During the long vacation of 1944 I spent such time as I could spare from my other duties in reading with some care Richard Price's book A Review of the Principal Questions and Difficulties in Morals. This was first published in 1758, and it reached a third edition in 1787. Price died soon afterwards, viz. in 1791. Until Ross published his book The Right and the Good in 1930 there existed, so far as I know, no statement and defence of what may be called the 'rationalistic' type of ethical theory comparable in merit to Price's. Price was thoroughly well acquainted with the works of other great English philosophers and moralists, such as Locke, Berkeley, Hume and Butler, and he develops his own views in conscious opposition to those of Hutcheson, the founder of the so-called 'moral-sense' type of ethical theory.

I had thought at one time of writing a critical account of Price's doctrines. But, when I began to do so, I soon found that it would be more profitable to treat independently and in modern terminology some of the questions with which Price was mainly concerned. Therefore my further references to Price will be only occasional and incidental; but I wish to make it plain that his book is the background of my paper, and that reading the former was the stimulus to writing the latter.

The topic with which I shall be primarily concerned may be called the 'epistemology of moral judgments'. This subject is or considerable interest in itself, and I think that it has been very inadequately treated by most writers on ethics. But it is important also for another reason. Questions of epistemology and of logical analysis are interconnected, and the answer which we give to a question of the one kind may have an important bearing on that which we should be inclined to give to a question of the other kind, e.g. I should be prepared to argue that, if ethical terms, such as right and good, are simple and non-naturalistic or are complex and contain a non-naturalistic constituent, then the concepts of them must be wholly or partly a priori. On the same hypothesis I should be prepared to argue that such judgments as 'Any act of promise-keeping tends as such to be right' must be synthetic and a priori. Now it is a well-known and plausible epistemological theory that there are no a priori concepts and no synthetic a priori judgments. If I am right, anyone who feels no doubt about this epistemological theory can safely reject the analysis of moral judgments which makes them contain non-naturalistic constituents. On the other hand, anyone who feels bound to accept that analysis of moral judgments will have to reject this epistemological theory.

In the discussion which follows I shall confine myself to the concepts right and wrong, in the specifically moral sense, and to judgments in which they occur as predicates. I think that most of what I say could be transferred mutatis mutandis to the concepts morally good and evil and to judgments in which they occur.

As Price points out, the words 'right' and 'wrong' are used in at least two different senses. This is made obvious by the fact that the sentence 'It is always right for a person to do what he honestly believes to be right, and wrong for him to do what he honestly believes to be wrong', is intelligible and would generally be admitted to be in some sense true. The two senses in which 'right' and 'wrong' occur in this sentence may be described as the 'subjective' and the 'material'. An act is subjectively right if the consequences which the agent expects it to have are such as he thinks would be materially right in the situation as he believes it to be. We shall be concerned here only with material rightness and wrongness. Let us call sentences in which the words 'right' or 'wrong', used in the material sense, occur as predicates 'deontic sentences'. An example would be 'Any act of promise-breaking tends as such to be wrong'.

I shall first distinguish certain alternative analyses which have been proposed for the situations expressed by deontic sentences and then I shall consider certain alternative theories which might be held concerning deontic knowledge or belief. In the course of the discussion I shall try to bring out the relations between the two sets of theories.

(i) Alternative Analyses of Deontic Sentences

When a person utters such a sentence as 'That act is right' he seems prima facie to be expressing a judgment, and in that judgment he seems prima facie to be ascribing to a subject a predicate which has no reference to his own or other men's sensations, emotions, desires, or opinions. But we know that such appearances may be misleading. Such sentences as 'This food is nice' and 'That thing is yellow' are of the same grammatical form as 'That act is right'. Yet everyone would hold that the predicate of the first refers to the speaker's sensations of taste, and many people would hold that the predicate of the second refers to the visual sensations of human beings. So the first question to be asked is this. Do sentences like 'That act is right' express judgments at all? If not, what do they express?

As is well known, there is a theory that such sentences do not really express judgments at all. It has been held that they express only certain emotions felt by the speaker, or certain desires of his, or certain commands. I shall call this the 'interjectional theory'. Price does not consider this extreme view. If it had been put to him, he would probably have regarded it as too fantastically absurd to be taken seriously. It is, indeed, a kind of theory which can be swallowed only after one has undergone a long and elaborate process of 'conditioning' which was not available in the eighteenth century.

Suppose that the interjectional theory is rejected. Suppose we hold that deontic sentences do express judgments of some kind, and that at any rate the fact that they are in the indicative mood is not misleading. The next suggestion is that the judgments which they express are really about certain human experiences, certain sensations or emotions or desires. I shall call this the 'subjective theory'. I shall now point out that it may take a great number of different forms, and shall try to classify them.

The factor common to all forms of the subjective theory is that there is a peculiar kind of experience which human beings are liable to have when they contemplate certain acts, e.g. acts of promise-keeping or of treachery, just as there is a peculiar kind of experience which they have when they look at certain objects, e.g. at snow or at soot. I propose to call this at present by the intentionally vague name 'moral feeling'. I use this term because it covers both sensation and emotion. Since deontic judgments take the two opposite forms 'That is right' and 'That is wrong', it must be assumed that moral feeling takes two opposite forms. There are analogies to this both in sensation and emotion. There are the opposed temperature-sensations of hotness and coldness, and there are the opposed non-moral emotions of love and hate. I shall speak of the 'pro-form' and the 'anti-form' of moral feeling, and will assume that the former is associated with judgments of rightness and the latter with those of wrongness. The first division of subjective theories is into sensational and emotional, according to whether moral feeling is held to be analogous to sensation and moral judgment to be analogous judgments of sense-perception, or whether the feeling is held to be a form of emotion and the judgments to be concerned with that emotion.

The next division of subjective theories is into what I will call the 'intra-subjective' and the 'trans-subjective' varieties. According to the first of these a person who judges that so-and-so is right is asserting something about his own moral pro-feelings only. He is not saying anything about the moral feelings of other men. According to the second variety such a person is asserting something about all men, or most men, or a certain restricted class of men and not only about himself.

Lastly, each of these two varieties of the subjective theory can be subdivided into what I call an 'occurrent' and a 'dispositional' form. On the occurrent form of the intra-subjective variety of the subjective theory a person who says that so-and-so is right asserting only that at this moment he is having a moral pro-emotion towards so-and-so. On the occurrent form of the trans-subjective variety of the theory he is asserting that all or most members of a certain class of men, e.g. most members of the Athenaeum, are at present having a moral pro-emotion towards so-and-so. On the dispositional form of the intra-subjective variety of the theory he is asserting that he has a disposition to feel a moral pro-emotion whenever he contemplates so-and-so or other acts like it. He may not be feeling such an emotion at the moment when he is saying that so-and-so is right. He might not be actually witnessing or thinking of such an act at the time; or, if he were, he might be in some special occurrent state, such as anger or jealousy, which is inhibiting or reversing his disposition to feel moral pro-emotion. On the dispositional form of the trans-subjective variety of the theory he is asserting that all or most men or all or most members of a certain class of men have a disposition to feel moral pro-emotion when they contemplate so-and-so or other acts like it. He might have strong reason to believe this even if he lacked that disposition himself. I have, e.g., strong reason to believe that most men have a disposition to like the taste and smell of apples, though I personally loathe them.

It appears then that there are at least eight possible species of the subjective theory, according as it is (i) sensation or emotional, (ii) intra-subjective or trans-subjective, or (iii) occurrent or dispositional. There are two remarks that I would make at this point.

(i) Even on the occurrent intra-subjective form of the theory such a statement as 'That act is wrong' could be questioned without accusing the speaker of lying about his own feelings at the time. But this could happen only in one way. The speaker might be mistaken about the kind of feeling which he is having when contemplating this act. He might think that he is having a moral anti-feeling when really he is having what Sidgwick calls a 'feeling of quasi-moral repugnance'. I have no doubt that such mistakes are often made by people, e.g., about their own feelings towards abnormal sexual desires and practices, (ii) I am inclined to think that the only form of the theory that is worth serious consideration is the trans-subjective dispositional form of it. But I should admit that it is not unplausible to hold that sometimes when a person says that so-and-so is right or that it is wrong he may be talking only of his own disposition to have a moral pro-feeling or anti-feeling when he contemplates such acts.

So far I have spoken only of singular deontic judgments, i.e. those of the form 'That act is right (or is wrong)'. But there are also universal deontic judgments, such as 'Any act of promise-keeping tends as such to be right' or 'Any act of deliberate deception tends as such to be wrong'. How would the subjective theory deal with the latter? Let us take, e.g., the trans-subjective dispositional form of the subjective theory and consider how it would deal with 'Any act of promise-keeping tends as such to be right'. It would say that this is equivalent to 'Any person contemplating an act which he believed to be one of promise-keeping would tend to have a moral pro-feeling in so far as he confined his attention to that aspect of the act.' No doubt this might require various qualifications, e.g., we might have to substitute 'any normal person' for 'any person' in order to allow for moral lunatics, and we might have to add 'provided he were in a normal state at the time' in order to allow for the possibility of his disposition to have a moral pro-feeling being inhibited or reversed if he were in a state of rage or of jealousy. But the general principle is clear enough.

Next let us suppose that all forms of the subjective theory are dismissed. We should then have to accept some form of what I will call the 'objective theory'. According to this a deontic judgment ascribes to an act a certain quality or relation or relational property which has no reference to the feelings or desires or opinions of the speaker or of anyone else concerning that act. Such judgments would be significant and might be true even if no human being had ever had moral feelings of any kind.

No doubt the objective theory might take many different forms. But for our purpose the most important principle of division is the following. Let us describe an 'ethical sentence' by enumeration as any sentence in which the words 'right' or 'wrong', 'ought' or 'ought not', 'morally good' or 'morally evil' or mere dictionary equivalents of them occur. Now, if the words 'right' and 'wrong' denote opposite forms of a certain objective characteristic, the following possibilities are open about that characteristic, (i) It may be simple and therefore indefinable as, e.g., the quality of sensible yellowness and the relation of temporal precedence are. (ii) It may be complex and therefore definable. If so, it may be definable (a) only by means of ethical sentences or (b) without the use of such sentences. The following alleged definitions of 'right' would illustrate these two possibilities. The first would be exemplified if 'right' could be defined only as 'what it is fitting to approve' or only as 'what is conducive to morally good experiences'. The second would be exemplified if 'right' were definable as 'conducive to social stability' or as 'productive of a balance of pleasant experiences'. I propose to give the name 'naturalistic' to (i) all forms of the subjective theory, and (ii) any form of the objective theory which holds that 'right' and 'wrong' are definable without the use of ethical sentences. I shall give the name 'non-naturalistic' to any form of the objective theory which holds that 'right' and 'wrong' are either indefinable or definable only by means of ethical sentences. For the present purpose it is not important to consider whether this use of 'naturalistic' and 'non-naturalistic' agrees exactly either in extension or in intension with Professor Moore's usage. Before leaving this topic there is one further remark to be made. I think it is fair to say that most competent persons who have reflected on this subject in recent years would agree that the only alternatives worth serious consideration are some form of either (a) the interjectional theory, or (b) the dispositional variety of the subjective theory, or (c) the non-naturalistic variety of the objective theory. Perhaps I should add that under the head of 'competent persons' in this connection I do not include the eminent natural scientists who from time to time take a holiday from their professional labours in order to instruct us in ethical theory.

(2) Alternative Epistemological Theories ofDeontic Cognition

I shall begin by considering singular deontic judgments, i.e. ones of the form 'That act is right (or is wrong)'. Presumably those moralists who hold a moral sense theory intend at least to assert that these judgments are analogous in certain important respects to judgments of sense perception, such as 'That thing is yellow'.

Now the first thing to notice is that two very different accounts may be given of such judgments as 'That thing is yellow'. These may be described as the naively realistic account and the dispositional account. I will now explain these terms.

(i) I think that the plain man in his plainer moments uncritically takes for granted that the very same sensible quality of yellowness which is presented to him when he looks at a bit of gold in white light literally pervades the surface of that bit of gold, not only when he is looking at it in white light, but also and in precisely the same sense when no one is looking at it and when it is in the dark. He believes that looking at the thing and its being illuminated by white light serve only to reveal to him the yellowness which has been there all the time in precisely the form in which it is now presented to him. This is what I call the 'naively realistic interpretation'. Price seems to have thought that this, or something like it, is what plain men believe. He also thought that this belief is not only mistaken, but can be seen to be internally inconsistent by anyone who reflects carefully on the natures of sensible yellowness and of material objects. I must confess that I cannot see this myself.

(ii) A person who makes the judgment 'That thing is yellow' may be expressing only his belief that it would present a yellow appearance to any normal human being who might at any time view it in white light. No doubt a person who accepts the naively realistic interpretation also believes this conditional proposition. But this belief is certainly not the whole of what he expresses by saying 'That thing is yellow', and it might not even be a part of it. It might be for him only a very obvious and immediate consequence of what he expresses by that statement. I give the name 'dispositional account' to the view that the whole meaning of such judgments as 'That thing is yellow' is a conditional proposition of the kind which I have just enunciated.

The next point to notice is this. If a person believes that a certain thing would present a yellow appearance to any normal human being who should at any time view it in white light he does not generally accept this conditional proposition as an ultimate fact. He generally amplifies it as follows. He ascribes to the thing a certain intrinsic property, and he ascribes to each human being a certain other intrinsic property correlated with the former. Let us call these respectively the 'objective' and the 'subjective correlate' in the perception of yellowness. It is held that when and only when a certain relationship is set up between a human being and this thing the subjective correlate in the person and the objective correlate in the thing together cause the thing to present a yellow appearance to the person.

This is common ground to the holders of the naively realistic and of the dispositional account. But there is a profound difference between them in point of detail. On the naively realistic interpretation the objective correlate just is that quality of sensible yellowness which, according to that theory, is spread out over the surface of the thing ready to be presented whenever the appropriate revelatory conditions are fulfilled. The subjective correlate just is the power of prehending the yellowness of yellow things when such conditions are fufilled. That power is activated whenever a person who possesses it stands in a certain bodily and mental relation to a thing which possesses yellowness.

On the dispositional interpretation the objective correlate is generally held to be a certain kind of minute structure and internal agitation in a thing which is not itself literally and non-dispositionally coloured. Again, the subjective correlate is not now the power of prehending the objective correlate. We have no such power. It is the capacity to have sensations of a certain kind, called 'sensations of yellowness'; and these are not prehensions of a quality of yellowness inherent in the thing perceived. There is no such quality. That power is activated whenever a person who possesses it stands in a certain bodily and mental relation to a thing which has this peculiar kind of minute structure and internal agitation.

I do not think that anyone who accepted the dispositional interpretation would give the name 'yellowness' to that minute structure and internal cogitation of a colourless object which, according to him, is the objective correlate of sensations of yellow. He would confine the name 'yellow' to (a) the peculiar sensible quality of certain sensations, e.g., those which he has when he looks at the yolk of an egg in white light, and (b) the dispositional property which certain things have of giving rise to such sensations in a normal human observer when he views them in white light. If he were wise, he would distinguish these two usages of the word as 'sensible' and 'physical' yellowness; or he might prefer the more general phrases 'occurrent' and 'dispositional' yellowness. To the minute structure and internal agitation which are the objective correlate of the perception of things as yellow we might give the name 'physical correlate of yellowness'.

We can now see that the moral sense theory of singular deontic judgments might take two entirely different forms, viz. a naively realistic one and a dispositional one. Both would start from the common ground that there is a peculiar kind of experience which human beings are liable to have when they contemplate certain acts, and that this can take either of two opposite forms, viz. a pro-form and an anti-form. Both would hold that this experience is of the nature of feeling, where 'feeling' is used to include both sensation and emotion as distinguished from thought. From this common basis they diverge as follows:

The naively realistic form of the moral sense theory would take moral feeling to be like what visual sensation is supposed to be on the naively realistic view of visual perception. When a person contemplates a certain act and has a moral pro-feeling in doing so that feeling either is or involves a prehension by him of a certain characteristic, viz. rightness, in the act; and that characteristic belongs literally and non-dispositionally to the act quite independently of whether anyone happens to contemplate it or to have a moral pro-feeling when doing so. (I have used the alternative phrase 'is or involves a prehension' rather than the simpler phrase 'is a prehension' because it might well be held that a moral feeling is never just a prehension of the objective rightness or wrongness of a contemplated act, but is always such a prehension qualified by a certain kind of emotional tone.)

I am fairly certain that the adherents of the moral sense theory did not interpret it in this way; for they did not, I think, put a naively realistic interpretation on visual sense-perception. But some of them may quite likely have thought that plain men mistakenly put this interpretation both on such judgments as 'That act is right' and on such judgments as 'That thing is yellow'. On the other hand, I suspect that Professor Moore when he compared intrinsic goodness with yellowness in Principia Ethica, was tacitly assuming something like the naively realistic interpretation of both such judgments.

The dispositional form of the moral sense theory would take moral feeling to be either (a) a special kind of emotion or (b) a sensation analogous to those of taste or smell and not to those of sight. I suppose that hardly anyone would put a naively realistic interpretation on such perceptual judgments as 'That is bitter' even if he were inclined to put such an interpretation on judgments like 'That is yellow'.

Starting from this basis the theory might take the dispositional form in one or other of its main varieties. The feature common to all of them would be that the moral feeling which a person has when he contemplates an act neither is nor involves a prehension by him of an independent non-dispositional characteristic of rightness inherent in that act. On the trans-subjective variety of this theory a person who says that an act is right means, roughly speaking, no more than that any normal person who should contemplate this act when he was in a normal condition would have a moral pro-feeling. On the intra-subjective variety of the theory the speaker would mean the same kind of thing with 'he himself' substituted for 'any normal person'. I have little doubt that most upholders of the moral sense theory meant to assert the trans-subjective variety of the dispositional form of it. But they did not always make this clear to their readers, and perhaps they were not always clear about it themselves.

It is perhaps worth remarking that the moral sense theory might conceivably take the occurrent intra-subjective form. It might allege that, when a person calls an act right, all that he means is that his present contemplation of it is accompanied by a moral pro-feeling. I think that this form of the theory is so obviously inadequate that supporters of the moral sense doctrine can hardly have meant to assert it. But some of them may have incautiously made statements which would suggest that this is what they meant, and their opponents may sometimes have found it convenient to seize upon these as readily assailable Aunt Sallies. It seems to me that the only two forms of the moral sense theory that are worth serious consideration are the naively realistic form and the trans-subjective variety of the dispositional form. I shall now consider them in turn.

(2.1) Naively Realistic Form of the Moral Sense Theory.

The only kinds of sense-perception which can with any plausibility be interpreted in a naively realistic way are visual and tactual perception. Therefore the naively realistic form of the moral sense theory will have very little to recommend it if singular deontic judgments differ from judgments of visual and tactual perception in just those respects which make a naively realistic interpretation of the latter plausible. It seems to me that the relevant differences are profound and that the analogies are superficial.

(i) In stating the moral sense theory I have so far used the intentionally vague phrase 'having a moral pro-feeling or anti-feeling when one contemplates an action'. If singular deontic judgments are to be analogous to judgments of visual or tactual sense-perception, this must be held to be analogous to having a sensation of yellowness when one looks at the yolk of an egg or having a sensation of coldness and hardness when one touches a block of ice. Is there any such analogy?

We must begin by distinguishing two cases, viz. (a) where one person makes a deontic judgment about an act done by another, and (b) where he makes such a judgment about an act done by himself.

(a) One person never can perceive the act of another, if by 'act' we mean something to which moral predicates can be applied. He can perceive only some bit of overt behaviour on the part of another, e.g., writing a cheque and handing it over to a third person. That bit of overt behaviour may be an act of forgery or of paying a debt or of subscribing to a charity or of bribing an official. As a subject of moral predicates it is a different act according to the different intentions with which it is done. Now one person can contemplate another's intentions only in the sense of making them objects of thought and never in that of perceiving them.

I think that this suffices to wreck the moral sense theory in its naively realistic form as applied to singular deontic judgments made by one person about the acts of another. Even if a naively realistic account of such judgments as 'That thing is yellow' were acceptable, there would be no analogy between them and such judgments as 'That act is right' when the judger and the agent are different. For 'that thing', e.g., a certain bit of gold, is perceived by the person who makes the judgment that it is yellow. The thing is perceived; it is perceived as yellow; and the sensation of yellowness is an essential constituent of the perception of the thing. The naively realistic account of the situation is that the percipient is acquainted with the surface of the thing, and that the latter reveals to the percipient through his sensation of yellow that objective non-dispositional quality of yellowness which it possesses independently of human observers and their sensations. This account is here prima facie highly plausible. But 'that act', if done by another, is not perceived except as a bit of overt behaviour. In respect of those characteristics which make it a possible subject for moral predicates it can only be conceived. The moral feeling, even if it be a sensation and not an emotion only, is not an essential constituent of the perception of the act as a bit of overt behaviour; only visual sensations are essential constituents of that perception. And finally the relation of the moral pro-feeling or anti-feeling to the conception of the act as, e.g., one of debt-paying or one of bribery cannot possibly be like the relation of a sensation to a perception of which it is a constituent, e.g., the relation of a sensation of yellowness to the visual perception of a thing as yellow.

(b) When a deontic judgment is passed by a person on one of his own acts the above criticism does not hold. In performing an act a person is or may be directly aware of his own intentions. He knows it directly as an act of intended bribery or forgery or debt-paying or whatever it may be, and not merely as a bit of overt behaviour of a certain kind. Similarly, in retrospection a person generally knows by personal memory what were his intentions in his own past acts. No doubt introspective self-perception and personal memory are very different in important respects from sense-perceptions. But they agree with it, and differ from one's awareness of the experiences of another person in being ostensibly instances of direct acquaintance with particulars. It seems to me then that, if the Moral Sense Theory in its naively realistic form is to be defended, it must be confined in the first instance to deontic judgments made by a person about his own acts. We might suppose that he derives his notions of rightness and wrongness from perceiving those characteristics in certain of his own acts by means of moral sensations. Once he has acquired the notions in this way he can proceed to apply them to the acts of other persons; although he cannot perceive these and therefore cannot perceive their rightness or wrongness, but can have only conceptual cognition about them.

Now I think that there is a very serious objection to this view. It is certain that I have moral pro-feelings and anti-feelings both when I introspect or remember certain acts of my own and when I conceptually cognize the similar acts of other persons. Now I cannot detect any relevant difference between my moral feelings in the two cases. But, as we have seen, it is impossible in the latter case to hold that there is any analogy to visual sense-perception as interpreted by the naively realistic theory. It is impossible to hold here that the moral feeling is a state of acquaintance with an objective characteristic of rightness or wrongness in the cognized act. Therefore it seems unreasonable to suppose that the precisely similar moral feelings which one has when introspectively perceiving or remembering one's own acts is susceptible of a naively realistic interpretation.

I pass now to another profound prima facie difference between singular deontic judgments and judgments of visual or tactual perception. If I judge that a certain act is right or that it is wrong, it is always sensible for anyone to raise the question 'What makes it right or makes it wrong?' The answer that we expect to such a question is the mention of some non-ethical characteristic of the act, e.g., that it is an act of promise-keeping, of giving a false answer to a question, and so on. Let us call these 'right-inclining' and 'wrong-inclining' characteristics. Now the connection between the presence of any of these non-ethical characteristics and the tendency of an act to be right or to be wrong seems to be necessary and self-evident, not causal and contingent. (I say the 'tendency to be right or to be wrong' and not just 'rightness' or 'wrongness' for a reason which will be familiar to all readers of Ross's ethical writings. One and the same act may be, e.g., an act of truth-telling and one of betrayal. It is not self-evident that such an act is resultantly right or resultantly wrong. But it might well be held to be self-evident that it tends to be right in respect of being an act of truth-telling and to be wrong in respect of being one of betrayal, and that it would be right if it had no wrong-inclining characteristic and would be wrong if it had no right-inclining characteristic. These points were made clearly enough by Price, but have since been made much more clearly by Ross.)

Now the fact which I have just mentioned is relevant to both forms of the moral sense theory, but for the present we are concerned only with the naively realistic form of it. If I look at a thing and judge it to be yellow, it is not particularly sensible to ask 'What makes it yellow?' The question is sensible only if it is interpreted causally, e.g., in some cases the answer might be that it contains saffron. And a more ultimate answer would be that it has such and such a minute structure and internal agitation. Now on the naively realistic theory the thing is pervaded literally and non-dispositionally by an inherent quality of yellowness; and there is no self-evident necessity for all things which have a certain kind of minute structure and internal agitation and only such things to be pervaded by yellowness. It is simply a contingent general connection between two sets of properties of a material thing, viz. certain geometrical and kinematic properties, on the one hand, and a certain objective colour, on the other. The connection between being an act of promise-breaking and tending to be wrong does not seem to be in the least like this.

It is worth while to remark before leaving this topic that, even if our cognition of the rightness or wrongness of acts were analogous to visual or tactual perception interpreted in the naively realistic way, it is quite certain that our cognition of right-inclining and wrong-inclining characteristics is not. Such characteristics as being an intentional breach of promise, an intentional return of a borrowed article, and so on, are highly complex relational properties. They can be cognized only conceptually; it is non-sensical to suggest that they could be cognized by anything analogous to sense-perception or to introspective self-perception.

On the other hand, the fact, if it be a fact, that the connection between certain non-ethical characteristics and the tendency to be right is necessary and self-evident is not in itself a reason for denying that rightness and wrongness are cognized by something analogous to sensation interpreted in a naively realistic way. For the connection between having shape and having size is necessary and self-evident, and yet both these characteristics are cognized by visual sense-perception.

I think that the upshot of this discussion is that there is little to be said for and much to be said against the moral sense theory in its naively realistic form as applied to deontic judgments. We can therefore pass to the dispositional form of the theory.

(2.2) Dispositional Form of the Moral Sense Theory.

I do not think that we shall be unfair to the theory if we confine our attention to the trans-subjective variety of it and if we assume that moral feeling is of the nature of emotion rather than sensation.

I shall begin with some general remarks about emotion, (i) An emotion, e.g., an experience of fearing or hating, as distinct from an emotional mood, such as a state of apprehension or of crossness, is always directed to a cognized object. This may be real or hallucinatory, e.g. one may be afraid of a real man who is pointing a revolver at one or of an hallucinatory appearance of such a man in a dream. Again, if the object be real, it may be correctly or more or less incorrectly cognized e.g. one may be afraid of a real physical object which one sees when crossing a field in twilight and takes to be a man pointing a revolver at one, and this object may really be a harmless scarecrow.

(ii) We must distinguish between what I will call 'mediated' and 'unmediated' emotions. Sometimes when a person feels a certain emotion towards a certain object he has an experience which may be described as feeling that emotion towards that object in respect of certain characteristics which he believes (rightly or wrongly) that it possesses. In that case I shall say that his emotion is mediated by this belief about the characteristics of the object, and I shall call these characteristics the 'mediating characteristics' of the emotion. Often, however, the emotion is not felt in respect of any characteristic which the experient believes the object to have. In that case I shall say that the emotion is unmediated. If I am angry with a person, e.g., I may feel this anger in respect of some fault which I believe (rightly or wrongly) that he has committed. But I may feel angry with a person, and still more obviously I may dislike him, just directly and, as we say, 'for no assignable reason'. This is an example of an unmediated emotion.

(iii) Presumably every occurrence of any emotion, whether mediated or unmediated, has a total cause. In many cases, no doubt, an essential factor in that cause is the presence of certain characteristics in the object. I will call these 'evoking characteristics'. In the case of a mediated emotion the evoking and the mediating characteristics may be, and no doubt often are, wholly or partly the same. But very often they must be different; for the object often does not really have the characteristics which the experient believes it to have and in respect of which he feels his emotion towards it.

(iv) It is commonly held that certain kinds of emotion are in some sense 'appropriate to' objects which have certain characteristics, and that they are 'inappropriate to' objects which lack these or which have certain others, e.g., fear is held to be appropriate only to objects which are dangerous. Again, it is held for a given degree of dangerousness there is, within fairly narrow limits, a fitting degree of fear. To fear objects which are not really dangerous is described as 'irrational'; and to fear intensely objects which are only slightly dangerous is described as 'inordinate'.

It is a well-known fact that if a person begins by feeling an unmediated emotion towards an object he is very liable to go on to ascribe to that object such characteristics as would make the emotion appropriate and to ascribe to those characteristics such a degree as would make his emotion ordinate. A very familiar example of this is provided by persons who are jealous of others. Lastly, if a person feels a mediated emotion towards an object in respect of a characteristic to which that emotion is inappropriate, he is very liable to divert his attention from this fact and to ascribe to the object another characteristic in respect of which the emotion would be appropriate. These tendencies, which have been perfectly familiar to playwrights, preachers and plain men throughout the ages, have been hailed as great discoveries of modern psychology under the name of 'rationalization'.

We are now in a position to consider the trans-subjective dispositipnal form of the moral sense theory. In essence the theory is that such judgments as 'That act is right (or is wrong)' are analogous to such judgments as 'That food is nice (or is nasty)'. The correct analysis of them is some variant on the formula 'That act would evoke a moral pro-emotion (or anti-emotion) in any human being who might at any time contemplate it'. There might have to be qualifications about the individual being 'normal' and being 'in a normal state', but we need not trouble about them at present.

Now this form of the theory does avoid the first objection which I made against the naively realistic form of it. It does not have to assume that one person literally has knowledge by acquaintance of the intentions of another. It does not have to assume that the experience of having a moral feeling when contemplating an act of one's own is fundamentally different in kind from that of having a moral feeling when contemplating a similar act of another person. For we can and do have emotions towards objects which are cognized only conceptually, and we can and do feel such emotions in respect of characteristics whose presence is only conceived and not perceived.

It seems to me that the main difficulties of the theory can be summed up in the following three questions: (i) Can it deal with the fact that judgments like 'That act is right' seem always to be grounded upon the supposed presence in the act of some non-ethical right-inclining characteristics, such as being the fulfilment of a promise? (ii) If so, can it deal with the further fact that the connection between a right-inclining characteristic and the rightness which it tends to convey seems to be necessary and synthetic? And (iii) can it deal with the fact that it seems not only intelligible but also true to say that moral pro-emotion is felt towards an act in respect of the characteristic of rightness and moral anti-emotion in respect of the characteristic wrongness? I shall take these three questions in turn.

(i) I think that a fairly plausible answer, so far as it goes, can be made to the first question. We shall have to say that the right-inclining characteristic which is the ground of the judgment 'That act is right' just is the mediating characteristic of the moral pro-emotion which is felt towards such acts. To say that every moral judgment is founded upon some non-ethical characteristic of the act which is its subject will be equivalent to saying that every moral emotion is a mediated emotion. Such characteristics as being an act of promise-keeping will be mediating characteristics for moral pro-emotion; such characteristics as being an act of lying or of deliberate cruelty will be mediating characteristics of moral anti-emotion.

It should be noticed that the theory can account quite plausibly for the facts which Ross describes under the head of his distinction between 'prima facie duties' and 'a duty proper'. (I prefer to use the phrases 'components of obligation' and 'resultant obligation'.) An act is known or believed to have various characteristics, e.g. to be an act of truth-telling, a breach of confidence, and an optimific act. The first and the third of these features give rise to components of obligation of various degrees of urgency towards doing it; the second gives rise to a component of a certain degree of urgency against doing it. According to circumstances the resultant obligation may be to do it or to avoid doing it. Now it is a perfectly familiar fact that an object may have several characteristics, and that it may call forth an emotion of one kind in respect of some of them and an emotion of the opposite kind in respect of others; so that the emotion towards the object as a whole may be predominantly of the opposite kind. The present theory would say that we tend to feel a moral pro-emotion of a certain strength towards the act in respect of its being one of truth-telling and in respect of its being optimific; that we tend to feel a moral anti-emotion of a certain strength towards it as being a breach of confidence; and that our moral emotion towards it as a whole is the resultant of these two tendencies, and may be either predominantly pro or predominantly and according to circumstances.

(ii) The second question is much harder. It is alleged, e.g., that the proposition 'Any act of promise-keeping tends as such to be right, and any act of promise-breaking tends as such to be wrong' is necessary, self-evident, and synthetic. On the present theory of deontic judgments this would be equivalent to something like the following proposition: 'It is necessary, self-evident and synthetic that any human being who should contemplate an act which he believed to be one of promise-keeping would tend to feel a moral pro-emotion towards it, and that he would tend to feel a moral anti-emotion towards any act which he believed to be one of promise-breaking'.

Now it might be objected that the latter statement is certainly false. It is a purely contingent fact that human beings have a disposition to feel moral emotions at all. They might have been as devoid of them as they are of a disposition to have special sensations in presence of magnets. Moreover, granted that they do have such an emotional disposition, it is a purely contingent fact that moral emotions are mediated in the particular ways in which they are. It is quite conceivable that the belief that an act is one of promise-keeping should have mediated a moral anti-emotion, and that the belief that it is one of promise-breaking should have mediated a moral pro-emotion; just as it is conceivable that men should have liked the taste of castor oil and disliked that of sugar. In that case, on the present theory, promise-breaking would have tended to be right and promise-keeping to be wrong; just as castor oil would have been nice and sugar nasty.

So the objection comes to this. If the present form of the moral sense theory were true, certain propositions which are in fact necessary and knowable a priori would have been contingent and knowable only empirically. Therefore the theory is false. I am sure that this is the most important of Price's objections to the moral sense theory, though I have developed it in my own way. What are we to say about it?

It is plain that there are only two lines of defence open to the present form of the moral sense theory, (a) One is to argue that propositions like 'Any act of promise-keeping tends as such to be right' are not necessary. (b) The other is to argue that propositions like 'Any human being who should contemplate an act which he believed to be one of promise-keeping would tend to feel a moral pro-emotion towards it' are not contingent. Let us consider the two alternatives in turn:

(a) I think that this line of argument would divide into two parts, which might be called the 'offensive' and the 'defensive'. The offensive part would take the opposite view as a hypothesis and try to show that it is untenable. The defensive part would try to explain why certain propositions which are in fact empirical and contingent appear to many people to be a priori and necessary.

(a, a) The offensive part may be put as follows: What precisely do our opponents maintain? If we may take Price as their ablest representative, they seem to assert something like the following doctrine. Suppose that a person reflects, e.g., on the situation of being asked a question and on the notions of responding to it by a true answer and responding to it by a false answer. Then he will find it self-evident that the former kind of response has a certain relation of 'moral fittingness' and that the latter has an opposite relation of 'moral unflttingness' to such a situation. This relation of moral fittingness or unflttingness is held to be unique and unanalysable. And the process of recognizing that it necessarily holds between certain kinds of response and certain kinds of situation is held to be analogous to that of recognizing that certain mathematical terms, e.g. stand in certain mathematical relations.

Now the objection which will be made by supporters of the moral sense theory is twofold. It will be said that the doctrine just enunciated involves a priori concepts and synthetic a priori judgments, and that neither of these is admissible. We will take these two points in order.

If there is a simple unanalysable relation of moral fittingness or unfittingness, it is certainly not manifested to us by any of our senses. We literally see that one coloured patch is surrounded by another; we literally hear that two notes, sounded together or in very close succession, concord or discord with each other; and so on. In such cases we presumably derive our ideas of the relation of surrounding and the relation of concording or discording by comparison and abstraction from such sensibly presented instances of terms standing in these relations. It is plain that we do not acquire the idea of moral fittingness or unfittingness in this way. Nor do we derive the idea from instances of terms presented to us by introspection as standing in that relationship. Introspection presents us with certain of our own experiences as standing in certain temporal relations, e.g., as being in the same specious present and partly overlapping in time, and so on. Again, since the relation of moral fittingness or unfittingness is held to be simple and unanalysable, the idea of it cannot be one which we have constructed in thought from elements presented separately or in different contexts by sensation or introspection or both. (The idea of the complex relationship of a colonel to the subordinate officers of his regiment, e.g., is no doubt reached in some such way as this.) But it is held by many philosophers to be a fundamental epistemological principle that every idea is either derived by abstraction from instances presented in sensation or introspection or is an intellectual construction from elements so derived. If this principle be admitted, it is impossible that we should have any conception of the relations of moral fittingness and unfittingness as described by such moralists as Price.

For my part I attach very little weight to this argument. I can see nothing self-evident in what I will call for short 'Hume's epistemological principle', and I am not aware that any conclusive empirical evidence has been adduced for it. It seems to me to be simply a useful goad to disturb our dogmatic slumbers, and a useful guide to follow until it begins to tempt us to ignore some facts and to distort others. I am inclined to think that the concepts of cause and of substance are a priori or contain a priori elements; at any rate I have never seen any satisfactory account of them in accordance with Hume's principle.

The second point in the offensive part of the argument is this: Suppose, if possible, that 'right' and 'wrong' are simple unanalysable notions, as Price, e.g., held them to be. Then any proposition which asserts a connection between some non-ethical characteristic, such as promise-keeping, and tendency to be right must be synthetic. Now a proposition may be synthetic and contingent or analytic and necessary, but it is an admitted general principle that no proposition can be both synthetic and necessary. Therefore the combined doctrine that 'right' and 'wrong' are unique unanalysable notions and that such propositions as 'Any act of promise-keeping tends as such to be right' are necessary must be false.

Such an argument would have different effects on different persons. Suppose that A and B are both quite convinced up to a certain moment of the truth of a certain general principle, and suppose that at that moment C brings to their notice an apparent counter-instance. If each is to be self-consistent, something will have to give way in each of them. But it need not be the same something. A may remain completely certain of the general principle; he will then have to maintain that the instance is only apparently contrary to it and explain why it seems to be so. B may find it impossible to doubt that the instance is contrary; he will then be forced to give up the general principle and explain why it seemed evident. These are the two extreme possibilities. Between them are numberless possible intermediate alternatives, where the person concerned is led to feel some doubt of the unqualified truth of the principle and some doubt whether the apparent counter-instance really conflicts with it. Speaking for myself, I occupy one of these intermediate positions. As for Price, he would have been completely unmoved by this kind of argument. For he held, in full knowledge of Hume's doctrine and in conscious opposition to it, that there are plenty of synthetic necessary facts in other departments beside that of morals. For these reasons I think that it is rather futile to rely on a general argument of this kind.

(a, b) The defensive part of the argument might take the following line. Civilized men throughout human history have been assiduously conditioned in infancy and youth by parents, nurses, schoolmasters, etc., to feel moral pro-emotions towards acts of certain kinds and to feel moral anti-emotions towards acts of certain kinds. Moreover, if we consider what kinds of acts are the objects of moral pro-emotions and what kinds are the objects of moral anti-emotions we notice the following facts about them. The former are acts whose performance by most people on most occasions when they are relevant is essential to the stability and efficient working of any society. The latter are acts which, if done.on many occasions and by many people, would be utterly destructive to any society. On the other hand, the former are acts which an individual is often strongly tempted to omit, and the latter are acts which he is often strongly tempted to commit. This is either because we have strong natural impulses moving us to omit the former and to commit the latter, or because the attractive consequences of the former and the repellent consequences of the latter are often remote, collateral, and secondary. It follows that any group of men in which, from no matter what cause, a strong pro-emotion had become associated with acts of the first kind and a strong anti-emotion with acts of the second kind would be likely to win in the struggle for existence with other groups in which no such emotions existed or in which they were differently directed. Therefore it is likely that most of the members of all societies which now exist would be descendants of persons in whom strong moral pro-emotions had become attached to acts of the first kind and strong anti-emotions to acts of the second kind. And most existing societies will be historically and culturally continuous with societies in which such emotions had become attached to such acts. These causes it might be argued, conspire to produce so strong an association between such emotions and such acts in most members of every existing society that the connection between the emotion and the act seems to each individual to be necessary.

No doubt this line of argument will produce different effects on different persons. For my own part I am inclined to attach a good deal of weight to it.

(b) I pass now to the second kind of defence which might be made for the dispositional form of the moral sense theory. This is to contend that the proposition about human emotional dispositions which, according to the theory, is equivalent to 'Any act of promise-keeping tends as such to be right' is necessary. It might be thought that this contention is so palpably absurd as not to be worth putting forward. But I believe that a case can be made for it, and I propose to make it.

We must begin by noting that the proposition which is equivalent to 'Any act of promise-keeping tends as such to be right' could not with any plausibility be taken to be the crude unqualified proposition 'Any human being has a disposition to feel a moral pro-emotion whenever he contemplates an act which he believes to be one of promise-keeping'. So far from being necessary the latter proposition is not even true. To make it true it will have to be qualified somewhat as follows. We must substitute for it the proposition 'Any normal human being has a disposition to feel a moral pro-emotion towards any act which he believes to be one of promise-keeping if he contemplates it when he is in a normal state'.

Now it might be argued that, when the proposition is thus qualified, it is necessary. For, it might be said, it has then become analytic. It is part of the definition of a 'normal' human being that he has a disposition to feel moral emotion, and that he will feel that emotion in its pro-form towards acts which he believes to be ones of promise-keeping, of truth-telling, or beneficence, and so on. And it is part of the definition of 'being in a normal state' that when one is in such a state this moral-emotional disposition will not be inhibited altogether or excited in abnormal ways.

No doubt the immediate answer which an opponent of the moral sense theory would make to this contention is the following: He would say that such propositions as 'Any act of promise-keeping tends as such to be right' are not only necessary but synthetic. The defender of the dispositional form of the moral sense theory has shown that, on his analysis, they would be necessary only at the cost of showing that they would be analytic. This answer is correct so far as it goes, but I think that the defender of the moral sense theory could rebut it as follows.

The fact is that it is often by no means easy to say whether a proposition is analytic or not. The analytic propositions of real life are not like the trivial examples in logic-books, such as 'All Negroes are black' or 'All right angles are angles'. The following are much better worth considering, e.g. 'The sun rises in the east'. 'A freely suspended magnet sets itself with its axis pointing north and south', and 'Pure water boils at 100° C under a pressure of 76 centimetres of mercury'. The first of these is analytic if 'east' and 'west' are defined by means of the sun, and synthetic if they are defined by means of the magnetic or the gyroscopic compass. The second is analytic if 'north' and 'south' are defined by means of the magnetic compass, and synthetic if they are defined by means of the sun or the gyroscopic compass. The third might be taken as a definition of '100° C'. But if that term were defined in some other way, e.g., thermo-dynamically, as on Lord Kelvin's absolute scale, it might be regarded as an analytic proposition about pure water. For an important element in the definition of 'pure water' is that it has a certain boiling point under certain standard conditions.

Two important points emerge from these examples. The first is that the same type-sentence may express both an analytic and a synthetic proposition, and that a person who uses several tokens of this type even in a single discourse may sometimes be expressing the analytical and sometimes the synthetic proposition. The former is necessary and the latter is contingent. It would not be surprising if a person should sometimes become confused in such cases and think that every token of this type expresses one and the same proposition which is both synthetic and necessary.

The second point is this. Such an analytic proposition as 'Pure water boils at 100° C under a pressure of 76 centimetres of mercury' has at the back of it a whole system of interconnected empirical generalizations, apart from which it would never have been worth anyone's while to formulate it. It would take me far too long even to begin to state a few of these empirical generalizations. It will suffice to say that they are all represented in the various qualifications which make the proposition 'Pure water boils at 100° C under a pressure of 76 centimetres of mercury' analytic.

Now it might be suggested that facts like these throw some light on the alleged synthetic necessity of such propositions as 'Any act of promise-keeping tends as such to be right', and on the claim of defenders of the dispositional form of the moral sense theory that the equivalent propositions about human emotional dispositions are necessary because analytic.

The proposition 'Any act of promise-keeping would tend to call forth a moral pro-emotion in any normal human being who might contemplate it when in a normal state' is obviously rather like the proposition 'Any sample of pure water boils at 100° C under the normal atmospheric pressure, i.e. 76 centimetres of mercury'. Just as the latter is analytic, but is founded on a whole mass of interconnected empirical generalizations, so is the former. I will now try to justify this statement.

It is an empirical fact that the vast majority of men have a disposition to feel moral emotions, and that the minority who lack it differ in many other ways from the majority of their fellows. It is an empirical fact that there is very substantial agreement among men in the kinds of act which call forth moral pro-emotion and in the kinds which call forth moral anti-emotion. The small minority of men who habitually feel moral pro-emotion where most of their fellows feel moral anti-emotion, or vice versa, are generally found to be odd and abnormal in many other ways. There is, in fact, so high a degree of positive association between moral and non-moral normality that it would make very little difference in practice whether we defined a 'normal' man solely by reference to his moral dispositions or solely by reference to his non-moral dispositions, or by reference to a mixture of both. But the proposition that any normal human being would tend to feel a moral pro-emotion towards any act which he believed to be one of promise-keeping would be synthetic if one defined 'normality' solely by reference to non-moral dispositions, whilst it might well be analytic if one defined it wholly or partly in terms of moral dispositions.

Again, there is a very high degree of positive association between the tendencies to feel moral pro-emotion towards acts of promise-keeping, of truth-telling, of beneficence, etc.; and there is perhaps an even stronger degree of positive association between the tendencies to feel moral anti-emotion towards acts of treachery, of unfairness, of cruelty, etc. Therefore it would make little practical difference which of these mediating characteristics was included and which was omitted from the definition of 'normality'. Now, if the tendency to feel moral pro-emotion towards any act which is believed to be one of promise-keeping were 'included in the definition of 'normality', the proposition that any normal man would tend to feel such an emotion towards such acts would be analytic; whilst, if this were omitted and 'normality' were defined by reference to some of the other mediating characteristics of moral emotion, this proposition would be synthetic.

It therefore seems likely that, if the analysis which the dispositional form of the moral sense theory offers for such propositions as 'Any act of promise-keeping tends as such to be right' were correct, a sentence of this type might often express a proposition which is analytic and necessary and might as often express one that is synthetic and contingent. If so, it is not unlikely that a confusion should arise and that it should be thought that every such sentence expresses one and the same proposition which is both necessary and synthetic.

It remains to say something of the qualification 'when in a normal state', which has to be added to make the statement universally true, and which at the same time makes it more nearly analytic. It may be compared to the qualifications about the water being pure and the barometric pressure being normal in my example about boiling-point.

At the back of this qualification lie certain negative and certain positive empirical facts. It is found that a person who generally does feel moral pro-emotions towards acts of certain kinds and moral anti-emotions towards acts of certain other kinds will on some occasions not do so. He may feel no moral emotion; or perhaps on very exceptional occasions the normal form of his moral emotion may be reversed. These are the negative facts. The positive facts are certain empirical generalizations about the kinds of occurrent conditions under which such inhibitions or reversals of moral emotion tend to take place. 'Being in a normal state' is then defined in terms of the absence of such conditions, e.g. not being angry with or jealous of or frightened by the agent whose act is being contemplated. Now, although one has at the back of one's mind a fairly adequate but rather confused idea of these negative conditions, only one or two of them will be explicitly before one's mind on any particular occasion when one uses the expression 'in a normal state'. According as one or another is in the foreground on a given occasion the same sentence may express an analytic or a synthetic proposition.

I suggest, then, that defenders of the dispositional form of the moral sense theory might attempt in some such ways as these to rebut the objection that, whilst propositions like 'Any act of promise-keeping tends as such to be right' are necessary and synthetic, the propositions which it asserts to be their equivalents are either contingent or analytic.

(iii) The third difficulty which the moral sense theory, in the form of it which we are considering, has to meet is this. It might be alleged that the mediating characteristics in respect of which a person feels moral pro-emotion or anti-emotion towards an act which he contemplates are the supposed rightness or wrongness of the act. Suppose, e.g., that a person feels a moral anti-emotion when he contemplates an act which he believes to be one of promise-breaking. Then, it might be said, he does so only in so far as he believes promise-breaking to be wrong. Suppose that he believed the act to be one of promise-breaking but did not believe that such acts tend to be wrong. Then, it might be alleged, there is no reason to think that he would feel a moral anti-emotion towards it.

Let us begin by considering what view a Rationalist, like Price, would take on this question of the mediating characteristics of moral emotion. I think that the following is fair statement of his position. It is a necessary proposition that any rational being who contemplated an act which he believed to be one of promise-breaking would tend to feel towards it a moral anti-emotion. But, though true and necessary, it is not self-evident. It is a logical consequence of two more fundamental propositions, each of which is self-evident. They are these: (a) It is self-evident to any rational being that any act of promise-breaking tends as such to be wrong. (b) It is self-evident that any rational being who contemplated an act which he believed to be wrong would feel towards it a moral anti-emotion.

We have already considered what the supporters of the moral sense theory might say about the first of these propositions. What are we to say about the second? It seems to me that everything depends here on how much we put into the connotation of the phrase 'rational being'. On a narrower interpretation of that phrase proposition (b) is synthetic but contingent, on a certain wider interpretation that proposition becomes necessary but analytic. Sometimes the one interpretation and sometimes the other is at the back of one's mind without one realizing the fluctuation, and so one is inclined to think that proposition (b) is both necessary and synthetic.

A 'rational being', on the narrowest interpretation, means roughly one who is capable of comparing, abstracting, and forming general notions; who is capable of seeing necessary connections and disconnections between terms and between propositions; and who has the power of making inferences, both deductive and inductive. I call this the 'narrowest' interpretation, because it takes account only of cognitive characteristics and leaves out emotional and conative ones. The next stage in widening it would be to include in the definition of a 'rational being' what I will call 'purely intellectual' emotions and conations, e.g. intellectual curiosity, taking pleasure in neat arguments and displeasure in clumsy ones, desire for consistency in one's beliefs, and desire to apportion the strength of one's beliefs to the weight of the evidence.

Let us say that a person who had the cognitive, conative and emotional dispositions which I have just enumerated would be rational 'in the ethically neutral sense'. Suppose that Price were correct in thinking that moral fittingness and unfittingness are relations which hold of necessity between certain types of response and certain types of situation. Then a person who was rational in the ethically neutral sense would in principle be capable of having ideas of right and wrong and of making moral judgments. (I say 'in principle' because (a) he would, by definition, have the general capacity to see necessary connections between terms and between propositions, whilst (b) it might happen that his insight in this particular department was lacking, as that of some rational beings is in the department of mathematical relations.) But, so far as I can see, there would not be the slightest inconsistency in supposing that a being who was rational in the ethically neutral sense, and did in fact have the ideas of right and wrong and make moral judgments, was completely devoid of specifically moral emotion and conation. The fact that he knew or believed a to be right and b to be wrong might arouse in him neither moral pro-emotion towards the former nor moral anti-emotion towards the latter, and it might not evoke in him the slightest desire to do a or to avoid doing b or vice versa. I cannot see any logical impossibility in the existence of such a being; whether it would involve a conflict with some of the de facto laws of psychology I do not know.

Now the vast majority of the beings whom we know to be rational in the ethically neutral sense do in fact feel moral pro-emotion towards acts which they believe to be right and moral anti-emotion towards those which they believe to be wrong, and they are in fact to some extent attracted towards doing the former and repelled from doing the latter. Moreover, it is logically impossible that these specifically moral emotions and desires should exist in a being who was not rational in the ethically neutral sense; for their characteristic objects can be presented only by a process of reflective thinking. The wider interpretation of the phrase 'rational being' includes these specifically moral conative and emotional characteristics in addition to those which constitute the definition of 'rational' in the ethically neutral sense. It is, of course, logically impossible that a person who is rational in this widest sense should fail to feel moral pro-emotion towards what he believes to be right and moral anti-emotion towards what he believes to be wrong. But this is a merely analytical proposition. It is synthetic and contingent that a person who is rational in the ethically neutral sense should be so in the wider ethical sense also. But the fact that rationality in the ethically neutral sense is almost invariably accompanied in our experience by the additional features which convert it into ethical rationality and the fact that the latter logically entail the former produce a confusion in our minds. We are thus led to think that the proposition that any rational being would feel a moral pro-emotion towards any act which he believed to be right and a moral anti-emotion towards any that he believed to be wrong is both necessary and synthetic.

So much for the Rationalist account of moral emotion and its mediation by the characteristics of rightness and wrongness. What can the moral sense theory, in its trans-subjective dispositional form, make of the alleged facts?

On the face of it this theory is presented with the following difficulty. Suppose that we try to combine the alleged fact that rightness and wrongness are the mediating characteristics for moral emotion with the analysis of moral judgments given by the theory in question. Then we seem to be committed to the following proposition: 'A person will tend to feel a moral anti-emotion towards an act which he believes to be one of promise-breaking so far and only so far as he believes that most persons when in a normal condition would feel such an emotion in contemplating such an act'. Now this has a. prima facie appearance of circularity; and, even if it be neither logically nor causally circular, it certainly does not seem very plausible.

The first remark that I have to make is that the objection just stated rests on a premise which is plausible but false. It tacitly assumes that, if the correct analysis of the proposition 'S is P' is 'S is p₁-and-p₂', then anyone who is believing the former proposition is ipso facto believing the latter. Now there may be some sense of 'believe' in which this is true; but there certainly is an important sense in which it is false. It is quite obvious that a number of persons who accept different and incompatible analyses of a proposition may all believe it; and therefore there must be a sense in which some at least of them believe it without ipso facto believing the proposition which is its correct analysis. This is particularly obvious in the present case. Nearly everyone believes that acts of promise-breaking tend as such to be wrong; but some of these persons think that wrongness is a simple characteristic, others think that it can be analysed in one way, and others think that it can be analysed in various other ways. So, even if the correct analysis of 'x is wrong' is 'Any normal person who should contemplate such an act as x when in a normal state would feel a moral anti-emotion towards it', it does not follow that the correct analysis of 'A believes that x is wrong' is 'A believes that any normal person who should contemplate such an act as x when in a normal state would feel a moral anti-emotion towards it'. So it is not fair to say that the moral sense theory must hold that anyone who feels a moral anti-emotion towards an act in respect of his belief that it is wrong is ipso facto feeling that emotion in respect of his belief that any normal person who would feel such an emotion if he were to contemplate such an act while in a normal state.

I suppose that this argument would be generally admitted as applied to the case of a person who did not accept, or did positively reject, the analysis of moral judgments proposed by the moral sense theory. But it might be said that it will not apply to the case of a person who accepts that analysis. I think, however, that even this could be questioned. A person may have assented to a certain analysis of a proposition when the question of its analysis and the arguments pro and con were before his mind. He may continue to accept it, in the dispositional sense that he would assent to it again at any time when the question was raised for him. But during the intervals he may often have the experience of believing the proposition without thinking of the analysis of it which he has accepted. Therefore it seems to me that even an adherent of the moral sense theory might often feel a moral anti-emotion towards an act in respect of his belief that it is wrong without ipso facto feeling that emotion in respect of the belief that it has those characteristics which he holds to be the correct analysis of 'being wrong'.

So much for the dialectics of the matter. But what is really happening when a person is said to feel a moral pro-emotion or anti-emotion towards an act in respect of his belief that it is right or that it is wrong? We must begin by distinguishing what I will call 'first-hand' and 'second-hand' emotion. Suppose that a certain word has been very often used in connection with objects towards which a certain kind of emotion has been felt and that it has seldom or never been used except on such occasions. Then this word may come to act as a stimulus calling forth this kind of emotion. When the emotion is evoked in this way I call it 'second-hand'.

Now there is no doubt that a great deal of moral emotion is, in this sense, second-hand. And there is no doubt that the words which have come by association to act as evokers of second-hand moral emotion are the words 'right' and 'wrong'. When a person is said to feel a moral emotion towards an act in respect of his belief that it is right or that it is wrong what is really happening is very often the following. He knows or believes that acts of this kind are commonly called 'right' or called 'wrong'. He repeats these words sotto voce to himself or has auditory images of them when he thinks of the act in question; and by association they evoke a second-hand moral pro-emotion or anti-emotion towards the act. Plainly there is nothing in this to cause difficulty to the supporters of the moral sense theory.

But of course this does not cover the whole field. There is first-hand moral emotion; indeed, if no one had ever felt a first- hand emotion of a given kind, it is difficult to believe that anyone could now feel a second-hand emotion of that kind. What is happening when a person is said to be feeling a first-hand moral emotion towards an act in respect of his belief that it is right or that it is wrong? I can give only a very tentative answer to this question, based on my own imperfect introspection of a kind of situation with which I am not very familiar.

It seems to me that in such cases I do not first recognize or think that I recognize a quality or relation of rightness or wrongness in the act, and then begin to feel a moral pro-emotion or anti-emotion towards it in respect of this knowledge or belief. What I seem to do is to consider the act and its probable consequences under various familiar headings. 'Would it do more harm that good? Would it be deceitful? Should I be showing ingratitude to a benefactor if I were to do it? Should I be shifting on to another person's shoulders a burden or a responsibility which I do not care to bear for myself?' In respect of each of these aspects of the act and its consequences I have a tendency to feel towards the act a certain kind of moral emotion of a certain degree of intensity. These emotional dispositions were largely built up in me by my parents, schoolmasters, friends and colleagues; and I know that in the main they correspond with those of other persons of my own nation and class. It seems to me that I call the act 'right' or 'wrong' in accordance with my final moral-emotional reaction to it, after viewing it under all these various aspects, and after trying to allow for any permanent or temporary emotional peculiarities in myself which may make my emotional reaction eccentric or unbalanced. By the time that this has happened the features which I had distinguished and had viewed and reacted to separately have fallen into the background and are again fused. They are the real mediating characteristics of my moral pro-emotion or anti-emotion; but I now use the omnibus words 'right' or 'wrong' to cover them all, and say that I feel that emotion towards the act in respect of my belief that it is right or that it is wrong.