1. Freedom and Scepticism
Ethical theory has often appealed to the idea that as agents we have a power of self-determination – a power to determine our action for ourselves. The immediately natural conception of self-determination seems to be as freedom, a power that leaves it up to us whether we do A or refrain – a power of control over which actions we perform. Central to the idea of freedom, then, is power over alternatives. This involvement of alternatives is picked out by the ‘up to me whether’ construction, which is completed by specification of alternatives by way of actions and outcomes within my power; freedom is the power to determine for ourselves which alternative occurs.
Belief in our possession of such a power is especially naturally reported by talk of what we do being ‘up to us’. And put in such terms, belief in the power is very general. It is easy to think of its being up to me whether I go out or stay in, raise my hand or lower it.Footnote 1 But there has long been scepticism about the reality of such a power among philosophers. This scepticism often begins nowadays from the supposed link of our conception of freedom to incompatibilism. Incompatibilism about freedom is the doctrine that the possession and exercise of the power is incompatible with the causal determination of our action by prior occurrences outside our control. Sceptics such as Galen Strawson have viewed incompatibilism as supposedly both implied by our present concept of freedom, and dooming that concept to non-application.
Strawson allows for a weak kind of compatibilistically acceptable self-determination involving a power of our passive motivations to determine how we act. This power is what Thomas Hobbes called voluntariness – the power of motivations, of desires or other pro attitudes towards action, to cause the action motivated. Hobbes took voluntariness to constitute the very definition of human action, which by its very nature (in his view) involved doing things as an effect of our wanting to do them. Strawson goes further, and takes voluntariness to provide not just action, but a metaphysically unproblematic form of self-determination – a power on our part to determine action:
A naturalistic explanation of this sense of self-determination would connect it tightly with our sense, massively and incessantly confirmed since earliest infancy, of our ability to do what we want in order to (try to) get what we want, by performing a vast variety of actions great and small, walking where we want, making ourselves understood, picking up this and putting down that. We pass our days in more or less continual and almost entirely successful self-directing intentional activity, and we know it. Even if we don't always achieve our aims, when we act, we almost always perform a movement of the kind we intended to perform, and in that vital sense (vital for the sense of self-determining self-control) we are almost entirely successful in our action.
This gives rise to a sense of freedom to act, of complete self-control, or responsibility in self-directedness, that is in itself compatibilistically unexceptionable, and is quite untouched by arguments against true responsibility based on the impossibility of self-determination. But it is precisely this compatibilistically speaking unexceptionable sense of freedom and efficacy that is one of the fundamental bases of the growth in us of the compatibilistically impermissible sense of true responsibility. To observe a child of two fully in control of its limbs, doing what it wants to do with them, and to this extent fully free to act in the compatibilist sense of the phrase, and to realise that it is precisely such unremitting experience of self-control that is the deepest foundation of our naturally incompatibilistic sense of true-responsibility-entailing self-determination, is to understand one of the most important facts about the genesis and power of our ordinary strong sense of freedom.Footnote 2
Bernard Williams too talks of our primitive conception of self-determination as being Hobbesian voluntariness, or ‘action unimpeded’ in the execution of one's will; a conception which he describes as ‘seed’ to the ‘plant’ of any more developed or demanding conception of self-determination as incompatibilist freedom.Footnote 3
The supposedly original conception of self-determination locates the power as a power of our motivations, conceived as passive in that they precede any doing by us, to cause actions that would satisfy them. And of course this conception of self-determination is entirely compatibilist. The operation of the power is not only consistent with the causal determination of our action by occurrences outside our control, but actually involves just such causation. The introduction of an incompatibilist conception of the power requires, then, a wider transformation in our understanding of it. The power must no longer attach to passive desires, but be exercised by us independently of our desires. And if action must by its very nature involve motivating attitudes, then these attitudes cannot be passive determinants of what we do, but arise as themselves aspects or parts of what we do. The motivation for our action is relocated to a mythical and inherently self-determined faculty of will – a faculty that involves inherently active motivations of decision that operate apart from ordinary desire.
We tend to think we have a will (a power of decision) distinct from all our particular [passive] motives.Footnote 4
If freedom as we have currently come erroneously to conceive it, as a power exercised apart from ordinary desires, does not really exist, how has such a conception of power entered our psychological belief? An obvious line of thought involves an error theory – the intrusion of defective ethical conceptions into psychology. The culprit is some erroneous model of blame and moral responsibility. We begin with incompatibilism, not about freedom, but about moral responsibility. The actions for which we are morally responsible must not be determined causally by prior conditions for which we are not responsible. So a new power is introduced to base that moral responsibility that likewise rules such causal determination out – a power of libertarian or incompatibilist freedom.
This theory of a development in the understanding of self-determination – a development from an unobjectionably compatibilist ‘seed’ to an extravagantly incompatibilist ‘plant’ – faces a number of questions. First, the transformation is hard to understand as a development of one conception of self-determination from another, just because the initial power, Hobbesian voluntariness, looks so unlike genuine self-determination. Instead of determination of action by the self, we have determination of an action by a passive motivation – a desire that happens to the agent and that is not of their own doing. Indeed as we shall see, Hobbes certainly did not suppose that voluntariness involves a form of self-determination – a power by which we determine for ourselves what we do. He saw voluntariness as nothing more than the determination of action, as one motion in matter, by passions as just other motions in matter. For reasons that we shall be examining, he denied the very possibility of self-determination, and gave it no role in his ethical theory.
Secondly, we might question the centrality given to incompatibilism as an element of our ordinary conception of freedom. Incompatibilist intuitions are very widespread. But they are not universal. And they arguably have to do, not with the nature of freedom as a power in itself, but with the power's relation to – its compatibility with – another case of power, namely antecedent causation. More central to our conception of freedom or control, and what distinguishes it as freedom, just is its involvement of alternatives – the very up-to-us-ness of how we act. The particular way in which alternatives are involved in our idea of freedom, it will emerge, distinguishes freedom as a power very sharply from Hobbesian voluntariness. But the way we understand freedom to involve alternatives may have very little to do with incompatibilism. And it may not be easily explicable just in terms of some intrusion of ethical theory into psychology. The way freedom involves alternatives certainly feeds into our ethical belief. But it may not be entirely ethical in origin.
2. Self-determination and moral responsibility
We can criticize someone ethically or by standards of reason without implying their possession of some power of self-determination. We can criticize someone as selfish without implying any power on their part to determine that they are selfish or unselfish. Similarly we can criticize someone for some failure of reason without implying that they had a power to determine for themselves whether the failure occurred. Failures of theoretical reason, for example, may occur without any power on the believer's part to determine or prevent their occurrence. A degree of selfishness, or some imperfection in one's capacity for reason, can perfectly well be built in.
For this reason, philosophers from Hobbes and Hume to our own day have striven to detach ethics from any presupposition of some metaphysically problematic power of self-determination by assimilating moral blame, the holding of someone as morally responsible, to some more general form of ethical criticism. For Hume, such blame is no more than a pronouncedly negative evaluation – an evaluation that can as much be of passive attitudes and dispositions as of actions. For Scanlon, moral blame is rational criticism that is distinctive only as coming with a sting, a special significance for us, not as presupposing any distinctive, moral responsibility. And precisely because it occurs simply as rational criticism, Scanlonian blame, too, can be for attitudes as much as actions:
…“being responsible” is mainly a matter of the appropriateness of demanding reasons…For this reason, one can be responsible not only for one's actions but also for intentions, beliefs and other attitudes. That is, one can properly be asked to defend these attitudes according to the canons relevant to them, and one can be appraised in the light of these canons for the attitudes one holds. The “sting” of finding oneself responsible for an attitude that shows one's thinking to be defective by certain standards will be different in each case, depending on our reasons for caring about the standards in question. But the basic idea of responsibility is the same.Footnote 5
But moral blame is more distinctive than that. It involves more than criticism of someone as defective in their actions or attitudes. Blame does not just report a deficiency in the person blamed. It further states that this deficiency was the person's fault – that they were ‘to blame’ for it. The attribution of something not only as a fault, but as someone's fault, the fault of the person blamed, is essential to anything recognizable as genuine blame.
For suppose someone is subject to ordinary rational criticism. Suppose, for example, that they have committed some error of reasoning. It is always a further question whether that they made this error was their fault. Are they responsible and to blame for the fact that they made it? – or did they make the mistake through no fault of their own? They were certainly being foolish or less than sensible; it is, after all, their reasoning which was bad. But we can still ask whether it was through their own fault that they reasoned incorrectly. The question of one's responsibility for one's attitudes remains open, even when one's rational appraisability for those attitudes is admitted. In which case Scanlon must be wrong. The kind of responsibility assumed in blame does not reduce to the appropriateness of rational appraisal.
R.M. Adams similarly confuses moral blame with more general ethical criticism. In his ‘Involuntary sins’ Adams has suggested that when people are criticized for being selfish, such criticism amounts to blame:
Perhaps for some people the word ‘blame’ has connotations that it does not have for me. To me, it seems strange to say that I do not blame someone though I think poorly of him, believing that his motives are thoroughly selfish. Intuitively speaking, I should have said that thinking poorly of a person in this way is a form of unspoken blame.Footnote 6
But the selfishness of someone's motivation does not of itself settle the question which is raised in blame – namely their responsibility for the motivation they possess. Their selfishness is one thing. It is still a further question whether their possession of such a character is their fault. There is no inconsistency at all in criticizing someone as having a selfish character, while wondering or doubting whether their possession of this character really is their fault. Just as someone can be criticized as poor at reasoning without this being supposed to be their fault, so they can be criticized as selfish without this being supposed to be their fault.
In moral blame, the agent is not merely criticized for some fault in their actions and attitudes. The fault is put down to them as their fault. And this putting a fault down to someone as their fault is essential to the content of blame. And it is at this point that a power on the agent's part comes in. Why should a fault on the agent's part be their fault unless they had a power to determine its occurrence – or a power to prevent its occurrence which, though possessed by them, they failed to exercise?
3. The Nature of Power
Many philosophers write as if power were a rather uniform phenomenon. It is often claimed, for example, that power is by its very nature a causal phenomenon – a claim that, as we shall see, was importantly made by Hobbes, but which is still very current today:
In the first place, the notions of power or disposition are already causally laden notions and it can thus reasonably be argued that unless one already has a grasp of causation, one cannot have a grasp of power. Powers, indeed, are often called causal powers.Footnote 7
Certainly, one very intuitive case of power is the very familiar kind that is involved in causation, and that is possessed and exercised, not by causes and effects indifferently, but specifically by causes. This is the power of stones to break windows or the power of fire to boil water – the power that causes have to produce their effects. But it is useful to step back, and raise the question why causation itself is so widely viewed as involving power – and, more specifically, a power possessed and exercised by causes over what they affect?
Power involves a kind of capacity. Causal power constitutes, after all, a capacity to produce effects. But, of course, it is not the mere presence in them of a capacity that makes it true that causes possess power. And that is because the idea of a capacity extends far wider than that of power. For example, there are capacities not to cause and affect, but to be affected. But the capacity to be affected hardly constitutes any kind of power over anything, and the process of being affected is hardly the exercise of power. The contrary is true: to be affected is to be subject to power that is possessed and exercised by something else. Contrast my view with John Locke's.Footnote 8 Locke divides power into active and passive. Active is defined as the power to make a change, passive is the power to receive it. As an account of power this is certainly defective. For, of course, Locke ignores powers to prevent change from occurring. But more importantly, Locke's ‘passive power’ involves the opposite of any exercise of genuine power. It is a form of powerlessness – subjection to the power of another.
Power, then, is a very special capacity. And what, I conjecture, is common to power in all its forms is a capacity to produce or, at the upper limit, to outright determine the occurrence or non-occurrence of outcomes. It is this capacity to produce or determine outcomes that causes possess, but which their effects lack. Causes determine the occurrence of their effects, and not vice versa.
If a power is a capacity to produce or determine outcomes, a causal power must be a capacity to produce or determine outcomes causally. And that might suggest the possibility, at least at the conceptual level, of power that is not causal. A power that is not causal is going to be a power the exercise of which determines outcomes, but without determining them causally. And that in turn raises the question of whether the power involved in self-determination is conceived by us as a straightforwardly causal power. Is it conceived as determining outcomes as ordinary causes do?
4. Hobbesian Scepticism about Self-determination
Thomas Hobbes denied the very possibility of self-determination. He mounted his assault on the very idea of self-determination as part of a radical programme to detach ethical and political theory from reliance on the notion. The moral and psychological theory for this programme was principally expounded by Hobbes in a dialogue – the Questions concerning Liberty, Necessity and Chance – with an opponent, a defender of Aristotelian scholasticism, the Anglican Bishop Bramhall.Footnote 9
How does Hobbes propose to detach ethics from self-determination? Some of the time, Hobbes does what Hume will do later as well – which is to treat moral blame as no more than negative evaluation:
[Why do we blame people?] I answer because they please us not. I might ask him, whether blaming be any thing else but saying the thing blamed is ill or imperfect...I answer, they are to be blamed though their wills be not in their power. Is not good good and evill evill though they be not in our power? And shall I not call them so? And is that not praise and blame? But it seems that the Bishop takes blame not for the dispraise of a thing, but for a praetext and colour of malice and revenge against him that he blameth.Footnote 10
In other contexts Hobbes seems to allow for a distinctive responsibility for how we act:
The nature of sin consisteth in this, that the action done proceed from our will and be against the law.Footnote 11
But the responsibility here involves a kind of legal responsibility – according to a view of that responsibility which avoids appeal to self-determination. Holding someone responsible, in Hobbes's view, seems to involve no more than holding them to sanction-backed directives on the voluntary – something that presupposes no more than their rational responsiveness to such directives. To be morally responsible, on this model, we have merely to be legally governable. But, for Hobbes, that only requires that we be capable of performing or avoiding actions on the basis of a desire so to do, as a means to avoiding sanctions. And this presupposes nothing more than action in the form of Hobbesian voluntariness – something that Hobbes thought had nothing to do with self-determination.
Freedom, for Hobbes, consisted not in a power of agents over alternatives, but in something quite different: namely, in an absence of obstacles to the satisfaction of an ordinary causal power – the power of a motivation to cause its satisfaction. Freedom consists, for example, in the absence of external constraints, such as chains, or sanction-backed laws, that might prevent my desires from causing movements by me that might satisfy them:
Liberty is the absence of all impediments to action, that are not contained in the nature, and in the intrinsecal quality of the agent.Footnote 12
Hobbes's opponent Bramhall was effectively a spokesman for the ethical and psychological theory of the late scholastic Francisco Suarez. And it is Suarez who is the ultimate target of much of Hobbes's writing in this area. In Suarez the idea of freedom really is the idea of a special kind of power – a power that, though still for Suarez a form of causation, is causation of a quite distinctive kind. Freedom is causal power in what he describes as contingent form.Footnote 13 As a free agent I am not a necessary cause as causes in wider nature are – a cause that under any given circumstances can operate in only one way. A massive brick that strikes a window can determine but one outcome – that the window breaks. Whereas, by contrast, I have a power, freedom, by which in one and the same set of circumstances I could equally well determine any one of a range of alternative outcomes. So under a given set of circumstances I have the power, say, to lower my hand or to raise it – and my nature as possessor of the power leaves it contingent how I will exercise it, and so which action I shall perform.
Hobbes denies that such a contingent power is possible, because it is unrecognizable as causal power. For Hobbes's scepticism about freedom is based on a clear view of the only form that power can take in nature. The only possible form that power, the capacity to produce or determine outcomes, can take, in Hobbes's view, is as ordinary causation – the kind of power that bricks, or motions involving them, possess and exercise to break windows.
It is tempting to think that Hobbes's problem with freedom is mainly with what I shall call multi-wayness. Freedom or control of what we do involves alternatives. To have control of whether one does A is to be capable of determining either that one does A or that one refrains. And it is very natural to view this control as a single power that could be employed in more than one way – hence multi-wayness – to produce either the outcome that I do A or the outcome that I refrain. That is the nature of control as a power: to leave it up to me which I do, and to be employable in doing either. Hobbes's case, on this reading of him, is simply that there cannot be such a thing as a multi-way power – a power that can, under a given set of circumstances, be used in more than one way, to produce one of a variety of outcomes.
However we should beware of this tempting assumption. It should not be assumed that freedom, understood as its being up to me to determine a range of alternatives, need involve multi-wayness as just defined – a single power employable in more than one way, to produce any one of these alternatives. And in any case, I shall suggest, even if freedom did not involve multi-wayness, it would still involve a form of power which Hobbes denied.
Moreover, it seems there could be cases of multi-way power that were not at all like freedom, but much more like (possibly slightly unusual) cases of ordinary causal power. True, much ordinary causation seems not to be multi-way – as the case of the brick hitting the window reminds us. Causation here seems to take one-way form. In a given set of circumstances, when the massive brick hits the window, the brick or its motion can exercise its power to produce but one effect – that the window breaks. But need this be true universally? Can there not be probabilistic causes with a power that could, under certain circumstances, operate in more than one way, to produce a range of outcomes? Perhaps the power of one particle to accelerate another could produce in the other particle, with some probability, one acceleration; or perhaps, with another probability, another slightly different acceleration instead. This would still be recognizable as ‘ordinary’ causal power. And it would not involve the causing particle's possession of freedom. It would not be up to the particle which acceleration it produced; that would not be something that the particle ‘determined for itself’.
Hobbes was, of course, a determinist. Probabilistic causation is not a possibility on his metaphysics of causation. He thinks that a cause's power operates, under any given circumstances, to produce but one outcome. But the issue of multi-wayness – the possibility of a causal power's operating under given circumstances in more than one way, to produce more than one possible outcome, is not what is fundamental to Hobbes's scepticism about the very reality of freedom, or indeed of self-determination in any form at all. Hobbes's scepticism has more to do with something that can be detached from multi-wayness, and that radically distinguishes freedom from ordinary causation. I shall call this factor contingency of determination; and it has to do with how the possessor of a power, such as a cause, determines an outcome when it does.
In Hobbes's view, if an entity has the power to determine a specific outcome, and the circumstances required for the successful exercise of the power are all met – then the power must be exercised. The determining entity's very presence, with its power, must necessitate the occurrence of the outcome it has the power then to determine. It follows on this view that an entity cannot really possess the power to determine, under one and the same set of circumstances, more than one alternative outcome. For an entity really to be capable of determining each outcome, Hobbes argues, it must simultaneously produce each outcome. Referring, abusively, to Suarez's contingent cause as an ‘indetermination’, Hobbes writes:
But that the indetermination can make it happen or not happen is absurd; for indetermination maketh it equally to happen or not to happen; and therefore both; which is a contradiction. Therefore indetermination doth nothing, and whatsoever causes do, is necessary.Footnote 14
Suarez was right about one thing. Contingent determination is part of our ordinary understanding of freedom, and distinguishes freedom from ordinary causation. In the case of freedom, the power-bearer may have the power to determine the occurrence of a particular outcome, and all the circumstances required for the power's successful exercise may be met – without the power being exercised to produce that outcome. Freedom can involve the power to determine alternatives, only one of which can actually be produced, only because this is so.
Suppose by contrast an ordinary cause has under given circumstances, the power to produce a range of possible effects. The cause is probabilistic: any one of these effects might with some probability occur, or it might not. In such a case the cause does not count as determining the effect that it produces. A probabilistic cause at most influences the occurrence of that effect, but without determining it in a way that removes all dependence of the final outcome on simple chance. Whereas we do think of the free agent as determining that he does what he does, but without the action's performance being guaranteed just by his presence as a free agent with the power then to determine it.Footnote 15
Contingency of determination distinguishes a free agent from any cause – including a probabilistic cause. But so too does something else – something which involves not the power's relation to outcomes, but the agent's or power bearer's relation to the power.
Consider again ordinary causes. Either their operation is predetermined by the very nature of the power and the circumstances of its exercise: in those circumstances their power is to determine one particular outcome, an outcome which they will then produce. Or, as in the case of probabilistic causes, how the cause will operate is undetermined, that is, dependent on mere chance. But what seems importantly to distinguish freedom, as ordinarily conceived, is that this is not so. It is neither predetermined nor merely chance and undetermined which way a free agent exercises their power. The agent determines for himself how he exercises his power. It is up to the agent whether he exercises his power to produce this outcome or that. If the power of freedom is indeed multi-way, a power employable in more than one way to produce more than one outcome, then in relation to that power there is what we might term a freedom of specification: it is up to the bearer which outcome the power is exercised to produce.
Hobbes is very well aware of this element to our conception of freedom as a power. The idea of the agent's determining his exercise of the power is arguably central to self-determination – to the very idea of determining outcomes for oneself. In Hobbes's view, this idea of a determination of how the power is exercised is viciously regressive.
And if a man determine himself, the question will still remain what determined him to determine himself in that manner.Footnote 16
The very idea of self-determination, for Hobbes, is incoherent. And that is because it viciously involves the idea of a power to determine, the exercise of which has first to be determined.
But it is not obvious that Hobbes is right about the regress. The regress is vicious only if the way in which the exercise of the power is determined – to produce this outcome or that – involves a prior exercise of power distinct from the exercise of the power determined. But this is not obviously what we ordinarily suppose.
There is in the case of freedom a conceptual distinction between (a) the power's relation to outcomes – the power can operate to produce more than one outcome – and (b) the power's relation to me, namely that I determine for myself what way it operates. But we do not suppose there to be any corresponding ontological distinction between distinct exercises of power – one exercise of power by me to produce outcomes, and a prior exercise of power to determine that exercise. Multi-wayness and determination of the mode of exercise by me are simply conceptually distinct features of a single exercise of control. In exercising control over outcomes I ipso facto determine for myself how the control is exercised. That is what control is – a power to produce outcomes the manner of exercise of which I determine for myself. In one and the same exercise of power I produce one outcome rather than another, and I determine how the power is exercised.
Scepticism about freedom has often been directed at freedom conceived in incompatibilist terms. We saw this in Galen Strawson; but such scepticism was strongly expressed long before, by writers such as David Hume. Hume thought that incompatibilist or libertarian freedom was impossible, and impossible because by detaching freedom from prior causal determination – from prior necessity – it reduced freedom to nothing more than the operation of chance. To remove prior necessity – to reduce the influence on the agent of prior causes or remove causal determination – is to leave the final outcome dependent on mere chance.
… liberty, by removing necessity, removes also causes, and is the very same thing with chance.Footnote 17
Certainly with ordinary causes, if it is not determined in advance what effect a given cause will produce, the outcome must indeed depend, to a degree, on simple chance. If causation is the only power in play, take away prior necessity and you certainly are left with mere chance – chance and nothing else. So to the extent that a cause is merely probabilistic, what effect it will produce depends to a degree on mere chance. But to suppose that in all cases the alternative to necessity is mere chance is to assume that there can be no such power as freedom as we ordinarily understand it – a power involving contingency of determination. For even if the outcome is not already causally predetermined – so that it is initially chancy how the agent will act – freedom, as ordinarily understood, may prevent the final outcome from depending on simple chance. Freedom allows the outcome still to be determined – by the agent. It is arguable, then, that the real target of Hume's scepticism is not freedom conceived in incompatibilist terms, but freedom in a form that involves contingency of determination.
And contingency of determination is not a specifically incompatibilist notion. To say that a power involves contingency of determination, is not itself to say anything about the power's compatibility with causal determinism. All that contingency of determination expressly asserts, is that an agent might possess the power to determine an outcome in the circumstances – and yet still not exercise the power to produce that outcome. It is quite another question whether, compatibly with his possession of the power, the agent's exercising or failing to exercise it could itself be causally determined.
Where freedom is concerned, there are two forms of scepticism. There is scepticism from the supposed conceptual truth of incompatibilism. But there is also scepticism from freedom's basic identity as a power over alternatives distinct from ordinary causation. The second scepticism denies the very possibility of such a power, not because of any incompatibilist theory of it, but because as ordinarily understood, as a power over alternatives, freedom is too radically unlike the causation found in wider nature. It is this second form of scepticism that may prove the most serious. Indeed, it looks as though, as in Hume's case, some of the first kind of scepticism might really depend on the second. Freedom is indistinguishable from chance only if there can be no such thing as a power that is distinct from ordinary causation – a power to determine alternatives that can operate even in cases where the final outcome is initially undetermined causally.
5. Freedom and Reduction
We have been examining freedom as a basis of moral responsibility, and what seems distinctive, and metaphysically problematic perhaps because so distinctive, about power so conceived. But need freedom be so metaphysically distinctive? What of theories of freedom that instead of distancing freedom from ordinary causation, appeal to ordinary causation to explain what freedom is, and especially its involvement of alternatives?
Classical post-Hobbesian English language compatibilism adopted such an account. In that intellectual tradition, self-determination as power to determine alternatives, and so to determine otherwise, was not dismissed as Hobbes dismissed it, but was instead simply constructed from Hobbes's theory of action – out of voluntariness. Hobbes had rejected this as a possible theory of self-determination: but the English language philosophical tradition that followed him, and which was deeply influenced by him in other respects, did not. Was this whole compatibilist tradition misconceived; or should it have followed Hobbes's more radical example of abandoning self-determination altogether?
Classical English-language compatibilism explains freedom in terms of a combination of distinct cases of voluntariness as involving ordinary one-way causal power. So the account of freedom is ontologically reductive. The power to determine alternatives is not a single multi-way power that can be used to determine more than one outcome, but is explained as constructed out of a combination of distinct one-way powers. And the account is conceptually reductive. No appeal is made to any conceptually primitive notion of freedom or up-to-us-ness to characterize the agent's relation to the power.
For it to be true that it is up to me whether I do A, on this theory, the following must then be true:
if I were motivated to do A, that would causally determine me to do A.
if I were motivated to refrain, that would causally determine me to refrain.
The ontological reduction of freedom's involvement of alternatives in terms of a combination of distinct cases of one-way causal power is very important in its implications for ethics and moral responsibility. It explains why in our own day so many philosophers, and principally Harry Frankfurt, have come so readily to assume the irrelevance of freedom to moral responsibility. Defining freedom in terms of voluntariness profoundly changes our view of what is going on when an agent exercises freedom to determine what he does, and in a way that detaches moral responsibility from freedom as a power to determine alternatives by way of action – as involving a power to act otherwise.
The distinctive feature of freedom or control as a power over action, we have supposed, is its involvement of alternatives. To be free is to have control over whether one does A or not. And I have said that it is natural to understand this involvement of alternatives in terms of multi-wayness – a single power that could under given circumstances be used to produce more than one action or outcome. One and the same power could be exercised either to do A, or to refrain. To do A through exercising this power is to do A through the exercise of a power to do otherwise – a power, control of which action one performs, that could equally have been used to omit doing A. To possess that power of control with respect to an action's performance is, equally, to possess it with respect to the action's omission.
Whereas voluntariness is obviously quite different. To do A voluntarily is to do A because one decides or wants to. But this capacity to do A on the basis of wanting to do it would in no way be involved in refraining to do A. Voluntarily to refrain from doing A would involve the quite distinct power to refrain from doing A on the basis of wanting to refrain. And the powers really are distinct, in that each power can be possessed without the other. To use Locke's example: I can possess and be exercising a power to stay in my room on the basis of wanting to; but, unbeknown to me, the door may be locked, and I altogether lack the power to leave should I so want.Footnote 18 Moreover, as distinct, the two powers are exercised quite separately: each of these two powers is exercised without any exercise of the other – we obviously cannot at one and the same time both be doing A on the basis of our power voluntarily to do A, and refraining from doing A on the basis of our power voluntarily to refrain.
Voluntariness can only be used to provide an account of freedom, then, by appealing to an agent's possession of both these two distinct voluntary powers – both a power to do A voluntarily and also a power voluntarily to refrain; and by then claiming that the agent is exercising his freedom whenever he is exercising one of these powers. Instead of a single power that is inherently multi-way, we have a combination of two distinct one-way powers. But if freedom is indeed just a combination of two distinct powers for voluntariness, then freedom will surely drop out as a distinct condition on moral responsibility. And that is because, on this classic reductive account of freedom, the only power of self-determination we ever exercise is voluntariness. Freedom as a power to do otherwise is left a power that is never exercised at all.
On the classical English-language compatibilist reduction, whenever I do A, the power of self-determination which I am exercising is the power to do A voluntarily. But this, very evidently, is not a power to act otherwise. Any power to act otherwise – to refrain – that I may possess is quite distinct. It is the power voluntarily to refrain from doing A. And even if this power does happen to be possessed, it is certainly not being exercised. It is quite inert. Its absence would make no difference to the power I am actually exercising, since the two powers are distinct and independent of each other. In which case, the presence or absence of this unexercised power to act otherwise must be irrelevant to my moral responsibility for what I do. How can moral responsibility ever depend on a kind of power that is never actually exercised to determine what we do? But that, on this reductive account of freedom, is precisely what the power to do otherwise becomes.
Our moral responsibility for action depends on the fact that we ourselves determine how we act. The question then is what kind of self-determining power we really exercise. For that will provide the true basis of our moral responsibility. Is it that we are exercising a power to act otherwise? Or is that we are acting as we will and because we so will? Which matters – control or voluntariness? The idea of freedom as a complex case of voluntariness is an attempt to combine both conceptions. But it is a deeply unstable compromise, and control is surely going to be the loser. And this is because the power to act otherwise is never actually being exercised to determine action – only a power to act as one wills. The power to act otherwise is present, but as a dummy that plays no active role at all. Why make moral responsibility depend on it, if it is irrelevant to any power that the agent actually exercises over how he acts?
It is this transformation in the theory of freedom that left it so easy for Frankfurt to think that moral responsibility must be independent of any freedom to do otherwise:
When a person acts for reasons of his own, and is guided entirely by his own beliefs and preferences, the question of whether he could have done something else instead is quite irrelevant to the assessment of his moral responsibility. Analyses purporting to show that agents do invariably have alternatives are simply not to the point, when there is no reason to suppose that having those alternatives affects the decision or conduct of the agents in any way.Footnote 19
It might seem initially puzzling why Frankfurt should say this. Suppose the power I exercise over how I act, and that bases my moral responsibility for how I act, is an inherently multi-way power of freedom that could equally be used to do A or to refrain. Then surely the presence or absence of an alternative to whichever I actually do, which comes with the presence or absence of the power to determine that alternative, would imply a difference to my actual decision or conduct in one crucial respect. It would betoken the very presence or absence of power on my part to determine what I do. But now it is clear why Frankfurt is denying this. For he is assuming that as a power over alternatives, freedom is not an inherently multi-way power.
He is assuming that the apparently multi-way power of freedom decomposes into two distinct cases of one-way power: a power to do A and a further and distinct but otherwise like kind of power to refrain. This one-way power that is used to construct and compose freedom and that, in effect, replaces it, is supposed to be recognizably a power of self-determination, sufficient to base the agent's moral responsibility for what he does. And when we exercise our power to determine our actions for ourselves, it is only ever this one-way power that we exercise. There really is no inherently multi-way power of self-determination that could be exercised instead. Granted this account of self-determination, the presence or absence of a freedom to act otherwise would make no difference to the power being exercised by the agent over what he does – and so it would indeed be of no relevance to the agent's moral responsibility for what he does.
Of course it is far from obvious that the power by which we actually determine what we do really is a composite of distinct one-way powers. Whether we are incompatibilists or compatibilists, why not suppose that freedom is what it appears to be – a multi-way power that leaves it up to us, within our control, which action we perform? And in fact our conviction that freedom is of immediate relevance to moral responsibility suggests that our understanding of freedom is of a single multi-way power. As far as intuition is concerned, its being up to me which action I perform provides immediate support for how I act being my responsibility; and my lacking control over which action I perform provides equally intuitive support for my not being responsible. And that suggests that our ordinary conception of freedom does not allow for its reduction into distinct cases of one-way power.
Frankfurt has of course claimed to prove that moral responsibility does not depend on any freedom to act otherwise. In his ‘Moral responsibility and alternate possibilities’ he asks us to consider examples where, he argues, an agent Jones is deprived of the freedom not to perform some action A, not by anyone else actually intervening to make him do A, but by the mere possibility of their so intervening.Footnote 20 Someone very knowledgeable and powerful, Black, is monitoring Jones to check that he really is going to do A. Should he show any sign of acting otherwise Black would certainly act to ensure that Jones did indeed still do A. Hence, thanks just to Black's potent but inactive presence, Jones lacks the freedom not to do A. But Black does not have to intervene, since Jones goes ahead and does A ‘off his own back’ and independently. In fact he does A in just the same way that he would have done had Black not been present, so that he remained free not to do A. Jones does A anyway, independently of Black. Since Black's presence and the concomitant lack of a freedom to do otherwise make no difference to what Jones does and why he does it –
Indeed everything happened just as it would have happened without Black's presence in the situation and without his readiness to intrude into it.Footnote 21
– they can make no difference to Jones's moral responsibility. The presence or absence of freedom must in itself be irrelevant. Jones must be morally responsible whether or not he is free to act otherwise.
Frankfurt is taking cases where his opponents would already accept that the agent is morally responsible – which ex hypothesi are cases where the agent is free to act otherwise; and hopes to show that these cases can be transformed into examples where the agent is not free to act otherwise, but without change in any features of the case that might be relevant to the agent's moral responsibility. The removal of the agent's freedom to act otherwise must make no difference to how the agent acts and to the power deployed by him to determine how he acts. That's why Frankfurt is so keen to emphasise that, with the removal of Jones's freedom
Indeed everything happened just as it would have happened without Black's presence in the situation and without his readiness to intrude into it.
If Black's potent but inert presence does indeed make no difference, someone who believes in Jones's moral responsibility given his freedom to do otherwise and Black's concomitant absence, will remain committed to admitting Jones's moral responsibility even given Black's presence – and even given the consequent lack of any freedom to act otherwise. Having admitted Jones's moral responsibility initially, they must go on admitting it even when Black is introduced into the story.
But this strategy will only work if the removal of a freedom to act otherwise does make no difference to how the agent acts and to the power deployed by him to determine how he acts. And that will only be true if agents do determine how they act by exercising power in one-way form, and not in a form that is inherently multi-way. The problem for Frankfurt is that those who believe that moral responsibility does depend on a freedom to do otherwise cannot accept his view that human self-determination is exercised only in one-way form. They are committed to maintaining the opposite – that human self-determination is only ever exercised through a power of freedom that is inherently multi-way. Why else would moral responsibility depend on the freedom to do otherwise? For if self-determination were exercised in one-way form, then the irrelevance of a freedom to do otherwise to moral responsibility would follow immediately – as we have already seen, and without having to appeal to complicated Frankfurtian thought-experiments.
Frankfurt's opponents cannot be expected to admit that the presence or absence even of a Black who does not actually intervene is irrelevant to Jones's moral responsibility. For in so far as the presence of Black, willing and able to intervene, and with consequent control over Jones's action, did imply the absence in Jones of any power to act otherwise, Black's presence would – in their view – also mean that Jones lacked any power to determine for himself how he would act. And lacking that power, Jones would lack moral responsibility for his actions. Thanks to what Black's intentions and ability to determine Jones imply for Jones's own powers, Jones simply cannot be determining his action as he might in a world where Black were absent.Footnote 22
The work is all done, then, by Frankfurt's assumption that self-determination is exercised only in one-way form. And the Frankfurt cases are irrelevant to the argument. They do no actual work themselves. For once we grant that self-determination is only ever exercised in one-way form then, as we have seen, Frankfurt's claim of the irrelevance of freedom to moral responsibility follows immediately, and without the need to appeal to complex Frankfurt-cases. On the other hand, without this assumption, the Frankfurt-cases cannot be used to prove freedom's irrelevance anyway. For Black's presence or absence can no longer be assumed to imply no difference in Jones's power to determine for himself what he does. When Frankfurt expressly asserts
Indeed everything happened just as it would have happened without Black's presence in the situation and without his readiness to intrude into it.
that claim can only be true if the power exercised over Jones's action to determine what he does is one-way and not multi-way. And that is just the point at issue.
6. One-way Self-determination
The classical compatibilist reduction of freedom to a complex form of voluntariness looks unhappy – but not just because it denies multi-wayness. The denial of multi-wayness makes the reduction implausible as an account of the power of freedom that we ordinarily take ourselves to possess, because inconsistent with the ethical significance we ordinarily accord the freedom to do otherwise. But the reduction contains other features that are problematic in more fundamental ways – that make the reduction implausible as an account of self-determination in any form at all.
There are two related difficulties. The first is that a compatibilist conception of moral responsibility is being built into the theory from the outset. Self-determination is being explained in terms of the determination of action by some entity that is distinct from the agent, and that, as prior to what the agent is responsible for, is not itself the agent's responsibility. But this obviously presupposes that the self-determined character of action must be compatible with the action's determination by factors for which the agent is not responsible. The problem is clear. Apparently competent users of concepts of moral responsibility and blameworthiness disagree about whether compatibilism about moral responsibility is true. But that makes it implausible that compatibilism about moral responsibility, even if it proves to be true, should follow immediately and trivially from our very concept of the power that leaves us responsible.
The worry is reinforced, and in a way that does not involve any assumption of incompatibilism, by the distinctive nature of moral blame – the criticism that asserts moral responsibility. As we have already noted, ordinary rational criticism operates at just one level. It criticizes the person by reference to some deficiency in a state or occurrence in the life of that person. We criticize someone as irrational because they hold or are disposed to hold attitudes that are irrational. The criticism is of defective states and of the person just as possessing those defective states. Blame, on the other hand, does not simply criticize someone for deficiencies in their attitudes or other states but puts their possession of such states down to them as their fault. This supposes something more – a problem not simply with events and states in the agent's life, but with the agent as determiner of those events and states. But explaining such determination in terms of voluntariness removes this distinctive element. The supposed problem with the agent as determiner is turned back into a problem that primarily involves an event or state within their life, such as a motivation, and the agent just as possessor of the event or state. In which case, what is left of the idea that the agent is especially responsible, in a way that goes beyond ordinary criticizability?
But none of this is necessarily to reject one fundamental theoretical ambition of Frankfurt's, which is to establish that moral responsibility can be detached, at least conceptually, from being based on any power to determine alternatives by way of action. Frankfurt may not have proved that, as things are, our moral responsibility does not actually depend on a freedom to act or determine otherwise. But it may still be at least conceptually possible that moral responsibility should have been independent of such a freedom.
Our ordinary conception of self-determination takes it to involve a multi-way power – a power to produce more than one outcome – and this multi-wayness is fundamental to the ethical significance of its involvement of alternatives. It is this multi-wayness, if anything does, that leaves moral responsibility dependent on a freedom to act otherwise. But multi-wayness is not obviously essential to self-determination, nor what distinguishes self-determination most fundamentally from ordinary causal power. For we saw that probabilistic causation could perhaps involve multi-way causal power: one and the same power could operate, under a given set of circumstances, to produce a variety of possible outcomes, such as different rates of acceleration. Even the thought, arguably essential to self-determination, that the bearer of the power must be the self, the agent, and not a mere occurrence, does not obviously differentiate self-determination from ordinary causation. Ordinary causal power has often been understood to be a power not, or not simply, of occurrences, but a power of substances too. The cause of the window's breaking may be understood to be not just the event of the brick's hitting it, but the brick itself, when it hits the window.
What is most distinctive of self-determination as a power, and most sharply distinguishes it from ordinary causal power? What is distinctive flows from the basic thought that in exercising the power the agent must really be determining outcomes for himself. The agent is the determiner of what he does, and in a way that does not subordinate his role either to prior determining factors for which he is not responsible or to simple chance. It is this aspect of self-determination that makes incompatibilism about the power, and about the moral responsibility that rests upon it, at least an initial theoretical possibility, and so – as it clearly is – a matter for real debate, rather than something ruled out by the very nature of the power. The alternative to the action's determination by factors for which the agent is not responsible is not, as Hume alleged, the operation of mere chance.
This is missing from ordinary causation. To the extent that an ordinary cause determines an outcome, this involves prior factors, the circumstances under which the cause is found together with its own causal power – what in the circumstances it is capable of determining – ensuring that it operates to produce that very outcome. How the cause will operate is already predetermined, and not left up to the cause. If the cause's operation is not so predetermined – if what outcome it operates to produce is not already settled by its powers and circumstances – then the cause can at most influence and not determine what will happen. And its operation, far from being determined by it or by anything else, will depend on mere chance.
What prevents this being true where self-determination is concerned? We have seen that two related factors are crucial. There is contingency of determination, which leaves it open whether an agent with the power to determine a given outcome will so exercise his power. The agent's exercise of action-determining power is not ipso facto predetermined by factors for which he need not be responsible. The second factor is freedom – not in relation to outcomes, but in relation to the agent's very exercise of power. The agent's exercise of power is not undetermined, but up to the agent, so that the agent determines that exercise for himself.
This involvement of the agent as determinant of his own exercise of power is what Hobbes objected to in his attack on the very idea of self-determination. But we can detach this aspect of self-determination from multi-wayness – the identity of freedom as a power to determine more than one outcome. For there could be a one-way power the exercise of which was up to the agent. Obviously if the power were one-way – under given circumstances there is only one outcome its exercise can produce – it could not then be up to us how we exercised it, to produce this outcome or that. In relation to a one-way power, there could be no room for what we have termed a freedom of specification. But it could still be up to us whether we exercised the power at all. There is room, then, for our possession, even in relation to a one-way power, of a freedom of exercise – a freedom to use the power or not. Self-determination could involve the exercise of a power to determine but one outcome, where it was up to the agent, not which outcome he determined – that is fixed by the one-way nature of the power – but whether he exercised the power at all.
If self-determination did take this one-way form, we could still possess freedom as its being up to us which actions we perform. For alternatives to be left up to the agent it would be enough that the agent should possess each of a power to determine that he does A and a distinct power to determine that he refrains – and a freedom of exercise in relation to each of these powers. That would leave it up to the agent which power he exercised and so whether he did A or not – just as freedom requires. But since no multi-way power would be involved in the determination of action, no power to do otherwise – to determine alternative actions or outcomes – would ever be exercised by the agent in the determination of what he actually does. In which case the power to do or determine otherwise would remain an inert extra. It might be lacking without making any difference to what the agent is actually doing, or to the power actually exercised by him to determine what he does. In which case the power or freedom to act otherwise – freedom as a power over alternatives by way of action – would be irrelevant to moral responsibility, just as Frankfurt supposes.
Such an account of freedom as power over alternatives by way of action would be ontologically reductive. Freedom would no longer be characterized as a multi-way power – a power to determine more than one outcome. It would be decomposed into a combination of distinct one-way powers. But the theory of freedom would no longer be conceptually reductive. A primitive notion of freedom would still be deployed within the theory, to characterize the agent's relation to the power itself, as involving a freedom of exercise.
We now see that we can appeal to one-way self-determination to provide a theory of freedom that is very close to our ordinary understanding of it, but which abandons the multi-wayness that is so clearly an element of that ordinary understanding. On this theory the bearer of the one-way power is the agent, and not some other entity for which the agent has no responsibility; and it is genuinely up to the agent whether he exercises the power. The power of self-determination involved is plainly sufficient to base moral blame as we ordinarily understand it. So it looks as though there is certainly no conceptual dependence of moral responsibility on a freedom of alternatives at the point of action. It is conceptually perfectly open that an agent be morally responsible for what they do, without possessing a freedom to act otherwise. On that specific issue, Frankfurt seems to be right.
Frankfurt supposes, though, that from the conceptual possibility of moral responsibility without the freedom to do otherwise, it follows that our moral responsibility is actually independent of the freedom to do otherwise. But that is a simple mistake, and does not follow. Our moral responsibility may still actually depend on a freedom to do otherwise, but without this following from the very concepts of moral responsibility and self-determination. For it might be a non-conceptual or a contingent truth, but a truth nonetheless, that the only power that we actually possess and exercise to determine how we act is a multi-way power, and even one additionally involving a freedom of specification in relation to the manner of its exercise. In the absence of this power, we would be simply incapable of self-determination in any form. In the absence of a freedom to do otherwise, we would not be morally responsible at all. There is nothing in Frankfurt's arguments to rule out the possibility that we are actually capable of self-determination only in this multi-way form.
Our belief in freedom does seem to be a belief in our possession of a multi-way power to determine alternatives. But because it is at least conceptually open that we could be morally responsible without such a power, this raises a doubt about the hypothesis that entirely locates the origin of our belief in multi-way freedom in a theory of blame. The hypothesis, a sceptical one, was that our belief in multi-way freedom is not based on anything genuinely in human nature, or even on the way that human nature and human action is represented in experience to us. Instead it comes from an intrusion of ethics into psychology. We postulate the power only because forced into it by our belief in our moral responsibility and by our practice of blaming people, and putting their faults down to them as their fault.
This hypothesis is put in doubt because it now appears that we could intelligibly hold people morally responsible for what they do without any belief in freedom as a multi-way power. We have just established that we could operate a theory of moral blame and moral responsibility that justifiably puts faults in the agent down to the agent as their fault, and even left room for an incompatibilist conception of that responsibility and the power that based it, but which did not involve a conception of that power as multi-way, and that consequently did not treat the agent's responsibility as depending on the power to do otherwise. Belief in the multi-wayness of self-determination and in a freedom of specification in relation to that power seems strictly inessential to our ordinary understanding of moral blame. In which case it may well come from elsewhere.
The problem of free will and moral responsibility is often thought to be a conceptual problem. Our concept of moral responsibility is supposed by itself to fix the nature of the power of self-determination that bases it; or the concept of freedom as involving a power over alternatives is supposed to fix the properties of the power in other respects, such as its compatibility with determinism. But now we see that this may not be so. The concept of moral responsibility places some demands on self-determinationFootnote 23 – but not to the extent of demanding power over alternatives by way of action. And the very idea of power over alternatives does not even fix the composition of that power.
We might wonder then if the relation of freedom to other powers, such as prior causation, is conceptually determined either. After all, why should relations of compatibility or otherwise between powers always be conceptually determined? We would not usually assume that for distinct cases of causal power. But that is another topic.Footnote 24
