1. We have dealt with emotivism at length because it is the most arresting and influential ethical theory of the past quarter century. In the pure form in which we have discussed it, however, there are probably few moralists who would still subscribe to it. The decline of its popularity is largely due to the recognition of three deficiencies in it.
In the first place, its denial that there is such a thing as moral judgment at all was in too flagrant an opposition to ordinary meaning. This was admitted by some emotivists themselves.
‘Certainly’ wrote Mr Ayer, ‘the view, which I still wish to hold, that what are called ethical statements are not really statements at all, that they are not descriptive of anything, that they cannot be either true or false, is in an obvious sense incorrect. For, as the English language is currently used—and what else, it may be asked, is here in question?—it is by no means improper to refer to ethical utterances as statements… when someone wishes to assent to an ethical verdict, it is perfectly legitimate for him to say that it is true, or that it is a fact, just as, if he wished to dissent from it, it would be perfectly legitimate for him to say that it was false.’1
If Mr Ayer continues to hold an emotivist view,
‘this does not mean that all ethical statements are held to be false. It is merely a matter of laying down a usage for the words “proposition” and “fact”, according to which only propositions express facts and ethical statements fall outside the class of propositions.’2
Where does this leave us? It is ‘legitimate’ for the plain man to go on talking as if moral judgments were true or false; it is also legitimate or the emotivist to talk as if they were not. But the plain man's usage is legitimate only in the sense that it reflects the ordinary use of language; Mr Ayer's usage is legitimate in the sense that it reflects the truth, namely that moral ‘judgments’, however taken by the plain man, are cognitively meaningless. It seems to me idle in such a case to say, as Mr Ayer does, that he is merely recommending a new usage. The new usage implies that plain men and most moralists have been fundamentally mistaken about what they meant when they made moral judgments. One can state this issue, as one can any other issue, as if it were a verbal question, the question in this instance whether a moral judgment shall be called a proposition. But the issue of moment here is not what it shall be called, but what it is, and behind Mr Ayer's new usage is an ethical theory which he admits to run counter to the common one.
But to those among the analysts who had a higher veneration for common usage, this was taking too large a liberty. Moore had said, and the later Wittgenstein more emphatically, that one must not play fast and loose with actual meaning; and if it could be shown that we actually did, in our moral statements, mean to say something true, to argue about moral beliefs and contradict them and make inferences from them, then emotivism could not offer itself as a correct analysis of common meaning. We must amend it into closer conformity with this meaning.
2. Secondly, it soon became plain that the emotion of the emotivists was by no means a simple affair. One of the complaints against them was that they seemed to reduce the moral attitude to one of mere liking or aversion, whereas moral approval and disapproval, even if emotions merely, had qualities of their own. Indeed, on further examination they looked less and less like emotions at all. What was expressed in moral judgment might be any one of a wide variety of attitudes—an entreaty, for example, an excuse, a wish, or some form of command. These were not primarily cognitive states; so those who placed them at the heart of the moral judgment were still aligning themselves on the whole with emotivism. But they insisted, and with justice, that they were not mere or pure emotivists. They had subjected the non-rational attitudes expressed in such judgment to a more discriminating analysis.
3. Thirdly, it was widely held that emotivism had gone too far in its dismissal of argument in morals. Stevenson's kind of emotivism did not permit one to say that any kind of conduct was, strictly, more reasonable than any other, or to defend any moral pronouncement by logically relevant means. All one could do if one differed from another was to bring causal pressures to bear upon him—including, to be sure, the recital of facts—in the hope that they would change his attitude. But since he had asserted nothing, there was nothing for one's arguments to be relevant to; it would be absurd to try to establish logically one's own emotion, or to disprove his. This situation was felt by later analysts to be awkward in the extreme. It was too plain for denial that men did argue about moral issues, and that at times they succeeded in justifying their views by evidence universally felt to be relevant. To be sure, one could never prove a moral judgment in the same conclusive fashion as a statement in geometry. But to say that repaying a creditor was no more reasonable, in the sense of justifiable by reasons, than cutting his throat was an outrageous position.
To anyone studying the place of reason in moral judgment, this last line of retreat from emotivism is, of course, particularly interesting. Unfortunately, we cannot attempt to deal with it fully. Though the important books on ethics in recent years have been all too few, the spate of articles has been prodigious, and (apart from the dreariness of writing that protests continually its concern with verbal usage), to deal with the nuances of the many minute departures of recent years from emotivism would hardly be possible. We shall merely indicate with brief comment a few of the more significant lines which the retreat has taken.
4. One mild rising against emotivism we have already noted. Mr Edwards has seen that such terms as ‘good’, ‘right’, and ‘ought’, usually express more than a feeling or an attitude. What more they express depends, he holds, on the context of their use. If I speak of a good (or nice) steak, I mean that it is tender, done to the degree of rareness that I like, and so on. If I speak of a man as good, I mean that he is kind, gentle, free from envy, and so on. The term ‘good’ both expresses my feeling and refers to these qualities. Since, then, in calling something good, I am asserting certain qualities to belong to something, I can defend my judgment in the same way as if I had said it was square or made of wood. If challenged, I shall point to the existence of the denoted qualities; the best conceivable reason for my judgment's truth is the presence in the object of the qualities my words refer to.3
5. This is a move in the right direction. It formally concedes that value judgments are really judgments, not interjections merely. But its break with emotivism is too half-hearted to be a real advance. The emotivists had already owned that when we use value terms it is because of the recognized presence in the object of certain qualities, but they had declined to include these qualities in the meaning of the terms. Mr Edwards’ new proposal is that they should be so included. But is the advance more than verbal? The emotivists had distinguished between descriptive and emotive meanings and said that we could give reasons for the first, but not for the second. Mr Edwards would extend the meaning of ‘good’ to include both these types. But within this extended meaning he would draw precisely the same distinction as that already drawn by the emotivists, and would hold, as they do, that for the descriptive meaning we can give reasons, and for the emotive, not. And since the value meanings fall exclusively on the emotive side, he stands, as regards reason in moral judgment, precisely where the emotivists did.
He is reluctant to think this, and holds that he has made a genuine advance. Stevenson had held that if one offered reasons for a moral statement, they had a causal relevance only; they might tend in fact to win over an opponent, but since one had submitted no proposition, one had said nothing to which one's ‘reasons’ could be logically relevant. Edwards holds that on his view they would be relevant, since the meaning of ‘good’, ‘right’, and ‘ought’ will now contain a factual component, and for this factual part of the statement reasons may be offered.
Where is the advance? The emotivists do not deny this. Edwards would be going beyond them only if he showed that logically relevant reasons could also be offered for the value component in the statement. This he cannot do, for he concedes as completely as they do that the value side of it is emotive purely, and that one can offer no logical grounds for an interjection. He thinks that ‘Stevenson's theory concerning the nature of moral disputes is altogether mistaken’,4 since it implies that ‘settlement’ of them means merely the achievement of similar attitudes, and ‘reasons’ are merely causes acting upon the emotions. But when one examines the new senses of ‘settlement’ and ‘reasons’ which he proposes to recognize, one finds that they apply as exclusively to the factual side of moral statements as Stevenson himself contended. If one wants to attack the value side of an opponent's statement, there is still only one way to do it. The statement, on that side, expresses nothing but an attitude; attitudes cannot be proved or disproved by reasons; they must be worked on through causes; and the most effective causes are often such as sensible men would repudiate as irrelevant. Mr Edwards is rightly repelled by the emotivists’ assimilation of ethical reasoning to propaganda, and does his best to avoid it. I share to the full his repugnance to it. What I cannot understand is how or why so clear-headed a writer can have supposed himself to have escaped from it.
6. A position similar to Mr Edwards’ has been suggested by Mr Urmson in an able and much-discussed article ‘On Grading’.5 Mr Urmson holds that to call things good or bad is to grade them, and that we shall get needed light on how these terms are used in morals if we consider their use in the professional work of grading, for example the grading of apples. If we do, we find that for admission into each grade there is a set of criteria; to be labelled as ‘super’, for example, the apple must be of a certain size, ripeness, and colour. Whether we speak about a good apple or a good gun, a good cricketer or a good man, we are grading the object before us, and in doing so we are making conscious or unconscious use of certain criteria. Mr Urmson thinks that in all these cases we are using the word ‘good’ in the same sense,6 but that the criteria used in applying it differ from case to case. Ordinarily when we disagree about the goodness of something, we can discuss the matter profitably, because we do not really differ either as to the meaning of ‘good’ or as to the criteria for applying it, but only as to whether one or more of these criteria are present. If you call Jones a good man and I disagree, you can point out to me that Jones does in fact pay his debts and is in fact kind to his children, despite what I have heard to the contrary; and when I am set right as to these facts, I shall presumably ‘grade’ him as you do.
7. But suppose we have different criteria for calling things ‘good’; what then? Suppose that a sensualist and a scholar are taking about the value of knowing Latin, or the things to which such knowledge gives access. Or suppose a communist and a democrat differ as to whether Jones is a good man. Here the very criteria on which they would base their use of the word may be different. ‘When there are differences of opinion about what grading criteria to adopt in a given situation, is there not a right and wrong about it… or are we to say that the distinction, for example, between higher or lower, enlightened and unenlightened moral codes is chimerical?’ ‘Certainly’, Mr Urmson writes, ‘we do not want to say this’. But if we really differ about our criteria for higher and lower, how can we proceed? ‘Of course we cannot, when debating what criteria to use for moral grading, grade the criteria morally’,7 since it is precisely the criteria for such grading that are under dispute. Can we cut the knot by recognizing one code as more enlightened than another? But then ‘enlightened’ is again a grading word. Mr Urmson seems almost ready to give up the struggle. ‘Clearly when we debate which of two moral codes is more enlightened there is no ultimate court of appeal, no umpire, unless some agreed revealed religious code is treated as a dens ex machina’; and in this he plainly does not believe. But is it the case that we really do differ in our criteria for enlightenment? Here Mr Urmson finds grounds for hope. No doubt a complete and definite list of such criteria would not be agreed upon. But would any sober person deny that a culture is enlightened so far as it contributes to the health, wealth, and happiness of those who live under it? Mr Urmson appears to think not. On such criteria for the goodness of competing codes men generally do seem to agree.
8. Now granting that in grading codes we use the same ultimate criteria, what is the relation in a given case between these criteria and the label ‘good’ which they carry with them? Granting that under code A there is more health, wealth, and happiness than under B, are we to say that these things entail the greater goodness of A, either analytically or synthetically? Or are we to say that they merely cause in fact an emotion of approval? To say that the criteria abc are the goodness, or entail it analytically, is naturalism. To say that they entail the goodness synthetically is, in substance, intuitionism. To say that they merely give rise to a favouring emotion is emotivism. Mr Urmson rejects all three. He thinks they all miss the relation between the criteria and the label. The affixing of the label ‘good’ is not the making of a judgment either analytic or synthetic, nor is it ‘a squeal of delight’; it is an act, a doing something, the making of a choice, together with the committing of oneself to similar choices in like situations.
9. What is interesting about this theory is its evident retreat from emotivism toward the view that moral judgment can be defended by reasons, combined with a clinging to the emotivist view that ‘x is good’ is not a judgment at all. Consider the first point. If the communist and the democrat differ as to whether a man or a culture is good, they can compare their criteria for assigning this label. If they find among these criteria one that is admitted by one party but not by the other, they can discuss whether its inclusion is enlightened or not. And on the criteria for enlightenment, they will substantially agree; if the presence of the disputed factor clearly involves more health, wealth and happiness, the disputants will—if candid—agree that it should be retained. Mr Urmson thus believes that the choice between good and bad is not arbitrary but objectively based, and that it can be defended by insights that are all but universal. A belief that health, wealth and happiness are bad can be set down by virtually universal agreement as the mark of an unenlightened mind. Suppose we ask why—as Urmson does not. Surely the answer is that it is so obviously false. We should say that anyone who really thought health, wealth and happiness to be bad or of no account was either stupid or mentally twisted. Our major criterion for being enlightened is the power to see what is true. On this interpretation of his view, Mr Urmson is moving back precipitately toward a cognitive ethics. There is hope of agreement because when it comes to the ultimate issues of good and bad, enlightened minds will render identical verdicts, and they will do so because they will apprehend a common truth.
10. Unfortunately, what he offers us with one hand he takes away with the other. Having raised our hopes by suggesting that ‘good’ and ‘bad’ are assigned on objective grounds and can be defended in the last resort by appeal to criteria of enlightenment which are the same for all men, he goes on to say that the application of the terms is not a rational or cognitive process at all, but an act, all too vaguely conceived. But how can enlightened insight be more relevant to an act than to an emotion? An act is not true or false; no evidence can prove or disprove it, or render it more or less probable. Hence Mr Urmson's position remains to me less than clear. The statement ‘happiness is good’ seems to be a rational statement, in the sense that it shows an enlightened solution of the criteria for goodness. But this same statement is not really a statement at all, but an act of choice which, as non-cognitive, is neither rational nor irrational. How these two sides are connected I do not see.
One suspects some want of consistency. ‘X is good’, as the affixing of a label, is taken to be not a judgment, but an act of choice. The statement that criteria abc are the right ones for the affixing of the label does seem to be a judgment, which can be argued for by showing that its rejection would be unenlightened. But is not the selection of the criteria abc as the ‘right’ ones as much an act of choice as the selection of ‘good’ as the right label for them? If ‘x is good’ is not a judgment, but an act, so is ‘abc are the right criteria for good’, and then there would seem to be no room for rationality anywhere in the original statement or in the statements supporting it. On the other hand if ‘abc are the right criteria’ is a judgment, though a selective one, then there is no reason why ‘x is good’ should not be one too. Both statements are equally judgments, and equally capable of truth or falsity. Such criticism may, indeed, rest on misunderstanding; Mr Urmson presents his view only in brief outline. In any case the theory is interesting as typical of post-war linguistic ethics. There is a manifest desire to escape emotivist irrationalism, coupled with an equally manifest desire to refuse moral judgments the position of judgments at all. Its equilibrium is therefore unstable.
11. The same double desire is evident in the ethics of another writer of the linguistic school, Mr R. M. Hare. His book on The Language of Morals is an admirably clear and careful statement, and fortunately for the interest of his readers, it is more concerned with matters of substance than would be gathered from his insistence on talking about ‘the peculiarities of sentences’ and ‘the behaviour of words’. We may well ask the same two questions about his position that we asked about Mr Urmson's: What is it that is expressed in such judgments as ‘X is good’? and, in what sense can they be supported by reasons?
Mr Hare loses no time in answering the first question. There is a fundamental difference between what is expressed by descriptive and by prescriptive language, and a statement of ‘good’, ‘right’, or ‘ought’, is of the latter kind. A descriptive statement is one that makes a true or false assertion, such as ‘you are going to shut the door’. The simplest form of prescriptive statement is an imperative, such as ‘shut the door’. Are all moral judgments imperatives? Mr Hare seems to hold that they are. He is not an emotivist; ‘x is wrong’ is not an expression of felt disapproval merely. It expresses, rather, something like ‘Don't do x’. The distinctive feature of moral judgment is that of prescribing or guiding choices. When I say ‘x is wrong’ or ‘you ought not to do x’, I am uttering a kind of general instruction or imperative designed to prevent people from choosing x. More surprisingly Mr Hare treats the judgment of good also as essentially imperative. At first, indeed, it looks as if he were going to deal with it in a different way, for in discussing it he takes as his motto the definition in the Oxford English Dictionary: ‘Good… the most general adjective of commendation…’, and says that the one constant office of the term, which holds equally whether we talk of ‘a good fire-extinguisher’, ‘a good sewage-affluent’ or ‘a good man’, is this one of commending.8 But it soon turns out that commendation itself is a kind of imperative; ‘to commend is to guide choices’.9 Its primary work is to urge upon others or myself the choosing of one kind of thing or action rather than another.
12. This being Mr Hare's answer to our first question, how would he answer the second? Can we support imperatives by reasons, or call one more rational than another? He is eager to show that we can, and he uses two principal means of doing so. The first is to revive the distinction between the meaning and criteria of ‘good’ with which Mr Urmson has made us familiar. He holds that though the primary meaning of ‘good’, that is, its commendatory one, remains the same through every context, its secondary meaning that is, the criteria for applying it, vary from case to case; ‘good fire-extinguisher’ does convey information, and a different kind of information from ‘good sonata’ or ‘good man’. And regarding the facts thus secondarily asserted when we call something good, it is plainly possible to offer evidence, just as it was in the theories, so similar at this point, of Mr Edwards and Mr Urmson.
Mr Hare's own device is more original. He holds that the imperative itself can be analysed into two parts, and that one of these parts turns out to have most of the logical features of an ordinary indicative statement; it can be defended, contradicted, and used as a basis for inference. ‘You are going to shut the door’ is an assertion. ‘Shut the door’ is an imperative. But they clearly have something in common: both of them are about the same thing, namely your shutting the door in the immediate future. We could recast them in the forms:
Your shutting the door in the immediate future, yes.
Your shutting the door in the immediate future, please. The first part of these sentences, which is the same for both, Mr Hare calls the phrastic; the second part, which varies, he calls the neustic. ‘The utterance of a sentence containing phrastic and neustic might be dramatized as follows: (1) The speaker points out or indicates what he is going to state to be the case, or command to be made the case; (2) He nods, as if to say “It is the case”, or “Do it”. He will, however, have to nod in a different way, according as he means one or other of these things.’10
Now if we fix our eyes on the common phrastic above, we see that it gives to the imperative containing it all that is necessary to serve as a basis of rational discussion. The imperative can be contradicted, as in ‘Your not shutting the door in the immediate future, please’; it can be supported by reasons, as when we say the room is too cold; it can be declared unreasonable, as when we say the door is fixed, and so cannot be moved. Indeed a logic of imperatives that is very like the logic of indicatives has been shown to be possible.11 In reasoning about imperatives, the most important point to bear in mind, however, is that though facts may be introduced in support or objection, no imperative conclusion can be drawn from factual statements only; there must be at least one imperative in the premises. From the premises, ‘You ought to prevent needless pain’, and ‘This is a case of needless pain’, it follows that you ought to prevent it. From the minor alone, ‘This is a needless pain’, no statement of obligation follows.
13. Mr Hare's ethical theory is the most completely elaborated that we have so far had from the linguistic moralists. Is his conception of moral judgment as an imperative an advance over the emotivists’ view of it as an interjection, or Mr Urmson's as an act of choice? A little, perhaps. But it has failed to satisfy some of those who most firmly agree with him that moral ‘judgments’ are not really judgments. Mr Braithwaite, for example, has pointed out that to recast ‘Lies are always wrong’ into ‘No lies being told by anyone, past, present or future, please’, is not plausible.12 We plainly do make moral judgments about the past—a fact, indeed, that we have found fatal to both the instrumentalist and emotivist theories. I suspect it is also fatal to imperative theories. For we cannot intelligibly be said to command something to be done in a past that we know to be unalterable. Nor again can we sensibly enjoin an avoidance of lying on persons we know to be beyond our reach, like our descendants of a century hence, though our moral judgment is clearly meant to be universal. Mr Hare himself suggests a further difficulty which, though he does his best to answer it, seems to me unanswerable.13 Is it not possible to say to someone meaningfully, ‘You ought to do x, but don't’? Here the moral statement and the command, far from being identical, are in conflict with each other. I may be convinced, for example, that you ought to make your will in favour of Jones, but want very much that you should enrich me instead, and may do all I can to that end. The difficulty is the fundamental one that appraising an action as right or wrong is simply different in intention from the attempt to guide future choices. ‘If I censure someone for having done something’, says Mr Hare, ‘I envisage the possibility of him, or someone else, or myself, having to make a similar choice again; otherwise there would be no point in censuring him.’14 No practical point, perhaps, but is my only interest in making the judgment the interest in getting something done? If it were, the censure spoken silently, of a man I was sure was incorrigible, would not really be censure at all, since it is not an attempt to get either him or myself to act. Yet it may carry a full and unequivocal moral meaning. Both censure and commendation may retain their moral point when a command would have no occasion and no sense.
14. Does Mr Hare deal more successfully with the other issue of importance, as to the support of moral judgment by reason? His thesis is that we can get at the neustic, so to speak, through the phrastic. Let us see how one would go about it to support a moral judgment. We have a friend who never opens a newspaper. We say to him, ‘You really ought to look at a paper now and then.’ ‘Why should I?’ he asks. ‘Because you ought to take some interest in politics.’ ‘Why should I interest myself in politics?’ ‘Because you are supposed to be able to cast an intelligent vote.’ ‘But why should I trouble to vote?’ ‘Because if people don't trouble to vote, they can't make a democracy work.’ ‘Why should they care whether democracy works?’ ‘Because if it doesn't they'll soon find themselves with something very much worse. Some set of persons, hungry for power, will probably take over, and exploit them unmercifully; they will cut down their right to criticize at home and involve them in trouble abroad.’ Here arguments, more or less relevant, are offered for an imperative. Mr Hare would no doubt remind us of two points about them. First, the arguments are never merely factual, for no ought follows from an is; one premise at least must itself be an imperative. Secondly, every moral judgment is really a judgment of principle. In saying to my friend ‘You ought to read the papers’, I am saying by implication, ‘Anyone circumstanced as you are should read them.’ What these requirements amount to is that if I am to argue for my judgment, I must bring it, as a particular case, under an imperative principle, which may be very sweeping and complex. In the present case the major premise is something like, ‘Whatever steps are necessary to avoid the destroying of civil liberties and the mutual trust of nations should be taken.’ Then—with the help of subsumptions—‘Your reading the papers is such a step; therefore you ought to take it.’ A fully adequate major premise would involve the description of a whole way of life. For
‘a complete justification of a decision would consist of a complete account of its effects, together with a complete account of the principles which it observed, and the effects of observing those principles… Thus, if pressed to justify a decision completely, we have to give a complete specification of the way of life of which it is a part. This complete specification it is impossible in practice to give; the nearest attempts are those given by the great religions… Suppose, however, that we can give it. If the inquirer still goes on asking “But why should I live like that?” then there is no further answer to give him… We can only ask him to make up his own mind which way he ought to live; for in the end everything rests upon such a decision of principle.’15
15. This is admirable—up to the last step. Mr Hare is surely right that we do argue in this way for our moral judgments; we do plainly argue from principles, and we may see, if we are fortunate, that these principles hang together in some large view of how life should be lived. But when we come to a conflict between two conflicting ways of life, is that the end? Is the choice between them a matter, not of rational judgment, but of a non-rational decision or act of will? Mr Hare thinks it is. If a Greek of the fourth century BC were to meet a Hebrew prophet of the same period, they would soon find themselves in conflict about ultimate and intrinsic values—about the value of knowledge, of holiness, of athletic sport, of the experience of beauty, of contrition for sin. Regarding such differences Mr Hare Would say, in terms as unqualified as the emotivists’, that to talk of an objective decision, of either party's being really right or wrong was meaningless. Even in theory, there is no ground on which a judge could stand in arbitrating the difference. All that either party could do would be to announce a commitment, to issue an imperative to himself and others which, as essentially an expression of preference, was beyond rational vindication or criticism. At this point the linguistic moralists of Britain make a curious rapprochement with the existentialists of the continent. The ultimate act of moral choice is, for both alike, an act of will responsible to nothing beyond itself.16
The linguistic moralists consider that they have left emotivism far behind. A moral judgment as Hare conceives it can be supported by arguing from its coherence with an extended system of principles, and one can say to anyone who disputes it, ‘you must either accept the judgment or reject all that it implies’. This looks like an impressive advance. The trouble is that having carried us this grateful distance beyond emotivism, these moralists pull us back at the last step with an irrationalism that calls the whole advance into question. For if the imperative ‘you ought to read the papers’ rests on a wider system, and this system in turn on a non-rational commitment, then all the commitments that depend on it are similarly in the end indefensible.
Why not be content with this? We gave our answer in the last chapter. What we have here is a free use of reason on the lower slopes with a surrender to unreasoned decision at the summit. On this summit the imperativist and the emotivist join hands. We have seen what to think of emotivism. We could not believe that when the adherents of different ways of life, the communist for example and the democrat, differed about the value of individual freedom or happiness, they did not differ in belief at all. We have the same trouble with imperativism. What we are concerned about is whether our moral judgments are really judgments, and it makes very little difference whether, if this is denied, it is on the ground that they are interjections or on the ground that they are commands. Emotivism and imperativism alike deny the chief thesis we are concerned to support, and we reject them for much the same reason. Both hold that the ultimately differing parties—the communist and the democrat, the Greek and Hebrew—are not really differing in opinion or belief; that neither denies what the other is saying; that neither is making a mistake; that neither is nearer the truth than the other, because in such conflicts there is no truth to be found. On such a theory, as we have seen, neither can make a value judgment of the past without a distortion of meaning; neither, strictly speaking, can offer arguments that are relevant to the final decision; neither can distinguish, in the end, propaganda from moral reasoning. If Mr Hare met an opponent of the Christian way of life, in which I take it he believes (or, rather, to which he has made a non-rational commitment), he would no doubt set out in his able fashion its doctrines of love, sin, and forgiveness, and its attitudes toward art, sex, and science, exhibiting these as some sort of connected whole. Having done so, he could only say, ‘Take it or leave it.’ His subscription to the Christian way of life as a whole is neither more nor less reasonable than its rejection; since it is not a judgment but an act of will, it of course does not follow from anything that has been presented, nor does his opponent's opposite subscription; and if he wants to change his opponent's attitude, his only recourse, so far as I see, is to psychological pressures. These things I cannot accept. The ultimate reason is merely that a judgment, even about intrinsic value, is too plainly a judgment—not an expletive nor yet a command. The case for this has been stated in part in the last chapter. The rest of the case lies in an alternative account of the meaning of ‘good’, which will presently be offered.
16. Among the linguistic moralists, the writer who has gone most extensively into the way moral judgment makes use of reasons is Mr Stephen Toulmin. To him this is the most interesting question in ethical theory. It is plain enough that moral judgments are not, like scientific judgments, true or false in any straightforward sense; their function is ‘to alter one's feelings and behaviour’.17 Yet we must face the arresting and undeniable fact that we do offer reasons for them, and that these reasons are confidently appraised as relevant or irrelevant, good or bad, and often as conclusive. This fact is enough in itself to impose a veto on both subjectivist and emotivist ethics. The subjectivist says that ‘x is right’ means ‘I approve of x’. But it would be absurd to offer arguments in favour of the thesis that I am having a certain emotion. The emotivist says that ‘x is right’ is an exclamation evincing feeling and not reporting it. But again it would be absurd to adduce arguments for an exclamation, whereas everyone does offer them for moral judgments. Mr Toulmin would say nothing so monstrous as that common sense and common usage might be mistaken in these matters; for, lively as his sense of humour is, he takes with great seriousness the tables of the law delivered to Wittgenstein about the inviolability of usage. People do in fact argue about their moral judgments; hence any theory that would forbid their doing so must be in error.
It might be inferred from these linguistic loyalties that when people said ‘x is good’ Mr Toulmin would take them to mean ‘x is good’ in the most straightforward sense. But not so. It is curious to recall, in passing, what moral philosophers have done when confronted with this simple statement. Moore argued that it should be taken at face value, as a judgment ascribing a quality, and often true. Westermarck agreed in this interpretation, but since the truth of the judgment, so interpreted, collided with his ethical theory, he held that common sense was on this point universally mistaken. Mr Ayer, though insisting that philosophy consisted of definitions and that its definitions were to be tested by reference to usage, also admitted that ‘x is good’, was commonly meant to assert what was true; but since the acceptance of such truth would collide with his ethical theory also, he ‘recommended’ a different use of ‘good’. Mr Toulmin, much influenced likewise by Wittgenstein, takes still another course. He holds that ordinary men do not really mean to ascribe a property at all, as the objectivists have claimed. He offers, so far as I can discover, three arguments for this view.
17. In the first place, the sort of property objectivists have talked about neither exists nor can be conceived, and it is queer to charge the plain man with making daily use of something so elusive. Mr Toulmin runs over the possible types of property and finds that goodness is not among them. A property must be one of three kinds. It may be a ‘simple quality’ like the red I now see in a flower; or it may be a ‘complex quality’ like 257—sided, whose presence can be checked only by some appropriate routine like counting; or it may be a ‘scientific quality’ like ‘radiating such-and-such types of electromagnetic wave’. Properties of the first two of these types are directly perceivable; those of the third are not. Mr Toulmin proceeds to argue, successfully I think, that goodness, not being directly perceivable, is not a property of the first type or of the second. The argument that it is not of the third type never materializes,18 though one would have supposed that it was precisely in this category of the three mentioned that goodness was most likely to fall. But even if it falls in none of these categories, the argument will not carry much weight with those against whom it is chiefly directed. Moore, Broad, Ross, and Ewing all take goodness as belonging to none of these types, but as being, or at least involving, a non-natural character. Mr Toulmin's argument is of the form: all properties are of kinds a, b, or c, goodness is none of these; therefore it is not a property at all. Such an argument is of no weight unless the major is shown to be exhaustive. And those who have held that goodness belongs to none of these types but to a fourth type, d, will not be much moved.
18. Secondly Mr Toulmin argues, the plain man has been misrepresented. Moralists of traditional types have assumed without inquiring that he agreed with them, that because in saying ‘x is good’ he used an adjective, he must mean by it what adjectives usually mean, namely some kind of attribute. But if we take the trouble to ask him whether he means what Moore, for example, makes him mean, his answers give no colour to this interpretation. Suppose, says Mr Toulmin, that I am talking to a Moorean philosopher about a common friend who by general agreement is a man of the highest character.
‘“Surely,” I may say, “if ever a man knew what goodness was, he does!”
“I imagine that he does,” the philosopher will say.
“And yet,” I may reply, “I have asked him whether, when making up his mind what to do, he is conscious of observing any ‘non-natural property’, any ‘fittingness’, in the action he decides on, and he says he isn't. He says that he does what he does because there's a good reason for doing it, and that he isn't interested in any additional ‘non-natural properties’ of his actions”.’19
Surely this result is not very significant. Philosophers have found it hard themselves to say what they mean by good; and when the ordinary man is suddenly confronted by the meta-ethical question, which he has never thought of asking, whether he is referring to some kind of ‘non-natural property’, it is only to be expected that he should hesitate or say No. But an opposite result could no doubt be gained with equal ease by merely rephrasing the question. If one were to ask him whether in saying it was good to have an educated mind he meant to say something true, he would presumably say ‘Of course I do’; if then asked whether he could verify what he meant by means of eyes, ears, or hands, he would probably say ‘No’. Where meanings are so vague, and those who use them are so little practised in analysis, it is only too easy, with the help of some verbal legerdemain, to extract the answer one wants.
19. Thirdly, Mr Toulmin thinks that when objectivists have taken moral judgments to ascribe a character to an object, they have been moved by an argument which, if brought clearly to light, can be seen to be fallacious. They have had the experience of differing with others as to whether a proposed action was right, and they have supposed that this difference was possible on one assumption only. If two people differ about the colour of something, one saying it is red and the other denying this, what are they differing about? Obviously whether the quality redness does or does not belong to something. What are they differing about in the moral case? Surely, in the same way, over whether the quality rightness does or does not belong to an act. Thus if their difference is to be intelligible there must be a quality of rightness to which both alike are referring.
This conclusion, says Mr Toulmin, does not follow. There is another way of interpreting such differences. ‘All that two people need (and all that they have) to contradict one another about in the case of ethical predicates are the reasons for doing this rather than that or the other.’20 When a person differs from another as to whether an action should be done, what he is saying is that there are better reasons on his side. Mr Toulmin apparently holds that the meaning of ‘x is right’, so far as it has a cognitive meaning at all, is merely that there are valid reasons for doing x. The judgment seems to be at once expressive and assertive—to express approval—as yet undefined—and to assert that this approval has adequate reasons behind it. Mr Toulmin's third argument against objectivism is that this alternative account of moral judgment is a more plausible account than the one it offers us. We must therefore now ask how plausible it is.
20. If to say that something is right is (on the cognitive side) to say that there are valid reasons for it, we may well ask what is meant by a valid reason. How are we to know one when we see it? Mr Toulmin answers that there are only two kinds of valid argument for a moral judgment, each appropriate to its own set of circumstances. Suppose we are choosing between two courses of action one of which accords with the accepted rules of the community, while the other conflicts with them. In this case we should be offering a valid argument for the first if we showed that it did so accord while the second did not. Suppose on the other hand that an accepted rule itself is called in question, or that two accepted rules are found to conflict. It is clear that some further mode of validation must be found, for the question now is how we are to validate an accepted rule itself. Here, Mr Toulmin says, the only valid argument is one that appeals to ‘the general requirement that preventable suffering shall be avoided’.21 It is only ‘a natural and familiar extension’ of this requirement to say that we should aim also at the greatest general happiness.22
In Mr Toulmin's discussion, these two ways of validating judgments sometimes appear as methods employed at different levels of civilization, the appeal to established rule being the natural method in fixed and primitive communities, and the appeal to general happiness superseding it as societies become more ‘open’ and self-critical. But they are also discussed as if they were independent and equally valid types of justification. This seems to me unwarranted. The head-hunter would no doubt justify his headhunting by saying that it was the accepted practice in his community; the Moslem would justify his polygamy and the Christian his condemnation of it by the same sort of argument; and if the argument is valid at all, it is presumably valid universally. ‘Primitive ethics’, says Mr Toulmin, ‘is “deontological”,’ but it is clear that he does not mean that the rules it lays down are binding because self-evident, as Ross maintains, for many of these are admittedly mere taboos. And to say that an action may be validated as right by the appeal to an accepted taboo is most unplausible. Such an appeal can provide at best only a provisional justification which must be reviewed by a superior court before it can be granted real legitimacy.
21. Mr Toulmin virtually admits this. He goes so far as to say that it is the avoidance of needless misery that gives the notions of obligation, right, justice, and duty their meaning.23 The only conclusive validation for any course of action lies in showing that it leads to this result, or to its ‘extension’, the greatest happiness. The new interpretation of ‘x is right’ which is to supersede the confusion of objectivism is thus that it expresses an attitude (yet to be defined) combined with the assertion that it can be validly defended, which assertion means in turn that the action will conduce to happiness or the avoidance of misery. This latter part of the doctrine seems a very tame result of so much analysis and so much defiance of traditional ethics. It is essentially a return to Sidgwick. Of course Sidgwick's defence of utilitarianism is a powerful one, far more so than anything here attempted; but if it is to be made plausible at this time of day, it must be defended against many objections that are now too familiar to need repeating. Why, for example, are we not offering a valid argument for a course of action if we show that it would increase human knowledge or understanding, even if it left the amount of happiness unchanged?
22. So far, then, as Mr Toulmin would reduce ‘x is right’ to ‘good reasons of a utilitarian kind can be offered for x’, his account is both incomplete and something of an anachronism. We must now ask what it is for which these reasons are offered. One can hardly say, as Mr Toulmin does seem to be saying at times, that ‘x is right’ means nothing but ‘there are valid reasons for “x is right”’; we must know the meaning of that ‘x is right’ which the reasons are offered for.
It is difficult to discover what Mr Toulmin's answer is. Indeed he seems to give two different answers, both of which, unhappily, involve his theory in incoherence. On the one hand he insists that the judgment ‘x is right’ is really a judgment; the emotivists and imperativists are mistaken in making it the expression of an attitude merely; it is a statement about an object. What does it assert? It asserts that the act is worthy to be approved. In its objective force this statement is on a par with the statement that a conclusion is logically valid, which means not that someone believes it, but that it is worthy of belief.24 This contention calls for two comments. First, as Mr Broad has pointed out in an admirable review,25 ‘there is nothing particularly new or startling in this aspect of the theory. It has been very fully developed by, e.g., Sir W. D. Ross and by Dr Ewing’—and, he might have added, by himself. But, secondly, this view is inconsistent with Mr Toulmin's principal thesis, namely that ‘x is right’, on its cognitive side, is an assertion that there are valid reasons for doing x. He has insisted that when two people differ in opinion as to whether x is right, their difference is exclusively over whether there are valid reasons for doing it. This cannot be true if the judgment is to be interpreted in the present way. For two people might agree that x is right in the sense ‘worthy to be approved’ who disagreed flatly as to the reasons that would validate it, and even as to whether there were such reasons; and they might disagree over whether x was right while agreeing as to the kind of reasons that would justify it. Indeed it is absurd to talk of reasons at all unless there is a difference between these reasons and the conclusions for which they are offered. Mr Toulmin speaks as if the conclusion collapsed into the reasons for it.
But then on the other hand this whole view that when people differ on moral issues they are differing as to reasons seems to have been offered precisely to save the emotivist-imperativist thesis that moral judgments are not in the strict sense judgments at all. For if people could and did contradict each other over ‘x is right’ when used with its traditional meaning, there would be no occasion or warrant for this new doctrine that such contradiction concerned only their reasons. And if ‘x is right’ is now to be taken as the expression of an attitude, either of feeling or of command, then ‘the place of reason in ethics’ remains thoroughly obscure. For you cannot validate an emotion by reasons as you can a proposition; and as we have seen, you cannot prove or disprove a command. Mr Toulmin, like Mr Hare, is eager to show, as against irrationalists in ethics, that some moral judgments can be validated by reason, and others invalidated. But Wittgenstein had announced in passing that ‘there are no ethical propositions’,26 and Mr Toulmin feels constrained to accept this in the sense of holding that ‘ethical utterances’ are not descriptive, nor therefore true or false in the sense of corresponding to anything in the actual world.27 But the whole force of his insistence on ‘validating’ moral judgment through reasons was drawn from the underlying assumption that these reasons were relevant, and that ethical conclusions could be established or overthrown by them. He argued against the emotivists that they must be wrong because these reasons were constantly given, and felt to be relevant, though they could not be relevant to an emotion. If this argument is valid at all, it is equally valid against the position he is here urging.
23. We have selected a very few of the linguistic moralists for critical attention, and there are others that we might have taken equally well. They are obviously an able group. But for that reason we may perhaps be permitted a gesture of protest against what we conceive Wittgenstein to have done to their generation. In reading such men as Sidgwick, Moore, and Ross, we might and often did disagree with them, but we knew where they stood on fundamentals. They had an eye for the essential points, and argued about them with a force and lucidity that left us with an exhilarated feeling of getting clear on major issues. In reading the linguistic analysts, one has a feeling of being led over interminable flats to no firm conclusion and in no fixed direction. It is as if the very idea that there were any fundamental issues or crucial points had been repudiated. Every statement, we are told, has its own logic. Hence to talk about ethical statements generally is rash; and even to talk about ‘x is right’ or ‘x is good’ is misleading since it assumes that people mean by these words the same thing. It is safer to say that statement A is very like B, and B very like C, and C very like D, and D again rather like A, though in somewhat dissimilar respects from those in which A is like B; but to crystallize these likenesses into definite class statements is dogmatic and misleading. Mr Urmson thinks that ‘x is good’ is like a process of grading, but on such essential questions as what is meant by affixing the label and whether the criteria for affixing it can be rationally chosen, we found it hard to get clear guidance. Mr Hare is convinced that value judgments are like commands, and also that they are like judgments. At the all-important top level, the likeness to judgment seems to fail, and at lower levels, for example, in value judgments of the past, the resemblance to commands likewise fails. It is significant that Mr Hare should be less disturbed by the first dissimilarity than by the second. Mr Toulmin argues elaborately that moral judgments are like expressions of feeling, that they are like reports of feeling, and that they are like statements of fact, but also that in manifold subtle ways they are unlike all of these; and his ultimate view seems to be an unstable alliance between something like emotivism for the meaning of ‘x is right’ and utilitarianism as the way of establishing it.
The school has not yet touched bottom. Linguistic moralists still talk about moral judgments as such, presumably by dint of ancient custom. But if the later Wittgenstein is right, this is rash generalization. The ‘logic’ of ought-statements is so different from that of right-statements, and both from that of good-statements that it is shocking to lump them together; and the logic of ‘happiness is good’—a statement itself shockingly alien to common usage—is so different from that of ‘wine is good’ or ‘we had a good vacation’ that it is shockingly naive to assume a common denominator. The natural terminus of the movement is a situation in which its slogan is taken seriously and every statement (or sentence) is assigned a logic of its own. The assumption of older moralists that one can profitably study the meaning of good or the nature of obligation as such will have been exploded. What had been supposed to be major identities will have dissipated themselves in innumerable minute differences, and ethics will have become a level plain extending as far as the eye can see and littered, it is to be feared, with dry bones. If that consummation is not reached, it will be because the direction of the movement is discerned, and good sense calls a halt to it.
24. The revolt, when it comes, will have small patience with two tenets of linguistic ethics. One is that meta-ethics has no ethical implications and may be discussed in a logical vacuum, antiseptic to moral commitments. This doctrine, however popular, is untrue. To adopt certain meanings for ‘right’ and ‘good’ as the valid ones is also to elect a way of life. The person who believes that ‘right’ means ‘productive of the greatest pleasure’, as Mill did, will not only think differently; he will, if sincere, act differently, from the person who believes, as Paley did’, that it means ‘concordant with the will of the Christian God’, or, as Kant did, that it means ‘consistent in principle’ or, as Nietzsche did, that it means ‘conducive to the interests of me and my class’. To be sure, we have said more than once that the correctness of an analysis is not to be judged by the desirability of its consequences. But in dealing with fundamental notions like right and good, correct analysis itself requires that one should see the implications of one's theory for men's actual valuations. If defining ‘right’ in a certain way would commit one to calling right a range of acts that are generally condemned as wrong, that is evidence against the correctness of the analysis. Meta-ethics cannot be pursued responsibly in divorce from ethics.
25. There is another favourite of the linguistic moralists that is also destined, one suspects, for a short life. This is the attempt to resolve ethical discussion into a discussion about language. The attempt has not yet been carried remorselessly through, and if it were, ethics would have few readers. What students of ethics have always discussed, and what they are still chiefly interested in, is types of conduct and character, qualities of motive and attitude, the tests for right and wrong, the ends men have proposed to themselves, the comparative values of these ends, and the nature of value itself. These are of course not primarily questions about the usage of words. Like every problem whatever, they may be discussed, not wholly unprofitably, as if they were questions of language. Just as one can say that Hume, in reflecting on how the billiard balls acted on each other, was really concerned with the word ‘cause’, so one can say that Mill, in discussing the test of right action, was really concerned with the word ‘right’. But this is hardly even a half truth. For (a) it was his concept of right, his special theory about what in an action made it worthy of approval, that alone gave his words the slightest importance; (b) he could have set out the same theory in another language, or in other words of the same language; (c) if the actual usage of words is taken as a guide in these or other matters, it is likely to prove inadequate, erroneous, and contradictory,28 as linguistic moralists have had to admit; witness Mr Ayer's concession that emotivism conflicts with ordinary usage and his ‘recommendation’ of an unorthodox usage none the less; (d) in conceptual analysis, language must follow the distinctions of thought, not thought the set of labels that language happens to provide. If these things are true, then (e) to call the study of ethical theory a study of the language of morals is tiresomely misleading. It suggests that the study is far more trivial than in fact it is, and more so than the linguistic moralists themselves in practice take it to be.
So far as one can see, these moralists have achieved no clear view jointly or singly, as to the meaning of ‘good’, ‘right’, or ‘ought’, or as to what moral judgment expresses, or how it may be validated. Acute as they are in dissecting and differentiating the meanings of words, their writing on ethics has tended to be academic, unimaginative, and flat. Their somewhat disdainful references to emotivism, like their similar references to logical positivism, convey a deceptive impression of the distance they have travelled from the doctrines prevalent in the thirties. They are still enmeshed by these doctrines; and they are unlikely to escape from them while they maintain their parochial reverence for the gospel according to Wittgenstein. Nevertheless in one respect we owe them much. They loosened the moorings that tied ethics for a time to emotivism. None of those we have studied has broken with it cleanly; none of them has conceded that a judgment of good, or right, or ought, is a judgment in the straightforward sense at all. But they are once for all beyond the doctrine that moral judgments are mere explosions of feeling. They admit that in some sense such judgments may be more or less rational, and that they may in some sense be rationally defended. Thus their work is more than a retreat from emotivism. It is an able attempt, however abortive, to find a place for reason in ethics.