From time to time we have had occasion to mention the close relation between belief and action, and we have seen more than once that the epistemology of belief overlaps with moral philosophy, or rather perhaps with practical philosophy, the philosophy of conduct or intelligent behaviour, of which moral philosophy is a part. (It is not obvious that everything we do must be either right or wrong. And even if it were, we could still consider some actions from a purely prudential point of view, in so far as they are directed to the fulfilment of the agent's own wishes, whether long-term wishes or short-term ones. To put it in a Kantian way, hypothetical imperatives play a very considerable part in our conduct, as well as categorical ones.)
It is not very surprising that there should be this close connection between practical philosophy and the epistemology of belief. After all, intelligent action consists very largely in acting ‘under the guidance of’ or ‘in the light of’ one's beliefs.
But there is one very obvious question which we have not yet considered. What is the nature of moral beliefs themselves? Perhaps this question should be raised in a wider form: what is the nature of valuational beliefs in general, of which moral beliefs are a special case? Not all value concepts are moral ones. We use and understand value words of many different sorts, not only ‘right’ ‘wrong’ ‘ought’ ‘virtue’ ‘vice’, but also ‘charming’ ‘disagreeable’ ‘interesting’ ‘fine’ ‘tragic’ ‘tedious’, and many more. And any of these words may occur in sentences which appear, at least, to express beliefs. ‘I am sure that The Mysteries of Udolpho is a very tedious book.’ ‘You have not read it, but I have; and I think (or my opinion is) that it is quite interesting.’ ‘My opinion is that the view from the top of Skiddaw southwards is one of the finest in England.’ Nevertheless, it would be generally agreed that moral concepts are the most important of all value-concepts; and in this discussion I shall confine myself almost entirely to moral beliefs.
The main question I wish to consider is just this: are there moral beliefs at all? It is perfectly obvious that there are moral sentences, and that these sentences are frequently used both in our public and overt discourse, and also in the private or inner discourse of each one of us, where words publicly spoken or written are replaced by verbal imagery of an auditory or visual or kinaesthetic kind. It is a mistake to suppose, as some philosophers seem to, that an intelligent being is always talking, if talking means ‘talking to someone else’, and that this is the only way in which intelligence, or intelligence at the conceptual level, could possibly manifest itself. This is one of the errors of a collectivistic and sociologically-minded age.
But though we all agree that there are moral sentences, and though we might all agree that there are moral judgements, meaning by this ‘whatever moral sentences are used to express’, there is profound disagreement about the analysis of moral judgements. For more than two centuries, ever since the Moral Sense theory was propounded by Hutcheson, a controversy has been going on about the analysis of moral judgements, and it is still unsettled. But at least it has become a little clearer what the issue is. It can be put quite simply in this way: does the distinction between ‘true’ and ‘false’ apply to moral judgements?
Suppose that a person has risked his life in order to save the life of someone else, and you say ‘his action was right’. Are you saying something which is either true or false? It is true or false that the action was done, that it was done at such and such a time and place, with such and such an intention. But is it true (or false either) that the action was right?
If we agree with you and say ‘Yes, it was right’, do we think that there is a properfy called ‘rightness’ which the action had, as it had the property of being done in Clapham yesterday afternoon and of occupying a period of 4 minutes? Suppose you gave a very detailed description of the action, a detailed description of everything the man did during those 4½ minutes and also of everything that was going on in his mind. Would you add at the end ‘There is another rather interesting fact which I have forgotten to mention. His action was right’? If a statement is true, there is a fact which makes it true, and if it is false there is a fact which makes it false.
It is a fact that the action was done in Clapham yesterday afternoon. This is what makes it true to say that it was done there at that time; and the same fact makes it false to say that it was done somewhere else or at another time, for example in Brighton or yesterday morning. Is it also a fact that the action was right? Are there moral facts? If we think that an utterance such as ‘this action was right’ is true or false, we must think that there are moral facts. If there are, such utterances as ‘this action was right’ express either knowledge or belief. In that case, there are moral beliefs and they are either correct or mistaken, as any other belief is.
Yet the notion of a ‘moral fact’ is rather a strange one. It is not easy to see what kind of a fact it could be, or in what sense it could be a fact at all. The view that Tightness and wrongness are properties which actions have is likewise a strange one. It is still odder, perhaps, to say that obligatoriness is a property which some actions have, and that you are ascribing this property to an action (or rather, to a possible action?) when you tell me that I ought to do it.
It is not very surprising, then, that some philosophers have doubted whether the disjunction ‘either true or false’ applies to moral judgements at all. What is surprising is that the view that they are neither true nor false was only suggested fairly recently. Even Hutcheson and Hume were not wholly clear about it. They could be interpreted as maintaining that there are moral facts, but that they are facts about our own feelings. Hume says, for example, that a virtuous action is one which pleases us after a particular manner. He is certainly denying that there is a property called ‘virtuousness’ which the action itself has. Nevertheless, if someone who says it is virtuous is saying that it pleases him after a particular manner, surely he is saying something true or false about himself? Autobiographical remarks can be true or false. Indeed, he is perhaps saying something true or false about the action too, namely that it has the causal property of exciting a special sort of pleasant feeling in him when he witnesses or recalls or is told about this action. Hutcheson's Moral Sense theory could be interpreted in a similar ‘autobiographical’ way.
But such an interpretation of either writer would not, perhaps, be altogether fair. Perhaps what they wish to say is that when we make a moral judgement about some action, we are expressing our feelings about the action or our attitude towards it, and not stating that we have these feelings or this attitude. It is rather as if we took off our hat to the action or to the person who did it, or saluted him or congratulated him by saying ‘Well done!’ The most appropriate name for this feeling or this experienced attitude is ‘approval’, or ‘disapproval’ when the words we say are ‘it was wrong’ (this would be rather like shaking our fist at the man). This seems to me the most plausible version of the doctrine that moral judgements are neither true nor false. One may call it the Approval Analysis of moral judgements, or the Attitudinarian Analysis.
There is however another version of the doctrine that moral judgements are neither true nor false, and here it is held that they are analogous to imperatives. On this view, when I say to you ‘This is the right thing to do’ it is rather like saying ‘Go on! Do it!’ And if I say ‘It was wrong to do that’ this is rather like saying ‘Don't do that sort of thing again’. I am not telling you that something is the case, as I should be if I were making a statement. It is more like telling you to do something, or not to do it.
But of course we cannot quite say that all moral judgements are commands. If we make the judgement ‘Brutus ought not to have killed Caesar’ or ‘his killing of Caesar was wrong’, it is too late now to tell him not to do it, or even to tell him not to do that sort of thing again. Moreover, if someone in Cambodia is now asking himself whether he shall pay his grocer's bill, we cannot very well tell him to pay it, though we can say that he will be acting rightly if he does pay it.
Nevertheless, it might be maintained that alt these utterances resemble commands much more than they resemble statements, and that we might describe them as injunctions. When we say ‘Brutus ought not to have killed Caesar’ or ‘it was wrong of him to do it’ we might be saying something like this: “Let no one assassinate any of his political opponents, ever. But Brutus did assassinate one of his, when he killed Caesar.’ And similarly: ‘Let anyone who owes money to anyone else pay it. So if there is a man in Cambodia who owes money to his grocer, let him pay it,’ (The word ‘let’ has the force of the Latin gerundive, faciendum.)
This version of the doctrine that moral judgements are neither true nor false may be called the Injunctive Analysis. There are several different varieties of it, and the one I chiefly have in mind is Professor R. M. Hare's in his well-known book The Language of Morals. But it must not be forgotten that the notion of an imperative plays a fundamental part in Kant's moral philosophy. Moreover, the theory that we mean by ‘right actions’ those actions which God commands, and by ‘wrong actions’ those which God forbids, is perhaps almost as old as Theistic religion itself. But though God's commands or prohibitions could not themselves be true or false, it would still be true or false that God has commanded us to do actions of one sort and has forbidden us to do actions of another sort. This theory, then, is irrelevant to our present discussion, though historically it may have helped to suggest the Injunctive Analysis of moral judgements by maintaining that moral judgements are at any rate closely related to commands.
The Approval Analysis
In these two lectures we shall mostly be concerned with the Approval Analysis or Attitudinarian Analysis, as it may also be called. Your lecturer must confess that he has a prejudice in its favour, because it seems to accord with his own moral experience: if indeed he has any, for the writings of some moral philosophers make him suspect that he is not a moral being at all. But perhaps there are some others who feel a kind of spiritual affinity with the eighteenth century British Moralists of the so-called ‘Sentimental’ school; and it may be worth while to try to reformulate the approval analysis in the terminology of our own time, and to consider some implications of it which its original inventors did not sufficiently emphasize.
But before we turn to this task, there is one general remark to be made. It would be misleading to say that according to this type of moral philosophy the function of the word ‘right’ is merely to express an attitude of approval and the function of the word ‘wrong’ is merely to express an attitude of disapproval. For the word ‘merely’ suggests that it does not matter very much what a man approves of, and that the difference between approving and disapproving is not particularly important. It should not be supposed (though perhaps it sometimes has been) that according to theories of this type ‘there is really no difference between right and wrong’ and that those who hold them do not take the moral life seriously.
On the contrary, if it is the function of the word ‘right’ to express approval and of the word ‘wrong’ to express disapproval, they have a very important function indeed, and it still makes a very great difference to our lives and to our relations with our neighbours whether we say of an action ‘it is right’ or ‘it is wrong’. Moreover, the difference which it makes is exactly as great as it is if moral judgements are true or false. On that view, there are moral facts, and a moral judgement is true if it corresponds with a moral fact and false if it does not. But even so, one might still fail to take such facts seriously. One might say ‘Oh well, no doubt it is a fact that this action of mine was wrong. But what of it? Why make such a fuss about it? Why does it matter?’
Whether we take moral distinctions seriously is decided by considering how we live—in our inner lives as well as our overt behaviour—and not by considering which analysis of moral judgements we subscribe to.
‘There Obviously Are Moral Beliefs’
But is it worth while to pay much attention to the doctrine that moral judgements are neither true nor false—either to the ‘approval’ version of it or to any other—if it entails (as it seems to) that there are no moral beliefs? Surely it is perfectly obvious that there are moral beliefs?
Still, we must not forget the distinction between belief ‘that’ and belief ‘in’.2 It is indeed obvious that there are moral beliefs-in. And these are what we are usually enquiring about when we ask what Mr. So-and-So's moral beliefs are. For example, he believes in telling the truth always, no matter what the circumstances are: or he believes in turning the other cheek, whatever the provocation. But when a man expresses a belief ‘in’ a type of action or a rule of conduct, he is expressing a valuational attitude. It need not be an attitude of moral approval. A man might believe in telling the truth always as being the most expedient policy, the one which pays best in the long run, despite some appearances to the contrary. But often a belief-in is, inter alia, an attitude of moral approval, and then a moral belief-in is at least a part of it. The same could be said of belief-in a person, for example the belief which most Englishmen had in Winston Churchill during the years 1940–45. This belief-in was amongst other things an attitude of moral approval, though it was also a trust in his intelligence, resourcefulness, energy and efficiency.
But is it not perfectly obvious that there are moral beliefs, and not just moral beliefs-in, but also moral beliefs-that? Surely it is true, for example, that moral judgements are made with different degrees of confidence? One may think that an action was wrong, without being sure that it was. One may be almost sure that it was wrong, but not absolutely sure; or again one may be absolutely sure, completely convinced, that it was wrong. To put it in another way, it clearly makes sense to ask the question ‘was it wrong?’ This question may be answered with different degrees of confidence, and we express these degrees of confidence in just the same way as the different degrees of confidence we may have when we believe that something is the case. Or again, we may be unable to answer the question; we may remain undecided between the two alternatives ‘it was wrong’ and ‘there was nothing wrong in it’, just as we are when we suspend judgement concerning the alternatives ‘it rained in Glasgow last Sunday’ and ‘there was no rain in Glasgow last Sunday’. The same belief-like expressions are appropriate when we make moral judgements about a person (as opposed to an action). ‘I think he is a kind person and an honest one too.’ To this you may reply ‘I am quite sure he is kind, and I think he is honest too, though I am not absolutely sure about that’.
What are we to make of such utterances, if we accept the view that moral judgements are neither true nor false? There is an analogous case which we might perhaps find useful. I ask a friend where he is going to spend the Easter holiday this year, and he answers ‘I think I shall go to Brighton, but I am not quite sure’.3 Is he expressing an opinion? Is he making a true or false prediction about his whereabouts a month or so hence, though he makes it with something less than complete conviction? Is his utterance comparable with another which he makes ‘I think the Easter weekend will be cold, as it often is’?
Now let us suppose that on Easter Monday I meet him in Yarmouth, and also that it is a fine warm day all over the country. I say to him ‘So your forecast about your whereabouts was false after all. You were mistaken when you said you were going to be at Brighton; you made a mistake about the weather too.’ But obviously when he said ‘I think I shall go to Brighton, but I am not quite sure’ he was not expressing an opinion at all (though he was, when he said he thought the Easter week-end would be cold). This is not because it is impossible to have opinions about one's own future actions. We do have them sometimes. I express such an opinion when I say ‘I think I shall be sending this book to the publishers about three years from now’. This is just a moderately confident prediction or forecast about myself. It may quite well be false. I may easily be dead three years from now; and even though I am still alive at that time, it is perfectly possible that the book will not be ready for the publishers by then.
But the man who said ‘I think I shall go to Brighton, but I am not quite sure’ was not making a moderately confident prediction about his own future, nor was he saying anything true or false. He was expressing an intention, but a provisional and revisable one, and the words ‘I think… but I am not quite sure’ were a way of indicating its provisional and revisable character. Of course, he could also have used these words ‘I think… but I am not quite sure’ if he had been expressing an opinion. But that is because opinions too have a provisional and revisable character.
In a similar way, one might be expressing a provisional and revisable disapproval when one says ‘I think it was wrong to do that, but I am not quite sure’ and a provisional and revisable approval when one says ‘I think it would be right to tell him, but I am not quite sure’. On the other hand, if you say you are quite sure it would be right to tell him—you have no doubt of it at all—we might hold that your attitude is one of firm and settled approval, and does not have the provisional and revisable character which mine has.
Such utterances as ‘I am quite sure he is a kind man, and I think he is an honest man too, but I am not quite sure about that’ might be elucidated in a similar way. One is here expressing two approval-attitudes about the same person (approving of him in two different respects). The first approval is firm and settled, the second is provisional and revisable. We are also able to give a meaning to the question ‘was it right?’ or ‘is he a good man?’, even if we do not think the answer to it is true or false. (The answer to the question ‘Where are you going for the Easter holiday?’ was not true or false either.) We can say that when one asks someone else, or oneself, ‘was that action right?’ or ‘is he a good man?’ the question expresses an undecided or indeterminate moral attitude.4 One has not made up one's mind whether to approve or to disapprove; much as a man might ask another man, or himself ‘Shall I go to Brighton?’ when he has not made up his mind whether to go there or not.
There is one other point which should be mentioned now (I shall say more about it later).5 Approvals and disapprovals have presuppositions, somewhat as beliefs-in have. As we shall see later, we approve or disapprove of an action, or a person, in respect of some characteristic which we believe it or him to have. This belief may be called the presupposition of our approving or disapproving attitude. This belief is a straightforward belief-that, about some matter of physical or psychological fact; and what is believed is true or false, in the straightforward sense in which it is true or false that it rained in Glasgow last Sunday. If the belief is mistaken, awkward questions arise about the approval or disapproval whose presupposition it is. But the point at present is that this belief, like any other belief about matters of fact, may be held with varying degrees of confidence.
If the belief is held with only a moderate degree of confidence and itself has a provisional and revisable character, the approving or disapproving attitude, whose presupposition it is, will also have a provisional and revisable character, at least if one is a reasonable person. If I think, but am not absolutely sure, that you intentionally caused great pain to another human being, I do disapprove of your action, but only in a provisional and revisable way. I say ‘I think it was wrong, but I am not absolutely sure’; but I say so because I think, without being absolutely sure, that it was intentional, and did cause great pain. In this case, the disapproval is provisional and revisable because the factual belief which is its presupposition is provisional and revisable, an opinion and not a conviction.
Nevertheless, even though one is absolutely sure about the physical and psychological facts—even though one knows what they are in a fairly strict sense of the word ‘know’—one's approving or disapproving attitude may still be provisional and revisable. This may easily happen when the action we are making a moral judgement about—whether past or proposed, our own action or another's—is one in which the agent is confronted with a conflict of duties. An obvious example is the action of telling a lie when the agent cannot otherwise keep a secret with which he has been entrusted. This action is wrong in one respect but right in another. This amounts to saying, on the theory we are discussing, that we approve of it in respect of one characteristic which it has, and disapprove of it in respect of another. The question is, whether to approve of it ‘on the whole’ or to disapprove of it ‘on the whole’. Whichever of these two attitudes we eventually adopt, it is likely to be a provisional and revisable one, expressed by saying ‘I think it is on the whole right but I am by no means sure’ or ‘I think it is on the whole wrong but I am by no means sure’, even though in this case we are in no doubt at all about the physical or psychological facts.
So far, we conclude that although we do use and understand such expressions as ‘I think it was wrong’, ‘I am sure it is right’, this does not by itself settle the question whether there are moral beliefs in the belief-that sense. Let us now try another approach to that question.
‘Ethics Without Propositions’
There is a well known paper by Professor W. H. F. Barnes called ‘Ethics without propositions’.6 Possibly the type of ethical theory he has in mind should rather be described as ‘Ethics without assertions’. Indeed, he himself suggests so. It cannot be literally without propositions, since we have to entertain propositions e.g. ‘an act of such and such a sort was done by So-and-So’ in order to approve or disapprove. Sometimes we not only entertain the proposition but also believe it; sometimes we know it to be true. I may believe on good, though not conclusive, evidence that A told a lie to B, and disapprove of this action which I believe him to have done. Again, I may approve of some action which I actually witnessed yesterday and now clearly recall, or disapprove of one which I did a minute or two ago myself. Sometimes, however, we just entertain the proposition in a neutral way, without either belief or disbelief; we may approve or disapprove of a proposed action which has not yet been done and perhaps never will be. Indeed, if we disapprove, and express our disapproval, this may be sufficient to prevent its being done.
Could we say, then, that such an ethical theory is at any rate without ethical propositions, though certainly not without propositions concerning the ‘natural’ (non-ethical) characteristics of actions? Not quite. For surely the propositions which are the objects of our approvals and disapprovals may themselves contain ethical concepts, and frequently do? ‘John has not done what he promised to do’ is an obvious example. To make a promise to do something is to lay a moral obligation on oneself to do it. The concept of a promise, then, is very far from being an extra-moral or morally-neutral concept. What should we have to do to turn this proposition into a purely factual one, describing the situation in entirely non-ethical terms? The question is not easy to answer, but perhaps we could say something like this: ‘John said to Tom the words “I promise to be back in half an hour”, well understanding what he said: and then he went away; and more than half an hour has elapsed, and he has not come back although it was in his power to do so.’
Again ‘A has taken B's watch’ might appear to be a completely non-ethical proposition. But the little apostrophe s in the phrase ‘B's watch’ conveys the idea of property, and this again is certainly not an extra-moral or morally-neutral concept. If something is a man's property, he has a right, within limits, to use it as he pleases. What should we mean by saying that he has this right, if we accepted the Approval Analysis? We should be expressing approval of his having the power, within limits, to use the object in question as he pleases, and disapproval of any action by any other person which restricts, or deprives him of, this power. It follows, that in saying ‘A has taken B's watch’ we are already expressing disapproval of what A has done: unless we add ‘with B's permission’, for one way in which B can do what he pleases with the watch is to lend it or give it to another person, and then he transfers his own power to use it as he pleases to someone else, either temporarily or permanently. What we shall have to say if we want to make a purely factual assertion is something like this. ‘A removed a certain watch from the place where it previously was, with the result that B involuntarily lost the power, which he had up to that time, of using it within limits as he pleased.’ But would this be adequate, unless we also said ‘and this power was one which we approved of his having’? For B might have had this power de facto for a long time, even though the watch was not his at all.
It is indeed exceedingly difficult to give an ethically-neutral analysis of the phrase ‘B's watch’. Some additional clause is certainly needed. But perhaps it might be suggested that we need not necessarily be expressing approval of B's power of using the watch, within limits, as he pleased. Might it be sufficient to say that such a power is in fact generally approved of? Conceivably we ourselves disapprove of what is called ‘the institution of property’ or take a neutral attitude about it, neither approving nor disapproving.
It is, of course, a matter of fact that approvals and disapprovals do exist, and it is an important fact too, which often has to be mentioned when one is trying to describe a social situation. A historian, for example, often has to mention the moral judgements which were made by the people he is describing, since much of their conduct would otherwise be unintelligible. But in so doing, he need not make any moral judgements himself. And certainly he need not himself approve of what they approved of (e.g. treating slaves as one pleases) or disapprove of what they disapproved of (e.g. refraining from taking vengeance for personal insults). What is necessary is that he should be capable of making moral judgements himself, and know from personal experience what it is like to approve of one type of conduct and disapprove of another. In short, the historian himself must be a moral being. Otherwise, he will not understand what he is talking about, and the moral judgements of the persons whom he is describing will be for him no more than curious noises which they made, inexplicably followed by wars, assassinations, rebellions and other catastrophic occurrences.
Perhaps, however, it is not very surprising that ethical words (those expressing approvals and disapprovals) should be so deeply rooted in our discourse that it is very difficult to get rid of them, and therefore very difficult to describe human actions in completely non-ethical terms. We do happen to be moral beings, after all. Or rather, we do not just happen to be. For if we were not, we should not be persons, though we might still be intelligent creatures with a human shape.
Some Logical Properties of Attitudes
The examples we have been considering ‘John has not done what he promised to do’ and ‘A has taken B's watch’ could also be used to illustrate another point, and a more important one. They show that even if moral judgements are neither true nor false, they may still have logical properties. Even in ‘Ethics without assertions’ moral judgements may still have an analytic character, as they may in objectivist Ethics which holds that they are assertions. On either theory, it is wrong by definition to break a promise when it is in one's power to keep it; and on either theory it is wrong by definition to take someone else's property without his permission. No doubt, in both these examples, it might sometimes be the lesser of two wrongs. Taking someone else's property without his permission, or breaking a promise which we had made, might be the only way of saving a man's life in an emergency. Nevertheless, there is still something wrong in both these actions, even though it is ‘on the whole’ or ‘on balance’ right to do them in some exceptional circumstances. That there is something wrong in both of them is a matter of definition. It follows from the definition of the term ‘property’ in the one case, and of the term ‘promise’ in the other. This is still so, if we say that ‘this is X's property’ just expresses our approval of his having the power, within limits, of using the object as he pleases and our disapproval of anyone who makes it impossible for him to use it thus; and likewise if we say that ‘these words consitute a promise’ just expresses our approval of the speaker's doing the action which the words describe (e.g. coming back half an hour from now) and our disapproval of his failure to do this action when it was in his power to do it.
But though a man who says ‘it is wrong to break a promise’ has expressed an analytic disapproval, as we may call it, he has nevertheless expressed something, and has conveyed something to his hearers. He has conveyed to us that he does have a moral attitude towards anyone who says ‘I promise to do such and such an action’ and then fails to do it, even though he still has the power to do it. Moreover, he has conveyed to us what that attitude is, and has shown us that he attaches the same meaning, or force, to the word ‘promise’ as we do.
It is possible, however, that an expression of approval might convey nothing. It might be completely empty or ‘vacuous’, and ‘vacuity’ too is an important logical property. Suppose I visit a strange country and wish to find out what the moral attitudes of the inhabitants are. I go to their wisest man and question him about it. He replies ‘There are no moral disagreements here, as I am told there are in your part of the world. Here we are quite unanimous in our moral judgements. We all approve of right actions and of good men; and we all disapprove of wrong actions and of bad men.’ Clearly he has told me nothing. He might as well have said ‘All of us here approve of some actions and some men, and all of us disapprove of other actions and of other men’. The most I could learn from what he has said is that these people do have moral attitudes of some sort; and I was sure of this already.
It is worth while to notice that other sorts of ‘non-assertive’ utterances can be vacuous in a similar way. ‘Do what you please’ is a vacuous command. Though it has the verbal form of a command, there is nothing which it tells us to do, at any rate if we take it to mean fac quod tibi placet. (If we took it to mean ‘Do what gives you pleasure’ it could conceivably be disobeyed, and would therefore not be vacuous.) Again, suppose someone comes to me for advice and I say to him ‘Just do what you think best’. My advice is empty, because it gives him no guidance at all about what he is to do. It is in fact a polite way of refusing to give him what he asked for.
Self-contradictoriness is another important logical property. If moral judgements are true or false it is analytically false, self-contradictory, to say ‘there is nothing wrong in breaking a promise’. It cannot be analytically false to say this, if the attitudinarian moralists are right, because it cannot be false at all. Nevertheless, anyone who says it is still making an utterance which is in some sense self-stultifying or absurd. It is not open to us to reply ‘Oh no, that cannot possibly be true’. The speaker never claimed that it was, if the approval analysis of moral judgements is correct. But it is open to us to reply ‘Oh no, we cannot possibly agree with that’, because there is an obvious impropriety in the way he has expressed himself. And the impropriety is a logical one. By saying ‘there is nothing wrong in’ he has expressed absence of disapproval; and then, by completing his sentence with the phrase ‘the breaking of a promise’ he has expressed disapproval, since we mean by ‘a promise’ a form of words such that we disapprove of the speaker's subsequent failure to do the action described in them, if it is still in his power at that time to do the action described.
Logical Properties, Continued: Generality
Other logical properties come to light when we notice that it makes sense to ask why one approves (or disapproves) of such and such an action or of such and such a person. We approve of what John did, because he had promised to do it, and we approve of any action which is the keeping of a promise, or in so far as it is. I approve of you for giving a lift to a lame man and going ten miles out of your way to take him where he wanted to go, because I approve of anyone who takes some trouble to help anyone else who is in need of help. We disapprove of torturing suspected persons in order to get information from them, because it is cruel, and we disapprove of any cruel action even when it is done in order to achieve a result which in itself we approve of, such as the more efficient detection of criminals.
We see from these examples that there can be general approvals and disapprovals. We can and do approve not only of this person or this action, but of a class of actions or persons; and the same applies to disapproval.
It may be noted that many other human attitudes (and the verbal expressions of them) can have this same property of generality: for example, liking and disliking, admiring, being afraid of, even wishing. One may like cats: not just this cat or that one, but cats in general or cats as such. One may be afraid of Alsatian dogs, not just of the particular one which lives at the house round the corner, but of Alsatian dogs as such. These attitudes are relatively permanent ones. But even a very temporary one, such as wishing for a cup of tea now, may still have something general about it. Within quite wide limits any cup of tea will do. I am quite willing to drink this one in the Refreshment Room at the railway station, because it is at any rate a cup of tea.
Indeed, it may seem plausible to suggest that all the attitudes we have are in some respect or degree general. But this is not true of all the attitudes we can have towards other persons. There is one important exception, the attitude which may be called unconditional love. According to Christian ethics it is the best of all possible attitudes. So anyone who is discussing moral questions should say at least a few words about it, though I myself say them with fear and trembling.
It might be described as a non-general pro-attitude; and just for that reason, there is something paradoxical and mysterious about it, though it does exist, or at least approximations to it do. It differs from what one might call ‘ordinary liking’. For in ordinary liking there is an answer to the question ‘What is there about So and So that you like?’ or ‘in respect of what characteristic of his do you like him?’ But here there is no answer. (You would still be ‘for’ this other person, no matter what he does and no matter what changes he undergoes, whether for the worse or for the better.) Or if there is an answer, it will have to be ‘in respect of his haecceitas’. You are ‘for him’ just as being the individual that he is: just this person. And the notion of haecceitas is a very mysterious one indeed.7
But so far as one can see, nothing like this is true of approval and disapproval. Unconditional love is not itself a moral attitude. It is indeed the object of a moral attitude: it is something of which we approve very highly. But it is not itself a kind of approval. Indeed, it is quite compatible with disapproval. We may suppose that St Paul loved his Corinthian converts in this unconditional way, but he certainly disapproved of some of the things they did. Perhaps the most appropriate thing to say of unconditional love is that it transcends the sphere of morality altogether. For when we approve of some particular action, we do approve of it as being an instance of some type or kind of action; and when we approve of a particular person, we approve of him in respect of some characteristic or conjunction of characteristics which he has (usually dispositional ones); and these are general characteristics which other persons may have, such as courage or benevolence or trustworthiness. The same applies to disapproval. Approval and disapproval do have the logical property of generality, as many of our other attitudes have.
An Analogue of Deductive Reasoning
We may now notice another logical property which approvals and disapprovals have. It is closely connected with this property of generality. Something analogous to deductive reasoning is possible in the sphere of approval and disapproval. Perhaps we may not wish to call it deductive reasoning, since expressions of approval of disapproval are neither true nor false. But at any rate we can speak of logical derivation here, and it is at any rate analogous to deductive reasoning of the syllogistic sort.
That is what makes it possible to answer the question ‘Why do you approve of this particular action?’ or ‘Why do you approve of this particular person?’ (and also ‘Why do you disapprove?’). We have seen already that such questions can often be asked and answered. We answer, first, by expressing approval of all actions of a certain kind or sort, and secondly by expressing our belief that this particular action was an instance of that kind or sort. We approve of all actions in which one person helps another person who is in difficulties,8 and we believe that William did an action of this kind at 2.15 p.m. this afternoon, when he helped a blind man to cross the High Street. So we approve of the particular action which he did at 2.15 p.m. this afternoon. The same applies to our moral attitudes to a particular person. We disapprove of Archibald because he is unreliable; he hardly ever does what he has undertaken to do. We disapprove of all persons who seldom do what they have undertaken to do, and we believe that Archibald is a person of that sort.
This derivation-procedure is analogous to the syllogistic deduction ‘All whales are mammals, and this creature is a whale; so this creature is a mammal’. The analogy is not, of course, complete. According to the theory we are considering, the major, premiss in our moral derivation-procedure is neither true nor false. It consists just in expressing an attitude to a type or sort of actions (or persons). The conclusion too is neither true nor false. Of the three distinguishable stages in our derivation-procedure—major premiss, minor premiss, and conclusion—it is only the minor premiss which is true or false. It is either true or false that William helped a blind man to cross the High Street at 2.15 p.m.; and it is either true or false that Archibald seldom does what he has undertaken to do.
Nevertheless, this derivation-procedure does resemble deductive reasoning in a very striking way, and it seems quite proper to apply the word ‘logical’ to it. In the example about helping the blind man, we are committed to approving of this particular action. And the commitment is a logical one. We cannot consistently refuse to approve of this particualr action, if we approve of all actions in which one person helps another person who is in trouble, and believe that this particular action was an action of that kind.
The derivation-procedure which we have just considered depends on the relation between a type or sort on the one hand, and a particular instance of that type or sort on the other. This relation could also be described as the relation of class-membership. But there is another logical derivation-procedure which depends upon the relation of class-inclusion, i.e. the relation between a wider class and a narrower class included in it. This relation too enables us to express derivative or consequential approvals and disapprovals, and to answer the question ‘Why do you approve (or disapprove)?’
For example, a celebrated Oxford philosopher, one of my own teachers, once expressed his strong disapproval of pictorial advertisements. When asked why he disapproved of them, he answered that such advertisements were ways of telling a lie without making a statement. The poster we see at the railway station advertising the seaside town of Snoring juxta Mare does not actually say ‘the sun always shines there’, but the picture conveys that impression and is intended to convey it.
We may analyse the philosopher's derivation-procedure as follows. In his major premiss he expresses disapproval of all voluntarily deceitful behaviour; in his minor premiss he expresses his belief that all displaying of pictorial advertisements is voluntarily deceitful behaviour; and in his conclusion he expresses his disapproval of all displaying of pictorial advertisements.
Of course, we may question his minor premiss, which expresses a straightforward ‘belief that’, and is either true or false. We may doubt whether such advertisements are intended to deceive, and also whether they do in fact deceive anyone. But we cannot object to the derivation-procedure which he used. It is a perfectly valid (logically-proper) way of explaining to us why he disapproved of all pictorial advertisements.
Underivable Approvals and Disapprovals
It could fairly be said, I think, that in using either of these derivation-methods we are justifying our approvals and our disapprovals, whether of persons or of actions, as we justify our assertions and our denials by deductive arguments: and this, despite the radical difference there is between assertions and denials, which are true or false, and approvals or disapprovals, which are neither.
Finally, we may notice that in both cases alike the process of justification must stop somewhere; and this is another important logical feature which they have in common. If you ask me why I approve of charitable actions as such, I can only answer ‘I just do approve of them, that's all. And surely you do too?’ If it is said that this approval of mine is analytic, on the ground that the word ‘charitable’ already expresses approval, I have to reformulate my answer, but I still cannot justify it. ‘I just do approve of anyone who loves his neighbour and tries to promote his neighbour's happiness or decrease his neighbour's misery. And surely you do too?’ Suppose you disagree and say ‘such people are silly and sentimental’, thereby expressing disapproval; or perhaps you say ‘people are welcome to behave in that way if they wish, of course, but there is nothing either good or bad about it’ (expressing a neutral attitude, neither approval nor disapproval). Then there is nothing more that I can do to justify my own attitude. Nor would there be anything more I could do, if I were asked why I disapprove of causing pain to other sentient beings merely for the sake of one's own pleasure. Here again, I could only say ‘I just do disapprove of it, that's all; and surely you do too?’
But of course the process of justifying an assertion or denial by deductive argument has to stop somewhere too. At some stage or other, we encounter indemonstrable premisses, as we encounter ‘underivable’ approvals or disapprovals. We certainly encounter such indemonstrable premisses if we hold an ‘objectivist’ view of ethics and maintain that moral judgements are, affter all, true or false. Suppose someone says that causing pain to other sentient beings merely for the sake of one's own pleasure is something which has an objective property of wrongness, regardless of anyone's approval or disapproval, and we ask him to justify what he has said. He can only reply ‘I just see (or ‘I just know intuitively’) that it is so. And surely you can see it for yourself?’ There still have to be what one might call ‘ultimate’ or ‘underivable’ moral judgements, whether we accept or reject the view that moral judgements are true or false.
Our conclusion so far is that ‘ethics without propositions’ (or ‘without assertions’ if we prefer) is certainly not ethics without logic. I have been speaking throughout about the approval analysis of moral judgements; but the same is true of the injunctive analysis. There is a ‘logic of imperatives’, as Professor R. M. Hare has shown. It would seem that there is a logic of approvals too; at any rate, we can say that approvals and disapprovals have logical properties. The one with which we have so far been most concerned is generality. We have seen that some approvals and disapprovals are general ones. Could it be said that all are? It could be said, at any rate, that all approvals and disapprovals have something general about them, even when what is approved of or disapproved of is a particular action or a particular person. They always have at least ‘implicit’ generality.
This is because we approve (or disapprove) of a particular entity as having such and such a characteristic, or in respect of some characteristic which it has, or in that it has some characteristic; and the same applies to disapproval. It is conceivable (and in many cases it is true) that other entities also have this same characteristic. I approve of your sending a cheque for £5 38. 2d. to your bookseller. This is a particular action, done by a particular person, on a particular day. But I approve of it as being the payment of a debt. And this characteristic which it has is one which many other actions may have, actions done by other persons at other times and places. I approve of A as being a kind and friendly person and of B as being an honest one, and these are characteristics which other persons may have. I disapprove of myself as being afraid of Alsatian dogs, and afraid of many other things too which no man whom I approve would be afraid of. And this again is a characteristic which other persons may share.
This is the explanation of something which might otherwise puzzle us, the fact that we can both approve and disapprove of the same action, or person, at the same time. I approve of your action as being the payment of a debt; I disapprove of it in that it is a very belated payment, since you received the bill two and a half years ago. I approve of your making that remark to Robert just now, in that you were telling him the truth; I also disapprove of your making it, in respect of your unkind or discourteous way of putting it. We approve of King Henry VIII for his courage and his patriotism, but disapprove of him for his cruelty.
It would seem, then, that even in a particular approval or disapproval, where we approve (or disapprove) of just this action, or this person, there is something which might be called ‘implicit generality’. In approving of this particular action in respect of the characteristic C which it has, we are implicitly approving of any other action which has the characteristic C.
Approvals ‘Presuppose’ Beliefs about Matters of Fact
We must notice, however, that the characteristic in respect of which we approve of something are not merely characteristics which it does in fact possess. We must know or believe that it does possess them. It may be a fact that hundreds of Eskimos are at this moment engaged in telling the truth to other Eskimos, but it is impossible for me to approve of what they are doing, since I am completely unaware of it. Even if I were on the spot and actually heard one Eskimo telling the truth to another, it would not be in my power either to approve or to disapprove of what he was doing, because I should not be able to understand what he was saying. To approve or to disapprove of something ‘as having’ the characterisitc C, one must either believe or know that it does have the characteristic C; and this is true whether the something is a person or an action. Moreover, in actual fact the entity in question need not have the characteristic C at all. It is enough that I believe that it has, whether on good evidence or not, whether correctly or incorrectly. And this belief, of course, is not itself a moral belief, nor is it a belief ‘in’. It is a straightforward belief ‘that’ concerning some matter of physical or psychological fact.
If this belief is incorrect, it does not follow that the moral judgement I make is false. On the view we are discussing, it cannot be either false or true; what I am doing, when I make it, is just to approve, or to disapprove. Nevertheless, there is clearly some sense in which it is a mistake to approve, or disapprove, of X, if X does not really have the characteristic C in respect of which we approve or disapprove. It may be a very grave mistake too. It is no light matter to disapprove of a man for something which he did not do. Or if the word ‘mistake’ be thought too strong (on the ground that it suggests that there are ‘moral facts’ after all) at any rate we can say that such a disapproval is misguided or misplaced.
The same point could be put in another way by using the word ‘presupposition’ which I used earlier: although moral judgements cannot (on this view) be true or false, they nevertheless have ‘presuppositions’9 which are true or false. The same could be said of other ‘non-assertive’ attitudes, such as wishing and hoping, to which the distinction between truth and falsehood does not apply. If someone wishes to visit Utopia, or hopes to live there some day, it is a vain wish or a vain hope, because there is no such place. We are under a misapprehension if we cherish such a wish or such a hope; it has a false presupposition.10
Moreover, an approval or disapproval may be questionable, or doubtfully acceptable, because it has a questionable presupposition, even though it itself is neither true nor false. The fourth century saint, St Martin of Tours, cut his cloak in half and gave one half to a beggar. He was a serving Roman soldier at the time, wearing his armour. (At least, that is how he is commonly represented, and we will assume that the representation is correct.) It was his military cloak which he cut—and then gave away half of it—and he cut it with his sword which was also part of his military equipment. We approve of his action highly, in that it was an act of Christian charity. But perhaps we also disapprove of it, in that he was bisecting what was not his to bisect, and giving away what was not his to give. The cloak, we say, was not his property, after all. It was part of his uniform, like a British soldier's greatcoat, and it was the Emperor's property, as the British soldier's greatcoat is the property of the Queen. St Martin has no right to do as he liked with it. The sword was not his property either, any more than the British soldier's bayonet is. He was not at liberty to use it just as he pleased, and certainly he was not at liberty to use it for cutting a piece of the Emperor's property in half.
But of course the disapproving part of our two-faced attitude is highly questionable, even though the approving part is not. The presupposition which our disapproval has—the belief that in the fourth century Roman Empire a Roman soldier's military uniform and equipment was the Emperor's property and not the soldier's own—is highly dubious; and it is even dubious whether the concept of ‘a uniform’ existed at that time at all.
In this lecture I have been trying to show that the difference between an ‘attitudinarian’ analysis of moral judgements and an ‘objectivist’ analysis, in which they are held to be true of false, is not quite so clear-cut as it looks. Whether we accept the one analysis or the other does not matter quite so much as we might suppose. Certainly approvals and disapprovals are neither true nor false. Nevertheless, they still have logical properties, as assertions and denials have. It looks as if the question whether there are moral beliefs11 cannot be answered with a straightforward ‘Yes’ or ‘No’.
In the next lecture, I shall begin by considering the logical concept of inconsistency, and shall try to shew that this too applies to approvals and disapprovals, as it does to assertions and denials. I shall then go on to say something about the relation between moral beliefs and feelings, and shall try to illustrate my remarks by asking you to imagine an intelligent being who has no feelings.
These two lectures on Moral Beliefs are a somewhat expanded version of the Burtwood Lectures delivered at Cambridge in 1965, under the auspices of Corpus Christi College, Cambridge. I should like to express my gratitude to the Master and Fellows of the College for inviting me to deliver these annual lectures in that year, and also for allowing me to use a revised version of two of the orally-delivered Gifford Lectures for this purpose.
See Lecture 9, below.
In the United States, I am told, it is quite common for a person who is offered a choice between two cakes to say ‘I believe I'll take this one’.
Cf. the question ‘What am I to do?’ which is an expression of indecision. The answer to it (usually) is an exhortation or a piece of advice, and neither of these is true or false.
See pp. 398–9, below.
Aristotelian Society Supp. Vol., 1948, pp 1–39.
Or perhaps the answer to ‘what is there about him that you have a pro-attitude to?’ is ‘everything’. But this suggests the equally mysterious doctrine of ‘individual essences’, propounded by Leibniz.
If one burglar helps another burglar who has got into difficulties while climbing down a drain-pipe, do we approve of his action? Yes, in so far as it was an act of kindness or beneficence. But it had other characteristics as well. It was part of a burglarious joint-enterprise, which we disapprove of.
See p. 384–5 above: I have borrowed this useful technical term from Professor P. F, Strawson, though he has not, I think, himself used it in this ‘Ethico-logical’ or ‘Ethico-epistemological’ context.
Cf. ‘They were afraid where no fear was’. They really were afraid, of course. To call it an ‘imaginary fear’ is therefore misleading. The fear actually existed and actually had effects. But it had a false presupposition. They believed, mistakenly, that the enemy were attacking them.
Pp. 377 et seq., above.