This book has been printed digitally and produced in a standard specification
in order to ensure its continuing availability
Great Clarendon Street, Oxford OX2 6DP
Oxford University Press is a department of the University of Oxford.
It furthers the University's objective of excellence in research, scholarship,
and education by publishing worldwide in
Oxford New York
Auckland Bangkok Buenos Aires Cape Town Chennai
Dar es Salaam Delhi Hong Kong Istanbul Karachi Kolkata
Kuala Lumpur Madrid Melbourne Mexico City Mumbai Nairobi
Sao Paulo Shanghai Singapore Taipei Tokyo Toronto
Oxford is a registered trade mark of Oxford University Press
The moral rights of the author have been asserted
Database right Oxford University Press (maker)
All rights reserved. No part of this publication may be reproduced,
stored in a retrieval system, or transmitted, in any form or by any means,
without the prior permission in writing of Oxford University Press,
or as expressly permitted by law, or under terms agreed with the appropriate
reprographics rights organization. Enquiries concerning reproduction
outside the scope of the above should be sent to the Rights Department,
Oxford University Press, at the address above
You must not circulate this book in any other binding or cover
and you must impose this same condition on any acquirer
MOORE'S ARGUMENTS AGAINST CERTAIN FORMS OF NATURALISM
by C. L. Stevenson 16
A REPLY TO MY CRITICS
by G. E. Moore 33
THE NATURALISTIC FALLACY
by W. K. Frankena 50
GOOD AND EVIL
by P. T. Geach 64
GEACH: GOOD AND EVIL
by R. M. Hare 74
by Philippa Foot 83
How TO DERIVE 'OUGHT' FROM 'IS' by John R. Searle 101
THE PROMISING GAME by R. M. Hare 115
THE INTERPRETATION OF THE MORAL PHILOSOPHY OF J. S. MILL
by J. 0. Urmson 128
INTERPRETATIONS OF MILL'S 'UTILITARIANISM'
by J. D. Mabbott 137
Two CONCEPTS OF RULES by John Rawls 144
EXTREME AND RESTRICTED UTILITARIANISM
by J. J. C. Smart 171
NOTES ON THE CONTRIBUTORS 184
INDEX OF NAMES 188
Most of these articles are reprinted without revision, or with only small alterations. They are not, of course, to be taken as necessarily reflecting their authors' present opinions.
THE ARTICLES reprinted here centre round two topics lately much debated : firstly the nature of moral judgement, and secondly the part played by social utility in determining right and wrong. Both these debates go back to the eighteenth century, for at that time philosophers divided for and against the moral sense and intellectualist theories of moral judgement, and at the end of the century Bentham laid down that the principle of utility was the foundation of moral good.
The later articles in the volume (numbers IX—XII) are quite simply about utilitarianism, so their relation to the past is clear. Numbers I—VIII are less obviously related to the subject of eighteenth-century battles ; but nevertheless the connexion is close. Like ourselves Hume and his contemporaries were concerned with the possible, or impossible, objectivity of moral judgements. In what, they asked, did the virtuousness of virtuous actions consist? How was it apprehended? Was it rather judged of or felt? Did we know what we ought to do by the intellect or by a moral sense? Was there, indeed, anything there to be known, or did moral discourse rather express our feelings than speak of our discoveries about virtue and vice? Hume himself decided that the search for objective moral properties was vain, and argued that in calling an action virtuous we say nothing but that we feel a pleasing sentiment of approbation in contemplating it, a theory that seemed to explain how moral judgements were linked to action. For we shall naturally be concerned to do, and to promote, what affects us in this pleasing way, whereas if the morality of actions were said to lie in something of which our reason tells us it would have to be shown why this discovery should have a necessary influence on the will.
One might say that the problems that trouble us at the present time are precisely Hume's problems. More directly, however, it has been Professor G. E. Moore who has set the stage for us, and the name of Hume does not even appear in the index of Moore's Principia Et/flea. It is as if we have started with Moore and then worked back from him to Hume. Let us first then say something about the immensely
2 THEORIES OF ETHICS
influential arguments advanced by Moore in 1903.1 Moore's central thesis was that goodness is a simple non-natural property discovered by intuition. The rest of his ethics was built on this foundation, since Moore believed that other moral judgements, for instance those about right action, were related to the basic intuitions of goodness, right action being action that produced the greatest possible amount of good. This last belief made Moore into a kind of utilitarian, but it is not this part of his theory that has interested people most. What seemed particularly important, at least in succeeding generations, was his idea about the judgement that got the whole thing going. Moore insisted that these judgements were objective, and he explained that they were made by intuition. He is therefore called an intuitionist, sharing this label with philosophers such as Prichard and Ross who agreed that moral intuition was the basis of moral judgement, even if they disagreed about where the intuitions came in. An intuitionist is one who believes that in the end we must 'see' that certain things are good, or right, or obligatory. Up to a certain point, they say, one may argue a question of morals, showing that individual cases fall under particular principles by the nature of the facts; but in the end one is driven back to a point at which one can say nothing but 'I see it to be so'.
The difficulties in this position are by now clear enough, and he would be a brave man who would assert, like Ross in the middle thirties, that 'every ethical system admits intuition at some point'. For moral intuition, unlike the ordinary kind of intuition that tells one what someone else is thinking or feeling, is supposed to be the `apprehension' of a quality whose presence cannot be discovered in any other way. Now of course if one knows intuitively that, for instance, someone who is showing no obvious signs of it is angry, then one says 'I just know'. But one learns what it is to be angry from the other cases, and in principle one can put a check on one's intuitions by taking evidence that would settle the matter. Thus one discovers whether, or in which cases, one can trust one's intuitions, and some people's intuitions are demonstrably better than others because more closely correlated with the independently established facts. This independent check is just what is lacking for the supposed moral intuitions, and attempts to assimilate the two by talking, for instance, about intuitions that 'stand the test of time' or those of 'more highly
'G. E. Moore, Principia Ethica. (See Bibliography ior all publications whose details are not given in the footnotes to this Introduction.)
developed peoples' are simply a cheat. For what tells us that the correct moral intuitions may not be those that we think of first but later abandon (`first thoughts are best')? What tells us that primitive peoples do not have a faculty of correct moral intuition that civilization tends to destroy?
It seems, then, that the intuitionist's talk about 'apprehension' and `seeing' is unjustified given his own premises; so also his claim that the man who 'judges' on the basis of his 'moral intuition' is putting forward an opinion about an objective matter. For without any method that could, even in principle, decide between conflicting intuitions we seem to have 'the mere trappings of correction'. I say `I am right and you are wrong' and 'I was wrong when I said . . .', but these sentences merely express a reaction. And if they merely express a reaction we are not far from the subjectivist theories that Moore and the other intuitionists rejected.
Why then, given the difficulties, did Moore support a theory of moral intuition against those who, like Hume, saw moral judgements as an expression of the speakers' feelings or attitudes? Moore's arguments against these theories are the subject that he and Professor C. L. Stevenson are debating in the first and second essays in this volume.
Moore had argued' that a man who says that a certain action is right or wrong cannot simply mean that he has a feeling of approval, or any other feeling or attitude, towards it. For, he says, that would imply that when one speaker says 'X is right' and another 'X is wrong' X would be both right and wrong; and when a man says at one time 'X is right' and at another 'X is wrong' that same individual action X would be at one time right and at another wrong. Stevenson counters by suggesting that 'X is right' means 'I now approve of X', arguing that if consistently applied this would have neither of the consequences suggested by Moore. Thus we cannot say, with Moore, that `If "X is right" said by A is true, then X is right' and if "X is wrong" said by B is true X is wrong' for the conclusions, when translated, becomes 'I approve (disapprove) of X' and I may be a different person from A or B. Moore has, however, a third argument which Stevenson is prepared to allow some plausibility. He says that a subjective 'attitude theory' fails to provide for the disagreement that clearly exists between two speakers who say 'X is right' and 'X is wrong' respectively. For if each is speaking only of his own feelings how will they contradict each other? Perhaps one has the feeling and one does not. Stevenson's
'Ethics, ch. iii.
4 THEORIES OF ETHICS
reply is that there is indeed no logical incompatibility between the two statements: that the speakers do not necessarily hold contradictory beliefs. Nevertheless there is disagreement between them, since their attitudes are opposed. It is the expression of opposed attitudes that gives the opposition between A's 'X is right' and B's `Xis wrong', and it is in this way only that they need 'disagree'.
Stevenson is here drawing on the theory of the emotive meaning of ethical terms, which goes back to the discussions of the Vienna Circle in 1918-19, and was stated clearly by Ogden and Richards in 1923 when they wrote, in The Meaning of Meaning, that in moral language `. . the word "good" serves only as an emotive sign expressing our attitude . . . and perhaps evoking similar attitudes in other persons or inciting them to actions of one kind or another'.' The theory had also been advanced by Professor A. J. Ayer in Language, Truth and Logic, but had nowhere been expounded in such detail as in Stevenson's articles in Mind for 1937 and 1938, and he was further to develop it in Ethics and Language published in 1945. There he says that a word's emotive meaning is what makes it suitable for such 'dynamic' purposes as expressing our own attitudes and changing those of other people, rather than the 'descriptive' purpose of communicating beliefs. The emotive meaning of a word is its tendency to produce affective responses in the hearer, and to be used as a result of affective states in the speaker.
Confronted by the suggestion that ethical disagreement might be merely disagreement in attitude, Moore, who characteristically said that this possibility had 'simply not occurred' to him, agreed that his arguments had been inconclusive. The cause of ethical objectivism now seemed in a bad way. As put forward by Moore it had involved the dubious notion of ethical intuition, and the arguments in its favour had collapsed. Meanwhile it was Moore himself who had attacked the other foi in of objectivism that might have held the field. For he had insisted that no definition of goodness could exist to link the property with provable matters of fact. Thus it was impossible to say, for instance, that 'good' just meant productive of happiness, when it might have been provable that certain things were good. Moore said that such theories committed the 'naturalistic fallacy', and this time he found the emotivists on his side.
That Moore's arguments against naturalism are inconclusive is the thesis of the third article in this volume, which is largely taken up with the discussion of what those arguments were. Moore thinks that no one has the right to put forward propositions such as 'pleasure
and pleasure alone is good' on the basis of a definition: such statements must always be synthetic and never analytic. But what exactly is the 'fallacy' supposed to be? Professor Frankena considers three possible suggestions: (i) that the mistake is that of defining a non-natural property like goodness in terms of a natural one, (ii) that it is the mistake of defining one property in terms of another, and (iii) that it is an attempt to define the indefinable. Frankena argues that whichever version we take Moore has failed to show that any mistake is involved and has in fact simply begged the question. To establish (i) he would have to show that goodness is a non-natural property, which he simply states. For (ii) he would have to show, for each example, that goodness was 'some other thing' from the property with which it was being equated, and this he does not do. To establish (iii) he would have to prove that goodness is a simple and therefore indefinable property, which is something he asserts without proof.
Frankena says, and is surely right in saying, that Moore believed the naturalistic fallacy to be committed by any definition of good; but it is not this that later writers have in mind when they think of Moore as the great opponent of naturalistic ethics. They themselves are interested in ruling out a certain type of definition, and go back to what Moore said about the impossibility of identifying natural with non-natural properties. Unfortunately, however, Moore had never succeeded in explaining what he meant by a 'natural' property; the nearest he came to it was in saying that the goodness of a thing did not belong to its description as its natural properties did. So it did not seem clear what kind of definition was to be excluded. Stevenson, however, claimed that his theory of emotive meaning showed the truth for which Moore had been groping. The point was not that goodness was a special kind of property, for it was no property at all; rather there was a special kind of meaning belonging to ethical terms, and the defective definitions were those that omitted this emotive element in the meaning of 'good'. Thus Moore's non-naturalism could be defended while his intuitionism was undermined. We may notice that the emotivists and intuitionists are alike in one important way: both deny that moral propositions are open to ordinary kinds of proof. The intuitionist says that in the end one must say 'I just see that it is so', while the emotivist admits that he will be driven back to the expression of his fundamental attitudes. For both the argument will come to an end once all the facts have been exposed.
For a number of years it was emotivism, and theories related to
6 THEORIES OF ETHICS
emotivism, that held the centre of the stage. Of related theories the most influential was that developed by Professor Hare, which came to be known under the label of 'prescriptivism'. Hare replaced Stevenson's 'emotive meaning' with his own 'evaluative meaning', explaining that words like 'good' and 'ought' were used 'evaluatively' (to make 'value judgements') when they were used 'with commendatory force'. As so used they entailed imperatives, for Hare says that he makes it a matter of definition that if someone is using the judgement 'I ought to do X' as a value judgement he must recognize that . if he assents to the judgement he must also assent to the command "Let me do X" Thus a man who uses the word 'good' evaluatively must accept a first person imperative, and behind each particular imperative will lie a general 'quasi-imperative' addressed, as it were, to all persons at all times. Hare is not denying that words such as `good' and `ought' can be used `non-evaluatively', but the definition is supposed to have some rough correspondence with what we would mean by a value judgement in everyday life. With evaluative meaning Hare contrasts descriptive meaning, but like Stevenson gives no account of this side of the dichotomy. To be descriptive a word must be non-evaluative, so that he says there must be 'definite criteria for its application which do not involve the making of a value judgement'. A word may have both descriptive and evaluative meaning, but is called a 'descriptive word' only if there is no evaluative element.
Thus equipped Hare proceeds to launch a full scale attack on ethical naturalism, defining a naturalist as one who tries to equate value words with those whose meaning is 'purely descriptive' and who therefore believes he could deduce an ethical conclusion from descriptive premises. The price of naturalism, says Hare, is the loss of the commendatory or action-guiding force of the ethical terms, and one of the great advantages that he claims for his own theory is that it shows how moral judgement is necessarily connected with choice. Indeed both Stevenson and Hare seemed to have provided for the necessary connexion between morality and the will on which Hume had insisted. In Stevenson the connexion between moral judgement and action was built into the theory of emotive meaning: an emotive word expresses the speaker's attitudes, which the listener is thereby invited to share, and since an attitude is 'marked by stimuli and responses which relate to hindering or assisting whatever it is that is called the "object" of the attitude' this means that the use of an emotive term tends to express the speaker's willingness
'R. M. Hare, The Language of Morals, p. 168.
to do certain things, and to influence the hearer in a similar direction. Hare, as we saw, had connected the evaluative use of language with the acceptance of first person imperatives, and quasi-imperatives addressed to the world at large. He could, therefore, claim that on his theory value judgements were essentially 'action guiding', bearing this relation both to the speaker's own actions and to those of other people. Picking up Hume's assertion that moral judgements are necessarily practical he connects this with Hume's famous dictum about the gap between 'is' and 'ought'. An 'ought' cannot be deduced from any descriptive statement, since 'oughts' have this special connexion with the direction of choices, and 'is' statements do not.
It is this position that Hare is defending against Professor Geach in the fifth article reprinted here. In his attack Geach had argued against Hare's explanation of the 'action-guiding' function of the word 'good', and so against his theory of evaluative meaning. Geach agrees with Hare that 'good' is an action-guiding word, for it belongs to the idea of goodness that normally and other things being equal people should choose things that they call good. But this does not mean that when used in its normal sense the word must be used 'for commending'. On any particular occasion the direction of choices may simply not be in question, and the word is not then used in any special way. Thus there is nothing to stop an expression of the 'good F' type from having a straightforward descriptive sense.
Geach does, however, see a difficulty in his own position. For suppose that the expression 'a good action' has a fixed descriptive meaning, and we can pass, say, from the fact that an action is an act of adultery to the fact that it is a bad human act. How are we to get from the supposedly descriptive sentence 'adultery is a bad human act' to the imperative 'you must not commit adultery' ? Why should the thought that it is a bad action deter anyone from doing it? Geach replies that 'although calling a thing "a good A" or "a bad A" is not of itself something that touches the agent's desires, it may be expected to if the hearer happens to be choosing an A'.1 And what a man cannot fail to be choosing is his manner of acting; so to call a manner of acting good or bad cannot but serve to guide action.
Not surprisingly Hare finds this reply quite unsatisfactory. He argues that if 'man' and 'action' are taken as functional words like `knife' and 'soldier' then 'good knife' and 'good soldier' will of course have a 'fixed descriptive meaning'. But then it will no longer be true that what no one can fail to be choosing is his manner of
'See p. 71.
8 THEORIES OF ETHICS
acting. For it could very well be that some individual had no interest in doing those things which make a man a good man, when he would choose his actions under some other heading and by other principles of choice. Geach has, therefore, failed to account for the fact that moral judgement, unlike some others of the 'good A' form, must be action guiding for each man whatever his particular desires. Hare himself had guaranteed this by insisting that 'good' when used evaluatively carries in its meaning a commitment to choice. Before a functional word such as 'soldier' it is not so used, or rather its evaluative import is neutralized by the word 'soldier'. For this word introduces a special point of view from which a choice would be made, so that it is as if one were saying w trying to be a good soldier that is what I must do'. An expression which is, as a whole, evaluative, is one that carries an actual, not a hypothetical, rule of action, and this is what a moral judgement must surely be.
The problem that was vexing Geach was also troubling the present editor when she wrote the article 'Moral Beliefs', printed as number VI in this volume. In the first half of the article I had argued against the idea of an evaluative element in the meaning of the word 'good' which should be independent of its descriptive meaning, saying that we cannot make sense of the notion that a man is thinking 'this is a good action' if he brings the wrong sort of evidence to show that it is a good action. Nor will it necessarily help to appeal to feelings that he has, for there are some feelings that cannot be attributed to a man unless he has the right thoughts. This part of the article suggested that the expression 'a good action' had a fixed descriptive meaning, or at least that it was fixed within a certain range.
Now, this, though it has been in fact rejected by emotivists and prescriptivists, who think it a contingent matter if our evaluative terms possess a fixed descriptive meaning, is not right at the centre of the dispute between the two parties. For the anti-naturalist could agree that an expression such as 'good action' had a fixed descriptive meaning while still arguing for an extra 'volitional element' in value judgements. Perhaps a man who calls an action a good action must apply certain descriptions to it but also have certain feelings or attitudes, or accept particular rules of conduct ? How else is the action-guiding force of the word to be maintained? In the second part of the article I suggested that it could perfectly well be given by the particular facts with which the goodness of a good action is connected. For some facts about a thing are such as to give any man a reason for choosing that thing. The difficulty was, of course, to show that