METAPHILOSOPHY Vol. 4, No. 1, January 1973

‘TWO-DOMAINISM AND HUMAN FREEDOM

ROBERT YOUNG

In his book Free Action (London, 1961), A. -- I. Melden argues

. . . logical incoherence involved in the supposition that actions, desires, intentioas, etc., stand in causal relations, either in the Humean sense or in any sense in which the term ‘causal’ is employed in the natural sciences . . . determinism, if it employs this sense of cause, is not false but radically coldusled. So it is with indeterminism and libertarianism which grant to determinism the intelligibility of employing the causal model. . . . Given this fatal blunder, actions degenerate into mere bodily happenings . . . (p. 201). It is evident that the underlying claim here (and elsewhere in

the book) is not a version of libertarianism, the doctrine that, as a matter of contingent fact, deterministic accounts of action cannot be provided for at least some actions. Rather it is the stronger claim, which I will label (D) for convenience :

(D) it is conceptually, or, in principle, impossible to provide deterministic explanations of action. That is, given that ‘there is a deterministic account of x’ and

that ‘y is an action’, it is held to be conaeptually impossible for x to be identical with y.

In this paper my purpose is to consider whether there are any consequences for the problem of freedom and determinism which stem from the views of an influential group of writer’s on the philosophy of a d o n (of whom Mielden is a leading representative). I t is perhaps going too far to ‘speak of this group as a ‘school’, but its members display a clear community of me+Aodological and doctrinal aharacteristics. Foremos’t among the matters in which they ape agreed for the purposes of this paper, is their support for principle (D). I t will be convenient to label these writers ‘two-domainists’. This is probably the least misleading of the epithets applied to them in the literature. Its reasonableness is evident when onle notes that these writers characteristically distinguish between persons as physiological organisms and as beings who act (agents). Persons as ageats are

23

that there is a

24 ROBERT YOUNG

creatures to whom it is essential that teleological concepts apply and these concepts are not reducible, either logically or empirically, to physicalistic discourse anld modes of explanation. The methods of explanation and the concepts employed in these two domains ar8e incompatible. Their opponents, of whom I am one, I shall call ‘causlal theorists’. ‘Causal theorists’ stress the essential identity in method and modes of explanation between the physical sciences and the social.

Therie may well be readers who consider that there is by now such an imposing array of counters to the views of ‘two- domainists’ as to render this paper superfluous. The justification I would claim for the paper in the face of such a consideration is twofold. First, even though some of what I have to say (in section I1 and I11 particularly) has been said elsewhere, no-one has attempited comprehensively to refute the ‘two-domainist’ case. And that case is wide-ranging. In those places at which I rehearse counters employed by other writers I do so usually because of the power of their counters and always because I see no point in thnowing away ladders one has climbed.

The second justification is more important. Support for ‘two- domainism’ remains at a very high level in the literature. Some writers are so casual about appealing to the tenets of ‘two- domainism’ as to make one woinder whethler they have ever read any counter-arguments let alone been persuaded by them. I cibe just two examples in very recent blooks. Both R. S. Downie in his Roles and Vdues (London, 1971) and W, D. Hudson in his Modern Moral Philosophy (New York and London, 1970) appear not to recognize what I have called the imposing array of counter-arguments to ‘twodlomainism’. The manner of $heir appeal to ‘twodomainism’ as the resolution of the problem of freedom and determinism certainly conveys that impression. The fact that Hudson’s book, in particular, will be widely used provides a justification then, I would urge, for again showing that ‘twodomainism’ does not dispense with that hoary problem.

The paper has the following structure. In section I, I seek briefly to sketch ce&ain points of agreement about human action as between ‘two-domainists’ and ’causal theorists’. The aim is to furnish a helpful context for consideration of the matters they dispute and to throw the disputed claims into clearer relief. In sections 11-IV I formulate and consider the arguments advanced by ‘two-domainists’ in support of (D). It remains open, of course, that some additional and more success- ful arguments might a t some future time be used to establish

‘T WO-DOMAINISM’ AND HUMAN FREEDOM 25 (D). In addition, therefore, to attacking the extant arguments of ‘two-domainists’, I suggest Wherever possible positive argu- ments in favour of a deterministic theory of action. Finally, in section V, I try ta show that a successful defence of (D) would not reveal that the doctrine of determinism can have no bearing on freedom of action. That is, that it may be an open question whether the doctrine of determinism as applied to human decisions and actions is true or false, but not whether the doctrine in such a context is radically confused.

I I t is generally agreed among partiles to the controversy about

action that there is an intuitive distinction to be made between cases of genuine action and mere happenings. The existence of such a distinction does not preclude the existence of borderline cases which pose difficulties in classification. There is no doubt that one aim in developing explicit theories for handling the easily classifiable cases has been the diesine to go on to explain the more difficult ones as economically as possible. This pro- cedure might be called ‘the spoils to the victor’ approach. There has been, therefore, rough agreement on the pre-analytic force of the distinction mentioned and a degree of procedural agree- ment. The disagreement which remains, concerns the nature of the characteristics possessed by all those things which, on intuitive grounds, unhesitatingly qualify as actions, but which are missing in mere happenings.

Furthermore, the ‘two-domainists’ and their opponents agree in rejeicting several attempts at characterizing the distinction. There are three such attempts to which I wish to draw attention. Firstly, it is agreed that action cannot be distinguished from mere happenings by equating aation with bodily behavimour simpliciter1 or, secondly, with bodily behaviour plus a non- causally re’lated concurrent mental event. In neither case is there a sufficient condition for action. As regards the formier, persons undergo twitches and nervous spasms, can have their limbs moved by machines or other persons, can perhaps be sraid to act by refraining from bodily movement or by emitting to make some movement. An appeal to some such idea as ‘negative behaviour’ in these latter instances would involve one in the

11 have added this qualification because some of the ‘two-domainists’ claim the identification is, on occasion, possible provided we take into account the context of rules and practices in which the movement occurs. More of this in section W .

26 ROBERT YOUNG

difficulty of explicating that idea as well and might require us to say (absurdly) that people are always acting. In none of the situa- tions referred to a moment ago is bodily movement a sufficient condition for acting. As regards the second characterization, it is likewise not the case that the occurrence of the behaviour and the mental event (such as a pro-attitude) is logically sufficient for the performance of the action. A person might have the strongest relevant pro-attitude toward kicking a certain doctor in the shin, and actually do so as a consequence of the doctor’s tapping the person’s knee during a medical examination. (The kick results from a. reflex movement of the patient’s foot. Such a reflex movement, even accompanied by the mental state, is not sufficient to characterize action as against a mere happening.)

The third characterization which ‘two-domainists’ and their oppmmts are united in rejecting is the volitional theory. Granted that neither of the charaicterizationis considered above picks out actions, it might be contended that what is essential to the contrast between actions and mere happenings is the uoZunta$ nature of action. Advocates of the volitional theory have con- tended this and claimed that a bit of behaviour is voluntary, if, and only if, i t is caused by the occurrence of a special sort of event, namely, a volition. Thus the defining feature of an action is that it is a voluntary event and what makes an event volun- tary is that it is caused by a person’s act of will or volition. The term ‘volition’ is explained by reference to such concepts as deliberation, choice, effort and trying.

I regard the rejection of the volitional theory as an area of agreement in a qualified way. Even though opponents of two- domainism have not been as adamant in their rejection of the volitional theory, they have been loath to accept it, albeit for different reasons. ‘Two-dlomainists’ object to it in principle. ‘Causal theorists’ in general wish to eliminate volitions in favour of wants, beliefs and so on. The latter have been more sympa- thetic to the theory perh’aps because many woul’d say that ‘A did x (intentionally)’ entails ‘A trield to do x’, whlere the ‘trying’ caused the doing of x.

With this qualification let us reconsider why the volitional theory has been rejected. A twofold attack has been waged against the volitional theory. I t has, firstly, been attacked on

2According to Ryle, The Concept of Mind (London 1949), pp. 67ff, this sense of ‘voluntary’ is not the normal one but a device of philosophers. For reasons that will emerge I will substitute the tam ‘free’ for his normal sense.

‘TWO-DOMAINISM’ AND HUMAN FREEDOM 27 empirical grounds. The contention here is that people some- times act without any introspectable antecedlent deliberation, choice or effort. I am not particularly happy with some argu- ments offered in support of this contention even though I think the contention itself is very likely well-directed. Thus when Richard Taylor argues along the following lines in Action and Purpose he is less than c~nvincing.~ Hle claims that if I experi- ence something, then I will be immediately aware of it; and if I am not so aware of it, then I haven’t experienced it. But this seems dubious for it is a t least plausible that a person can experienue wm&ling without .being immedia\tely aware, or aware at all, that he is experiencing it. Volitions might be more like unrecognized desires than like pains. The conflicting testi- mony from introspective evidence certainly ensures that the empirical attack is not of ‘knock-down’ force.

The second part of the attack consists in Rxlgs claim that the volitional theory lea& to an infinite regress and a vicious one a t that. Suppose that there are volitions which cause bodily and mental behaviour. Then either these volitions are caused or they are not. If they are themselves caused the question arises concerning the cause of the volition’s cause and so on ad infiniturn. If volitions are, however, not themselves caused but are uncaused causes, it seems they are just things that persons make happen. But this won’t do for they were introduced in order to avoid saying that in acting we just make something happen. Thus either the problem has just been shunted back one step (and hence the analysis is inadequate) or a new mystery just as perlexing as the notioa of acting has been introduced.

The conclusion of the objection is that if volitions are the causes of (voluntary) actions then either a vicious regress results, or, no advance is made in the analysis of human action.

But there seems to be good reason for rejecting the dilemma by breaking the first horn (the regress). A non-vicious regress can be shown to result from supposing that every volition is caused. What distinguishes action from mere behaviour, if one takes this line, is that the causal chain aissociated with acting includes a volition, whereas the causal chain associated with mere behaviour does not. In short, even if every event, including volitions, is caused, this is still no reason for claiming that a person would have to do an infinite number of things in acting voluntarily. Provided the causal chain incorporates the volition

3Englewood Cliffs, New Jersey, 1966 pp. 66ff. A similar argument is employed by Melden. o p . cit., pp. 47ff.

28 ROBERT YOUNG

the behaviour is voluntary action. This is an appropriate place to point out that since wants are not acts, the regress objection does not even get off the ground if volitions are replaced by wants.

I have already noted in passing that a distinction should be drawn between voluntary action (in the sense used by adherents to the volitional theory) and free action. It will become clear later in the paper that it is free action that neecds to be con- sidered in relation to determinism, though voluntariness is a necessary condition of freedom of action. The volitional theory, whether or not it is adequate, should be viewed as an attempt to account for the distinction between (voluntary) action and mere happenings-not as an attempt to give the final word on the problem of freedom and determinism.

We have now exhausted the agreements and can look a t the substantial disagreements between ‘two-domainists’ and ‘causaI theorists’. ‘Two-domainists’ contend that what is essential to human action cannot be discovered by revising the volitional theory (or its like), for it is so radically misconceived as to lead one into a philosophical cul-de-sac. We ‘understand’ human actimon by appeal to the logical connecti’ons that such action has with the complex of deslires, intentions, reasons, conventions and rules that are inextricably involved in social life. The under- standing which deterministic or causal explanation can offer is only relevant to bodily behavior or movement, not action. The attempt by the volitional theory (or any causal theory) to replace the former by the latter is a conceptual error doomed a priori to failure. The volitional theory reliies aIlegedly on the mistaken notion that motives, desires, intentions, and so on are interior events, processes or states of the person. The having of intentions (and other teleological features) is what differentiates actions. But this is by no means to revive the volitional theory with its talk of present mental causes, because intentions and so on are neither present, nor mental,4 nor are they the causes of action.

‘Causal thelorists’ on the other hand deny that the voldtilonal theory must be abandoned as wrongheaded even though they agree that it is itself mistaken.

11 I will begin now to formulate some of the claims made by

‘two-domainists’ and denied by ‘causal theorists’. These claims 4May Brodbeck, “Meaning and Action”, Philosophy of Science, 1963, pp. 309-24,

has remarked on the ‘incongruity’ of this claim. See further p. 311.

‘TWO-DOMAINISM’ AND HUMAN FREEDOM 29 and their supporting arguments constitute the bulk of the defence of principle (D). Attacks on them are, therefore, attacks on supporting evidence for (D) and in turn on (D) itself.

Charles Taylo? has written :

The element of ‘purposiveness’ in a given system, the inherent tendency toward a certain end, which is conveyed by saying that the events happen ‘for the sake of‘ the end, cannot be identified as a special entity which directs the behaviour from within, but consists rather in the fact that in beings with a purpose an event’s being required for a given end is a sufficient condition of its occurrence.

The sentiment here and elsewhere is that :

I. The form of a teleological explanation of behaviour (one in terms of purposes, or, in the case of human beings, intentions) is not logically reducible to the form of a causal explanation.

At the same location, Taylor argues that the distinctively teleological aspect of action comes out in the following. Assume an agent A and assume that B is a piece of his behaviour to be explained. Then if A has the goal G, and the state S of A and the environment E of A is such that G is achieved only if B occurs, then B occurs.

It has been pointed out firstly that thlis does not seem $0 be correct if only for the reason that it is not the fact that G occurs only if B occurs which makes A do B, but rather the fact that A believes that G occurs only if B 0ccurs.G

Furthermore, what A does is not merely a function of his goals but of which goals he wants to satisfy most in the situation. If we bear that in mind we need to revise the formulation given above and we sbould substitute for it something like this: if in environment E and in state S, G is the goal most desired by S and S believes that doing B is the only way to get S, then S does B. But it is immediately obvious that this formulation could be accepted by a ‘causal theorist’. Indeed it will become evident as

5The Explanation of Behaviour (London, 1964), p. 10. Unlike Melden, op. cit., pp. 105ff and R. S. Pegters, The Concept of Motiucrtion (London 1958) pp. 27ff. Taylor considers the choice between teleolcngical and causal or physicalistic explana- tions is not just a conceptual issue but depends on both empirical investigations and analysis.

SCf. C. Landesman, “The New Dualism in the Philosophy of Mind’’, Review of Metaphysics, 1965 pp. 329-45 (esp. pp. 336ff).

30 ROBERT YOUNG

we proceed that this sort of formulation is characteristically adolpted by ‘causal theorists’. It might be objected, though, that this revised formulztion is not applicable to those systems to which we cannot justifiably ascribe beliefs and pro-attitudes which, neverthelless, appear to be teleological. Even if i t were not plausible $0 suggest that we could reduce apparently tele- ological to non-teleological explanations, that is, to ones making no mention of purposes, the objection could bring little comfort to quite a few of the ‘twodomainists’. Some of them claim that teleological concepts apply to non-humans (perhaps excepting the higher animals) only analogically. Peters’, S. Hampshire8 andzLW%hg all stress that intentional action is $he province of those animals which can express intentions linguistically or which can follow rules. This seems far less plausible than the possibility of reduaing the apparently teleological to non- teleological explanations.

Perhaps ‘twodomainists’ might well concede under such pressure that the revised formulation is compatible with its being the premise of an ordinary causal explanation. This would not Rquire their acceding to the further claim that the substance of the formulation is compatible with its being the premise of an ordinary causal explanation.

I11 In line with this last remark it hasl been claimed that:

11. The intentions and desires which bring about behaviour are not its causal antecedentsc They may be reasons for actions but not causes.

This particular thesis has’ been defended by a number of crucial supporting arguments. Indeed the bulk of the ‘two- d~o~mainist’ case rests on the seven arguments which I shalI consider in this section. The counter-attack against several of these arguments is formidable and I will draw on that attack in contending that reasons for actiions can be causes.

Some examples of the opposing positions follow briefly. Melden has written in Free Action :

I have contended nlot only that the intention cannot function as a Humean cause of the action of raising the arm, on the ground that if it did it could not possibly explain the action

70p. cit., P. 98. Whoupkt and Action (London, 1960) p. 98. 9The Zdea of u Social Science (London, 1958) p. 52.

‘TWO-DOMAINISM’ AND HUMAN FREEDOM 31 in the sense in which actions are explained by intentions, but also that the intention cannot be identified with any of the items that cross the agent’s mind during the incident (p. 93).

Gharks-Xagkr extends the claim when he says in The

We could not say that the intention was the causal antecedent of the behaviour. For the two are not contingently connected in the normal way. We are not explaining the behaviour by the ‘law’, other things being equal, intending X is followed by doing X, for this is part of what we mean by ‘intending X’, that, in the absence of interfefing factors, it is followed by doing X. I could not be said to intend X if, even with no obstacles or other countervailing factors, I still didn’t do it. Thus my intention is not a causal antecedent of my behaviour (p. 33).

On the other hanjd Donald DaYidsonl’ has argued that the following are both necessary conditions for defining the relation of reasons to the actions they explain:

C,. R. is a primary reason why an agent performed the action A under the description d only if R consists of a pro-attitude of the agent toward actions with a certain property, and a belief of the agent that A, under the description d, has that property. C,. A primary reason for an action is its cause.

Before looking at the seven arguments advanced in s u p r t of 11, it may be helpful to attend to some prima facie evidence about the sort of explanation we are giving when we explain what an agent does by referring to his reasons (pro-attitudes and beliefs). It seems plausible to claim that such explanations are prima facie causal explanations. First, idioms which are used elsewhere of causal relationships are often used in speaking of an agent’s reason for action. For example, the use of ‘because’ and ‘caused’ is commonplace.

Second, by altering what a person wants or believes, or by getting him to adopt new wants or beliefs, it is often possiblle to influenae his action and behavior. Consider the case of a robber

Explarxztion of Behaviour :

lo“Actions, Reasons and Causes”, ]ournal of Philosophy, 1963 pp. 685-700. With this article one should compare W. D. Gean, “Reasons and Causes”, Review of Mefa- physics, 1966 pp. 667-88 and R. Brandt and J. Kim, “Wants as Explanations of Action”. Journal of Philosophy, 1963 pp. 425-35 for sihilar views.

32 ROBERT YOUNG

who convinces his victim that he should hand over his wallet rather than suffer physical violence. I would claim that such behaviour is voluntary action though it is not free action. A number of writers on causation have suggested that causal relations are often marked out by the fact that some factors can be oontirolled by manipulating others. William Dray, for instance, claims in his book Laws and ExpEanation in History (Oxford, 1957) that we cannot consistently say “that x does not cause y though by manipulating x we can control y” (p. 94). If this plawible point goles through, support is thereby given to the view that there is a causal relationship between wants and bdiefs and the actions they explain.

Third, as Gean has pointed out in the papm previously cited, statements of an agent’s reasons seem capable of supporting oounterfactuals such as “If he had nlot knlown (or believed, or wanted etc.) that x, ceteris paribus, he would not have done y”. The ability to support counteirfactua1,s parallels standard causal explanations.

Even though a prima facie case can be made out along the Iines considered, we cannot regard the relationship between (primary) reasons and the actions they explain as causal unless it is possible to defeat the arguments we will now scrutinize.

11.1. It is argued that the factors appealed to in the claim that a (primary) reason for an action is its cause, namely beliefs and pro-attitudes, are not events and therefore cannot be causes (since they age not of the right logical

One preliminary p i n t is that there is an ambiguity connected with sipeaking of believing and wanting. We may be speaking of what is believed or wanted. In this sense there is no question that these are not causes. Alternatively we may be speaking of the believing or wanting. ‘Causal theorists’ are interested in this latter sense. Strictly then, the question is whether in the latter sense we are dealing with the causes of a person’s voluntary actions.

Let us first allow that only events may be properly spoken of as causes. Even so the argument is unconvincing. We oould with Davidson point out that:

States and dislpositions are not events, but the onslaught of a state or disposition is. A desire to hurt your feelings may spring up at the moment you anger me; I may start wanting

type).

‘TWO-DOMAINISM’ AND HUMAN FREEDOM 33 to eat a mlelon just when I see one; and beliefs may begin a t the moment we notice, perceive, learn or remember some- thing.ll

T’hat is, we may say that wanting and belicewing are not them- selves events, but the coming into such a state is an event and thus a change of antecedent conditions. Davidsp claims that very often (maybe even typically) we are no€% possession of knowledge of the event even though we are sure there was one. Our ignorance of the event or sequence of events does not make us prepared to conclude that there was no such event.

We could on the other hand take a somewhat different tack while still assuming that only events can be causes. We might urge that the having of wants and beliefs constitutes part of the antecedent causal conditions and &us. part of the causally determining set of factors associated with action. This, of course, raises the tricky and controversial question of causes versus causal conditions-a matter of contention since Mill.

But we could mount a second objection to this argument by challenging the contention that only events can be causes. We frequently refer to states, standing conditions and even the failure of events to occur, as causes. Why could the having of intentions, for instance, not bei considered part of the causal conditions or causal factors? This seems reasonable when we recognize that the having of an intention consists in having (primary) reasons for action which become thte reasons why the action is done just in case it gets done. Describing a person’s intentions requires making reference to the desires and beliefs actually (causally) effective in him.

This first argument seems unsuccessful.

11.2 It is also argued that explanations in berms of (primary) reasons cannot be causal because there is a logical connection between the factors of wants and beliefs and th’e action to be explained,. Since a causal relation- ship is a contingent one, explanations in terms of (primary) reasons cannot be causal.

This is one variant of the so-called ‘argument from distinct existences’. I noted above that a statement of intention is an expression of a perslon’s effective desires and beliefs. It may be that recognition of this has led some philosophers to assert the

llop cit., p. 694.

C

34 ROBERT YOUNG

non-contingency of the connection between intentions and action.

Melden argues in Free Action : If the relation were causal, the wanting to do would be, indeed it must be, describable independently of any reference to the doing. But it is logically essential to tbe wanting that i t is the wanting to do something of the required sort with the thing one has. Hence the relation between the wanting to do and the doing cannot be a causal one (p. 128). M. F. Cohen12 claims, . . . the motive alnd bellief taken together determine in an a priori fashion what action can be said to follow from the desire. Hence the connection between the motive explanation and the statement of the action which it explains is not empirical but analytic, and the motive explanation fails to satisfy the contingency principle of empirical causality.

It might be well to point out that this argument does not really depend on a Humean analysis of causation as is often stated, but upon a contingent analysis of causation.

There are some very serious worries about this argument. Firstly, we may surely characterize a cause in such a way that it is logically and not merely contingently connected wi@h its effect. As Medlin13 has pointed out there is, for example, a logical link between a child and its parents in that a person could not be a parent unless he or she had had a child and also that a person could not be a child unless it had had parents. Yet no-one is going to deny that there is a causal relation between parent and child.

Secondly, there is an oddity in the idea that causal relations are empirical rather than logical. Davidson14 ’has noted that such a claim surely could not mean that every true causal statement is empirical. For suppose it is true that ‘A caused B’, it follows then that the cause of B=A and by substitution we have the analytic statement ‘the cause of B caused B’. The truth of a causal proposition in other words is dependent on what events are described, the analytic or synthetic status of the proposition on how the events are described.

l%“Motives, Causal Necessity and Moral Accountability”, Australasian Journal of Philosophy, 1964 pp. 322-34.

13Brian Medlin, “Materialism and the Argument from Distinct Existences”, in J. J. MacIntosh and S. C. Cuval (eds.), The Business of Reason (London, 1969) pp. 168-85.

14op cit., pp. 695f.

‘TWO-DOMAINISM’ AND HUMAN FREEDOM 35 Nevertheless, it might well be maintained, as DayjdzKhim-

self recognizes, that a reason explains an action only when the descriptions are appropriately fixed, and the appropriate descrip- tions are not logically independent. Clearly there would be a logical connection between reason and action if we supposed that to say a man wanted to do x meant that he would perform any action he believed would achieve x. But this would trivialize the ability of (primary) reasons to explain actions. There is no logical OY factual difficulty in supposing that a person might, for example, want to eat, know that his wife has put the food he sees on the table there for him to eat and yet not eat. Nor will it do to say that for one to want to do something means that he will at least try to do it unless he wants to do something else molre. As a definition of ‘wanting’ such a proposal would be circular. Wanting requires a t most a dispositional analysis involving the idea of the thing wanted.

Furthermore, if central state materialism is true there are strong grounds for believing that such an analysis will depend in turn on talk of states. Whether That theory is true or not, enough damage has already been done to the argument under consideration to undermine its support for 11. There is one final criticism I should raise.

Even if one admitted a logical connection between the concepts of, say, wanting and doing, one could justifiably contend that the possibility of causal explanations of actions requires only that the events in question are contingently connected. If it is objected that warnings are not events and so cannot be causes of actions we have just come full circle back to argument I.

Even though the ‘argument from distinct existences’ seems to me to fail there is perhaps a modicum of truth contained within it. At least some philosophers have thought that it must be possible to characterize the cause truly in such a way that, under that description, it is only contingently connected with its effect.15 But, whether or not this is so, it by no means serves to establish 11.2.

11.3 A closely related argument is th’e following. A statement of what the agent wants and believes does not explain by telling us how he comes to act as he does, but instead i t explains by enabling us to more fully charac-

l5David Pears argues in “Are Reasons for Actions Causes?” in A. Stroll (ed.) Epistemology (New York, 1967), that there is a counter to this claim. The counter comes from those fairy stories which treat wishes as causes and describe a wish simply as concentrated willing.

36 ROBERT YOUNG

tarize the act; i t tells us what kind of act is done. But if this is so we do not have two logically separate factors (the cause and the effect) but only one under two different descriptions.

Thus in Free Action Meldcen .- asserts:

In any simple causal explanation of one event by reference to another, it is not the identity or the character of the effect that is a t issue, but the conditions in which it occurs-how it came to be. . . . As the alleged cause of the action, it cannot serve further to characterize the action. As motive it must- for it tellls us what ‘in fact the person was doing (p. 88). Let us allow for the sake of argument that there is a type of

explanation which consists in saying what (kind of) act the agent is performing. Melden’s own example of a driver who signals by raising his arm, illustrates this type of explanation. In answer $0 the question “Why did he put his arm out like that?” we might reply, “To signal a turn”. His raising his arm is, in the circumstances, signalling. I t is dear that this sort of explanation is not causal explanation. But, even so, several objections can be levelled at this argument.’6

Firstly, Melden’s example is SIO well-chosen as to hide thfe fact that there are probably only a few special cases which fit the pattern outlined. There seem to be other cases in which stating one’s reasons does not enable anyone to say more fully what kind of action one’s action is (e.g. picking up my umbrella as I go out in order to avoid getting wet because I believe it will rain).

Secondly, even if we settlle the question about d a t sort of action an agent has performled we have not ruled out thereby, questions about why an agent has done such a thing. We can, for instance, ask why an agent signalled his turn. Just saying that his act was an act of signalling tells us nothing about why hce signalled. That seems to be a matter of interest, too!

Thirdly, saying that a man is signalling a turn by means of extending his arm, probably depends on the possibility of explanation by reasons. If he did not want to signal or did not know (believe) that extending one’s arm was the means of signalling, it is plausible to doubt that he signals at all.

Hence, if explanation in terms of primary reasons is not reducible to mere specification of the kind of act being per-

16The following appraisal of the argument owes much to Gean, op. cit., pp. 679ff.

‘TWO-DOMAINISM’ AND HUMAN FREEDOM 37 formed, then no grlound has been offered in my view for claiming the presence of a Icogical (and non-causal) relation between ‘he raised his arm’ and ‘he wanted to . . . and believed . . . ’.

11.4. It has been charged that explanations in terms of (primary) reasons cannot be causal since they fail to imply generalizations to the effect that in relevantly similar circumstances, the same result will occur. That is, the covering-law analysis of explanation adopted in the physical sciences is said to be inappropriate in the social sciences.

Some examples. M. F. Cohen argues, in the article previously

. . . in neither first nor third person statements does a motive explanation of an action rest upon an empirical generalization about what the agent (or others) would do in like circum- stances when possessed of such a motive (p. 329). . Hart ---- and gonor6 in Causation in the Law (Oxford, 1959) also

The statement that one person did something because, for example, another threatened him, carries no implication or covert assertion that if the circumstances were repeated the same action would follow (p. 52). Covering-law theorists assert that a satisfactory explanation

meets the formal condition that the explanandum is logically deducible from the explanans, wbere the explanans incorporates a general law and a statement of initial conditions and from these initial conditions the explanandum follows invariably (or usually) in accordance with the law. I have included the weaker sense here because some covering-law theorists allow non- deductive probability relations between explanans and explanan- dum. In addition to the formal condition mentioned previously, there is also an epistemological condition implicit in the model because of its talk of law or regularity.

Once again Davidson’s pioneering article (previously cited3 has made the task of replying much easier. He has pointed out that ignorance of the sort of laws required by covering-law theorists does not in fact inhibit valid causal explanation.

Provided we know directly which f a c t d s ) cause an event we have adequate evidence for the existence of a causal law cover- ing the case. He goes on to p i n t out that we are usually far

. * - - cited, that:

allege that:

38 ROBERT YOUNG

more certain of a singular causal connection than we are of any causal law governing the case. Consider: (a) ‘A caused B’ entails some particular law involving the predi-

(b) ‘A caused B’ entails that there exists a causal law instantiated

Each of these formulations sustains causal explanations in terms of laws. But only the second, (b), fits most causal explana- tions and it fits explanations in terms of a man’s (primary) reasons equally well.

Perhaps an example will help clarify the point. I drop a vase kom a tall building and assert that dropping it caus’ed it to break. In so asserting I imply that given similar circumstances, the same result would occur. But if I were to drop an exactly similar vase under rtelevanfly similar circumstances and it failed to break, what then? Assuming causal statements do imply generalizations I could only say that there was some difference in the two cases. Why couldn’t this much be true of explanations of ‘human behavior? Why couldn’t it be the case that if the relevant psychological and physiological facts are not duplicated for an agent, then ceteris pribus that accounts for a difference in his behaviour? Especially since human beings are much more idiosyncratic things than vases.

11.5. I t is argued (e.g. by Hart __-- and Hcmap6) that the kind of knowledge one has of one’s own reasons in acting is incompatible with the existence of a causal relation between reasons and action. The privileged position one has in knowing one’s intentions does not come by induction or observation and these are the ways in which one comes to know about causal relations. Furthermore (and this ties in with argument 11.4 above), because of the agent’s conclusive knowledge of his own reasons, the way in whi& we support or refute reason-explanations shows that tbey do not imply generalizations. The only relevant generalizations are ones constructed out of instances in which we recognize that the agent was acting for a reason and this was known independently of such generalizations.

Four things can be said in reply. Firstly, even though an agent’s testimony is normally conclusive, it is also possible for

cates used in the dlescriptions ‘A’ and ‘B’

by some true dlescriptions of A and B.

‘TWO-DOMAINISM’ AND HUMAN FREEDOM 39 his testimony to be mistaken without this leading to doubt about his sincerity or memory. A person may be deceived about h’is own motives. The fact that a person may be deceived does not show that first person reports are not normally “conclusive” but it does show that we cannot analyze reasion-explanations or the generalizations which may be used to support them, in terms of first-person reports.

Secondly, even if we grant:

(a) that the agent’s testimony is conclusive in the required sense, and,

(b) that the only relevant generalizations are ones constructed out of instances in which we recognize that the agent was acting for a reason and this was known independently of such generalizations,

this still would not shlow that explanations in terms of reasons cannot be causal. Davidson,17 for instance, points out that it may only show that in thle case of knowledge of one’s reasons, we know that a causal relation exists on the basis of one instance. In such cases we can be said lm knlow that generalizations to the effect that in relevantly similar circumstances the same result will occur, are true on the basis of one instance.

Thirdly, there appear to be cases of straightforward causation in which one person is in a better position to know the cause of something’s happening than is anyone else, and in which his honest testimony is as conclusive as that of the agent in giving his reasons for action. Some writers have appealed to cases which are inconclusive because they involve the idea of ‘mental causation’. A better illustration might be found where, say, a student says “Dropping that chemical into the liquid made it explode” when he didn’t know the nature of the interacting chemicals but simply observed the two things come into contact and immediately explode.

Fourthly, there are cases in which we cannot ever rely upon the agent’s statement of his reasons at all, namely when we are concerned with higher animals (or young children) not able to use language to tell us what wants and beliefs moved them to act as they did. In such cashes we probzdbly appeal to generaliza-

17op cit., p. 699. See also R. Hancock, “Interpersonal and Physical Causation”, Philosophical Review 1962, pp. 369-76.

40 ROBERT YOUNG

tions about the species or to familiarity with a particular animal (or child). We do, in other words, seem to use generalizations to confirm and disconfirm explanations in terms of reasons and, furthermore, sudh generalizations cannot normally be reduced to instances in which the agent’s reason is determined independ- ently of the generalizations. One would need to argue for the existence of some differentiating factor between higher animals and humans to prevent this conclusion’s being extended to human action. Some ‘two-domainists’ are not loath to do this, though! (See the authars cited in notes (7)’ (8) and (9)J

11.6. In support of thesis 11, it is argued that the logic of the causal relation is different from that of the reason-action relation. The causal relatilon is transitive, but the latter is either intransitive or non-transitive.

This argument seems to be quite widespread but has not been oriticized by causal theorists to my knowledge. I will cite only one example and refer to several others. Alan White in his introduction to The1 Philosophy of Action (Oxford, 1968Fwrites :

. . . if my curiosity to see what would happen can be said to have m d me to press a button and pressing a button caused the death of some people in a room, thien my curiosity might be said to have mused their death; but though curiosity to see what happened was my reason for pressing the button, it was not my reason for killing the people (p. 17).18

However, this argument trades on a misconception. Those who use it seem to have been misled into thinking that there is an entailment relation ‘between the (primary) reason being the cause of the action under any description of thle action and its being a reccson for the action under any description, Clearly, however, opponents of the ‘two-domainists’, n,eed not claim there is an entailment. On the contrary they claim that something is a (primary) reason for a oertain act only under a particular description of the action. A dcxription, namely, relating pro- attitude and belief in such a way as to constitute a ‘sufficient’ condition of the action’s occurrence. Thus in a case such as that

1Kf. also R. J. Richman “Reasons and Causes : Some Puzzles’’, Australasian Journal of Philosophy, 1969 pp. 42-50; M. Brand (ed.) The Nature of Human Action (Glen- view, Illinois, 1970), p. 12.

‘TWO-DOMAINISM’ AND HUMAN FREEDOM 41 of White’s the agent had no reason for the action of killing the pe~ple.‘~

- I1 .7. Finally, it is claimed that reason-giving explanations

have a special function. They justify an action or show its appropriateness in the circumstances. Teleological explanations, in other words, reveal the rationality of what an agent does or intends to do. Thus discourse about reasons belongs to the language of elvaluation not to the objective, value-free description of natural events. Hence, reasons cannot be causes.

But even if the premise, that the special function of reasons is to provide a rationale for action, is true, it certainly does not follow that reasons cannot be causes. Although being a cause may not be sufficient for something to be a reason, its being a cause does n’ot preclude it from being a reason either! Thus if it is the case that

(1) x has a (primary) reason for doing some action; (2) x does that action because of the reason,

the question comes back again to the force of the “because”. This argument does not seem to me to be able to rule out the possibility that only the causal relation renders the “because” intelligible.

I conclude, therefore, that none of the seven arguments I have considered, has the force credited to it by ‘two-domainists’ and hence that neither individually nor collectively do they establish the argument with which we began in this section. Hence there is nothing in the arguments which establishes a logical incon- sisQency or obsarity in the view that teleological explanations of action are causal explanations. We have; seen to date that neilther the form, nlor the substance, of teleological explanations preclude their being causal explanations. Perhaps this seems a relatively weak point at which to arrive. Not so, however, for ‘two-domainisrn’ seeks to deny the possibility we have to date defended. Whether that possibility is actualized does not enter into consideration of the truth of principle (D).

19Daniel Bennett, “Action, Reason and Purpose” Journal of Philosophy, 1965 p. 87 has argued that making causing intentionally and cawing rationally, relative to the expression “under the description” has two counter-intuitive consequences. It Seems to mean that agency entails a language and that an action would only be done for a reason if it were actually described. These are matters worthy of attention but are beyond the scope oi this paper. The important point for our purposes is one Bennett recognizes, namely that the description-relative device overcomes the argument considered above.

42 ROBERT YOUNG

IV After such a long and compljex argument, a reader might be

forgiven for thinking that ‘two-domainism’ has been laid to rest. But such a reader would have reckoned without the extraordin- ary ingenuity of ‘twodomainilsts’. Thus, even though the major part of the attack on (D), and defence of the possibility of a deterministic account of action has been rehearsed, there are yet a couple more claims with which we must deal.

Not only have ‘twodomainists’ argued that action is to be explained non-causally; they have sometimles claimed action is exempt from causal categories altogether. Thus

111. Action (intentional behavior) cannot be given a causal explanation a t all because no explanation of an action can be correct which is incompatible with the claim that it was brought about by an intention.

Ckarles Taylor writes in The Explanation of Behaviour :

If a given piece of behaviour is rightly classified as an action, then we cannot account for it by some causal antecedent, where the law linking antecedent (E) to behaviour (B) is not itself conditional on some law or rule governing the intention or purpose. For if the law linking E to B were not dependent on some law linking E and the intention, I, to do B, then E-B would hold whether or not E-I held. But then B would occur on E whether the correspmding intention was present or not. And then, w e n when it is present, it cannot be said to bring about the behaviour so long as this is done by E. Thus to account for B in terms of E would be to offer a rival account, to disqualify B ar, an action. (pp. 34-5).

On Taylor’s - ”II__.._ view, if B is an action, the most that can be hoped for in the way of a causal explanation is a linkage E-I-B where E is a cause of I. But since I, by I1 above, cannot be a cawe of B (only a reason), then though B is explained by I it is not causally explained. Hence actions cannot be causally explained.

We have already seen that I1 has not been established, so this thesis, which is dependlent on the truth of thesis 11, is in bother. Furthermore, there is to my mind, very grave doubt about the sense of the claim that I non-causally brings about B. If I could understand what that might mean I concede that there may well be no incompatibility or contradiction in the claim that E causally brings about I (and hence B) while I n m -

‘TWO-IIOMAINISM’ AND HUMAN FREEDOM 43 causally brings B about, too. I oonfess that I do not, however, understand.

This argument is allied with IV. Actions are not reducible to bodily movements simpliciter,

from which one can infer that actions are not the same as bodily movements simpliciter.

I t has been held that ‘two-domainists’ are guilty of a contra- diction in claiming that actions cannot be caused but that movements can, since they acknowledge that movements are sometimes identical with actions.2o At least one other writer has attempted to absolve ‘two-domainists’ of this charge.2l I will endeavour to show that there is a contradiction; that ‘even if there were no Teal contradiction there would still be other problems; finally, that if ‘two-domainists’ care to adopt a differ- ent way of speaking they can avoid the contradiction. I will try to show these three things in turn.

Melden writes in Free Action :

. . . although the bodily happening needs to be distinguished from the action of raising the arm, the former, in appropriate circumstances, is the very same event as ?he latter (p. 74) (my italics).

When Melden here speaks of “appropriate circumstances” I think he is referring to the social institutions and practices that, according to him, determine what action any given movement is. Just in passing it is worth pointing out that against Melden, “appropriate circumstances” must not only exist but be known (or believed) to exist by the agent. If this is so the knowledge or belief then looks like a cauml factor in his action. To return to the chief issue, the “rules” referred to by proponents of this argument, are quite independent of the causes of the movement (which are, say, brain states for argument’s sake$. A certain movement, which is numerically identical with posting a letter, is the posting only because the existence of letter-boxes and letters presupposes a host of social insititutions and practices. Thus knowing a person’s brain states cannot license a direct inference to the claim that he is posting a letter, for there are no laws which could so link brain states an,d I~atter-postings. In conceding this I do not wish to deny that there can be laws which link brain-processes, desires, beliefs, hand movements,

2oCf. Davidson, op cit., p. 700. 21Cf. Landesman, o p d., pp. 344f.

44 ROBERT YOUNG

letter-boxes and letters with the posting of a letter. The exist- ence of these complex laws is a matter for empirical investiga- tion. Opponents of ‘two-domainism’ need only claim that nothing ‘two-domainists’ have said preicludes the possibility of there being such laws.

I think all can agree that it is perfectly possible for a bodily movement which brings about a certain result to be the very Same thing as an action which is directed to that result. What Melldea seems to be claiming though is that the same movements might constitute distinct actions (depending on context, rules being followed and so on).

But this won’t help avoid the contradiction. For if there is any degree of seriousness in talk about identity then what goes for x goes for y if x and y are really identical. Thus if causal talk goes for bodily movements i t does so for actions. Furthermore, it seems pretty likely to me that it will also be apposite for institutions and practices.

Before attempting to spell out a way out of the above contra- diction it should be pointed out that if, in fact, ‘twodomainists’ are not guilty of a genuine contradiction in the claim under consideration, some of them may be guilty of vicious circularity. I suspect that for at least some of the ‘two-domainists’, an action is performled only if a rule is followed. But the notion of rulefollowing must surely be understood in terms of action. To obey a rule is to act in a prescribed way. Any expliaation of action in terms of ‘rule-following behavior’ would thus be viciously circular. Furthermore it may well be false that actions can only be performed if rules are followed. Not only can one act deliberately contrary to established practices, traditions and rules, but it seems plausible to say that one can act in new, innovating ways which are not rule-governed.2a

To return to the matter of the possible contradiction. It seems to me that if ‘two-domainists’ were to drop talk of identity in favour of some notion such as ‘constituting’ (or perhaps even of extensional equivalence, ~ I ~ o u g h I am miore doubtful of this) the point about the role of institutions in relation to some actions could be retained without succumbling to a contradiction. The bodily movement could be the outcome of antecedently sufficient causal factors but be constitutive of an action only under a certain (different) description which referred to the institutions, rules and so on.

BCf. M. Brand, o p . n’t., p. 17.

‘TWO-DOMAINISM’ AND HUMAN FREEDOM 45 V

I want briefly to make one point in this section. Whatever way ‘two-domainists’ jump on the matter just discussed it seems that they must face the spectre of determinism. I pointed out right at the beginning of the paper the belief among ‘two-domainists’ that thley are able to avoid doing that.

Consider a bodily movement, BM. Either there are ante- cedently sufficient (causal) conditionis for the occurrence of BM or there are not. If there are, then given the causal condition BM occurs. Suppose now that BM occurs under such circum- stances that it is an action. If it is the case that the presence of antecedently sufficient conditions neaessitates a particular move- ment in such a way that the movement cannot occur freely, saying that that movement occurs under certain rulies, customs, practices and so on cannot affect the necessity of the outcome. Appeal to these rules and so on may avoid contradiction but it cannot, as far as I can see, allow one to !sidestep the issue of the compatibility of freedom and determinism. These rules can only classify or categorize action, if they can do anything at all. Adherence to argument IV (filled in as sympathetically as above) is a natural fruit of the rest of the ‘two-domainist’ case. It appears, however, to undermine it! That says something very instructive about ‘two-domainism’, I suggest.

VI I have argued against various theses which are supposed to

It is conceptually or, in principle, impossible to provide deterministic explanations of action.

Not one of these theses (or their supporting arguments) appears to have come off unscathed. I have concentrated on arguments rather than on the methodollogy which leads to these arguments. In fact, even though I have concediad a good deal for the sake of argument, this methodological foundation may well be even more shaky than some of the argument^.^^ However, for my purposes, the arguments themslelves are sufficiently shaky to allow me justifiably to conclude that recent work in the philo- sophy of action, rather than showing the problem of freedom

ground principle (D), that is, the principle that:

23May Brodbeck, op a t . , and Ruth Macklin, “Explanation and Action: Recent Issues and Controversies”, Synthese, 1969 pp. 388-415, each try t o cast doubt on the methodology usually employed by ‘two-domainists’.

46 ROBERT YOUNG and de$erminism is a pseudo-problem, does something quite different. It strengthens the claim that since voluntary action is best understood as occurring when the causal chain associated with behaviour includes the (primary) reason, voluntary action is a necessary condition of free action. Thus, if the most plaus- ible understanding of voluntary action emerges from considering such action as the outcome of certain causal antecedents rather than of others, this augurs well for the compatibilist position regarding freedom and determinism. However, it only ‘augurs well’, that is, it is in line with, but by no means establishes that position. The schema is obviously equally acceptable to thie hard determinists.

My main concern at the moment, though, is to stress that a satisfactory philosophy of action is very much groundwork in relation to the problem of freedom and determinism. Despite the putative consequences for that problem claimed as tribute by the ‘twodomainists’ the upshot of my paper is that the problem, in fact, remaimW

FLINDERS UNIVERSITY OF SOUTH AUSTRALIA

MI am grateful to D. M. Armstrong and S. G. O’Hair for helpful discussions on the issues of this paper.