CANADIAN JOURNAL OF PHILOSOPHY Supplementary Volume 35. Guidance and Belief

Transcription

1 CANADIAN JOURNAL OF PHILOSOPHY Supplementary Volume 35 Guidance and Belief There is a difference between those things one does that manifest agency and those things that merely happen to one or that are the effects of one s agency. My typing these words manifests my agency is an action of mine whereas growing older is merely happening to me and making sounds as I type is but an effect of my action. Actions are sometimes but not always done for reasons and are characteristically but perhaps not invariably known by the agent without observation or inference. I m typing in order to get my paper done on time, and I don t need to look to know that this is what I am doing, though I wouldn t know without hearing it that I am making noises as I type. In this paper I will offer an account of these facts about action. The account starts from the idea that an agent performing an action guides what she is doing. The account I will develop is meant to contrast with what is now the dominant view, which aims to understand action and acting for a reason in causal terms. This dominant view has three parts. One is a causal theory of action. It holds that whether what a person is doing manifests agency depends on what caused her to do it, on the events that brought the action about. A second part is a causal theory of acting for a reason. It holds that when a person is doing something for a reason, those reasons caused her to be doing it. On one version of the dominant view, all actions are done for reasons, so that a person is performing an action only if her reasons made her do it. The third part of the dominant view is a causal conception of beliefs, desires, and intentions. It holds that these are internal bodily states fit to cause actions, and so fit to be reasons for action. My typing these words is an action, on this picture, because my doing it was caused by the internal psychological states that are the reasons for my 63

2 typing. 1 This package of views, linking a causal conception of action and of acting for a reason with a causal conception of belief and other psychological states, forms a familiar and compelling picture. But the conception of action at the core of this picture is not mandatory. Some theorists hold that whether something is an action depends on what else is true of the agent at the time of the action and not on the action s causes. Harry Frankfurt (1978) holds that an agent is performing an action when she is guiding what she is doing. Elizabeth Anscombe (1963) claimed that an agent is doing something intentionally performing an action when she knows independently of observation what she is doing. Though there are important differences in their views on action, Frankfurt and Anscombe agree that whether an agent is performing an action depends on what else is true of the agent at the time of the action, and not on what caused her to perform it. 2 This non-causalist view of action does not fit comfortably with the other two parts of the dominant picture. If something s being an action is not a matter of its causes, then why should an action s being done for a reason be a matter of its causes? And if an action s reasons are not among its causes, then why should we think that beliefs, desires, and intentions must be internal bodily states fit to cause actions? Indeed, why even think that beliefs, desires, and intentions 1 Two additional theses are usually added to this package. One concerns the identity of actions and holds that actions are in every case identical with bodily movements. The other thesis concerns explanation and holds that reasons both cause and explain actions. I won t discuss either additional thesis here in any depth. 2 One striking difference is that whereas Frankfurt was a compatibilist about free will and determinism, Anscombe is not. It is not clear to me whether Frankfurt would agree with Anscombe that actions are essentially known without observation. He says that his construals of having reasons and of rational behaviour do not include any essential reference to beliefs (Frankfurt 2002, 62). While he allows that spiders and snakes are capable of rational behaviour, he is reluctant to ascribe beliefs to them, saying that they are not sufficiently self-conscious to understand or to articulate for themselves what they are up to. I agree that spiders and snakes cannot articulate what they are up to, but I deny that it follows that they do not know or have beliefs about what they are up to. But I am also not sure what substantive matter hangs on this. It might be a merely terminological choice about how to use the word belief. 64

3 Guidance and Belief are reasons for action in the first place? Abandoning a causal conception of action thus makes room for us to consider alternative conceptions of acting for a reason and of the nature of belief, desire, and intention. Here is the alternative I propose. An agent is performing an action when she is guiding what she is doing, regardless of what caused her to be doing it. Guiding what one is doing is a matter of being ready to adjust one s performance in the face of obstacles and to move on to the next phase or part of the action when the time is right. To say that someone is performing an action is thus to say something about her guidance dispositions. Guidance dispositions are also relevant to understanding acting for a reason and the nature of belief. To describe an agent s reasons for doing something and to say what she believes is to say something, not about internal states fit to cause actions, but about her guidance dispositions, about how she is likely to act and react in various situations as she goes about performing her actions. One virtue of this proposal is that it provides a straightforward explanation of Anscombe s insight that agents know what they are doing without the need for observation or inference: the dispositions involved in guiding an action are the very same as those involved in being aware of it. Guidance is the practical face and belief the theoretical face of a single phenomenon. This proposal can be a genuine alternative to the dominant view only if its central notion of guidance is not itself a causal notion. It is very natural to think that for an agent to guide what she s doing is for her (or for her internal states) to be ready to cause changes in the flow or direction of her behaviour. This would be to think of our relation to our actions as like that of a captain to his ship, or a driver to her car. Some causal theorists have even adopted a guidance condition, so that actions, on their view, are propelled and then guided in a certain way by beliefs in conjunction with desires and/or intentions. So unless guidance can be understood in some other way than as causal intervention, a guidance account of action and belief will constitute no real alternative to the dominant causal view. Fully developing and defending a guidance conception of action, belief and reason that does not collapse into a causal account is beyond the scope of this paper. My aim here is to sketch the outlines of such 65

4 a conception and to show that it constitutes a viable alternative to the traditional causal conception. 3 I will start, in section 1, but offering a dispositional way to understand the thought that agents guide what they are doing when they act. In section 2, I use this notion of guidance to ground an account of belief. In section 3, I show that these notions of guidance and belief can help us to understand what it is to act for one reason rather than another and why it is that agent s tend to know independently of observation what action they are performing. I conclude, in section 4, by rejecting reasons to think that the central notion of guidance must ultimately be understood in causal terms. 1. Action and Guidance When one performs an action one guides what one is doing. Guiding what one is doing involves being ready to respond to obstacles that threaten to block one s performance and being ready to move on to the next phase or stage of the action when the time is right. Guiding what one is doing thus requires having what I will call guidance dispositions. This means that action statement both report actual events and predict possible ones. To say that Jones is making tea reports an actual event, since the statement is false if no event of tea-making is in fact underway, and it predicts possible ones, since the statement is false if Jones is not disposed to do what it takes to get the tea made. The aim of this section is to elaborate on these points. Let me start by saying something about how I am using the word action. 4 (I will sometimes say that an agent performing an action is doing the thing intentionally.) I use it to mark those cases where an agent s doing something manifests her agency. I have in mind a contrast between things that an agent herself does and things that merely happen to (or within) her, or that she does but only as a consequence of and not as a manifestation of her agency. So there are in fact many 3 For recent non-causalist views of action, see Scott Sehon (2005) and Fred Schueler (2003). One difference between their views and mine is that, as far as I can see anyway, both Sehon and Schueler hold that actions are invariably done for reasons. In the cited works, both offer non-causal, teleological accounts of the explanation of actions in terms of reasons. I won t say much here about explanation. 4 This usage is described by Frankfurt (1978). 66

5 Guidance and Belief contrasts here. Though examples seem easy to come by, there are also a variety of borderline cases, and it is not easy to find a general characteristic. My typing these words is an action, whereas my getting older and my falling toward the centre of the Earth are not, even though I am doing both things. They are merely happening to me. So the notion of action I have in mind is meant to capture a contrast between being active and being passive. But there are other contrasts in the neighbourhood. My cat is climbing the stairs and this is an action. But in doing so she is also moving dust particles, and this is not an action, though she is not exactly passive with respect to it. Moving the particles is a consequence of her agency, not a manifestation of it. So action in the sense I have in mind is also meant to contrast with those things an agent does but only as a byproduct of exercises of her agency. Part of the problem of action, of course, is to try to bring some clarity and system to these contrasts. Two more contrasts are worth mentioning here. Actions are sometimes intended, in the sense that the agent planned in advance on doing them. But actions are not invariably intentional in this sense. My cat s climbing the stairs is an action, even if, as seems reasonable given her limited cognitive capacities, it is not something that the cat intended or decided in advance to do. The fish s swimming in the tank is an action, but not one that was planned or intended in advance. So while all intended actions are of course actions, not all actions are intentional, in this sense. A related point is that actions are sometimes done for a reason or with a purpose in mind, but this too is not invariably the case. 5 My pacing back and forth as I lecture is an action, but not one done for a reason. 6 My cat might be climbing the stairs to get to the litter box, but she might also just be climbing them for no particular reason. Her climbing would be an action would be 5 What I have called the dominant view tends to neglect such cases of action, focusing instead on actions done for a reason. This focus might be due to the fact that the original dialectic concerned the explanation of actions in terms of reasons, with causalists and non-causalists in opposing camps on this issue. I won t say much here about how to explain actions, whether done for reasons or not. 6 Sergio Tenenbaum calls such actions non-intentional but voluntary (2007, 90 99). What is important for my purposes is the fact that they manifest agency. I won t discuss voluntary action here. 67

6 something that she is doing, as opposed to something that is merely happening to her or that is a mere consequence of her agency even though she is not doing it for a reason. That action statements report actual happenings is not controversial. 7 It is entailed by the standard view that an action just is an event caused in a certain way. But it is worth noting that it does not follow that action statements report current bodily movements. For an action statement may be true even when the agent is not engaged in any movement relevant to the action, and even when she is motionless. 8 I may be making a pot of tea as I sit reading my book, even though none of my current movements are parts of my tea-making activity. My tea-making does not call for me to move. I am waiting for the water to boil. I think that a good deal of what I am now doing (writing a book on belief, paying down my mortgage, training for a running race, and so on) does not at this moment call for me to move. So the fact that action statements report current happenings does not mean that action statements report current bodily movements. Still, even with this qualification, the idea that to say what someone is doing is to report an actual event or happening is not controversial. 9 It is also not controversial that an action statement predicts possible happenings, though this is perhaps less often noted. An action statement involves two sorts of predictions, one concerning how the 7 I should note that some theorists hold that actions are not themselves events. On one such view, actions are the causings of bodily behaviours, and causings are not themselves events. See Bach (1980), and Alvarez and Hyman (1998). But this view is compatible with the claim that an action statement reports an occurrence of an event, though on this view the event reported would not be the action reported. I won t consider this view here. 8 Falvey (2000) characterizes this as the broadness of actions describable with the progressive aspect. The relevance of this to an account of action is stressed in Thompson (2008). 9 Another terminological note. One might want to reserve the word event for something that comes to pass as opposed to something that is merely underway. It is often noted that something that is underway might never come to pass: I was making a pot of tea even though no tea-making ever happened, because I changed my mind or was prevented from finishing. One might want to describe this by saying that no tea-making event occurred, even though a tea-making was underway. I won t take a stand on this issue. 68

7 Guidance and Belief agent would respond to obstacles to the successful performance of the action and the other concerning the progress of the action through its typical phases or stages. That an action statement involves these predictions reflects the fact that in performing an action an agent guides what she is doing. Harry Frankfurt introduced the idea that in acting an agent guides what she s doing by considering when the complexity in an agent s behaviour indicates that the she is performing an action: Complexity of bodily movement suggests action only when it leads us to think that the body, during the course of its movements, is under the agent s guidance. The performance of an action is accordingly a complex event, which is comprised by a bodily movement and by whatever state of affairs or activity constitutes the agent s guidance of it. Given a bodily movement which occurs under the agent s guidance, the person is performing an action regardless of what features of his prior causal history account for the fact that this is occurring. He is performing an action even if its occurrence is due to chance. And he is not performing an action if the movements are not under his guidance as they proceed, even if he himself provided the antecedent causes in the form of beliefs, desires, intentions, decisions, volitions or whatever from which the movement has resulted. (Frankfurt 1978, 73) On Frankfurt s view, an agent is doing something intentionally performing an action only when she is guiding what she is doing. To say that an agent is guiding what she is doing is to say something about her dispositions, about what she would do in various situations. It is thus to make a kind of prediction. Frankfurt says that action is intentional movement, by which I take it he means intentional bodily movement. I agree that actions often involve bodily movements and it may be that only a being able to perform bodily movements could be an agent. But, as I noted above, one can be performing an action without bodily movements. And there are cases where the agent s guidance seems to be aimed at something other than bodily movement. Guiding the steeping of the tea is part of what it is to make tea. After I pour the water into the teapot, I m ready to intervene if needed to ensure that the steeping is not interrupted (by my cat, say) and to make sure that it does not proceed too long. My guidance 69

8 dispositions are aimed, it seems, not at my body s movements, but at the changes in the tea. This is so even though the interventions that would manifest this guidance would involve guided bodily movements. So the guidance involved in action is not, as Frankfurt seems to suggest, limited to the guidance of bodily movements. There are two confusions about guidance to be avoided here, though the first is probably harmless. One is to think that it is the action that the agent guides. The idea, rather, is that it is what the agent is doing, not the action, that the agent is guiding. As the cat climbs the stairs, she is guiding her progress up the stairs, and the fact that she is guiding it is what makes what she is doing the performance of an action it is what makes it a manifestation of her agency. To guide an action would be, on this usage, to guide a guided doing. It might make sense to think of guiding someone else s action, as when a piano teacher guides a student s playing. But in such a case the student s playing the guided action would not be a manifestation of the teacher s own agency. It is best, I suggest, to think of an action as guided doing. Still, this confusion is probably harmless, in the way that speaking of the justification for a piece of knowledge is harmless, when strictly speaking it is the belief that is justified, not the knowledge. A second confusion is perhaps more significant. It is to think that an agent s guidance of what she s doing is itself an action or that it somehow involves her acting on her bodily movements or on what she s doing. Guiding what one is doing is not something that one does: it is, rather, a way of doing something. To say that an agent is guiding what she s doing is to describe her dispositions, not to describe something else that she is doing over and above the action she is performing. The passive formulation, a doing that is under the agent s guidance is better than the active, guiding what she is doing, at capturing this aspect of the account. A related confusion would be to think that guiding what one is doing must involve acting on one s bodily movements. This is the confusion of thinking of our selves as like a captain who guides his ship s course by acting on the rudder, or like a driver who guides her car s movements by acting on the steering wheel or the brakes. I take it that an agent is not related in that way to her bodily movements when she is performing an action. Guiding what one is doing is not itself an action, nor is it a matter of acting on bodily movements. Rather, guiding what one is doing 70

9 Guidance and Belief is a matter of how one is disposed to move or act in various situations. The trick, of course, is to understand how guidance, understood purely dispositionally, can make the difference between something that manifests agency and something that does not. How can adding something that seems essentially passive yield agency? I think this question is at the heart of the problem of agency and I will return to it below. Though guiding what one is doing is not itself an action, it can and perhaps always does involve being ready to perform actions. As the cat intentionally climbs the stairs she is ready to move to the side should she encounter an obstacle, and were she to move to the side this would likely be something that she would do intentionally; it would also be guided behaviour. So the dispositions that constitute her guidance may include dispositions to perform actions of one sort or another. This is, I think, the typical case. But from the fact that guiding one s behaviour includes being disposed to perform actions it does not follow that guiding one s behaviour is itself an action or that it somehow involves acting on one s behaviour. Guiding what one is doing involves two sorts of dispositions. First, it requires being ready to make adjustments in response to possible obstacles. Frankfurt drew attention to these cases. Here is how Kent Bach, another of the first theorists to draw attention to the need for guidance in an analysis of action, put the point: [S]ome actions are more routine than others, requiring little if any attention, no decisions or flexibility during performance, hence minimal awareness. Indeed, if the action is not only routine but repetitive, maximal efficiency may require minimal awareness too much attention may slow one down and cause unwanted variations in performance. On the other hand, a novel, intricate action, like attempting an untried surgical technique, may require continuous, detailed, and accurate awareness of what one is doing. On must be alert to the slightest unforeseen contingency and be ready to alter one s course of action accordingly. (Bach 1978, 368) If Jones is turning on the light intentionally, then he must be ready as he raises his hand towards the switch to adjust his movements in whatever ways would be needed in order to get the light switched on. 71

10 He must, for instance, be ready to use more force if he finds that the switch is stuck. A cat intentionally climbing the stairs must be ready, if the situation calls for it, to adjust his path up the stairs if he encounters an obstacle. Of course, one possible response in the face of an obstacle is to abandon the action altogether. So whether an agent is doing something intentionally whether his doing it counts as an action of his depends on whether he is ready to adjust what he s doing in response to unforeseen contingencies. Because unforeseen contingencies are always possible for any action, guidance is essential to action. Guiding what one is doing also requires a second kind of disposition. It requires being ready to perform the next phase or part of the action when the time is right. This is clear in cases when the action is a complex one. Making an omelette involves breaking eggs into a container of some kind, mixing them, heating a pan, pouring the eggs into the pan, etc. When we say that Sarah is making an omelette, we are saying that she is more or less ready to perform each stage or phase when the time is right. Some actions involve a strictly ordered sequence of parts or phases, while others do not. But all actions, even such a simple one as switching on the light, involve parts or phases. Turning off the light requires first reaching up to the light switch, touching the switch, pushing it, etc. All actions, even the simplest ones, take time and so consist of parts or phases. This means that performing an action always requires being ready to perform the next part or phase when the time is right. Performing an action requires having guidance dispositions, but this does not mean that we can specify for each action type a set of dispositions that is necessary and peculiar to its performance. There are many reasons for this. For one thing, there are usually different ways to perform an action of a certain kind even in a given context, and the range of potential obstacles may vary from one circumstance to another. For another thing, some agents are better than others at performing certain actions, and this is a matter of training and experience among other things. Knowing this may influence what dispositions we expect an agent to have when we say that they are performing some action. So what dispositions we expect an agent to have when we make an action statement is a context-dependent matter. Let me summarize what I have been saying about guidance and action. To say that an agent is doing something intentionally is both 72

11 Guidance and Belief to report an actual event and to predict possible ones. Performing an action requires being ready to adjust one s behaviour as needed in order to get the job done, either because obstacles have emerged or because it is time for a next phase or stage in the action. These guidance dispositions typically include dispositions to perform various actions. Thus, to say that an agent is doing something is not to make a purely categorical statement about the occurrence of some event, nor is it to make a purely dispositional claim about the agent of the event. It is, rather, to make what Ryle once called a mongrel statement, one that both reports an actual event and predicts possible ones Guidance and Belief I want to turn now to belief. To say what an agent believes is to describe her conception of things, to characterize the world as she takes it to be. An agent s conception interacts with her actions in different ways. For one thing, an agent s conception of her world may include the fact that she is performing some action. It may even include facts about why and how she is doing it. I think that doing something intentionally requires believing that you are doing it, and in the next section I will consider how we should understand this requirement. But an agent s beliefs are also relevant to her guidance of what she is doing, and understanding this gets us to the nature of belief itself. The aim of this section is to develop this idea. All sides can agree that to say what an agent believes is to say something about what she would do in various situations. Here is one way to put the thought: to believe that p is to be disposed to act as if it were the case that p. But it would be good to elaborate on acting as if it were the case. The fact that action requires guidance suggests the following elaboration: to believe that p is to guide what one is doing in ways that would be effective if it were the case that p. According to this proposal, the dispositions associated with belief are guidance dispositions. I said that an action statement is a sort of mongrel statement involving both a report and a prediction. The same, I think, is true of belief ascriptions. In saying what someone believes, we are not just 10 Gilbert Ryle (1949, 141). 73

12 predicting possible occurrences, for her current actions are also being guided in accordance with that conception. To say what an agent believes is to say something about the path that she is in fact taking. In describing an agent s beliefs we are identifying a world in which she is doing certain things, for certain reasons, and in certain ways. In fact, we identify a space of possible worlds, differing one from another in terms of possible obstacles and effective responses. In identifying this space, we are characterizing the path that she would follow in a world that is as she takes the actual world to be. Belief ascriptions are thus also a sort of mongrel statement: they both predict possible occurrences (how she would behave in counterfactual worlds) and report on actual ones (how she acting in the actual world.) The guidance account of belief helps us to understand why citing a belief might be relevant in explaining an agent s action even if, as I think, citing a belief is not citing a cause of the action. To ascribe a belief is to say something about the path along which the agent is guiding what she s doing. Suppose we see Jones on the neighbour s front porch, pressing a button. When asked why he s doing that, we explain that he is trying to find the neighbours and believes that the button is a doorbell. In saying that our agent believes that the button is a doorbell we are saying that he is guiding his activity that of seeing if the neighbours are at home in a way that would be effective in a world in which that button is a doorbell. We are saying something, not about the causes of his behaviour, but about the path that behaviour is in fact taking, and the paths it would take in various possible situations. We are, in other words, saying something about how he is going about doing what he is doing and about what he would do if various obstacles arose. On this account, citing an agent s belief can explain his actions by providing a richer understanding of the how and why of that action. This result should not be surprising if, as I have suggested, both action-ascriptions and belief-ascriptions are mongrel statements, both reporting an actual occurrence and predicting possible ones. If this is right, then it is a confusion to think that in ascribing beliefs we are saying something altogether different about an agent than what we are saying in ascribing actions. One might have thought that in ascribing an action we are only mentioning an event, whereas in ascribing a belief we are only mentioning a disposition. But if such ascriptions are 74

13 Guidance and Belief both mongrel statements, both reporting an actual occurrence and predicting possible ones, then there is no such deep distinction in subject matter between belief ascriptions and action ascriptions. Both sorts of ascriptions describe the very same phenomenon, though from a different perspective or with a different focus. Action reports tend to focus on the actual occurrence while belief reports focus more on the possible ones, though both say something about both kinds. If this is the right way to think of belief ascriptions, then it ought not to be surprising that belief ascriptions can explain actions without citing causes. Fully developing this conception of belief is beyond the scope of this paper. In the remainder I want to explore the idea that agency can be grounded in mere dispositions. I realize that this idea can seem puzzling, even paradoxical. Dispositions can seem to be the paradigm of passivity, the polar extreme of action. Some theorists have argued, for different reasons, that guidance itself must be understood in causal terms if it is to ground agency. I don t agree and think that the urge to appeal to causation may rest on confusion about agency and possibility. In the next two sections, I will argue that the guidance account of belief and action can help us to understand fundamental features of agency: acting for a reason and the special awareness agents seem to have of their own actions. I will then end the paper by considering, and rejecting, reasons to think that guidance must in the end be understood in causal terms Guidance and Acting for a Reason The guidance accounts of action and belief can help us to understand several key marks of agency. First, it allows us to distinguish what an agent is doing intentionally from what she is doing knowingly but non-intentionally. Second, it allows us to understand what it is for an agent to act on one reason rather than another. Finally, it allows us to understand the special sort of awareness that an agent seems to have 11 I have not offered a reductive account of action. Guidance dispositions include dispositions to perform actions. One might think that this threatens a vicious regress, one that could only be stopped by something categorical (as opposed to dispositional) or by something dispositional that is not a disposition to act. I won t consider this threat here. 75

14 of her own action. Each of these is an essential mark of agency. So a guidance account of action can help us to understand at least some of what seems central to agency. In this section I consider the first two points, and I turn to the third point in the next section. Whether an agent is doing something intentionally depends on her guidance dispositions. This means that guidance dispositions are relevant to determining what actions an agent is performing. Whether my cat is climbing the stairs intentionally depends (inter alia) on whether she is prepared to do what it takes to accomplish that to act to overcome obstacles should an obstacle emerge and to move on to the next stage or phase of the action when the time is right. That guidance dispositions are relevant in this way is now, I think, fairly widely accepted among theorists of action. But it is perhaps not as widely recognized that guidance dispositions are also relevant to marking off those things that an agent is doing intentionally from those things that she is doing knowingly but non-intentionally. As I type these words I am making noises, and I know that I am. But I am not making them intentionally, whereas I am typing intentionally. The relevant difference is that, while I am ready to make adjustments in my behaviour to ensure that my typing gets done, I am not ready to make adjustments to ensure that the noise is being made. Indeed, given that my daughters are asleep in the next room and might be woken by the noise, I d be happy if I could type without making the noise. The air-force bomber who knowingly kills the civilians while intentionally destroying the factory is not killing them intentionally unless he is ready to adjust his behaviour as needed to ensure that they are killed and this is so even if he knows that he is killing them by bombing the factory. So a guidance account of action can help us to distinguish an agent s actions from things that she is doing knowingly but that are not actions. A guidance account of action can also help us to understand what it is for agents to act for reasons. By acting for a reason I have in mind cases when an agent does something with a further goal or aim in mind. We ordinarily describe such cases using a sentence like: S is doing X in order to Y. I put the kettle of water on to boil in order to make tea. The cat is climbing the stairs in order to get to her litter box, or because she wants to get to her litter box. (There are important differences here, but since they won t matter for my purposes, I will set them aside.) 76

15 Guidance and Belief One point to keep in mind is that something s being an action is not the same as its being done for a reason or on purpose, and the guidance account of action can help us to understand this. A fish that swims from one side of her tank to another is manifesting her agency, but she may not be swimming for any reason. When I pace back and forth, my pacing is something that I am doing and not merely something that is happening to me, or a consequence of my actions. But I may be pacing for no reason at all. I m just pacing. So what an agent is doing might manifest her agency even though she has no further goal or aim in mind. The guidance account of action shows why, since in these cases the agent is guiding what she s doing, even though she s doing it for no reason. She s guiding it, and this manifests her agency, even though she is not doing it for a reason. It may be, of course, that not all of an agent s actions can be done for no reason. This certainly seems to be so in the case of mature humans. It is hard to make sense of the possibility of a person going through life or even through a significant stretch of time performing actions but all for no reason. But it is hard to know what to make of this fact, if it is a fact. It may point to something fundamental about agency itself. But it may instead point to the fact that mature humans have or develop rational capacities capacities to plan or to predict the future, say whose normal operation makes non-rational actions rare even though these capacities are not strictly needed for mere agency. In any case, it is clear that something s being a manifestation of agency is not the same as its being done for a reason, and the guidance account of action can help us to see this. The guidance account of action can also help us to understand what s going on in cases when an agent does act for a reason. More specifically, guidance dispositions are relevant to determining what reason an agent is acting on when she has several different reasons for doing what she is doing. 12 Suppose that Jones has two equally good reasons for marching: by marching he would obey his commander 12 Davidson (1963) had charged that a non-causal account of action explanation could not mark this difference. On his view, the best way to understand what it is for an agent to have acted on one set of reasons rather than another is for that first set to have caused the action. The response I develop in the text is also found in Sehon (2005). 77

16 and he would impress his girlfriend who is watching from the sidelines. Whether in the end he is marching to follow orders or to impress his girlfriend depends on how he would behave in various possible situations. Would he continue to march about once his commander is satisfied, or would he stop marching in that case, even if his girlfriend had not yet seen his marching? How he would act in these different cases is relevant to determining which reason is the one he is acting on. So guidance dispositions are relevant to determining what reasons an agent is in fact acting on in cases when she has more than one reason to act. It is worth noting that an agent may perform some action for many reasons at the same time, or even change reasons in mid-stream. Maybe Jones is ready both to do what it takes to satisfy his commander and to do what it takes to impress his girlfriend. If you ask him why he s marching, he ll mention both considerations. Perhaps as his marching progresses he ll drop one of those reasons in favour of the other. This would be reflected in a change in his dispositions. Or perhaps he ll adopt new reasons for marching, perhaps deciding that improving his own skills is at that moment more important than either obeying his commander of impressing his girlfriend. Again, such a change in his reasons would be reflected in a change in his dispositions. His action would remain the same, though his reason for performing it would change, and this would be marked by a change in how he is disposed to act in various situations. A guidance account of action can make perfectly good sense of cases where an agent acts on several distinct reasons or changes reasons in mid-performance. I have been arguing that guidance dispositions are relevant to explaining why an agent is performing an action. These dispositions are also relevant to determining how she is performing it. There are usually different ways to perform some action. Suppose that there are three ways to turn off the light. Since each of the ways of doing it is itself an action, each way of turning off the light requires a different set of guidance dispositions. So whether Jones is doing it one way or another will depend on her dispositions. This means that what guidance dispositions an agent has is relevant to determining not just why she is performing an action but also how she is performing it. 78

17 Guidance and Belief 4. Guidance and Agential Awareness I want to turn now to a further mark of agency, the fact that performing an action seems to require believing independently of observation that one is doing it. 13 It is not perfectly clear just what the requirement is here or what the relevant notion of observational independence is. But it does seem that an account of action and belief ought to help us to understand why there appears to be some such requirement. My aim in this section is to develop the following account of this: action seems to require belief because the dispositions that constitute an agent s guidance of her action also constitute her belief in it. The first point to make is that when an agent is performing an action the dispositions associated with her belief that she is are guidance dispositions. Consider again the idea that to believe that p is to guide one s actions in ways that would be effective if it were the case that p, and suppose that I am making a pot of tea. If I believe that the water is boiling, then I am ready to guide my tea-making in ways that would be effective in a world where the water is boiling. That s clear enough. But now consider my belief that I am making tea. This is a belief about an action that I am performing. What dispositions are associated with this belief? Well, my believing that I am making tea requires my being ready to guide my tea-making (which is an action) in ways that would be effective if I were making tea. What could this readiness be other than my being ready to do what it takes to make the tea? In other words, the dispositions involved in believing that I am making tea, when I am making tea, would seem to be the very dispositions associated with guiding my tea-making. This suggests that when someone is performing an action and believes that she is, that belief involves the very same dispositions as the action itself. But it also suggests more than this. It suggests that when an agent is performing an action, she will have the very dispositions that we would expect her to have if we took her to believe that she was doing it. The guidance dispositions I have as I make the tea are (or anyway, 13 That it does was a widely held view until the mid-1960s. It plays a central role in Anscombe (1963), Hampshire (1959) and Melden (1961). Davidson (1963) argued against it and these arguments seemed to hold sway until the idea was revived in Velleman (1989). There is now a large literature, some of it listed in the footnotes below. 79

18 seem to include) the dispositions that I would have (or would be expected to have) if I believed that I was making tea. In a way, this is not surprising, since the account of belief I have offered defines belief in terms of guidance. But this is not an objection, since any plausible account of belief will define it in terms of action. The point is that in the special case of an agent s beliefs about her own actions, the dispositions involved in that belief seem to be guidance dispositions. When viewed from the perspective of action, these dispositions seem practical, but when viewed from the perspective of belief, they look like cognitive dispositions. If this is right, then it suggests that action and belief, and perhaps even freedom and knowledge, are but two sides of a single coin. Exploring this is something I will save for another occasion. So a guidance account of action and belief can help us to understand why it might seem that performing an action requires belief in it. But it can also help us to understand why this belief might seem to be independent of observation. 14 It is worth distinguishing two ideas here. One concerns justification and holds that for an agent to be doing something intentionally her justification or evidence for believing that she is doing it must be independent of observation. The second idea concerns acquisition and holds that the fact that the agent has the belief must be independent of observation if she is to be doing it intentionally. I want to set aside the idea about justification and focus on the one about acquisition. The point I want to make is that if the dispositions associated with performing an action are or include those associated with believing in it, then this offers a simple explanation of the fact that such beliefs seem to be independent of observation, at least in terms of acquisition. An agent s guidance dispositions as she performs an action are independent of observation. It is true, of course, that guiding what one is doing requires being ready to respond to observations. I am ready to lift the kettle and pour the water into the pot once I see that the water has come to a boil. We can put this by saying that guidance dispositions are (or anyway include) sensitivities to what is observed. Doing something intentionally requires being ready to respond to the 14 This idea has roots in the works of Anscombe (1963) and Hampshire (1959). Recent discussion of it is in Moran (2004), Hursthouse (2000), and Falvey (2000). 80

19 Guidance and Belief passing scene. But guidance dispositions are independent of observation in the sense that the agent acquires the dispositions as she begins the action and loses them when she stops the action, and this correlation is independent of what she observes. It is what an agent is doing that determines her guidance dispositions, not what she is observing, and this determination is not undermined by the fact that observations may inform her actions. I will retain my dispositions to do what it takes to make the pot of tea so long as I am still making it, regardless of what I may observe, unless I decide to stop doing it. But if, as I have suggested, the dispositions associated with guiding an action are or include those associated with belief in it, then the fact that guidance dispositions are independent of observation in this sense would explain the fact that the belief seems also to be independent of observation. When an agent is performing an action, it will seem that she believes that she is and that this belief is independent of observation, since it is acquired once the action starts and disappears only when the action ends, regardless of what the agent may observe. I start believing that I am making a pot of tea as soon as I start to make the pot of tea and I stop believing it as soon as I stop what I m doing. I did not acquire my belief that I was making a pot of tea by observing my behaviours or movements or anything else. Likewise, I won t change my mind about whether I am making tea by observing my behaviours or movements, at least not so long as I am still in the process of making it. If I stop making a pot of tea, which I might do if I decide to make coffee instead or if I realize that I don t have enough time to make tea, then I will stop believing that I am. My belief is independent of observation in the sense that its duration coincides with changes in my activity and not with changes in my observations. It starts and stops with the action itself. The guidance account of action and belief thus promises a very elegant explanation of the fact (if it is a fact) that performing an action requires believing independently of observation that one is doing it. Some have tried to explain the belief requirement in terms of a special sense of agency, a perception-like capacity that produces and sustains the required belief, but I find the proposed capacity obscure and anyway this proposal would leave it a mere coincidence that acting requires practical and theoretical dispositions that are observationally 81

20 independent. 15 Others have tried to explain the belief requirement in terms of intention, either a prior intention or an intention-in-action, but like the appeal to a sense of agency this suggestion would leave it a mere coincidence that action requires observationally independent dispositions of both a practical and theoretical nature. 16 Another suggestion is that the action and believing share a cause, perhaps a prior intention, or a set of beliefs and desires, or a process of deliberation, or an act of willing. 17 But this couldn t be the general story since not all actions are antecedently intended, or result from deliberation, or are willed. My cat did not intend or deliberate about climbing the stairs, but she climbed them intentionally all the same, guiding and believing in her actions as she did. And even if this story could explain why the acting and believing begin together, it is not clear how it could explain why they persist together. I submit that the guidance explanation I have sketched promises a simpler and more satisfying account of the fact that performing an action seems to require special agential awareness. My proposal offers a neat explanation of the belief requirement: since acting requires guidance, if guidance constitutes belief then acting also requires belief. And it offers a neat explanation of the required belief s observational independence: since the guidance starts and stops when the acting starts and stops, if guidance constitutes belief then that belief will also start and stop with the action s start and stop (unless sustained in some other way). This proposal does not involve positing special perception-like capacities; it does not rely on ascribing essentially practical mental states like sustaining intentions; and since it does not imply that action is always a result of deliberation, it respects what is common across human 15 For a sample of recent views on this, see the essays in Roessler and Eilan (2003). 16 Moser and Mele (1994). Bach holds that guidance dispositions are associated with non-conscious mental representations, what he calls executive representations. He considers the idea that these representations might also be called intentions but dismisses it because, he says, calling non-conscious representations intentions would be Pickwickian (1978, 371). 17 For a development of the view that the belief is produced by deliberation, see Rödl (2007) and O Brien (2007). For a different version of the same idea, see Velleman (1989) and Setiya (2008). 82

21 Guidance and Belief and non-human action. Most importantly, it provides a satisfying explanation of the coincidence of theory and practice, a coincidence that will seem mysterious so long as guidance and belief are viewed as distinct phenomena. 5. Guidance and Causal Mechanisms I have argued that a guidance account of action and belief can help us to understand several key marks of agency. But it is central to this account that whether something is an action has nothing to do with its causes. It has seemed to many that only a causal account can explain agency. Causation can enter analyses of action and belief in various places and for various purposes. Frankfurt (1978) held that the notion of guidance is itself a causal one, to be understood in terms of an underlying mechanism set to produce changes in the agent s behaviour. Rowland Stout (2005), who follows Frankfurt on this point, has argued that only if such a causal mechanism is present can it be correct to say that the agent is the one guiding what she is doing. His hope is to rule out cases of deviant causation, where the agent s dispositions are grounded in external causal powers and not in the agent s own active powers. A similar hope prompted some causal theorists to say that the dispositions that indicate guidance are grounded in intentions (or intention-like) states that causally sustain the action and keep it on its path. I doubt that the concepts of action and belief require such causal mechanisms, and I am sceptical that appeal to them can help with cases of causal deviance. The aim of this section is to explain why. In denying that something s being an action is a matter of its causes, I do not mean to deny that the bodily movements involved in action are as likely as any other events to have causes. The movements of the cat s legs as she climbs the stairs are caused by something, and even if the physiological and physical story about these causes is not yet very well understood, I am not sceptical that some story involving physiological states, mechanisms, and properties can explain those movements. So the causal scepticism I want to develop in this section is not aimed at the movements themselves. Rather, what I doubt is, first, that the concepts of action and belief must be understood in terms of causal states or mechanisms and, second, that appealing to such states 83