The ‘circularity’ problem about promising has surfaced again in a recent article.Footnote ¹ The problem may be stated as follows. The giving and exchange of promises involves relations of mutual trust among the participants. Why, having promised you to do X, should I do X? The simple answer is that this is what I promised. But what if I thought you were not counting on me to do what I promised? In that case my reason for keeping the promise would be weakened, and perhaps extinguished. What, I might ask myself, would be the point of keeping the promise? You, on the other hand, have a reason to count on me to do X: namely, that I promised. But this reason would be weakened, and perhaps extinguished, if you thought that I might think you were not counting on me to do X. But again, if I thought you might think this, then I would have reason to think you were not counting on me to do X. And then, again, my reason for doing X would be weakened and perhaps extinguished.

In ordinary life such reflections occur, of course, only in special cases. (The initial step - the doubt whether you are counting on me - is not uncommon in suitable situations.) But if one really thought the matter through as I have presented it, should this not undermine the mutual trust on which the making of promises depends? The fact is that it does not. People - including readers of this article - who have thought it through, go on making and accepting promises to one another just as they did before. They do not become more suspicious or less trustworthy than they were before.Footnote ²

The problem may be posed in Kantian terms: How is the practice of promising possible? Or more precisely, how is it possible even among those who have reflected? The problem can also be seen as an instance of a more general one: How is it possible for people to trust one another? Now there must be something wrong with the circularity argument, since its conclusion is blatantly contrary to the fact that promising is (even among those who have reflected) a common occurrence. But what is wrong with the argument? And how is the possibility of promising to be explained?

According to T.M. Scanlon, the problem can be solved by pointing out that in making a promise ‘I do several things’:

In another passage he writes that what ‘I lead you to believe’ is that ‘I know and take seriously the fact that … it would be wrong of me to make the promise if I did not intend to do what I said' (308).Footnote ³ It is through these (implicit) assurances of ‘taking seriously’ that the problem of circularity is supposed to be avoided. But it is not clear how such assurances could make the required difference. For if I cannot be sure that you are counting on me to do what I promised, how can I be sure that you are counting on me to take seriously what I said I would take seriously?

A different solution has been offered by Kolodny and Wallace, who maintain, in opposition to Scanlon, that promising is a ‘convention or social practice’. They claim that failing to keep one's promise ‘might … undermine the expectations that constitute [this] practice’, which ‘has great instrumental value to the participants’ (149). Such a person would be ‘free-riding on the contributions that others have made to the practice’. Now it is true that the prohibition against ‘free-riding’ is independent of the obligation of promising, and not exposed to the problem of circularity. And if the motive for keeping one's promises could be explained in terms of free-riding, then the problem of circularity would be solved. But can it be explained in that way?

Let us consider a straightforward case of free-riding. Suppose - adapting an example from Hume - that A and B are rowing a boat. The boat has two pairs of oars and B is sitting behind A. It is a hot day and rowing is hard work. How pleasant it would be for B to quietly raise his oars out of the water and to enjoy the scenery as a free rider! But B knows that it would be wrong to make A bear the whole burden of propelling the boat, and that he would not be making his due contribution if he behaved in that way. Here is another example: Let us assume I am living in a country where public revenues are spent, by and large, on activities and services that are approved by the population, including myself. Then it would be wrong for me to fail to pay my taxes, leaving it to others to bear the burden of financing those activities and services. Here again the metaphor of free-riding is appropriate.

But is it also appropriate for the case of keeping one's promises? Suppose I promise to do you a favour. If I keep that promise, I do not do it to contribute my share to an activity in which I co-operate with others; I do it simply because that is what I promised you. My obligation would be to you, and not to fellow participants in a co-operative activity. In giving that promise I would have bestowed a right on you in particular: namely, the right to receive what I had promised. And in the case of non-fulfilment I would be exposed to a justified complaint from you, and not from fellow participants in a co-operative activity.Footnote ⁴

The true solution of the problem lies, I believe, in doing justice to the role of natural inclinations in linguistic behaviour, which includes telling the truth as well as keeping promises.

Telling the Truth and Keeping Promises: What Comes Naturally

Human beings, like other animals, have the ability to communicate with one another; this is an important factor in the competition for survival. There is a difference, however, between human language and the communication systems of other animals. Human beings are responsible for what they say: responsible in the literal sense of being expected to respond, and respond in a suitable way, to such questions as ‘Why did you say that?’, ‘Why did you tell me that p when p is false?’; and, in the case of promising, ‘Why didn't you do what you promised?’ And someone who understood what he was saying when he made an assertion or gave a promise must understand that the questioner would have a right to complain if suitable responses to such questions were not forthcoming.

Now the responsibility of telling the truth to the best of one's knowledge is essential to the very existence of assertion. It must be the case that normally people tell the truth, that this is expected of them and that they may expect complaints and resentment if they are found to tell lies. If telling the truth were not the norm, if people could not be blamed for telling lies, then the utterances in question could not be regarded as assertions, and there would be no such speech-act as assertion. (This is not to deny that there might still be useful signals, oral or other, as given off by non-human animals.) But if the expectation of truthfulness is a necessary condition of the existence of assertion, how does this expectation come about? How is it that people normally tell the truth to the best of their knowledge? They do not do it with the aim of making assertion possible. Do they do it because they fear the resentment that may result from lying? But this resentment would be justified only if or because the norm of truth-telling is already in place. There must be some independent, natural inclination to tell the truth. And so there is. Telling the truth is the natural, spontaneous way of making assertions. It is what we do unless we have reason to do otherwise. If someone asks me how to get from A to B, I tell him the truth to the best of my knowledge. Why not? If I tell you how I spent the week-end, I tell you the truth unless I have some reason for concealing it. Why should I do otherwise? This is not because I am a virtuous person with a special respect for truthfulness; it is simply what comes naturally.

What is the use of language? One use, obviously, is instrumental. Language is a means of acquiring and conveying knowledge, making demands, devising plans and so on. But another, major function of language is that of a social cement. This is observable when people chat in a ‘useless’ way, exchanging remarks and questions because they feel like it and not for any reason other than that. It is the human way of being in touch with each other and comparable, in this respect, with the noises made by social animals and birds; and in this situation there is simply no point, in the absence of special reasons, in deviating from the truth.

What comes naturally, in this respect, is also what comes originally when we first learn to speak. As Wittgenstein observed, when a child learns to use such words as ‘chair’ and ‘red’, he does not begin with ‘That is probably a chair’ or ‘That seems red’ (Zettel 411–8). Similarly, a child does not begin with telling lies about chairs before it learns to say truly ‘That is a chair’. Telling lies, like doubting, must be a further stage; it is part of our emergence from an original age of innocence. A child in the course of learning to speak might be corrected for using the word ‘chair’ incorrectly (perhaps applying it to other items of furniture), but this would not be a case of lying.

With this emergence we become doubly responsible: in a moral but also in a non-moral sense. A person who is found to have lied is expected to justify or excuse his behaviour. That is the moral sense. But he is also expected to have a suitable reason, which need not be morally respectable, for having lied. Without such a reason his behaviour would be unintelligible. It is true that there are people who lie without any reason - without any advantage to themselves or others; but this is an abnormal condition. When we say that there is something wrong with such a person, we do not mean this in the moral sense of ‘wrong’. Such lying is pathological and not wicked. Such people are liable to get into trouble which is not of the kind that criminals get into. The trouble occurs because their behaviour leaves us puzzled. We cannot relate to them in the normal, satisfying way in which we relate with normal people.

It might be objected that the original learning situation is not a reliable guide to the use of language thereafter. Granted that the child originally learns to speak the truth, might he not, when more sophisticated, come to tell lies with equal readiness whenever it is to his advantage? It is true, of course, that children at that stage sometimes tell lies when it is to their advantage; but it cannot be true, in general, that they do so with equal readiness. This is because a child grows up under the guidance of parents (or other adults), and it is in their interest to inculcate the habit of truthfulness in the child. And the same is true of the habit of doing what one has promised. It is not surprising, then, that these ways of behaving are, by and large, natural and spontaneous. It is also in the course of this inculcation that dishonest behaviour comes to be associated with feelings of shame and remorse, for these feelings, and not only the behaviour, are inculcated by the parents' treatment of the child.

There is more to this inculcation than merely a fear of unpleasant resentment in the event of misbehaviour. A child who has mastered the concept of a reason must come to recognize that the resentment would be justified - that his failure would be blameworthy and a suitable object for remorse. And this also covers cases in which a lie or the breaking of a promise are unlikely to be found out - where the chance of being rebuked by the victim or by others is remote or non-existent. Thus the motive for keeping my promise need not vanish if I learn that this is so; for in that case I would still be aware that a complaint from the victim would have been justified, and this is enough to make a failure to keep the promise a suitable object for remorse. And hence the reply ‘Because I promised’ can be perfectly reasonable in such cases. Here we see the importance of hypothetical reflection in human motivation. What is important is not merely what will or may happen, but also what would have happened if … (if the victim were, or had been, able to confront me).

I said that it is in the parents' interest to inculcate such behaviour and feelings in the child, but this does not mean that they do it because it is in their interest. The parents' behaviour is itself a spontaneous expression of their feelings towards the child's behaviour and not a result of instrumental reasoning: it is natural for a parent to express resentment and displeasure towards a child that tells lies and does not keep his promises. And the same is true of other people - adults and children - in the child's environment.

These natural attitudes and developments were not recognized by Hume in his discussion ‘Of the obligation of promises’, where he tried to explain the origin of this and other moral concepts. There is, he wrote, ‘naturally no inclination to observe promises, distinct from a sense of their obligation’. But this sense of obligation, he argued, could not be accounted for by human nature; it must be due to a convention. ‘Fidelity is no natural virtue, and … promises have no force, antecedent to human convention’. He explained the advantages of such a convention. An exchange of promises between two self-interested parties, he pointed out, may be to the advantage of both, and this is often the case. Hence it comes about that ‘there is a certain form of words invented … by which we bind ourselves to the performance of any action. This form of words constitutes what we call a promise …’ (T 519–22).

Hume's account, as has often been noted, gives rise to a number of problems. How, where and by whom, we may wonder, was that invention achieved? Would there be no promises if the invention had not been achieved? Again, if the obligation of promise-keeping was introduced as a matter of convention, why should those who were not party to the convention be under an obligation to abide by it? There is also a problem of circularity (not quite the same as that discussed in the present article), which affects those who might have been party to it. They must be understood to have agreed to the convention - to have undertaken to abide by it. But such an undertaking would itself be nothing other than a promise - a promise to use that ‘form of words’ so as to bind oneself ‘to the performance of any action’.

There is no need, however, to posit the introduction of a convention to account for the sense of obligation to keep our promises. The inclination to do so, and the remorse that we are apt to feel when we fail to do so, can easily be accounted for (as I have argued) by the natural circumstances of human development.

To say that keeping promises is what comes naturally is not, of course, to deny that people often break their promises, especially when it is strongly in their interests to do so. But this does not entail a problem about the possibility of promising. Let us consider Hume's example of the two farmers in this light. ‘Your corn is ripe today; mine will be so tomorrow.’ It would be to our mutual advantage if I helped you with yours today, on condition that you do the same for me tomorrow. But ‘I have no kindness for you, and know you have as little for me’, and so ‘both of us lose our harvests for want of mutual confidence and security’ (T 520–1). According to Hume, as we have seen, the remedy lies in the introduction of a convention, involving a ‘special form of words’.

Now it is true that in a situation of mutual mistrust it may be difficult and perhaps impossible to gain the benefits of co-operation. This is indeed what often happens, a prominent case being that of international relations. But this does not affect the possibility of promising in situations in which people are not in a state of mutual mistrust; in which trust is the natural attitude and to be expected unless there are special reasons to the contrary. It is also true that even under these conditions people may break their promises if keeping them would be unduly inconvenient. In many cases, however, this is not unduly inconvenient, and no ‘sense of obligation’ is needed to motivate the keeping of such promises. Why should I not meet you outside the library if that is what I have promised? If it turned out to be difficult to do so, then a sense of obligation might be needed to overcome a temptation to break the promise. But in straightforward cases the difficulty and the overcoming would be in the opposite direction: it would be breaking the promise, and not keeping it, that would go against the grain. And this can be so even when there is some inconvenience in keeping the promise.

Such inclinations make it reasonable for us to expect that people will, on the whole, tell the truth and keep their promises; and if these expectations are disappointed, then we have a right to complain to the person who caused the disappointment. If, on the other hand, one were asked to give a promise the keeping of which would be unduly troublesome, then one might refuse to give it. ‘This is too much to ask’, one might say in such a case. In such cases, moreover, the person demanding the promise should be aware that it might be foolish to rely on it even if it were given. And if this were the usual situation, then there would be no way out of the reciprocal mistrust that was supposed to give rise to the problem of circularity. There is, however, no reason to assume that this is the usual situation.

Let us contrast the natural situation, as I have described it, with one in which there are artificial arrangements. Consider again the predicament of the two farmers described by Hume. Such predicaments can be avoided by the introduction of enforceable laws. If these are in place, then the farmers can draw up a contract, whereby A, having profited from B's help today, would be forced to help the latter with his corn tomorrow - or failing that, to compensate him for his loss. No such threats are needed, however, in ordinary promising-situations such as occur among members of a community in which there are natural relations of trust - where people who ‘give their word’ may be expected to keep to it.

I have spoken several times of what may be ‘expected’. Now this word can be used in a natural or in a moral sense. The first occurs when we speak of normal (=expected) weather conditions; the second involves responsibility. Which of these was meant when I said that people are, on the whole, expected to tell the truth and keep their promises? Both senses are involved. It is because speaking the truth is the natural, spontaneous way of making assertions that assertions are, in the natural sense, expected to be true unless there are special reasons to the contrary. But then it follows that if an assertion turns out to be false, we are liable to feel disappointed, to blame the person who is responsible for deceiving us and to expect him - in the moral sense - to answer for his behaviour. (The answer might be either that he thought he was telling the truth or that he had a justification for telling a lie.) And similarly, it is because doing what one said one is going to do is the natural, spontaneous way of behaving, that this behaviour is what is expected, in both senses of the word.

The responsibility of keeping a promise is different, of course, from that of telling the truth, but they are connected. To make a promise is, at the same time, to express an intention (it would be self-contradictory to say ‘I promise to do X but don't intend to do it’); and in expressing an intention one may or may not be telling the truth. But the natural inclination to tell the truth is present here as elsewhere. If I told you what I intend to do tomorrow, then, in the absence of special reasons, I would be inclined to tell you the truth. And then there is a further (logical) connection: that between intending to do X and doing it. This again is the normal case: it is what we expect unless something intervenes.Footnote ⁵

The Importance of Saying ‘I Promise’

I said that in making a promise one expresses an intention. But promises are often made by expressing an intention. According to a common but erroneous view, the expression ‘I promise …’ is essential to the making of promises - as implied by Hume's supposition that ‘there is a certain form of words invented …’. But is the expression ‘I promise …’ the normal or paradigmatic way of making a promise? Its use is not as common as might be supposed. It occurs typically in formal contexts (as in ‘I promise to pay the bearer …’, ‘I promise to serve my country’, etc.) and in cases where emphasis is required (as when one is challenged ‘Are you sure you won't let me down?’ and the answer is ‘No, I promise'; or again, ‘Do you promise to do X?’, etc.). The word ‘promise’ is, moreover, used less in making promises than in complaining about the breaking of them - as when we say ‘But you promised.’

That saying ‘I promise …’ is not a sufficient condition for promising is fairly obvious. The person to whom the promise is addressed must speak English, etc. He must also have an interest in what is being promised. Thus I cannot promise you a ride in my boat if you don't care, and I know you don't care, for boating; I cannot promise to meet you if I know that it is, and will be, all the same to you whether I turn up or not.

Is the locution ‘I promise’ a necessary condition of promising? Consider the following conversation:

Was the promise given when A made his original statement, or only when he said ‘I promise’? Clearly A was committed to some obligation when he told B that he would meet him, and had heard B's response. If he failed to appear, B would be entitled to rebuke him, and A would have to understand this. (If he did not, then he could not have understood what he and B had been saying.)

Did the words ‘I promise’ import a new obligation? Suppose that A, having said ‘I'll meet you’, and knowing that B was counting on him, failed to appear. Then B could properly complain that A had ‘broken his promise’. It would be absurd for A to try to defend himself by saying ‘But don't you remember? I only said I would meet you; I didn't actually promise’. To say that someone gave a promise is not to say that he actually used the word ‘promise’ (or any equivalent word).

It might still be said that the words ‘I promise’ would heighten one's commitment, importing a new degree of obligation if not a new obligation. This would not affect the main question - whether the word ‘promise’ is necessary for the giving of promises. But let us see what is to be said for the new claim. It is true that, in the example I gave, the words ‘I promise’ would have served to heighten A's commitment. This would also come out in what B might say if A had broken his promise. ‘I specially asked’, B might say, ‘and you said “I promise”’. No doubt the words ‘I promise’ can be used to emphasize one's commitment; but this can also be done in a variety of other ways. Instead of ‘I promise’, A might have said ‘I shall definitely be there’, ‘I won't let you down’, ‘I know it's important to you’ etc.

The difference between ‘I promise’ and ‘I shall’ is not one of degrees of obligation; it is rather that these expressions make sense in different circumstances. A person may or may not be giving a promise in saying ‘I shall …’; whether he does, will depend on circumstances such as B's interests. What is special about ‘I promise’ is that it is used only in obligation-creating circumstances - those in which ‘I shall …’ and ‘I intend’ would also have the force of a promise. To say ‘I promise’ is to presuppose that these circumstances are present, and in this it differs from expressions of intention.

It might be thought that ‘I intend …’, unlike ‘I promise …’, is essentially a ‘psychological’ statement, whose main function is to inform others of what one has in mind; and it is true that this can be its function. A friend may be curious to know what I intend to do in my study leave, and I may tell him what I have in mind. In this case, as in most other statements of fact, I would be under an obligation to tell the truth; but this would not be like the obligation of promise-keeping. In other cases, however, ‘I intend …’ has the force of a promise. If I tell a student that I intend to see him at three o'clock, then I may lay myself open to the complaint ‘But you promised …’ if I am not there at three o'clock.

Imagine a society in which, without ever using such a word as ‘promise’, people sometimes tell others their intentions, knowing that the latter will count on them, etc. When, in such circumstances, one person fails to do what he said, the other would be likely to resent it. Let us call this a promising-situation. If the latter person could use our language, he might complain ‘But you promised …’. But in the absence of such a word, he could express the same complaint in other words. Thus he might say ‘But you said you would …’, ‘You knew I was counting on it’, etc. - in short, the very things that we, with our language, might naturally say in such a case. And then the offender might be expected to ward off such complaints by means of excuses, just as he would in our society. In that society, just as in ours, the complaints would be regarded by others as reasonable, and so would the expectation of excuses. Thus the essential elements of the language-game of promising would be present in that society no less than in ours.

This being so, however, it is likely that a word or expression corresponding to our ‘promise’ would come into use. This is because the relevant situations occur frequently in human life, and statements of intention made in such situations have a special significance. The use of a special expression that indicates (or presupposes) that a situation is of that kind is then a natural development in the economy of language.

A similar development can be traced in the case of apologizing. This practice could go on (by expressing regret, etc.) even if the word ‘apologise’ did not exist. But given that it does go on, and that the relevant situations are common and important in our lives, the appearance of a special word is a natural development of language; and the expression ‘I apologise’ is then available as a way of apologising.

Another difference between promises and statements of intention is that the former are, as Austin put it, ‘performative’. By saying ‘I promise …’, I bring it about that I have promised; but by saying ‘I intend …’ I do not bring it about that I intend. I can say ‘I intend …’ without really intending, but I cannot say ‘I promise …’ without really promising. Nevertheless, as we have seen, ‘I intend’ and ‘I shall’ are commonly used to make promises.

The Place of Promising in Human Life

Promising-situations often develop naturally, and may be hard to avoid even in casual everyday conversation. Alec, having been asked how he is getting on, etc., tells Belinda what he is going to do tomorrow. So far, there is no promising-situation and no promise has been given; Alec has stated an intention, that is all. But now it occurs to Belinda that she might have more than a casual interest in what Alec is going to do. ‘Oh - as it happens, I need … And since you are going to …’, etc. Now the situation has developed into a promising-situation. It is true that Alec, being cautious of committing himself, may try to reverse or prevent this development. Perhaps he will tell Belinda not to count on his doing what he said - that he might not do it in spite of having stated that he would. Such disclaimers are often made, but can we suppose that they would always be possible?

What is Alec supposed to say now? Having told Belinda that he is certain to be there, how can he advise her not to count on his being there? The paradox is the same as might arise with any other assertion. Thus, if Alec told Belinda that Lima is the capital of Peru, and that he is quite certain of this, then he could not go on to tell her that Lima might not be the capital of Peru - that she should not count on it. Let us suppose, however, that Alec is determined to ‘opt out’ of the promising-situation that has arisen.

Let us consider what might be involved in trying to banish the making of promises from our society. Suppose that a law to this effect were enacted by the government. From next Monday, nothing that anyone says can be taken to create an obligation of promise-keeping, and the complaint ‘But you promised’ will be forbidden. Suppose now that you ask me to come and give a talk, or you offer to do X for me if I will do Y for you, etc.; and I agree. But then we remember the new law. What now? Would we not help ourselves by evading that law? Perhaps there would quickly come into being a special sign to indicate ‘I mean it’ - in other words, that I would recognize your right to complain if I failed to do what I said.Footnote ⁶

Similar remarks apply to the related practice of lending and borrowing. I need to use a shovel, and you have one but I haven't. Please may I borrow yours? You can spare the shovel and would like to let me have it; but you want it back in due course. Given the resources of human language, all this can easily be arranged. And in a society in which such situations arise with some frequency, the relevant concepts are virtually unavoidable.

It might be objected that even if all this is in place, the essential thing - the element of obligation - has still not been accounted for. Why, it may be asked, should A feel he is under an obligation? Would he really be under an obligation? Here it is important to distinguish questions of psychology from questions of logic. That A is under an obligation - that he has reason to do what he promised - is a matter of logic; but how he feels about that obligation is a question of personal character.

What, after all, does being under an obligation amount to? Suppose an anthropologist observed that, in a particular tribe, it sometimes happens that B is unpleasant to A, because A did not do what, in a promising-situation, he had said he would do. Then the descriptions ‘He promised’ and ‘He broke his promise’, ‘B reminded A of his obligation’ and ‘B blamed A for not doing what he promised’ would be appropriate ways of rendering into English what was going on in that situation. And likewise, if A responded by saying that he was prevented from doing what he had said, this would be appropriately described as a case of providing an excuse for breaking his promise. To question whether in such cases A would really be under an obligation would be like questioning whether the pieces of paper we use as money are really money – as if there were some standard of reality transcending the role that these artefacts and practices play in our lives.

Promises, Institutions, Games and Rules

My account of promising in terms of natural inclinations and expectations is contrary to a widely held ‘institutional’ or ‘rule-governed’ view of promising. Such a view was expressed by Rawls. According to him, promising is

A similar view was put forward by Searle, who also compared promising with the playing of games. He saw here the answer to the question ‘How can making a promise create an obligation?’ This question, he wrote, ‘would be [similar to] “How can scoring a touchdown create six points?” As they stand both questions can only be answered by citing a rule of the form “X counts as a Y”’. He regarded the obligation of promise-keeping as an ‘institutional fact’, depending for its existence on ‘constitutive rules’ comparable to those that are constitutive of baseball and other games.

The assumption that the ‘institution’ of promising is akin to a game has been made by both opponents and defenders of the ‘is/ought’ principle. When Searle tried to show how ‘ought’ can be derived from ‘is’ in the case of promising, he invited the response that such derivations are valid only for those who choose to participate in the ‘institution’ or ‘game’ of promising. But this choice or ‘endorsement’, it was argued, would amount to the introduction of an ‘ought’ premise, and in that case the ‘ought’ conclusion would be drawn without contravening the is/ought principle. (Hare's response was appropriately entitled ‘The Promising Game’.)Footnote ⁷

But do we really have a choice in this matter? Here the comparison with games breaks down. It is up to me whether to participate in a game, but how would this work in the case of promising? Perhaps one could avoid using the expression ‘I promise …’, but this, as I have argued, would not exclude the making of promises. Could one avoid making promises, with or without this expression? Only - as my illustrations have shown - by cutting oneself off from large areas of normal human conversation.Footnote ⁸

It is typical of games and institutions that certain actions or utterances have a different meaning, or none at all, outside the relevant game or institution. Thus the term ‘off-side’ owes its significance to the game of which it forms a part; and the verdict ‘not guilty’, spoken by a suitable person in a court of law, has a significance that it would not have outside that institution. One might think that the words used in giving a promise acquire their significance through being used ‘within the institution’ of promising, and that their force is, in this way, institution-dependent. But if this were true, then the use of those words outside the ‘institution’ of promising should have different significance – or, perhaps, none at all. But if A tells B that he will meet him tomorrow, knowing that B will count on him, etc., then his words cannot fail to amount to a promise. There is no institution such that, outside it, these words would mean something other than a promise.

Games and institutions are governed by rules. The game of chess, for example, is defined (‘constituted’) by a set of rules, so that a person who played with substantially different rules would not be playing chess. But there is no counterpart to this in the ‘game’ of promising. Again, the rules of a game can be altered by choice. Such rules are devised, adopted and altered for various reasons; and so are the rules of institutions. But there are no such possibilities in the case of promising.

One may be ignorant of the rules of a game or institution, and in learning them one may acquire new concepts, such as ‘off-side’, ‘check-mate’ and ‘strict liability’. But there is nothing of this kind in the case of promising. Perhaps it will be said that the rules of promising are learned at a very early age. What is true is that we acquire the concept of promising in the course of learning to speak; but this is not like learning the rules of a game or institution.

What, after all, are the rules of promising supposed to be? The ‘essential’ rule, according to Searle, is that ‘the utterance of Pr counts as the undertaking of an obligation to do A’ (op. cit. 63); while according to Rawls, the ‘basic’ rule is that ‘if one says the words “I promise to do X” in the appropriate circumstances, one is to do X, unless certain excusing conditions obtain’ (op. cit. pp. 344–5.). Now it is true that if one makes a promise (with or without uttering ‘I promise’) in the appropriate circumstances, then one is under an obligation to that effect; but this is not because of a rule. Suppose I were asked why I am under an obligation to do X: then the answer ‘Because I promised’ would be perfect. No further reason, such as might be supplied by citing a rule, would be called for.Footnote ⁹ And this is not so in the case of games. Why must a player leave the field when he has been caught out? ‘Because that is a rule of the game.’ There is no such question and answer in the case of promising.

Could there be a society without promising? Games and institutions vary from one society to another. The fact that we have cricket and parliamentary democracy is no reason for supposing that these things exist wherever people live in societies. But could there be a society without promising? I have argued that a promise is essentially a matter of telling someone, in a suitable situation, what one intends to do. Could there be a society in which there are no such situations? This depends on what we mean by ‘a society’. Let us assume that this includes the use of language. But here again we must be clear about what is meant by ‘a language’. The communication resources (‘languages’) of non-human animals, wonderful as some of them are, would not be relevant. The language that is needed must include, obviously, the resources for saying what one will do in the future; and it must also be possible to compare what a person said he would do with what he is doing, or failing to do, at the appropriate time. Let us also add the use of questions and hypotheticals. These are important because such questions as ‘Will you come tomorrow?’ are typical ways of eliciting a promise; and so is the hypothetical ‘If I help you today, will you help me tomorrow?’. Finally, some basic facts of human nature must be present - that people have hopes and desires, and are capable of feeling and expressing resentment if these are frustrated by others. Given these conditions, it is hard to see how promising could fail to occur.Footnote ¹⁰