Causal theories of action

12. BEHAN MCCULLAGH CAUSAL THEORIES OF ACTION (Received 23 January, 1974) The causal theory of action which Professor Donald Davidson has elaborated over the last few years at present faces a kind of counterexample which he admits proves his theory to be inadequate. 1 Davidson has argued 2 that a person's behaviour is an action if (1) there is at least one description of it which corresponds to what the agent intended, and (2) his behaviour was caused by the wants and beliefs by which he would rationalize it ("in the sense that their propositional expressions put the action in a favorable light, provide an account of the reasons the agent had in acting, and allow us to reconstruct the intention with which he acted" 3). Examples can be constructed, however, in which these conditions are satisfied but the behaviour in question is dearly not an action of the person who performs it. Davidson offers the following: 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.4 The trouble in cases like this is that the person's behaviour is not caused by his beliefs and wants in the way actions normally are. In this example, the thought that he might easily rid himself of the weight and danger of the other m a n on the rope caused a state of shock or fear in the climber which caused him to loosen his hold. Davidson, like Professor A. I. Goldman before him, believes that for his theory of action to exclude such cases it must specify in more detail what the appropriate causal sequence should be, i.e. the second of the two conditions stated above must be amplified. Goldman has said that the 'characteristic way' in which wants and beliefs lead to intentional acts can be described only by specifying the neurophysiological processes it involves, since although this 'characteristic way' can be 'felt' it can not, in his opinion, be acPhilosophical Studies 27 (1975) 201-209. All Rights Reserved Copyright 9 1975 by D. Reidel Publishing Company, Dordrecht-Holland 202 C. BEHAN MCCULLAGH curately delineated in terms of conscious experiences. The description of these processes, he added, is a job for special scientists, not philosophers. 5 Davidson is inclined to agree with Goldman, but points out the psycho-physical laws of the sort Goldman envisaged would be impossible to formulate. 6 So it seems that his theory cannot be modified to exclude these counter-examples. I wish to suggest that a causal theory of action can be formulated which will exclude these abnormal, uncharacteristic cases; and that the new formulation will involve important modification of the first as well as the second of the conditions which Davidson stipulated for a person's behaviour to be called an action. If we accept Professor A. C. Danto's analysis of basic and non-basic actions, 7 then every action either is a basic action or has one or more basic actions among its components. I suggest as a start, that a piece of behaviour is a personal action if the following are true of what appears to be the basic action(s) involved in it: (1) at the time of doing it the agent wanted, intrinsically or extrinsically, 8 to do it; (2) there was no other psychological state present, for example a state of shock, sufficient in the circumstances to have caused the behaviour; and (3) there was no physiological state present, like that which causes a nervous twitch, besides those states related to the relevant want, sufficient in the circumstances to have caused the behaviour precisely as and when it occurredP In Davidson's example the act of letting go the rope was the apparent basic action, and it was a component of the apparent act of killing the climbing companion (assuming that he died as a result of the rope being released). The act of letting go the rope would indeed have been a personal action if the above three conditions had been true of it, and in that case the act of killing the climbing companion would have been a personal action, perhaps an unintentional one, too. In fact, however, the second condition was not satisfied, and there is considerable doubt whether the first was. The climber let go the rope because he was 'unnerved', presumably shocked. It is doubtful that he even wanted to let the rope go, for although he wanted "to rid himself of the weight and danger of holding another man on a rope," and knew "that by loosening his hold on the rope he could rid himself of the weight and danger," we are not told that he wanted to do what this want and belief imply, namely to C A U S A L T H E O R I E S OF A C T I O N 203 loosen his hold. On reflecting upon all the likely abhorrent consequences of such an action, he would probably have felt strongly averse to it, despite the immediate relief he could have expected from it. Surely it was the thought that he might thereby kill his companion that unnerved him. But even if he had wanted to do this basic action, this want would not by itself have been sufficient to ensure that the action's performance was a personal action, for it was possible that this want caused the action by a 'wayward causal chain', to use Davidson's phrase, perhaps by causing him to shake with fear of discovery and so inadvertently to release the rope. Or at that instant a falling rock hitting his hand, or an uncontrollable muscular spasm, might have caused him to let go. However, when all three conditions stated in the last paragraph are satisfied of what appears to be a basic action, one may call it, and any action of which it is a component, 1~ a personal action. But as they stand, these conditions are too restrictive. Might not the want to do something be followed by the formation of an intention to do it, or even by deliberation and a decision to do it, and these psychological states be the cause of appropriate behaviour? In such cases, assuming that intentions and decisions differ from wants, condition (2) would not be satisfied; but we do not want to deny that such behaviour could be personal action. Are intentions and decisions different from wants? Davidson is inclined to speak of an intention as that which a person wants to do in acting. 11 But Professor J. W. Meiland has pointed out sufficient differences between wants and intentions to justify us taking them to be different psychological states. 19 For instance one can want to do one thing at a certain time more than anything else, yet not intend to do it because one does not believe it possible to do it. So the first of our three conditions should be altered to admit intentions as possible causes of actions. Notice that, allowing for extrinsic wants, it is inconceivable that a person could intend to do something and not want to do it. 13 An intention is probably not a kind of want, but a decision apparently is a kind of intention - an intention formed as a result of deliberation. 14 Our theory will accommodate behaviour caused by intentions and decisions, therefore, if the first condition is made to read: (IA) at the time of doing it the agent wanted, intrinsically or extrinsically, to do it, and perhaps intended to do it. Then condition (2) will no longer exclude cases 204 r BEHAN MCCULLAGH which we want to call actions. Condition (3) must be slightly ammended to mention intentions and well as wants: (3A) there was no physiological state present, besides those related to the relevant want and intention, sufficient in the circumstances to have caused the behaviour. It might be thought that the example which Davidson finds embarrassing could be accommodated another way. Some might wish to argue that the climber's state of shock was not a separate psychological event sufficient in the circumstances to have caused him to let go, but rather a state which was manifest in the letting go. If this were so, condition (2) in our theory would not suffice to exclude this case from the class of personal actions. Instead perhaps one should stipulate that not only is a personal action caused by wants and intentions but that it must be within the agent's control, so that if he had wished not to perform it at the time then he could have forborn. 15 Then the climber's letting go was not a personal action because it was an action beyond his control. But far from being at variance with our theory, the suggestion that for a person's action to be personal it must be under his control is both endorsed and clarified by it. For to say that had a person wanted not to do something he could have forborn is to say that at the time there were not any psychological or physiological states present, besides his desire and perhaps intention to do it, sufficient in the circumstances to have made him do it had that desire and intention been absent. So those cases of uncontrolled behaviour which we would want to deny are personal actions are ones in which either condition (2) or condition (3A), or both, are not satisfied. Note, however, that there is a kind of uncontrolled behaviour which we are content to call personal behaviour when the three conditions (1A), (2) and (3A) apply, namely behaviour caused by uncontrollable wants. Those who believe that psychological states such as shock, fear and anger do not cause uncontrolled behaviour but are merely manifest in it will not be happy with our analysis of control in this context, nor will they be easy to convince. The debate as to whether human dispositions can be causes, in particular whether reasons, like wants, intentions, and beliefs can be causes, has run a long time. There is no denying that all dispositions are manifest in any behaviour they are said to cause, but the causal claim seems to be justified in as much as they are identifiable C A U S A L T H E O R I E S OF A C T I O N 205 psychological states preceding that behaviour, but for which the behaviour would not in the circumstances have occurred. Davidson and Goldman believed that the only way their causal theories of action could be amended to exclude 'wayward causal chains' was by specifying in more detail the normal way in which wants cause actions. In fact, however, the difficult cases they mentioned can be excluded not by further specifying this causal connection, our condition (1A), but by requiring that alternative causes be excluded, our conditions (2) and (3A). This requirement accords with our intuitions. We feel confident that our wants and intentions caused our basic actions only when we know of no other state or event which would have caused them had our wants and intentions been absent. Our confidence in their causal efficacy does not rest upon detailed knowledge of how the connection was mediated. To know that our wants and intentions caused our actions, therefore, we need no law specifying in further detail the connection between them. We need only know the various sorts of psychological and physiological states which can suffice to cause behaviour of the kind in question, so that we can check that all but the appropriate wants and intentions and their corresponding physiological states are absent. Wants are only contingently, never universally, necessary for the movements they cause. If our judgements about whether a bodily movement is a personal action or not are made in this way, then they are defeasible, to adopt Professor H. L. A. Hart's term. 16 The presence of appropriate wants and beliefs provides primafacie reason for calling a movement an action; but the claim that it is can be defeated by showing other exceptional psychological or physiological causes were present. It is unlikely that we can list all the kinds of alternative psychological and physiological states sufficient to produce any particular kind of behaviour, even though we may able to recognize their presence easily enough. The concepts we have of these psychological and physiological states at present permit only vague discrimination. And while we cannot list all the alternative possible causes, we cannot state the necessary and sufficient conditions for wants and intentions causing movements. As things stand we may think we have observed no alternative causes of a movement to be present but be proved wrong by someone who knows more of them we do. For example, I might think I voluntarily withdrew my hand from a painful 206 c. BEHAN MCCULLAGH electric current, only to be told by a physiologist that its withdrawal was an automatic physiological response. If Davidson's theory of action is to be amended to exclude the unwanted instances we have been discussing, his second condition, that the behaviour must have been caused by the wants and beliefs which rationalize it, must be replaced by the three we proposed. Once these have been adopted, his first condition, that for a piece of behaviour to be a personal action there must be at least one description of it which corresponds to what the agent intended, is found to be misleading. It is not enough that any description of an action correspond to the agent's desire in acting. For unless the basic action were caused by a desire for it, any non-basic action of which it was a component would not necessarily be a personal action, even though, as in Davidson's climbing example, there was a description of the non-basic action (being rid of the weight and danger of holding his companion on the rope) which the agent wanted to be true, and this want caused it to become true. Only when the basic action involved in a non-basic (in this case what Danto calls a 'mediated') action is wanted and caused by that want, is the non-basic action personal. And if the basic action is personal, the non-basic actions it mediates are personal too, whether intended or not. One can usefully distinguish actions, personal actions and intentional actions. Personal actions are a sub-class of actions, and intentional actions a sub-class of personal ones. An action in general is the bringing about of a change of state in one thing by another. There is more to it than that, for as D. Hirschmann has shown one cannot mention just any cause as the agent of a change. 17 But inanimate as well as animate objects can do things: stones can break windows, and dogs can bite postmen. And I would add that human limbs and organs can do things too: fingers can tap, teeth chatter and eyes blink without the person concerned having wanted them to or even been aware that they were doing so. Indeed a whole human body, if tackled onto it in rugby, can smash a corner-post without that person having done anything. This is one reason for preferring to speak of personal rather than human action. Another is that the word 'personal' reminds us that the actions which people, as opposed to just their bodies, do are the bringing about of changes by mental states or events, beliefs, wants and sometimes intentions (whose ontological interpretation can be left open). Now although C A U S A L T H E O R I E S OF A C T I O N 207 all changes mediated by personal basic actions may, strictly speaking, be called personal actions too, not all of them need be intentional. The intentional ones are only those which refer to consequences of the basic action which the agent wanted and expected or even just expected, and which were brought about in the manner the agent wanted or expected, is Basic actions are themselves intentional only if they have at least some of those features the agent wanted and expected them to have. This last point is at variance with Davidson's constant assumption that all actions caused by primary reasons are intentional under some description. Thumbsucking, for example, might be caused by a desire for oral satisfaction and the knowledge that thumb-sucking will provide it, yet on occasion be unnoticed by the agent and so be intentional under no description whatever. If this is true, then Davidson's first requirement is false. Movements can be actions though they be entirely unintentional. If Davidson's conception of a personal action was too broad, Professor A. C. Danto's is too narrow. Davidson thought that anything we do is our action if it is intentional under any one description; Danto prefers to say that only those things we do which we intend to do are our personal actions. "Either we do everything which can be truly said of what we do... or reference to our representations defines the limits of practical.., responsibility. I opt for the latter? u" This preference is reflected in his analysis of 'm makes happen the event a by doing a', where a is a mediated action, the first truth-condition of which CA-F) is that 'm intends that a happen'. 2~This means that to speak of a person's unintentional action is a contradiction in terms. But we all do speak of them, and even Danto himself does, writing of unintentionally killing someone by running him over in a car. 91 What seems to have misled Danto is the thought that the actions which people do must all be ones for which they can be held responsible. Notice how in the quotation he casually equates what people do with 'the limits of practical responsibility'. From his discussion it is clear that he identifies the area of practical responsibility with that of moral responsibility, for in support of his option he appeals to the apparent injustice of the Greek gods in punishing Atrius and Oedipus for the unintentional bad actions they did. Atrius unwittingly ate his children, and Oedipus unsuspectingly made love to his mother. It would certainly be unjust to blame people for all the unintended aspects and consequences of their 208 c. BEHAN MCCULLAGH actions. These are truly infinite. But this is n o reason for saying people do n o t perform u n i n t e n t i o n a l actions. It only points to the fact that they should n o t be held m o r a l l y responsible for them. But can we really say that people do actions referring to any consequence of their basic actions, n o matter how r e m o t e ? This seems to be as u n c o n v e n t i o n a l as D a n t o ' s claim that personal actions c a n n o t be u n i n t e n t i o n a l . Indeed it is, b u t the n u m b e r of actions we ascribe to people is n o t limited to those which are intentional. It is limited, rather, b y the significance of their basic action in causing the c o n s e q u e n t events. T h e less significant it is, the less inclined we are to refer to the c o n s e q u e n t event as something the agent did. There is n o clear line dividing the actions we do from their consequences; the difference is a m a t t e r of degree. L a Trobe University NOTES 1 'Freedom to Act', in Ted Honderich (ed.), Essays on Freedom o f Action, Londons 1973, pp. 153-155. 2 Especially in 'Actions, Reasons and Causes', The Journal of Philosophy LX (1963), and in 'Agency', in R. Binkley et al. (eds.), Agent, Action, and Reason, Oxford, 1971. 3 Honderieh, op. eit., p. 147. 4 Ibid., pp. 153-154. Another example is given by A. I. Goldman, A Theory of Human Action, Englewood Cliffs, 1970, pp. 60-61. 5 Goldman, op. eit., pp. 62-63. 6 Honderich, op. cit., pp. 154-155. 7 See especially Analytical Philosophy of Action, Cambridge, 1973, pp. 28-30. 8 ,,... x wants 0 (purely) extrinsically if and only if his sole reason for wanting it is that he believes realizing it is necessary to (will, may) realize something else that he wants; whereas x wants 0 (purely) intrinsicallyif and only if he wants it solely 'for its own sake'..." Robert Audi, 'Intending', The Journal of PhilosophyLXX (1973), 389. 9 These are not sufficient conditions for predicting that a person will do what he wants. For prediction one must also know that he believes he can do it, that he actually can do it, and that he wants to do nothing else more strongly at the time. But given that something has been done, these eonditions are sufficient to justify its being called a personal action. xo Of course eomposite actions, as Danto defines them (op. eit., p. 29), those "consisting of basic actions performed in sequence or together, as in a dance", are only personal if all the basic actions which constitute them are. 11 "For example, suppose a man saws a piano in half because he wants to throw the piano out of the window and believes that sawing the piano in half will promote his enterprise... We can see that the intention with which he sawed the piano in half was to get the piano out the window." Honderich, op. cit., pp. 147-148. 12 The Nature of Intention, London, 1970, Chapter 6. C A U S A L THEORIES OF A C T I O N 209 a Robert Audi, in the article cited above, argues convincingly that intending entails wanting, though he is unable to prove the point. See pp. 389-391. 14 See Meiland, op. cit., Chapter 4. 15 See Danto's discussion of forbearing, op. cit., pp. 164ff. 16 In 'The Ascription of Responsibility and Rights', Proc. Arist. Soc. 49 (1948-49). On whether ascription of agency is ascription of responsibility, see below. 17 'Inanimate Agency', Proc. Arist. Soc. 72 (1971-72). 18 Danto, op. cit., pp. 24-25 and 12-13, argues that mediated actions are intentional only when the events caused by the basic action are not caused accidentally or fortuitously, though he adds " I propose no analysis of chance occurrences here". However it is not their fortuitousness but their unexpectedness which makes certain mediated actions unintentional; for in a way every causal sequence is fortuitous. 19 1bid., p. 21. 20 1bid., pp. 7-8. 21 Ibid., p. 24.