C. D. Broad

Published in Philosophy 21 (1946).

Ethics, in the sense in which that word is used by philosophers, may be described as the theoretical treatment of moral phenomena. I use the phrase 'moral phenomena' to cover all those facts and only those in describing which we have to use such words as 'ought', 'right and wrong', 'good and evil', or any others which are merely verbal translations of these.

      Moral phenomena fall into three distinct, though closely interconnected groups, viz. moral judgment, moral emotion, and moral volition. Suppose that I know or believe that I ought to keep a promise, though it might be more convenient to break it; that it is wrong to inflict useless pain on an innocent person, though it might be pleasant to score off him in public; that love is a good emotion and jealousy an evil one; and so on. These bits of knowledge or belief are instances of moral judgments. Suppose that I believe myself to have behaved wrongly on a certain occasion and that I feel remorse or self-disapproval, as distinct from mere fear of punishment or embarrassment at being found out, on that account. These feelings will be instances of moral emotion. Suppose, finally, that I have to decide between two alternative courses of action, one of which I believe to be right, and the other of which is pleasanter in itself or more attractive in its probable consequences. In so far as I am influenced in my decision by the thought that one of them is right and that the other would be wrong, and by the desire to do what is right as such, this is an instance of moral volition.

Analysis of Moral Judgments

The first and most fundamental problem of Ethics is about the nature of moral judgments and the concepts 'ought', 'right', 'good', etc., which are the most characteristic elements in them. Suppose I assert, deliberately and reflectively and not merely talking like a parrot, that on a certain occasion ought not to have broken a promise which he had made to B. Then, prima facie, the following things seem to be true:

  1. That in uttering this sentence, which I will call a 'moral sentence in the indicative mood', I am asserting an opinion (correct or incorrect) which I hold, and am not merely expressing an emotion which I feel.
  2. That the opinion which I am asserting is not merely about my own feelings or wishes or beliefs. In saying that A ought to have kept his promise to B. I seem to be asserting about A and B and their relationship something which is no more about me and my attitude towards them than if I had asserted that A is B's second cousin.
  3. That what I assert about A's breach of his promise to B, viz. that it was wrong and ought not to have happened, is something unique and peculiar, though perfectly familiar and intelligible to everyone. It cannot be expressed by any form of words which does not contain the words 'right' or 'ought' or some others which are obviously mere verbal translations of them.

      Now all these prima facie appearances have been questioned on more or less plausible grounds by competent moral philosophers, and this has led to some of the most fundamental discussions in ethics. I will now say something of the various alternative views which have been held on these points.

1. The Interjectional Analysis.

The most radically sceptical view is that what appear to be moral judgments are not really judgments, i.e. assertions of knowledge or opinion, at all; but are merely expressions of a certain kind of emotion. It is alleged that, when a person utters such a sentence as, 'This is wrong' or 'That is evil', he is really only expressing a certain kind of anti-emotion towards this or that. It is true that he uses a form of sentence which inevitably suggests that he is asserting an opinion and not merely expressing an emotion. For the sentence is of the same grammatical form as if he had said 'This is triangular', which is certainly an assertion of opinion. But, it is alleged, the grammatical form is misleading in the case of moral sentences in the indicative.

      I propose to call this theory the interjectional analysis. On this view there are no moral judgments; there are only what might be called 'ostensible moral judgments'. Moral sentences in the indicative mood are really interjections, like 'Hurray!' or 'Blast!', masquerading as assertions of opinion.

2. The Autobiographical Analysis.

Suppose we reject the interjectional analysis. Suppose we hold that, when a person utters a moral sentence in the indicative, he really is making a moral judgment. Then the next most sceptical view is that what he is asserting is simply that he feels a certain kind of emotion, pro or anti, towards the subject which he pronounces to be right or wrong, good or evil. I shall call this the autobiographical analysis.

      It must be noticed that it could take two different forms.

  1. It might be held that, when I judge that so-and-so is right, what I am asserting is simply that I here and now am feeling towards so-and-so a certain kind of pro-emotion. If so, my judgment is analogous to 'This butter tastes nice' uttered by a person while eating that butter.
  2. It might be held that what I am asserting is that I have a disposition to feel this kind of pro-emotion towards such persons or acts or situations as so-and-so. If so, my judgment is analogous to 'I like butter'. A person might truly say that he likes butter even if, on the occasion when he said so, he found the taste of butter repulsive because, e.g., he was feeling bilious.
I shall call these two forms of the autobiographical analysis respectively the occurrent and the dispositional form.

      It might be thought that there is no difference between the interjectional analysis and the occurrent form of the autobiographical analysis. This would be a mistake. There is a difference between merely expressing an emotion by means of an exclamation, e.g., ejaculating 'Damn!' when one is annoyed at losing one's collar-stud, and asserting that one is feeling such and such an emotion towards such and such an object, e.g., saying, 'I am annoyed at finding that I have lost my collar-stud'. An animal, e.g., can express an emotion of anger by snarling, but it cannot make the judgment which a man would express by saying, 'I am angry with so-and-so'. On the interjectional analysis to utter a moral sentence in the indicative is like expressing a feeling of annoyance with so-and-so by exclaiming, 'Damn you!'; on the occurrent form of the autobiographical analysis it is like stating that one is feeling annoyed with so-and-so.

3. The Statistical Analysis.

Suppose next that both forms of the autobiographical analysis are rejected also. It is still possible to suggest an analysis in terms of pro-emotion and anti-emotion. The suggestion would be that, when I judge that so-and-so is right, what I am asserting is that all or most men, or all or most members of some more restricted class, e.g. Englishmen or Etonians, have a disposition to feel a certain kind of pro-emotion towards persons or acts or situations like so-and-so. On this view moral judgments may be compared to such a judgment as 'Jazz music is popular'. This might be truly asserted by a person even if he were himself indifferent to jazz music or heartily disliked it. I shall call this the statistical analysis.

      Before going further I want to make two remarks about the three alternative kinds of analysis which I have been describing.

  1. All three of them arc stated in terms of certain emotions which a person may feel towards himself or towards another person or towards an action or a relationship. They may therefore all be described as emotional-reaction theories.
  2. The interjectional analysis and the two forms of autobiographical analysis agree with each other and differ from the statistical analysis in the following respect. The former may be described as intra-subjective. For, according to them, when a person utters a moral sentence in the indicative, what he is doing is either to express an emotion which he is feeling or to make an assertion to the effect that he is feeling a certain emotion or has a disposition to feel it. The statistical analysis, on the other hand, may be described as trans-subjective. For, according to it, when a person utters such a sentence he is asserting something about a whole class of persons which may or may not happen to include himself.

4. The Objective Analysis.

Finally, let us suppose that all forms of emotional reaction theory are rejected. Then we must hold that a person who makes a moral judgment is ascribing to its subject a certain property which would belong to it even if no one had ever contemplated it or felt any kind of emotion towards it. On this view A's judgment that Bís act of telling a lie on a certain occasion was wrong is comparable, in this respect at any rate, to a person's judgment that the weather in Cambridge on a certain day was rainy. He may indeed have been influenced by his emotions to make this assertion; but what he asserts is not about the emotions of himself or anyone else towards the weather in Cambridge on that day. Let us call this the Objective Analysis of moral judgments.

Naturalistic v. Non-naturalistic Theories

The question of analysis brings us to another question which is closely connected with it. Are moral predicates, such as right, ought and good, unique and peculiar; or can they be completely analysed and defined in terms of non-moral predicates? Theories which answer this question in the affirmative are called naturalistic; those which answer it in the negative are called non-naturalistic. The following would be typical examples of naturalistic theories. 'Better conduct means conduct that comes later in the course of evolution and is more complex and unified than earlier conduct of the same kind'. 'Right action means action which tends to promote the stability and increase the complexity of society'. 'To say that a person ought to do so-and-so means that, if he does not, he will be punished either in this life by his fellow-men or in the next by God'.

      It should be noticed that, if any form of the emotional reaction analysis be true, the question is answered automatically in favour of naturalism. Ethics becomes a branch of psychology. Nevertheless, there would remain a somewhat similar question even for those theories. It would take the following form. 'Is the emotion which we express, or assert ourselves to feel or to have a disposition to feel, or which assert that most members of a certain class have a disposition to feel, when we utter a moral sentence in the indicative an emotion of a quite unique kind? Or is it just a combination of emotions, e.g. fear, love, hope, etc., each of which can occur in non-moral contexts?'

      If the objective analysis be correct, the question of Naturalism v. Non-naturalism remains quite open, and special arguments are needed to answer it.

      The importance of the question is this. If Non-naturalism be true, Ethics is an autonomous science with an irreducibly peculiar subject-matter, though it will still have very intimate connections with certain other sciences, such as psychology sociology, etc. But, if Naturalism be true, Ethics is not an autonomous science; it is a department or an application of one or more of the natural or the historical sciences. Now the reduction of a plurality to a unity is a source of intellectual satisfaction, and therefore philosophers have a strong motive for trying to produce a workable naturalistic theory.

Right-making and Good-making Characteristics

We pass now to another very important problem. It may be introduced as follows. If a person says of anything that it is right or that it is wrong, it is always sensible to ask, 'Why? What makes it right or makes it wrong, as the case may be?' The sort of answers that one expects to such questions are: 'Because it will relieve pain', 'Because it is a breach of promise', and so on. Similar remarks apply, mutatis mutandis, to good and evil. If anything is said to be good or to be evil, it is always sensible to ask what makes it so. The sort of answers which one expects to get are: 'Because it is an act of courage', 'Because it is a feeling of pleasure at another man's misfortune', and so on.

      We may generalize this as follows. Moral characteristics are always dependent upon certain other characteristics which can be described in purely neutral non-moral terms. Let us call those non-moral characteristics whose presence in anything confers rightness or wrongness on it right-making and wrong-making characteristics. And let us define good-making and bad-making characteristics in a similar way.

      We will be in with right making and wrong-making characteristics. On the face of it there is a whole mass of these. E.g. being a breach of promise, being a deliberately misleading answer to a question, being an intentional infliction of needless pain, and dozens more, are characteristics which may plausibly be said to make an act wrong.

      Now an extremely important question is whether we can discover any kind of systematic unity among all these various right-making and wrong-making characteristics. Can we reduce them to a few fundamental ones? Can we perhaps reduce them all to a single fundamental one? Moral philosophers have naturally tried their hardest to do this, since it would plainly be tidier and more satisfactory to the intellect if it could be done.

      When we reflect on this problem we notice the following fact. At first one is inclined to say that every lie is as such wrong, that every breach of promise i as such wrong, and so on. But one soon finds that there are cases where this is not plausible; e.g. is it certain that a lie told' to an invalid or a breach of promise to a child is wrong when the results of telling the truth or keeping the promise would be extremely bad for him? Again, there are cases where any possible action will, e.g., be either a lie or a breach of promise. Suppose, e.g., that A has told me a secret on my promise not to reveal it, and that B afterwards asks me a question which I can neither answer truly nor refuse to answer at all without revealing the secret. Then whatever I may do in response to B's question will be either a breach of my promise to A or a lie told to B. But we are not prepared to say that whatever I do in such a situation will be wrong. On the contrary, w e should hold that in some cases it would be my duty to tell the truth to B and thus break my promise to A whilst in others it would be my duty to keep my promise to A and thus deceive B.

      For such reasons it is necessary to modify our notion of right-making and wrong-making characteristics and to talk instead of right-tending and wrong-tending characteristics. An intentionally deceptive answer to a question tends as such to be wrong, and so too does a breach of promise. If an act were nothing but an answer to a question, it would be right if true and wrong if false. If an act were nothing but the keeping or the breaking of a promise, it would be right if it were the former and wrong if it were the latter. But, if an act is both a true answer to a question and a breach of a promise, we can say only that it tends to be right in the former respect and tends to be wrong in the latter. The right act in such circumstances will be the one that makes the best compromise between the various moral claims on the agent, after allowing due weight to the relative urgency of each claim. We might compare the claims which arise from various right-tending and wrong-tending characteristics to forces of various magnitudes and directions acting on a body at the same time. And we might compare what I will call the resultantly right course of action to the course which a body would pursue under the joint action of such forces. Looking at the situation from the point of view of the agent, we can say that each right-tending and wrong-tending characteristic imposes on him a component obligation of a certain degree of urgency; and that his resultant obligation is to make the best compromise that he can between his various component obligations.

      When we consider the various right-tending and wrong-tending characteristics we find that they can be divided into two great groups, which I will call teleological and ostensibly non-teleological.

      1. One characteristic which tends to make an act right is that it will produce at least as good consequences as any alternative open to the agent in the circumstances. And one which tends to make it wrong is that it will produce less good or more evil consequences than some other act open to the agent. We can sum this up by saying that the property of being optimific is a very important right-tending characteristic. I call it teleological because it refers to the goodness of the ends or consequences which the act brings about.

      2. Now there are also many characteristics which are certainly right-tending or wrong-tending but are not prima facie reducible to the property of being optimific. No doubt truth-telling and promise-keeping do in the end and on the whole lead to better consequences than lying and breach of promise. But most people do not feel that this is the reason why truth-telling and promise-keeping tend to be right. They feel that the mere fact of being asked a question or having made a promise imposes on one an urgent component obligation to answer truly or to perform what one has promised, quite independently of whether the consequences will be good or bad. I therefore call these right-tending and v. wrong-tending characteristics ostensibly non-teleological.

      We have already seen that various ostensibly non-teleological right-making characteristics may lead to conflicting component obligations. It is also true that the ostensibly non-teleological obligation to tell the truth, e.g., may conflict with the teleological obligation to produce as much good and as little evil as possible. Consider, e.g., the following as a case. A commanding officer knows that one of his subordinates, who has been killed, has displayed disgraceful cowardice. No one else knows this or will ever do so unless the officer divulges it. The dead man's mother asks the officer leading questions about the circumstances of her son's death. If he tells the truth the mother will be made miserable for life and no one will be a penny the better. If he tells a suitable lie the mother will retain her ideals and be made happy and no one will be a penny the worse. Here there seems to be a plain conflict between the teleological obligation to produce as much good and as little evil as possible and the ostensibly non-teleological obligation to answer questions truly.

      Ostensibly non-teleological obligations can be subdivided into two groups, which I will call non-distributive and distributive. Truth-telling is an example of the former. Distributive obligations are concerned with the right distribution of benefits and disadvantages. Suppose that I am the sole executor and trustee under the will of a certain rich man. He has made two wills. In the first he has distributed his property more or less equally among a number of needy and deserving persons and institutions. In the second he has left the whole of it to a worthless rich relative. I am the only person now alive who knows that the second will has been made, and I could safely destroy it and carry out the provisions of the first. It is obvious that by doing this I should produce more good and less evil than by divulging the second will. Nevertheless I am under an extremely urgent ostensibly non-teleological obligation to distribute the property in accordance with the testator's second will, whilst my purely teleological obligation would be to distribute it in accordance with the first.

      Now much the most important attempt which has been made to reduce all the many and various right-tending characteristics to a single one is the theory called Utilitarianism. According to this one's only ultimate obligation is teleological; the only ultimate reason why an act is right is that it is optimific, and the only ultimate reason why any act is wrong is that it would produce less good or more evil consequences than some other act open to the agent in the circumstances. All ostensibly non-teleological obligations, whether distributive or non-distributive, are secondary and derivative from the one teleological obligation to act optimifically. The only reason why there is a component obligation to keep promises, to answer questions truly, and so on, is that on the whole such action will secure the best consequences in frequently recurring kinds of situation, such as having made a promise, being asked a question, etc. Suppose that a situation should occur in which, when all the remote, secondary, and collateral consequences as well as the immediate ones have been taken into account, the result of telling a lie or breaking a promise would be better than that of telling the truth or keeping the promise. Then it will be right to lie or to break one's promise and wrong to tell the truth or keep faith.

      It is plain that, if Utilitarianism can be made to cover the facts without distorting them, it has several advantages.

  1. It gives us the intellectual satisfaction of reducing a litter of disconnected grounds of obligation to a single one.
  2. To many people it does seem difficult to believe on reflection that it can ever be right to do what will have worse consequences when one could have done something else which would have better consequences.
  3. Utilitarianism gives a plausible explanation for the various degrees of urgency of the various ostensibly non-teleological component obligations; and it provides, in theory at any rate, a rule for compounding such obligations when several of them co-exist and conflict with each other.

      I will now say something about good-tending and bad-tending characteristics. The general principles are the same as in the case of right-tending and wrong-tending ones. It is plain that prima facie there are a number of different characteristics which tend to make a person or an experience or an action good, and a number which tend to make it bad. Now several of these may be present together in a single subject; and the question whether it is resultantly good or resultantly bad, and, if so, to what degree, will depend on the nature and the proportion of its various good tending and bad-tending characteristics.

      Here again there is naturally a strong desire among philosophers to try to reduce the litter of various good-tending characteristics to a single good-making one. The best known effort in this direction is the theory know n as Ethical Hedonism. This theory involves the following propositions.

  1. Nothing is either good or bad in the primary sense except actual experiences.
  2. The only characteristic of an experience which makes it good is its pleasantness, and the only one which makes it bad is its unpleasantness.
  3. The degree of goodness of a pleasant experience depends jointly on its duration and on the degree of its pleasantness.

According to this theory anything other than an experience which is called 'good' is so called in a secondary and derivative sense, viz. in so far as it contributes or tends to contribute to the occurrences of pleasant experiences and the non-occurrence of unpleasant ones.

      If both Utilitarianism and Ethical Hedonism could be accepted, we should have introduced the greatest possible unity into the region of moral phenomena. Unfortunately each of them seems to be too simple to cover the facts without distorting them.

      Before leaving this part of the subject I will make two remarks connecting it with the topic of analysis which I discussed earlier.

      1. Suppose that a person has persuaded himself that there is one and only one right-making characteristic, e.g. that of being optimific, or one and only one good-making characteristic, e.g. pleasantness. Then he is very liable to make the following mistake. He is apt to think that he has proved that 'right' means optimific or that 'good' means pleasant, i.e. that he has provided an analysis of rightness or of goodness, as the case may be. All that he has really shown in the first case, e.g. is that, if the words 'right' and 'optimific' are names of two different characteristics, then these two mutually involve each other. That is quite different from showing that the two words are really names for the same characteristic, and that what is meant by 'optimifics is the analysis of what is meant by 'right'. The distinction can be made quite clear by a simple non-ethical example. To be an equilateral triangle means to be a plane figure hounded by three equal straight lines. To be an equilateral triangle means to be a plane figure with three angles, all of which are equal. Evidently these are two different characteristics. But they mutually involve each other; for any figure which has either property necessarily has both. It seems not unlikely that many people who have thought they have given a naturalistic analysis of moral judgments have made this mistake; and that really they have done no more than to produce reasons for thinking that there is one and only one right-making or good-making characteristic, and have then proceeded to identify rightness or goodness with this.

      2. In discussing right-tending and good-tending characteristics I have spoken in terms of the objective analysis of moral judgments. It is important to notice that the same problem exists in a slightly modified form if we accept the emotional-reaction analysis. In that case what we have called a 'right-tending' or a 'good-tending' characteristic will be one which tends to call forth the peculiar emotion in its pro-form. What we have called 'conflicts of component obligations' will depend on the fact that the same act may have features which call forth the pro-emotion and others which call forth the anti-emotion. What we have called 'resultant obligation' will be connected with one's total emotional reaction to an object which has some features that tend to call forth the pro-emotion and others which tend to call forth the anti-emotion.

Intention and Rightness

When a person performs a deliberate action he does so in view of his knowledge and beliefs about the present situation and with certain expectations about the consequences which will ensue. These two factors are closely connected; for his expectations about the consequences are in part determined by his knowledge or beliefs about the present situation. I shall say that an act is intentional in respect of

  1. all those features and only those which the agent knows or believes to be present in the initial situation, and
  2. all those consequences and only those which he expects to follow.

Now a person's information on both these matters will always be incomplete and it may be in part mistaken. No man can foresee the very remote consequences of an action; and anyone may be mistaken about some of its immediate consequences, either through miscalculation or through inadequate or inaccurate information about present circumstances. Suppose, e.g. that a person receives a letter purporting to come from his old nurse and that he is moved to send her a postal money-order in the belief that she is in want and with the expectation that it will enable her to buy comforts. It may be that in fact the nurse has died, that the letter has been written in her name by a dishonest relative, and that the money will be spent by him on drink. What this man intended to do was to bring relief to his old nurse; what he in fact did was to enable a dishonest stranger to get drunk.

      Now, if we consider the agent's intention in this example, we are inclined to say that he acted rightly. But, if we consider the actual facts of the situation and the consequences, we are included to say that he acted wrongly and that the right action would have been to refuse to send money and to have reported the matter to the police. Thus we are faced with the problem of the relation between intention and rightness or wrongness.

      This question may be approached in the following way. Any act which can be called 'right' or 'wrong' can be viewed from two standpoints, viz. that of the agent who does it and that of the patient who is affected by it. In general these will be different persons, though there are special cases in which the agent and the patient are the same person at an earlier and a later stage of his life. Now in considering whether an act is right or wrong we must view it, so to speak, from both ends, i.e. in relation to the patient and in relation to the agent. In relation to the patient an act is right if and only if it fulfils his claims on the agent, or, as we say, 'gives the patient his rights in the matter concerned'. From this standpoint the agent's intention is irrelevant. In relation to the agent an act is right if and only if it is done with the intention of fulfilling the patient's claim and giving him his rights in the matter. From this standpoint anything in the actual consequences which is outside or contrary to the agent's intention is irrelevant.

      I propose to call any act which in fact fulfils the claims of the patient upon the agent materially right, regardless of whether the agent intended it to have this consequence or not. I propose to call any act which was untended by the agent to bring about the fulfilment of the patient's claims formally right, regardless of whether it does in fact have that result or not. A perfectly right act in a given situation would be one that was both formally and materially right. It would be an act which was intended by the agent to give to the patient his rights and which did in fact do so. Owing to incomplete or incorrect information on the part of the agent, or to defects in his powers of inference, it may happen that an act which is formally right is materially wrong, or that one which is formally indifferent or wrong is materially right. It should be noticed that the notion of material rightness is, in a certain sense, more fundamental than that of formal rightness. For what is formally right for the agent to do is to try to secure to the patient what is materially right for him to have done to him.

      There remains, however, a further serious complication to be considered. So far I have supposed that the agent makes no ethical mistakes. I have supposed only that he may have incomplete or inaccurate information about matters of fact and may make mistaken inferences on such matters from his information. I have assumed that he knows what ought to happen to the patient if his factual information were adequate and accurate. But of course the agent may be ignorant or mistaken about ethical matters too.

      Suppose, e.g., that a person is brought up in a community in which it is held to be a duty to carry on a family vendetta, and that he accepts that opinion. Let us assume, for the sake of argument, that it is mistaken, and that it is wrong to kill a member of another family simply because one of his ancestors killed one of one's own family. Suppose that this person is in a situation in which he can either kill a certain member of the family or let him escape. Whichever alternative he chooses we are inclined to say that he acts wrongly. If he kills the patient, he intentionally does to him what he believes ought to be done to him, but this is in fact what ought not to be done to him. If he lets the patient escape, he intentionally does to him what he believes ought not to be done to him, but this is in fact what ought to be done to him.

      It is plain that we are here concerned with yet another sense of 'right' and wrong'. I propose to call it subjective rightness and wrongness. An act is subjectively right if and only if the effects which the agent expected it to have on the patient are those which he believed that the patient is entitled to have produced in him.

      The relations between the various senses of 'right' which I have distinguished may be summarized as follows.

  1. A person could be sure of doing a perfectly right act only if both his relevant factual and his relevant ethical beliefs were complete and correct and if he had made no mistakes in his inferences. It is therefore plain that, if a person ever does a perfectly right act, it is largely a matter of luck that he does so.
  2. A person could be sure of doing a formally right act, even if his factual information were incomplete or inaccurate and he made mistakes in his inferences, provided that the effects which he thinks his act would have upon the patient are such as the latter really would be entitled to if his nature and situation were as the agent believes them to be. Therefore is when an agent's relevant ethical information is incomplete or incorrect it is a matter of luck if he performs a formally right act.
  3. A person could be sure of doing a subjectively right act, no matter how inadequate or inaccurate his factual and his ethical beliefs might be or how mistaken he may be in his inferences provided only that the effects which he thinks his act will have on the patient are such as he thinks that the latter would be entitled to if his nature and situation were as the agent believes them to be. It is therefore plain that a person who is ignorant, stupid, and misinformed about facts, who is incapable of drawing reasonable inferences, and who is insensitive or crazy in his opinions about what is materially right and wrongs may perform acts that are subjectively right. So it is not surprising that such acts may inflict the most terrible wrongs on those whom they effect.

      The problems which we have been discussing arise because we fail to distinguish these three senses of 'right' and 'wrong', and use these words in a vague way to include them all, sometimes having one meaning predominantly before our minds and sometimes another.

Motives and their Ethical Function

Among the characteristics which an agent believes an action to have, and among the consequences which he expects to follow from it, some will attract him towards doing it, some will repel him from doing it, and others will leave him indifferent. Suppose, e.g., that a person contemplates throwing a bomb at a ruler in a public procession. He may expect that the effects will include the death of the ruler, the death or injury of a number of innocent bystanders, and the breakage of a number of windows in the neighbourhood. The first part of the expected consequences may attract him, the second may repel him, and the third may leave him indifferent. A person's total motive in doing a certain action consists of all that he believes about the action itself and all that he expects about its consequences, which either attracts him towards or repels him from doing it. The former constitutes his total motive for doing it, and the latter his total motive against doing it. If, in fact, he does it, he does it because of his motives for doing it and in spite of his motives against doing it. Suppose, e.g., that the anarchist in my example is in general a humane man and that he decides to throw the bomb at the ruler. Then his motive for doing so is the attractive belief that it will kill the ruler; his motive against doing so is the repellent belief that it will kill or injure innocent bystanders; and he acts because of the former and in spite of the latter motive.

      It is plain that there are two aspects to any motive, viz. a cognitive and a conative-emotional aspect. The cognitive aspect of a motive is the fact that it is a belief about the nature of the action or an expectation about its consequences. The conative aspect is the fact that the agent has a certain disposition to be attracted or repelled which is excited by this belief.

      When we know what was a person's intention in doing an action and what consequences in fact followed fiom it, we are in a position to judge whether it was subjectively right, or formally right, or perfectly right, without needing to know anything about his motives in doing the action. But it is quite obvious that a man's motives in doing an action have a very important bearing on some kind of moral judgment which we make either on the agent or on the action. This fact is indicated in ordinary speech by such phrases as 'He did the right thing from the wrong motive'.

      Suppose, e.g., that a man performs an act which is intended to secure the just punishment of a criminal. He will forsee that the criminal will suffer directly and his family and friends indirectly, so this must be included as part of his intention. Now it may be that the belief that the law will be vindicated, that other men will be deterred from committing similar crimes, and that the criminal may be reformed, is an attracting one; that the belief that the criminal and his family will suffer is a repelling one; and that the agent acts because of the former and in spite the latter. If so, we should be inclined to say, not only that his action was right, but also that his motives in doing it were good. But it may be that the belief that the law would be vindicated, other men deterred, and the criminal perhaps reformed, exercised no attraction on the agent. He had, perhaps, had a quarrel with the criminal or was jealous of him; and what attracted him was his belief that the criminal and his family would suffer. If so, the action would still be right in any of the senses which we have considered, but we should certainly say that the agent's motive in doing it was bad.

      I have no doubt that the words 'right' and 'wrong' have, in addition to the ambiguities which we have already cleared up, the further ambiguity that they are sometimes used to include a reference to this agent's motives and sometimes used without such a reference. I think that it is on the whole more convenient explicitly to exclude reference to motives from our description of right and wrong action. One important reason for drawing the line at this point is the following. A person can choose which of several alternative possible actions he will do. But he cannot, in the same sense, choose which of several alternative motives shall attract him towards or repel him from doing a certain action. Now the predicates 'right' and 'wrong' are commonly understood to be confined to that which is directly dependent on a person's volition, in the sense in which his actions are so and his motives in acting are not.

Specifically Moral Motivation and Emotion

It seems, prima facie, that human beings have a great many different desires, and that these canno all be reduced to a single head. Naturally attempts have been made to do this. The most celebrated of them is the theory called Psychological Hedonism. This asserts that the only ultimate objects of desire for any person are to get and to prolong pleasant experiences and to avoid and cut short unpleasant ones. It is not generally admitted by competent authorities that this theory cannot be maintained, and that such plausibility as it has depends upon certain verbal ambiguities.

      Now, prima facie, there appears to be among our other desires and aversions one which is specifically moral. It seems that, if one believes that a certain course of action would be right, that belief stirs a certain conative disposition in one and is a motive for doing it. If, on the other hand, one believes that an action would be wrong, that belief stirs the same conative diposition and is a motive against doing it. These desires and aversions are often opposed to very strong non-moral desires and aversions, and they feel very peculiar in comparison with the latter whether they happen to oppose them or to reinforce them. For this reason they are commonly marked out by the name feelings of obligation, and some philosophers have thought it inappropriate to classify them as desires and aversions. For my part I see no objection to classifying them in this way, provided that one does not lose sight of their peculiarities. Therefore I propose to describe this peculiar kind of desire and aversion as the desire to do what is right as such.

      Now the following questions arise at this point.

  1. Is there really a desire to do what is right as such, or is the opinion that there is mistaken? Is it the case that, whenever a person thinks that he is attracted towards a course of action by the belief that it would be right or repelled from it by the belief that it would be wrong, he is really being attracted or repelled, not by these beliefs, but by beliefs about certain non-moral features of the act or of its consequences?
  2. Supposing that there is a desire to do what is right as such, is it ever sufficient to determine one's actions, or does it always need to be supported by some non-moral motive, such as desire for praise or fear of punishment?
  3. Supposing that this desire exists and is sufficient to determine one man's action in the absence of opposing motives, is it ever sufficient by itself to overcome opposing motives when they are present? Or must it in such cases always be reinforced by some non-moral motive?
  4. Supposing that Question 3 is answered in the affirmative, is there any sense, and if so what, in which we can say that the desire to do what is right as such always could have overcome all opposing motives, even when it did not in fact do so?
  5. Is it essential for the validity of moral judgments that Question 4 should be answered in the affirmative? And, if an affirmative answer is relevant to the validity of some but not all moral judgments, which are those to which it is relevant?

      It will be seen that Questions 4 and 5 bring us to the problem of Free-Will v. Determinism and its bearing on morality.

      As regards (Question 1 it is important to notice and to avoid the following very common fallacy. Suppose it could be shown that what we take to be the desire to do what is right as such has developed, either in the history of each individual or in that of the human race, on regular principles out of desires which were all purely non-moral. (More or less plausible attempts to show this have been made, e.g., by certain psychoanalysts, on the one hand, and by certain sociologists, on the other.) It would be a fallacy to conclude that what we take to be the desire to do what is right as such is not what it appears to be, but is really just one or a combination of purely non-moral desires. An account of the stages out of which something developed in a regular way is one thing, and an analysis of it as it is when fully developed is another. But it is very common to confuse the two and to imagine that one has shown that the end-term of such a process just consists of the earlier terms in a disguised form.

      This fallacy is often made plausible by the use of question-begging epithets for describing the earlier phases in such a process of development. Thus, e.g., some psychoanalysts describe an emotion which is supposed to occur in babies at the pre-moral stage by the name 'feeling of guilt'. Now the phrase 'feeling of guilt', if taken literally, means an emotion which a person feels towards himself in respect of his belief that he has done something morally wrong. It is therefore quite meaningless to suggest that anyone who has not already got the notion of right and wrong can literally have a feeling of guilt. The phrase 'feeling of guilt' must therefore be used in some unexplained metaphorical sense. But the use of it to describe the pre-moral stages illegitimately helps the suggestion that the end-term contains nothing that was not present in the earlier phases.

      This brings us to the general notion of specifically moral emotion. By this I mean emotions which appear prima facie to be felt towards persons or actions in respect of certain moral characteristics which they are believed to have. Such emotions may be either reflexive or non-reflexive. The former are felt by a person towards himself or his own actions, e.g. feelings of guilt, of remorse, of self-approval, etc. The latter are felt towards another person or his actions, e.g. feelings of moral approval or disapproval felt by one person for the acts of another.

      The only remark that I wish to make here about them is that their apparent existence presents a considerable difficulty to any form of the emotional attitude analysis of moral judgments. According to such analyses, to be right or to be wrong consists in being the object of moral approval or disapproval, as the case may be, to some person or class of persons. But, prima facie, an action becomes the object of a feeling of moral approval or disapproval to a person only in so far as he already believes it to be right or to be wrong, as the case may be. There is certainly the appearance of a vicious circle here, and it remains to be seen whether supporters of the emotional attitude type of analysis can show that this appearance of circularity is delusive.

Epistemological Questions

The last set of problems which I wish to mention can be stated as follows. How do we come to have ideas of specifically moral terms, such as right, ought, morally good, and so on? And how do we come to know or believe propositions connecting non-moral characteristics, such as truth-telling or promise-breaking with moral characteristics such as rightness or wrongness? These may be described as epistemological questions. It is plain that the answers to them will be closely bound up with the answers to the questions how moral judgments should be analysed.

      Suppose, e.g., that the interjectional analysis were correct. Then there are no moral judgments and therefore no moral predicates. The first question would then have to be transformed into the following. How do we come to make the mistake of thinking that we are ascribing to subjects predicates of a peculiar kind when in fact we are merely expressing certain emotions towards objects? The second question would have to be transformed somewhat as follows. Is it just an ultimate fact about human nature that most people tend to feel a certain kind of emotion when they contemplate, e.g., an act of promise-breaking; or is this explicable by general psychological principles and the particular influences to which most people are subjected in early childhood?

      Suppose, next, that the interjectional analysis is false, but that it were true that moral concepts, such as right and ought, are definable in terms of certain kinds of pro-emotion and anti-emotion. Then the origin of such concepts would presumably be like that of our concepts of other psychological terms. We should feel these pro-emotions and anti-emotions on certain occasions, we should introspect them and compare and contrast them with other experiences which we have and introspect, and then by a process of abstraction, we should form the idea of their characteristic emotional quality. Then, finally, we should define 'right' and 'ought' in terms of emotions which have this quality. Moral concepts would in fact be empirical in origin.

      Now, if this kind of analysis of moral judgments were correct, what we have called 'right tending' and 'wrong-tending' characteristics would be those characteristics which tend to evoke pro-emotions or anti-emotions of a certain specific kind towards persons or actions which are believed to possess them. So the second question would reduce to the following. How do we come to know or to believe that such and such non-moral characteristics of persons or actions tend to evoke in those who believe them to be present such a pro-emotion or anti-emotion? Presumably the answer would be that we derive such beliefs by generalizing from our experience. We observe that a belief that an act has a certain non-moral characteristic, e.g., that it is an intentionally misleading answer to a question, is regularly accompanied by an anti-emotion of a specific kind towards the act in question. And we base on this an inductive generalization. Such beliefs would in fact be empirical and inductive in origin.

      Even if all forms of the emotional attitude analysis were rejected and some form of the objective analysis were accepted, it would still be reasonable to hold that both moral concepts and moral judgments are of empirical origin, provided only that a naturalistic form of the objective analysis is adopted. But, if we feel obliged to accept a non-naturalistic theory of moral judgments and concepts, the case is altered.

Let us define an 'empirical concept' as the concept of a charactcristic which is either

  1. manifested to us in sensation or introspection, or
  2. is definable in terms of such characteristics together with the notions of cause or subtance or both.

(The concepts of sensible redness and of anger, e.g., come under the first heading; those of physical redness and irascibility, e.g., come under the second.) If we adopt this definition, it seems certain that the concepts of moral characteristics, such as right, ought, and morally good, cannot be empirical unless those characteristics are naturalistic. Therefore any one who accepts a non-naturalistic account of moral characteristics is almost certainly committed to the proposition that moral concepts are non-empirical. Now many philosophers accept, either as self-evident or as a postulate, the principle that all concepts are empirical. If one is quite sure of this epistemological principle, one will have to reject the non-naturalistic account of moral characteristics, no matter how plausible it may seem on other grounds. If one is quite sure of the non-naturalistic account of moral characteristics, one will have to reject this epistemological principle, no matter how self-evident it may seem or how useful it may be as a postulate. If, on the other hand, one is not quite sure of either, the conflict will tend to diminish one's confidence in both.

      Again, it seems plain that, if right and good are non-naturalistic characteristics, the propositions connecting them with right-tending or good-tending non-moral characteristics, such as promise-keeping or tendency to promote happiness must be synthetic. Now to many people it seems that such propositions as 'Any act of promise-keeping tends as such to be right' are necessary and self-evident like the axioms of pure mathematics. But it is also a very widely accepted epistemological principle that there can be no synthetic necessary propositions. There are, according to this principle, synthetic propositions and there are necessary propositions; but the former are all contingent and empirical, and the latter are all analytic. Now a person who holds that moral characteristics are non-naturalistic seems committed to holding that such propositions as 'Any act of promise keeping tends as such to be right' are either

  1. contingent empirical generalizations, or
  2. synthetic necessary propositions.

The former alternative conflicts with the prima facie appearance that these propositions are self-evident and necessary; the latter conflicts with the epistemological principle that all necessary propositions are analytic. Thus he must either reject the principle or try to show how it is that such propositions appear to be necessary and self-evident although they are in fact contingent and empirical.

      I have now completed my account of what seems to me to be the main problems of Ethics. I have confined myself to stating alternatives and indicating the connections and disconnections between them. This is not very exciting, but I think it is a necessary preliminary to anything more positive.

Transcribed into hypertext by Andrew Chrucky, Jan. 2, 1999.
Go to Philosophical Alternatives from C. D. Broad.