Philosophical Writings Vol. 43 No.1 Special Issue: Proceedings of the British Post-Graduate Philosophy Association Annual Conference 2014 HOW NOT TO NATURALISE ACTION University College London Abstract: The standard story of action the attempt to analyse action, or intentional action, in terms of mental causation has long been beset by the problem of so-called deviant causal chains. In fact, these present a challenge to all theorists standard or otherwise who want to understand the connection between an agent s desires, intentions and beliefs on the one hand, and their actions on the other, as a causalexplanatory one. I argue that the proper resolution of this problem involves understanding the mental states at issue as dispositions characteristically manifested in intentional or goal-directed action, so that the intentional non-intentional distinction maps more or less cleanly onto the manifestation non-manifestation distinction. Because dispositions are defined in terms of their characteristic manifestations, though, this means that the states in question are not explanatorily or ontologically prior to goal-directed action and are themselves to be understood in teleological terms. I argue on this basis that a standard story -type analysis cannot succeed as a naturalising or reductive account of the nature of intentional action (or, a fortiori, of action). Keywords: intentional action, mental causation, mental states, dispositions, causal action Date Submitted: 26/09/2014 Date Accepted: 02/11/2014 Vol. 43 pp. 33-44
HOW NOT TO NATURALISE ACTION 1. Introduction Some philosophers are concerned about how to find a place for action in the explanatory order of the world (Velleman, 1992, 465). The problem is often framed using Wittgenstein s question, what is left over if I subtract the fact that my arm goes up from the fact that I raise my arm? (1968, 621), inviting the thought that the challenge is to work out what needs to be added to an arm-rising in order to make it into an arm-raising. 1 One influential approach has been to claim that something such as an arm s going up constitutes an action in virtue of being caused by certain mental states of the agent. As Velleman characterises the theory, it says that There is something that the agent wants, and there is an action that he believes is conducive to its attainment. His desire and his belief jointly cause an intention to take [the action], which in turn causes the corresponding movements of the agent s body. (Velleman, 1992, 461) Such a conception of the nature of action has been so influential that it has come to be known as the standard story. In this paper, I will address a variant of the standard story, namely the causal theory of intentional action, understood as an attempt to reductively analyse intentional action in terms of prior concepts of action, causation, and mental states. As I understand it, such a theory aims to (at least partly) naturalise intentional action: to explain how it fits into the efficient-causal order of things; specifically, to analyse it without reference to final causes. I ll argue that the causal theory is not the right means to that end. My argument is simple: the best treatment of deviant causal chains, a problem that has dogged all theories of this sort, involves understanding the mental states at issue as dispositions that are characteristically manifested in intentional, or at least goal-directed, action. Such a conception of these mental states is independently plausible, but is made especially attractive by its power to resolve the problem of causal deviance. However, since dispositions are defined in terms of their manifestation, these states are not ontologically or conceptually prior to goaldirected action, and must themselves be understood teleologically. So, if intentional 1 It was not, I take it, Wittgenstein s intention to invite this thought. 34
How Not to Naturalise Action action is to be naturalised, it will not simply be in terms of being caused in the right way by these mental states. I will focus mainly on states of desire, but suggest that the argument will also apply to intention and belief. 2. Causal deviance One can formulate a variety of causal theories about action, with different targets in mind, and it s often unclear who exactly holds exactly which version. A whole array of causal theories, for instance, are attributed to Davidson, but often at the risk oversimplifying his views. Davidson argued that something is an action just in case it is intentional under some description (2001, essay 3), that someone acts intentionally just in case they act for a reason, and that when someone acts for a reason their action is caused by a suitable belief desire pair (2001, essay 1). However, he suggested in later work that such a causal relation could not provide a reductive account of what it is to act intentionally (2001, essay 4). Many philosophers inspired by Davidson, however, do think that a reductive causal analysis is available. One much-discussed kind of account aims to analyse action in terms of bodily movement, claiming that bodily movements constitute actions in virtue of being suitably caused. Views of that kind won t be my primary target, but my arguments should apply to them a fortiori. 2 My target is the somewhat less ambitious attempt to analyse intentional action in causal terms. A theory of this sort can take action as primitive and claim that what makes an action intentional is being caused by the relevant mental states. If the primitive actionconcept is non-teleological then intentional action will have been, at least for our purposes, naturalised. 3 As it stands, though, even this causal theory is only schematic. If it is to provide a non-circular analysis of its target, and to reduce final to efficient causes, it must be more specific. First, it must specify what sorts of mental states or events are the right ones to cause an action so as to make it intentional. This is fairly simple. The right kinds of states are typically taken to be desire, intention or belief. They will 2 See e.g. Hornsby (2004) for criticism of the standard story ; Hornsby (1997, pt. 2), Owen (1980) for problems regarding bodily movements 3 One example of a non-teleological account of action is that given in Alvarez and Hyman (1998). Setiya (2009) appears to favour a reduction of intentional action along these lines. 35
also have to be states with contents relevant to the action in question, such as intentions to perform that action or to achieve some further end served by the action, or desires for some goal and beliefs about how to achieve it. In Davidson's terms, they must be states that rationalise the action. The second gap for the causal theorist to fill is to say what it is for the states to cause the action in the right way. This is rather more difficult, because of the problem of deviant causal chains the reason why we talk about the right kind of causation in the first place. It has long been recognised that whatever mental states you plug into the formula, it s possible to construct cases in which the states cause the agent to perform an action of the appropriate sort, but cause it in the wrong way such that the action is unintentional. The classic example is Davidson s: A climber might want to rid himself of the weight and danger of holding another man on a rope, and he might know that by loosening his hold on the rope he could rid himself of the weight and danger. This belief and want might so unnerve him as to cause him to loosen his hold, and yet it might be the case that he never chose to loosen his hold, nor did he do it intentionally. (2001, 79) Similar cases can be constructed for states of belief and intention. The import of such examples is that if a causal account of intentional action is to be given, we need to be able to specify in a principled way what the right kind of causation is, so as to rule such cases out in a non-circular, non-ad hoc fashion. Many attempts at this task have been made, but, up until now, none has been ultimately satisfactory. 4 Thankfully, recent work on dispositions offers a novel and plausible solution to the problem. As we shall see, however, it's no solution for our causal theorist. 3. Desire & dispositions We can begin approaching a resolution of the problem of deviant causation by noting the grounds on which we actually distinguish right from wrong causal chains. It's a matter of teleology. We judge that Davidson s climber s action wasn t caused in the right way by their desire because we judge that the climber doesn t 4 Mayr (2011, chap. 5) gives an overview; see also e.g. Sehon (1997). 36
How Not to Naturalise Action loosen their hold with the aim of ridding themselves of the weight and danger. This notion of acting with an aim is exactly the one that our causalist wants to naturalise. The causal theorist, then, has to assume that the right kind of causation can be specified without appeal to this notion. But unfortunately (for them), the most promising specification of caused in the right way succeeds precisely by retaining this teleological idea. Hyman, in a recent paper drawing from contemporary work on the metaphysics of dispositions, argues that the problem of deviant causation can be resolved as follows. Desires are dispositions of agents, and an agent s action is caused in the right way by a desire just in case it is a manifestation, rather than a mere symptom, of that disposition. Since dispositions are defined in terms of their manifestation, the criterion for right causation is given in the definition of desire. 5 Hyman s proposed (rough) definition is that a desire is a disposition manifested in two main ways: first, by purposive or goal-directed behaviour, specifically, behaviour aimed at satisfying the desire, in other words, at getting what it is a desire to have, or doing what it is a desire to do; and second, by feeling glad, pleased or relieved if the desire is satisfied, and sorry, displeased or disappointed if it is frustrated. (2013, 3) Explanations of action in terms of desire are causal because desires, as dispositions, are causal factors: an explanation of an action in terms of a desire is of the same general sort as one of a glass's shattering in terms of its brittleness. If we understand desires in this way, then the problem of deviant causation is shown not to be a special problem for the analysis of intentional action, but rather an instance of the general problem of reducing the manifestation of dispositions to causal chains of events. [W]ith certain exceptions, argues Hyman, every disposition can be connected to the kind of occurrence that normally manifests it by a deviant or freakish causal chain (2013, 12). He gives the following example: A man might take a soporific drug before driving, and the drowsiness induced by the drug might make him crash the car and knock himself unconscious. If this happened, he would lose consciousness because he took the drug, but the exercise of its virtus dormitiva would be preempted by the crash. (2013, 12) 5 See Wittgenstein (1969, 245) for the distinction between symptoms and criteria of dispositions. 37
Many similar examples can be given. Indeed, Davidson s climber is an example of the same sort: the climber s dropping his companion is caused by a mere symptom of his desire to rid himself of the weight, namely his nervousness at having such a desire; it is caused by the desire, but does not manifest the desire. Had the climber dropped his partner intentionally, he would have dropped him with the aim of ridding himself of the weight, and his action would thus have manifested his desire to rid himself of the weight, and so been caused in the right way by the desire. Hyman s account, then, gives us a principled way to distinguish right from wrong causation: an action is caused by a desire in the right way to make it intentional only when the action manifests the desire and is done with the aim of satisfying the desire. Hyman's account of desire seems to me basically correct, but it s beyond the scope of this paper to offer a full defence of it. That said, I think we have already seen one significant point in its favour, namely that it enables us to give a straightforward explanation of the proper causal connection between a desire and an action done with the aim of satisfying that desire. The right kind of causation is a matter of the action s manifesting the desire. This is significant progress although not towards the kind of analysis envisioned by the causal theorist. I ve suggested that we can specify a type of causal connection between mental state and action that is not subject to deviant causal chains. This seems to be pretty good evidence that this is the connection we are after that it is the right kind of causation. It shouldn t be surprising that identifying the right connection between state and action was in part a matter of giving an account of the nature of the relevant states. Our problem, after all, was how states (of a certain kind) cause actions (in the right way). How a type of state can cause an action obviously depends on what the state itself is like. With this in mind, it's perhaps surprising how easily the question of the nature of these states tends to be brushed over in discussion of the standard story. 6 We're now in a position to say something decisive about the unfilled gaps in the causal theory. However, we don't end up with a naturalising or reductive account of intentional action. The theory we get is that desires are dispositions 6 Cf. Steward (1997). 38
How Not to Naturalise Action characteristically manifested in, among certain other things, actions aimed at satisfying the desire. Dispositions are defined in terms of their characteristic manifestation so, on the present view, desire is defined in terms of goal-directed action a teleological notion. The right kind of causation is a kind of final causation. Moreover desire is, on this view, ontologically dependent on goal-directed action: what it is to desire to φ depends upon what it is to do something with the aim of φ- ing. 7 So an analysis in which desire occurs in the analysans has not reduced final to efficient causes. It might be tempting to think that we could get back to a naturalising theory by eliminating the notion of goal-directed action from the definition of desire, replacing it with bodily movement, or with action (unqualified). Since desires would then be dispositions defined in part in terms of (unqualified) actions, we would seem to have the potential for a naturalising account that defined intentional actions as just those actions that manifest desires. Unfortunately that won t do. For one thing, it doesn t look like an adequate account of desire, since in eliminating goal-directedness from the definition of desire, you eliminate your own ability to explain why desires explain actions. Why do you go downstairs when you hear the knock? Because of your desire to find out who is at the door, which is a disposition to what? A disposition to do certain things in certain circumstances, perhaps. But why are you disposed to act in these ways in these circumstances? Because these actions seem to you to be a means to finding out who is at the door. If our aim is to eliminate the placeholders ( right kind etc.) from our theory of action, we haven t succeeded by shifting those placeholders into our definition of one of the key terms in that theory. For much the same reason, eliminating goal-directedness from the definition of the disposition eliminates our ability to make a principled distinction between right and wrong causal chains. The dispositional account of desire enables us to make this distinction in a principled way because dispositions are defined in terms of their characteristic manifestations. The distinction between an action caused in the right way and one caused in the wrong way just is the distinction between a manifestation and a symptom of the state. The reason it distinguishes intentional actions from unintentional actions or mere bodily movements is that the state is manifested in the 7 See Kalderon (2012) on the notion of ontological dependence. 39
former and not in the latter. If the desire to rid myself of my companion s weight is, say, a disposition to do things that seem conducive to ridding myself of the weight, rather than a disposition to do things with the aim of ridding myself of the weight, then I could in principle manifest that desire by dropping my companion without doing so intentionally. The account would thus fail to close the door on deviant causal chains. 4. Intention & belief I ve so far only discussed the attempt to analyse intentional action in terms of desire. The very fact that the problem of deviant causation crops up just as readily for belief and intention as it does for desire in itself suggests, in light of the points already made, that belief and intention should also be understood in dispositional terms. However, given that causal theories commonly appeal not only to desire but to belief and intention too, a full account of where those theories go wrong will need to directly address these states as well. To show how this might be done, I'll now make some brief suggestions about how intention and belief might be understood as dispositions manifested in the kind of action the causalist wants to explain. It seems likely that in so far as there is a kind of mental state called intention that causes intentional actions, it will look very much like desire, at least on the broad philosophical conception of desire that we've been working with. Of course intention has special features that desires in general lack. On one account, intentions are states that an agent can form by making a decision to perform the relevant action, which (unless revised) will lead the agent to perform the action directly (Holton, 2009, 2), and which are stable, in the sense that they have a tendency to persist once formed. We might add that an intention to φ won t only lead the agent to φ if they happen to be lucky enough to end up in a position to φ, but will also lead them to take steps that seem conducive to their getting into such a position. My intention to make a cake is expressed just as directly in my buying flour as it is in my creaming the butter and sugar. Generally, your intention to achieve some end motivates you to take the means to that end as well as to consummate the end. Given these observations, it's plausible that the definition of intention will have a certain amount of overlap with that of desire: an intention to φ is at least in part a (stable) 40
How Not to Naturalise Action disposition to φ, which can (but needn't) be formed by deciding to φ, and which is manifested in behaviour aimed at φ-ing (including the act of φ-ing itself). This (partial) definition, like that of desire, involves the notion of goal-directed action, and as such it is no more suitable for a naturalistic account of intentional action. There may be a lot more that we want to say about the nature of intention, but none of it seems likely to overturn this point. Belief needs to be treated slightly differently, not least because belief is best suited for a slightly different sort of causal theory. The concepts of acting intentionally and acting for a reason are often conflated, but they ought not to be (see Alvarez, 2009), and the notion of belief is clearly more intimately connected with the latter than with the former. Belief seems more appropriate to be employed in a causal theory of acting for a reason or, better, acting on a ground 8 than of acting intentionally. When one acts on a belief, one s belief (i.e., the thing one believes) is the ground on which one acts. As before, it s tempting to say that what makes it the case that one acts on the ground that p is that one s action is caused in the right way by one s belief (i.e., one s state of believing) that p. And as before, deviant causation rears its ugly head. (It should be easy enough to see how; I won t bore you with examples.) Again, I suggest that the solution is to define belief as a disposition manifested in the relevant kind of action: to believe that p is, at least in part, to be disposed to do things on the ground that p (where do things covers thought and feeling as well as bodily action, and the ground that p is non-factive). If that s correct, then the right kind of causation to connect a belief to an action done on grounds of the content of that belief is straightforwardly analogous to the one for desire: the action is caused in the right way when it manifests the belief in question. Again we don t have a naturalising analysis, because again the mental state in question is defined in terms of the thing the causalist wanted to naturalise, namely (in this case) the notion of acting on grounds. 8 I say ground because I take reason, or at least one important sense of reason, to be factive. A ground here is the sort of thing one can act on even when, as it turns out, one is mistaken. 41
5. Conclusion Where do we go from here? We ve seen good reason to think that an analysis of intentional action in terms of causation in the right way by desire attempts to analyse intentional action in terms of something desire which is not more fundamental than intentional (or at least goal-directed) action, and which is (derivatively) teleological in nature. We ve also seen that matters are likely much the same with respect to the states of intention and belief. Is there any hope, then, for fitting intentional action into the efficient-causal natural order? The intimate connection between intentional actions and, on the one hand, actions simpliciter and, on the other, mental states, made it seem that the causation of actions by states was the most promising way to explain intentional action in terms that appeared more naturalistically tractable. This thought looks to have been misguided. What, then, should we say about the project of naturalising intentional action? One point to note is about mental states. We ve seen that defining desire as a kind of disposition manifested in goal-directed action provides a strong solution to the problem of deviant causal chains. This seems like pretty good evidence that such a definition is correct. I ve also suggested that similar accounts may be given for belief and intention. Such analyses explain why we make an intuitive distinction between right and wrong causal chains, and also explain how action-explanation in terms of those states is causal as well as teleological (or rational in the case of belief). These accounts, if correct, have significant implications for the project of naturalising the mental. Perhaps most notably, they show that finding a place for these states in a naturalistic worldview will depend in part on doing the same for dispositions and powers in general. A second point a suggestion, really is that properly naturalising the mental phenomena discussed here will not be a matter for simplistic conceptual analysis. The mental states of interest are typed partly in terms of their connection with intentional action, but it also seems doubtful that we could make sense of the idea of intentional action in the absence of such states. It looks as if they all come as a package. I see no reason to think that any of these mental concepts should be reducible to any of the others. However, there also seems little reason to doubt that just as we can cast light on the dispositions of inanimate objects by 42
How Not to Naturalise Action investigating their structure, constitution, and so on these particular dispositions (desire, intention, belief) of objects of this particular kind (human agents), as well as their characteristic manifestation in goal-directed behaviour, will be illuminated by just the sort of work currently being undertaken in cognitive science and the philosophy of cognitive science. In particular, naturalistically-inclined work on mental content, action-guiding representations and the like look like much more promising avenues for understanding how intentional action and the mental states that it manifests fit into the explanatory order of the world than the causal theory of action does. I don t know whether we should expect intentional action to be reduced through such work, but we can certainly hope to learn a lot more about how it works than we ever did from the standard story. So, to conclude: naturalising action isn't as straightforward as the causalist envisaged but we can at least see a way forward. edgar.phillips.13@ucl.ac.uk REFERENCES Alvarez, M. (2009). Acting intentionally and acting for a reason. Inquiry, 52(3): 293 305. Alvarez, M. & Hyman, J. (1998). Agents and their actions. Philosophy, 73(2): 219 245. Davidson, D. (2001). Essays on Actions and Events (2nd ed.). Oxford: Oxford University Press. Holton, R. (2009). Willing, Wanting, Waiting. Oxford: Oxford University Press. Hornsby, J. (1997). Simple Mindedness: In Defense of Naive Naturalism in the Philosophy of Mind. Cambridge, Mass.: Harvard University Press. Hornsby, J. (2004). Agency and actions. In J. Hyman & H. Steward (eds.), Agency and Action (1 23). Cambridge: Cambridge University Press. Hyman, J. (2013). Desires, dispositions and deviant causal chains. Philosophy 89(1): 83 112. Kalderon, M. (2012). Experiential pluralism and the power of perception [PDF document]. Retrieved from: https://www.academia.edu/1495929/experiential_pluralism_and_the_power_of_p erception Mayr, E. (2011). Understanding Human Agency. Oxford: Oxford University Press. 43
Owen, D. W. D. (1980). Actions and bodily movements: Another move. Analysis 40(1):32 35. Sehon, S. R. (1997). Deviant causal chains and the irreducibility of teleological explanation. Pacific Philosophical Quarterly 78(2):195 211. Setiya, K. (2009). Reasons and causes. European Journal of Philosophy 19(1):129 157. Steward, H. (1997). The Ontology of Mind: Events, Processes, and States. Oxford: Clarendon Press. Velleman, J. D. (1992). What happens when someone acts? Mind 101(403):461 481. Wittgenstein, L. (1968). Philosophical Investigations (3rd ed. G. E. M. Anscombe, Trans.). Oxford: Basil Blackwell. Wittgenstein, L. (1969). The Blue and Brown Books: Preliminary Studies for the Philosophical Investigations (2nd ed.). Oxford: Basil Blackwell. 44