Universal Prescriptivism


R. M. Hare


Originally published in Peter Singer, A Companion to Ethics (Blackwell Publishers, 1991).



Universal Prescriptivism is best seen as an attempt to locate both the faults and the true insights in other current ethical theories, to remedy the faults while preserving the insights, and so to provide a synthesis between them. The expression 'ethical theory' covers attempts to say what we are asking when we ask moral questions. What do we mean by the words or the sentences that we use in moral discourse; what is the nature of the moral concepts or of morality? If successful, these attempts will have implications for another, epistemological, question which also belongs to ethical theory: how should we set about answering our moral questions rationally? Or can there be no rational way - is it just a matter of how we feel or what the current mores dictate? On the other hand, if there can be rational discussion of moral questions, does it demand that there be a truth about them, or a set of facts, that can be discovered?


The main division is between descriptivist and non-descriptivist theories. These are distinguished in various more or less misleading ways (Hare, 1985b). It is said that descriptivists hold that moral judgements can be true or false, while non-descriptivists deny this. But since there is a perfectly good sense, as we shall see, in which non-descriptivists can use the term 'true' of moral judgements, this way of speaking obscures the issue. So does the use of the terms 'cognitivism' and 'non-cognitivism', implying that the former does, and the latter does not, allow that we can know that some moral judgements are true. For again, there is a perfectly good sense in which non-descriptivists can allow this, as we shall see. Equally misleading is the oncological way of putting the distinction, by saying that descriptivists claim that there are moral qualities or facts existing in the world, whereas non-descriptivists deny this; for as soon as we start asking what it is for a moral quality or fact to exist in the world, we get lost.


Both non-descriptivism in all its varieties, and the descriptivist theories to be discussed, are semantical, not ontological theories. So-called ontological theses in ethics (e.g. 'ontological naturalism') may be substantial moral claims about what is or is not right, etc. (e.g. that what maximizes happiness is always, or perhaps even necessarily, right). The question of what they mean, which is our present topic, is a different one. On the futility of ontological disputes in ethics, see Hare, 1985b; . . .


We can avoid these difficulties if we put the distinction in terms of a theory of meaning which has been very popular: the truth-condition theory. This is not the same as the old 'verification theory' advocated by some logical positivists; but it [451] shares some of its insights. According to this theory, to understand the meaning of a sentence, as used to make a statement, is to understand the truth-conditions of the statement, i.e. what has to be the case for it to be called true. Those who hold that this is true of all sentences may be called descriptivists tout court.


Descriptivism of this sweeping sort is obviously false - Austin even called it 'the descriptive fallacy' (Austin, 1961. p. 234; 1962, p. 3). For there are certainly sentences and utterances whose meaning is not determined by truth-conditions.  Imperatives are the obvious example: in order to know what the request 'Shut the door' means, we do not have to know, and cannot know, its truth-conditions, because it does not have any. But maybe one can safely be a descriptivist with regard to large classes of sentences. It can perhaps be said that sentences expressing typical ordinary statements of fact do have their meaning determined by the truth-conditions of the statements they express, so perhaps we can happily be descriptivists with regard to such sentences. But since not all sentences are like this, as we have seen, the question arises, for any given class of sentences, whether their meaning is wholly determined by truth-conditions or not - i.e., whether they can be classified as purely descriptive.


The word 'purely' is important. It is possible for part, but not the whole, of the meaning of a sentence to be determined by truth-conditions. We may call such 'mixed' sentences 'descriptive' in a weak sense, but not in the strong sense here being used. In the strong sense a sentence is not descriptive (i.e. not purely descriptive) unless its meaning is wholly determined by truth-conditions. Ethical descriptivism is the view that this is true of sentences expressing moral judgements.  Ethical non-descriptivists, including prescriptivists, can readily admit that there is an element in the meaning of moral judgements (the descriptive meaning) which is determined by truth-conditions; but they differ from descriptivists in thinking that there is a further element in their meaning, the prescriptive or evaluative, or in earlier writers the emotive, which is not so determined, but expresses prescriptions or evaluations or attitudes which we assent to without being constrained by truth-conditions.


Kant was saying the same thing in other words when he spoke of the autonomy of the will: 'the property the will has of being a law to itself (independently of any property belonging to the objects of volition)' (Kant, 1785, BA87=440). To adopt an attitude, evaluation or prescription is a function of the autonomous will, constrained only, in Kant's words, 'by the fitness of its maxims for its own making of universal law' (Kant, 1785, BA88=441 - see below).


Having distinguished descriptivist from non-descriptivist ethical theories in general, we can now proceed to subdivide each of these, in order to put prescriptivism in its proper slot. Descriptivist theories can be divided broadly into naturalism and intuitionism. Both terms can be misleading, but they will serve.  The dispute between these is about whether or not the truth-conditions or moral judgements, which according to descriptivism give them their meaning, are to be determined by definitions (or. more loosely, explanations of meaning) which refer only to non-moral truths or properties. Naturalists think that this is possible; intuitionists, by contrast, think that no such definitions or explanations can [452] capture the meanings of the moral words. Note that the dispute between naturalists and their intuitionist opponents is not the same as that between descriptivists and non-descriptivists: it is a dispute within descriptivism. Non-descriptivists reject naturalism because they reject descriptivism of all kinds. They can use such of the intuitionist arguments against naturalism as are valid; but the main impetus of their attack is independent of these.


The main impetus comes from the recognition that both forms of descriptivism, in different ways, are destined to collapse into relativism. Though relativism has its supporters (among them, we are often told, most of the callower among American students - though this is an exaggeration), it is certainly not what most descriptivists are trying to establish. They set out with the aim of showing, rather, that there can be rational moral enquiry, yielding conclusions to which we must all in reason assent. It is the fact that both kinds of descriptivism signally fail, and are bound to fail, to show this that condemns them even by their own standards.


Naturalism collapses into relativism in the following way. If the meanings of moral words are explained in terms of truth-conditions, then what will ultimately determine the truth or falsity of moral judgements will be the particular truth-conditions accepted in a given society as determining the meanings of the moral words. So (to take an over-crude example) if we explain the meaning of 'ought' in the sentence 'Wives ought to obey their husbands in all things' by saying that the statement it expresses is true if and only if some conjunction of non-moral statements is true, we shall then have to specify this set of non-moral statements.  Perhaps we shall say that the statement is true if and only if wives obeying their husbands would contribute to the stability of society. Now it may be that in a given society it is generally accepted that one ought to do what would contribute to the stability of society. That, perhaps, is one of the moral principles that the society believes in. But this is a principle which could be rejected by feminists. They might think that although wives obeying their husbands would indeed contribute to the stability of society, they ought not always to obey, because husbands sometimes make demands that wives ought to disobey even at the cost of impairing stability.


What is happening in this example is that a substantial moral principle, that one ought to do what would contribute to the stability of society, has got promoted into an analytic truth, true in virtue of the meaning of 'ought'. But it is not an analytic truth. If it were, then the anti-feminists would win the argument, because the feminists, in contending that wives sometimes ought not to obey although this would destabilize society, would be contradicting themselves, by saying something which the very meaning of 'ought' establishes as false. If 'ought' has its meaning fixed by truth-conditions, and if the truth-conditions are those accepted by native speakers of the language, and if the native speakers of this language (apart from a very few deviants) use 'ought' in such a way that what destabilizes society must be something that one ought not to do, then one cannot in logical consistency say what the feminists are saying. We can see from this over-simple example that the effect of naturalism is to force everybody to embrace the accepted mores on pain of self-contradiction; and this is relativism. For a fuller [453]account see Hare, 1985a, 1986; . . .


The other variety of descriptivism, intuitionism, collapses into relativism in an even simpler way. Intuitionism is the view that the truth-conditions of moral judgements, which give them their meaning, consist in conformity with the data on which we have to base our moral reasoning, and with which its conclusions have to square; and these data are the common moral convictions that all morally educated people have. Since these convictions will vary from one society to another, the effect of intuitionism is, again, to anchor our moral reasoning to something relative to particular societies. True, there are convictions which are common to most societies: but there are others which are not, and no way is given by intuitionists of telling which are the authoritative data. To revert to our example, if it is a universal conviction in a society that wives ought to obey their husbands, then the feminists' case is ruled out of court; but if there were a society, as there may come to be, in which it was a universal conviction that wives have no such duty, then the anti-feminists' case will similarly be ruled out of court. Relativism is again the result.


One further kind of descriptivism may be mentioned here, namely subjectivism. This term is used very loosely, but here we shall be using it strictly for that kind of naturalistic descriptivism which holds that the meaning of 'ought' and other moral words is to describe the attitudes or feelings of people - for example to attribute to people in general, or to the speaker of the sentence, an attitude or feeling of approval or disapproval towards a certain kind of act. This is intended to be understood as a statement of non-moral, psychological fact about the speaker or about people in general - a fact which could be discovered by observation or reported introspection. Thus the theory is a naturalistic one. But the fact in question (which establishes the truth of a moral judgement, and is thus its truth-condition) instead of being, as in our previous example, one about what would happen in society if wives disobeyed their husbands, is a subjective fact about what people disapprove of. (James Rachels uses the term simple subjectivism to refer to this theory; . . .)


The words 'objective' and 'subjective' have a clear use here; we are distinguishing between two kinds of facts: facts about what will actually happen in society if wives disobey, and facts about what people think. The former are objective facts, the latter subjective. To define the moral words in terms of subjective facts would be one kind of naturalistic descriptivism. It is helpful, however, to notice that there is no important distinction between this kind of subjectivism and intuitionism; for intuitions and moral convictions are also subjective facts – facts about what people think; and therefore many of the objections that are commonly accepted against subjectivism apply equally to intuitionism. That explains why intuitionism collapses into relativism; intuitionists are appealing for support to nothing objective, but only to their own and other people's thoughts, and these will vary from one person and society to another.


When used outside this context, the words 'objective' and 'subjective' can be a source of confusion (Hare, 1976). In particular, if various kinds of non-[454]descriptivism are said to be 'subjectivist', this can only be in a quite different sense.  They do not make moral judgements equivalent to statements of psychological fact, because they do not make them equivalent to any kind of factual statement.  True, they agree with subjectivists in rejecting both objectivistic naturalism, and the objectivist claims of intuitionists; but that is all they have in common.  Subjectivism, in the sense used here, and prescriptivism (or for that matter emotivism) fall on opposite sides of the main division of ethical theories into descriptivist and non-descriptivist theories (see above). It therefore risks serious confusion to use the term 'subjectivist' to apply to both.


Non-descriptivism likewise can be subdivided. The earlier versions, mostly forms of emotivism, were essentially irrationalist. Having rejected the view that moral judgements are equivalent to statements of non-moral facts (naturalism) and the view that they are sui generis statements about moral facts discernible by intuition or appeal to convictions (intuitionism), they concluded too hastily that one cannot reason about moral questions: moral judgements are the expressions of irrational or at least non-rational attitudes of approval or disapproval. They concluded this because they added an additional premise which is false, namely that the only questions one can reason about are factual ones. A reading of Kant with his 'Practical Reason' (1785, BA101=448), or even of Aristotle with his 'phronesis' or 'practical wisdom' (which he says is epitactic or prescriptive - Nicomachean Ethics, 1143a8) should have cured them of this mistake.


The genesis of prescriptivism lay in the realization that this premise i's false. It has been generally recognized that there is something unsatisfactory about emotivism. Those who are beguiled by the false premise just mentioned have reacted by reverting to some form of descriptivism. Prescriptivists. on the other hand, have reacted in a more positive way by looking for a kind of non-descriptivism which would not be open to the charge of irrationality - a charge which most moral philosophers want to escape. They claim to have found it by showing that there are rules of reasoning which govern non-descriptive as well as descriptive speech acts.


The standard example is again imperatives: if (as is certainly the case) there can be logical inconsistency between contradictory prescriptions, someone who wants the totality of the imperatives, or in general prescriptions, that he (or she) accepts to be self-consistent will have to observe the rules which govern consistency. What these rules are should therefore be the main concern of the practical philosopher.


Some emotivists became irrationalists because they both assimilated moral judgements to imperatives and made a mistake about imperatives which is still too common, namely to think that they get their meaning from their causal properties. This may be termed the 'verbal shove' theory of the meaning of imperatives. It involves a failure to distinguish between what Austin calls perlocutionary acts (what one is doing by saying things) and illocutionary acts (what one is doing in saying them) (Austin, 1962). Verbal shoves and psychological prods are not part of the meaning of either imperatives or moral speech acts (Urmson, 1968, p. 130 ff.; Hare, 1971, s.f.).


[455] If one thought that imperatives owed their meaning to their causal properties, one might easily become an irrationalist about imperatives and thus about moral judgements, if these were prescriptive. But if this mistake is avoided, the thesis that moral judgements are a kind of prescriptions (perhaps not identical with the simple imperative kind, and certainly more complex in their logic) is consistent with there being reasoning-rules that govern moral thinking.


Prescriptivists have looked for such rules. They claim to have found them in a combination of the rules that govern ordinary simple imperatives, and a further set of rules that govern 'ought' and other deontic modal words like 'must' in its moral sense, which stands to imperatives in much the same relation as modal indicatives to non-modal ones (Hare, 1981, p. 23). These words are not peculiar to moral discourse, so it may be that further rules are needed for moral discourse in particular. But that can be left open for the present (see Hare, 1981, p. 52ff.).


In any case some differentia is needed to distinguish the rules common to all prescriptions, which govern imperatives as well as 'ought'-statements, from those which are peculiar to 'ought'-statements; otherwise we shall be left making no logical distinction between two classes of speech acts which are clearly different in their meaning, and therefore in their logic.


The most-discussed kind of prescriptivism, known as universal prescriptivism, finds this differentia in what has been called the universalizability of 'ought'-sentences and other normative or evaluative sentences. Most descriptivists too acknowledge this feature of moral judgements. One cannot with logical consistency, where a and b are two individuals, say that a ought, in a certain situation specified in universal terms without reference to individuals, to act in a certain way, also specified in universal terms, but that b ought not to act in a similarly specified way in a similarly specified situation. This is because in any 'ought'-statement there is implicit a principle which says that the statement applies to all precisely similar situations. This means that if I say 'That is what ought to be done; but there could be a situation exactly like this one in its non-moral properties, but in which the corresponding person, who was exactly like the person who ought to do it in this situation, ought not to do it', I contradict myself (Hare, 1963, p. l0ff.). This would become even clearer if I specified my reasons for saying why it ought to be done: 'It ought to be done because it was a promise, and there were now conflicting duties'.


Three warnings are necessary here to avoid confusions which have been too common. First, the 'situation' is to be taken as including the characteristics of the people in it, including their desires and motivations. If, therefore, the speaker says that a ought to do something to c, but that b ought not to do the same thing to d, because the desires of c and d are quite different, he is not offending against universalizability, because the different desires make the situations different. Bernard Shaw said 'Do not do unto others as you would they should do unto you. Their tastes may not be the same' (Shaw, 1903, p. 227); but this is not an objection to universalizability. If I ought to tickle one child's toes because it loves it, it does not follow that I ought to tickle another child's toes, however similar, if the second child hates it.


[456] Secondly, universality must not be confused with generality (Hare, 1972, p. 1ff.) The principle involved in an 'ought'-statement may be a highly specific, complex and detailed one, perhaps too complex for formulation in words. It does not have to be very general and simple. Complaints against universalizability, that it makes us the slaves of very simple general rules, therefore miss their target. To use an example which gave trouble to Kant: my moral principles do not have to be as general as 'Never tell lies'; they can be more specific, like 'Never tell lies except when it is necessary in order to save an innocent life, and except when . . ., and except when ...' (Kant, 1797). In a morally developed person the exceptions may get too complex to be formulated in words. But see below for the value, in our human situation, of general (i.e. not too specific) principles.


Thirdly, there can be universal relations as well as qualities (many-place as well as one-place predicates). Such is the relation mother of. The statement that everyone ought to look after his (or her) mother in her old age is therefore a universal statement, and the statement that a ought to look after his mother (but has no such duty to look after other people's mothers), is universalizable. The same can be said about the statement that I ought to keep my promises but not other people's. It is therefore no objection to the thesis of universalizability that there can be duties that one owes just to one person, provided that that person can be specified in universal qualititative or relational terms. It is no objection, even, that one can have the relation in question only to one person. 'Mother of’ is an example.


This point can be related to the previous one by citing a famous though much misused example given by Jean-Paul Sartre (1946, p. 40). During the Nazi occupation of France, a student came to seek Sartre's advice. The student's dilemma was whether to join the Free French forces to fight against the evil of

Nazism, or to stay with his widowed mother, who depended on him. Sartre uses the case to suggest that universal principles are useless in such situations, since each case is unique. Sartre himself seems to confuse universality with objectivity, though they are obviously different concepts.


But apart from that, his student did not have to find for himself any very simple, genera! principle. Perhaps he was the only person who had ever been in that particular complex situation. But he ought to have been able to form for himself a principle (a highly specific one) which he could accept for situations just like his. Perhaps other such situations would never occur. But one can conceive of hypothetical situations as like to his as one wishes; and he would be committed, on pain of self-contradiction, to admit that if they occurred, the same thing should be done in them. That our moral judgements have to extend to identical situations, hypothetical as well as actual, is an important tool in moral argument (Hare, 1981, p. 112ff.). But none of this implies that Sartre's pupil would be entitled to reproach or interfere with, unasked, someone else in the same situation who acted otherwise; he might think it an impertinence not to keep his thoughts to himself.


Universal prescriptivists hold, then, that 'ought'-judgements are prescriptive like plain imperatives, but differ from them in being universalizable. The task of explaining what prescriptivity is (the feature that 'ought'-statements share with [458] imperatives) can be attempted here only in the sketchiest way.  A speech act is prescriptive if to subscribe to it is to be committed, on pain of being accused of insincerity, to doing the action specified in the speech act, or, if it requires someone else to do it, to willing that he do it.


Prescriptivism thus falls within the class of ethical theories known as 'internalist': those which hold that to accept some moral judgement is eo ipso to be motivated in a certain way. This must not be confused with the view that for a moral judgement to be true is for someone to be motivated in some way; this would be a form of subjectivism in the sense used above. Internalist theories are contrasted with externalist theories, which hold that one can accept a moral judgement independently of one's motivations. So, for example, one may, according to externalists, say without contradiction or even pragmatic inconsistency 'I ought, but I have absolutely no inclination to'. By 'pragmatic inconsistency' is meant the logical fault that we all find in the statement 'He has been here already, but I don't believe it'.


Internalists and prescriptivists have been attacked on the ground that they make it impossible consistently to say 'You ought, but don't', or to think that one ought, but not be at all disposed to. The opposite side of this coin is that, as the 'but' shows, we all feel that there is something - even something conceptually and not just morally - wrong with people who say this sort of thing, which there would not be if externalism and descriptivism were correct. If someone were agonizing about what he ought to do, the agony would evaporate if he came to think the answer to his question quite irrelevant to his motivations or to what he actually did. This problem, sometimes called the problem of akrasia or weakness of will, is outside the scope of this article (see Hare, forthcoming).


We are now in a position to take up again the question of whether moral judgements can be called true or false. Imperatives clearly cannot. We can most easily sort out the problem by reverting to the notion of descriptive meaning explained earlier. Moral judgements have this because of their universalizability.  To make one is, as we have seen, implicitly to invoke some principle, however specific. In any moderately stable society, the principles that people accept and invoke in their moral judgements will be fairly uniform and constant. As a result, when one person says that somebody did what he ought in the circumstances, anybody who knows the circumstances and shares these commonly accepted moral principles will assume that, if he did what he ought, what he did was in accord with them; so he will think he knows what in particular the speaker was saying he did. If, then, it turned out that the person did not do that, he will say that the speaker was speaking falsely. And so he was, according to the commonly accepted descriptive meaning (i.e. truth-conditions) of the word 'ought' in that society.


This is not inconsistent with prescriptivism - which is why the 'true-or-false' characterization of descriptivism is so misleading. For prescriptivists too can give a limited role to truth-conditions in determining the meanings, in a given society, of moral words. If, as descriptivists think, the descriptive meanings of moral judgements, as so determined, were their entire meaning, then relativism would [459] be the consequence. For we can know the descriptive meaning of 'ought' in a given society; but it does not follow that it will have the same meaning in another society. A visitor from Saudi Arabia, hearing an Australian say that Tom's wife did what she ought, might assume that he meant that the wife obeyed. But the Australian might have meant that she disobeyed, because the subjection of women is not one of the principles accepted in Australia; Australians have another principle, that one ought to keep one's own end up.


So, although there is a perfectly good sense in which Saudi Arabians, speaking among themselves, can attribute truth or falsity to each others' moral judgements on the basis of the descriptive meanings of the moral words that they all accept, this attribution will break down when they are talking to Australians, It would break down even if they were speaking in Saudi Arabia to an Australian feminist missionary. The difficulty could be concealed by saying that both mean by 'ought' something rather vague, for example that to do it would make people happy. But if they try to establish the truth-conditions or descriptive meaning of this, they will disagree as before, and communication will again break down; for we may suppose that what counts as domestic happiness in Australia and Saudi Arabia is very different. The only way a descriptivist can get out of this difficulty is to say either (if he is a naturalist) that 'ought' and 'happy' have different meanings in the two places, and therefore different truth-conditions, or (if he is an intuitionist) that, because the convictions of people in the two places differ, there is eo ipso also a difference in how wives ought to behave; and both these ways of escape lead to relativism (Maclntyre, 1985; Hare, 1986).


The fact that moral judgements have descriptive meaning, and can therefore be said, within the limits just explained, to have truth-conditions and to be true or false, can be used to shed light on the much-disputed question of whether 'ought' can be derived from 'is' (moral judgements from non-moral facts). Given these features of moral judgements, it is easy to see why people should have thought that they can be derived from non-moral descriptive judgements, either deductively with the aid of a naturalistic definition, or by appeal to some substantial synthetic a priori moral principle grasped by intuition. For, as a Saudi Arabian might say, it is obvious that if a wife disobeys her husband (fact) she does what she ought not (moral judgement). Either this is true in virtue of the meanings of the words, given that disobedience by wives disrupts society (naturalism), or it is obvious anyway to those who have been properly morally educated (intuitionism).


This obviousness is reinforced if, as will be the case in stable societies, the moral education has instilled not just a certain use of language, nor just consistent behaviour in accord with the current mores, but deeply held convictions and feelings that such behaviour is required - that to depart from it goes against conscience. It is easy to see how people become naturalists or intutionists, and in both cases think that some non-moral factual premises make some 'ought'- conclusions inescapable. Prescriptivists have to deny this. because they hold that moral judgements commit the speaker to motivations and actions, but non-moral facts by themselves do not do this. The moral judgement, therefore, introduces a further element into the thought (the prescriptive or motivative element) which [460] is not there in the bare description of the facts. But the prescriptivist, if he is to win over the opposition, will have to provide an explanation, not merely of why people should think that moral judgements are inescapable given the -facts, but of how (by what rational process) we can arrive at a prescriptive moral judgement on the basis of the given facts. Is this possible for prescriptivists, who think that moral judgements are prescriptions, and so more than factual? It will not seem possible to those who think that only facts can be discovered by reason; but this is an error, as we have seen.


Kant, who understood that it is an error, gives some hints on how to proceed.  He says 'Act only on that maxim through which you can at the same time will that it should become a universal law', applicable, that is, whatever role you yourself occupy in the resulting situations (Kant, 1785, BA 82=421). If moral

judgements have the features of prescriptivity and universalizability that prescriptivists say they have, this method is imposed on us by the logic of the moral concepts. What maxims we can adopt, or what moral judgements we can accept, will then depend on what we are prepared to prescribe for all like situations (whether, for example, we were the unfaithful husband or his deserted wife).


'Imagining ourselves in others' positions' is a difficult operation which presents both practical and philosophical problems. The practical ones are evidence only that moral thought just is difficult: humans cannot do it at all well. The remedy for this incapacity we shall discuss below. The philosophical difficulties are too big a subject for this short article. They concern the problem of how to compare the strengths of the preferences of different people with each other and with our own preferences; the problem of 'other minds' to which philosophers have devoted so much attention; the problem of whether it makes sense to imagine myself being someone else (would it still be me?), and of what I am constrained to say about that person's situation when I do it (Hare, 1981, Ch. 7 and refs.).


A possible move for one who is looking for the necessary constraints on moral thinking is to say that unless I treat the person, in whose place I am imagining myself being, on equal terms with myself, showing him equal concern, I am not really imagining him as being me. This entails treating his preferences as of equal weight with my own present preferences, and thus forming preferences for the hypothetical situation in which I am he, equal in strength to those which he actually has.


This is what is involved in following the Golden Rule, doing to others as we wish others to do to us, and loving our neighbours as ourselves. It is also implicit in Bentham's maxim 'Everybody to count for one, nobody for more than one' (cited in Mill, 1861, Ch. 5 s.f.). The Kantian method we have been outlining is consistent with a form of utilitarianism (though not, we must add, exactly Bentham's form, because that is put in terms of pleasure, whereas Kant's theory is put in terms of will).


It is wrong to think, as many do, that Kantianism and utilitarianism have to be at odds. To treat a person 'never simply as a means but always at the same time as an end' requires, as Kant himself says on the next page, that 'the ends of a subject who is an end in himself must, if this conception is to have its full effect [46] in me, be also, as far as possible, my ends’ (1785, BA 69=430 f.). An end is what is willed for its own sake; so we are, according to Kant, to give equal respect to everybody's wills-for-ends, including our own; and this is what utilitarianism also binds us do. This involves, in a harmless sense, treating the ends of many people as if they were the ends of one person (myself). But this does not involve failing to 'take seriously the distinction between persons' (Rawls, 1971,pp. 27, 187)- a distinction of which Kant and the utilitarians are well aware.


Kant, though he may not have realized it, runs up here against the same difficulty that has often been alleged against utilitarianism, that it leads us to moral conclusions which seem counter-intuitive (e.g. that punishment of the innocent would be right, given assumptions which in practice would seldom be

satisfied). For this might, on such rare occasions, be what someone would do who was making other people's ends his own and so trying maximally to further them.  To punish an innocent man might avert some major disaster to almost everybody's ends. The solution is that actually suggested by many utilitarians, which involves a division of moral thinking into two levels - an idea that reminds us of the distinction made by Plato (Meno, 98b) and Aristotle (Nicomachean Ethics, especially Book VI) between right opinion or desire and practical understanding or wisdom (phronēsis). 'Critical' and 'intuitive' thinking are convenient names for these two levels. If we were perfect moral thinkers we might use the Kantian-utilitarian method, i.e. critical thinking, always. But if humans did this it would lead them astray: they would not have enough time or information, and would be at the mercy of self-deception and special pleading; and as a result they would often pretend to themselves that the conclusion which suited their own interests was the one demanded by the method.


Humans, therefore, would be well advised to school themselves to have the good dispositions or virtues which will lead them, on the whole, to do what an unbiased and otherwise perfect critical moral thinker would bid them do – if necessary without too much thought, if thought is inopportune. In other words, they should cultivate the same intuitions as intuitionists appeal to, coupled with strong inclinations to follow them, and with other morally desirable feelings (love, for example) which will reinforce them. Only when these general dispositions conflict (as they sometimes will) shall we be driven to do some critical thinking, and even then we shall doubt our own powers.


However, if it comes to deciding what intuitions and dispositions to cultivate, we cannot rely on the intuitions themselves, as intuitionists do. When we have the leisure and are free from selfish bias, we should think critically about which are the right or the best ones, as judged by the extent to which, in general, their cultivation fulfils people's ends. And over the ages the wise have done this; so there is a presumption that the moral convictions shared by thinking people are the right ones to have. But it is only a presumption: some of them may not be the right ones. Is it right, for example, to think that it is morally quite legitimate to eat non-human animals? If we doubt whether our predecessors are right, we are constrained to do some critical thinking ourselves: but even then it will be as well to be humble and not too self-confident. The 'wisdom of the ages' has some [462] authority simply because it is the result of the thought of a great many people in diverse situations.


Descriptivists draw comfort from this 'wisdom of the ages'; they claim that they know that its deliverances are correct. And indeed in a sense they do know; they have learnt, and not forgotten, that certain kinds of act are right and others wrong. But before they draw too much comfort, they should have a conversation with an Afrikaner who knows that it is wrong for blacks to claim equality with whites, or a Muslim fundamentalist who knows that it is right to stone adulteresses.





Aristotle: Nicomachean Ethics.


Austin, J.L: Philosophical Papers (Oxford: Oxford University Press, 1961).


----------: How To Do Things with Words (Oxford: Oxford University Press, 1962).


Hare, R.M.: Freedom and Reason (Oxford: Oxford University Press, 1963).


----------: 'Wanting: some pitfalls'. Agent, Action and Reason, eds. R. Binkley et al. (Toronto: Toronto University Press and Oxford: Basil Blackwell, 1971). Reprinted in his Practical Inferences (London: Macmillan, 1971).


----------: 'Principles', Proceedings of Aristotelian Society, 73 (1972/3). Reprinted in his Essays in Ethical Theory (Oxford: Oxford University Press, 1989).


----------: 'Some confusions about subjectivity', Freedom and Morality, ed. J. Bricke (Lawrence: University of Kansas. 1976). Reprinted in his Essays in Ethical Theory (Oxford: Oxford University Press, 1989).


----------: Moral Thinking (Oxford: Oxford University Press, 1981).


----------: 'How to decide moral questions rationally', in Italian, Etica e diritto: le vie della giustifcazione razionale, ed. E. Lecaldano (Laterza, Rome, 1985). English version in Critica 18 (1987), reprinted in his Essays in Ethical Theory (Oxford: Oxford University Press, 1989). (1985a).


----------: 'Ontology in ethics', Morality and Objectivity: Essays in Memory of John Mackie, ed. T. Honderich (Routledge, London, 1985). Reprinted in his Essays in Ethical Theory (Oxford: Oxford University Press, 1989). (1985b).


---------: 'A reductio ad absurdum of descriptivism', Philosophy in Britain Today, ed. S. Shanker (London: Croom Helm, 1986). Reprinted in his Essays in Ethical Theory (Oxford: Oxford University Press, 1989).


---------: Essays in Ethical Theory (Oxford: Oxford University Press, 1989).


---------: 'Weakness of will'. Encyclopedia of Ethics, ed. L. Becker (New York: Garland, forthcoming).


Kant, I.: Grundlegung zur Metaphysik der Sitten (1785). Refs. of form BA88=441 are, respectively, to pages of earliest editions and of royal Prussian Academy edn, as given in margin of translation by H.J, Paton under title The Moral Law (London: Hutchinson, 1948).


---------: 'On a supposed right to tell lies from benevolent motives', Kant's Critique of Practical Reason and Other Works on the Theory of Ethics, trans. T.K. Abbott, 6th edn (London: Longmans, 1927), p. 361.


MacIntyre, A.: 'Relativism, power and philosophy', Proceedings of American Philosophical Association 59 (1985).


Mill, J.S.: Utilitarianism (1863); ed. H.B. Acton (London: Dent, 1972).


--------: A System of Logic (1843); 5th edn (London: Parker, 1862), last chapter of last volume.


Plato: Meno.


Rawls, J.: A Theory of Justice (Cambridge, Mass.: Harvard University Press, 1971).


Sartre, J.-P.: L’Existentialisme est un humanisme (Paris: Nagel, 1946); trans. P. Mairet, Existentialism and Humanism (London: Methuen, 1948).


Shaw, Bernard: Maxims for Revolutionists, appendix to Man and Superman (London:

Constable, 1903).


Urmson, J. 0.: The Emotive Theory of Ethics (London: Hutchinson, 1968).



Further reading


Seanor, D. and Fotion, N.: Hare and Critics (Oxford: Oxford University Press, 1988).