Why reason internalism does not support moral internalism

Why reason internalism does not support moral internalism Chung-Hung Chang Department of Philosophy National Chung Cheng University Abstract Moral internalism and reason internalism are two distinct but closely related theses. It is possible to argue for moral internalism by seeking support from reason internalism. The author repudiates the possibility on the ground that reason internalism itself is implausible. I. Bernard Williams s internalist theory about reasons and the debate between moral internalism and externalism...4 II. A primary objection based on an observation on the relation between reason and motivation...7 III. Responses and rejoinders...9 A. Isn t it irrational not to act on reason requirements?...9 B. Does browbeating uninterested agent matter?...11 IV. Neither desires, nor facts only associated with desires, can provide reason in the ethical realm...14 A. A reason for an action is a fact that justifies the action...15 a. Why the distinction between motivational and justificatory reasons has its bite in the explanation of intentional actions...15 b. Motivational reasons are not psychological and they are not distinct from justificatory reason...18 B. Facts only associated with an agent s desires in her motivational set cannot be her reasons to act in an ethical matter...23 Conclusion...27

Recent debate between moral internalism and externalism begins its life in Thomas Nagel s The Possibility of Altruism, 1 according to which: Internalism is the view that the presence of a motivation for acting morally is guaranteed by the truth of ethical propositions themselves. On this view the motivation must be so tied to the truth, or meaning, of ethical statements that when in a particular case someone is (or perhaps merely believes that he is) morally required to do something, it follows that he has a motivation for doing it. Externalism holds, on the other hand, that the necessary motivation is not supplied by ethical principles and judgments themselves, and that an additional psychological sanction is required to motivate our compliance.(nagel, 1970) Two distinct internalist theses about moral judgement can be distinguished here. 2 Moral Requirement Internalism When in a particular case someone is morally required to do something, it follows that he has a motivation for doing it. Moral Belief Internalism When in a particular case someone believes that he is morally required to do something, it follows that he has a motivation for doing it. Put more precisely, they are: Moral Requirement Internalism * Necessarily, if it is morally right for an agent A to φ in C, he is motivated to φ in C. Moral Belief Internalism * Necessarily, if an agent A believes that it is morally right that he φs in C, then he is motivated to φ in C. 1 This is so because many recent works on the topic are influenced by Thomas Nagel, although it is William Frankena who first identifies the centrality of internalism to debates in metaethics. See Nagel (1970) and Frankena (1958). 2 Here I follow Alfred Mele s terminology. See Mele (1995). Cf. Darwall s distinction between existence and judgement internalism in Darwall (1983) and David Brink s distinction between agent and appraiser internalism in Brink (1989). 2

It should be clear that Moral Internalism that plays a central role in our ethical thinking is not in the form of Moral Requirement Internalism *. For, to be sure, one can be morally required to do something without realising or believing that he is thus required. The agent s ignorance or his mistaken view does not exempt him from his moral duties. In these cases, there is nothing wrong if the agent is not motivated to fulfil what she is required to do. Moral Requirement Internalism * is thus too strong to be true. We will fare better if we take it that Moral Internalism that plays a central role in people s ethical thinking is in the form of Moral Belief Internalism *. In the sections to come, when I mention moral internalism, I mean Moral Belief Internalism *. Moral Belief Internalism *, though, is not a plausible position. There is already abundant literature arguing against it. 3 I would like to make my contribution by arguing that a commonly accepted thesis that seems to support moral internalism is actually implausible. To see this point, let us be clear that Moral Belief Internalism * can be inferred from the following two theses: Moral Belief Rationalism: If an agent A believes that it is morally right that he φs in C, then he has reason to φ in C. And Reason Internalism: If an agent A has reason to φ in C, then he is motivated to φ in C. In contrast, the best version of moral externalism that appeals to me is in the form of Moral Requirement Externalism: For an agent A, even if it is morally right that he φs in C, he may not be motivated to φ in C. It can be inferred from the following two theses: 3 It has been argued by moral externalists that their opponents fail to explain why people with mental defects might believe it is morally right for them φ but, nonetheless, feel no thrust to do so. For discussions, see Stocker (1979), Brink (1986, 1989, 1997), Zangwill (2003), Shafer-Landau (2003, Chapter 6), Svavarsdóttir (1999). 3

Moral Requirement Rationalism: If for an agent A it is morally right that he φs in C, then he has reason to φ in C. And Reason Externalism: Even if an agent A has reason to φ in C, he may not be motivated to φ in C. As I see it, the latter argument for Moral Requirement Externalism has better chance to be a sound argument. However, there has been strong support for Reason Internalism in the debate between internal and external reasons. To prove my claim, I will argue against Bernard Williams 4, a main proponent of internal reasons, and show why we should accept Moral Requirement Rationalism and Reason Externalism and, consequently, reject moral internalism in metaethical debates. I. Bernard Williams s internalist theory about reasons and the debate between moral internalism and externalism Williams states his theory about reasons as follows: A has a reason to φ only if there is a sound deliberative route from A s subjective motivational set (which I label "S," as in the original [1979] article) to A s φ-ing. (Williams, 2001, p. 91) This is a necessary condition for there being a reason. It does not claim that A has a reason to φ whenever he goes through a sound deliberative route from her S to his intention to φ. It only claims that, whenever A has a reason to φ, A must have gone through a sound deliberative route from her S to the intention to φ. Williams consciously leaves aside whether having a sound deliberative route is a sufficient condition for there being a reason. This formulation also implies reason internalism as described above. For, first, a sound deliberative route is meant by Williams to associate with various deliberative 4 For Williams s exposition of his theory, see Williams (1979, 1989/1995, 1995, 2001). 4

capacities a person may use in a decision-making process, such as assessing alternative means for fulfilling an end, scheduling the fulfilment of various ends, weighing ends against each other and determining their relative importance, etc. (Williams, 1979, p. 104) It means that the person in Williams formulation of his internalist theory is an agent. Second, Williams insists that if something can be a reason for action it could be someone s reason for acting on a particular occasion. This means that a reason must be able to figure in the explanation of an agent s action and therefore that anything independent of the agent s motivation for action cannot be his reason. As Williams puts it, if it is true that A has a reason to φ, then it must be possible that he should φ for that reason; and if he does act for that reason, then that reason will be the explanation of his acting... When the reason is the explanation of his action, then of course it will be, in some form, in his S, because certainly and nobody denies this what he actually does has to be explained by his S. (Williams, 1989/1995, p. 39) Williams is fairly permissive about what an agent s subjective motivational set could include. He mentions not only desires, but also dispositions of evaluation, patterns of emotional reaction, personal loyalties, and various projects, as they may be abstractly called, embodying commitments of the agent. (1979, p. 105) The point of his allowing the range of psychological states is that, once an agent deliberates soundly from them to the intention of φ-ing and proceeds to φ, these psychological states can play a role in the explanation of the agent s φ-ing. This brings us to the third and final point: these psychological states contribute to the explanation of an agent s action. It may be that these psychological states are constitutive part of the agent s reason for φ-ing, such that whenever the agent wants to φ, it is because this psychological part of reason causes or gives rise to her intention to φ. Or it may be that these psychological states are closely associated with an agent s reason for φ-ing, such that whenever we want to explain why the agent has reason to φ, we need to cite these psychological states. Whatever is the case, these psychological states contribute to the explanation of an agent s action, simply because they are relevant to the agent s motivation. Therefore, if an agent A, after a sound deliberation, comes to believe that she has reason to φ, she is necessarily motivated by 5

the reason. Williams s internalist theory thus implies Reason Internalism. Consequently, it provides service to the aforementioned argument for moral internalism. Surely, Williams s internalist theory can also be used to refute the argument for moral externalism. It will be used to deny Reason Externalism as follows: (Cf. FitzPatrick, 2004) 1. Suppose there were external reasons. That is, suppose there are reasons independent of an agent s motivational sets. 2. If R is a reason for A to φ, then it must be possible for A to come to be motivated by R to φ through sound deliberation. 3. Since there are external reasons, it must be possible for some consideration R to be a reason for A to φ, even though A could not be motivated by R to φ through any sound deliberation process starting from S. 4. This means that it must be possible for A to come to be motivated by R to φ through other sound deliberative processes that start not from A s existing S. 5. But this is impossible: there are no motivationally efficacious, rational deliberative processes that are not starting from the existing elements of an agent s S. 6. Therefore, the supposition that there are no external reasons is false. Whether Williams s theory is supportive to moral internalism and damaging to moral externalism depends on the truth of his theory. If Williams is right, moral internalism is more likely to be true. If Williams is wrong, moral externalism would still be an open option. I shall argue that the latter, rather than the former, is the case. Williams s theory the thesis that, if R is a reason for A to φ, it must be possible for A to come to be motivated by R to φ through sound deliberation is suspicious. The rationale for this thesis is, presumably, Williams s insistence that, if something can be a reason for action, it must be able to be someone s reason for acting on a particular occasion. Fair enough! But the question is: why should we take this idea as 6

fundamental in our attempts to understand the concept of reason? Why should we not take the following as fundamental: if something can be a reason for action, it must be able to justify the action that is performed on it? Let us consider this question by pondering about a trivial connection between reason and motivation. II. A primary objection based on an observation on the relation between reason and motivation It should be granted at the outset that there is reliable connection between reason and motivation. Even external reason theorists need to recognise that, if R is really a reason for a virtuous agent A to φ, then if A is considering the matter aright, she will come to believe that R is a reason for her to φ, and be motivated accordingly by R to φ. Why? Because we normally take it that a person s virtue involves systematic connections between her reason judgements and her subsequent attitudes and actions. The inclusion of the terms aright and virtuous in this statement guarantees the statement s truth. If R is really a reason for A to φ and A fails to be motivated by R, then either A does not consider things aright, or she is not virtuous; if A satisfies both of the conditions, then she must come, sooner or later, to be motivated to act on the reasons. This connection between a virtuous person s reason and motivation, however, does not imply that for considerations to qualify as genuine reasons they must be capable of motivating virtuous agents. A virtuous person may be ignorant of relevant facts, or he may be misled to consider things in awry ways. Similarly, for considerations to qualify as genuine reasons they do not need to be capable of motivating an agent. An agent, even when he considers things most clearly, may not be interested in doing things that do not contribute to his own welfares. Therefore, an agent, virtuous or not, may have reasons to do things he is not motivated to do. Surely, sound deliberations can help the virtuous come to be motivated by real reasons, if by sound deliberations we mean to refer to some cognitive processes that are capable of unveiling all the relevant facts to the virtuous. By contrast, sound deliberations can get a rational but selfish agent come to be motivated to do what she 7

has reasons, but feel uninterested, to do, only when we use the term to refer to something that can not only unveil all the relevant facts but also turn a rational but selfish agent into a virtuous. Now, there is evidence in Williams s writings that by sound deliberations Williams means to refer to something capable of unveiling all relevant facts. In his discussion of someone wanting to drink a gin and tonic but falsely taking a glass of petrol as gin, Williams claims that the agent does not have reason to mix the stuff with tonic and drink, despite the agent thinks he has reason to do so. Williams denies the agent s reason judgement on the ground that the agent does not deliberate correctly. (Williams, 1979, pp. 102-3) To deliberate soundly, it seems then, one needs to be able to get rid of false beliefs. Moreover, there is also evidence in Williams s writings suggesting that by sound deliberations Williams means to refer to something that can unveil an agent s hidden desires. Williams remarks that an agent s false belief about what he has reason to do may be due to his ignorance of some desires in his motivational set. (Ibid., p. 103) This means that an agent may be able to re-discover his desires by going through a sound deliberation. Nonetheless, none of these observations suggest that Williams take it that sound deliberations are able to convert a rational but selfish agent into a virtuous one. A rational but selfish agent is someone who cares his own interests much more than what are morally right. That is to say, he does not have much de re desires to do the morally right. He may have de dicto desires to be moral, though. However, his de dicto desires to be moral are so weak that, when his interests runs in conflicts with those de dicto desires, he immediately gives up his desires to be a morally respectable person and opts for what can best satisfy his personal interests. In contrast, a virtuous agent is not only someone who has de re concerns about what are morally right, he is also someone who has strong de dicto desires to be moral. Indeed, his de dicto desires to be moral are so strong that, when moral requirements run in conflicts with his own concerns or interests, he is always readily prepared to do what is morally right. Sound deliberations that can help an agent find his hidden desires and/or all of the facts relevant to a state of affairs are not deliberations that can strengthen an agent s weak de dicto desires to be strong enough to the extent that he would give up his desires in 8

the conflict with moral requirements. Sound deliberations in the sense as intended by Williams cannot convert a rational but selfish agent into a virtuous. In view of this observation on the difference between a virtuous and a rational but selfish agent, I believe that we should reject Williams s claim that for some consideration to be a reason for an agent to act on it, there must be a sound deliberative route from his subjective motivational set to his φ-ing. Indeed, by triggering a virtuous person s psychological readiness for fulfilling his moral obligations, a revealed moral requirement may be able to motivate him and give him a reason to move. But it is unlikely that there is any psychological mechanism of a rational but selfish agent that can be triggered by a moral requirement in conflict with his interests, to the effect that he would be motivated to act accordingly. Still, we tend to think that a rational but selfish agent has reason to do what he is morally required to do, in spite of the fact that these requirements may conflict with his interests. We take it that his desires in his subjective motivational set do not justify his denial of the reason judgement. Indeed, our judgement that a rational but selfish person has reason to be more considerate to others is well-grounded. 5 And even the selfish person may agree that he has reason to do what is morally required. It is just that he probably would respond to those who remind him of his moral obligations in a provoking tone: So what? He agrees that he has reason to be more considerate, but he does not have the desire to do so. That is what embarrasses him when prompted. It seems then Williams s theory of internal reasons is wrong. III. Responses and rejoinders A. Isn t it irrational not to act on reason requirements? Critics may find fault in my observation on the relation between reason and motivation. It may be said that irrationality is a matter of conflicts between an agent s reason judgments and her intentions to act. If an agent judges that there is a reason for him to do something, say, he is morally required to φ, but does not come to be 5 For further discussions of these claims, see 14ff. 9

motivated to φ, he is irrational in the first place. There is no such an agent who is rational but so selfish that he is not motivated at all to act on the reason judgements he makes. Therefore, my charge that an agent may know correctly what he has reason to do but feel not interested at all to make a move is in itself inconceivable. It is a necessary connection between an agent s reason judgements and his motivations. However, this response is too hollow to accept. It is granted that people do use rationality and irrationality in the way as described by the response. However, this by no means implies that the use of the term is primary. Nor does this imply that the use is beyond doubt. Think of those who act contrary to their reason judgements say, agents who fail to stop drinking while knowing clearly that they should not drink anymore because otherwise they would be too drunk to fulfil an obligation they are required to do. 6 Their failing to stop drinking may be due to their recklessness, weakness of the will, or compulsion to have another round. They are irrational, according to the response, because they all fail to form an intention to stop drinking. However, calling all of them irrational does not seem very appropriate. At least, it does not faithfully represent the variety of our reactions to them: we usually blame more the reckless than the weak, and we do not usually blame the compulsive. We blame the reckless because they disregard their reason judgements and indulge themselves in having another drink. We blame the weak because we expect them to exercise more of self-control. We do not blame the compulsive because they cannot exercise self-control and thus are compelled to have another go. Thus, calling all of them irrational obfuscates the difference between our various reactions to those who fail to form an intention to act on their reason judgements. Calling all of them irrational also conceal the fact that we may still regard some of them as rational and responsible. Therefore, I believe that we should refrain from doing so. The first response does not really succeed. 6 This example is discussed by Gary Watson (1977). 10

B. Does browbeating uninterested agent matter? A more substantial response comes from Williams himself. Williams considers an example of a man who should be nicer to his wife but cannot be less concerned. Williams thinks that we can say of the man as ungrateful, inconsiderate, hard, sexist, nasty, selfish, brutal and many other unfavourable assessments, and we can say it would be better if he were nicer to wife, but we cannot say the man has reason to be nicer to his wife. Williams asks: The question is: what is the difference supposed to be between saying that the agent has a reason to act more considerately, and saying one of the many other things we can say to people whose behaviour does not accord with what we think it should be? As, for instance, that it would be better if they acted otherwise. (Williams, 1989/1995, pp. 39-40) Williams s point is this. There is a gap between blaming someone as morally wrong and ascribing him a reason to behave morally. Williams believes that if an agent has appropriate motivation in their subjective motivational set but do not carry out the motivation in the right way on particular occasions, it is fair enough to blame him. He is, and should be, blamed. For he fails to act on the reason that arises from motivations in his subjective motivational set. However, if an agent does not in his subjective motivational set have any appropriate motivations no de re concerns about his wife, nor de dicto desires to be moral, nor other relevant dispositions or desires (such as disposition to avoid others blame, or desire to be respected by people, etc.) blaming the agent for not acting on reasons whose existence the agent denies, according to Williams, is tantamount to browbeating him. And, unfortunately, browbeating an agent does not change the fact that he has no reason to perform the act he denies he reason to do. If we invite Williams to extend this point to the example of a rational but selfish agent, Williams would disagree with our intuition that a rational but selfish agent has reason to be more considerate to others, just like he disagrees that the man in his example has reason to be nicer to the wife. Those who think a rational but selfish agent has reason to be more considerate and blame him for not acting on the reason are, Williams would say, simply browbeating the agent. And, unfortunately, browbeating the agent does not give him any reason to be more considerate to others. 11

We can rephrase the objection as follows: 1. Ascribing to an agent a reason to φ while the agent can legitimately deny the ascription is browbeating the agent. 2. An agent can legitimately deny the ascription of a reason to φ if he finds that no sound deliberative route departing from her subjective motivational set could lead to the conclusion that he has reason to φ. 3. A rational but selfish agent may find it that no sound deliberations departing from her self-interested set of minds can lead to the conclusion that she has reason to be more considerate to others. 4. Therefore, a rational but selfish agent can legitimately deny the ascription of a reason for him to be more considerate to others. 5. Therefore, anyone who judges of a rational but selfish agent that he has reason to be more considerate to others is browbeating the agent. It should be noted that the soundness of this objection relies mainly on premise 2, which is actually an application of Williams s internal reason theory. As Williams s internal reason theory is targeted by our primary opposition, this argument cannot be a proper response to our opposition. Perhaps we should see this argument not as something to defeat our opposition, but as something to prompt us to see why our intuition about a rational but selfish agent is not appealing to friends of Williams. Having the possibility of browbeating people, the argument aims to show, our intuition cannot be beyond doubts. The best way to reply to this worry is to show why passing the judgement that a rational but selfish agent has reason to be more considerate to others does not browbeat him. We can argue, pace Williams, (a) that the agent s subjective motivational set does not justify his denial of the reason judgement we pass on him, and (b) that our request of him being more considerate is well-grounded. If this can be done successfully I shall argue this is the case in next section we will be able to pass a justified judgement on the rational but selfish agent without browbeating her. We will then be able to claim that he has a reason to be more considerate to others. Yet, we need to understand that the best reply is not the only reply. Even if what I 12

shall say in next section is too controversial to be true, the possibility of browbeating someone should not tempt us away from our intuitions about a rational but selfish agent. To see this, consider an example originally proposed by Thomas Scanlon. (Scanlon, 1998b) O Brien is someone who sets great value on being a gracious host but, because of his lack of social sensitivity and other cognitive and affective abilities, he does not have very good judgment about what this involves. As a result, O Brien sometimes behaves ridiculously, but he does not have even the faintest idea of this fact. Moreover, given deficiencies O Brien has, if we inform him of the fact, not only that he cannot appreciate the fact appropriately, he will also feel that he is browbeaten. He simply does not have the abilities required to go through sound deliberations that would lead him to want a change in his behaviour. What should we say of O Brien? Surely, we should say that O Brien has reason to change his behaviour. Even reason internalists have to agree with this intuition. After all, O Brien has desires to be a gracious host in his motivational set. If he were able to deliberate soundly i.e. deliberate in a way that can unveil his hidden desires as well as other relevant facts, without being constrained by his various deficiencies he would see that he has reason to change his behaviour. If so, reason internalists have to admit that the fact that O Brien feels browbeaten by our judgement that he has reason to change his behaviour does not change the fact that he has reason to change his behaviour. Thus, we have a counterexample to respond to Williams s complain. A rational but selfish agent might indeed feel browbeaten by our judgement that he has reason to be more considerate to others. However, his feeling so does not change the fact that she has reason to be more considerate. Williams s complain simply does not hold. Faced with this difficulty, Williams tries to defuse the problem in his 2001 article. There Williams distinguishes the judgements of the form A has reason to φ from the judgements of the form the reason for A to φ applies. (pp. 95-6) As I understand it, A has reason to φ means that if A deliberates as good as he can from his existing motivational set, he would recognise some considerations as reasons to φ. In contrast, the reason for A to φ applies means that if A or other agents deliberate competently from A s subjective motivational set, it would turn out that some considerations are 13

reasons for A to φ. Williams s intention is that, although the reason for O Brien to change his behaviour applies, he does not have reason to change his behaviour. Thus, our intuition about O Brien misfires. However, the distinction between A has reason to φ and the reason for A to φ applies seems to have led Williams to retreat to a place where he previously denies. Initially, Williams claims that [i]t is important that even on the internalist view a statement of the form A has reason to φ has normative force. (Williams, 1989/1995, p. 36) He explicates this point by saying this: A has a reason to φ means more than A is presently disposed to φ. One reason why it must do so is that it plays an important part in discussions about what people should become disposed to do. One example of this, which is uncontentiously related to questions raised by the internalist view, is given by advice in the if I were you... mode. Taking other people s perspective on a situation, we hope to be able to point out that they have reason to do things they did not think they had reason to do, or, perhaps, less reason to do certain things than they thought they had. (Ibid) However, his new definition of A has reason to φ in the discussion of O Brien has given up the normative force completely. It means nothing more than A would be disposed to φ if he engages in a deliberation. No other people s perspective is included in the new definition. Consequently, I do not think the new definition represents an appropriate concept of reason. I will argue in next section, a reason for an action must be a fact that can justify the action, not only to the agent, but also to others. IV. Neither desires, nor facts only associated with desires, can provide reason in the ethical realm I claimed earlier that we can prove (a) that a rational but selfish agent s subjective motivational set does not justify his denial of the judgement that he has reason to be more considerate to others and (b) that our judgement that he has reason to do so well-grounded. Let us now flesh out these claims. 14

A. A reason for an action is a fact that justifies the action The first thing to say is that desires in an agent s motivational set do not constitute any reason for action by default and that a reason for an action is a fact that justifies the action. To see this, we need to see that the commonplace distinction between motivational reason and justificatory reason is not a distinction between two different kinds of reasons. Let us begin our discussion with why the distinction has its bite in the explanation of intentional actions. a. Why the distinction between motivational and justificatory reasons has its bite in the explanation of intentional actions In the opening paragraph of his influential essay, Davidson asks, What is the relation between a reason and an action when the reason explains the action by giving the agent reason for doing what he did? (1963, p. 685) His answer is that the reason rationalizes the action. But what does this mean? Davidson labels as primary reasons those that rationalise an action. He gives a necessary condition for being a primary reason in the spirit of Hume as follows: C1. R is a primary reason why an agent performed the action A under the description d only if R consists of a pro attitude of the agent towards actions with a certain property, and belief of the agent that A, under the description d, has that property. (Ibid, p. 687) According to C1, the primary reason for an action consists in the agent s pro attitude or, in a broad sense of the commonplace term, desire and her relevant belief. A primary reason rationalises an action just in case the pair of psychological states rationalises the action. There are three possibilities concerning how the pair of psychological states rationalise an action. Either (i) that they explain the action, or (ii) that they justify the action, or (iii) that they explain as well as justify the action. As a prominent proponent of Hume s theory of reason, Davidson does not explicitly state his choice. This leaves room for interpreters to pursue different readings of Hume s theory. Thomas Nagel once criticises the Humean idea that desires can provide reason for action by inviting us to consider prudential reasons one can have now or in the future. If the Humean idea is right, Nagel thinks, prudential reasons must be reasons 15

stemming from one s present desires for the satisfaction of future interests and desires. (Nagel, 1970, p. 36) For example, I know that I may be assailed by a desire to eat Indian cuisine next Wednesday. Despite I do not have such kind of appetite now, I do have desire to satisfy my future desire. Therefore, I have a reason to schedule my diary to fit in a Wednesday feast of Indian meal. This reason is a prudential reason. Nagel asks us to think of a slightly different scenario. Suppose I do not now have desire to satisfy my future desire that would assail me next Wednesday. Granted that my future desire will provide me with reason to look for a good Indian restaurant next Wednesday when the desire takes place, I do not at present have a prudential reason to do so, due to my lack of relevant desires. Therefore, I now have no reason to do what I will have reason to do. This, Nagel thinks, is an unacceptable result. For it violates a constraint of the concept of reason, the Generality of Reasons: [i]f there is a reason to do something on a particular occasion, it must be specifiable in general terms which allow that same reason to be present on different occasions, perhaps as a reason for doing other things. (Ibid, p. 35) Equally unacceptable is the possibility that I may have reason to do what I will have no reason to do. Yet, if we agree with the idea that desires can provide reason for action, the possibility seems inevitable. Consider a case in which I, under strong emotion against those who make fun at a working class s vulgar talkings, feel a strong desire to beat them up. I know that I will feel no such desire later on after cooling down. However, on the Humean story, it seems that, at the moment I feel the desire, I have a reason to do what I shall not have reason to do, namely to beat up those who discriminate against the working class. The Humean story is therefore deemed to be a failure. It is obvious that the force of Nagel s objection rests on the thesis of the generality of reasons. As I see it, the thesis actually grows out of two other theses. The first thesis is: The Generality of Justification If one is justified to do something in a particular situation, then he is justified to do the same thing in all like situations. The second thesis is: 16

The Justifiability of Reason If one is justified to do something in a particular situation, then he has reason to do the thing in the situation. It is clear that the conjunction of both theses entails the truth of the Generality of Reasons. Therefore, I take it that a reason for φ is something that justifies φ-ing. In terms of Davidson s terminology, I take it that that a reason rationalises an action just in case it justifies, rather than explains, the action. Michael Smith understands so well that the force of Nagel s objection comes from the conjunction of the Generality of Justification and the Justifiability of Reason. When he attempts to contribute a response to Nagel, he chooses not to challenge the truth of the Generality of Justification, nor that of the Justifiability of Reason. Instead, he chooses to accuse Nagel of misreading Davidson s point. According to Smith, Davidson s formulation of Hume s theory is concerned with how an action is explained by our agency. In other words, he is concerned with how an action is motivated and consequently carried out. It is then not surprising that Davidson s concept of primary reasons focus only on the agent s psychology. From the beginning, no efforts have been made to capture primary reasons justificatory force. Therefore, it is unfair to charge Davidson of overlooking the justificatory role a reason might entertain. 7 To put it in more details, Smith s response is this. If we distinguish carefully motivational reasons from justificatory reasons, and if we take it that Davidson is only concerned with explanation of an action, we can see that Nagel s objection is not actually on target. For although it is not acceptable to say that I have now no justificatory reason to do what I will have justificatory reason to do, nor that I have justificatory reason to do what I will have no justificatory reason to do, it does make 7 In a trivial sense, every reason justifies. One who mistakes a bottle of petrol for a mix of gin and tonic may feel motivated and thus, in Smith s view, has a motivational reason to drink the bottle to quench his thirst. This motivational reason justifies the agent s drinking the bottle of petrol in a subjective sense: the agent believes that he would be warranted to quench his thirst if he drinks the bottle. However, this subjective sense of justifiability is too trivial to play a role in our analysis of reasons for action. 17

sense to say that I have now no motivational reason to do what I will have motivational reason to do, or that I have motivational reason to do what I will have no motivational reason to do. For example, it is perfectly all right that I have no desires and therefore no motivational reasons now to look for a good Indian restaurant, in spite of the fact that I will have such a desire and will therefore have such a motivational reason next Wednesday to look for a good Indian restaurant. It is also perfectly all right that I have now a desire and therefore a motivational reason to beat up people who are making fun at working class, in spite of the fact that I may have no such desires and therefore no such motivational reasons to do so. The Generality of Reason holds only in the case of justificatory reasons, but it does not hold in the case of motivational reasons. If we accept the distinction between motivational and justificatory reasons, and if we accept Smith s reading of Davidson s theory, we should agree with his criticism on Nagel: he conflate[s] the claim that an agent has a motivating reason to φ with the claim that she has a [justificatory] reason from the perspective of rationality to φ. The Humean is making only the first claim, not the second. (Smith, 1994, p. 100) The distinction between motivational reasons and justificatory reasons, thus, has its bite on solving problems arising from the Generality of Reasons. b. Motivational reasons are not psychological and they are not distinct from justificatory reason The problem with the seemingly perfect story provided by Smith begins with this question: does motivational reason provided by desires really deserves to be called reason? If, as Davidson claims, a motivational reason is a combination of desires and beliefs and if, as Smith claims, motivational reasons are something entirely different in kind from justificatory reasons, I believe we have to say: this kind of purely motivational reason does not deserve to be called reasons. To see this, consider a fair lady who knows very well that she does not have reason to buy a fashion handbag displayed in the windows of Harrods but feels strongly disposed to buy it. She knows she has no reason to buy it, because she knows that she already has many similar handbags, that she cannot afford the cost, that she 18

actually likes the badge rather than the bag, that it is of no practical use, etc. Suppose, in spite of her strong disposition, she turns away from the bag, suffering not much from after-effect disappointment. Should we say she has reason to buy the bag? It seems that she has no reason to buy the bag. After all, it is a fact that she believes she has no reason to buy it and she suffers not much from her decision. Thus, it seems that the lady s earlier disposition to buy the bag, in conjunction with the relevant belief that if she buy the bag she can fulfil her desire, does not constitute a reason for her to buy the bag. Unfortunately, Smith s story cannot accommodate this simple intuition. Under the clear-cut distinction between motivational reasons and justificatory reasons, friends of Smith would have to complicate the example and say implausible things like this. Although the lady has motivational reason to buy the bag, she has justificatory reason not to buy it. On balance of her motivational and justificatory reasons, the lady decides not to buy the bag. The lady makes this decision only because her justificatory reason has outweighed her motivational reason. She turns away from the bag only to act on the balanced-out reason. As I see it, this complication is implausible for two reasons. First, it does not fit the lady s psychology. The lady thinks that she does not have reason to buy the bag. She does not that her justificatory reason not to buy the bag outweighs her motivational reason to buy the bag. Second, there seems to be no point in calling the lady s earlier disposition to buy the bag a reason. By calling something a reason for an action, we mean that the thing justifies the action and/or explains why the thing is done. Yet, the lady s disposition does not justify nor explain her not buying the bag. It is better to call the disposition a desire rather than a reason to buy the bag. This example gives us a counter-example to Smith s idea that there are motivational reasons provided by desires distinct in kind from justificatory reasons. This example should remain forceful even if we imagine the lady suffers serious or long-term after-effect regrets from her decision and then goes back to the shop to buy the bag. For if the lady feels enormous regrets, and if she is not irrational, then her regrets must be because she discovers now that she actually has reason to buy the 19

bag. 8 For example, she may realise that she did not understand how much she liked the bag when she made the decision, that she overestimated her abilities to cope with the disappointments resulting from the decision, and so on. Still, the lady s previous undisclosed desire, even in conjunction with correct beliefs about herself, does not provide motivational reason distinctive in kind from justificatory reason for her to go back to Harrods and buy the bag. What provides the lady reason to buy the bag is the fact that the bag fulfils her desires and makes her happy, rather than her feelings about, and desires for, the bag. More importantly, the fact that the bag fulfils the lady s desires and maker her happy not only motivates but also justifies her to buy the bag. It is not a motivational reason distinct in kind from justificatory reason. One might ask: if the fact that the bag fulfils the lady s desires and makes her happy is associated with the lady s desire, why not take the lady s desires as her reason to act? Isn t it true that what will fulfil the lady s desire is specified or determined by her desire? Why not take the final determining factor, rather than the fact determined by the final factor, as the reason for action? Two responses are in place. First, to say that the lady s desires identify, specify, or determine the fact that the bag is valuable to her is not to say that the lady s desires give rise to the bag s value. The bag must have certain features that can be identified by the lady s desires as valuable. Therefore, it is at least equally plausible to regard the fact that the bag has these features as the determining factor and hence the reason for the lady to buy it. Second, whether it is the lady s desire, or the fact that the bag has certain features specified by the lady s desires as valuable, that should be regarded as her reason for buying the bag, does not depend on which of them is the determining factor. It depends on which of them can let us have a consistent view about reasons. Our earlier discussion of the lady has suggested that there is no point in taking an agent s desire to be reason. Our discussion has also suggested that it is because the lady does not find valuable any features of the bag earlier, so she does not find any reason to buy the bag. To be consistent with these claims, we should take it that the fact that the bag has certain features that can be identified as valuable by the lady s desires, rather than the lady s desires per se, is the reason for her to buy the bag. Our 8 If the lady still thinks she has no reason to buy one, but cannot help the feeling of great regrets, she is irrational. 20

claim that there is no motivational reasons distinctive in kind from justificatory reasons remains intact. Understandably, friends of Smith may not be convinced. It is easy for them to find an example to cast doubts on the view I am proposing. Think of someone who wants to purchase an original Picasso. Suppose after sound deliberations she buys a painting in front of her. She believes that it is a genuine Picasso, but in fact it is a fake. 9 What should we say of the agent s action? I believe we should say this: the agent does not actually have reason to buy the painting, because the painting in front of her does not really have features a genuine Picasso should have. However, friends of Smith would say: granted so, we can still ask a further question, namely, for what reason the agent sets out to buy the fake Picasso? This question is important, they would say, because an answer to this question is essential to our attempts to understand why the agent performs the action she performs. Saying that the act is performed out of no reason, just as we did earlier, does not help at all. The intention of Smith s friends is surely to press us to agree with the distinction between motivational and justificatory reasons. However, I do not think that their argument is convincing. Here is a rephrase of their argument: 1. To understand why the agent buys the (fake) painting, we need to answer the question: for what reason she buys the painting. 2. The reason for the agent to buy the painting is not the fact that the painting is a Picasso, because there is no such a fact. 3. The only reason for the agent to buy the painting is the combination of her desire to buy a Picasso and her (false) belief that the painting in front of her is a Picasso. 4. The combination of the agent s psychological states does not justify the agent s buying the painting. 5. Therefore, the combination of the agent s psychological states is her 9 This example is given by Smith (1994). For the sake of discussion, I adopt it with necessary modifications. 21

motivational, rather than justificatory, reason to buy the painting. The problem with this argument lies in its first premise. It is not true that in order to understand why the agent buys the fake painting, we need to know her reason for buying it. In fact, knowing her motivation to buy the painting is good enough to give us an explanation of her action. We can say, The agent desires to buy a Picasso, but she does not know that the painting is fake. Therefore, although she has motivation to buy the painting, she does not actually have reason to buy it. With the distinction between motivation and reason made, this explanation perfectly explains why the agent buy the fake painting. Friends of Smith want to press us to agree that the agent s motivation constitutes her motivational reason to φ. But why should we agree with this? I have been arguing, first, that an agent s motivation to φ does not necessarily justify her to φ and, second, what deserves to be called a reason must be some fact that can justify the act at issue. I conclude that a reason for action has to be justificatory in nature. If a reason is also motivational, it must be because the reason happens to motivate the agent to act. It cannot be motivational without being justificatory. I find no rationale to agree with Smith that an agent s motivation constitute her motivational reason to act. Smith s friends need to convince me that there is indeed such a rationale. Otherwise, I would take it that their distinction between motivational and justificatory reasons fails. As a matter of fact, I suspect that the whole argument given above is led astray by the formality of how the first premise is presented. To see this, we should notice that asking For what reason the agent buys the fake painting? is not the only way to set off to understand the agent s action. We can also set off for the journey by asking Why the agent buys the fake painting? The difference between the two can be minute, especially when they are used loosely in daily life. However, if we are to be strict in using our language, there is an important difference between the two: while the former question demands a reason to be the answer to the question, the latter does not. As I see it, this difference translates into this contrast: while the former asks a specific question about what fact could justify the agent s buying the fake painting, the latter question ask a more general question it only demands a loose explanation of how the action is brought about. We can answer this question by providing a reason-explanation. Alternatively, we can answer this question by providing a 22

motivation-explanation. In this sense, the why-question is broader than the for-what-reason-question. In sum, if we take the difference between the two questions seriously, we should modify the first premise of the argument as follows: 1 * To understand why the agent buys the (fake) painting, we need to ask why the agent buys the painting. Then, it becomes clear that the combination of the agent s desire to buy a Picasso and her belief that the faking painting is a genuine one is only an answer to the why-question. It is not an answer to the for-what-reason-question. Without distinguishing carefully the two different questions, Smith s friends are led astray to take it that the combination of the agent s psychological states is an answer to the for-what-reason-question and, consequently, the combination qualifies to be a reason. With the distinction carefully made, I believe we should conclude that the argument is an unsound argument. An agent s psychological states cannot be her motivational reason to act, because an agent s psychological states can never justify one s action. B. Facts only associated with an agent s desires in her motivational set cannot be her reasons to act in an ethical matter So far I have argued that the combination of an agent s psychological states, her desires and her beliefs, does not constitute any reason for her to act upon. A reason for an action is a fact that justifies the action. If a reason can be called a motivational reason for an action, it is because the reason not only justifies but also happens to motivate the action. These points are enough to lead us to reject an interpretation of Williams s theory of internal reasons. According to the interpretation, desires in an agent s motivational set constitute part of the agent s reason for φ-ing, such that whenever the agent wants to φ after sound deliberations, this psychological part of reason causes or gives rise to her intention to φ. For, as I have claimed, the psychological part can never be a reason. However, these points are not enough to reject a different interpretation of Williams s theory. According to the interpretation, desires in an agent s motivational set do not constitute part of the agent s reason for φ-ing, but they are very closely associated with the agent s reason for φ-ing, so closely indeed that whenever we want 23