Intentions and Two Models of Human Action

Chapter 7

Intentions and two Models ofHuman action

thomas Pink

1. Introduction

I wish in the following paper to compare, in outline at least, two rather different models of intentional action. The first model is voluntariness-based. this model teaches that intentional human action occurs as an effect of rationalizing pro attitudes towards its performance, and essentially consists in doing what we want or will because we want or will to do it. this model is a familiar feature of modern english-language action theory. It goes back to thomas Hobbes, though the best-known modern defender of it is davidson. as we shall see, the model has been further elaborated by writers such as Harry Frankfurt and david Velleman to provide a theory not only of action but also of self-determination. the second model is practical reason-based. this model views intentional action as consisting, not in any voluntary or willed effect, but in a distinctively practical or action-constitutive exercise of rationality. this model reached a highly developed form in medieval and early modern scholasticism; and it provided the target against which Hobbes first developed his rival voluntariness-based theory.1

In what follows I shall first establish some common ground about what action might involve, and then show how the two models exploit that common ground in quite different ways.

2. Motivation and Voluntariness

It will be useful to begin with some general claims about agency on which many action-theorists of otherwise widely differing views could still agree. one such claim is this. Fully successful agency contains two elements: motivation and voluntariness.

Motivations are content-bearing pro attitudes – attitudes such as intentions or desires or evaluative judgments which motivate us towards and leave us favouring

1 For a general account of the two models and of the history behind them, see my Free Will: A Very Short Introduction (oxford, 2004). For more detailed historical discussion see my “suarez, Hobbes and the scholastic tradition in action theory”, in thomas Pink and Martin W. F. stone (eds), The Will and Human Action: From Antiquity to the Present Day (London, 2004), and “action, Will and Law in Late scholasticims”, in Jill kraye and Risto saarinen (eds), Moral Philosophy on the Threshold of Modernity (dordrecht, 2005). the argument of this paper is developed further in my The Ethics of Action: Self-Determination (oxford, forthcoming).

Reasons and Intentions154

doing various things, be it crossing the road or planning a summer holiday. Whereas voluntariness, or action performed voluntarily, is what is then done, in response to the real or at least apparent desirability of doing it, and on the basis of such pro attitudes towards doing it. so, for example, I count as crossing the road voluntarily when I cross the road on the basis of my desire or intention to cross the road.

Let me emphasize that here I am using “voluntariness” as a term of art, to refer to action performed on the basis of a pro attitude towards performing it. I do not intend “voluntariness” to mean “action” simpliciter. the relationship between voluntariness as I understand it and agency is a matter for debate – not a trivial equivalence. Voluntariness must indeed be one form taken by action. But action may yet prove to take non-voluntary form too.

notice that action performed voluntarily may of course be mental as well as bodily. When in my mind there forms an image of a scene, I may perfectly well be entertaining the image voluntarily, on the basis of a desire or intention so to imagine the scene.

Motivation Voluntariness

Pro attitudes to doing a(e.g. Intentions or desires to do a, judgments that it would be good to do a)

cause doing a.

Content-bearing attitudes; their contents determine agency goals.

Voluntary in that a occurs as effect of pro attitudes to its performance.

Goal-directed? Goal-directed through being a voluntary effect of pro attitudes.

Table 7.1

Motivations play two roles in successful agency. they have contents – and these contents specify the goals towards which the agent is acting, and in terms of which his agency can be made intelligible. Why is he crossing the road? He is crossing the road intentionally, and he is crossing it in order to get to the other side. and part of what makes this true is that the pro attitudes on the basis of which he is acting include a pro attitude towards crossing the road, which is based in turn on a pro attitude towards getting to the other side. He thinks that if he crosses, he will indeed get to the other side; and so intending to get to the other side, he forms an intention to cross. these intentions, which provide the agent’s motivating pro attitudes, deliver through their contents the goals towards which his voluntary agency, indeed his agency as a whole, is directed. again, when I voluntarily imagine a scene, my goals in so doing will be given by the pro attitudes on the basis of which my imagining occurs: they may be a pro attitude towards visualizing the scene from a certain perspective, given a belief that if I do so I will remember some element of it, and a pro attitude towards so remembering.

secondly, motivations provide the effects through which voluntary agency takes place. to count as acting on the basis of his pro attitudes to crossing the road and getting to the other side, the agent must be crossing the road through the effects of

Intentions and Two Models of Human Action 155

those motivations. the motion of his limbs which takes the agent across the road must not be occurring by pure chance; and it must not be occurring as the effect of some quite different cause (such as an earth slippage). His limbs must be moving as an effect of his intentions to cross the road and get to the other side.

the voluntary, by contrast, has a certain deliberative and teleological priority over motivation. the move through motivation to voluntary action is a process which can involve reasoning or deliberation, and has a goal on attaining which that reasoning is focused. and the goal of the process is not the possession of any particular motivation, but rather the successful performance of voluntary action – which is why the fully successful completion of intentional agency must always involve the voluntary.

For it is with how we voluntarily act, and the desirability of that, which we as practical deliberators are principally concerned, and not with what prior motivations we hold. In general we worry about whether, say, to go on holiday, and what the best holiday would be, not about whether to desire or intend to go on holiday. We deliberate about how voluntarily to act; and it is centrally on the basis of that deliberative concern with the voluntary that our motivations to action are formed.

so pro attitudes, be they motivating judgments, or desires, or intentions, all have one important thing in common. they are attitudes which, when based on deliberation, are based on deliberation about their voluntary objects, and about the desirability of these. Pro attitudes must, by their very nature, occur in a way which is responsive to whether or not their objects are desirable.

It follows that motivating pro attitudes towards the voluntary are not formed voluntarily themselves. We form pro attitudes non-voluntarily, by directing ourselves at their objects as desirable (or at least not undesirable), and so in a way responsive to how desirable those objects are. We do not form pro attitudes in response to the desirability of holding those attitudes themselves. any further ends that move us to form a pro attitude towards doing a are those ends relevant to the goodness or desirability of doing a, not those ends relevant simply to the desirability of holding the pro attitude. We are moved to desire to do a by the ends which we expect doing a to further – not by the ends which we expect simply desiring to do a to further.

3. The Non-Voluntariness of Decisions and Intentions

amongst the pro attitudes which explain our actions are intentions and the events of decision by which intentions are formed.2 It seems that, like other pro attitudes, decisions are taken and intentions are formed non-voluntarily. It seems that they are not taken or formed voluntarily, in response to their own desirability and on the basis of our wanting or deciding so to decide or intend.

2 strictly speaking, not every intention-formation is a decision. suppose you ask me a question, and I form the intention of replying. If not replying to you was an option which I did not even entertain, I cannot be said to have decided to reply to you. I cannot be described as having taken a decision about whether to do a if my formation of the intention to do it was unaccompanied by any entertainment of alternatives. But the points which I make about decisions and their status as intentional actions apply, I think, to intention-formation generally.


If decisions themselves were taken voluntarily – if they were voluntary actions exactly like the voluntary actions which they explain – then a decision could itself be taken just on the basis of a prior pro attitude towards taking it, such as a desire or a decision or intention to take it. If decisions were voluntary actions then they too would be subject to the will just as are the actions which they explain. Yet it is utterly and abundantly clear that our decisions to act this way rather than that are not so subject to the will. they cannot be taken just on the basis of prior desires or decisions so to decide. For example, I cannot decide today that at precisely 2 p.m. tomorrow I shall then decide to lose weight – and expect so to decide tomorrow just on the basis of today’s decision so to decide. I cannot take a specific decision to act just on the basis of having decided to take that particular decision. decisions and intentions to act are not themselves subject to the will and voluntary as are the actions which they explain. as Hobbes rightly observed, using “willing” to mean deciding or intending: “I acknowledge this liberty, that I can do if I will, but to say, I can will if I will, I take to be an absurd speech.”3

one can of course take and execute decisions to make one’s mind up – to decide one way or the other. I can decide today that at precisely 2 p.m. tomorrow I shall make my mind up then about whether to lose weight – and thus arrive at my final decision tomorrow on the basis of today’s decision to decide. But that in no way shows or even suggests that the final decision taken at 2 p.m. tomorrow, whether it proves to be a decision to lose weight or, alternatively, a decision not to lose weight, is taken voluntarily. For though I can decide in advance that it is then when I shall make up my mind, voluntarily postponing my final deliberation and decision until that time, I cannot effectively decide in advance which decision I shall take. I cannot decide in advance what I shall eventually decide. If I do finally decide to lose weight, that decision will be taken non-voluntarily – that is, in response to the merits as I then see them of its object, the option of losing weight. the decision to lose weight will not be taken voluntarily, on the basis of an earlier decision to take that particular decision.

the clear non-voluntariness of decisions and intentions has a rationale – a rationale which becomes evident when we consider the function of decision-making. the whole point of bothering to take decisions about how we shall act, after all, is to ensure that we end up performing the right voluntary actions. decisions have a reason-applying function in relation to voluntary action. the point of taking decisions about which action to perform is to apply practical reason as it refers to the voluntary actions decided upon – to ensure that one performs voluntary actions which reason supports or at least does not oppose.

this reason-applying function is a fundamental feature of the common-sense psychology of decision-making. It shows up in the way we naturally tie the justifiability of deciding to do A to the justifiability of remaining motivated to do A thereafter and of finally doing A. We never ordinarily agree with a decision to do A while leaving the question of whether to do a still open. to endorse someone’s decision to do a as the right decision is always to endorse their doing a thereafter as the right action. to agree with how someone has decided to act is – by clear implication – to agree with

3 thomas Hobbes, “of Liberty and necessity”, in david daiches Raphael (ed.), British Moralists (Indianapolis, In: Indiana university Press, 1991), pp. 61–2.


their going on to perform the action which they have decided upon.4 and that is not surprising, given that the function of decisions is to apply reason as it concerns the voluntary; for if that is the function of a decision, a decision to perform some voluntary action A can only be justified if doing A thereafter would be justified too.

Voluntary actions are justified by features of them which make performing them good or desirable. to argue someone into, say, crossing the road, you need to make out that crossing the road would be desirable, either in itself, or as a means to further ends. It follows that, in so far as decisions serve to apply reason as it concerns the voluntary, the function of decisions is to be responsive to how desirable their objects are. that means that justification for deciding to do A is tied to the desirability of subsequently doing A: for the decision to be justified, subsequently doing A must be on balance desirable.

this tie between the rationality of a decision and the desirability of its object rules out decisions being voluntary themselves. take kavka’s well-known toxin puzzle.5 this involves the offer of a $1 million prize for taking a particular decision to do a, a prize to be won just for taking that decision and irrespective of whether a is subsequently done. the action a decided upon is, considered overall, mildly undesirable: in kavka’s example, it involves taking a mild toxin one or two days later.

If the decision to do a were voluntary, then one could take that decision at will, in direct response to the desirability of taking it. one could take that decision simply because one wanted or had decided to, because taking it would win the prize. now to exercise any such capacity for voluntariness, if one did possess it, in such a prize-winning way, would surely be perfectly rational. even if the prize-winning decision did lead one to enact it – to take the mild toxin – the toxin’s mildness would guarantee no lasting damage, and that damage would be far outweighed by the huge sum won through the decision itself. the decision is clearly a desirable one; and so if decisions were taken voluntarily, in response to their desirability, taking this particular prize-winning decision would surely be fully justified. But, in such a case, once the prize was won there would be every justification not to retain the decision thereafter or enact it: in itself the toxin would be unpleasant, and nothing would be gained by taking it, the prize having already been secured. taking the toxin would be wholly undesirable. In which case, a voluntary decision to take the toxin could be taken, and taken with full justification, in order to win the prize; but though the decision was justified, acting as decided, taking the toxin thereafter, would not be justified at all.

If decisions were voluntary, the justifiability of deciding to do A would no longer imply that subsequently doing A was justified too. To endorse someone’s decision to do a would no longer be to endorse their doing a thereafter. since, in common sense psychology, the rightness of a decision does imply the rightness of the act decided on – to endorse someone’s decision is to endorse their doing what they have decided – decisions can’t be voluntary.

4 assuming, of course, as I am throughout, that we are considering appraisals of the decision and of the act decided on made from the same information base. Having endorsed someone’s decision we can, later on, perfectly well come to advise them not to act as decided. But that is because new information has come in meantime – information which, if available earlier, would have prevented our original endorsement of the decision too.

5 kavka, “the toxin Puzzle”.


the non-voluntariness of decisions and intentions clearly has the same general basis as the non-voluntariness of other motivations, such as desires. decisions are the formations of content-bearing attitudes – of intentions to act. and, as with desires to act, the object of a decision or intention to act is a voluntary action – some action which, when we decide to perform it, we are responding to in a way appropriate to, and to be justified in terms of, the action’s desirability. If decisions are non-voluntary, that must be because, as much as any desire-formation, they are content-bearing events directed at their objects, the voluntary actions decided upon, as actions which it would be desirable to perform.

4. What is an Action?

We have so far examined two fundamentally distinct components to the process of successfully performing action – motivation and voluntariness. But now the question arises as to precisely what occurrences in this process count as actions, and on what grounds. Voluntariness seems to be a case of action, on anyone’s view. But what makes it so? – and do any of the non-voluntary motivations behind voluntariness arise as action too?

on any view, an action is a kind of event. to perform an action, to do something, is always to make a change. But what kind of event is an action?

actions seem distinguishable in one central way – by their relation to an object. an action is an event which has an object – an object at which the agent is directing the action’s performance; and this object is what motivates the action’s performance, being the goal at which the agent is aiming, and which he is using the action to attain. this goal-directedness, this property of being done by its agent as a means to an end, is something which all actions have in common. and it distinguishes events which are actions from events which are not.

events which are non-actions are not directed at attaining goals. either events which are non-actions are not directed at objects at all – as when, through a reflex, someone’s legs just randomly move. or if they are object-directed, these objects are not goals; they are not objects which the event is being used by the agent to attain. thus, to take examples to which I shall be returning, the event of forming a belief or a desire has an object – what the agent is starting to believe or desire. But that object, what is believed or desired, is not in general a goal which the belief- or desire-formation is being used to attain. In forming a belief I am in general simply responding to the likelihood that the belief ’s object is in fact true. I am not forming the belief in order to make that object true. and in forming a desire I am simply attracted by the desire’s object; I am not yet doing anything to bring it about.

an event’s being directed by someone at attaining a goal – its being used by that someone as a means to an end – is clearly sufficient for agency. To employ a means to an end, to do a for the sake of attaining B – that is certainly to perform an action. It is arguable too that if we take a broad enough view of the means–end relation, so that means can constitute as well as cause the attainment of their ends, to act is always to employ a means to an end. this is so even if the end is simply what I am doing, so that the object of the action is its own performance. suppose I am quite deliberately


and intentionally drawing a doodle. suppose too that there is no further goal at which the action is being directed. that does not leave the doodling without any goal at all. there is still something for the sake of which the doodling is being done. Insofar as the doodling is being done by me quite deliberately and intentionally, it is clearly being done by me for its own sake. so I am doing what I do towards a goal – that goal being precisely the doing of what I do.

Since goal-directedness in an event is clearly sufficient for agency, and arguably necessary too, central to any model of agency must be its model of goal-directedness.6 so where is intentional goal-directedness, and therefore intentional agency, to be found? as we shall see, there are at least two quite different models of intentional goal-directedness – the voluntariness-based model and the practical reason-based model; and these consequently give two quite different accounts of what counts as intentional action and why.

5. The Voluntariness-Based Model of Action

there is one point at least where most philosophers have seen goal-directed agency as occurring. this is through the effects of our prior motivations, in successful voluntariness. Crossing the road must be something which I can do, and do intentionally, as a means to some end. What gives my crossing of the road its goal-direction when it has it? as we have seen, the goal-direction comes from prior content-bearing motivational causes of my crossing the road – pro attitudes towards crossing the road and towards, say, getting to the other side. so one form of goal-directedness is constituted causally. In the case of a voluntary action such as crossing the road, to cross the road intentionally as a means to ends is to cross the road voluntarily, as an effect of pro attitudes towards crossing the road and towards further goals to be attained through crossing the road. Goal-directedness in voluntary form arises as an effect of prior pro attitudes to its performance.

so one important model of action – the voluntariness-based model – models goal-directed agency in general on goal-directed voluntariness. this is the view of agency which we get in writers such as Hobbes or davidson – and in many other english-language action-theorists. according to this voluntariness-based model, goal-direction, and so intentional agency, implies voluntariness, and exclusively occurs as an effect of prior motivations, in the form to be found on the “Voluntariness” side of the diagram above. Intentionally to do a as a means to some end requires that there be some psychological attitude causing me specifically to do A – namely a pro attitude towards doing a, whether for its own sake or as a means to further ends.

The passivity of motivation

Goal-directed action, according to this model, is located in the voluntary, and occurs as an effect of motivating pro attitudes. and if the model is not to lead to a vicious causal regress, these action-causing motivations must not have to arise out of a further case of

6 those who deny that goal-directedness is essential to action can still take what follows as enlightening about the nature of one central and characteristic form of action – action in goal-directed form.


action in turn – a case of action which, on the model, would ex hypothesi have to be caused by yet further pro attitudes towards its occurrence, and so on ad infinitum. Whether they are cognitive “besires” – motivating pro attitudes which are beliefs – or mere non-cognitive yens or urges, the causing pro attitudes in terms of which goal-directed action is defined, and out of which it arises as an effect, must ultimately be passions in the following sense. they must all be pro attitudes which are formed passively or other than through action. and so I shall use the term “passions” as a general term for referring to the pro attitudes which the voluntariness-based model uses to define action.7

In any case, it looks as though the pro attitudes in terms of which actions are explained are going to be passive anyway, according to the model – precisely because coming to hold those pro attitudes is something that we do non-voluntarily. and that is an intuitive enough conclusion, at least where certain pro attitudes such as desires are concerned.

Consider desire-formation. suppose I am deliberating about an offer which I have been made; and this process of deliberation leaves me forming an increasingly strong desire to accept it. that I form this strong desire to accept the offer, rather than remaining indifferent to it, or forming a strong desire to refuse – this is hardly going to count as an action of mine. once I start thinking about the offer – once I have deliberately performed that action – coming to want to accept it is not something further that I intentionally do. Instead the want is surely something which comes over me – a state which my understanding of what the offer involves just brings upon me.

Desires are passions. Coming to hold a specific desire, such as a desire to accept an offer, is something which happens to one, rather than something which one deliberately does. and the connection between goal-directedness and action gives part of the story why. Forming a desire to accept the offer is not an action because, as we have already observed, it does not constitute the adopting of means to pursue any end. In so far as I want to accept the offer I am of course attracted to accepting it. But that attraction is all my attitude involves. the desire-formation is not an action because in forming the desire I am simply responding to its object, what I want, as something desirable or good. I am simply attracted by it. In forming the desire I am not yet pursuing that object as my goal. the object of a desire is not a goal which the desire is being formed in order to attain.

But if desires are not formed in order to ensure that we subsequently act as desired – if simply in coming to want to accept the offer I am not pursuing any goal – what makes this true? the voluntariness-based model appears to explain. the desire is not being formed voluntarily, in response to its own desirability and on the basis of some pro attitude towards so desiring. as for desires, so, on the voluntariness-based model, for our non-voluntary motivating pro attitudes generally – whether cognitive or non-cognitive, whether mere desires or full-blown intentions. The forming of a specific motivation to do a rather than B is not something which ever directly counts as our own intentional doing – precisely because the pro attitude is formed non-voluntarily.

7 By an event’s being “passive” I mean merely that it is not an action; and by a pro attitude’s being a “passion” I mean merely that its formation is not an action. the words “passive” and “passion” sometimes have other implications, such as of causal inertness or of detachment from reason or cognition and the like. no such further implications are intended here.


If the occurrence of a desire is ever our deliberate or intentional doing, that can only be so indirectly – because the pro attitude is the effect of some prior voluntary action which can be deliberately used to cause that specific attitude. I can, for example, increase my desire for food by intentionally and voluntarily taking a run; or by concentrating my mind on inviting mental imagery of delicious tastes and smells. In these cases inherently non-voluntary feelings and motivations can, to a degree, be our own voluntary doing – but only indirectly, as intended effects of prior actions that were voluntary. But aside from the indirect manipulation of ourselves into holding them by prior voluntary action, these feelings and motivations can only be formed by us non-voluntarily. and so, on the voluntariness-based model, none of these feelings and motivations can count as actions on their own account.

as for desires, so for non-voluntarily formed pro attitudes generally. and this is why classic supporters of the voluntariness-based model such as Hobbes and davidson have been inclined to assimilate intentions to intuitively passive attitudes such as desires. For davidson, the forming of a particular intention is not a case of intentional action – any more than is coming to hold a particular desire. Why not? the same answer is given for intentions as applies to desires. In davidson’s view, forming an intention is no action precisely because the formation of particular intentions, like desire formation but unlike the performance of the actions intended, is non-voluntary: “the coming to have an intention we might try connecting with desires and beliefs as we did other intentional actions … But the story does not have the substantial quality of the account of intentional action.”8 Particular decisions to act aren’t taken voluntarily, in response to the desirability of so deciding – but non-voluntarily in response to the desirability of the action decided upon.9

Summary

on the voluntariness-based model of agency, then, it is passive motivational causes which ultimately supply, through their contents, the goal direction essential to agency. they make agency intelligible by giving “the reason why” the agent is acting or doing something – his aim is to do a and, possibly, to do various other things by doing a. It is essential to goal-directed agency that it occur as an effect.

It follows, according to this model, that our capacity for agency depends for its exercise on the causal powers of motivations that are passive. to act intentionally is to be caused to do what one wants by one’s prior passive motivation to do it. It follows that any overall obstacle to our desires and other passions effecting their satisfaction is an obstacle to our agency, because it is an obstacle to what agency essentially is – doing what we want or are passively motivated to do on the basis of wanting to do it.

8 donald davidson, “Intending”, Essays on Actions and Events (oxford, 1980), p. 90.9 see davidson, “Replies”, pp. 213–14. In fact, in his “Intending” davidson wants to

identify intentions with outright judgments that the voluntary action decided upon is good or desirable, desires being prima facie judgments to the same effect. and in his view the formation of none of these judgments is an intentional action, because all are formed non-voluntarily.


6. The Voluntariness-Based Model and the Agency of the Will

It is not clear that agency is restricted to voluntary actions, whether they be bodily such as crossing the road, or mental such as visualizing things or thinking and deliberating. For there are certain motivating attitudes prior to and explanatory of such voluntary actions – namely decisions to perform them and the states of intention formed by decisions – which seem themselves to arise in and through intentional agency.

Consider what in our lives is up to us – what we have control over. I do not simply have control over what I do voluntarily – over what I do on the basis of deciding and intending to do it. I seem also to have and exercise a prior and direct control over what I decide and intend. Just as it is up to me whether or not I cross the road, so it is up to me beforehand whether I first decide to cross the road. And my present control of that decision – its up-to-meness – can give me control now of whether later on I shall end up acting as decided. Which is precisely why I take a decision to cross the road, if I do – in order to ensure that in the future I act exactly as decided and cross the road.

now plausibly, our control is exercised in and through deliberate or intentional agency. For me to have control over whether I speak loudly or softly, that I speak loudly or softly must be something that I can intentionally do as a means to ends. Control exercised in and through decision-making is no exception. For me to have control over my present decisions, and through these decisions over my future voluntary actions, taking a specific decision to act must be something that I can intentionally do, and do as a means to causing my performance of the future voluntary action decided upon.

Goal-direction, the adoption of means to ends, must also be found in motivations of the will. Taking a specific decision, deciding to do A rather than B, must be a goal directed and intentional action – something I can deliberately do, and as a means to ends, the ends including that I subsequently act as decided.

How might a supporter of the voluntariness-based model seek to accommodate decisions as goal-directed intentional actions?

i. Decisions as straightforwardly voluntary actions

It might be tempting to return to the voluntariness-based model – to claim that what makes decisions actions is that decisions to do a are voluntary effects of prior pro attitudes towards taking them. that I decide to do a rather than B is my intentional doing, on this view, in that what motivates me to take the decision is nothing other than the fact that I want or intend so to decide.

some philosophers, such as david Lewis, would maintain that decisions can perfectly well be straightforwardly voluntary actions. We can take specific decisions at will, in response to the real or apparent desirability of taking them, simply on the basis of wanting or deciding so to decide.10

now anyone who teaches the voluntariness of decisions and intention-formations must pay a price – a price that Lewis himself is perfectly willing to pay.11 they must

10 see david Lewis, “devil’s Bargains and the Real World”, in douglas Maclean (ed.), The Security Gamble (ottawa: Rowman & allanheld, 1984), pp. 141–54.

11 Lewis, “devil’s Bargains and the Real World”, p. 143.


give up a fundamental intuition: that the point of deciding what to do is not to take the decisions we want to take, but to ensure that we do the right thing – that the function of the will is to apply reason as it governs the voluntary, so that the rationality of deciding to do a guarantees the rationality of doing a thereafter.

We have seen why this price must be paid. If decisions are motivated and justified by reference to their own desirability, and not by reference to that of their objects, the rationality of deciding to do a and that of subsequently doing a must fall apart. For the fact that a particular decision would be a desirable one to take does nothing to guarantee the desirability of acting as decided thereafter.

My view is that the price is simply not worth paying. For the intuition that decisions play an essentially reason-applying role in relation to the actions decided upon is basic to the common sense psychology of the will. It is at least as basic as the intuition that decisions are self-determined actions. and it implies that even if decisions are actions, they cannot be voluntary.

ii. The appeal to the voluntariness of deciding one way or the other

there is one thing at least which, it seems, I can do voluntarily – namely to arrive at a decision one way or the other about whether to do a. the intention on which a decision to do a is based may not be an intention to take that particular decision. But it may perfectly well be an intention to make one’s mind up – to decide, one way or the other, about whether to do a. so, as Robert kane, alfred Mele and Randolph Clarke12 have all suggested, cannot the status of deciding to do a as a goal-directed action be explained in terms of its being an effect of an intention or other pro attitude towards making up our minds – towards arriving at a decision one way or the other?

this proposal simply fails to address the problem. It does nothing to vindicate the voluntariness-based model, because it does nothing to establish the voluntariness of deciding specifically to do A. And it is the voluntariness of that which would need to be established if the voluntariness-based model is to accommodate our ordinary conception of how we ourselves can determine our own wills. For we naturally think that not only do we determine for ourselves whether we take a decision at all. We also determine for ourselves both what specifically we decide and, in so doing, too, how we shall act thereafter. Which means that our taking of these specific decisions must be something that we deliberately and intentionally do – and do as a means to ensuring that we perform the actions decided upon.

so it is not enough for it to have been our deliberate goal-directed doing that we took a decision at all. It must have been our deliberate goal-directed doing that we took this decision rather than that. But what we specifically decide, it seems, is no more subject to our prior will or decision than what we specifically want. At the level of specific decisions the voluntariness-based model, and its conception of what agency involves – acting as we decide or want – has not been shown to apply.

12 see Randolph Clarke, Libertarian Accounts of Free Will (oxford, 2003), p. 26; Robert kane, The Significance of Free Will (oxford, 1996), and alfred R. Mele, “deciding to act”, Philosophical Studies 100, no. 1 (2000), p. 205; alfred R. Mele, Motivation and Agency (oxford, 2003).


In any case, many of our intentions are not effects of prior intentions to make our minds up. Indeed, on pain of a vicious causal regress, not all our intentions could be so caused. sometimes we simply take a decision without having had any prior intention of making up our minds. We just make up our minds on the spot, as soon as options occur to us, without having had first to form a prior intention to arrive at a decision. But are decisions so taken the less our own doing for that? In such cases are we the less able deliberately to determine for ourselves which decision we take?

iii. Decisions as quasi-voluntary actions

Let us accept that we cannot take specific decisions just on the basis of deciding or wanting to take them, and in response just to the desirability of so deciding. as Hobbes put it so bluntly: the will is not voluntary.13 as much as desires, decisions and intentions are pro attitudes which must occur in a way responsive to the desirability of their objects. If then decisions are actions, as desire-formations are not, that cannot be because the voluntariness-based model of action in its classic form is true. If decisions are actions, that model must be false. But something like it, a heavily qualified version of the model, might still be true.

decisions may not be voluntary. We may not be able to decide as we decide or want to. But that does not mean that some sort of pro attitude towards the taking of specific decisions cannot play a part in their motivation. and perhaps it is only because of the motivating role played by such pro attitudes towards them that decisions do count as intentional actions.

What kind of pro attitude might move us to take a decision? It cannot simply be any pro attitude whatsoever towards taking the decision – otherwise we would return to the voluntariness-based model in its unqualified form. If decisions are to occur as responses to the desirability of their objects, decisions cannot be taken just on the basis of their own general desirability. But perhaps a decision is taken on the basis of its desirability in one specific respect – as a means to attaining its object. The pro attitudes which motivate a decision to do a, then, must always include a particular kind of desire so to decide. they must always include a desire so to decide which is pro tanto in a particular way – which is directed at the decision insofar as it provides a means to attaining the further desired end that a be done. What makes decisions to do a actions, and gives them their goal-direction – towards the goal that a be done – is their being motivated by desires to do a, and to take the decision to do a as a means to this end.14

13 “Can any man but a schoolman think that the will is voluntary?” thomas Hobbes, The Questions concerning Liberty, Necessity and Chance, Clearly Stated between Dr Bramhall Bishop of Derry, and Thomas Hobbes of Malmesbury (London: 1656), p. 256.

14 this in fact was the view of decision motivation defended in my previous book, The Psychology of Freedom. What such a theory of decision motivation needs, of course, is a story about why decisions should have to be motivated by these desires in particular – a story which is not ad hoc, but is based on a plausible account of what kind of event a decision is. In that book, I suggested that a decision is a special kind of action – a motivation-perpetuating action. a decision produces the action decided upon by perpetuating the prior pro attitudes which motivated it – by stabilizing and ensuring retention of the desires by which the decision was itself caused. these desires must therefore be ones which, if retained, would motivate the agent thereafter to act as


the voluntariness-based model says that actions are performed in response to the desirability of performing them, and on the basis of our desiring or deciding so to act. Put so simply, this model is false of decisions. But a qualified version of it is true. decisions are taken in response to their desirability in one particular respect – as means to one desirable end in particular, the end of acting as decided – and on the basis of desires to use them as means in just this way.

On this view, we have no unqualified capacity to decide as we desire or want. But certain desires to decide do still play a role in motivating our decisions. We can and do take decisions on the basis of wanting take them; but just provided that our desire to take those decisions is always for one reason in particular – to use them as means to ensuring that we later act as decided.

this theory of how decisions are actions is more faithful to common sense psychology than Lewis’s view – the view which treated decisions simply as voluntary actions. It respects decisions’ character as events that are by their very nature responsive to the desirability of their objects, the voluntary actions decided upon – as events which therefore cannot be straightforwardly voluntary themselves. Moreover, unlike the view of kane and Mele, it does something to address the point at issue – our capacity to determine how we decide by intentionally taking one specific decision rather than another. The status of specific decisions as goal-directed actions is made to rest not, irrelevantly, on the unqualified voluntariness of something else, but on a qualified or quasi-voluntariness of those decisions themselves. But the theory still faces serious criticisms.

The gratuitousness of appeal to desires to decide

the goal-directedness of what we do voluntarily does seem to come from prior pro attitude causes – from desires or intentions so to act – and from the contents of those attitudes. appeal to such causes, then, is central to understanding action in its voluntary form. But the same is simply not true of decisions. With decisions the postulation of causing desires to decide is not necessary in the same way. It is just not needed to make sense of decisions as having goal-direction in the same way as any action.

take some voluntary bodily action, such as deliberately crossing the road. If we are to make sense of that action as something goal-directed – as intentionally done as a means to an end – we shall need to appeal to some prior pro attitude cause. and it is clear enough why. Without such a pro attitude and its content, there would be no object at all at which the agent’s movement was being aimed. For it seems that the voluntary action involves no content and object of its own beyond that provided by the motivating pro attitudes which give rise to it. Which is why in seeing such a voluntary action as goal-directed, we naturally relate it to pro attitudes on which its performance is based; we naturally convey that goal-direction by referring, for example, to “the intention with which the action is being performed”.

decided. these desires must be desires to perform the action decided upon, and to adopt means to this end – means such as deciding so to act. this previous theory of mine about decision-making is radically different from that defended here. In this present paper it is denied that decisions need be motivated by prior pro attitudes at all, either towards acting as decided or to using the decision as a means to this end.


of course, not all that we do voluntarily lacks a content and object of its own. suppose that I imagine or visualize a scene in my mind’s eye, and do so voluntarily, on the basis of a decision to visualize it. there is here an object of the imagination – the scene which I imagine. But the object of an imagining is not by that fact a goal which the imagining of it is aimed at attaining. taken in itself, an object of an imagining is simply that; a state of affairs which I merely imagine, and not necessarily – indeed, not usually – a state of affairs that the imagining is directed at bringing about. Indeed, an imagining need not be a goal-directed action at all. It may be an event without any purpose, an event which has simply intruded itself passively into my mind. to show that the imagining is a goal-directed action and to determine what its goals might be, we have to leave the imagining’s own content, and find pro attitudes on the basis of which the visualization is occurring. It is these pro attitudes and their contents which give the imagining its goal-direction. these might be pro attitudes towards visualizing a scene of that kind, and perhaps towards attaining various ends thereby – such as remembering something experienced once before or even, in some cases, if the agent really thinks the visualizing can have this effect, making a scene of the kind to be imagined come true. It is the objects of these motivating pro attitudes which give my imagining its goal-direction – not the object which my imagining has in its own right.

as for imaginings, so for something else that we can do voluntarily – deliberating. When, for example, we deliberate practically, we entertain a sequence of thoughts – thoughts with objects of their own, namely the various options between which we are deliberating, conceived by us in the way that we believe them to be. now if in entertaining these thoughts I really am acting intentionally – I am intentionally deliberating about what to do – this whole process will have a further object. It will have an object in the sense of a goal towards attaining which all my thinking is being intentionally directed by me – the goal of finding out how best to act. But this goal and the direction of the thoughts towards it is not constituted simply by the fact that I am entertaining these thoughts with the objects which they have in their own right – the relevant options as I believe them to be. For the having of thoughts with such objects need not be my deliberate doing at all. the thoughts could simply be passing undirectedly through my mind. the goal, if there is one, at which these thoughts are being intentionally directed must come from some pro attitude on the basis of which the thoughts are occurring – such as a desire or intention to find out which option is best. So, just as before, the goal-direction of what we do voluntarily comes from the pro attitudes which cause the action and motivate it – and not from anywhere else.

trying provides another example of how the goal-direction of what we do voluntarily always depends on the content of a prior pro attitude on which the voluntary doing is based. trying, again, is something which we can do voluntarily, on the basis of a decision or desire to try. and when we try there is an object of trying, in the sense of something that we are trying to do. But that object of trying does not, simply as what is attempted, provide the trying with its goal. We have again to turn to the pro attitudes on which the trying is based. and those may not include pro attitudes towards doing what is attempted at all. suppose I wish to discover what happens when I try to raise my arm and fail. So I decide that I will tie my arm down as firmly as possible and then try to raise it. When, on the basis of that decision I try to raise my arm, my trying to raise it is intentional and has a goal. But that goal is not raising my arm. My goal in doing what I


do is quite opposite: it is that I try to raise my arm and fail. and that goal is a function again, not of what I am trying to do, but instead of the pro attitudes on which the voluntary attempt is based – in this case, an intention to try to raise my arm and fail.

Where actions of intentional road crossing, imagining, deliberating, trying and the like are concerned, the intentional pursuit of goals does seem to imply causation by prior content-bearing attitudes, and cannot be understood in other terms. the action’s goal-direction is not determined internally; that is, it is not determined by some object which the action has of itself and independently of causing pro attitudes. Instead the action’s goal-direction is provided from without, by the objects of pro attitudes which are causing it – prior pro attitudes by which the action is being motivated, and on the basis of which it is being performed.

decisions, on the other hand, are quite different. decisions are by their very nature goal-directed events. and some at least of their goals do seem to come from objects internal to the decisions – from the decisions’ contents – and not from any pro attitude cause. In which case decisions count as actions simply by virtue of being content-bearing events of a certain kind – not by virtue of any prior causes that they have.

on the face of it, it seems that decisions can perfectly well be taken without any sign that some prior pro attitude caused the decision – whether a pro attitude towards the decision itself or even towards its object. suppose one ruminative afternoon I decide to stand up to continue my walk, rather than stay sitting by the river or start to go home. until the moment of that decision there need have been no detectable pro attitude on my part whatsoever either towards going on with my walk or towards deciding to go on with it. Prior to the decision there need have been no evidence or hint of a passion or other prior inclination that was somehow causing or impelling me to take that particular decision, or indeed any decision at all. and that, I suspect, is precisely because there need have been no such prior pro attitude. sometimes we can just decide to do things, without any prior pro attitude pushing us so to decide.

not only can we take decisions without prior pro attitudes pushing us to decide as we do. It also seems that decisions so taken count as perfectly good actions nevertheless. suppose I do just decide to do something without any prior desire having caused me so to decide. Cannot my taking that particular decision still have been my own deliberate and intentional doing? should the apparent lack of any prior pro attitude pushing me into taking that decision lead me to doubt whether the decision really is my own doing – my own quite deliberately taken decision? a doubt so grounded seems absurd. the presence of such a pro attitude seems quite inessential to my belief that my own decisions are indeed my very own doing.

If this is right, decisions are indeed made actions through being the kind of content-bearing events that they are – not by virtue of prior causes that they have. and as actions, there is one immediate goal at which they must always be directed – namely, performance of the action decided on and intended. But this immediate goal is the very object which the decision has as a content-bearing event in its own right – the goal is simply what the decision is a decision to do, the object provided by the decision’s own content. so the immediate goal of a decision seems to come, not from the content of some prior pro attitude cause on which the decision is based, but from the decision’s own content and the kind of attitude to that content which a decision constitutes.


What makes decisions actions is not what causes they have – but the kinds of content-bearing events which they themselves are.

so, at any rate, I shall shortly be arguing when I develop the practical reason-based model. What distinguishes active goal-directed decisions from, say, passive goalless desire-formations, that model teaches, is not that decisions have causes which desire-formations lack. It is simply that decisions and desire-formations, taken in themselves, are different kinds of content-bearing event – different kinds of object-possessing event which are related to their objects in crucially different ways. these are differences in the internal constitution of the events themselves as content-bearing occurrences, and do not come from prior causes – though they may of course have implications for how the two kinds of event can be motivated and caused.

We have been considering a watered-down and qualified version of the voluntariness-based model. But the model in this qualified form is still objectionable. It is still tying the status of a decision or intention-formation as an action to something beyond the nature of the decision itself – to the decision’s being an effect of a prior pro attitude to its being taken. But by contrast to cases of genuine voluntariness, such as road-crossings, imaginings and the like, it really is not clear that our belief that our decisions are our own deliberate doing rests on any such belief about their causal history. decisions seem somehow to count as action, as our own deliberate doing, just through being the kind of content-bearing events that they are, irrespective of whether they have been caused by any desire or other pro attitude towards taking them.

The agency of decision-making is left qualified

those of us who do have an intuition that our decisions are actions – that what we decide is our own doing – do not normally see the agency of our decisions as a thinner or qualified version of the agency of the actions decided upon. That we take a decision to go to go for a walk, rather than deciding to stay at home – this is just as much something that we deliberately do, and just as much something which we determine for ourselves, as is our eventually going out rather than staying in. after all, normal, everyday decision-making seems to provide a locus of self-determination par excellence – a case of self-determination in its most vivid and immediately intuitive form. to decide what we shall do is, in general, precisely to determine for ourselves what we shall do. now if our ordinary decisions provide particularly clear cases of self-determination, and self-determination presupposes action – something self-determined must be our very own doing, and not something which merely happens to us – then our decisions must be as clear and unqualified cases of action as any. If the everyday taking of specific decisions to act offers a paradigm case of self-determination, how could the taking of those decisions exemplify the defining features of intentional agency, of what counts as our own deliberate doing, only marginally or qualifiedly? How could decisions exemplify those action-constitutive characteristics less fully than do the actions which decisions explain?

unfortunately, the agency of decision-making turns out, on the quasi-voluntary theory, to be just a heavily qualified analogue of agency in another more full-blooded form – the subsequent and fully voluntary actions which our decisions explain. and it is clear enough why. the quasi-voluntary theory still understands the capacity for agency exercised in decision-making as a capacity to do things on the basis of wanting


to do them. It is only because decision-making involves such a capacity that it counts as a case of genuine action at all. But, as the theory admits, the capacity for doing things on the basis of our wanting to do them which is exercised in our decision-making is limited or qualified – and deeply so compared to the capacity exercised in our fully voluntary action. For it still remains very much the case that, as Hobbes made so brutally clear, the mere fact that we want to take a given decision does nothing of itself to motivate us to take it; and decisions are never taken simply on the basis of our holding pro attitudes to taking them. Which is why decisions behave so much like other non-voluntary attitude-formations such as desire-formations, and so little like the genuinely voluntary actions which they explain. In the case neither of intentions nor desires are we generally able to form these attitudes on the basis of wanting to form them.

Like the voluntariness-based model proper, the quasi-voluntary theory still understands action in the same highly debatable way – as the exercise of an ability to do things on the basis of our wanting to do them. But at best and even on the quasi-voluntary theory’s own terms, decision-making involves such an ability only very restrictedly. such an ability is exercised in full-blooded form only in our performance of the voluntary actions decided upon. that leaves the agency of decision-making precisely a thinner, qualified version of the agency of voluntary action. If we want it to be anything more, we need a different model of agency entirely – not a model of it as doing what we want, but something quite different.

7. The Practical Reason-Based Model of Action

this alternative model of goal-directedness supports a theory of intentional action which is fundamentally non-causal. according to this theory, a capacity for goal-directed agency need not be a causal power of passions. to act intentionally need not be to do something on the basis of wanting to do it. We can explain what it is for someone intentionally to employ a particular means to an end – but without appealing to some passive cause which their goal-directed action must have had.

In deciding to do a, according to the model, I am intentionally forming an intention as a means to an end – the end of ensuring that I eventually do a. But where actions qua decisions are concerned, this goal-direction is to not be explained in terms of causation. It is to be explained rather in terms of the action’s having content and the way in which the action is directed towards that content. the goal-direction of a decision is a function, not of its actual causal history – there need be no actual cause that led the agent to decide to do a rather than B – but in terms of its constituting a certain kind of content-bearing event.

nor should this be a strange idea. after all, actions are said to have objects – just as much so as beliefs and desires. Just as we can talk of the object of a belief, we can talk of the object of an action – an object which is the action’s goal. and at least some goal-directed actions may come to possess their objects in the same way as do beliefs and desires and the events of their formation; not through being effects of prior attitudes directed at the same object, but through their possession of a content of their own. In other words decisions may count as actions through being events of attitude-formation in their own right – content-bearing events whose contents specify what object it is at


which they are directed, and whose kind determines in what way they are directed at that object. and this is the central insight of the practical reason-based model of agency.

How, in general terms, does the practical reason-based model conceive intentional action? the practical reason-based model of action says that to act intentionally is to pursue goals, and to pursue them on the basis of one’s capacity for rationality. Indeed, it goes further, and maintains that to act intentionally is precisely to exercise one’s capacity for reason or rationality – but in a practical or action-constitutive manner. and this is something which one can either do competently and with due care, and so rationally; or incompetently and recklessly, and so irrationally. the model has room for agency which is irrational.

Intentional action is a mode, a specifically practical mode, of exercising one’s capacity for rationality. By exercising a capacity for rationality I mean here, in this context, making, through one’s attitudes, a rationally appraisable response to an object of thought. the response might be purely mental or might involve bodily movement. either way, it will be a response directed at some object of an attitude, such as belief, desire or intention. one exercises rationality in this object-directed way when one forms attitudes in the first place. One’s response just is the formation of an attitude, such as a belief, desire or intention, directed at the relevant object. But one also exercises rationality in this way when one goes on voluntarily to do things on the basis of attitudes already formed – attitudes whose contents inform what one is doing, and provide it with an object at which it is directed.

In both cases we have an event or change directed at an object. the event or change, moreover, is one which is rationally appraisable, and which is to be justified and rationally appraised by reference to that very object – the standards of rationality and justification depending on what kind of response to the object is being made. What the practical reason-based of action does, is to define intentional action as one such kind of object-directed response – a response which is distinctively practical and action-constitutive in nature.

one can very well exercise one’s capacity for rationality is ways that are not practical or action-constitutive at all. there can be an exercise of one’s capacity for rationality which is purely theoretical. Here one responds to an object simply by coming to believe that it is true. and the standards which determine the rationality of such a response are going to be correspondingly theoretical in nature. For one’s response to be justified, and the belief to be formed rationally, the object, what is believed, must be sufficiently likely to be true.

But one can also exercise one’s capacity for rationality in a way that is fully practical – that really is action-constitutive. one does this when one responds to an object as one’s goal – as something which the exercise of one’s rationality, the very response to the object which one is making, is directed at attaining. For example, one can exercise one’s reason practically in this way when, on the basis of an intention to do a, one voluntarily does something as a means to ensuring that a is done. In performing this voluntary action, one is exercising one’s capacity for rationality. one’s voluntary action is a response directed at an object, the doing of a – an object provided by the content of the intention which is motivating the action. and one is responding to this object in a rationally appraisable way. the response is one for or against making which there can be justifications, and which can be made more or less rationally. And the response


is action-constitutive because the object responded to stands to the response as its goal – an object which the response is directed to attaining. What shows that the object is a goal? What shows this is the way the voluntary action’s rationality depends on this object at which it is being directed. the mode of dependence is precisely that appropriate to something done in order to attain the object. For the action to be performed rationally, that object must be sufficiently desirable, and the action must also be sufficiently likely to attain the object.

this is a fully practical or action-constitutive mode of exercising reason. the agent is using the exercise of his capacity for rationality in order to attain a goal. He is responding to an object of thought – and doing so precisely in order to attain that object of thought and make it real. and so the rationality of the response depends both on the sufficient desirability of the object to be attained, and on the sufficient likelihood that the response really will attain it. But this fully practical mode of exercising reason is not to be found in voluntary action alone. It is found in non-voluntary intention-formation too – and in equal and unqualified measure. Such a practical mode of exercising reason, I shall now argue, is also found when one decides or forms an intention to do a rather than B. For such an event is again an object-directed exercise of one’s capacity for rationality. It is a rationally appraisable response to an object – the object of the intention being formed. and again the object in question stands as a goal which one’s response to it is being used to attain. For the rationality of the intention-formation is determined in just the same way by its object as in the case of the voluntary action – by reference both to the desirability of the object and also to the sufficient likelihood that the response to it in question, the intention-formation, will help attain it.

so on this model of action, intentional agency can unproblematically occur in our decision-making and intention-formation – and as much so as in our voluntary action. decision-making no longer has to count as at best a marginal case of agency. For intentional agency is now held to occur as a goal-directed exercise of rationality. and this form of exercising rationality is as much to be found in our decisions to act, in as full-blooded and unqualified a form, as in the voluntary actions which those decisions explain.

Compare deciding to do a with another event in which reason is also exercised non-theoretically – the event of forming a desire to do a. If, as the practical reason-based model supposes, in decision-making reason is being exercised fully practically, in the sense of constituting fully intentional action, the same cannot be true of forming a desire. For while it is natural to assume that deciding or forming an intention to do a is an action – something which we deliberately do – it is equally natural to assume, as we have already seen, that forming a desire to do a is a passive occurrence. For we do not and cannot plausibly control directly what desires we form as we directly control how we decide to act. desires are states which are passive – which happen to us. Which is why the control which we have over our desires is always indirect, being exercised manipulatively through the effects of prior actions on what we want – as when we employ exercise as a means of increasing our desire to eat. But if taking a particular decision, as something which we can control directly, without having to use such prior manipulation, is an intentional, goal-directed action, whereas forming a particular desire


is not, wherein lies the difference? What makes intention-formation a case of action when desire-formation is merely a case of passion?

What is essential to the goal-directedness of a decision is the relation of the decision, as a content-bearing event, to the object which its content specifies. The decision is related to that object, namely the action or outcome decided upon, as to a goal to be attained through the decision’s occurrence. and what establishes that relation of the decision to its object as a relation of means to end to be attained thereby, is the way in which reason governs decision-making. For a decision to be taken rationally, it is not enough for the object of the decision to be desirable. There must also be some sufficient chance that taking the decision will lead to the attainment of its object. Reason treats decisions as goal-directed exercises of rationality – as events of exercising reason which are being employed as means to attaining their objects, and so as actions. and if reason treats decisions as just such goal-directed actions, that must be what decisions are.15

But what is true of decisions is not true of desire-formations. there is nothing irrational about coming to want something to happen while being very sure that if it happens, it will happen other than because one wants it to happen. We have and report such wants and desires all the time, and no one criticizes their rationality. Who do I want to win the cup? It is england that I want to win the cup – simply because, from my point of view, england’s winning would be highly desirable. Perhaps england is my national team; or perhaps it’s just that england’s winning would put my boss in a good mood. such considerations are quite enough to leave an england win a perfectly sensible thing for me to want. I need not also suppose, what is obviously false, that my wanting england to win will actually help england to win. What I happen to want will have no effect at all on england’s chances of winning. But that does not matter. since my desire for england to win is not something I am actually forming in order to get england to win, the fact that the actual outcome will be quite unaffected by the desire is quite irrelevant to the rationality of my forming it.

sometimes it is not just that I want something to happen which will happen, if it does, independently of my wanting it to. Its happening independently can be part of what I want, and very much matter to the desirability of what I desire. Parents, for example, might reasonably want their grown-up children to do the right thing – but to do the right thing for themselves, and quite independently of the fact that their parents want them to do it. suppose a parent is indeed sure that his grown-up children’s actions are by now entirely beyond parental influence. What the children end up doing is sure to

15 Let me emphasize that a decision’s status as an action has to come both from its mode of direction at its content and from its being an event or change. Compare decisions as goal-directed events with intentions as goal-directed states. Intentions are states which are directed at their objects just as are the decisions which form them – as goals to be attained through the intentions. But to be in a state of intending is not to be performing an action precisely because simply by intending one is not doing anything – one is not introducing any change. supporters of the voluntariness-based model, indeed all of us, need to make an analogous distinction in relation to the voluntary: between voluntary actions, which are events caused by a will that they occur, and voluntary states, which are states maintained on the basis of a will that they persist. to be in a latter state, such as deliberately remaining very still, is again not to be performing an action, precisely because nothing is being voluntarily done – no change is being voluntarily introduced. (My thanks to Michael Bratman for encouraging me to emphasize this point.)


be causally independent of that parent’s wants and attitudes. Given this belief, it is still both rational and natural for the parent very much to want his children to do the right thing – all the more so as this, if the child does it, will be its independent achievement. What, under the circumstances, the parent cannot rationally do is decide that his children will do the right thing. and that is precisely because reason governs the decision as a goal-directed action – as an exercise of rationality with a goal, namely ensuring the doing of the decision’s object, what has been decided. For it is always and obviously irrational to employ an action as means towards an end when one is sure the means employed will have no effect on the attainment of that end.

so the difference between the decision and the desire-formation is this. taking a decision is an object-directed exercise of one’s reason which is governed by reason as a goal-directed action – as a means employed towards an end, the end being that what is decided upon should occur. Hence a rationally taken decision must be a decision which is sufficiently likely to attain its object. The taking of a decision is a goal-directed doing, and is governed by reason accordingly. Whereas forming a desire is an object-directed exercise of one’s reason which is not governed by reason in the same way. In determining the rationality of a desire, we do not treat the object of that desire as a goal which the desire-formation is being used to attain. a rationally held desire need offer no prospect whatsoever of attaining its object. so while in decision-making, one is exercising reason practically or in a goal-directed and so agency-constitutive manner, in desire-formation the exercise of reason is not similarly practical.

often, when we take decisions between options, we assume that each possible decision is as likely to attain its goal, performance of the action decided upon, as any other. But this need not be so. Consider a case which I have discussed in greater detail elsewhere.16 I might presently prefer and want most of all to do a rather than B in the future – and given the risks involved in doing B, such a want may be very sensible. Perhaps I am a professional stuntman – and B is at some future date attempting a particularly dangerous stunt, while a is prudently refraining. But if, because of a likely intervening increase in my recklessness, there is at least a significant chance (if no more) of my subsequently abandoning any decision I now took to do a and of doing risky B anyway; and if, from the point of view of planning other matters, there is enough of a premium on settling now as surely as possible which of a and B I shall be doing – then notwithstanding the rationality of my present preference for doing a, it might be more sensible for me now to decide to do B. For that decision would be sure to attain its goal, which is causing me to act as decided. the point and purpose of a decision is to cause me to execute it. Which is why a decision that would be too ineffective at attaining this end can be a decision that I have reason not to take.

How likely a decision is to attain its object matters, then, to the rationality of a decision – as the likelihood of any action’s attaining its goal is relevant to the rationality of that action. But how likely a desire or preference is to attain its object is not similarly relevant to the rationality of forming that attitude. and so, precisely because of this, it can be rational to want and prefer things to happen which one cannot rationally decide should happen.

16 For a more detailed version of this example, see my “Purposive Intending”, Mind 100, no. 398 (1991): 343–59, and my The Psychology of Freedom, chapter 8.


this story told of why decisions are actions is entirely non-causal. the goal-directed exercise of rationality is constitutive of the decision event’s being an intentional action. But that goal-direction is not, as on the voluntariness-based model, a function of the content of a prior passive motivational cause. It is instead a function of the decision’s own content and of the mode of direction which the decision constitutes towards that content. decisions are content-bearing events of attitude-formation – they are events of intention-formation – whose occurrence, just as much as with belief-formations or desire-formations, constitutes a distinctive exercise, competent or defective, of the agent’s capacity for reason. and, in the case of decisions, the status of that object as a goal to be attained by the decision is given, not by any cause, but by the way that any decision’s rationality is determined in relation to its content and to the object which that content specifies. For the decision to be rational, the object of the decision has to be both sufficiently desirable, and sufficiently likely to be attained through the decision.17

decisions and intention-formations, then, are attitude-formations unlike any others. they are content-bearing events which are directed at attaining their own objects – at making their own contents come true. so their character as actions uniquely comes from their character as the kind of content-bearing events that they are. But that means that their status as actions is not derived from prior causes – just as the character of other events of attitude-formation is not derived from prior causes.

Belief- or desire-formations can surely sometimes occur uncaused – without there having been any prior cause to explain why it was a belief or desire with that specific content which arose; why it was to that conclusion rather than another that the agent leapt, why it was that urge rather than another which he came to feel. the character

17 as david Gauthier has reminded me, attitudes are often said quite generally to have aims. Beliefs are said to “aim” at truth; desires and intentions to “aim” at the good. so events of attitude-formation might also be seen as events with an aim – the same aim as the attitude formed. Does this mean that on my theory attitude-formation in general is a goal-directed action? I find talk of attitudes as having aims rather loose. to the extent that it is defensible, however, it must mean, not that attitudes or attitude-formations in general have goals as actions do, but something much weaker. It is simply that for each such attitude, there is a concept – be it a concept of truth or of goodness or of something else – which can be used to define a proper object of the attitude, and so what it is for the attitude to be rationally formed. thus beliefs are rationally formed only if their objects are likely to be true, desires and intentions only if their objects are likely to be sufficiently good or desirable. But this does not make belief- or desire-formations or attitude-formations generally all goal-directed actions. For in this sense of “aim”, goal-directed actions “aim”, that is, are properly directed, at something very special: not at truth, nor simply at the good – but at the good as something to be attained through the action. an action is rationally performed only if its object is both likely enough to be good, and likely enough to be attained through the action. that is what makes the object of an action a genuine goal – an end which the action is being used as means to attain. Belief may “aim” at truth. But belief-formation is not in general a goal-directed action because the rationality of forming a belief that p does not depend on the likelihood of some good’s being attained through forming the belief, but simply on the likelihood that the belief ’s object is true. so, for example, arriving at the truth about whether p may be desirable; but also – if the truth is unpleasant or trivial – it may not. either way, the issue is irrelevant to the rationality of forming a belief that p – which depends simply on whether p is likely enough to be true.


of the event as the kind of attitude-formation that it is – the event’s possession of content, and its mode of direction at that content – this is independent of prior causation. so too, decisions could perfectly well be taken and intentions formed uncaused – without there having been any prior cause pushing the agent to decide that way rather than another.

Given that decisions to do a are content-bearing events in their own right, it would be perverse to tie their very occurrence as such events with such content, to being effects of prior passive motivations towards doing a – motivations with exactly the same content, and which were pushing the agent to decide that specific way. Indeed, it would be more than perverse, but downright absurd. Content cannot generally have to be inherited from prior causes with the same content, otherwise content would never start. and so we should treat decisions on a level with other attitude-formations. We must regard a decision’s character, its being the particular content-bearing event that it is, as independent of the decision’s having a cause. and since it is precisely this character which makes it an intentional action, this means that intentional goal-directed agency can occur without any cause why the agent acted that particular way rather than otherwise.

suppose that an agent is capable of taking a decision, one way or the other, about whether to do a. If he does in fact decide to do a there need have been no prior cause pushing him to decide this way rather than the other. In taking the decision, he will have deliberately adopted a particular means to a particular end – but can have done so without any desire or other pro attitude having caused him so to act. there can, on this conception of agency, therefore, be action uncaused by any prior motivation. For if, as a matter of contingent fact, prior motivations do often influence our decisions, our decisions certainly do not owe their character as decisions or as actions to that fact.

of course, many decisions may in actual fact perfectly well have causes. In many cases, they may actually result from prior pro attitudes – if not to the decision itself, then to the decision’s object. But this fact is not essential to the nature of the decision as the specific kind content-bearing event which it is, governed by reason as decisions are – and so is not essential to its identity as an action. Which is why sometimes there is nothing to show that we have any pro attitude towards some voluntary action a until we actually do decide to do a. Go back to that ruminative afternoon when I decide to stand up to continue my walk, rather than stay sitting by the river or start to go home. We have seen that until the moment of that decision there need have been no detectable pro attitude on my part whatsoever towards going on with my walk. Prior to the decision there need have been no evidence or hint of a passion or other prior inclination that was somehow causing or impelling me to take that particular decision. and that is precisely because there need have been no such prior pro attitude. sometimes we can just decide to do certain things, without any prior cause pushing us so to decide. and when we do, our taking that particular decision can still have been our own deliberate and intentional doing.

so it seems wrong to explain our capacity for agency as a causal power of our passive motivations to produce effects leading to their satisfaction. a block to the satisfaction of our desires is not ipso facto a block to our action. What frustrated the satisfaction of our passive motivations or desires might be, not some obstacle from without, but our


own action – our very own decision unmotivated by any prior passion. What, for the voluntariness-based model is a sheer impossibility, was very much a possibility for past supporters of the practical reason-based model – for scholastic theorists of action such as aquinas and suarez.18 and it should also be a possibility for us.

8. Action and Self-Determination

Intentional action is no longer being explained as doing what one wants because one wants to do it – a model of agency which could apply, at best, only very qualifiedly to decisions of the will. Instead intentional action is being explained in terms which apply equally both to decisions to act and to the voluntary actions which those decisions explain. In either case, to perform an intentional action is to exercise one’s capacity for rationality – make a rationally appraisable response to an object of thought – in a distinctively practical way. to perform an intentional action is to make a certain kind of rationally appraisable object-directed change – a change which is goal-directed, being directed at its object as a goal which that change is being used to attain. the change may be made voluntarily – on the basis of a prior decision that it occur. or it may be made non-voluntarily – as in the initial taking of that decision.

Has something been left out of our account of intentional action? It might appear to some that something has been left out. For actions, it might be objected, are not merely goal-directed events. they are not merely events used to attain the objects at which they are directed. actions also, we naturally think, have a special relationship to their agent – a relationship which allows the agent to play a special role in the action’s performance. an action is an event which, in some sense, the agent himself determines to occur. and nothing so far has been said by the practical reason-based model to relate actions in that special way to their agents.

the practical reason-based model merely says that to act intentionally is to exercise reason in a practical way. But what distinguishes this mode of exercising reason from any other? not, it seems, any special relation to its agent. What distinguishes a practical mode of exercising reason from other object-directed exercises of rationality is the special way in which it is related to its object – as to a goal – and not any special way of being determined by the agent.

The voluntariness-based model is rather different. The terms in which it defines goal-directedness do not supply goal-directedness alone. they also involve a special relationship between agent and action. an action counts as a goal directed event precisely by virtue of the special way in which its agent determines it to occur. an action is an event which its agent causes to occur – and which he causes to occur by willing, by wanting or deciding, that it occur. actions are events which are determined to occur by their agent’s own will. It may therefore look as though the voluntariness-based model is doing more justice to something that makes action distinctive among other events – its agent-determined nature. However, as we shall now see, the advantage of the voluntariness-based model on this point is merely apparent.

18 see my “suarez, Hobbes and the scholastic tradition in action theory”.


First, it is true that, as we ordinarily conceive things, action can indeed be agent-determined – something that the agent has determined for himself. agents, or at least adult human agents, have a capacity for self-determination – a capacity which they can exercise in and through how they act. But this element of self-determination, though peculiarly linked to agency – self-determination is only ever exercised in and through how we act, not in and through what merely happens to us – is by no means essential to the very occurrence of action. For it seems that actions, even fully deliberate and intentional actions, can sometimes be imposed on their agent. an agent may, for example, be driven to act by a compulsive desire – a desire the compulsive nature of which precisely prevents its being the agent himself who determines how he will act. What distinguishes action from other events, then, is not being determined in any distinctive way to occur by its agent – that may or may not prove to be the case – but being an event which is goal-directed, and done as a means to ends. a theory of intentional action must therefore define action in terms which entail its goal-directedness. But it need not define action in terms which entail its being distinctively agent-determined.

secondly, goal-directedness and self-determination, even when both are present, are two quite separate features of agency. The first involves a distinctive relation between an action and its object, and second between an action and its agent. and these relations are obviously very different. they relate the action to two quite different things. It is simply not obvious why the story which we tell about goal-directedness, a relation which any action has to the object at which it is directed, should also be a story about self-determination and the special relation which actions can but need not have to their agents. there is no reason to expect a theory of action to do what the voluntariness-based model is attempting to do, which is to explain goal-directedness and determination by the agent in terms of one and the same factor – this one and the same factor, according to the story which the voluntariness-based model is telling, being voluntariness.

Thirdly, it is a fundamental mistake anyway to try to define self-determination in terms of voluntariness. We do have a capacity to determine for ourselves how we act – or so we ordinarily think. But it is not clear that this capacity need take the form of a capacity to act as we will. Historically, it is true, many philosophers have tried to understand self-determination precisely as a kind of voluntariness – as a capacity to do things on the basis of a will to do them. Harry Frankfurt is the most prominent defender of such a theory of self-determination today.19 the outline of the theory is familiar. First we explain what it is for an agent properly to be identified with his will – what it is for his motivating attitudes truly to count as his own. then, when conditions on the identification of the agent with his will are met, self-determination can be explained in terms of voluntariness. For the agent to determine that he performs an action is then simply for him to perform it on the basis of the will that is his own – and so on the basis of his willing, deciding or on balance wanting, so to act.20

19 see especially Frankfurt, The Importance of What We Care About.20 Velleman, “What Happens When someone acts?”, also follows Frankfurt in applying an

elaboration of a voluntariness-based account of action to provide an account of self-determined action – what Velleman terms “human action” par excellence. For Velleman certain kinds of pro


But this way of understanding self-determination is inviting only if the voluntariness-based model of action is true, so that all the intuitively self-determined actions that we perform are indeed performed by us voluntarily. the problem is that there are intuitively self-determined actions which we perform – actions which seem to us as self-determined as anything we do – which we do not appear to perform voluntarily; and which indeed we cannot possibly perform voluntarily, at least in any unqualified sense. these are our decisions to act – actions which we precisely do not perform simply on the basis of our deciding or wanting to perform them. the capacity for self-determination which we ordinarily suppose ourselves to possess cannot therefore be a capacity for voluntariness. It cannot be voluntariness if the intuitive paradigm of self-determined agency is to be found in a form of agency – that of the will itself – which is less voluntary than any other.

self-determination, as we ordinarily understand it, is not voluntariness or doing what we want, but something very different. our most natural conception of self-determination is instead a conception of it as freedom. By freedom I mean its being up to us or within our control which actions we perform. If I possess freedom, it can be up to me whether I raise my hand or let it fall, open my eyes or close them, go for a walk or stay at home. Which I do is within my control; and whichever I end up doing, I was free to do otherwise. the idea of freedom or control is the idea of a power over our action – a power which makes alternative ways of acting available to us.

these two conceptions of self-determination, as freedom and as voluntariness, are, it seems clear, quite different. Freedom is the idea of a control over alternatives, over which actions we perform. By contrast, the idea of voluntariness says nothing about having any such control over how we act. It is instead the idea of our actions having a certain special explanation. Voluntary actions are actions which are explained by a motivational response to the desirability of performing them – by, in other words, our own prior will to perform them.

and the case of decisions shows that it is in terms of freedom that we ordinarily understand self-determination, not in terms of voluntariness. If our decisions can be as self-determined as any actions which we perform, that cannot be because they are as voluntary as any actions we perform. For they are clearly not. so if we count as determining what decisions we take, it is not through taking decisions voluntarily, but through taking them freely. our decisions are determined by us in so far as when we take them, we are exercising a control over which decisions we take. We can determine how we decide, not because we can take decisions as we will or want – but because it is up to us which decisions we take, and we are free to act otherwise.

The voluntariness-based model certainly defines agency in terms of a special relation between actions and their agents. But this is not a point in its favour. actions, it is true, can stand in a special relation to their agents – when agents are genuinely self-determining in the exercise of their agency and it is those agents themselves who determine whether their actions occur. But such a relation is not essential to intentional action as such. and when it is to be found, the relation has nothing to do with voluntariness, and

attitude to acting voluntarily are supposed to play “the role of the agent” in determining his action; self-determined action is voluntary action motivated by these particular pro attitudes. But, as I shall argue, self-determination does not consist in any form of voluntariness.


everything to do with something very different – the exercise of freedom as a power over alternatives. to understand the common-sense psychology of self-determination, then, we need, not a theory of action as voluntariness, but a theory of action as a practical mode of exercising reason. and we need then to supplement that model of action with a theory of freedom as something exercisable in and through the practical exercise of reason. But such a theory of freedom is a task for another time.21

21 It is developed in my The Ethics of Action: Self-Determination.

Intentions and Two Models of Human Action

Documents

Transcript of Intentions and Two Models of Human Action