From Mind, 1975, pp. 481, 482-8, 488-95, 496-9. Reprinted by permission of
Basil Blackwell, Publisher.

---

THE purpose of this article is to draw attention to an important similarity between rules and decisions and to suggest a way in which both can be analysed in terms of reasons for action. Rules are not all of a kind. I shall confine myself to the kind of rules and principles with which moral, legal, and political philosophers have been most concerned. These may be called mandatory norms, and they can be affirmed in statements of the form ‘x ought to (or should or must) do A and C’. In the terminology coined by von Wright, ‘x’ designates the norm-subject, ‘A’ the norm-act, and ‘C’ the condition of application.! ‘Promises ought to be kept.’ ‘Parents should look after their children.” ‘Nobody should commit adultery.” ‘When insulted one should issue a challenge to a duel.’ These are well-worn examples of mandatory rules.

1. EXCLUSIONARY REASONS

When conflicting reasons bear on a problem we determine what ought to be done by assessing the relative strength or weight of the conflicting reasons. In the presence of conflicting reasons we say, the agent should act on the balance of reasons. He should act on the reason or combination of reasons which override those conflicting reasons which apply to the problem facing him. These notions are interdefinable. If p is a reason to do A and g is a reason for not doing A (or, as we could for technical convenience say, for doing not-A) then if p is stronger than q it is also the more weighty reason and it overrides g and, therefore, on the balance of reasons it prevails and the agent ought, on the balance of reasons, to do A.
We thus have at our disposal a set of interlocking notions which we commonly employ to describe both the way in which conflicts of reasons are resolved (“He was aware of conflicting reasons but thought that the need to look after his sick child overrides all other considerations’) and the way in which such conflicts should be resolved. The pervasive use of this terminology suggests that all practical conflicts conform to one logical pattern: conflicts of reasons are resolved by the relative weight or strength of the conflicting reasons which determines which of them overrides the other. So long as we are content to handle such conflicts in an impressionistic way there is nothing wrong in this suggestion. But if we are concerned to construct a logical theory of practical conflicts we have to recognize that not all conflicts are of the same type. My claim, yet to be explained and defended, is that we should distinguish between first order and second order reasons for action and that conflicts between first order reasons are resolved by the relative strength of the conflicting reasons, but that this is not true of conflicts between first order and second order reasons.
The distinction between first order and second order reasons for action has not been recognized or discussed by philosophers. This is no doubt due at least in part to the fact that it is not reflected in any straightforward way in our use of the expressions of ordinary language. Reasons of both kinds are referred to as ‘reasons’, ‘considerations’, ‘grounds’, ‘factors’, etc. The resolution of conflicts of reasons belonging to different levels, just like the resolution of conflicts of reasons belonging to the same level, is described in terms of one reason prevailing over, or overriding, or being stronger than the other. So long as we are content to rely on our intuitive grasp of the sense and use of such expressions the distinction between first order and second order reasons need not concern us. My claim is that a useful explication of the notions of strength, weight and overriding is possible but only at the cost of restricting their scope of application and that if we embark on such an explication the theory of conflict must allow for the existence of other logical types of conflicts and of conflict resolution.
I shall proceed by first providing an explication of the notions of ‘strength’ and of ‘overriding’. On the basis of these it will be possible to outline what I shall call the intuitive model of practical conflicts. I shall then provide two examples which do not conform to this model and finally I shall suggest briefly how the intuitive model should be modified to cope with them.
Let us say that p and q are conflicting reasons relative to x, A, and C, if and only if p is a reason for x to do A in C and q is a reason for x to do not-A in C.? Let us say that p overrides q relative to x, A, and C if and only if p and q are conflicting reasons relative to x, A, and Cand ifp&qisa reason for x to do A in C and it is not the case that p & q is a reason for x to do not-A in C.3 In other words p overrides q if they conflict and if the complex fact which is the conjunction of both is a reason for the action for which p is a reason but not for the action for which gq is a reason. The definition does not assume that of every pair of conflicting reasons one overrides the other. It may be that p & q is a reason both for doing A and for doing not-A or that it is a reason for neither. The definition does not and is not intended to help a man decide which reason overrides the other. All it does is explain what he is deciding when he decides that one reason overrides another.
If the reasons for x to do A in C override the reasons for x to do not-A in C then x has a conclusive reason to do A in C.* When a person has a conclusive reason to perform an action then on the balance of reasons he ought to do it. The intuitive model of the resolution of practical conflicts assumes that all practical conflicts are of the type described above and that one ought always to act on the conclusive reason. The intuitive model is represented by the following principle:

P1: It is always the case that one ought, all things considered,* to do whatever one ought to do on the balance of reasons.

I do not wish to chailenge the validity of (P1) directly. Instead I shall show that it is not normally applied to many quite common conflict situations.
Imagine the case of Ann who is looking for a good way to invest her money. Late one evening a friend tells her of a possible investment. The snag is that she has to decide that same evening, for the offer to make the deal will be withdrawn at midnight. The proposed investment is a very complicated one, that much is clear to Ann. She is aware that it may be a very good investment, but there may be facts which may mean that it will not be a good bargain for her after all, and she is not certain whether it is better or worse than another proposition which was put to her a few days before and which she is still considering. All she requires is a couple of hours of thorough examination of the two propositions. All the relevant information is available in the mass of documents on her table. But Ann has had a long and strenuous day with more than the average amount of emotional upsets. She tells her friend that she can’t take a rational decision on the merits of the case since even were she to try and work out the consequences of accepting the offer she would not succeed—she is too tired and upset to trust her own judgement. He replies that she can’t avoid taking a decision. Refusing to consider the offer is tantamount to rejecting it. She admits that she rejects the offer but says that she is doing it not because she thinks the reasons against it override those in its favour but because she cannot trust her own judgement right now. He retorts that this violates (P1) and is unreasonable. Her weariness and her emotional state are not reasons for rejecting the offer. They do not establish that it would be wrong or undesirable to accept it, or that to do so would be contrary to her interests, etc. According to (P1) she should examine the offer on its merits. (P1) does not entail that she should disregard her present mental condition. She must, in following (P1), recognize that her judgement may be affected by her mental state and correct it to prevent this from happening. Ann, however, finds that this will only make things worse. She certainly cannot trust herself in her present state to work out how her mental state might taint her judgement. She insists that though she is taking a decision against the offer, she can rationally do so not on the ground that on the merits the offer ought to be rejected but because she has a reason not to act on the merits of the case. This, she concedes, is a kind of reason not recognized in (P1), but that only shows that (P1) is not valid.
Ann’s case is interesting because she claims to be acting for a reason which is not taken into account in (Pl). It may be that she is wrong in thinking that she has a valid reason for her action, but since the reason she relies on is not uncommon it deserves a closer study. The special feature of her case is not that she regards her mental state as a reason for action, but that she regards it as a reason for disregarding other reasons for action. (P1) allows her to take her fatigue as a reason for going to bed. But she regards it as a reason (or proof of a reason) for rejecting a business proposition despite the fact that her weariness does not bear at all on the merits of the proposition. She claims to have a reason for not acting on the balance of reasons.° Her reasoning is typical of situations in which the agent cannot trust his judgement because he is under pressure of time or because he is drunk or subject to a strong temptation or to threats or because he realizes that he is influenced by his emotions, etc. But such reasoning is not confined to situations of such a nature.
Consider the case of Colin who promised his wife that in all decisions affecting the education of his son he will act only for his son’s interests and disregard all other reasons. Suppose Colin has now to decide whether or not to send his son to a public school. Among the relevant reasons are the fact that if he does he will be unable to resign his job in order to write the book he so much wants to write, and the fact that given his prominent position in his community his decision will affect the decisions of quite a few other parents, including some who could ill afford the expense. However, he believes that because of his promise he should disregard such considerations altogether (unless, that is, they have indirect consequences affecting his son’s welfare). Again, some will think that his promise isn’t binding, but that is besides the point. Our aim is simply to understand the reasoning of those who believe in such reasons, and it must be admitted that they are numerous. Colin’s promise, like Ann’s fatigue, does not affect the balance of reasons. It is not itself either a reason for sending his son to a public school or against doing so. Nor does it change whatever reasons there are. It does not mean that the consequences of Colin’s decision on his chances to write his book or on the decisions of other parents are no longer relevant reasons. They are but Colin has, or believes he has, a reason to disregard them and not to act on them. Colin, like Ann, believes that he has a reason for not acting on certain reasons and that means that he believes that he may be justified in not acting on the balance of reasons.
To explain Colin’s and Ann’s reasoning several new concepts have to be introduced. Let us say that a person does A for the reason that p if and only if he does A because he believes that p is a reason for him to do A. A person refrains from doing A for the reason that p if and only if it is not the case that he does A for the reason that p. In other words a person refrains from acting for a reason if he does not do the act or does it but not for this reason. ‘Refrains’ is used here on an extended sense which does not imply that the agent intentionally avoids acting for the reason. A second order reason is any reason to act for a reason or to refrain from acting for a reason. An exclusionary reason is a second order reason to refrain from acting for some reason. Both Colin and Ann believe that their reasoning is sound because they believe that they have valid exlusionary reasons on which their decisions are partly based. Both reject (P1) because it does not take account of exclusionary reasons. Exclusionary reasons are the only type of second order reasons with which we will be here concerned.
If p is a reason for x to do A in C and q is an exclusionary reason for him not to act on p then p and q are not strictly conflicting reasons. q is not a reason for doing not-A in C. It is a reason for not doing A in C for the reason that p. The conflict between p and q is a conflict between a first order and a second order reason. Such conflicts are resolved not by the strength of the competing reasons but by a general principle of practical reasoning which determines that exclusionary reasons always prevail, when in conflict with first order reasons. It should be remembered that exclusionary reasons may vary in scope: They may exclude all or only some of the reasons which apply to certain practical problems. There may, for example, be some scope-affecting considerations to the effect that though Colin’s promise apparently purports to exclude all the reasons not affecting his son’s interests it does not in fact validly exclude considerations of justice to other people. An exclusionary reason may also conflict with and be overridden by another second order reason. Only an undefeated exclusionary reason succeeds in excluding. If exclusionary reasons are ever valid then the following principle is valid:

P2: One ought not to act on the balance of reasons if the reasons tipping the balance are excluded by an undefeated exlusionary reason.

(P2) contradicts (P1) and if valid should lead to the modification of (P1). The introduction of exclusionary reasons entails that there are two ways in which reasons can be defeated. They can be overridden by strictly conflicting reasons or excluded by an undefeated exclusionary reason.’ It follows that if (P2) is valid then (P1) should be replaced with (P3).

P3: It is always the case that one ought, all things considered, to act for an undefeated reason for action.

2. DECISIONS

In what ways does a decision to perform an action differ from an intention to perform it? It is quite clear that often there is little difference. On many occasions ‘He intends to’ and ‘He has decided to’ are used interchangeably. Yet the central cases of decision differ from mere intentions in important respects and it is to these that we must turn.
Four features characterize fully fledged decisions:

(1) To decide is to form an intention. A decision may or may not involve a mental act of deciding. But even in those cases in which the decision is not crystallized in a mental act it is true that if a person decides at time 1, to do A then for some time immediately before f, he did not intend to do A and for some time after t, he does intend to do A.

(2) Decisions are reached as a result of deliberations. x decides to do A only if he forms the intention to do A as a result of a process of deliberation whether to do A or how to solve a practical problem, where the doing of A is regarded by the agent as a solution to the problem. In most cases a decision results from deliberating on the reasons for or against the action. But a person may decide to perform an action without having first considered the reasons for it, if he has considered some alternative solutions to a practical problem and if the moment the thought of the action occurs to him it appears to him as the appropriate solution to that problem.® ‘Decision’ is sometimes used to apply to an intention formed without deliberation, usually when the agent is aware of conflicting pulls. One may even talk of an unconscious decision. But the central cases are those of intentions formed on the basis of deliberation. Not every intention to perform an action is a result of a decision. It is the process of deliberation as well as the fourth condition discussed below which distinguishes decision-based intentions from other intentions.

(3) Decisions are taken some time before the action. Occasionally we speak of a decision which is immediately carried out. But normally one decides to perform an action some time in the future. It is characteristic of decisions that one can change one’s mind about them. In this decisions are similar to intentions and differ from straightforward cases of choosing. If Jones is offered a tray of different kinds of drinks and takes a martini it would be correct to say that he chose the martini but not that he decided to take the martini. ‘If Jones decided to take a martini, then we expect that prior to the action there was deliberation, or at least preference and resolution, and that between the time he decided and the time he acted we could correctly say that he intended to take the martini.9

(4) Decisions are reasons. The three features of decisions discussed above fail to account for some aspects of decisions. They do not explain why a decision is normally regarded as a stronger indication that the act will be done than an intention to do it which is not based on a decision. Nor do they explain why people often refuse to consider reasons for or against the action they have decided to take on the ground that the matter has already been settled by their decision. The explanation lies in the fourth feature of decisions: A decision is always for the agent a reason for performing the act he decided to perform and for disregarding further reasons and arguments. It is always both a first order and an exclusionary reason. Let me first argue that decisions are exclusionary reasons and then that they are first order reasons.

It should be remembered that a decision is reached only when the agent (1) reaches a conclusion as to what he ought to do and (2) forms the belief that it is time to terminate his deliberations. The first condition is not enough. Imagine a person who considers a problem for a while and then postpones the decision to the next day. At the time he concludes his deliberations for the day he may be as much in the dark as to which decision to take as when he began his deliberations. But it is also possible that he has already formed the view that the proper decision is to do A. That he hasn’t yet decided to do A is not due to any hesitation or uncertainty on his part. He simply wants to consider another argument which he has no time to examine today, or he may want to hear the view of a friend whom he will meet tomorrow. He may be quite certain that the further argument or any facts to which his friend may draw his attention will make no difference to his decision. The reason for which we say that he has not yet reached a decision (and therefore also that he has not yet formed an intention) is not any uncertainty about what to decide or what to do, but only that he genuinely believes that he should consider some further reasons or facts or re-examine his reasoning—just in case. Indeed the following day he may decide that it would be futile to re-examine his reasoning, or that there is no point to wait for the advice of his friend, etc., and decide without further deliberation to do A. His decision consists simply in bringing the readiness to continue deliberation to an end.
To make a decision is to put an end to deliberation. It is also to refuse to go on looking for more information and arguments and to decline to listen to them when they crop up in one’s mind or are suggested by other people. No doubt in most cases the refusal to re-open the case is not absolute. Usually it is accompanied by some unspecified rider—provided no new information becomes available, or, more strongly, provided no major change occurs, etc. Not all decisions are of the same strength, not all of them are subject to the same escape clause. But all of them are exclusionary reasons and it is this which distinguishes between them and mere intentions to act. An intention may often be less liable to change than a decision. But it is always (unless based on a decision) open to the competing claims of other reasons. To decide what to do is to rule out such a competition or at least to limit it.
Similarly though a decision is completely abandoned only when the agent abandons his decision-based intention, it is partly abandoned the moment the agent, still intending to perform the action, is ready to reconsider the case for doing it (unless, that is, his decision was never intended to exclude this factor). That explains why a person may refuse to discuss a problem with another on the grounds that he has already made his decision. That one has taken a decision means that one regards oneself as having an exclusionary reason to disregard further reasons or arguments. To convince another that we hold ourselves open to argument we have to make it clear that we are ready to change our mind and to do so is already partly to abandon the decision.
So far I have tried to show that decisions are exclusionary reasons in the sense that it is logically true that if x has decided to do A then x believes that his:decision is a reason for him to disregard further reasons for or against doing A. It is not part of my claim that all decisions are valid exclusionary reasons, but only that whoever makes a decision regards it as such. A decision is a valid exclusionary reason only if the agent is justified in treating it as such. Often he is not. However, few would deny that sometimes there must rationally be an end to deliberation and indecisiveness even before the time for action. Hence it is clear that some decisions are valid exclusionary reasons. Paradoxical as this may seem reason sometimes requires disregarding reasons for action.
It could be claimed that though it is a necessary truth that whenever a person makes a decision he believes that he has exclusionary reasons to disregard further reasons, it is not the case that he regards the decision itself as an exclusionary reason. This seems to me wrong. To believe that one has a reason not to consider the matter any further is to believe that one ought to decide. One may of course believe that one ought to decide without being able to do so either because one does not know what to decide or because one cannot stop one’s deliberations and form a firm intention. The interesting point is that having taken a decision a person may come to the view that it was a premature decision. He may become convinced not that the decision was wrong but rather that it was wrong to decide at that time. Nevertheless, since he has taken a decision, he now has an exclusionary reason not to reconsider the matter. That the decision was premature is a consideration which may lead the agent to reopen the matter for further consideration but this is never an automatic result. A decision to disregard a decision is itself a new step which should be based on reasons.
Furthermore a decision, like any other action, can be taken for a reason or for no reason. A person may take a decision even while believing that there is no reason for him to take a decision now. He may not believe that he has a reason to exclude other reasons from consideration and to bring his deliberation to an end and yet do so and make his decision. Once the decision is made it is a reason for him to avoid further consideration. If this were not the case it would have been impossible to make a decision without believing that one should decide. These facts show that the decision itself is an exclusionary reason.
The status of decisions as exclusionary reasons may be clarified by comparing them to promises. That a person promised to do A is a reason for him to do so. One should make a promise only if there are sufficient reasons to do so. But once a promise is made it is a reason for action even though it is a promise which should not have been made. Moreover, a person can promise knowing that he should not. Once the promise is made he has a reason to perform the promised act despite the fact that he made the promise knowing that he should not make it. The same is true of decisions. That a person has made a decision is for him an exclusionary reason not to consider further reasons. One should make a decision only if there are sufficient reasons to do so. But once a decision is made it is an exclusionary reason even though it is a decision which should not have been made. Moreover, a person can decide knowing that he should not. Once he has decided, he has an exclusionary reason despite the fact that he decided while knowing that he should not do so.
A promise is a reason which can be defeated by other reasons and the fact that it should not have been made may be relevant to whether or not it is defeated. And this too is true of decisions. Some will think that a promise is a reason only in virtue of a general principle that promises ought to be kept. We could similarly regard decisions as exclusionary reasons in virtue of a general principle that decisions ought to be respected. Both principles need to be spelt out. Both are based on the idea that people should have a way of binding themselves by intentionally creating reasons for action. The Promise Keeping Principle states that they create a reason to do A if they express an intention to another person to be bound to do A.’° The Decision Principle states that people can create an exclusionary reason to exclude further consideration by deciding, i.e. by meeting the first three conditions of a decision and forming an intention to stop deliberation. Both principles are sound practical principles, though both can be abused by making a promise or a decision which should not be made.
It should be remembered that the analogy I am discussing is between the formal features of promises and decisions. Materially they differ. Promises are designed to increase trust and predictability in inter-personal relations; Decisions—to enable people to settle matters in their own mind and put an end to deliberation. The reasons justifying them and determining their strength are entirely different. The formal analogy is, however, considerable. Its most important feature is that a person cannot make a promise without regarding it as a reason for him to behave in a certain way, nor can he make a decision without regarding himself as having an exclusionary reason. (In some institutional contexts such as the law one can both decide and promise without intending to bind oneself.)
That the analogy between decisions and promises is really a close one can be seen by comparing both with oaths and vows. These are often regarded as promises one makes to oneself. They can also be regarded as a kind of decision: a solemn and formal decision with very few escape clauses. I believe that the analogy between decisions and promises can be further explored. For example, both are content-independent reasons: regardless what you promise or decide to do you have a reason to do it because you have promised or decided. To examine the full scope of the analogy will involve establishing two things: (a) that promises are not only reasons to perform the promised act but also exclusionary reasons not to act for other, conflicting reasons; (b) that decisions are also reasons for performing the act one has decided to perform. This investigation cannot be undertaken here. I would like, however, to make a few inconclusive remarks in support of the view that decisions are reasons for performing the act decided upon.
Normally one decides to perform an action because one thinks that all things considered it ought to be done. In such circumstances it makes little difference whether the reasons for the decision are also regarded as the reasons for the action or whether the decision itself is regarded as the reason for the action and the reasons for it are only the reasons for the reason, so to speak, that is they are the grounds for which the reason is considered to be a valid one. Even in these circumstances the agent when asked for his reasons for performing or for intending to perform the action may refer to his decision. But more often than not he will refer to the reasons for his decision. It is tempting to regard such references to the decision as refusals to reveal the reasons or as assurances that the action was taken after serious consideration of the reasons. But an alternative interpretation is also available. We can regard the decision as the reason for the action and maintain that when the agent refers not to it but rather to the reasons for the decision this is because he knows that the purpose of the enquiry will be better served by stating the grounds on which his reason is based. My reason for not stealing may be that I believe that stealing is wrong but when asked why I did not steal I may say that God’s commands must be observed—which is, I am assuming, my reason for believing that stealing is wrong.
A man may decide to do A even though he does not believe that all things considered he ought to do A. He may, for example, believe that he ought to decide and that it does not matter what he decides or he may believe that he ought to decide and not know what to decide. In such cases it is clear that the decision is regarded by the agent as a reason for action. Before he decided he saw no reason why he should do A rather than not-A. Having decided, he has a reason to do A—namely his decision. This argument suggests that at least in some cases decisions are reasons for the act decided upon.

3. MANDATORY NORMS

In the numerous discussions of the nature of rules the one subject that is rarely discussed is the relation between rules and reasons. Yet the relations between these notions must be clarified if the nature of rules is to be illuminated. What ought to be done, all things considered, is to perform that action for which there is an undefeated reason for action. Since there is no doubt that rules are relevant to what ought to be done, there must be a logical relation between rules and reasons. The nature of the relation is rather elusive. Are rules general statements of reasons? Does the statement that it is a moral rule that promises ought to be kept mean the same as the statement that there are reasons to keep promises? This cannot be true. We can state that there are rules but rules are not themselves statements. Statements, unlike rules, can be true or false. Rules, unlike statements, can be valid or void, they can be practised or followed or violated. Both statements and rules can be made, but each is made in a different sense. To make a rule is to change the reasons that people have. By making a statement one may draw attention to reasons but does not.normally change them.
If to make rules.is to change the reasons that people have then rules must themselves be reasons for action.1! But obviously not every reason is a rule. So the question is what sort of reasons are rules? They cannot be reasons of great importance since some rules are of a fairly trivial nature— e.g. some rules of etiquette. Nor can they be reasons which result from the fact that someone in authority expresses the wish or his intention that some people shall behave in a certain way. Moral rules are not laid down by authority. Some have suggested that rules are social (or personal or institutional) practices. There is no doubt that the existence of some practices is a reason for action. One ought to drive on the left because everyone does so. The validity of rules of etiquette is based on the fact that they are practised. But not all rules are of this nature. Many moral rules are usually claimed to be of universal validity regardless of whether they are practised or not.
It may be useful to begin by noticing that the role mandatory norms play in affecting the behaviour of a person who believes in them is analogous to that of decisions. Suppose that I discover some mechanical fault in my car. I decide to take it to the garage the next morning, but to drive in it to a very important meeting today. It occurs to me that at the meeting some acquaintances of mine are likely to ask for a lift home. I do not want to take the risk of giving them a lift, even if they were ready to take it when told of the condition of the car. J know that if asked J will find it difficult to refuse. I therefore decide now not to take anybody in the car that day. I am taking a decision in advance, hoping that this will help me not to yield to requests when I am faced with them. When requested I will have to decide again. I cannot avoid this even if I want to. But this is a different decision—now I have only to decide whether to stick by my original decision. I am merely asking myself whether anything has happened which should make me reconsider the problem.
I may go further. I may, reflecting on the matter, decide on this occasion to make it a rule never to take anyone in my car when I suspect it has some mechanical fault. If so I am simply making a general decision. Of course even if I adopt the rule now I may have to decide again in the future what to do in particular cases, but my problem then will be different from what it would have been had I not adopted the rule. Having adopted the rule what I have to decide is whether to act on it in this particular case. What I am not doing is assessing the merits of the case taking all the relevant facts into consideration. I am not doing this for I have decided on a rule, that is, I have accepted an exclusionary reason, to guide my behaviour in such cases. I may occasionally, of course, examine the justification of the rule itself. If I re-examine the rule on every occasion to which it applies, however, then it is not a rule which J] have adopted. I may on the other hand examine the rule occasionally even when not confronted with a case to which it applies. This is the test by which to determine whether a person follows a rule. It makes clear one advantage of having rules—it makes possible greater flexibility in making decisions, it allows for decisions to be made and revised in conditions other than the occasion for action.
Given our analysis of decisions there is little surprise that if a person decides to follow a rule he has an exclusionary reason to behave in accordance with the rule—the decision is his exclusionary reason. A person may, however, come to follow a rule without having decided to do so. He may have been brought up from early childhood to believe in the validity of the rule and to respect it. He may have drifted into following the rule as an adult gradually over a period of time without ever really making up his mind to do so. It is quite clear that the role a rule plays in the deliberations and behaviour of a person who follows it does not depend on whether he came to follow it one way or the other. A person who follows a rule without having decided to do so may one day critically examine his practice and decide to continue to follow the rule. But the role of the rule in his practical reasoning will not change, now that he has decided to follow it, from what it was before. Indeed this is precisely the purport of the decision—that he should continue to follow the rule as before.
The conclusion we are driven to is that mandatory norms in general play the same role as decisions in the practical reasoning of those who follow them. A person follows a mandatory norm if and only if he believes that the norm is a valid reason for him to do the norm-act when the conditions for application obtain and that it is a valid reason for disregarding conflicting reasons and if he acts on those beliefs. Compare two cases. Jack has friends in France whom he hasn’t seen for many years. Jill has made a rule to spend her holidays in France. Both have a reason to spend the holidays in France this year, but there is a considerable difference between their reasons. Jack’s desire to see his friends will have to compete with all the other reasons relevant to the problem. The trip will be expensive. His wife doesn’t like his French friends and would rather buy a new washing machine. Going to France will mean that he won’t be able to attend an important conference in the U.S., etc. Jack will go to France only if he finds that on the balance of reasons that is what he ought to do. Jill also faces many conflicting considerations, but she has no intention to act on the balance of reasons. She has adopted a rule to spend her holidays in France and she did so precisely in order to spare herself the necessity of deciding every year what to do during the holidays. She adopted the rule in order not to have to act on the balance of reasons on each occasion. Having the rule is like having decided in advance what to do. When the holidays come she is not going to reconsider the matter. Her mind is already made up. Her rule is for her an exclusionary reason for not acting on conflicting reasons as well as a reason to go to France for the holidays.12
This is the benefit of having rules and that is the difference between mandatory rules and other reasons for action. Mandatory rules are regarded by those who follow them as both a first order reason for performing the norm-act and a second order reason for not acting on conflicting reasons. That is why they are adopted and followed whenever the agent believes that he has a reason for disregarding certain reasons: This could be to save time, to avoid the danger of succumbing to temptation or the agony of indecision. Nor are these the only kinds of grounds which can justify mandatory norms. People often believe they have a reason in a particular situation to follow standards laid down by authority regardless of the balance of reasons which apply to that situation. They think that being subject to authority means precisely that. It means that they ought to follow the authority even if it is wrong. One is not submitting to authority if one does what one has reason to do anyway. Similarly one may regard oneself bound to disregard conflicting reasons because one has committed oneself to do so. To be sure rules, though exclusionary reasons, do not exclude all contrary considerations. Depending on the grounds for the rules, they may exclude a wider or a narrower range of conflicting reasons. But all mandatory rules exclude some conflicting reasons and that is the difference between them and ordinary first order reasons.
Not every rule is a valid reason. The point I am concerned to argue is that a person follows a rule only if he believes it to be a combination of valid first order and exclusionary reasons. He may be wrong but for him the rule is a rule only in being such a combination of reasons in the validity of which he believes. To explain what rules are one must do more than explain what it is to follow a rule. We are normally interested in rules either because they are valid or because they are practised by a certain community (or individual or institution) or because they were prescribed by a présumed authority. When the existence of rules is asserted the intention is normally to assert either the validity of the rule (‘It is a moral rule that people should not kill other sentient beings’); or that it is practised (“The rule of vendetta still survives in many parts of the globe’); or that it was prescribed (‘It is a statutory rule of English Law …’). It is not possible here to provide an analysis of the 3 dimensions of rules. Let me just say that all three dimensions can be easily explained once the notion of following rules has been explained. A rule is practised by a person if and only if he follows it. It is practised by a society if the bulk of its members follow it. Institutionalized rules, such as legal rules, are also to be explained by reference to the practices of individuals—in particular the practices of the officials of those institutions. Similarly the notion of a prescribed rule can be explained by reference to following rules. A person makes a new rule only if he expresses an intention that certain people should follow a certain pattern of behaviour as a rule. Finally a valid rule is a justified rule, i.e. a rule the norm-subjects of which should follow. So we see that the notion of following a rule provides the key to the explanation of rules. And this justifies the conclusion, based on the foregoing analysis of following rules, that rules are combinations of first order reasons for the norm-subjects to perform the norm-act when the conditions of application obtain, and exclusionary reasons for not acting on some conflicting reasons in those circumstances. If this is so then the analogy with decisions can contribute considerably to our understanding of the function of rules and their role in practical reasoning.

01 G. von Wright, Norm and Action, 1963, chapter 5.
02 Many first order reasons which are normally regarded as conflicting are not conflicting reasons according to this definition. We could regard the definition as defining ‘strictly conflicting reasons’. Two first order reasons conflict if statements of these reasons themselves or in combination with other true statements conforming to some conditions entail the existence of strictly conflicting reasons. I shall continue to discuss only conflicting reasons as defined above. My conclusion can be easily adapted to conflicting reasons in the wider sense of the term.
03 This definition is based on Prof. Chisholm’s definition of one requirement overriding another, though it is not identical with it, cf. ‘Practical Reason and the Logic of Requirement’, p. 123 above.
04 Unless the reasons for doing A are cancelled. Suppose I make a promise and that later keeping the promise becomes impossible or that the man to whom IJ made the promise releases me from it. The reason I had to do the promised act was cancelled. It was not overridden. Neither the impossibility of keeping the promise nor the fact that I was released from it is a reason for not doing the promised act. I do not have to assess which of several conflicting reasons overrides the other. The reason J had was simply cancelled.
05 T am using ‘ought, all things considered’ to indicate what ought to be done on the basis of all the reasons for action which apply to the case, and not only on the basis of the reasons the agent in fact considered or could have considered.
06 In my example Ann has not formed any view on the balance of reasons. But this is immaterial. She might have formed the opinion that the offer is a good one and refuse it all the same. She may distrust her judgement and refuse to act on it.
07 Strictly speaking there is also a third way in which a reason can be defeated—it can be cancelled as explained in note 4.
08 For a subtle and imaginative analysis of the first two conditions see: Meiland, The Nature of Intention, London, 1970, pp. 55-65. Meiland does not, however, recognize the importance of the other conditions.
09 A. Oldenquist, ‘Choosing, Deciding, and Doing’ Encyclopedia of Philosophy, P. Edwards (editor in chief), New York, 1967, vol. 2. p. 98 (italics added).
10 Cf. my “Voluntary Obligations and Normative Powers’, Aristotelian Society Supplementary Volume 46 (1972), pp. 96-100.
11 Strictly speaking it is not the rule but the fact that there is a rule that .. . which is a reason. It is not, however, possible to explore this point here.
12 Cf. D. P. Gauthier, Practical Reasoning, Oxford, 1963, pp. 160-3 for a similar account of rules.