Department of Modern History, Politics and International Relations - Staff

Utilitarianism and Virtue

Utilitarianism and Virtue*

(Published in Ethics93 (1983), pp. 451-66.)

John Kilcullen

A line of thought suggested by certain passages in Mill's writings runs as follows.[Note 1] Virtue should be regarded as an end in itself outranking even happiness, because virtue so regarded guarantees certain modes of feeling and conduct, and the benefits resulting from this guarantee make up for what is lost in the odd cases in which virtue and happiness conflict. Notice that benefits result from the guarantee, not only from the conduct guaranteed. In this paper I will explore this theory in comparison with certain other versions of utilitarianism. I will illustrate my argument by reference to Mill, who, I believe, held substantially this theory; but exegesis is not my present purpose.

I. The utility of commitment

Virtue may guarantee fidelity to moral rules, or concern for ends, or both; it may be impossible or inexpedient to reduce concern for an end to rule. But for simplicity's sake I will set ends aside for the present and consider virtue as a guarantee (not an absolute guarantee) of conformity with rules. Other guarantees are possible, but for the present I will set these aside also. By virtue, then, I mean in part a settled disposition of character which causes a man or a woman to act according to certain rules.

There are many possible codes of rules. For each rule, or variant of a rule, there is another set of possible codes, consisting of all the other possible codes with this rule or variant included. Every possible priority rule or other device to resolve conflict between rules also gives rise to a new set of possible codes. Among these many possibilities three kinds are worth distinguishing: ( 1) codes which consist of some version of the utility principle alone or with "summary" rules, without other ("secondary") rules;[Note 2] (2) those which include a set of secondary rules comprehensive enough to decide every particular case arising in every department of conduct; (3) those which include some secondary rules but not an absolutely comprehensive set. The third category can be subdivided according to the treatment of cases not covered by secondary rules: (3.1) they may all be decided by direct application of the utility principle; or (3.2) some of them may be decided by the utility principle, the rest being left unregulated for the person concerned to act as he or she pleases; or (3.3) they may all be left unregulated. For a world like ours the theory I am exploring will probably favor 3.2 or 3.3.

An important feature of the present theory is the connection it makes between rules and guarantees. One possibility is to have no guarantee; the person will "adopt" or "acknowledge" or "subscribe to" some code, perhaps the utility principle alone, but there will be no lasting cause ensuring that it will be acted on throughout the future. Another possibility is to have one single virtue, general benevolence we might call it, guaranteeing conformity with the utility principle. Another is to have a complex of particular virtues, such as truthfulness and fidelity to undertakings, guaranteeing conformity with a correspondingly complex code.

Now imagine a utilitarian assessing the possible combinations of codes and guarantees to choose the one which from his or her point of view is best, that is, the one which maximizes general happiness. The following points would need to be considered. (a) Among the consequences of commitment to a code are the acts and omissions the code requires. They may be beneficial in most cases; but if the code includes secondary rules, there may be cases in which the code forbids the act with the best consequences or requires some other act instead, and then there will be some loss. (b) There are the costs of establishing the guarantee---of developing the virtue if one does not have it already, of preserving it, of manifesting it (since to gain benefits under d below the guarantee must be manifest). (c) Virtue may make some actions more satisfying to the agent, thereby increasing their good consequences,[Note 3] and other actions less satisfying. But in addition---and this is distinctive of the present theory---there are consequences which follow not from the acts which virtue guarantees, but from the fact that they are guaranteed. These are of two kinds: (d) if other people know they can rely on someone to act in certain ways, then they may omit costly acts of conflict or precaution, and they will be readier to cooperate with him and readier to act beneficially in ways which might have seemed too risky if he had been less predictable, and a sense of security, trust, and friendship are themselves good;[Note 4] (e) someone who can rely on his own future feelings and conduct can trust himself in things which lead into temptation, or require persistence. But (f) some of the consequences of reliability may be bad: one may trust oneself too much, or others may, or they may take advantage of one's predictability to do something harmful. Finally, (g) there are certain advantages in leaving people free to act as they please unregulated by moral principles, so for every code there is a cost arising from the loss of some liberty.

Which combination of code and guarantee is best? Consider the code consisting of the utility principle alone combined with no guarantee. The costs under a and b of the above headings are nil, but so are the advantages under other headings. Take d: if someone says he intends to act according to the utility principle alone, even assuming that we can believe that this intention is real, can we be sure it will be acted upon throughout the future? Only if we believe it to be rooted in some lasting disposition (we may already know him to be benevolent); but this is not the combination with no guarantee. So consider, second, the combination consisting of the utility principle alone guaranteed by general benevolence. The cost under b is higher, but there are more advantages under d. Someone committed to the utility principle is to some extent predictable; he will not be selfish, and if his beliefs about his situation and the likely effects of various actions are known it may be possible to predict what he will do. But reliable information about his beliefs may be difficult to get, since the utility principle may not in every case prescribe truthfulness and may even prescribe some deceptive stratagem. Consider, third, a code with secondary rules of truthtelling and keeping faith guaranteed by the corresponding virtues. There will be costs under a as well as b,f, and g, but there will be increased advantages under c, d, and e. In comparison with this, other combinations including more secondary rules would have the advantage of reducing communication costs (there would be less need for information about beliefs and for undertakings), but otherwise the advantages under d might not increase much while the costs under a, b, and g would.

Which combination is best depends upon the person's circumstances, or, I will say, on his or her "world"---that is, not the circumstances of some particular act but the context of a whole life. Utilitarians will not spend too much time in working out which combination of code and guarantee is best for their respective worlds since returns on this activity soon diminish rapidly. As I remarked earlier, for a world like ours I expect they would conclude in favour of a code of type 3.2 or 3.3,[Note 5] guaranteed by some particular virtues, notably truthfulness and fidelity to undertakings. If this is the conclusion, then utilitarians will do the actions which create, preserve, and manifest the appropriate virtues; they may have them in some measure already from upbringing, since a similar line of argument would justify the inculcation of virtue by a utilitarian parent or teacher.

I have imagined the utilitarian assessing not actions but possible combinations of code and guarantee. This assessment can be translated into terms of action; it is a decision whether to do the actions which cause virtue. The consequences will include other actions some of which might not satisfy the utility principle if assessed singly (see heading a above). Commitment to a code is created and shown in other ways besides acting in obedience to it, and some acts done in obedience to it do not show commitment because they are not seen. So when the benefits of the guarantee (mainly through effects upon confidence---heading d) are credited to the various acts from which they result, the appearance of loss in some cases of obedience may not be removed; even when effects on confidence are taken into account, some acts of obedience to the rules which virtue guarantees may be really less good in their proper consequences than other acts which could have been done instead, and therefore would not satisfy the utility principle if assessed singly. Yet the acts which cause virtue may satisfy the utility principle even when the losses from resultant acts are taken into account. Developing virtue is in this respect like an investment, in which present opportunities, and some future opportunities also, are passed up for the sake of a better total outcome. Now I am assuming that utilitarians can, without abandoning utilitarianism, approve investments and roundabout methods; that they can assess the actions making up an integrated course of action not singly but together, by the total outcome, allowing gains from some to offset (real) losses on others; that is, that they can withhold judgment from some acts, regarding them only as causes or consequences of others, even though in some sense they are independent particular acts.[Note 6] On this assumption the theory I am exploring is utilitarian.

Notice that the argument does not depend on the suggestion that a single virtuous act is useless or less useful unless it forms part of a practice, or unless other such acts will be done; "threshold effects" are not essentially involved.[Note 7] The argument is not that there should be a practice, but that someone, perhaps a single individual, should be believed to have a disposition. Once he or she has done and said enough to create that belief, the beneficial effects may follow even if the situation never again arises for the disposition to issue in action (as when the disposition is a deterrent), and even if no one else develops the same disposition (as when someone in a special position, such as a king, cultivates a disposition especially appropriate to that position, such as respect for the liberties of the subject).

II. Virtue or hypocrisy?

The benefits which compensate for the occasional sacrifice of happiness to virtue depend mainly upon other people's beliefs. Secret virtue would seldom satisfy the argument. Is it in fact an argument for a Machiavellian policy of seeming virtuous without being so? To seem to be committed to a code of secondary rules but always actually to do the act with the best consequences will be the best utilitarian policy, when it is possible. In some circumstances it may be possible, and then the argument supports hypocrisy. But if other people are shrewd judges of character, hypocrisy may not be enough to produce the beliefs from which the main compensating advantages result. To win the trust of experienced people it may be necessary to manifest a state of character which is: difficult to counterfeit; enduring; not much affected by calculations of the advantage of violating the code, or by the presence or absence of observers; and a strong enough influence to overcome contrary motives, at least usually. (Add these points to what "virtue" means in the present theory.) These conditions exclude the possibility of seeming to be committed to the code but actually doing the best act. If others are reasonably shrewd, to achieve the good consequences of being believed to be committed it may be necessary to be genuinely committed, so that one will actually obey the code even in some cases in which a forbidden act would have better results.

This should not be confused with the common act-utilitarian treatment of confidence. An act-utilitarian will follow a rule even when some forbidden act would have better consequences if this is necessary to preserve or foster other people's confidence in him, and if the benefits from their confidence will offset the difference between the forbidden act and the best act allowed by the rule; and then, of course, the forbidden act would not really have better consequences---its consequences only seem better when effects on confidence are overlooked. But if it is known or suspected that someone will break a rule in secret, or even openly when the advantages are great or the need f6r confidence small, then confidence in him must already be weakened, even if he has never actually broken the rule. To secure confidence it may be necessary to be genuinely committed by some enduring character trait not much affected by secrecy or by calculations of advantage. The argument, then, is not that it might be expedient to obey the rules when people are watching, but that it might be expedient to preserve or develop a state of character which guarantees that we will obey the rules even when no one is watching. Veracity is one of the first virtues likely to be justified by this line of argument, and as soon as it is, a Machiavellian pretense of other virtues is ruled out. But not altogether; self-deception and the subtler forms of hypocrisy remain a risk. To be conscious of the reactions of fallible observers must always be to some extent corrupting. If the only observer were God, friendly but infallible, there would be an incentive to searching self- criticism and genuine virtue. But even for religious people God is not the only observer who counts: they are concerned with witness, example, scandal; they make friends, allay fears, reassure the diffident, and engage in many other activities which involve winning trust. Every morality, no matter what its basis, encourages or allows such things. As long as it matters to any extent and for any reason what other people think, or what one's fallible self thinks, then hypocrisy and self-deception are a danger: consider George Eliot's Bulstrode in Middlemarch.

Veracity guarantees the genuineness of other virtues. There are also "secondary" or "auxiliary" virtues which protect the genuineness of virtue, some of which have no convenient name. There is frankness: a disposition to reveal one's true mind even without being directly questioned. Another is a certain independence of the opinions of others: a determination to satisfy oneself first, then those who share the same ideals, and then others. Another is a disposition to self-criticism, so that the concern to satisfy one's own judgment does not degenerate into self-deceiving narcissism. Unfortunately the secondary virtues are also corruptible. Frankness may be selective and consciously intended to win trust; selective self-criticism may be a diversionary tactic of self-deception; an attitude of indifference to the opinions of others may be assumed in the knowledge that it will make a good impression. None of the defenses is impregnable, although together they may reduce the risk. A shrewd judge of character will know to look for the auxiliary virtues and will know that even a substantially virtuous person may do a little faking. To be completely genuine the commitment must be to virtue for its own sake,[Note 8] and not for the sake of effects upon observers. Mill claims that a utilitarian can and should recognize the utility of desiring virtue disinterestedly as an end in itself. But the argument I am exploring, borrowed from Mill, paradoxically recommends disinterested commitment for its effects. This calls for a sharp distinction between justification and motive: the utilitarian needs to be motivated by a desire for virtue as an end in itself, while believing that this desire is justified by its consequences. Even if this is psychologically possible, as I believe it is, the argument will not justify completely genuine virtue. The more one's associates know about possible forms of deviousness, the more virtue will have to be genuine to win their trust. It might be possible to find less knowing associates, or to go in for things which do not need close contact or much trust; the argument justifies genuine virtue only when it is worthwhile to associate closely with shrewd judges of character, and since they remain fallible the justification even then remains incomplete. But those who hold theories which purport to justify completely genuine virtue do not themselves actually have it, because they too, for various reasons, care what fallible observers think. There is not much difference between a utilitarian who can never quite justify genuine virtue and a Stoic who can never quite attain it.

III. Virtue and other guarantees

Earlier I set aside concern for ends, and also other possible guarantees besides virtue; it is time to take these into account and to put the argument into a more general context. There is an obvious likeness between this theory and Hobbes's, which recommends not virtue but some political institution. My general term for what such theories are about is "guarantee," meaning some state of affairs from which it can be inferred that one will probably not do certain things (for example, because it is out of one's power, or difficult or costly); other examples are bonds, deposits, and the penalties which sanction a contract-anything which ties one's hands or closes off some options. Anyone who enters a legally enforceable contract, for example, voluntarily subjects himself to penalties if he acts in certain ways that might otherwise have served his purposes, and he does so because he thinks that benefits likely from the guarantee will make up in the end for the disadvantages of constraint. Setting up a guarantee belongs to a wider class of actions, those which form part of integrated courses of action.[Note 9] Now let us generalize Hobbes's argument. Universal selfishness is not an essential premise and a political institution need not be the conclusion. The war of every man against every man could arise among fanatical idealists if they disagreed (or merely thought some of them might think they might disagree) about their ideals or about the implementation of a common ideal. Unfanatical idealists might come into mild conflict or be less willing to cooperate, and all conflict and noncooperation results in evil or loss of good. So if any sort of practical disagreement is (or might be thought to be) possible, it may serve one's purposes (whatever they are) to offer some sort of guarantee. Some guarantees, though not all, provide sanctions or incentives to make some options less or more attractive to the calculating decision maker. Sanctions may be political, popular, religious, or moral, to follow Bentham's classification; they may attach to standing rules, or they may be attached ad hoc to particular actions by means of such devices as contracts, promises, and oaths. Hobbes's argument shows the utility of political sanctions for standing rules, Hume's argument about promises shows the utility of attaching the popular sanction to particular acts. The usefulness of the religious sanction is presupposed by the practice of law courts, in which jurors and witnesses take oaths; it has the advantage over political and popular sanctions that it checks even secret violations and violations by persons too powerful to fear the law or public opinion; God is mighty and knows even secret acts and thoughts. The moral sanction has similar advantages and acts also on people without religious beliefs.

The moral sanction is perhaps not a sanction at all. What the term suggests is that a person may be deterred from wrongdoing by thinking of the pangs of guilt to follow, as he might stay away from the dentist for fear of the pain. This seems unrealistic; he would feel guilty because he believes that the act is wrong, and this belief influences him now apart from anticipations of guilt. Similarly with the treatment of the pleasures of doing right as an incentive: if a person takes pleasure in doing what is right it is because he believes it to be right, and this belief may be enough to explain the action. Sanctions and incentives are brought into play by some other person, who can withhold them when it is expedient. Moral condemnation by one's own conscience is not like this; it would be strange to say, "1 see I have done wrong, but is it expedient to inflict moral sanctions on myself?"

Virtue, at all events, is not a sanction but a guarantee of another sort. A sanction operates by entering into calculation: if I do this thing, I will suffer penalties; is it worth it? Virtue preempts calculation. The virtuous person may know quite well how the calculation goes (in fact, I will argue later that virtue should not blind a person to consequences); he may know, for example, that the pangs of guilt would not qua pain outweigh the advantages of the act. But the decision does not depend upon calculation; he does not seriously consider violating the rules which virtue guarantees. The guarantee consists in being the sort of person who coolly disregards the advantages of wrongdoing. Similarly with secondary ends:[Note 10] someone committed to such an end does not calculate the pleasure or other results of achieving it, as if it were merely a means to those benefits, but treats it as an end in itself. In comparison with sanctions (including the moral sanction, if there is such a thing) virtue has the advantage of being relatively independent in any particular case of the state of one's feelings at the time. It is the empirical counterpart of Kant's Good Will, a settled resolve to do what is right because it is right independently of the incentives even of moral feeling:

A person of confirmed virtue, or any other person whose purposes are fixed, carries out his purposes without any thought of the pleasure he has in contemplating them, or expects to derive from their fulfilment; and persists in acting on them, even though these pleasures are much diminished by changes in his character or decay of his passive sensibilities,...[I]t is because of the importance to others of being able to rely absolutely on one's feelings and conduct, and to oneself of being able to rely, on one's own, that the will to do right ought to be cultivated into this habitual independence.'[Note 11]

(" Absolutely" is an exaggeration; every guarantee costs something, so the optimal guarantee will be less than absolute.)

Sanctions and incentives, once they exist, fit the scheme of act-by- act calculation,[Note 12] but virtue does not: its utility is as part of an integrated course of action. In some cases the sacrifice of happiness to virtue (under heading a) may be only apparent, since the pleasures of virtue and the anticipated pains of the moral sanction (heading c) may be enough to make the virtuous act best even if assessed singly. But if virtue is independent of pleasure and pain there may be cases in which this is not true; and yet, because of advantages under d and e, there may be a utilitarian justification for virtue, assuming that utilitarians may judge an integrated course of action by its total outcome. If they may not, then neither can they set up and maintain a system of sanctions and incentives: once such a system exists, act-by-act calculation may justify conformity, but to set it up needs acts which compete with other acts which must generally be better, if acts setting up the system must be judged singly and cannot be credited with the improved consequences of decisions made under its influence. The same reasoning applies to guarantees of any sort. So if utilitarians can approve the establishment of any sort of sanctions or guarantees, they can approve the cultivation of virtue. Now if virtue is a fixed purpose independent of pleasure and pain, we need not accept the description utilitarians sometimes give of the conscientious person who usefully shrinks from the pangs of guilt, recoils in horror, refuses to look, feels guilty about breaking the rules even when rationally convinced that an exception is justified---a poor anxious creature no self- respecting man or woman would wish to be. Virtue is a calm state of mind, a quiet purposiveness, a settled willingness to act on rules or toward ends critically accepted, not troubled by even the clearest perception of the advantages of an occasional deviation; in which there is nothing irrational, assuming compensating advantages from the guarantee such a disposition gives.

To return to the utilitarian assessment of possible combinations of codes and guarantees. What I said earlier was oversimplified. The utilitarian will consider, for some parts of the code or for all of it, guarantees other than virtue; perhaps in some worlds the best combination will not include virtue at all, though in our world I believe it will. I will use "moral guarantee" to refer to the moral sanction or virtue or both, and "morality" to refer to those rules and ends ("ideals") which are backed by the moral guarantee, or to those together with the guarantee. Now it may be best for guarantees to reinforce one another at some points---for example, for those who administer the political sanction to be held to their task by morality and public opinion, and for public opinion on their performance to appeal to morality. It might sometimes be best to guarantee important rules in every way, so that violations are illegal, unpopular, irreligious, and immoral. Some theologians and philosophers once laboured to attach the sanctions of morality and religion to every act of government and to persuade governments to sanction all the rules of religion and morality. But for this there is no utilitarian justification in our world, or, I believe, in theirs: motorists who exceed the speed limit need not believe they have done something immoral, though morality and other guarantees may be needed to keep the traffic police and courts to their proper task. If morality, public opinion and the law together guarantee certain basic rules, then other rules can be left to be guaranteed by morality alone, or by the law alone;[Note 13] and no rule need be guaranteed absolutely.

In a world like ours I would expect the best combination to include certain virtues. Some will orient a person toward secondary ends the pursuit of which cannot usefully be reduced to rule, or not completely;[Note 14] for example, frankness and generosity. Others will guarantee basic rules; for example, veracity, fidelity to undertakings, justice. Other rules will be left to law or public opinion. There will also be a measure of liberty. The best code is not likely to include enough secondary rules to decide every case. The remaining cases might be decided by direct application of the utility principle, but at least in some cases it might be better to leave people at liberty to do as they please. And it might be best if they guarantee to leave one another at liberty. So I expect that the best code for our world will include secondary rules protecting an area of liberty, and in particular a moral rule against condemning an act as morally wrong unless it violates some secondary rule of morality.[Note 15] The argument for such a rule is that, although in some cases something will be lost by not using morality to exact the best possible act, these losses are more than made up for by the results of guaranteeing a measure of liberty. How much liberty this rule protects depends of course on the comprehensiveness of the code of secondary rules of morality; the utility of liberty is a reason for keeping the code in check (see heading g, Sec. I above).

IV. Morality and expediency

I will say that an act is "morally wrong" if it deserves moral condemnation, "morally permissible" if it does not, and "a moral duty" if its omission would be morally wrong; these definitions seem close to ordinary usage, but they can be taken as stipulations. I will also stipulate (departing from ordinary usage) that something is "inexpedient" if another possible act would contribute more to the general happiness, and "expedient" otherwise. Now suppose that in a certain world the best combination of code and guarantees includes ( I) secondary rules which yield in some cases other decisions than the utility principle would if applied directly; (2) a secondary moral rule to the effect that only violation of a secondary moral rule warrants moral condemnation; (3) a rule making moral considerations practically conclusive, at least to the extent that what is morally wrong is simply not to be done no matter what reasons there are for doing it. Then there may be cases in which (a) it is a moral duty to do something inexpedient, or (b) something expedient is wrong and not to be done,[Note 16] or (c) it is permissible to do---and wrong for anyone to condemn morally---something inexpedient, even without the excuse that duty requires it.

So a gap opens between expediency and morality. The science of expediency (the "Art of Life," as Mill calls it) is architectonic, but moral rules have overriding authority in the particular case. That is, in choosing a code and its guarantees we are considering a question of expediency---the conclusion is not that we have a moral duty to guarantee a certain code, but that it is expedient to do so; but if the best code includes 1-3 above, then conclusions a-c follow, and in the particular case expediency gives way to morality. It is expedient that it should:[Note 17] the argument for 1-3 is that what is lost in the odd case when morality requires or allows the inexpedient is made up for by the benefits which result from guaranteeing some duties and some liberty. The judgment of expediency in the particular case, even though it is overridden, remains true: as I argued at the end of Section I, the act which morality requires may be really less beneficial in its proper effects than some other act that could be done instead. But to guarantee to obey moral rules which sometimes require really inexpedient acts, and to guarantee moral liberty for some such acts, may be expedient because of the guarantee's other effects.

Similar remarks apply to secondary ends. If the best combination of code and guarantees includes virtues which guarantee not obedience to rules but concern for ends, then some moral judgments will be expressed in terms not of right and wrong but of goodness or praiseworthiness,[Note 18] or of the specific virtue involved (generosity, etc.). A morally permissible[Note 19] act is morally good (praiseworthy, generous, etc.) insofar as it furthers an end supported by virtue. Just as secondary rules sometimes require an inexpedient act, so an act may be morally good as furthering an end other than the general happiness, with respect to which it may be inexpedient. So there is a gap not only between expediency and moral rightness but between expediency and moral goodness.[Note 20] The" Art of Life" recognizes the expediency of orientation to secondary ends, despite the fact that this will sometimes lead to inexpedient acts.

But all of this depends on suppositions of fact. In some worlds the best combination might not include secondary rules and ends and particular virtues; it might consist of the utility principle alone, guaranteed by universal benevolence. Then morality and expediency would coincide: the morally permissible act, the morally good act, and the expedient act would be identical.

V. Individuality, intelligence, and discretion

The argument I have been pursuing assumes that acts which make up an integrated course of action are to be judged not singly but by the total outcome; and I believe that for some worlds it will justify a course of action including some acts inexpedient in isolation (namely, some acts which develop virtue and some virtuous acts), losses on these being offset by the other consequences of virtue. If this is a version of rule-utilitarianism there are some obvious differences from other versions.[Note 21] It does not begin by postulating that particular cases are to be judged by secondary rules, and it is not motivated by a desire to approximate conventional morality; the use of secondary rules is a conclusion argued for on grounds of utility. It does not require every particular case to be decided by secondary rules; it may be expedient to leave some to be judged by the utility principle direct, or to free choice. Several other differences are summed up in the words of my heading.

First individuality: it is not necessary, on this theory, for everyone to adopt the same code, or to adopt the code everyone would adopt if they were rational and informed utilitarians or in some hypothetical situation. Everyone should offer the guarantees appropriate to his or her own actual and individual situation. The argument is that it is expedient for A to guarantee to others that he will obey a certain code, because then B and others will feel secure, refrain from conflict, cooperate, undertake beneficial projects of their own, and so forth, and the benefits will outweigh what it costs for A to provide the guarantee. Nothing in this implies that others should offer similar guarantees; the argument would have to be reiterated for B, C, etc., successively, and it might not be true of each of them that the benefits outweigh the cost of the guarantee. If there are specialized roles, some rules and some virtues (or other guarantees) may be more important for some than for others. If there is a diversity of moral traditions then the cost for some people to offer a certain guarantee (e.g., by some modification of character) may be higher than for others. For some rules part of the benefit hoped for from their adoption may be that others will adopt them too. Such rules might be adopted by contract, but they need not be: they might be adopted unilaterally by some who wished not to be objects of precaution, in the hope that others would follow suit for the same reason. In general the theory does not envisage unanimity.[Note 22] Mutual confidence requires knowledge of what rules each person can be relied on to keep, not that all keep the same rules, and confidence may be of value even if it is not mutual.

Second, intelligence. The theory does not require blindness to the undesirable consequences of the current code, or forbid revision. The morality one was brought up on is the starting point, but as circumstances change and consequences are reassessed the code should also change. Virtue preempts calculation in the sense that particular decisions do not depend upon calculation, but it need not prevent calculation. If it is indeed true that the benefits resulting from the guarantee compensate for what is lost in the odd case then the virtuous person will coolly disregard the loss, but if it ceases to be true, or never was true, nothing should prevent him from becoming aware of it. Now dispositions and other guarantees cannot be simply turned off or redirected at will, and anything that could be cannot serve as a guarantee. But in time, as experience presses in, perhaps with effort, and not without some outward signs of change, dispositions and other guarantees can be modified. Outward signs of change are important: we can rely on people whose dispositions we know may change provided we believe we would notice the change in good time. Truthfulness and fidelity to undertakings are important also, as basic parts of a flexible character: if we believe that someone will not deceive us about his moral code and will meet our legitimate expectations or compensate us if he cannot, then we can rely on him even though he regards his whole code as revisable, provided we believe that he is not likely to revise the rules of truthfulness and fidelity suddenly and drastically. Guarantees cannot and need not be absolute and eternal.

Third, discretion. On this theory individuals may make exceptions or on-the-spot revisions in especially hard cases which they could not reasonably be thought to have foreseen in adopting the code. The code will never be perfect, and unforeseen cases may arise in which the loss from obedience to the code may be extreme. The possibilities are: to revise the rule on the spot without prior notice; to obey it now but announce a change for the future; and to make an exception for this case.[Note 23] Virtue (fidelity to implicit undertakings) should guarantee that the first and third possibilities are not chosen too readily, because of effects on confidence; but in extreme cases the threatened damage may be too serious to obey the rule now and change it later, and then virtue should allow revision or exception without any residual feeling of guilt. Predictability in unpredictable cases is not necessary or useful. The optimal level of confidence is less than absolute; to guarantee never to make an exception might lead to excessively harmful acts in some cases, and the difficulty of foreseeing such cases would make people excessively reluctant to commit themselves to generally useful rules.

Rule-utilitarian schemes sometimes postulate the adoption of rules the universalisation of which would have the best consequences, or the rules which would be agreed upon by an intelligent community of utilitarians without moral preconceptions (except for commitment to the utility principle), or the rules as much like the code of the ideal community as the actual community could be persuaded to accept.[Note 24] The theory I have been exploring differs from schemes like these, I believe for the better. It envisages not merely the "formal adoption" or the "following" of rules but the cultivation of a character which is an effectual commitment; since the commitment has its roots in childhood impressions the rules must have some continuity with the customary morality, but they need not be identical with it; how it compares with the set of rules the universalisation of which would have the best consequences is irrelevant except insofar as one's adoption of it will actually lead to its adoption by others; the decision about rules is not made by or on behalf of the whole community but by an individual, who may or may not expect that others will make the same decision; the decision does not purport to be absolute and final but is subject to revision and exception.

VI. Conclusion

If the line of thought explored in this paper is sound, there is an argument in terms of its consequences for virtue: that is, for being the sort of man or woman who can be relied on to act in certain ways even when one can see advantages, even advantages to some ideal cause, in acting otherwise. This argument may be of some importance at a time when lying and cheating are believed to be commonplace in business, education, and government, and when in private life infidelity even for merely selfish reasons is often taken lightly.

Although the argument is available to utilitarians and other consequentialists, it is not specifically consequentialist: it is available also to those who deny (as I do myself) that every action is to be valued only for its effects, as long as they acknowledge that some actions are to be valued at least in part for their effects and that those may include actions which develop and preserve useful dispositions.

Links 2005

Kilcullen, "Mill on Duty and Liberty", Australasian Journal of Philosophy, Vol. 59, No.3; September 1981. Argues for an interpretation of "Mill's principle" and evaluates his arguments for that principle.
---------, "Reciprocity arguments for Toleration", from Sincerity and Truth: Essays on Arnauld, Bayle and Toleration (Oxford: Clarendon Press, 1988). An argument akin to that of "Utilitarianism and Virtue".
---------, Liberal Democracy. Discusses Mill's views on democracy.
---------, Free Enterprise and its Critics. Touches on Mill's views on political economy.

Recent literature on Utilitarianism

John Gray, "John Stuart Mill: Traditional and Revisionist Interpretations"
Sinnott-Armstrong, Walter, "Consequentialism", The Stanford Encyclopedia of Philosophy (Summer 2003 Edition), Edward N. Zalta (ed.)
Hooker, Brad, "Rule Consequentialism", The Stanford Encyclopedia of Philosophy (Spring 2004 Edition), Edward N. Zalta (ed.)
Lawrence M. Hinman, Ethics updates: Literature on Utilitarian Ethics

Return to Politics, Philosophy and Medieval Studies


* I wrote a version of this paper while a Visiting Fellow at the Australian National University. I am grateful for comments made on earlier versions by S. I. Benn, J. I. Kleinig, G. W. Mortimore, and J. C. Smart, and by referees.

Note 1. Cf. J. S. Mill, Utilitarianism, chap. 4, in J. S. Mill, Collected Works, ed. J. M. Robson et al. (Toronto: University of Toronto Press, 1963-), vol. 10, pp. 235, 239, and Logic, 6.12.7, in Collected Works, vol. 8, p. 952. In this as in other points of ethics and political theory J. S. Mill seems to have been influenced by Sir James Mackintosh; see his Miscellaneous Works (London: Longman, 1846), vol. 1, pp. 194-95, 202-3, 246-47, 364-65, and vol. 3, p. 101.

Note 2. By a secondary rule I mean any rule other than the utility principle and summary rules. A summary rule is one which gives the same directions in each case it covers as the utility principle would, or near enough for losses usually to be made up for by economies of calculation; see J. Rawls, "Two Concepts of Rules," Philosophical Review 64 (1955): 3-32, pp. 19-24. Whenever I refer to the utility principle understand "or summary rules."

Note 3. See R. M. Adams, "Motive Utilitarianism," Journal of Philosophy 73 (1976): 467- 81. The thesis of this article, in my terms, is that advantages under c may make up for losses under a; since these are advantages to the agent, the "motive" theory is available to egoists as much as to utilitarians.

Note 4. See Mill, Logic, 6.11.6 (1843 ed.), Collected Works, vol. 8, p. 1154; J. S. Mill to G. Grote, January 10, 1862, in Collected Works, vol. 15, p. 762; On Liberty, chap. 4, in Collected Works, vol. 18, p. 277.

Note 5. Mill favored a code with a limited set of secondary rules; see Logic, 6.11.6 (1843 ed.), in Collected Works, vol. 8, pp. 1154-55.

Note 6. See Adams, p. 473.

Note 7. See D. Lyons, Forms and Limits of Utilitarianism (Oxford: Clarendon Press, 1965), p. 72. Assuming that some actions may count only as consequences, the theory I am exploring satisfies Mill's remark that "the right way of testing actions by their consequences is to test them by the natural consequences of the particular action, and not by those which would follow if everyone did the same" (J. S. Mill to John Venn, April 14, 1872, in Collected Works, vol. 17, p. 1881). The "general consequences" test was a familiar idea by Mill's day (e.g., G. Berkeley, "Passive Obedience," in Works ed. A. A. Luce and T. E. Jessop [London: Nelson, 1953], vol. 6, pp. 17-46; W. Paley, Works [London: Rivington, 1830], vol. 3, p. 53; J. Austin, Lectures on Jurisprudence [London: Murray, 1879], vol. I, p. 111), and Mill himself sometimes uses it (Utilitarianism, chap. 2, par. 19, in Collected Works, vol. 10, p. 220, "Whewell," in Collected Works, vol. 10, pp. 181-82). But on the present theory it is irrelevant unless my action will cause everyone to do the same, or unless I am committed to a rule against actions that would not meet this test; such a rule might have value as a guarantee that I would not give myself secret privileges, but it would have heavy costs under heading a.

Note 8. Sir James Mackintosh noted the utility of auxiliary virtues ("useful habits which, being securities to Virtue, become themselves virtues") and the need to cultivate virtue for its own sake (Mackintosh, vol. 1, pp. 246-47, 195).

Note 9. Other members of this class include "bridge-burning" strategies, which also include acts inexpedient (or "irrational") considered singly. See T. C. Schelling, The Strategy ofConflict (Cambridge, Mass.: Harvard University Press, 1960), pp. 17-19.

Note 10. By a "secondary" end I mean anything other than the general happiness which is sought not merely as a means. The "primary" end is the general happiness, concern for which is guaranteed by the virtue of benevolence. On secondary ends see J .S. Mill, "Bentham; in Collected Works, vol. 10, pp. 95-96, 110-11, Utilitarianism, chap. 4, in Collected Works, vol. 10, p. 236, Logic, 6.12.7, in Collected Works, vol. 8, p. 952, and Autobiography, ed. J. Stillinger (London: Oxford University Press, 1969), pp. 85-86. Virtue is a secondary end, as well as a guarantee of obedience to rules and of concern for certain other ends; one of the things a virtuous person aims at for its own sake is virtue---to be more perfect and to persevere.

Note 11. Mill, Utilitarianism, chap. 4, in Collected Works, vol. 10, pp. 238-39.

Note 12. See R. E. Sartorius, Individual Conduct and Social Norm5 (Encino, Calif.: Dickenson Publishing Co. , 1975), esp. pp. 62-68. This and other act-utilitarian treatments of sanctions do not seem to consider the justification of the acts needed to set up sanctioning institutions.

Note 13. Notice that, on this account, that the act be morally wrong is not a necessary condition for the imposition of other sanctions. Legal condemnation does not presuppose moral condemnation.

Note 14. On duties of imperfect obligation, see Mill, Utilitarianism, chap. 5, in Collected Works, vol. 10, p. 247; and Immanuel Kant, The Metaphysical Principles of Virtue, trans. J. Ellington (New York: Bobbs-Merrill Co., 1964), pp. 48-49.

Note 15. As I interpret it, Mill's principle of liberty forbids penalties except for violation of some secondary rule-some "definite" or "assignable" duty, not some general duty to do as much good as possible. Mill does not specify the code of duties, assuming the substantial utility of the current code, but argues that "what are called" duties to self should be excluded, leaving only duties to others. See my paper, "Mill on Duty and Liberty," Australian journal of Philosophy 59 (1981): 290-300.

Note 16. According to D. Lyons, Mill values and prefers the more beneficent above the less beneficent act but does not hold that the latter must be wrong; see his "Mill's Theory of Morality," Nous 10 (1976): 101-20, pp. 103-4. Similarly L. W. Sumner outlines a theory according to which when good and right conflict we should prefer the best; in such a case it will be justifiable, though not a duty, to do what is wrong; see his "The Good and the Right," Canadian Journal of Philosophy 5, suppl. (1979): 99-114, p. 112. On my account the consideration that the act or omission would be morally wrong is practically conclusive (that is, conclusive for practice): if the act with best consequences is morally wrong it is not (morally) preferable and is not to be done; and the less beneficial act may be a moral duty, omission of which is wrong.

Note 17. Lyons asks ("Mill's Theory of Morality," p. 119): "If happiness is really Mill's ultimate end, how better to express it but by refusing to subordinate it to any conflicting values?" My answer is that the total happiness may be greater if we do not try to maximize the production of happiness in each single act but obey certain rules and seek certain ends in a principled way.

Note 18. On positive moral worth and moral praise of actions beyond duty, see J. S. Mill, Auguste Comte and Positivism, in Collected Works, vol. 10, pp. 337-39. On the relationships between "right and wrong," "good and evil," notice that between the members of each pair there is middle ground. Often "right" connotes duty ("I must do what is right..."); but there are morally permissible acts (not wrong) which are not duties. Similarly a morally permissible act may not be positively good or praiseworthy and yet not be evil.

Note 19. By 3, the judgment that an act is wrong is practically conclusive: it is not to be done no matter what reasons there are for it-even if it furthers some ideal end. It would seem inappropriate to call a morally impermissible act good.

Note 20. On this point my theory differs from that which L. W. Sumner presents, which treats goodness as equivalent to what I call expediency, conduciveness to the general happiness; see Sumner, pp. 111-14.

Note 21. On the points which follow, see D. Braybrooke's illuminating article, "The Choice between Utilitarianisms," American Philosophical Quarterly 4 (1967): 28-38; Braybrooke shows how a single basic utilitarianism becomes act- or rule-utilitarianism when applied in different worlds. Act-utilitarians might decide to individuate acts by criteria according to which what I call an integrated course of actions counts as a single act, in an ad hoc attempt to save the appearance of act-by-act assessment for all possible worlds. This would not show that rule-utilitarianism is just act-utilitarianism after all but reduce both to the original undifferentiated utilitarianism. But to me the distinction seems worth preserving.

Note 22. One of the reviewers for this journal draws attention to the advantages of committing oneself to the rules to which one's neighbours are committed and remarks that I exaggerate the utility of personal variation. I do not mean that variation itself has utility (though perhaps it has), merely that in some situations the commitments of individuals who calculate the utilities correctly will be various. Their calculations should take account of costs of communication (see Sec. I above), which are reduced by consensus. A person's "actual and individual situation" includes the commitments others have made or would make in response to his.

Note 23. According to Mill, the rules of justice are "more absolute" than any other rules, but even to these there are exceptions; Mill, Utilitarianism, chap. 5, in Collected Works, vol. 10, pp. 255, 259. Compare his comment on Berkeley's "rule-utilitarianism": ". ..The writer was misled by an exaggerated application of that cardinal doctrine of morality, the importance of general rules. As it was acknowledged that the cases in which it is right to disobey the laws or rebel against the Government are not the rule but the exception, Berkeley threw them out altogether, for his moral rules admitted of no exceptions" Dissertations and Discussions [London; Parker, 1875], vol. 4, pp. 183-84). Notice that it is not "essential ...that the exception should be itself a general rule" (Mill, "Whewell," in Collected Works, vol. 10, p. 183). True enough, when an exception is justified it will be justified in all relevantly similar cases, but there may be no utilitarian justification for including a rule to cover such cases in the code guaranteed by virtue; the guaranteed code need not purport to cover every particular case.

Note 24. See R. Brandt, "Toward a Credible Form of Utilitarianism," in Morality and the Language of Conduct, ed. H. Castaneda and G. Nakhnikian (Detroit: Wayne State University Press, 1963).