions established by sporting clubs, social clubs and universities, as well as being the basis for inflicting particular sanctions, are generally utilitarian in nature. In such contexts, if the rules were routinely enforced against

503

ANDREW B ~

violators, then if an agent who is authorised to enforce the rules departs, for whatever reasons, from the licensed sanction or penalty, and imposes a lesser penalty, it is natural to describe such a departure as 'an act of mercy'. This is true even though the penalty that was licensed for that particular offence was not grounded on retributive considerations or justified by the fact that the offender deserved it, as a matter of justice. This highlights an important point. An action is an act of mercy, not because it involves essentially a departure from justice or a deserved punishment, but because it involves a departure from some possible or likely burdensome course of events through the choice of a beneficial act-option over one more burdensome. Acts of mercy are not, therefore, essentially tied to retributive contexts. Mercy tempers cruelty, indifference, irrational rule-worship, apathy, bigotry and ignorance, amongst other evils, as much as justice. Since the context in which an act of mercy occurs is irrelevant to it being an act of mercy, the second difficulty apparently faced by the utilitarian dissolves. An interesting consequence of Primoratz' argument is that it seems to undermine the retributivist's capacity to accommodate mercy. Pure retributivism is a single principle theory. It maintains that what is retributively just is right. In selecting an action the retributivist (should she be true to her lights) is unable to recognise reasons that are not grounded in retributivist considerations. The retributivist must do what is retributively just, and if some action is right, then that is what she ought to do. Special considerations that are non-retributive in nature have no moral force for her. There is no significant sense in which the retributivist could say, 'I ought to do such and such, but special considerations persuade me to act differently on this occasion'. Mercy would always seem impossible (as any putative act of mercy would, if justified, be a departure f r o m nothing, since it would be an act of retribution) or wrong (since i t would be unjustified on retributive grounds). So, mercy is excluded as a possible and moral act-option for a pure retributivist. Primoratz and the other contemporary writers who believe that retributivism is unique in being able to accommodate mercy are mistaken, since it seems that retributivism cannot accommodate mercy at all.

504

MERCY,UTILITARIANISMANDREIRIBUI'WISM

Moreover, the usual solution to the tension between mercy and retributive justice is to argue that retributive justice does not require that a person 'get what he deserves', but merely permits or licenses a particular imposition. So, retributivism merely permits by not prohibiting mercy. However, this solution will not work. It .remains the case that mercy either will be unjustified on retributive grounds or be morally impossible. On the one hand, if some putative act of mercy were justified on retributive grounds, then it would be an act of retributive justice rather than a genuine act of mercy; thus, there would be no sense it which we could contrast an act of mercy with an act of retribution. On the other hand, if the act were not positively sanctioned by retributive justice then any agent committed to retributivism would find mercy morally impermissible. Mercy, as an action that is distinct from an act of retribution, is not possible for the retributivist unless she abandons retributivism and permits other reasons to count in justification. If she does this then the question naturally arises: Is she still a retributivist? Or is she the adherent of some hybrid theory of which retributivism is one component? Whatever the answer, mercy and retributivism have not been reconciled.

IV. Mercy and Utilitarianism Utilitarianism is usually divided into act and rule varieties. We shall examine the relationship of mercy to both varieties of utilitarianism before moving on to a general solution to the problem. Act utilitarianism may be characterised as the view that an action is right if it, of all the actions an agent could perform on that occasion, produces o r tends to produce the most valuable consequences. Prima facie, there is no room for mercy: either mercy involves remitting a penalty and doing less than is required, in which case it lacks utilitarian justification; or the putative merciful action is exactly what is required to promote utility and so it is, from the utilitarian's point of view, justified. But then the putative act of mercy fails to be a genuine act of mercy, since the agent has not refrained from inflicting any harsh treatment nor departed from inflicting some other merited and harsher treatment.

505

ANDRE-WBRIEN

While this argument appears plausible, it will not do. We recall that mercy was characterised as an act-option with particular properties. Now, an act utilitarian will be confronted with many different act-options and merciful act-options will be one of them. When an act utilitarian comes to determine which of the options facing her is right, she will be looking to see which of the options has consequences of the greatest value. Unless an option is available and capable of realisation it cannot be evaluated and ranked. So, if mercy is one act-option facing an agent and some other harsher treatment is another, both options must be open and available to be chosen if they are to be capable of evaluation. Although the act utilitarian must select that option that will maximise utility, the fact that one option is morally unavailable to her does not mean that she cannot be untrue to her lights and select the non-maximal option. In fact, there are many sorts of cases where there may be a temptation to select the harsher option or where one may believe that the likely beneficiary merits harsh treatment: duels and acts of revenge, or a belief that a wrongdoer is evil, being some examples. In the case of punishment, there may be no temptation, but there is still the harsher option, the likelihood of which increases as the utility of mercy becomes less clear. In all cases, however, there is a threat of harsh treatment (since harsh treatment is one option to be evaluated). It is within the power of the mercy-giver to depart from this option; and by selecting the more lenient act-option, albeit for reasons of utility, she does so. In doing this she prevents the option involving harsher treatment from being imposed, and because of this choice, the threat posed by the harsher option is alleviated. It is natural to describe such a choice as merciful and an agent who made that choice as having acted mercifully. It might be argued against this that a mercy-giver must, within the context of punishment, intend to inflict the harsher penalty, but decides for some reason not to. Since the utilitarian does not intend to inflict the harsher option, but merely examines the options open to her, and selects on the basis of utility the more lenient act-option, her action cannot count as an act of mercy. Underlying this argument is the assumption that an act of mercy necessarily involves intending a more burdensome thing then doing another more mild thing. This is

506

MERCY,UI'IL1TARIANISMANDREIRIBUIIVISM

wrong. A judge, for example, may have no idea what penalty is appropriate and so intend nothing. She may invite submissions from counsel as to the penalty to he imposed. Should she admit that the particular offence in question usually attracts a very harsh penalty but then go on to select the more lenient penalty urged upon her by counsel, and justify the selection on the grounds of utility, it would he natural to describe her action as 'an act of mercy'. It appears then that the act utilitarian can accommodate mercy. This solution, however, suffers the usual problems of act utilitarianism: the high decision costs that result from calculating over each case, along with the apparent licence for moral opportunism. For these reasons act utilitarianism does not engender confidence in the institutions in which it is used and is unsuitable for public policy purposes. The usual response to these problems has been to abandon act utilitarianism and adopt rule utilitarianism. Rule utilitarianism may be characterised as the view that the rightness of an option is not determined by its actual consequences, but by whether that option conforms to some rule that would, if generally adopted, produce or tend to produce when impartially calculated, on balance, at least as much good for the greatest number of people as any other rule, In general two reasons are advanced for adopting rule utilitarianism. First, unlike act utilitarianism, it saves on decision costs: an agent knows quickly and reliably whether an action is right or wrong in virtue of it falling under some rule. An agent does not need to calculate over individual cases. Second, rule utilitarianism can, if the rule is framed appropriately, provide a means of prohibiting victimisation of innocent people when it comes to inflicting punishment - the mere possibility of which is often considered a decisive criticism of act utilitarianism. In such a case the rule would specify that only the guilty be punished. The utilitarian reasons for this are clear. Cases of victimisation often come to light and do more damage to the legal system in the longer term than any short term benefits that may be produced. Second, it represents a guarantee to the subjects of the system that the innocent have nothing or little to fear. All people may walk a little taller and more confidently knowing that victimisation of the innocent is not

507

ANDREWBR1EN

sanctioned by the system. Not only are their lives improved, but the status and effectiveness of the legal system is enhanced by such guarantees. Can rule utilitarianism accommodate mercy? It seems not. Suppose the rule utilitarian adopts this rule: 'Every agent who performs an action of type A shall receive penalty P.' If this rule were generally adopted then a rule utilitarian who did what she ought to do could in no sense depart from some harsher course of events, and inflict a more lenient penalty; in other words, act mercifully. Unlike the act utilitarian who must determine which of the act-options open to her best promotes utility, the rule utilitarian has no range of actoptions to select amongst: there is only ever one action to perform, namely the one that accords with the rule. Departing for a moment from the context of punishment, it does seem true that both the act and rule utilitarian can act mercifully in other contexts, such as those involving charity. In these contexts the beneficiary is threatened with, or is in, some dire situation as circumstances are. If the utilitarian intervenes in the course of events and bestows a benefit upon some person in a dire situation, it would be natural to describe his action as an a c t of mercy. After all, without the intervention of the utilitarian the suffering would have occurred or at least continued unabated. To a c t mercifully in these circumstances (and remember, no claim is being made about the motivational state of the actor) is to intervene in an existing course of events and thus force a departure from them. In the case of the rule utilitarian, all her rule does is justify her act of mercy. For example, in such a case the rule utilitarian would apply a rule such as: 'Whenever I see suffering that it is within my power to relieve, and there are no reasons not to, I will relieve it.' All the act utilitarian does is calculate the value of the consequences of acting over the value of the consequences of refraining from acting and in that way justify her act of benevolence. Against this it might be argued that the utilitarian had to act as she did. The harsher option, namely allowing the person who was suffering to continue suffering, was never a real option. For this reason, the utilitarian's action fails to be an act of mercy. This is mistaken. It is true that the utilitarian had to act as she did and that

508

MERCY,UTILITARIANISMAND REIRIBUI1VISM

her action does not represent a departure from what she morally had to do. But the utilitarian's action is a departure from the existing course of events, namely the dire situation of the beneficiary. Her suffering would have continued had the utilitarian not intervened. The fact that the utilitarian felt compelled to act as she did, does not disqualify this action from being an act of mercy. What identifies an action as an act of mercy is not the motivation of the benefactor but the properties that the action itself has - as we saw in section I. It is true that the utilitarian's action is not a display of the virtue of mercy, since it lacks the compassionate motivation, disvaluation of harshness and high evaluation of gentleness, that are characteristic of that virtue; nor does such a description indicate anything about the springs for the action. We are concerned neither with the virtue of mercy nor the springs for mercy, but only with acts of mercy. On this account i t seems clear that the utilitarian can, when faced with the prospect o f bestowing a benefit in cases that do not involve punishment, perform such actions and be justified for doing so for utilitarian reasons. Let us return now to the context of punishment and a general solution to this problem. As is well known utilitarianism can be thought o f as a theory o f how actors ought to decide what to do; that is, a decision theory. Or it may be considered a theory about what makes an action right. On this view, it is a metaethical theory about the nature of rightness. As a metaethical theory it does not matter how or on what basis one decides to do what one does so long as, on balance, more happiness is produced than by any competing option. If the best results are produced by allowing agents to act on springs that are not directly utilitarian then they should be encouraged to do so. As John Austin the jurisprudent says: "Though he approves of love because it accords with his principle, he is far from maintaining that the general good ought to be the motive of the lover. It was never contended or concerned by a sound, orthodox utilitarian, that the love should kiss his mistress with an eye to the common weal." For this reason, any number of things - intentions, motives, virtues, whole social institutions, and for that matter acting out of a sense of mercy which directly maximise utility, are all permitted so long as they do in fact attain this end.

509

ANDREWBRIEN

Now, it should be clear that there is no necessary incompatibility between mercy and utilitarianism when the latter is considered a theory of tightness. Similarly, when utilitarianism is considered an account about how agents ought to deliberate there is no necessary incompatibility. On this view, agents deliberate over each and every decision; in other words, regarding utilitarianism as a decision procedure assumes act utilitarianism. 'Mercy', we have seen, is merely a name given to certain sorts of actions, and it has been shown to be compatible with act utilitarianism. Yet act utilitarianism, I have argued, faces a number of difficulties that render it unacceptable for use in institutional settings. The solution then is to consider utilitarianism as a theory of rightness. Given this, what is needed is a version of utilitarianism that retains the advantages of rule utilitarianism, such as the prohibition of victimisation and the savings made on decision costs, yet permits departures from the rules by institutional actors on appropriate occasions. It must furnish to institutional actors the capacity to calculate over individual cases when such calculation promotes utility. It must do this, however, without producing the decision costs that would attend determining if a particular occasion was the appropriate occasion and without producing the motivational redundancy that normally attends allowing departures from a rule. Motivational redundancy is the problem that is always thrown against any attempt to permit the departure from a rule that a utilitarian may adopt; namely, that such a strategy amounts to saying that one should follow the rule that has the best consequences unless it does not have the best consequences, in which case some other rule should be followed. Such a utilitarian would then find himself calculating over individual cases and the motivation for adopting the rule in the first place is lost. How might this version of utilitarianism be formulated? We recall that in order to be distinctively utilitarian a theory must honour the utilitarian credo that at all times public welfare be promoted, whether through individual action or through the existence and operation of an institution. The way the credo is put into practice is important only to the extent that the credo is in fact honoured and only in so far as that particular way is the best way to realise the

510

MERCY,IJIK.IrARIANISMANDRETRIBLrI'IVISM

credo. For this reason any strategy will be a variety of utilitarianism if it holds the basic good to be the promotion of public welfare. Given this, and the requirements that the formulation avoid the usual problems associated with utilitarianism that were discussed above, the most elegant formulation attains the utilitarian ideal by adopting and implementing two rules. (Perhaps this type of utilitarianism would be best called 'Two-rule utilitarianism'.) The first rule must determine the candidates for punishment and the second must determine the candidates to be exempted from the first rule. We can see how such an approach avoids the old chestnut of motivational redundancy. An agent is permitted to calculate over individual cases only if the escape rule is activated. This version o f utilitarianism is a type of 'restrictive utilitarianism.' This type of utilitarianism holds that while it may be appropriate to evaluate options in terms of them maximising value, it is not necessarily appropriate to select options on this basis. It may be more sensible to forswear or restrict direct evaluation by adopting a general policy that one adheres to, unless there are easily identifiable reasons not to. If this approach saves on decision costs, and prohibits actions that may prove to have a disutility attached, such as victimisation, and overall the system promotes public well-being, then it is to be preferred on utilitarian grounds. Restrictive utilitarianism, it should be clear, is justified for just these reasons. And they are reasons that a utilitarian can recognise whenever utilitarianism is considered a metaethical theory. Given this the rules could be formulated thus: R1. Agents who have been found by some established, public, and rational procedure to have performed actions of type A, and only such agents, should receive penalty P. The justification for such a rule is clear: a society is a happier place if victimisation is prohibited and the process of discovering wrongdoers is public, rational orderly and workable. The mere possibility of victimisation, as well as conducting in secret the process of determining who has been a wrongdoer, is sufficient to undermine confidence in the institutions of a society, which exist after all only for the public benefit. Moreover, notions of desert figure prominently in popular morality. Any policy that seeks to go outside the popular morality

511

ANDREWBRII~

and ignore the received opinion that no punishment should exceed the gravity of the offence, will lack popular support. Consequently, any rule that imposes burdens upon wrongdoers must incorporate to some extent the notion of desert and the limitations that this may place upon the sorts and severity of actions that are considered appropriate. Nevertheless, the penalty for A-type actions will be determined through research not only into what it is thought a person deserves for such an offence, but into the seriousness of the offence and the deterrent effect required. For these reasons, the subjects of a system administered by a rule such as R1 receive a double and self-validating guarantee. Not only do the administrators promise not to victimise the innocent or inflict upon wrongdoers burdens in excess of their desert, but since the process of determining who is a wrongdoer is public, the subjects can see that the promise is being honoured. Victimisation that promotes public welfare only succeeds if the innocence of the victim is concealed and the people can be made to believe that this person ought to be made to suffer. In the sort of system imagined here such concealment is impossible and the utility that is attached to victimisation dissolves, since no community, apart from perhaps a community of sadists, relishes victimisation. Now, a legislator would also know that, even if she is adept at framing laws, it is impossible to specify in advance all the vagaries of circumstance that arise in human affairs and which may dilute the guilt of the accused. She would also know that she cannot anticipate the various temptations which may urge a person to the commission of an offence, and that the utility of a law may change over time. Knowing that complete specification is well nigh impossible, the legislator would then adopt another rule, in effect an 'escape' rule, that would grant institutional actors the discretion to depart from the rigid application of the first rule. Such an escape rule may take this form: R2. In every case in which it has been established that a person has performed an action of type A and so falls under rule 1, and where there is good reason that can be found without great decision costs, to believe that keeping rule 1 will not maximise utility but breaking it will, then rule 1 ought to be broken and some degree o f individual calculation ought to be undertaken.

512

MERCY,LrlK/rARIANISMANDRE1RIBUnVISM

The second policy is in effect an escape policy. It permits a departure from the initial policy only when there were good and obvious reasons to do so and when this can be known without great decision costs. Such a 'policy-escape policy' combination pre-empts an agent's calculating over individual actions, while prohibiting moral opportunism. All an agent has to do in order to act rightly, is ascertain that some case falls under a particular policy and that the escape condition is not satisfied. If, on the other hand, the escape condition is satisfied, then it will be right to depart from the initial policy and, depending on the case, mercy (i.e. more lenient treatment than would be licensed by RI) may be in order. Now, when an agent decides that the escape condition has been satisfied, the decision between the competing act-options is resolved by a direct appeal to utility. So, in answer to the question, 'When ought I to be merciful?' an agent would reply, 'When it best promotes utility.' Therefore, an agent may act mercifully for utilitarian reasons. This relies upon an obvious distinction often overlooked in the contemporary d e b a t e - the distinction between an action and its justification. It should be clear that the motivations for adopting rule utilitarianism would be maintained, namely the saving on decision costs (since not every offender will activate the escape rule) and the prohibition of victimisation. Yet, institutional actors would be justified in not inflicting the penalty licensed by rule 1 when such a departure would promote utility, without the attendant decision costs. Of course, unless the utilitarian adopted the escape rule, mercy would not be justifiable or possible within the context of punishment for a rule utilitarian. Moreover, this strategy also neutralises the charge that the rule utilitarian is committed to 'irrational rule worship' in those cases where there is a great utility attached to not following the rule. In the case of punishment, it would allow the merciful remission of some standard penalty that a particular crime usually attracts, when the authority who has the power to break the rule is persuaded of the special nature of the case; i.e. when she comes to know that the initial rule will not in this case promote utility.

513

~

BRIEN

This points clearly to an important virtue of two-rule utilitarianism: it may be applied neatly to existing social structures. For example, in many jurisdictions the law specifies a maximum penalty that must be applied to the most serious of cases that fall under that law. The point of a specific law and the court proceedings concerned with ascertaining guilt and which generally surround a law's application, may be seen in terms of RI: it is a mechanism for determining who has performed actions of a specific type, actions that have a social disutility attached to them, and it grounds a rational, measured and orderly response to such actions. The barrister's plea in mitigation, on the other hand, could be seen as an attempt to 'trip' the escape clause and show why a particular case that obviously falls under R1 should not be treated to the full extent required by that rule. It might be argued that this can not count as a version of rule utilitarianism since the rule utilitarian necessarily must be committed to one rule only. This is wrong. There is nothing in the concept of rule utilitarianism that requires adherence to one rule only. Rule utilitarianism holds only that whatever actions are performed accord with s o m e rule. And it should be clear, whether R1 or R2 is appropriate on some occasion, a rule is still being obeyed. Further, especially within the area of public policy, in order to be a utilitarian of any sort, all one need be committed to, as we saw, is the view that whatever actions are performed or institutions or strategies adopted must promote general welfare. The way this desideratum is realised is not fixed and it is open to a person committed to the utilitarian ideal to judge not only actions, but rules, and combinations of rules or motives or intentions and so on. Moreover, it is clear that general welfare is promoted not by one single thing but through a complex interaction of individual actions, the existence of institutions, coercive mechanisms and so on. The problem for the utilitarian is two fold: first a co-ordination problem, not only in terms of individual actions and individual goals, but in terms of the tension that arises from the instability that individual actions produce and the stability that institutions promote. Second, an epistemological problem of determining what, on any occasion, the best bet is likely to be. It seems likely that overall the best option is a

514

MERCY,UTILITARIANISMANDREIRIBUTIVISM

society that allows not only individual actions, but also requires, in general, adherence to specific rules and institutions that operate in open and knowable ways. Given this, and the fact that single rule utilitarianism is clearly inadequate and that act utilitarianism is impossible to implement, the only type of utilitarianism that will allow the flexibility required while maintaining the advantages of rule utilitarianism is the two-rule variety discussed here. Another criticism is that R2 would not be the only other rule. Why could there not be yet another 'escape' rule, say R3, licensing a departure from R2 - and so on? The answer is clear: R2 justifies a departure from a general rule that governs the ordinary, run-of-themill cases, to specific calculation of individual cases. What further shift possibly could there be? The rules R1 and R2 guide the actions of agents so that when taken together the result has the best chance of delivering a result that is likely to maximise utility. There is, simply, no need for an R3 and the overall utilitarian goal of the system makes any more than two rules redundant. One result of two-rule utilitarianism seems to be that whenever an institutional agent performs an act of mercy she is obliged to do so. This leads directly to the two conundrums mentioned above: How then can mercy be supererogatory and can two-rule utilitarianism accommodate supererogatory actions suchas mercy is thought to be? A supererogatory action is one that a person has no obligation to perform and to whom no blame attaches if she fails to perform it. When then does a person have an obligation to perform an action? The standard response for a utilitarian would seems to be, 'whenever that action will, on balance, maximise utility'. Nevertheless, it may be the case that obliging a person to perform an action that would maximise utility may on occasion be counter-productive. The person may regard the obligation as too onerous and so may fail to comply. A higher rate of compliance, as well as the performance of actions that go some ways to promoting utility, may be achieved by not obliging that the action be performed. It seems then that whether a person has an obligation to perform an action not only depends upon the action maximising utility but also upon whether obliging that action itself leads to more of the desired actions being performed. On this view, supererogatory actions are those actions which would maximise

515

ANDREWBRIt~ 84

utility if it were not counter-productive to oblige them to be performed. In the case of two-rule utilitarianism a supererogatory action would be one where an institutional actor indulges in a certain amount of individual calculation when not required or beyond that which is required by institutional rules, so long as the effort expended does not adversely affect the overall utility of the system. So it seems that the common intuition that supererogatory action that are also acts of mercy can be accommodated by two-rule utilitarianism. What then of the conundrum created by the competing intuitions that mercy is both supererogatory, yet also appears on occasion to be obligatory? The answer is, of course, that sometimes an act of mercy may be supererogatory and at other times it is obligatory. Its deontic status depends upon whether it promotes utility to oblige the action or keep it optional. This rests upon a simple fact often overlooked. An act of mercy does not have, essentially, any particular deontic status. This much was clear from the analysis given above, which showed that it was possible to identify an action as an act of mercy without reference to it being obliged or optional. In other words, mercy is not essentially obligatory or optional and its deontic status is quite independent of those features that identify an action as an act of mercy. Hence the conundrum dissolves. The point of all this is that two-rule utilitarianism can accommodate merciful actions when they are supererogatory or when they are required. The question now is: does using the escape rule to remit a penalty that is justified by a different rule count as an act of mercy? Given the account of mercy already developed, it is clearly right to characterise such remissions as 'acts of mercy'. The offender was under the threat of some penalty that would have been applied had the escape clause not been activated. The intervention of the judge (who has the power not to impose the penalty should the escape condition be realised) in the usual course of legal events prevents the threat from being carried through. Even though the judge was on utilitarian grounds required t o remit the penalty, the harsher option was nevertheless a real option. It could logically have been selected, and without the application of the escape rule, which justified the judge's departure from the initial rule, it would have had to be selected. The harsher option did not occur

516

MERCY,UTILITARIANISMAND REIRIBUTIVISM

because of the judge's decision to realise the beneficial act-option. And that, we recall, is what mercy involves. An analogous situation occurs when, as it is claimed, God decides not to punish a wrongdoer. He must, if His action is to be a moral action, base His decision on good reasons; it cannot be grounded in mere whim, fancy or caprice, but must include, amongst other things, considerations of desert and reasons of compassion. Each case is taken on its merits, and though one may predict that God will always act mercifully, the harsher option is logically available, even if never chosen. God chooses the merciful act-option over the harsher alternative - every time. It may be thought that the utilitarian's action fails to be an act o f mercy because it lacks the spontaneous goodwill or compassion often associated with mercy, whereas God's action has these springs, and it is this that identifies God's action as merciful. This is wrong. This argument rests upon the assumption that the only motivations for an act of mercy can be goodwill and compassion. While they may motivate some acts of mercy, we have seen that they need not motivate all putative acts of mercy, in order for an action to be correctly called 'an act of mercy'. An agent may act mercifully when she encounters a person who is ground down by poverty or starvation or illness, and be motivated to do so because she believes that people should not suffer such indignities, or that such things are wrong; or she may be motivated to remit a debt out of self-interest, as was Draco. Compassion and goodwill play no motivating role in such cases. They are not essential to an a c t being correctly d e s c r i b e d as an act of mercy, since an action can, as we saw, be specified as an a c t o f mercy quite independently of its springs. Such an objection confuses the display of the virtue of mercy (where spontaneous goodwill, compassion, care and concern f o r another's welfare are essential elements of the virtue of mercy) with what it is for an action to be an act of mercy. Draco does not display the virtue of mercy when he relinquishes his claim over the tenant farmer, but he does a c t mercifully. Draco chooses for selfish reasons an act-option that benefits the tenant farmer rather than one that harms him. In the same way, it is right to say that a rule utilitarian has acted mercifully when, because the case satisfies an escape rule, she

517

ANDRE-WBRIEN

breaks some initial rule and selects a more lenient option and thereby treats an offender less harshly than in the circumstances would ordinarily occur. Acting mercifully for the utilitarian is a matter of choosing, on the grounds of utility, the merciful act-option over harsher options that are open to her. So, in answer to the question, 'When ought I to be merciful?', the utilitarian does not respond, as Smart suggests she would: 'I ought to do such and such, but special considerations persuade me to act differently on this occasion'. Rather, the utilitarian would say 'I could do such and such but utilitarian considerations persuade me that morally I ought to be merciful...' It is clear from all this that utilitarianism, of a sort that legislators would, I believe, need to favour, can accommodate mercy. Ironically, there is hope for the retributivist in this solution. Suppose that a retributivist is faced with various options, differing in severity, for punishing some offender. She may even be tempted to impose a harsh penalty. But after assessing the desert of the offender she decides a particular lenient penalty is morally appropriate, rather than one of the harsher penalties. This would count as an act of mercy for the reasons advanced above. The retributivist selected what was, from her point of view, the morally appropriate act-option and this happened to be the more lenient one. So, if it is right to describe as an act of mercy some action performed by a utilitarian, that was justified by its utility, but which happened to be one of the more lenient actoptions open to him, then it would, for analogous reasons, be appropriate to describe a similar action performed by the retributivist, only in this case justified by an appeal to retributivist considerations, as an act of mercy too. Aren't such actions what the utilitarian or the retributivist ought to be doing anyway, in which case such actions cannot really count as acts of mercy? This objection confuses the justification for an act with what it is for that action to he the action that it is. Mercy is, as I argued above, a certain sort of act-option, and in utilitarian or retributive contexts its selection happens to he justified in particular ways. What two-rule utilitarianism does is provide a basis upon which an agent may justify the selection of a merciful option over a harsher option. The objection overlooks the fact that an action may

518

MI/RCY,UTIL/TARIANISMAND RE'IRIBUrlVISM

have two descriptions. In virtue of being justified in a particular way, some action may be retributive or utilitarian in nature, and in virtue of being the more lenient option of a set of act-options that are available to an agent, it may also be merciful.

V. Conclusion We have seen how both the utilitarian and retributivist can accommodate mercy. Therefore, the widespread belief that it was only retributivism that could accommodate mercy was shown to be wrong. Why has it been a persistent belief that utilitarianism cannot accommodate mercy? It stems from the fact that mercy has often been seen to involve a departure from justice - or a departure from some standard that is justified. As we saw, this view of mercy is wrong. Mercy, it was suggested in section II, is a particular sort of act-option. It is one that addresses the dire needs of a person under a substantial threat or burden. It is not a matter of doing less than one is justified (all things considered) in doing, but rather, it involves performing a beneficial action that brings about a departure from some burdensome but possible and likely course of events; in the case of punishment, imposing less of a burden than it is possible and there is good reason to impose. On this view, an agent is justified in choosing the merciful act-option, rather than some other harsher act-option, when it best promotes utility. In doing so, the utilitarian departs from those other act-options that are open to her, but which fail to promote utility to the same extent. A plea for mercy made to a utilitarian authority would involve pointing out the special circumstances that surround a case, and which make mercy the act-option that best promotes utility. As is clear by now, the mistake that many modem writers make is to confuse the act of mercy with its justification. As soon as the two are separated the problem dissolves. A few final points. It might be argued that if this account o f mercy is true, then whenever the utilitarian selects an option she w i l l be acting mercifully. Therefore, the argument presented here proves too much. The argument runs like this. Utilitarians are always faced with a range of options that differ in severity. They are always constrained to select that option that on balance promotes welfare the most. Since there will always be a more severe option for someone

519

ANDREWBRIEN

whose welfare forms part of the overall calculation, in doing what she must do, the utilitarian will always be acting mercifully for someone whose welfare forms part of the overall calculation. This argument ignores two important points. The first is that an act of mercy alleviates some dire situation. It seems to be an essential element in any act of mercy that the situation alleviated was grave. I t is not an act of mercy on my part, for example, to satisfy some mild preference of yours (Chopin rather than Ravel), that if left unsatisfied would merely have produced some small discomfort. Unless a person was in a dire situation that was alleviated, then no act of mercy has occurred. The second point is that acts of mercy have particular targets. They are aimed at something, the dire need that this identifiable person has. The argument suggested that because someone somewhere was better off, this could count as an act of mercy. It can't, simply because such an action is impersonal; it is not aimed at a person in a dire situation. So, even though someone somewhere may be better off, when a utilitarian selects the option she does it will only be an act of mercy if the person who is at least partly the object of the action, has had a dire burden alleviated and it was intended that this burden be alleviated. It is true, however, that this account of mercy fails to show how the utilitarian can be justified in acting mercifully when such an action clearly fails to maximise utility and it normally would be obligatory. For the utilitarian such actions are certainly unjustified. The sort of case I have in mind might run like this. Suppose that Dr Smith decides to go to a leper colony and minister to its needs. Clearly, this can be characterised as an act of mercy. In doing so, however, he forsakes a brilliant medical career. As a result important discoveries of far greater benefit to the wider community than the benefit derived from ministering to the leper colony, are not made. Has Dr Smith in acting mercifully, acted wrongly? Intuitions run in both directions and that question cannot be answered here. But even if we believe, for reasons of utility, that Dr Smith did act wrongly, i t does not show that the utilitarian cannot accommodate all acts of mercy. Clearly those act of mercy which maximise utility would still be justified. Rather all this example shows is that utilitarianism, from a public policy perspective, can accommodate

520

MERCY,LrI'ILITARIANISMANDREIRIBUTIVISM

only an impoverished range of merciful actions, namely those actions that maximise utility and which it is useful to oblige or permit. A 11 sub-maximal merciful actions (that are not in the permissible category because requiting them would be counter-productive) would be unjustified. There is another, perhaps more intractable, problem that faces both the retributivist and the utilitarian - though not in the area of public policy but only within private morality. When we reflect upon the grounds for the high valuation traditionally accorded mercy, we can see that it is based upon the belief that it is a virtue and that it has intrinsic worth, rather than because it promotes utility or accords with retributive justice. These intrinsic reasons are often justification enough to perform a merciful action, since the mercy-giver's power is such as to render her immune to any ingratitude that the beneficiary may show. Merciful actions add to the mercy-giver's arete as a human being and as a member of society, though this would never be a reason for a mercy-giver herself to display the virtue of mercy. Now, an agent may be moved by the virtue of mercy to perform a merciful action. This action may itself be capable of justification on a utilitarian or retributivist ethic and from these points of view such an action would then be right. However, from the point of view of received moral opinion such tightness is only accidental since conventional morality does not ground the moral standing of mercy in either solely retributive or utilitarian considerations. For this reason, it would seem that the disposition to be merciful has value quite independently of its utility or its accord with retributive justice. What this suggests is that whilst the utilitarian or retributivist can perform merciful actions, she is unable, in any rich sense, to account for the moral value thought to be possessed by mercy, its praiseworthiness and moral goodness, which is also reflected in our ordinary moral perceptions; that is, the very property of some merciful actions that elevates them from the class of the ordinary and mundane and identifies them as expressions of the best part of the human spirit, MASSEY UNIVERSITY PALMERSTON NORTH NEW ZEALAND

521