Joseph Raz, Normativity: The Place of Reasoning, KING’S COLLEGE LONDON LAW SCHOOL RESEARCH PAPER NO. 2015-08; OXFORD LEGAL STUDIES RESEARCH PAPER NO. 3/2015; COLUMBIA PUBLIC LAW RESEARCH PAPER NO. 14-434 (2014).

Available at: https://scholarship.law.columbia.edu/faculty_scholarship/1899

---

An important aspect of the explanation of normativity relates it to the way Reason (our rational powers), reasons (for belies, emotions, actions, etc.)2 and reasoning, with all its varieties and domains, are inter-connected. The relation of reasoning to reasons is the topic of this article.3 It presupposes that normativity has to do with the ability to respond to reasons using our rational powers.4 The question is where does reasoning fit in? I will compare two sketchy accounts. What I call it the simple account takes reasoning to consist (roughly) in responsiveness to perceived reasons. It presupposes that intentions, attempts or actions can be conclusions of reasoning. Those who affirm that possibility often regard reasoning that has such conclusions as practical reasoning. Hence much of the paper will be about practical reasoning. I will illustrate ways in which the simple account is at odds with the concept of reasoning. The alternative sees reasoning as a search for a justified answer to a question, or for the justification of an answer, suggesting a different view of the place of reasoning in explaining normativity. I start by outlining this second account.

1. Reasoning is an Activity:(a) Criterion of Success

Reasoning is something we do. It takes time. It has a beginning and an end. It is up to us whether to do it or not. Even though we sometimes drift into reasoning, and sometimes find it difficult to stop, reasoning is an intentional activity. As with other intentional actions, while they can be intentional without being engaged in for a purpose, or undertaken for a reason, being intentional merely because their conduct is under our control, typically reasoning is an activity in which we engage for a reason. And typically the reason is to find an answer to a question. More accurately, reasoning is essentially an activity aiming to establish the justification, the case for its conclusion, undertaken in order to establish whether the conclusion is a correct answer to its question.5
The article will expose certain ambiguities in this formulation. But why take reasoning to be a search for an answer or a justification of an answer to a question? Why does it not simply establish that a conclusion follows from some premises? Some conclusion follows from anything one takes as a premise. Why reason from these premises rather than others? Given that it is an intentional activity the answer is that there is reason to engage in it (reason for the reasoner to engage in it now), which is that these premises help with establishing something that the reasoner has reason to establish, which is more or less the same as saying that the reasoner has reason to find a justified answer to a question. 6 One condition for the success of the reasoning is that the question is well-formed, meaningful. Another is that its conclusion answers the question. A third is that the reasoning shows the answer to be correct. Relevance to justifying an answer to the question establishes which premises to rely on.
‘Conclusion’ is broadly speaking the end, termination of something. The (“real”) conclusion of reasoning is its outcome; not necessarily its temporal conclusion. Often the reasoning continues after having reached the “real” conclusion, when, for example, one re-examines its justification (without changing one’s mind about anything at the end). Sometimes the “real” conclusion is known, and the reasoning aims to confirm it. The “real” conclusion is the answer to the question the reasoning is about. But answering a question need not involve reasoning. You ask me for my name or for the way water came to be on Earth, and I know the answers and give them to you, without having to reason in any way at all. Similarly, one may know the proposition that is the answer to a question without knowing that it is the answer. One may know that there was a drought last year without knowing that that is the answer to the question: ‘what caused the crop failure last year?’ Reasoning involves looking for an answer, and for a justification of its being the answer.
A “justification”, in this context, is an argument, namely a statement of a reason or set of reasons given in support of an answer to the question the reasoning is about. The reasons are expressed in propositions, and most commonly when we report on our reasoning after it concluded we express the argument (or parts of it) that justifies its conclusion. If, as we see it, the reasoning ended in failure we commonly report on it by stating what question we tried to explore and for what kind of argument we were looking. So the “real” conclusion is a proposition that is taken by the reasoner to be an answer to the question. It is the conclusion of the argument that is taken by the reasoner to justify belief in it. Belief in the conclusion of the argument is the (“real”) conclusion of the reasoning. The reasoning is abandoned and remains incomplete if the reasoner does not come to or reaffirm belief in the conclusion of its argument. For example, the reasoner may realise that his premises point to a certain conclusion, that P supports C, and yet stop short of believing that C. He has reached an interim conclusion, one that he believes, but has not yet answered his question (say, whether C?). Not endorsing/believing the conclusion of his argument, the reasoner has not yet reached the conclusion of his reasoning. He may continue it or abandon it.7
How much support must the argument lend to its conclusion to be valid (I use ‘a valid argument’ to designate that the support it gives its conclusion is adequate)? It must support it to a degree that would justify the conclusion of the reasoning, namely endorsement/belief in its conclusion, that is embracing the conclusion of the argument as an answer to the question the reasoning sought to answer. With this in hand we have the case for taking reasoning as aiming to establish the justification of an answer to a question. The question the reasoning is about is determined by the reasons for reasoning, and the question determines the strength of support the argument must provide for its conclusion, for the strength must be such as to justify endorsing it as the answer to the question.8
‘Justified in …’ does not entail ‘not justified not to …’. Is this consistent with the fact that to be valid an argument must establish that there is something amiss in believing/endorsing its premises without believing/endorsing its conclusion? It is, as I will illustrate in the case of justified belief, which I take to be an abbreviated way of saying that someone is justified in having that belief in the circumstances of the relevant time. There is a difference between justified action and justified belief. For a person at a point in time it may be justified to do or not to do any number of incompatible actions or activities. That is because any number of incompatible options may be supported by undefeated reasons. Each practical reason is a fact that establishes that there is some good in the action for which it is a reason. So that when two incompatible actions are supported by undefeated reasons there is some good in each of them, and nothing to make a stronger case for one than for the other (this is what being undefeated entails). Epistemic reasons do not show that there is some good in the belief they are reasons for, rather, they show that there are some indications that it is true. If there are also equally strong indications that it is false there is no reason supporting either that belief or its negation. Therefore, for an argument to justify a belief (for a person at a time) it has to establish that the case for the belief is stronger than the case for its contradictory. That there is some case for that belief and some case for various contrary beliefs does not justify any of them.
Some epistemologists have taken to using concepts familiar from practical contexts, such as ‘being entitled to believe that …’ or ‘being permitted to believe that …’. While these tendencies may speak of a growing interest in a general study of normativity, they overlook differences between normative domains. There is no sense in which a belief is permitted or prohibited (except by obnoxious laws or customs). Possibly, the ill-considered use of these concepts is encouraged by two considerations. First, the fact that the strength of the case required to justify the conclusion of the reasoning (the ‘degree of proof’ as lawyers say) is relative to what is at stake (criminal conviction or a private law remedy – to stay with the legal example). This appears to suggest that practical considerations are (sometimes) epistemic reasons. In fact it shows that they are among the considerations that determine the strength of the support the argument has to provide for its conclusion to be valid. In any case there is nothing here to suggest that the conclusion is “merely” permitted. Second, in some domains, there is a vague range of strength of arguments where suspension of belief in the face of them is no fault, even though forming or sustaining the belief on their basis is justified. Whatever the explanation of this latitude the phenomena to be explained do not suggest that people have within the range of latitude a permission to believe or not to believe. That would mean that they can choose whether to believe. But in these as in the general case we do not choose what and when to believe. Rather, people’s epistemic functioning differs regarding the strength of support that would lead them to have a belief, and within a certain range that functioning is rational.
In conclusion: We can of course assess an instance of reasoning by various criteria (was it efficient? elegant? etc.). The basic standard of success is (a) it was (at the start) reasonable to take the question to be well conceived; (b) the conclusion of the argument is an answer to the question (rather than being evasive, changing the question, etc.); (c) the argument associated with the reasoning is valid, it establishes that belief in its conclusion is justified; and (d) the reasoner concluded his reasoning by coming to believe, or confirming his belief in the answer to the question as a result of the reasoning.9
By this standard the success will be relative to the reasoner’s other beliefs and to his rational capacities (was successful given that the reasoner was a high school student, but would not have been had she been a research physicist). The standard relies on the semantics of questions and on the theory of valid arguments (deductive logic, non-monotonic logic, methods of experimental inquiry etc., including the considerations that establish what can and what cannot be a reason for what) to determine the relevant parts of the tests for success. As indicated I rely on only one aspect of the theory of valid arguments10: whatever kind of argument is concerned, it is common to all rules of valid arguments that believing or relying on the premises and rejecting or refusing the conclusion involves some defect, some imperfection. The right reaction, all things told, may be to believe in the premises and refuse the conclusion. But even so there is some epistemic imperfection in that condition. If one had more knowledge about the way things are, that outcome would have been avoided; the rational outcome would have required revising some of the premises or believing the conclusion.11

2. Reasoning is an Activity:(b)its Scope

The criteria of success in reasoning allow that the reasons for reasoning may vary, relating to a case for finding an answer to the question, confirming or refuting an answer one has, finding an argument justifying the answer, or confirming or refuting an argument one has. But, as not all reasoning is successful, the criteria of success presuppose another criterion determining what is reasoning.
There are two broad cases in which people reason unsuccessfully. They may reach a conclusion on the basis of a defective/invalid argument, or they may not come to a conclusion, but abandon the reasoning incomplete. A distinct instance of reasoning is incomplete if it does not end with an answer (true or false) to its question. One can of course break off with a view to continuing it some other time. It can be cut short when one realises that it may be dangerous, or otherwise undesirable, to have the answer or to continue with the inquiry, or for other causes.12
What makes unsuccessful reasoning (of either kind) reasoning is what makes successful reasoning reasoning. One distinct activity of reasoning is that of a person who takes it to be successful if, and because, it meets something like the criteria of basic success I outlined. These conditions of success are deliberately vague. They are meant to help us identify ordinary reasoning. Reasoning, being an intentional activity, cannot be successful accidentally (or rather what makes it reasoning is not that it is an activity that would have been accidentally successful reasoning had it been reasoning). And it can be reasoning even if unsuccessful. What makes the activity one of reasoning is the recognition by the reasoner that the activity is successful if it is successful as reasoning.13 But that cannot be a reference to an ideal or correct standard of success. People who have mistaken views about the standards of success for reasoning may still be reasoning, provided their understanding of these standards is not too remote from the correct ones. Therefore and roughly speaking, reasoning is an activity attempting to be successful by those standards; or alternatively, an activity of a person who takes its success to be determined in that way. That ‘taking’ is manifested in accepting that deviation from relevance and from the other rules governing arguments is a mistake, being willing to correct such mistakes, and by realising that one was mistaken if deviations from the rules come to one’s attention after the reasoning is completed.
The above criterion identifies instances of reasoning by their core. It does not determine their outer contours: what does and what does not belong to a single complete instance of reasoning? When does it begin or finish? There is no point in pursuing these questions to the bitter end. Regarding many mental acts there will not be a fact of the matter whether they belong with the reasoning or not. But some broad criteria of the scope of a complete instance of reasoning are part of our understanding of what reasoning is.
The reason for reasoning determines its scope. There are two levels of reasons involved: the reasons for seeking either a justified answer to the relevant question, or a justification for the answer; and the reasons for conducting the reasoning in the way it is conducted (for pursuing subsidiary questions, for collecting and assessing certain data, etc.). The reasons for the way to conduct the inquiry are governed by the reasons for having it, but follow general principles regarding the conduct of inquiries about issues of the relevant kind. As we saw, given that the reasoning aims at a justification of a conclusion that is the reply to the investigated question, it is natural that in reporting on it one would produce the argument to the conclusion. But, of course, the stages in the presentation of an argument, proceeding in an orderly way from premises to interim conclusions to further premises, etc., to the conclusion, are rarely if ever in the temporal order of the stages of the actual reasoning. Some of the stages gone through in the reasoning directly reflect stages in the argument (call them the primary stages). They involve activities such as coming to view one of one’s beliefs as a relevant premise in the argument, or postulating for the sake of the argument some assumption, or drawing an interim conclusion from some of the premises, etc. Others are connected to the primary stages, being ways of bringing them, or their content, to mind, focussing attention on them, being aids to seeing their relevance and interconnections. They involve free floating ruminations, searches for ideas, vaguely coming to feel that some ideas are unlikely to work, and more. All these activities happen in the shadow of pursuing a justified answer to a question, the pursuit that in its totality is one’s reasoning.
There is no reason to think that the loosely structured way in which we reason is disadvantageous, that we would have done better to regiment our reasoning and limit it to its organised primary stages. So the best we can do in determining the stages of the reasoning is to say that broadly speaking one complete instance of reasoning consists of the various activities and attitudes that are part of the search for the answer to the investigated question and for its justification.
What makes them part of that search? We can reject the suggestion that mental activities or processes that causally contribute to the conclusion of the reasoning are part of the reasoning. Not all phases of the reasoning are causally productive, or contribute to its conclusion. I do not mean that some of them lead to dead ends etc. Those can be constructive. I mean, for example, cases in which as the reasoning proceeds we forget what we did before and have to repeat our exploration. Furthermore, not all psychological processes that are causally productive towards the conclusion of the reasoning are part of it. One familiar and dramatic example is the sense people have that something happened during their sleep, or at a time when they put the problem aside and went swimming, something that suddenly opened the solution to their mental gaze. No doubt such things happen: psychological processes contribute to finding the argument and to leading to the conclusion, in ways that we are not aware of. But they are not part of the reasoning.
Mental activities and processes are part of the reasoning only if they include some conscious thoughts, and are governed in part by recognition that they strive towards an argument that would justify an answer to the relevant question. That recognition takes the shape of a feedback loop whereby steps that deviate from the goal are rejected. The feedback loop itself is not necessarily conscious, and the reasoner will often be unable to articulate its nature. But its operation constitutes the reasoner’s control of the course of reasoning, a loose control that allows for many activities and processes not consciously controlled and not strictly governed by reasons to proceed this way or that, but none the less directed towards the goal of the reasoning. Concurrent attitudes and activities that are not related in the right way to the goal of the reasoning are not part of it.
It follows that not all the activities that are part of one’s reasoning are strictly guided and organised by reasons. They are governed by the reason to look for a justification for an answer that led to the reasoning, and that means that the reasoner acknowledges that the process and its results are successful if they lead, reasonably efficiently, to a successful conclusion of the reasoning, i.e. if the reasoner acknowledges that the activity he or she is engaged in is subject to criteria of success like the ones enumerated above.

3. An Objection and the Simple Account

If reasoning is a search for a justified answer to a question its conclusion can only be a belief or a proposition believed. Rhetorical metaphors aside, a killer asked what did he do (he killed Jones) cannot reply ‘I answered the question what to do with Jones?’ But there is a tradition going back to Aristotle taking actions, attempts or intentions to be the conclusions of some instances of reasoning, often said to be practical reasoning. That cannot be shown to be wrong by taking one’s starting point to be that reasoning is a search for an answer to a question. And the argument I gave for that starting point may not be sufficient to settle the issue.
Moreover, the account I offered, it can be objected, arbitrarily discriminates between reasons for belief and reasons for actions, intentions or emotions. It takes reasons for belief to be followed by reasoning to the belief one has a warranted reason to have, but requires a different account for following reasons for actions, intentions or emotions. Why is not reasoning the way to follow them as well?
The focus of an account of reasoning should be, the objection runs, on elucidating the relations of reasons to reasoning. This suggests a simple alternative account of reasoning. It may not have been defended in quite that way by anyone14, but it will be useful to examine the simple account (as I shall call it) as a way of bringing to light an important difference between two approaches to the understanding of reasoning.
The simple account consists of two propositions:

SA (P1): Successful reasoning is recognizing that something is a reason and responding to it, in the way it makes appropriate.
If, having realized that all told I should take this medicine, I take the medicine, then I reasoned from the premise that I have reason to take the medicine and other relevant premises to the conclusion that was the taking of the medicine. If realising that today is Monday I come to believe that tomorrow is Tuesday then I reasoned from the premise that today is Monday and other relevant premises to the belief that tomorrow is Tuesday. Not all reasoning is successful. So the simple account contains a second part, which goes roughly as follows:

SA (P2): an activity that is taken by the agent to be successful if and because it is successful reasoning is reasoning.
The simple account offers an explanation of the relations between reasoning and normativity: reasoning is nothing but a way to respond to normative reasons. Generally speaking so long as one has the belief or emotion or performs the action for which one has adequate reason one is free from fault. 15 But that may be due to luck or coincidence. Only when one’s conformity to reason is due to reasoning, i.e. recognizing and following the correct reasons, does one display the skills and attitudes that constitute rational responsiveness to reasons, as one also does when one reasons from perceived reasons that happen (not because of the malfunctioning of one’s rational powers) to be mistaken.

4. Simple Objections:

The simple account encounters difficulties. Suppose that you ask yourself whether the Conservative Party will win the next general election. As you are deliberating a friend rings to ask you to meet him. You start preparing to go out. You were reasoning about the outcome of the election but did not come to any conclusion. Your reasoning was interrupted. Now suppose that you consider how to get to the airport the following morning. You conclude that you should leave at 6 a.m. to catch the 6:45 train. You fail to leave at 6. You did not interrupt your reasoning without concluding it. Your reasoning was complete, and your conclusion was not a mere interim one. Failing to leave at 6 was not a failure to complete the reasoning. It is natural to think of your reasoning as a practical reasoning: you were deliberating about what to do. If so then some practical reasoning, in some sense of the word, does not have actions as its conclusion.
Suppose that you are walking to work. It occurs to you that you will not have time to lunch at the cafeteria. You wonder whether to cross the busy street to get a bun. Will this make you late for work? Would it matter if you are 10 minutes late? You conclude that all things considered it would be best to get the bun. You turn towards the traffic light when you stumble, knock your head and are ferried to A&E. Did your reasoning whether to get the bun remain unfinished, being interrupted by your fall, just as your reasoning about the election remained unfinished, being interrupted by your friend’s calling you? No. You concluded your reasoning, and your not acting on it came later. It was not an interruption of your reasoning.
Perhaps, actions are not the conclusions of practical reasoning; perhaps its conclusions are intentions? That view does not conflict with the preceding observations. And it is quite natural to say that I concluded my deliberations, forming the intention to ϕ. But then, it may be that the intention merely followed the conclusion of the reasoning rather than being its conclusion. For other indications suggest that it is not the conclusion. For example, knowing that you were not sure whether to apply for a certain job I ask you: ‘Have you concluded what to do?’ and you may say: ‘I thought about it all day, and I know what I should do, but I am not sure what I will do’, suggesting that you concluded your reasoning, but not by forming an intention. After all I cannot follow up by asking you: ‘when will you finish your reasoning?’. It is clear that that is over. Only the problem of resolve remains. Or, when a friend who freely admits that he knows that he should give up smoking confesses that he cannot bring himself to decide to do so, I cannot reply with ‘I did not realise that your reasoning powers are so poor’. His failure is one of resolve not of reasoning.
These observations do not amount to a conclusive argument. Some people would deny that they are objections at all. They simply beg the question, they will say. The so-called objections presupposed that neither intentions nor actions can be the conclusions of reasoning. They did not establish that this is the case. This is not quite right, however. True, the objections presupposed something. They presupposed that we are familiar with the concept of reasoning, and, barring difficult cases, we know what is reasoning when we see it, even while we are unable to provide an account of what it is. They take the cases to illustrate straightforward situations regarding which the simple account is mistaken.
If intentions and actions cannot be conclusions of reasoning a natural assumption is that all reasoning concludes with a belief or beliefs. There are additional candidates. Could not a supposition be the conclusion of reasoning? Or could not the acceptance of a proposition (e.g. accepting someone’s innocence) be such a conclusion? I will not consider these possibilities. ‘Accepting that …’ is a mental act. If acts are not conclusions of reasoning neither is acceptance. Suppositions are different, being more like beliefs, but they seem to be subsidiary types of conclusions, mostly or always intermediate ones, and for present purposes can be left on one side.
Beliefs are the conclusions of at least some classes of reasoning. To remind ourselves: they need not be new beliefs. Reasoning can conclude in endorsing or reinforcing or weakening an existing belief. Reasoning to a belief (like many other ways of forming beliefs) is subject to the forms of interference familiar to anyone who considered the formation of intentions (and the processes leading to or frustrating the attempt to perform actions). Reasoning to a belief can be distorted by various forms of motivated irrationality (wishful thinking, rationalisations of desires for revenge, desires to please, etc.) or fall prey to other forms of distortion (anxieties, lack of resolve due to low self-confidence, low self-esteem, and others). They may lead to unsuccessful reasoning, and sometimes to incomplete reasoning. Whereas, on my account, reasoning that fails to lead to intentions or actions is not, in virtue of that fact, incomplete. That intentions are subject to akrasia and similar distortions does not establish that they cannot be conclusions of reasoning. But if they cannot, we must accept this asymmetry as a feature of reasoning.
The simple objections suggest that the simple account may be closer to the truth regarding reasons for beliefs than regarding other reasons. But such a restriction appears arbitrary and unmotivated. It is therefore not surprising that there are reasons to doubt it. The simple account, restricted or otherwise, purports to offer a sufficient condition for reasoning: Given that reasoning has to do with rational reaction to reasons, that is the only part of it that is in doubt. But, if F is a reason for the agent to R, not all ways of coming to R when taking F to be a reason for it are cases of reasoning. For example if I believe that John gave me a present for my last birthday because I remember his doing so, no reasoning need have been involved in forming the belief, even though that I remember him giving me the present is a reason to believe that he did. If sound this point refutes the simple account. It does not provide a sufficient condition of reasoning to a belief, any more than of reasoning to an intention.16 This, however, does not show that intentions or actions cannot be the conclusions of reasoning. We need to examine it further.

5. Practical Reasoning:

We are trying to establish whether only beliefs can be the conclusions of reasoning in order to understand the role of reasoning in an account of normativity. One theoretical case for taking actions or intentions to be possible conclusions is that otherwise one cannot explain practical reasoning. If only beliefs can be conclusions what is practical about practical reasoning? This section will challenge this case.
One answer is that practical reasoning is reasoning in search of an answer to a practical question. ‘What is to be done?’ may stand as the prototype, though variations range wider than in time (what was to be done? etc.), modality (what is one permitted to do? Must do? etc.), relevant agent (what is the government to do?) or circumstances (what is to be done if things are so and so?). Reasoning aiming to answer some other questions may be classified together with the above. For example, reasoning about whether it would be cowardly or disloyal or mean or vain to act in a certain way, or to have certain feelings.
Making the classification of reasoning as practical turn on the question explored, is in line with other classifications of reasoning, as economic, or educational and so on. Practical reasoning in that sense does not seem to be governed by special rules of inference. It is modal and defeasible (non-monotonic) but so is much ordinary reasoning. It deals with concepts, such as rights and duties, which have their own concept-specific transformation rules. It is common that any sphere of discourse or learning supplements the general rules of inference with concept-specific transformation rules. On this view, practical reasoning is ordinary reasoning regarding a particular range of questions.
Does that show that its conclusions are beliefs and not intentions, for example? It does not. The issue is not terminological. There is nothing inappropriate in using ‘practical reasoning’ as the name of the class of reasoning I characterised. But perhaps there are cases of reasoning (perhaps they are a subclass of practical reasoning as defined) whose conclusions are (the formation of) intentions. Call them P2 reasoning, and those that are practical according to the account I just gave, P1 reasoning. There are many cases of P1 reasoning that cannot end with the formation of an intention, i.e. the premises that lead to their conclusion do not warrant forming an intention.17 For example, I may reason now what to do now, and I may reason tomorrow about what I should have done now. Both episodes of reasoning may be identical in all respects (allowing for modulation of temporal reference) except that, if any reasoning can conclude with an intention, only my current reasoning can conclude with an intention to do something now.
Reasoning triggered by a ‘what am I to do?’ question can conclude with the belief that I have an adequate reason to ϕ. It will guide me in my subsequent thoughts and decisions. But neither it, nor the premises that led to it, require forming an intention to ϕ. For example, there may be quite a number of incompatible acts that I have adequate reason to perform, but it may well be irrational to form intentions to perform each action that I know to be supported by an adequate reason, and so far as my reasoning goes nothing wrong in not forming an intention to ϕ.
In other cases, even though it may appear that an intention can be formed when the reasoning concludes with belief that one ought to ϕ, no intention need be formed. Suppose that I conclude that I ought to do something for my child when he reaches maturity (he is now 3 years old). My conclusion notwithstanding I form no intention to do so, not because of any doubt, or weakness, but simply because it is not necessary. According to the definition proposed above I engaged in and concluded my practical reasoning on the issue (it is a P1 reasoning). According to the intention as conclusion view I engaged in theoretical reasoning only (it is not a P2 reasoning).
I hesitated in introducing this example, for future intentions present more complex features than is often appreciated. Most relevant to our concern is that while future intentions can be formed without the agent as much as noticing the fact, when they are consciously and deliberately formed, forming them is an act that is justified when supported by adequate reason. As argued by Ulrike Heuer, the existence of a reason to perform an action at some time in the future is not a sufficient reason to form now an intention to do so.18 Additional considerations are needed to establish a case for forming the intention. So, that I ought to or must ϕ in the future does not, without further premises, warrant forming now an intention to do so, and needless to say forming such an intention would not be a valid conclusion of the argument that concludes that I ought to ϕ in the future. Moreover, when the extra considerations are available, the conclusion will often be that forming an intention is permissible, rather than required. An intention to do something is not a valid conclusion of a reasoning that establishes no more than that forming or having the intention is permissible.
Such cases illustrate the range of types of reasoning often thought of as practical that do not warrant an intention as a conclusion, even if intentions can be conclusions of reasoning. The following cases illustrate a more far-reaching point: They concern cases in which I ought to ϕ, but have no reason to form an intention to ϕ and have an adequate, perhaps conclusive reason not to intend to ϕ.
For example, suppose that I ought not to act disloyally, even when doing so would benefit the person to whom I would be disloyal. My reasoning led me to this conclusion. I never act disloyally in such circumstances, but I never form an intention not to do so. Why should I? After all I am not in the least tempted to act disloyally in that, or almost any other situation. The thought of doing so never enters my mind. Forming the intention not to be disloyal appears to me to be demeaning, to be saying to myself that I need to resolve not to be disloyal, otherwise there is a risk that I will be. According to the proposed classification I engaged in practical reasoning, and I live by it. If I were open to temptations of disloyalty I would have had a reason to form an intention not to be disloyal. But as I am not, I have no such reason, and have an unopposed reason not to form such an intention. My conduct is affected by the belief with which my P1 reasoning concluded. But I did not engage in P2 reasoning.
Nor are omissions the only cases of this kind. I may have a reason to perform an act that I will indeed perform, yet it may be demeaning or otherwise undesirable that I should form an intention to do it. It is said that Kant decided that a daily walk was what his health needed. So each day, come rain or shine, at precisely 3.30pm, he would emerge from his lodging, and walk up and down the street. Legend has it that so punctual and reliable was his walking routine, the neighbours used to set their clocks by him. Enabling the neighbours to set their clocks was, no doubt, a good thing, and Kant may well have become aware of the facts. Yet, I would understand someone in his position who sees no reason to, and does not form the intention to go out punctually at 3:30. He grants that he ought to go out at 3:30 (until he gives adequate notice that he will not). But given that he would turn out at 3:30 precisely, without intending to turn out at 3:30 precisely (his daily routine naturally leads him to do so), he may well reject the idea that he is a local clock.19
In such cases we have reasoning that rightly concludes that one ought to ϕ, and yet one has no reason to intend to ϕ and there are adequate reasons not to intend to ϕ. Both the reasoning to the conclusion that one ought to ϕ, and the reasoning to the conclusion that one may not, perhaps even should not, intend to ϕ are practical in the P1 sense, and guide the reasoner’s conduct. But neither is P2, nor can they be turned into valid P2 reasoning. Yet it seems odd to conclude that there is no practical reasoning that applies to these cases. So, possibly even advocates of intentions as the conclusions of some practical reasoning would allow that not only reasoning concluding with an intention can be practical.
None of this argues that intentions cannot be the conclusions of some cases of reasoning. However, the examples undermine the thought that reasoning cannot be practical, or cannot guide people’s conduct unless its conclusions are intentions or actions. The loyalty case shows that if there are cases of reasoning concluding with an intention they are not more practical (namely have no greater influence on one’s conduct) than at least some cases of reasoning ending with a belief about what one should do. An argument against intentions as conclusions of reasoning has to do more. It has to be grounded in an understanding of the nature of reasoning.

6. The Role of Reasoning in our Normative Functioning:

Just as successful reasoning to a belief establishes that there is something amiss, some imperfection, in retaining all the premises and not believing the conclusion, so to have an intention as its valid conclusion the reasoning has to establish that there is something amiss in retaining all the premises and not having the intention. Reasoning that an intention is permissible concludes with a belief to that effect. The permitted intention cannot be its conclusion. There is no blemish in not taking advantage of a permission. Therefore, for an intention to be a conclusion of reasoning, that reasoning must show that there is a conclusive reason to form or to have that intention. It follows, I will argue, that if an intention is the conclusion of a valid reasoning, the argument associated with the reasoning must include an interim conclusion that there is a conclusive reason to form or to have that intention. Borrowing Broome’s style of presenting the argument that underpins reasoning that concludes with an intention, my contention is that a valid reasoning to an intention is supported by a valid argument a fragment of which has an intermediate conclusion (or premise) of the form:

(IC) There is a conclusive reason to intend to ϕ. And whose conclusion has the form
(AC) I shall ϕ (this being an expression of an intention). And people reason validly to the intention to ϕ when they form (or maintain) the intention because they take the argument to show that there is a conclusive reason for it.

This contention relies on two claims: (1) the argument that underpins the reasoning must include the interim conclusion (IC). (2) To reason validly to the intention, reasoners must believe or endorse (IC). Some people doubt one or both of these claims. They do not deny that one can reason to (IC) and from it to the intention. But they deny that believing/endorsing the interim conclusion and reasoning from it are necessary steps in reasoning to an intention.
The argument from (IC) to (AC), if valid, is valid in virtue of an inference rule sanctioning that transition. Call it the Intention Derivation Rule (IDR). Those who deny my first claim, namely that in reasoning to an intention (IC) is a necessary step, cannot rely on IDR to validate the argument that they have in mind, the one lacking (IC). What alternative inference rule do they have in mind? It cannot be the inflationary rule saying something along the lines: ‘given reasons for and against an intention form the intention’. What then is it? When we reason whether to intend or to act (the more common way of becoming aware of reasons to intend), the argument we come to rely on is normally complex. It identifies reasons for and against the intention, determines their relative stringency and strength by exploring some of their implications, on the basis of which it determines that there is a conclusive reason for the intention. We expect many rules of inference or other transition-sanctioning rules to be employed in the course of such arguments. It is unlikely that they can be lumped together into one, but even if they could that one would include as a phase in the argument the determination of a conclusive reason for the intention. That has already been established above. There is no rule or set of rules of inference that avoids IDR. Therefore it is impossible to avoid (IC) as a step in such an argument. That disposes of the objection to my first claim.
The objection to the second claim must therefore allow that (IC) is a necessary step in the argument, for it must allow that the move to an intention relies on IDR. Needless to say, it must also allow that the formation of an intention is the conclusion of reasoning only if one forms the intention because of a realisation that the underpinning argument requires it. However, the objection proceeds, that realisation need not take the form of a belief in (IC). One can rely on (IC) without believing it. How so? To reiterate, the intention can be a conclusion of reasoning only if it is formed because the reasoner takes the reasoning to have established a conclusive reason for the intention. Does that not mean that the reasoner came to believe that there is a conclusive reason? True, it may well be that the thought that there is a conclusive reason did not occur, that the reasoner did not consciously think: ‘there is a conclusive reason … ’.20 But most of our beliefs never featured in our thoughts. Determining when one has a belief is a difficult task, but given that the reasoner relies on there being a conclusive reason because his or her argument establishes that there is one I see little doubt that he or she believes that there is such a reason.
The objections having failed, it turns out that any reasoning concluding with an intention divides into two stages: first, reasoning to a belief that there is a conclusive reason for an intention, followed by a second stage consisting of forming the intention on the basis of that belief.
The emerging picture appears to confirm the earlier surmise, namely that most of what we may have in mind when thinking of practical reasoning is reasoning about a specific domain, or domains, and like any other reasoning its conclusion is a belief. The conclusions of much, though by no means all, such reasoning may be properly followed by the reasoner forming an intention. That would be the case when the reasoning showed that there is an adequate reason for having or forming such an intention, and the reasoner chose to form that intention. Often, the reasoner will not form the intention, and there may be nothing amiss with that choice. Only in a proper subset of cases of such reasoning will the appropriate conclusion indicate that there is a conclusive reason for having or forming an intention.21 In some of those cases the reasoner will not be able to form the intention, but normally forming it would not be impossible. Indeed it will be what one should do after concluding the reasoning.
It seems clear by now that the simple account cannot be correct. Reasoning is not the way in which we intentionally respond to reasons. We had a counter example in section 4 showing that not all beliefs formed for reasons are formed through reasoning, and the cases in which people form intentions because they have adequate reasons, but no conclusive reasons, to have them add counter examples. We still do not have an argument to establish that intentions cannot be the conclusions of reasoning. But we edged further in that direction. The last section removed or weakened the theoretical need to suppose that intentions can be conclusions of reasoning. The current section shows that there is an oddity in supposing that they are.
That is because successful P2 reasoning consists in nothing more than reasoning from ‘there is a conclusive reason to intend to ϕ’ to intending to ϕ. By now it may well appear that the step from that belief to the intention is not one of reasoning. Why not? It is not that reasoning requires greater complexity. I do not know of a measure of complexity that would establish this contention. Nor is it that there cannot be reasoning from a single premise. There are cases in which it is possible to believe the premise without believing the conclusion, and yet reflection on the premise, attending to the premise, may convince one of the conclusion. Furthermore, even if one believes in both premise and conclusion one may be unaware of the connection between them, but become aware of it when reflecting about, attending to, the premise. In such cases one would be reasoning from the premise to the conclusion. Finally, if the transition from ‘there is a conclusive reason to intend to ϕ’ to intending to ϕ is not one of reasoning, that is not simply because the conclusion is not a belief.
The problem is not directly with the content of premises or conclusions. It has to lie with the kind of transition that constitutes reasoning, and identifying it is a central part of an account of the nature of reasoning. If certain alleged conclusions (e.g. actions), or combinations of premises and conclusions are ruled out that is because the reasoning-transition cannot obtain there. The remarks about reasoning from a single premise indicate the direction of travel: reasoning is an activity whose success depends on coming to realise that some items (premises) support others (conclusions). Not only the existence of the support relationship is essential to it, but the coming to realise that the relation obtains, not having been aware of it or certain of it before, and responding to it by adopting the conclusion. Reasoning is a special case of responding to reasons, responding by discovering, by realising that C, or that P supports C.22
Anticipating an objection: realizing is not the same as coming to believe. It refers to an experience: in extreme cases we refer to it as the Eureka moment. Reasoning is not a recitation of the argument that supports its conclusion. It is an activity that leads to a realisation that the conclusion is a well-supported answer to the question one is considering, and this is confirmed by our implicit knowledge of what reasoning is. Given that what one is led to when realizing something is a belief, it follows that the conclusion of reasoning is a belief.
Therefore, reasoning presupposes the possibility of believing the premises without believing the conclusion. If that possibility does not exist reasoning to that conclusion is not possible, and of course not necessary. And the possibility has to exist for me if I am to be able to reason from those premises to that conclusion. I cannot, for example, reason from the fact that my brother was named Ben to his then having that name. The existence of the required support relationship between items is not sufficient for the possibility of reasoning from one to the other. P&Q entails P, but I cannot reason from P&Q to P.
We had evidence, through generalizable examples, that actions are not conclusions of reasoning. Now we have an explanation and a general argument: when we reach the conclusion of reasoning the transition that takes us to the conclusion is in the nature of a realisation. But there is none in the transition from a belief in an adequate or even conclusive reason for an action to its performance. We can come to realise that circumstances call for an action. But that means realising that there is a case, perhaps a conclusive one, for the action. Not that the action is there. When we act for a perceived reason, there was nothing we did not realise about the connection between that we ought to ϕ and ϕ-ing that led us to ϕ-ing. If we know that we ought to ϕ (or have a case to ϕ) then we know that ϕ-ing is what we ought to do (or have a case for doing). There is, of course, a relation between acting for a reason and believing that one has that reason. To take that to show that the action is a conclusion of reasoning is to reduce reasoning to responding to a perceived reason. It fails for not every action for a reason involves reasoning.
The same argument excludes the possibility of intentions as conclusions of reasoning. If the reasoning of some people established that there is an adequate or a conclusive reason for them to intend, it left nothing (relevant) for them to discover or realise. If they form that intention they do so for a reason, but no realisation leads to that formation, only their prior knowledge that they have reason to intend.
So understood reasoning is an internal process in which our thoughts adjust our thoughts.23 Reasoning relies on, but is distinct from processes or activities in which our thoughts are formed by our interaction with the world, or which set us to interact with, to impact on, the world. For example, recognition and memory are not reasoning: seeing Martin in a crowd and recognising him; recalling that I was at this restaurant where I am now yesterday; remembering that the alarm going off is a reminder that I should get up are examples of thoughts or beliefs formed by being impacted upon by the world. But while these beliefs may trigger reasoning, or just feature as a premise in some reasoning, they are not the conclusions of reasoning. Similarly doing something upon coming to the view that one should is a matter of setting oneself to affect the world or actually doing so, a reaction that may be justified by reasoning but is not itself part of that reasoning.
Intentions belong with actions and not with beliefs; both involve the will. Embedded intentions, e.g. the intention that makes my drinking a cup of coffee or running, intentional, are aspects of actions whose existence is inseparable from the actions they make intentional. If actions cannot be the conclusions of reasoning it would be surprising if embedded intentions could be. Independent, or future directed intentions, can have separate existence, and the argument here offered, as well as other of their features, suggests a need for an explanation of the way they are close to actions, a topic for another occasion.
Reasoning is the handmaiden of normativity. In as much as features of the world make certain responses, emotional, cognitive or active, appropriate, where we have the capacity to respond to them through the use of rational powers, they belong to the normative domain. Reasoning is the reason-guided mental activity of finding out how we should orient ourselves towards the world. Practical reasoning consists of those reasoning activities that aim to determine how we or others should act in the world. The acting, including the intentions with which it is done, is not part of the reasoning, but is determined by it, at least when we react rationally.

1 I am grateful for comments by John Broome, Ram Neta, Ulrike Heuer and participants in M. Alvarez & C. Littlejohn’s seminar ‘Practical Reasoning’.
2 I discuss normative reasons and people thinking about normative reasons. Many are moved by normative considerations and reason about their implications without using or even having the concept of a reason. They use other concepts (duty, ought, what is virtuous, desirable, advantageous, prudent, moral, profitable, divinely commanded etc. These differ from “reasons”, but all imply that there is a reason, or even that the fact that they apply is a reason. When I say that a person relies on, believes in, or concludes that there is a reason I mean that they have a belief that entails that there is a reason.
3 In such discussions we struggle to clarify our meaning, because the words we use or could use have more than one meaning, or a more general meaning, so that only their use in some contexts is relevant to the exploration of normativity. Unless otherwise indicated, I use ‘reasons’ to mean normative reasons, namely features of the world that, given their context, make certain responses appropriate for certain people. The reasoning I explore is neither system 1 nor system 2 thinking as explained by D. Kahneman: THINKING, FAST AND SLOW (Macmillan 2011) 20- 29, probably overlapping system 2.
4 See FROM NORMATIVITY TO RESPONSIBILITY (OUP 2011), especially chapter 5.
5 In part this view is shared (on different grounds) by P. Hieronimy (‘The Use of Reasons in Thought (and the use of earmarks in arguments)’, ETHICS 124 (2013) 114; ‘Reflection and Responsibility’, PHILOSOPHY AND PUBLIC AFFAIRS 42 (2014) 3).
6 Cf. P. Boghossian (‘What is inference’ PHILOSOPHICAL STUDIES 169 (2014) 1) who characterises S infers from p to q as S judges q because S takes the (presumed) truth of p to provide support for q (4). But later he amplifies that inference is an activity with a purpose: ‘it is something we do with an aim – that of figuring out what follows or is supported by other things one believes’ (5). Most if not all reasoning has other aims. But he recognises the basic point: it is an activity undertaken for a reason, though his ‘taking’ seems to be a feature of responding to reasons generally, and is not confined to responding to reasons in reasoning.
7 To avoid tedium I will often assume that the context makes clear whether ‘conclusion’ refers to the conclusion of the argument or of the reasoning. Similarly, ‘argument’ will be used to refer both to the reasons and to the propositions that state them.
8 In other words, the strength varies with the question and with the case for answering it.
9 Two other conclusions to the argument bring it to a close: that the question cannot be answered, or that the answer cannot be known;
10 Thus avoiding questions about the distinction between premises and rules of transformation or inference, the identification of tacit premises and presuppositions, the question of the validation of rules of inference, and much more.
11 Perhaps the imperfection consists in not conforming to an epistemic reason, meaning that if the argument is valid then one has reason either to believe/endorse the premises and believe/endorse the conclusion or to revise at least one of the premises. That reason need not be conclusive. If it is overridden it should not be followed, but that would leave an epistemic reason that has not been conformed with. I am sceptical about this as the explanation of the imperfection. As noted, in general, overridden epistemic reasons – unlike overridden practical reasons – do not leave a remainder that makes the situation imperfect.
12 Reasoning can also change course in midstream, abandon or suspend progress with the original question and take on another.
13 This way of identifying which activities are reasoning conforms to Boghossian’s taking condition: ‘Inferring necessarily involves the thinker taking his premises to support his conclusion and drawing his conclusion because of that fact’ (op.cit. 5), but interpret the ‘taking’ to include a reference to independently sound standards of success, and to be manifested in a myriad of beliefs and dispositions, which together constitute an intentional stance.
14 In his ‘account of the nature of practical reasoning’ Dancy argues for a view similar to the simple account. He writes: ‘is there a suitable normative relation in which both belief and action can stand to the considerations rehearsed in the reasoning that leads to them, and to which they are a response?’ His affirmative reply is ‘when someone deliberates well and then acts accordingly, the action done is the one favoured by the considerations rehearsed in the deliberation, taken as a whole. It is a response to those considerations as together calling for or favouring it. And this is perfectly analogous to theoretical reasoning, when someone forms a belief as the belief most favoured by the considerations adduced as premises’. (‘From Thought to Action’, OXFORD STUDIES IN METAETHICS, Ed. R. Shafer-Landau, 9 (2014) 4). Dancy prefers not to express an opinion as to whether beliefs and actions can be conclusions of reasoning, finding that to be an unhelpful question. Though unless he is willing to contend that reasoning has no conclusions I think that only the characterisations of reasoning and its conclusions that he considered are unhelpful.
15 A reason to ϕ is an adequate reason to ϕ iff it is neither defeated by any conflicting reason nor undercut or cancelled by anything. It may not defeat all the conflicting reasons. In that case it implies that ϕ-ing is free from fault, and also that one has conclusive reason to conform to one of the undefeated conflicting reasons (whereas being permitted or being free from fault is no reason for anything). Some people think that adequate but not conclusive reasons are rare, some suggest that it is a failure in Reason, that it strives but fails to establish a conclusive reason. Some even think that that is my view (see Dancy). I see no case for these views.
16 In the preceding section, when introducing the simple account, it was suggested in its favour that it avoids discriminating between epistemic and other reasons, the suggestion being that on my account we come to believe for reasons always through reasoning to that belief. It is now clear that that is not an implication of my account.
17 One person’s reasoning about what another person is to do is a P1 practical reasoning, but it cannot warrant as a conclusion the forming of an intention by the reasoner that what ought, let us say, to be done is that someone else should take a certain action. Some people see the use of the first person pronoun in the reasoning as essential to its being practical. See A. Müller, ‘Radical Subjectivity: Morality v. Utilitarianism’ RATIO 19 (1977) 115, and J.M. Finnis FUNDAMENTALS OF ETHICS (OUP 1983) 114. Hence my examples will all be of first person reasoning.
18 See U. Heuer ‘Intentions and the Reasons For Which We Act’, PROCEEDINGS OF THE ARISTOTELIAN SOCIETY 114 (2014), 291.
19 Several of the points made in the paper seem to challenge J. Broome’s principle of enkrasia that in its simple form suggests that one is irrational if believing that one ought to ϕ and that one is able to ϕ only by intending to ϕ, one does not intend to ϕ (J. Broome, RATIONALITY THROUGH REASONING, (Wiley 2013), 170-173, 288-290). In fact these points do not conflict with his principle so long as it is confined to cases in which one believes that one has a conclusive reason to ϕ that one can conform to, but only by intending to ϕ (and that one can so intend). What these observations do is illustrate how limited the application of the principle is, and that (partly as a result) it is not the key to the rational connection between belief and action and intention. Clearly, the argument of this paper contradicts Broome’s contention that practical reasoning has intentions as its conclusion. Part of the aim of the paper is to distinguish between conditions of rationality and the rules governing reasoning.
20 Something like this is Dancy’s objection: ‘I can adduce considerations, deliberate, and act accordingly without needing to form an intermediate conclusion that this or that course of action is the one I have most reason to pursue. The notion of a reason need not appear explicitly in my thought, because to respond to something as a reason is not, and does not require, believing it to be a reason.’ Op.Cit. 11. He is right about the notion of a reason not having to appear explicitly, and about people responding to reasons without having the concept of a reason (see fn. 2 above). But these observations do not constitute an objection to the view I express in the text above, and which I defended also in FROM NORMATIVITY TO RESPONSIBILITY in the passages to which Dancy objects. To object he needs to claim that one’s reasoning can validly conclude with an action without an intermediate conclusion that entails that one has a conclusive reason to perform the action. It need not be that belief. It can be a belief that one would be wrong not to perform the action, that one must perform it, or any belief that entails the existence of a conclusive reason.
21 Some people maintain that as inevitably a reason for an action, even a conclusive one, is a reason for any action of a class of actions with some property, that alone shows that actions cannot be the conclusions of reasoning. I express no view on that issue. It is not to be confused with the general case in which the reasons for several incompatible options are undefeated. Dancy, misguidedly, suggests that reasons for belief are also reasons for a believing belonging to a class of possible believings (its content is determined, but the identity of the state, disposition or attitude of believing is not). That is not right because an adequate reason to believe is a reason to believe from the time you should become aware of it. There is no doubt some leeway as to what that time is, but it is not indeterminate in the way that the particular identity of the act that you do when following a reason is underdetermined by the reason.
22 No intuitionistic element is smuggled in here, as the realization does not underwrite its own success.
23 Mental acts, including the acts and activities that constitute reasoning, are not themselves the conclusions of reasoning either. They are guided by beliefs about, say, how the reasoning should continue.