Reprinted in Broad's Critical Essays in Moral Philosophy, ed. David R. Cheney, 1971 from Festskrift tillagnad Karl Olivecrona, Stockholm: AB P. A. Norstedt & Soners Forlag, 1964.
This paper constitutes a revised version of Broad's earlier paper 'Imperatives, Hypothetical and Categorical', The Philosopher, Vol. 2, September 1950, pp. 62-75.
OBLIGATIONS, ULTIMATE AND DERIVED
C. D. Broad
I shall call any sentence in the indicative mood in which the word 'ought', or any obviously equivalent word or phrase, such as 'is under an obligation to', 'has a duty to', etc., occurs as the principle verb, a deontic indicative. Examples are: 'I ought to go to the dentist', 'You ought not to eat peas with a knife', 'He ought to make an allowance to his old nurse', 'Persons who have borrowed money ought to repay it to the lender at the agreed date', 'There ought to be laws against cruelty to animals', and 'A fountain-pen ought not to be continually making blots'.
The first point to be noticed is that these sentences may be divided into two classes in the following way. Some of them assert of a person that he ought (or ought not) to do so-and-so. Others assert of a conceivable state of affairs that it ought to be, or of an actual state of affairs that it ought not to be. We can thus distinguish two important classes of deontic indicatives, viz. 'ought-to-do' ones and 'ought-to-be' ones. (It may be remarked that the sentence about fountain-pens falls somewhere between these two classes; for it asserts of a set of inanimate objects that they ought not to behave in a certain way.)
In each of these two classes of deontic indicatives we can distinguish broadly between those in which 'ought' occurs in a specifically moral sense and those in which it occurs in a non-moral sense. In the sentences 'He ought to make an allowance to his old nurse' and 'There ought to be laws against cruelty to animals' the word 'ought' is plainly used in a specifically moral sense. In the sentences 'You ought not to eat peas with a knife' and 'A fountain-pen ought not to be continually making blots' it is plainly used in a non-moral sense. I have no doubt that there are plenty of marginal cases, between these two extremes, where one might reasonably hesitate to say whether the sense in which 'ought' is used is or is not specifically moral.
In what follows I shall be concerned wholly with ought-to do indicatives, in which the grammatical subject is the name of a person or a class of persons. I shall not be considering ought-to-be indicatives here.
I will begin with some preliminary remarks about ought-to-do sentences, (i) The word 'ought' in English has certain grammatical peculiarities which are not, I think, of any philosophical significance. (i) It cannot be used in the the future sense. But one can see that this is of no significance, if we substitute the phrase 'to be under an obligation'. One can say, e.g., 'When you become a parent you will be under an obligation to support your children'. (ii) When used of the past, the words 'ought' and 'ought not' have certain linguistic suggestions, which are also without philosophical significance. If one says 'X ought to have done so-and-so', there is a strong suggestion that he omitted to do this. Similarly, if one says 'X ought not to have done so-and-so', there is a strong suggestion that he did that action. All these irrelevant suggestions can be avoided by substituting the phrase 'to be under an obligation'. One can say that X was under an obligation to do so-and-so on a certain past occasion, without suggesting that he failed to do it; and one can say that X was under an obligation to do so-and-so, without suggesting that he did it. (iii) It should perhaps be added that, in ordinary English speech and writing, 'to be under an obligation to do so-and-so' does not mean the same as 'to be obliged to do so-and-so'. The normal meaning of the latter is to have no option but to do so-and-so.
(2) In 'ought-to-do' indicatives the grammatical complement to the word 'ought' or 'ought-not' is a name or a description of what I will call an agibile, i.e. a possible act of a certain kind. This agibile is supposed to have been, to be now, or to be going to be completely in control of the agent's will at the time referred to in the sentence. By this I mean that it is assumed that it would have been or will be enacted, if and only if the agent had decided or shall decide to enact it, and had set himself or shall set himself to carry out his decision.
(3) Kant said, truly I think, that we use 'ought-to-do' in regard to an agent only if we conceive there to be an actual or possible conflict of motives in him concerning the agibile in question. It always suggests that the agent may have to 'force himself to enact a certain agibile; and that, unless he makes and keeps up a certain special effort, he will either do nothing relevant or will enact some other alternative which is somehow easier or more attractive or less repulsive to him. This brings out a difference between 'ought-to-be-done', even in its most strictly moral sense, and 'morally right'. No doubt there is a close connection between the two. On one interpretation of 'ought', what a person morally ought to do in any situation is what would in fact be morally right for such a person to do in such a situation. On another interpretation of 'ought', what a person morally ought to do is what he believes to be morally right, for such a person as he believes himself to be, to do in the situation as he believes it to be. But, on either alternative, it is one thing to say that he morally ought to do so-and-so, and another thing to say that so-and-so would be morally right for him to do. In making the deontic statement we imply or suggest that he has a desire to do what is right as such, that he has other desires or inclinations which may conflict with this, and that he may need to make a special effort in order to do what he believes to be right.
This point may be brought out (as Kant remarks) by noticing that, whilst we should say that God always acts rightly, we should hesitate to apply the word 'ought' to him. For we assume that in him there would be no motives or inclinations which might possibly conflict with the desire to act rightly as such.
This reference to an actual or possible inner conflict extends to cases where little, if anything, specifically moral is involved in 'ought-to-do'. Take, e.g., the case of a person who has a decayed tooth which occasionally gives him severe pain. He may consider the question simply from the point of view of his own interests in the most narrowly hedonistic sense. He may be quite convinced that it would pay him very well, from that point of view, to go to the dentist and perhaps suffer a short bout of severe pain, in order to secure permanent freedom thereafter from toothache in that tooth. Even so, it is very likely that he will have a considerable internal struggle, and will not go to the dentist unless he takes himself in hand and forces himself to do so. A man in that situation would be very likely to say to himself 'I ought to go to the dentist', and a friend would be very likely to say to him 'You ought to go to the dentist'.
There is, indeed, one circumstance which gives a moral tinge even to such sentences as 'I [or you] ought to go to the dentist'. We approve, in ourselves and in others, the capacity and the act of overcoming one's own laziness, fear of immediate pain or unpopularity, or desire for immediate passive satisfactions, in order to carry out one's more far-reaching desires and purposes. For the possession and the exercise of that capacity is a necessary condition of all serious achievement, whether morally good, bad, or indifferent. We are thus inclined to feel and to express a kind of qualified moral approval of it even when it issues in acts which are morally indifferent, e.g. acts of far-sighted prudence, which cost an effort. We do so even when it issues in acts which we morally condemn. That is, perhaps, what lies at the back of the paradoxical admonition: Si peccas, pecca fortiter [If you sin, sin boldly].
That completes my preliminary remarks. I will now turn to a more detailed discussion of ought-to-do indicatives concerning persons. Let us say that such sentences express 'obligations of activity'. We can then begin by classifying obligations of activity on two independent principles, viz. (i) the nature of the activity, with which they are concerned, and (2) the nature of the obligation asserted.
(1) Human activities may be divided into practical and theoretical. So we have, corresponding to this division, obligations of practical and obligations of theoretical activity.
Suppose I were to say: 'You ought to try to produce as much good and as little evil as you can' or 'You ought to keep your promises'. These would express obligations which you are under as a being engaged in the practical business of co-operating or struggling with others, and affecting yourself or them for good or ill by your actions. So they are examples of obligations of practical activity. Suppose, on the other hand, I were to say: 'You ought to accept the conclusions which follow logically from the premises which you accept' or 'You ought to proportion the strength of your convictions to the weight of the relevant evidences available to you'. These sentences would express obligations, though not perhaps specifically moral ones, which you are under as a thinking being engaged in the theoretical activity of exercising your intellect. So they are examples of obligations of theoretical activity. Of course, theoretical activities may have practical consequences, and they are often pursued primarily in view of such consequences. But that does not affect the validity or the importance of the distinction.
Now I think there is a fairly close analogy between the two kinds of obligation. We have seen that obligations of practical activity presuppose an agent who has the desire to do right as such, but also has other desires and inclinations, which may conflict with it and may induce him to enact one of the wrong agibilia instead of the right one. Similarly, the obligations of theoretical activity presuppose a thinker who has the desire to think reasonably, but also has the prejudices and lazinesses, which may conflict with it. These may induce him to accept one of the propositions under consideration which he is not logically justified in accepting, or to believe one of these propositions more strongly or less strongly than the available evidence logically justifies him in doing. In each case a specific effort needs to be made and kept up, if the agent is to do what he ought. The obligations of practical activity presuppose that it is, in some sense, within the agent's power to enact the right agibile, in spite of the inclinations which conflict with his desire to do what is right as such. Similarly, the obligations of theoretical activity presuppose that it is, in a like sense, within the agent's power to suspend judgment when the evidence is inadequate, in spite of his desire to make up his mind. And they presuppose that it is, in that sense, within his power to proportion the strength of his convictions to the weight of the available evidence, in spite of his prejudices and his intellectual laziness.
(2) Let us now consider the classification of obligations of activity in respect of their intrinsic nature. In that respect they can first be divided into two fundamentally different classes, viz. teleological and ostensibly non-teleological.
The sentence 'You ought to try to produce as much good and as little evil as you can 'expresses an obligation which is explicitly teleological. One peculiarity of such an obligation is this. It contemplates a certain possible state of affairs, and it considers the agibile as a factor contributing to bringing it into existence or to prevent its coming into existence. Or, again, it contemplates a certain actual state of affairs, and it considers the agibile as a factor contributing to prolong it, or to cut it short, or to modify it in certain ways.
But that is not enough to make an obligation explicitly teleological. An essential condition is that this possible or actual state of affairs, and these possible modifications in it, are regarded from a certain point of view, viz. in respect of their value or disvalue, whether moral, aesthetic, hedonic, or otherwise. In so far as other aspects of them are considered, these are taken into account only as being good-making or bad-making characteristics. The ground alleged for the obligation to enact a certain agibile is simply and solely that it will produce or prolong or improve a good state of affairs or that it will avert or cut short or improve a bad state of affairs. And the ground alleged for the obligation to avoid enacting a certain agibile is exactly similar mutatis mutandis.
I have used the phrase 'ostensibly non-teleological' obligations, in the above dichotomy, because some philosophers have denied that there are any obligations which are really non-teleological. Utilitarians hold that all ostensibly non-teleological obligations are derivative; and that they can and must be derived, so far as they are valid at all, from the one ultimate teleogical obligation to produce as much good and as little evil as possible. But many persons would not be prepared to take this view about such obligations as truth-speaking, promise-keeping, etc. And any honest and intelligent Utilitarian would admit that these obligations are ostensibly non-teleological, though he would hold that that appearance is misleading and would try to explain how it may have arisen.
I think that there is no difficulty in defining the notion of a genuinely non-teleological obligation, whether or not there be in fact any such obligations. I propose to take the alleged obligation to answer a question truly, if at all, as a plausible prima facie example. The essential point is that the obligation to act in the specified way in situations of the specified kind is alleged not to be grounded on the goodness or badness of the consequences which acting (or failing to act) in that way would contribute to produce. The obligation is alleged to be grounded either (a) on certain intrinsic qualities of such action, or (b) on certain of its non-causal relations to the situation in which it is done, or (c) on certain features in its consequences other than their goodness or badness.
Consider the sentence 'You ought to answer truly, if at all, when asked a question'. Let us suppose, as most unsophisticated persons do, that it expresses a non-teleological obligation. The sentence might, of course, be interpreted in at least two different ways. It might mean (i) that you ought to give what you think is the true answer, whether or not you think that doing so will produce a true belief as to the quaesitum in the mind of the questioner. Or it might mean (ii) that you ought to give an answer which you think will produce a true belief as to the quaesitum in the mind of the questioner, whether or not you believe that answer to be true. On the first alternative, the obligation is not concerned with the consequences of the act at all, but with a certain quality or relational property of it, viz. its being an utterance of what the answerer takes to be the truth about the subject of the question. On the second alternative, the obligation is indeed based on a certain feature of the intended consequences of the act, viz. the truth of the belief to be produced by it in the mind of the questioner. But that does not make the obligation teleological, in the sense defined above. For the question whether it is good or bad for the questioner to have a true belief on the subject of his question is held to be irrelevant to the obligation to answer in such a way as to produce in him a true belief.
Having dealt with the classification of obligations into practical and theoretical, and into teleological and non-teleological, I will now pass to another important distinction among them. An obligation of activity may either be restricted or unrestricted in its range of application. A restricted obligation is concerned with a certain specific type of situation, e.g. that of being asked a question. The deontic indicative here asserts that any person (or any person of a specific kind) who is acting in response to such a situation ought to act in a certain specific way. An example would be: 'Whenever a person is asked a question he ought, if he has the relevant information, either to return a true answer or to decline to answer'. An unrestricted obligation is supposed to apply equally in any situation in which a voluntary action is to be done. An example would be: 'Whenever a person acts he ought to try to produce as much good and as little evil as he can'. Another example would be: 'A person ought never to treat another in a way in which he would not be willing to be treated by another'. It is evident that, if there be any unrestricted obligations, they will all be extremely abstract; and that, taken by themselves, they will give a person very little positive guidance as to what he ought to do in any particular situation.
We can now consider another important division of obligations of activity, viz. into those which are ultimate, and those which are derivative, for a given individual at a given stage of his development. An unrestricted obligation of activity is ultimate for an individual, if it seems evident to him on inspection that anyone performing a voluntary action in any situation is under an obligation to act in the way specified. A restricted obligation of activity is ultimate for an individual, if it seems evident to him on inspection that anyone (or anyone of a certain kind), placed in a situation of the kind specified, ought to act in the way specified. An obligation of activity is derivative for an individual, when (a) it does not seem evident to him on inspection; and (b) seems to him to require, and to be capable of, being established by deductive reasoning.
A plausible example of an unrestricted obligation of activity which is ultimate for many individuals is the Utilitarian principle: 'In all his actions a person ought to try to produce as much good and as little evil as possible'. Another plausible example is the obligation of theoretical activity expressed by the sentence: 'When called upon to make up his mind as to the truth or falsity, probability or improbability, of several alternative propositions, a person ought to proportion the strength of his convictions to the weight of the evidence available to him'. The former is specifically moral; the latter perhaps is not. So far as I am aware no reason can be given for either of these alleged obligations. And to most of us no reason seems to be needed, for it seems self-evident on inspection that people are under these obligations.
Let us now consider some of the ways in which an obligation can be derived. I am inclined to think that it is a true general principle that no obligation can legitimately be derived, unless some other obligation is already presupposed as a premise. You cannot legitimately infer a deontic proposition from nothing but non-deontic premises; though you can (and perhaps must) use non-deontic premises, in conjuction with deontic ones, in your derivation. Again, I am inclined to think that a specifically moral obligation can be legitimately derived only from premises which include some deontic proposition asserting a specifically moral obligation. I would admit, however, that these impressions of mine (and especially the second of them) may be mistaken.
Leaving these general preliminaries, let us now consider the derivation of specifically moral obligations. To illustrate some of the more important types of derivation let us take as an example the proposition: A person, when asked a question, ought never to give an answer which he believes to be false. To some people this might appear self-evident on inspection. For any such person it would state an ultimate non-teleological obligation of limited range. But there are many people who are not in that position, and yet would accept it as stating a derived obligation. There seems to be at least two alternative possible ways in which it might be derived.
(i) Suppose that an individual finds self-evident the following unrestricted non-teleological deontic proposition, viz. A person ought never to treat another in a way in which he would not be willing to be treated by another. (This is one form of the 'Golden Rule'.) Suppose further that he knows or believes that no-one is willing to be told a lie in answer to a question. These two propositions together entail that a person ought never to give an answer which he believes to be false to a person who puts a question to him. On the two suppositions which I have made, this proposition would state, for the individual supposed, a derived non-teleological obligation of restricted range.
(ii) Let us next suppose, instead, that an individual finds self-evident the following unrestricted teleological deontic proposition viz. In all one's actions one ought to try to produce as much good and as little evil as possible. Suppose, further, that he knows or believes that telling lies in answer to questions always produces less good or more evil in the long run than telling the truth or declining to answer. These two propositions together entail that one ought never to give an answer which one believes to be false to a question which is put to one. On the two suppositions which I am now making, this proposition would state, for the individual supposed, a derived teleological obligation of restricted range.
It is worth remarking here that it is conceivable that one and the same individual might accept both pairs of premises, for they are not obviously inconsistent with each other. In that case one and the same deontic indicative sentence would express for him an obligation of restricted range which could be derived both teleologically and non-teleologically. It might even happen that such a person also found the deontic proposition under discussion self-evident on inspection. In that case the same sentence would express for him an ultimate obligation of restricted range. He would not need to derive it; but he might well be strengthened in his conviction by seeing that it can be derived, both as a teleological and as a non-teleological obligation.
Let us next consider what Kant called 'hypothetical imperatives' and 'imperatives of skill.' The former presuppose a desire which is assumed to be common to nearly all men at nearly all times, and therefore not to need explicit mention, e.g. desire for good health, long life, prosperity, etc. The latter presuppose a desire which is peculiar to a particular individual or class of individuals on a particular occasion or set of occasions, e.g. a desire in Mr Jones to kill his wife, or a desire in cooks to make an apple-pie. Two things seem plain about these deontic indicatives. One is that, if they express obligations at all, these are not specifically moral ones. The other is that they are in some way concerned with what I will call 'obligations of consistency'.
An obligation of consistency may be either practical or theoretical, and in neither case would it seem to be specifically moral. I would formulate the obligation of practical consistency as follows: 'A person who intends a certain end ought either to cease intending it or to take the most efficient means open to him to attain it. He ought not both to go on intending it and to do acts which would make it impossible for him to attain it'. I think it is important to formulate the obligation in this disjunctive way. For that makes it clear that the 'ought' and 'ought-not' now under discussion are concerned with consistency or inconsistency between (a) continuing to intend a certain end, and (b) acting or failing to act in certain ways which are relevant to the attainment or to the non-attainment of that end.
Let us now consider the derivation of a 'hypothetical imperative', in Kant's sense of the word. We will take as an example: A person ought to take exercise and not habitually to overeat'.
It is assumed as a factual premise that taking exercise is a necessary condition for keeping in good health, and that habitual overeating is a sufficient condition for failing to do so. From this factual premise and the obligation of practical consistency we can infer the following proposition: 'A person who intends to keep in good health ought either to give up that intention or to take exercise, and he ought not both to go on intending to keep in health and habitually to overeat'. Now it is assumed that all men intend to keep in good health. On that assumption, we can substitute the phrase 'a person' for the phrase 'a person who intends to keep in good health'. It is further assumed that the intention to keep in good health is a standing intention, which a person cannot or will not abandon, though he can and often does act in ways which he knows to be inconsistent with fulfilling it. On that assumption, the only way for anyone to be practically consistent in this department of his life is to take exercise and not habitually to overeat. So we reach the conclusion that a person ought, in order to be practically consistent, to take exercise and not habitually to overeat.
Let us next take what Kant would call an 'imperative of skill'. We will suppose that Mr Jones has formed the intention to kill his wife, and that much the most efficient way open to him for securing that end is to put arsenic in her tea. From these premises, together with the obligation of practical consistency, we can infer the proposition: Mr Jones ought, in order to be practically consistent, either to give up his intention to kill his wife or to put arsenic in her tea'. If he cannot or will not give up his intention, you can say that he ought, in order to be practically consistent, to put arsenic in his wife's tea. But it is essential to add explicitly the qualification 'in order to be practically consistent'. For it is only in relation to it that the obligation exists. In the moral and the legal senses of 'ought' he ought not to do this. He ought to give up his intention to kill his wife.
It is sometimes alleged that what Kant calls 'hypothetical imperatives' and what he calls 'imperatives of skill' do not really express deontic propositions at all. It is suggested that they are simply equivalent to non-deontic sentences expressing causal propositions of a certain kind. Thus, e.g., it might be alleged that the sentence: 'You ought, unless you give up your intention to kill your wife, to put arsenic in her tea', is simply equivalent to the sentence: 'The most efficient means available to you for carrying out your intention to kill your wife is to put arsenic in her tea'. That seems to me to be a mistaken view, though I am willing to admit that such deontic sentences may sometimes be used to mean more than this. But, in general, I think that the causal proposition is only the factual ground for a derived deontic proposition. The latter has also a deontic ground, viz. the ultimate, though not specifically moral, obligation of practical consistency. I take it that Kant would have agreed at least with the negative part of my contention. For he calls such propositions 'imperatives', and surely a mere causal proposition could not be called an 'imperative' in any sense of that word.
That brings me to the next question which I propose to discuss. As we all know, Kant gave the name 'imperatives' to what is expressed by ought-to-do indicatives about persons. How far is that nomenclature illuminating, and how far is it misleading? That question, in various forms, has been much discussed by many able philosophers in recent years. I fear that all that I can say about it here is somewhat platitudinous.
If the name 'imperatives' is taken literally, the doctrine would be that an ought-to-do indicative about persons expresses neither more nor less than what would be naturally expressed by a corresponding sentence in the imperative mood. E.g. the sentence 'You ought not to steal' would express and convey exactly what is expressed and conveyed by the sentence 'Do not steal!', i.e. an order issued by one person and received by another. Another possible alternative interpretation of the doctrine would be the following. Such a deontic indicative does indeed express or convey something which is not expressed or conveyed by any sentence in the imperative, viz. some characteristic kind of information, whether true or false. But it also expresses or conveys a command, as an ordinary sentence in the imperative would do. And its specifically deontic character, which distinguishes it from such other sentences in the indicative as, e.g., 'Stealing will probably land you in jail sooner or later', is bound up with this imperative function. A third alternative would be a more cautious modification of the second. Instead of saying that a deontic indicative derives its specifically deontic character from expressing or conveying a command, it might be said that such sentences function in certain respects in a way analogous to ordinary imperative sentences. There are unlikenesses as well as likenesses, and the likenesses are not exact. Perhaps that is all that Kant wished to imply by his use of the word 'imperatives'. And, however that may be, perhaps that is all that can be maintained. I will now make some comments on this.
(1) I think that the first alternative can be rejected at once, at any rate as regards specifically moral ought-to-do indicatives about persons. In the case of a literal imperative, e.g. the military command 'Form fours!', there is no sense in asking whether what it expresses is true or false. The only sensible questions which can be raised about a literal imperative are such as the following. Was it actually uttered, and, if so, was it meant seriously? Granted that that was so, is there any doubt as to precisely what was commanded? Granted that there is no doubt on that point, was the person who uttered the imperative entitled to issue orders on that subject to the person or persons whom he addressed?
Now, in the case of a moral deontic indicative, e.g. 'You ought not to give an answer which you know to be false, when asked a question', it seems obvious that we can sensibly raise a question which does not fall under any of the above headings, viz. Is it in fact true, or is it false, that a person ought never to act in that way in such a situation? Conversely, it seems plain that certain of the questions which can be raised about literal imperatives do not arise in regard to moral deontic imperatives. A person may, e.g., fully believe that he ought not wittingly to give false answers to questions put to him; and yet he may deny that anyone has actually forbidden him to do so, and deny that anyone would be entitled to issue orders to him on that subject.
(2) Among literal imperatives we must distinguish two different kinds, which may be called 'violent' and 'legitimate'. The imperative 'Stand and deliver!', issued by a highwayman at the point of his pistol to a wayfarer, is an example of the former. The imperative 'Form fours!', issued by an officer to a company of his own men, whom he is drilling, is an instance of the latter. Now there seems to be very little analogy between a violent imperative and what is expressed or conveyed by a specifically moral deontic imperative. On the other hand, analogies between legitimate imperatives and specifically moral deontic indicatives tend, for the following reason, not to throw much light on the nature of what is expressed by the latter. A legitimate imperative is one issued by a person who has a right to give orders about a certain matter to certain persons; and it is issued to one of those persons who is under an obligation to obey him in such matters. An officer stands to his men in a certain relationship, which gives him a right to command them in certain of their actions, and places them under an obligation to obey his commands in respect of those actions. Thus the notion of a legitimate imperative presupposes the notion of rights and correlative obligations. So there is a kind of vicious circularity in claiming to illuminate the notion of moral obligation by reference to legitimate imperatives.
(3) I think that there are at least two causes which make it seem plausible to assimilate deontic indicatives to literal imperatives:
(i) There is at least one genuine likeness between the situation in which a person stands when he literally receives a command and that in which he stands when he believes himself to be under a moral obligation to act in a certain way. In both cases the act is not one which he would do simply because he likes doing it, as he might dance a jig or whistle a certain tune simply because he felt so inclined. Again, it is not one which he would do as an obvious means to securing some immediate satisfaction or cutting short some unpleasant experience, as he might eat if he felt hungry or if he were offered some food whose taste he knew that he liked. On the contrary, the act commanded and the act judged to be obligatory are often alike in being irksome or positively unpleasant in themselves. They are often alike in that they involve forgoing some immediate satisfaction, or bringing on oneself some pain or loss or unpopularity, or incurring some danger. They tend, in fact, in both cases to be acts which, as we say, 'go against the grain'. And the more they do so, the more fully does the agent realize, in the one case that he is being commanded, and in the other that he is under an obligation. This is certainly an important analogy.
(ii) For Jews, Christians, and Mahometans, at any rate, some of the most important negative obligations, i.e. duties of omission or avoidance, are formulated in the so-called 'Ten Commandments' as literal imperatives, issued by God, and promulgated on his behalf to men by his prophet Moses. That no doubt makes it easy for those brought up in any of those religions -- and all contemporary Europeans and Moslems have been brought up in societies which are rooted in them and are still haunted by the ghosts of them -- to identify what is expressed by a deontic indicative, e.g. 'A person ought not to steal', with what is expressed by a literal imperative uttered by God, e.g. 'Thou shalt not steal!' or 'Do not steal!'.
But, even if we were to accept the story of the alleged events on Mount Sinai in the most literal sense, the identification would be quite unwarranted. At most it might be held that the only ground for the deontic proposition that a person ought not to steal is the fact that God has issued the command 'Thou shalt not steal!'. Now that fact is not something that can be expressed by a sentence in the imperative mood; it is the alleged historical fact that a certain command has been issued by a certain person on a certain occasion. Moreover, it is not really possible to hold that the historical fact alone could be the ground for the deontic proposition that a person ought not to steal. If a similar command had been issued, e.g., by Moses on his own authority, no one would suppose for an instant that the fact that he had issued it would be a ground for the corresponding deontic proposition. An essential premise would be that the command was issued by God; and that we, as his creatures, stand in such relation to him that we have a duty to obey his orders. Unless you add this latter deontic premise, the Ten Commandments would be nothing but violent imperatives. In that case, though it might well be prudent to obey God's orders, in view of his overwhelming power, there would be no more question of moral obligation than there is in handing over one's purse to a highwayman at his command.
(4) We may note, further, the following prima facie differences between literal imperatives and what is expressed by deontic indicatives:
(i) A person does not literally issue orders to himself. But it is just as intelligible for a person to say of himself: 'I ought to make an allowance to my old nurse' as it is for him to say of another: 'You ought to make an allowance to your old nurse'. Attempts are sometimes made to evade this difference by representing statements of the form: 'I ought to do so-and-so' as expressing commands issued by a man's conscience or his higher self to his lower self. This way of speaking involves personifying one's conscience or one's higher self, and treating one's lower self as another person. That is harmless enough in practical life. But, if taken seriously, it is plainly mythology of a dangerously misleading kind. Here, at least, the example of Kant should be regarded as a warning and not as an encouragement.
(ii) A person does not literally command or forbid an action which he knows or believes to have already been done or to have been already left undone. But it is quite common and intelligible to say: 'A did x, but he ought not to have done it' or 'A failed to do y, but he ought to have done it'.
(5) It seems prima facie that certain deontic indicatives, so far from expressing commands, state the ground for issuing certain legitimate commands. Suppose I utter to someone the literal imperative: 'Pay me £2 19s 4d immediately!' Let us assume that I am not an armed robber issuing a violent command. The other man may reasonably ask me: 'Why should I do so?' Then it would seem that I should be giving a reasonable ground for my demand, if I could truly say: 'You promised to pay me that sum at the present date and time, and you know very well that you ought to keep your promises.'
On the whole then it seems to me that the unlikenesses between what is expressed by deontic indicatives and what is expressed by literal imperatives are at least as striking and important as the likenesses. What I think must be admitted, however, is this. A person's belief that he is under an obligation to enact a certain agibile on a certain occasion (provided that the belief is occurring as an actual experience, and is not existing merely as an unactivated disposition) does exert on him (provided he be mentally normal and that he has been subjected to a minimum of moral training) a felt impulse towards enacting that agibile. If we substitute the words 'refrain from enacting' for 'enact', and 'repulsion against enacting' for 'impulse towards enacting' in the above sentence, the resulting sentence is perhaps even more obviously true. Suppose, now, that the agibile which one believes oneself to be under an obligation to enact is in other respects distatsteful to one, or that some of the alternative agibilia are in other respects attractive to one. Then, I think, it must be admitted that the felt impulse towards enacting the agibile which one believes that one ought to enact, does resemble, perhaps more closely than anything else that one can recall, the experience which one has on receiving a command from a person, whom one recognizes to have the right to issue it to one, to do something which is in other respects distasteful.