Front. Philos. China 2011, 6(3): DOI 10.

1007/s11466-011-0144-4

RESEARCH ARTICLE

John HYMAN

Acting for Reasons: Reply to Dancy

Abstract This paper argues that we need to distinguish between two different ideas of a reason: first, the idea of a premise or assumption, from which a persons action or deliberation can proceed; second, the idea of a fact by which a person can be guided, when he modifies his thought or behaviour in some way. It argues further that if we have the first idea in mind, one can act for the reason that p regardless of whether it is the case that p, and regardless of whether one believes that p; but if we have the second idea in mind, one cannot act for the reason that p unless one knows that p. The last part of the paper briefly indicates how the second idea of a reason can contribute to a larger argument, showing that it is better to conceive of knowledge as a kind of ability than as a kind of belief. Keywords reason, action, knowledge, explanation 1 In Acting in Ignorance, Jonathan Dancy argues that some explanations are factive while others are not. He also says that knowledge is factive and that seeing and proving are factive. The term factive was originally applied to verbs, such as know. Some philosophers also describe predicates, such as x knows that grass is green, and mental states, such as knowledge, as factive. When seeing, proving and explanations are added to the list, it becomes difficult to understand what kind of property factivity can be. Let us say that a sentence-forming operator O on one or more declarative sentences is factive if, and only if, the statement Os1 sn cannot be true unless the statements s1 sn are true. If this is what factivity is understood to be, it is not just a brute fact about language. Dancy says, Knowledge is factive, because he knows that it is raining but it is not is uninterpretable. But this gets things the wrong way round.
Received October 12, 2010 John HYMAN () The Queens College, Oxford University, Oxford OX1 4AW, UK E-mail: [email protected]

John HYMAN

He knows that it is raining but it is not is uninterpretableor cannot be truebecause he knows that is factive; he knows that is factive because one cannot falsely or erroneously know that something is the case; and one cannot falsely or erroneously know that something is the case because knowing that something is the case is a relation between knowers and facts or truths. What about explanations? The idea that explains why is not factive is absurd. If p explains why q, it is the case that q because it is the case that p; in short, q because p. (Dancy acknowledges that because is factive.) Equally, if p explains why not-q, it is not the case that q because it is the case that p, in short, not-q because p; if not-p explains why q, it is the case that q because it is not the case that p, in short, q because not-p; and if not-p explains why not-q, it is not the case that q because it is not the case that p, in short, not-q because not p. Why is explains why factive? Because explanation is a relation between facts or truthsthe relation of making intelligible or making understood. Non-facts or non-truthscall them fictionscannot make facts or truths intelligible or understood. For example, if it is a fiction that Tom lost his phone, that Tom lost his phone cannot make the fact that he failed to call Sam intelligible or understood, although it can seem to do so, if it seems to be a fact. Linguistic intuitions, our inclinations to interpret a sentence in one way or another, are a poor guide to factivity, because given the right context, or the right priming, we can interpret the same sentence in different ways. In order to decide which phrases are factive and which are not, we need to think about the kinds of activity, mental act, etc., they are about. Why does Dancy claim that some explanations, and in particular some explanations that identify an agents reason for doing, feeling, hoping, etc. (or not doing, not feeling, not hoping, etc.) are not factive? The answer has to do with what he thinks reasons for doing something are. He says in several places that a persons reason reveals the favourable light in which the action appeared to him, but he also acknowledges that this is not a satisfactory explanation as it stands because it does not apply to a persons reason for feeling something. For example, a reason for feeling sadsay, being bereaveddoes not reveal the favourable light in which being [sad] appears to the agent. But a reason for feeling something is not any different from a reason for doing something. In fact, the same reason may be a reason for both: Being bereaved may be a reason for feeling sad, and also a reason for cancelling a trip. What Dancy offers as an alternative at that point is that a reason is a consideration that, for the agent, renders [feeling something] an appropriate response. I do not intend to claim ownership of the term reason, and if this is what we agree to call a persons reason, then of course Dancy is right to deny that a reason must be true. But this does not show that one kind of explanation is not factive. Suppose (taking one of Dancys own examples) my reason for going to

Acting for Reasons: Reply to Dancy

the stationin Dancys sense of the term reasonwas that my daughter was arriving on a train. What explains my going to the station was not that my daughter was arriving on the train. What explains my going to the station was that this was a consideration that made my going to the station an appropriate thing to do, from my point of view. Factivity is preserved. Dancys mistake is the result of combining three ideas: (i) that giving (stating, identifying) a persons reason can explain his action; (ii) that a persons reason is a consideration that makes an action appropriate, from that persons point of view; (iii) that a consideration that makes an action appropriate, from that persons point of view, need not be true. These claims are all true, if we adopt Dancys conception of a reason, and if they are interpreted with sufficient care. Dancys mistake is to assume that when we explain a persons action by giving his reason, in his preferred sense of the term reason, the explanans is the reason itself. The truth is that what Dancy calls the reasonthe consideration that made the action appropriate, from the agents point of viewcan be false. But if we explain a persons action by giving the agents reasonin this sense of reasonthe reason is not the explanans. The explanans is what is stated before the phrase explains why or after the word because. And this is not that my daughter was arriving on the train; it is that my daughters arriving on the train was a consideration that made going to the station appropriate, from my point of view. I shall therefore set aside the claim that some explanations are not factive, and focus on the idea of a reason. At this stage, I shall make two comments about Dancys conception of a reason, both of which will be substantiated in due course. First, a consideration that makes an action appropriate, from the agents point of view, need not be true. But it need not be something the agent believes to be true either. It is helpful to call it a premise, because it is clear that we can act on premises we do not believe. For example, I can act on the premise that a student felt too ill to come to a lecturein other words, I can assume or accept that the student felt too illeven if I do not believe this to be true. Therefore if we adopt this conception of a reason, we should not claim that I need to believe my daughter is arriving on a train in order to go to the station for that reason. I may believe she is dreadfully unreliable, but feel obliged to assume that she is arriving on the train, just in case she is. Second, as I shall argue in more detail in due course, we sometimes use the term reason to express a different idea: the idea of a persons being guided by a fact. This is a metaphor, of course, but it is a perfectly familiar one to English-speakers, and it is not difficult to explain. If someone is said to have been guided by a certain fact, this means that he took it into consideration, when he modified his thought or behavior in some way, or decided what to think or

John HYMAN

what to do. This, I suggest, is how we should understand the idea of a fact being a persons reason. Furthermore, if we understand the idea of a fact being a persons reason in this way, then the explanans can be the persons reason. For example, if my daughter was in fact arriving on the train, and if the fact that she was arriving on the train was my reason for going to the station, then the fact that she was arriving on the train explains why I went to the station. Therefore, I went to the station because she was arriving on the train. Perhaps it was a faint awareness of this form of explanation that led to Dancys mistake about factivity. Perhaps his assumption that the agents reason is the explananswhich on his conception of a reason it is notwas an unwitting acknowledgement of this other conception of a reason. Be that as it may, I shall argue that Dancy has confused these two closely related but distinct conceptions of a reason: the idea of a fact being a persons reason for his action and the idea of a person acting on a premise or assumption. 2 If we survey the literature in the theory of action, the confusionor to put it neutrally the disagreementabout what reasons are is wider than this. For example, Davidson says that reasons consist of mental states or dispositions; whereas Kenny and Audi both claim that reasons can be goals. (Audi adds that an agents goal is something that he or she wants (Audi 1993, p.15f), such as to die with dignity or to get home before dark.) These claims are inconsistent, and in my view both are wrong. But the mistakes these authors make are understandable, because explanations which mention reasons are often equivalent to ones that mention mental states or goals. For example, Jamess reason for going to church was that it would please his mother and James went to church because it would please his mother mention Jamess reason, whereas James went to church because he wanted to please his mother mentions a mental state, and James went to church to please his mother refers to a goal. For most purposes, the explanations that mention reasons are equivalent to the ones that do not. Even so, reasons are not mental states or goals. To see why, we only need to remind ourselves that a reason is the kind of thing that can occur in a piece of reasoning as a premise. (This does not mean that they are nothing more than premises, as we shall see.) Therefore, if I reason thus: p therefore, q my reason for concluding that q is p. Every reason, without exception, can be a premise, whether it is a reason for doing something, for believing, hoping or wishing something, for loving or hating someone; or for not doing, believing,

Acting for Reasons: Reply to Dancy

hoping or wishing something, or for not loving or hating someone. Mental states and goals cannot be premises, therefore they cannot be reasons either. It is true that some premises describe or identify mental states or goals. For example, James wants to please his mother describes Jamess mental state, and Jamess goal is to please his mother identifies his goal. And the both can evidently be premises: {James wants to please his mother/Jamess goal is to please his mother} therefore, James will probably go to church. But Jamess mental state and his goal themselves cannot be premises. Because premises can be true or false, they can be asserted or denied, and they can stand in logical relations to other truths or falsehoods, whereas mental states and goals cannot do any of these things. Perhaps it is also true that a persons reason for doing or not doing or believing or not believing something etc. must be the content of his mental statefor example, perhaps it must be something he believes, or something that he knows. But if that is so, the mental state is not his reason. His reason is the content of the state. 3 In a few articles published since 1999, I have argued that the fact that p can only be a persons reason for doing something if she knows that p (Hyman 1999; Hyman 2006; Hyman 2010). This is part of a larger argument, in which I try to show that it is better to conceive of knowledge as a kind of ability than as a kind of belief. Knowledge, I argue, is the ability to do things for reasons that are factsthe ability we exercise when we are guided by the facts. In this paper, I mainly focus on the claim that the fact that p can only be a persons reason for doing something if it is a fact she knows. Dancy argues in Acting in Ignorance that what he calls the Knowledge View is false. The Knowledge View, he explains, is the view that one can only run for [the] reason [that the train is leaving] if one knows that the train is leaving. He attributes this claim to me, but in fact this is not the claim I have defended. The claim I have defended, is that a fact can only be a persons reason for doing something if she know that fact. This is consistent with denyingas Dancy doesthat reasons must be truths, because it is consistent with denying that reasons must be facts. The idea is that if a persons reason is a certain fact, for example, if Jamess reason for going to church is the fact that it will please his mother, then James must know that going to church will please his mother. It is not enough for him to believe this; it is not enough for him to believe it truly; it is not enough for him to believe it truly and with a justification; it is not even enough for him to believe it truly and with a justification with no false lemmas; it

John HYMAN

has to be something that he knows. Why cannot the fact that p be a persons reason for doing something unless she knows that p? Why is it not sufficient to believe that p? In Acting in Ignorance, Dancy argues that it is sufficient, and an earlier article defends the same idea: Suppose that I am meeting my daughter off the train. I do not know that she is on the train, and I know that I do not know this, but I do believe that she is on the train. So I head off to meet her. What would I say if asked what my reason is for meeting this train? I might say My daughter is on it. But I would be more likely to say I think my daughter is on it, or My daughter should be on it, or My daughter is probably on it. There is some pressure to avoid offering the bald claim that my daughter is on the train as my reason if I do not know this to be the case. But what if my response to this is to specify my reason thus: I think my daughter is on the train? Are we to say that I have now shifted my ground to something I know? In line with a certain approach to Moores paradox, I think that this remark is not to be understood as autobiographical, but as a more guarded version of the claim that she is on the train. It is not, then, that pressure can be put on me to restrict myself, in giving my reasons for doing what I do, to things that I take myself to know. The pressure is more to express, in the way that I give my reasons, the degree of confidence with which I hold them (Dancy 2008, pp. 262279). This passage is condensed, but the main idea appears to be that if someone says My reason for meeting the train is that I believe my daughter is on it because he does not know that his daughter is on the train, this is not because a fact cannot be a persons reason unless she knows that fact. His reason is that his daughter is on the train, even though he does not know that she is, but he states his reason with a qualification indicating his degree of confidence in its truth. As I acknowledged in the first section of this article, Dancys main idea would be plausibleit would be plausible that something can be a persons reason if he merely believes it, but does not know itif by a persons reason we meant the premise or assumption on which he acted. Because not knowing that p does not prevent me from assuming that p, that is, from taking the proposition that p as a premise. We can reason from false premises, and if a premise is false, it evidently cannot be known. For example, one can reason from the premise that the train from Paris will arrive on platform eight, even if in fact it will arrive on platform one. If we can reason from false premises without knowing them, then we can reason from true premises without knowing them as well. For example, one can reason from the true premise that the train from Paris will arrive on platform one, even if the timetable one consults is out of date. Therefore a person can reason from a premise that he merely believes, but

Acting for Reasons: Reply to Dancy

does not know. One can also, however, reason from a premise, whether true or false, without believing it to be true. This is what one does when one argues by reductio. It is something one does if one accepts the students implausible excuse for failing to attend a lecture, and decides how to deal with the student on this basis. It is even something one may be obliged to do. For example, if a judge directs the jury in some matter of fact, which he can do in the English legal system, the jury is obliged to deliberate accordinglyin other words, it is obliged to assume the thing in question, say, that the goods were stolenregardless of what its members happen to believe. In Acting in Ignorance, Dancy says that one can run for that reason [i.e., that the train is leaving] so long as one believes the train to be leaving even when it is not in fact leaving. But according to his own conception of a reason, he should not insist on the requirement that the agent believes his reason to be true. As we saw earlier, his claim that a reason for doing something specifies the attractions the action had for the agent is unsatisfactory. But if we say instead that a reason for doing something is a consideration that makes it seem appropriate to the agent to do the thing in question, then in the case of the jury, for example, the proposition that the goods were stolen, which the judge directed the jury to accept, can be their reason for returning a guilty verdict, even if they believe it to be false. Thus we do not need to know our premises: we do not need to know what we assume. Dancy is right about that. And we do not need to believe our premises either. This suggests that what we normally understand by a reason is not merely a premise or assumption. For if the normal idea of a reason were simply this and nothing more, it would not be contradictory or absurd to say that my reason for going to the station was that my daughter was arriving on a train, although I did not believe she was arriving on a train, or that I went to the station because my daughter was arriving on a train, although I believed she was not arriving on a train. But these things certainly appear to be contradictory or absurd, although no doubt (as noted in the first section) we can interpret them in ways that soften or cancel this appearance. I shall make two further comments about the passage quoted. First, it is difficult to see how Dancys way of interpreting My reason for going to the station is that I believe my daughter is arriving on a train can apply to the case where I say my reason is that my daughter is probably arriving on a train, or that she should be arriving on a train. He is right in claiming that if I say My reason for going to the station is that I believe my daughter is arriving on a train, I have not shifted my ground to something I know. He is right, because My reason for going to the station is that I believe my daughter is arriving on a train does not mean that I am being guided by a fact about my state of mind. But if someones reason is that it is probable that p, or that it should be the case that p,

John HYMAN

why should we should deny that he is being guided by a fact about what is probable, or should be the case? It is perfectly normal to be guided by facts of this kind. Second, Dancys comment about Moores paradox is unconvincing for two reasons: (i) Moores idea is that p, but I do not believe that p is paradoxical or absurd, even though both conjuncts may be true. Dancy seems to be saying that the paradox arises because saying that one believes or does not believe something is a guarded or qualified assertion or denial of the thing one says one believes or does not believe, rather than a description of ones state of mind. Therefore p, but I do not believe that p combines an unqualified assertion that p with a guarded or qualified denial that p. One could say it is a guarded or qualified contradiction. But why should we choose between the claim that saying one believes something is a guarded or qualified assertion, and the claim that it is a description of ones state of mind? Why can it not be both? Many statements have this kind of dual use. For example, if one says That was a nasty thing to do, one normally describes the thing as nasty, but one also normally expresses disapproval, and if one says p, one normally asserts that p, but one also normally implies that one believes that p. Why should not I believe that p have a similar dual use? (ii) There is a simpler explanation of Moores paradox, which does not transfer to statements giving reasons. The paradox arises from the fact that assertion implies belief: if one asserts that p, one implies that one believes that p. Hence, p, but I do not believe that p is an implicit contradiction. That is why the paradox only occurs in the first person: there is nothing wrong with saying p, but he does not believe that p. By contrast, the absurdity of I am going to the station because my daughter is arriving on a train, but she is not arriving on a train arises from the fact that explanation is factive, therefore it is preserved in the third person: He is going to the station because his daughter is arriving on a train, but she is not arriving on a train. 4 As we have seen, we have the idea of reasoning from a premise, which may be known, or merely believed, or not believed, and which may be true or false. But as well as the idea of reasoning from a premise, we also have the idea of being guided by a fact. As noted earlier, if someone is said to have been guided by a certain fact, this means that he took it into consideration, when he modified his thought or behavior in some way, or decided what to think or what to do. This, I suggest, is how we should understand the idea of a fact being a persons reason.

Acting for Reasons: Reply to Dancy

One reason why the metaphor of guidance is helpful is that it draws attention to the similarity between awareness of facts and awareness of things. Think of a cat stalking a bird. The cats awareness of the bird is expressed in the way it modifies its behavior in response to way the bird modifies its behavior. The bird hops this way, the cat turns this way; the bird flutters across the courtyard, the cat advances a few paces; and so on. The cats movements are responsive to, are guided by, the bird. Or think of a traveller followingbeing guided bya guide. The guide takes the left fork, so the traveller takes the left fork; the guide pauses, so the traveller pauses; and so on. Being guided by facts is comparable. Whether one is guided by facts or by things, ones thought and behavior are responsive to what one is guided by. But as Wittgenstein pointed out, this is not like a train being guided by the rails, because the way knowledge gets expressed depends on ones aims or purposes, and because being guided by the facts is not passive and constraining, any more than reading is, although the readers thought (if he is reading silently) or speech (if he is reading out loud) is guided by the words she sees on the page. A person can reason from a premise that he merely believes, but does not know. But he cannot be guided by a fact he does not know, any more than a cat can stalk a bird it cannot see (or smell, etc.) or a traveller can follow a guide he cannot see (or hear, etc.). The relationship between knowledge and mere true belief is similar to the relationship between perception and veridical hallucination. If Aaron sees Moses taking the left fork, and follows him, Aaron is guided by Moses; whereas if Aaron cannot see Moses, but he hallucinates Moses is taking the left fork, he is not guided by Moses, even if he follows the same route. Thus being guided by a fact is different from merely reasoning from a premise or assumption. If we think of the case where a fact is a persons reason as a case where a person is guided by a fact, it appears that it has to be a fact he knows. For example, Jamess reason for going to church cannot be the fact that it will please his mother unless James knows that it will. His reason may be that he believes it will please his mother, or that it will probably please his mother, or that someone told him it would please his mother. But the fact that it will please his mother cannot be his reason, unless it is a fact he knows. Furthermore, if James merely believed that going to church would please his mother but did not know that it would, we can say that he went to church because he believed that it would please his mother, but we cannot say that he went to church because it would please his mother. But if he knew that it would please his mother, we can say either that he went to church because he knew that it would please his mother or that he went to church because it would please his mother. In this kind of explanation, knowledge is transparent: we can look straight through it to the fact. That is why I said earlier that in this kind of explanation, the explanans can be the persons reason, and that is why Prichard

10

John HYMAN

was right, in the remark quoted in my article How Knowledge Works and in Dancys Acting in Ignorance, to equate doing a duty because it is a duty and doing a duty really in consequence of knowing it to be a duty: [According to a certain view about duties] we can never, strictly speaking, do a duty, if we have one, because it is a duty, i.e. really in consequence of knowing it to be a duty. At best, if we have a duty, we may do it because we think without question, or else believe, or again think it possible that the act is a duty (Prichard 2002, p. 87). The reason for this difference between what we can say if James knew that going to church would please his mother and what we can say if James merely believed that going to church would please his mother, I suggest, is that knowledge is the ability to be guided by the facts. Whether James goes to church because he believes that it will please his mother or because he knows that it will please his mother, he is influenced by his state of mind: his believing something in one case, and his knowing something in the other. However, since knowledge is the ability to be guided by the facts, in the latter case, and in the latter case alone, we can explain why James went to church not only by identifying the state of mind that influenced him, but also by identifying the fact that guided him; not only by saying that he went to church because he knew that it would please his mother, but also by saying that he went to church because it would please his mother. Of course both explanations are factive, but only the fact that functions as the explanans in the second explanation is the agents reason. And of course a reason, in this sense, is not merely a consideration that makes an action seem appropriate to the agent; it is a fact that guides the agent when he acts. 5 Finally, I shall make a brief remark about the intellectual gain this conception of a persons reason yields. I have just mentioned the most substantial gain. The claim that the fact that p can only be a persons reason for doing something if she knows that p can be part of larger argument, shows that it is better to conceive of knowledge as a kind of ability than as a kind of belief. The general idea that knowledge is a kind of ability can be found in remarks by Wittgenstein and Ryle (Wittgenstein 1958, p. 150; Ryle 1949, p. 129). But it has received vastly less attention than the idea that knowledge is a kind of belief, and the question of whether it is possible to analyze the concept of knowledge has been mistakenly equated with the question of whether it is possible to analyze the concept of knowledge in terms of the concept of belief. As soon as we conceive of knowledge as a kind of ability, the task of defining

Acting for Reasons: Reply to Dancy

11

knowledge looks very different, and more tractable. For instead of asking what we need to add to believing to get knowing, this new conception of knowledge encourages us to ask how knowledge is exercised or expressed, because this is always how we define abilities. And actually it is not very difficult to say how knowledge is exercised or expressed. Knowledge gets exercised or expressed whenever a modification of a persons thought or behavior is informed by a fact he knows, in other words, whenever a fact a person knows is one of his reasons for modifying his thought or behavior in some way. This theory of knowledge can be summarized in the claim that knowing that p is being able to be guided by the fact that p; or simply that knowledge is the ability to be guided by the facts. I defend this theory of knowledge in the articles included in the list of references. References
Audi, R. (1993). Action, Intention and Reason. Ithaca, NY: Cornell UP Dancy, J. (2008). On How to ActDisjunctively. In: A. Haddock & F. Macpherson (eds.), Disjunctivism: Perception, Action, Knowledge. Oxford: OUP Hyman, J. (1999). How Knowledge Works. The Philosophical Quarterly, 49: 433451 Hyman, J. (2006). Knowledge and Evidence. Mind, 115: 891916 Hyman, J. (2010). The Road to Larissa. In: M. de Gaynesford (ed.), Special Issue: Agents and Their Actions, 23: 393414 Prichard, H. A. (2002). Duty and Ignorance of Fact. In: J. MacAdam (ed.), Moral Writings. Oxford: OUP Wittgenstein, L. (1958). Philosophical Investigations. Oxford: Blackwell Ryle, G. (1949). The Concept of Mind. London: Hutchinson