Show Summary Details

p. 1048. Freedom and its place in naturelocked

  • Thomas Pink


‘Freedom and its place in nature’ assesses the randomness problem. According to scepticism, there is no middle ground between predeterminism and uncontrollable randomness. Libertarianism posits that events can be causally undetermined but still controlled. Thus, freedom is a power to determine events — a special kind of causal power. However, this reduction of freedom to a special case of causal power cannot explain how we can decide to frustrate our desires. Libertarians see freedom as the vehicle for the exercise of freedom, not as an effect of it. Libertarianly free action must be causally undetermined, but it is more than random chance. Scepticism has no convincing argument against libertarian freedom.

Is freedom a causal power?

The exercise problem may now have been solved. Action is no longer to be understood, in terms that threaten libertarian freedom, as an effect of passive desires. Action can perfectly well take uncaused form. Actions can occur as uncaused decisions, without being any the less genuine and deliberate actions. But the randomness problem still remains. Even though directed at definite goals, as uncaused or as causally undetermined our actions could still be merely chance performances. What intentions we formed could still be random and not an exercise of genuine control. We need to show that libertarians can distinguish causally undetermined freedom from the operation of mere chance; how by tying freedom to lack of determination by prior events libertarians are not at the same time equating it with randomness.

According to the sceptic, remember, libertarian freedom does come to no more than randomness. For there are only two alternatives. Either an action is causally predetermined – which libertarians insist would exclude freedom. Or, to the extent that its occurrence is not predetermined, it must be occurring by pure chance. In which case, the sceptic now insists, genuine freedom is again ruled out. There is no middle way.

p. 105I have already raised the question why should anyone suppose this. It could be that there is a third possibility – that, though causally undetermined, the event is not occurring by pure chance or randomly, and this because the event is occurring through the exercise of our freedom. Though causally undetermined, the event is not occurring by pure chance, because we are exercising control over whether it occurs. We need then to distinguish two kinds of causally undetermined event. There are those events that are causally undetermined and nothing more. These events are genuinely random in that their occurrence is dependent on mere chance. Obviously in the case of these events, any involvement of freedom in their occurrence is excluded. And then there are those events that are undetermined causally but still controlled. Here we do not have mere chance or randomness because something more is involved; and that something more is the exercise of freedom. The agent is controlling whether or not the event occurs.

In fact, there is a very important reason why someone might want to exclude this third and last possibility – that the event is causally undetermined but controlled. As we shall see, this last possibility is ruled out if we make one crucial assumption: that freedom, if it exists at all, must be a kind of causal power. And I think it has been precisely this assumption that has generated the randomness problem. So let me say something more about this assumption, about what it involves and why anyone should make it.

Clearly freedom – our capacity to control how we act – is a power of some kind. After all, freedom leaves how we act up to us. Freedom leaves, as one might put it, how we act ‘within our power’. Our control of our actions is just that: an action- and event-determining power, a power that we have to determine how we will act.

Freedom, then, is a kind of power to determine events. It is a power to determine whether a given event occurs or whether it does not. It is a special power which, it appears, only rational beings such as we humans can possess, and which can be exercised only over and p. 106through action – through how we act. Freedom is a power to determine which actions we perform. The question arises then how this power relates to other powers in nature.

For there is another power to be found in wider nature – a power possessed not just by rational agents such as humans but even by inanimate objects such as sticks and stones. This power is causal power: the power to produce effects. And undoubtedly this power is importantly related to freedom. For freedom is a power that can certainly be extended through causation. Add causal power to some already existing freedom or control, and you get yet more control. Suppose I already control whether or not I perform a given action, such as whether I flick a switch. Then my control can extend even further through the causal power of this action, through its possible effects. Perhaps flicking the switch would cause the lights to go on or off. In which case, thanks to the action having this power to affect the lights, my control of whether I perform it also gives me control of whether the lights go on or off. The causal power of an action that I already control gives me yet more control – over everything which performing that action can affect.

It is tempting to suppose that the relation between freedom and causation could be even closer. Perhaps freedom is not only extended through causation. Perhaps freedom just is a kind of causation. In which case from the very outset the exercise of freedom consists in the production of effects. Any event over which we exercise control must occur as an effect that we have caused. Which means that there will really only be one event-determining power in the world, not two. There will simply be causal power, of which human freedom will be no more than a further instance.

We can see how satisfying this idea might be. It appeals to the profound need we have to simplify in our theorizing about the world. We want to explain as much as possible in terms of as little as possible. We want to reduce the apparently rich plethora of kinds of things and phenomena within the world down to the smallest p. 107possible range of fundamental elements – a simple and economical range of elements out of which the rich complicated whole can somehow be shown to be constructed. So why not replace what appear to be two distinct powers, freedom and causation, with only one more fundamental power, causation, of which freedom will turn out to be but a special case?

This option looks all the more attractive when we consider all the difficulties and doubts to which the very idea of freedom gives rise. What better way could there be to resolve these doubts and difficulties than by revealing freedom to be just the same familiar power as the power of stones to break windows, or the power of fire to boil water?

But if we do make this identification, the equation of lack of causal determination with a dependence on mere chance or randomness will follow. For mere chance or randomness just is sheer lack of causal determination. It is what you get from lack of causal determination and nothing else – including the absence of control. And if control is only ever exercised as a kind of causation – as a kind of causal power – then to the extent that events are causally undetermined, undetermined by any causal power, so to that same extent their occurrence will also be uncontrolled. And that does indeed leave their final occurrence, both causally undetermined and uncontrolled as it is, dependent on mere chance or randomness.

On the other hand, it really is not obvious that freedom is a kind of causal power. After all, sticks and stones do not possess freedom, though they or events involving them clearly can produce effects. Indeed most things that have causal power, which produce effects, lack control over how that power is exercised. A stone thrown at a window does not control whether or not it breaks the window. Once thrown against the window with a given force, there may be only one effect that it can have – that the glass shatters. Freedom is quite different. Freedom or control is inherently a power that can be exercised in more than one way – to determine either p. 108that a given action occurs or that it does not. We have control over which actions we perform, whereas ordinary causes lack control over which effects they produce.

On any view, to exercise freedom is to determine for oneself whether or not a given action occurs. Freedom is a power to determine. Many philosophers go further, though. They just assume that to determine and to determine causally must come to the same thing. In which case, on their view, what is causally undetermined would also have to be undetermined by us, and so outside our control. But this ignores the possibility that freedom might be a non-causal power; that in exercising freedom we might be determining how we act non-causally – and so in quite a different way from the way in which a stone determines that a window is broken.

There is a clear limit to the simplicity in nature. There must be fundamental distinctions, otherwise everything would be exactly the same – which very obviously is not the case. It is as important to do justice to the differences as to draw out the similarities. We do not usefully explain anything just by insisting, dogmatically, that things that are evidently different are really just one and the same thing. The power of freedom and causal power may provide a case in point. These two powers do look very different; and perhaps that is just because they really are distinct.

It is important to stress this point because the English-language tradition in philosophy has been so prone to neglect it. In the last 50 years, it has been particularly prone to ignore the point where things involving the mind and mentality are concerned. The mind, and especially the highly developed human mind, is clearly a remarkable and distinctive thing. In the mind we find things and phenomena that seem quite different from anything else in wider nature. We find consciousness, we find understanding, we find rationality or the capacity to respond to reasons – and we find freedom, the capacity to control the will and the further voluntary p. 109actions that depend on the decisions of our will. Or so we think. But rather than recognize and appreciate this distinctiveness, many philosophers have been immensely uncomfortable about it. They have taken the path first opened up by Thomas Hobbes – the path of naturalism, which tries to deny or abolish the evident differences between humanity and wider nature.

One option is that of naturalist elimination. This option is simply to deny that things such as consciousness, understanding, rationality, and freedom really exist. But the other, more subtle option is that of naturalist reduction. This option is to admit that each of consciousness, understanding, rationality, and freedom really does exist, but to make out that it is really nothing more than a special case of some other supposedly less problematic feature found more generally in wider nature. We try to characterize all that these things involve in other terms – terms borrowed from wider nature and not trivially equivalent to the phenomenon that we are trying to characterize. With freedom the claim will be this: that freedom is really nothing more than a kind of causal power.

Compatibilist naturalism and freedom as a causal power

Reductivist accounts of freedom as a kind of causal power have a native home. Their origins lie in the Hobbesian tradition. Compatibilists of this naturalist kind have been particularly keen to explain what freedom is in other terms and to view freedom as nothing more than a kind of causation. These compatibilists have claimed that freedom, the power we have over our actions, is a causal power of our desires. For they have assumed Hobbes’s theory of action. They have assumed that to perform an action is just to do something voluntarily, on the basis of a prior desire to do it – this desire being an antecedent and entirely passive cause of how we act. In which case it becomes very appealing to reduce freedom, the power that we have over our actions, to no more than the causal power of our desires and appetites to lead us, marionette-like, p. 110successfully to do what they motivate us to do. Why should freedom not come to just this causal power? After all, on the Hobbesian theory of action the exercise of this very causal power is all that action involves. Action just is being led to do what we want by the fact that we want to do it. Unless our desires do produce such effects on what we do, action cannot occur at all – and so freedom cannot be exercised either. Any obstacle to the satisfaction of our desires just is, by its very nature, an obstacle to the exercise of freedom. So why not define the exercise of freedom as consisting, purely and simply, in such desires unobstructedly causing their satisfaction? Freedom just is the power of our desires to cause us to do as desired.

But this compatibilist reduction is quite unacceptable. It conflicts with the common-sense notion of an action which, as we have seen, does not define action in general as an effect of wants and desires. Goal-directed action can perfectly well take a form that need not be caused by desires or indeed by any other prior motivation. Goal-directed action can take the form of uncaused decisions to act. This means that we can no longer define freedom as a power of our desires to cause their satisfaction. For a block to the satisfaction of such desires need no longer be a freedom-threatening obstacle to our own action. What might prevent the satisfaction of our desires could be nothing other than our very own deliberate decision – such as, for example, our decision just not to perform the low action that all our desires and appetites are inclining and tempting us towards. The block to the satisfaction of our desires could lie in our very own deliberate action. We could be deliberately frustrating our desires by the contrary exercise of our own will. But if it is, not some external obstacle, but our very own deliberate action that is frustrating our desires, what would be inherently freedom-threatening about that?

Action, we now see, is not by nature an effect of our desires. It is instead something very different. Action is really a capacity deliberately to determine which, if any, of our desires are to be satisfied. And so our freedom, our control over how we exercise this p. 111capacity for action, is correspondingly a control over whether and which of our desires are to be satisfied. Our use of this power generally to frustrate our desires is not a loss of our freedom, but one possible way of exercising it.

Freedom, our power over our own action, cannot plausibly be identified as a causal power of prior desires or other passive motivations to determine how we act. And that is because we can deliberately use our own action, the very action through which we exercise our freedom, to frustrate our desires. If freedom is a causal power at all, it must be a causal power of a quite different kind.

Libertarianism and freedom as a causal power

Libertarians, of course, should never define freedom as a causal power of desires or other passive motivations to determine how we act. And there is a very obvious reason why. Since such a causal power is something that, in strong enough form, could perfectly well remove libertarian freedom, it and libertarian freedom must be quite distinct. For libertarians the causal power of any prior occurrence or happening over how we act is a potential threat to our freedom. So freedom can never be identified with such a causal power.

Yet many libertarians have worried that, if freedom is not some kind of causal power, there is nothing else for freedom to be. Action that was causally undetermined would not be determined or controlled in any other way. It would simply be action that was random. So these libertarians too have sought to identify libertarian freedom as a causal power – but as a causal power of a different kind.

There is only one other kind of causal power for our freedom to be. Not a causal power of any antecedent event or happening, not even that of a prior event or happening in our own minds such as a desire or feeling. Freedom must instead be a causal power attaching to and exercised directly by our own selves. Freedom must be a causal p. 112power, not of some prior action-influencing event in an agent’s life, but of the agent himself. And, since freedom is inherently and essentially a two-way power, a power to do or to refrain, this causal power must be similarly exercisable in more than one way. The causal power that constitutes our freedom must be both a power to produce a given effect and to prevent its occurrence. And so we arrive at one very influential and popular theory of libertarian freedom: the theory that appeals to what philosophers term agent-causation. Freedom is supposed to be a special two-way causal power over actions, a causal power possessed and exercised, though, not by any unfree event but by free agents themselves. Freedom is supposed to be an agent-causal power.

Why should libertarians want to characterize freedom as an agent-causal power? Remember that the agent-causal theory is really doing two jobs. First, it is reducing freedom to another kind of power, revealing it to be but a special case of a phenomenon, causal power, found more widely throughout nature. So we have the satisfaction of locating even something so (supposedly) unnatural and exceptional as libertarian freedom as really just another part of nature. But secondly, the theory is also solving the randomness problem. If when we exercise freedom, we as agents causally determine how we act, then our libertarianly free action cannot be random. Free action cannot be random because it is causally determined – not by some prior freedom-threatening happening but by us ourselves as free agents. For one thing is clear and agreed on by everyone. Causal determination precludes all randomness in what is causally determined. The exercise of libertarian freedom is being very clearly distinguished, then, from randomness.

So when some action A occurs through an agent’s exercise of his freedom, the agent himself is operating as a cause. Not any mere event or happening, not any desire or motivation, but the agent himself is causally determining whether or not he does A.

p. 113

The occurrence of action A is not random, because it is causally determined. But its occurrence can still be an exercise of the agent’s control because it is determined, not by an uncontrolled cause, but by the agent himself, as a freely operative cause.

It is of course important to this story that the agent’s causal power must be exercised freely. The agent must control how he exercises his agent-causal power; he must control whether he causes or prevents A’s doing. Otherwise, though causally determined by the agent himself, A’s doing would be determined by a cause outside the agent’s control, and libertarian control over whether A is done would be removed. But the claim that the agent controls his exercise of agent-causal power is anyway hard to deny. For that causal power is supposed to be nothing other than the agent’s freedom; and how can freedom be exercised anything other than freely?

So should we view libertarian freedom as a form of causal power, as a causal power of agents? The issue is complex, and the arguments on both sides many. But there is an obvious consideration why we should not – which suggests that libertarian freedom cannot be any kind of causal power.

Remember that the agent-causal theory of freedom is supposed to do two jobs. It is supposed to give freedom a satisfactory location and identity within wider nature. It is supposed to reassure us that freedom is just another case of a very familiar kind of thing, causation. And it is supposed to resolve the randomness problem, by explaining how the exercise of libertarian freedom is distinct p. 114from randomness. I shall suggest that neither of these jobs is being done.

We do nothing to assuage worries about what libertarian freedom might come to, or to explain how it excludes pure chance, if we simply label it with another name – a name shared by a phenomenon that, in reality, is very different. That, I think, is all we are doing if we refer to libertarian freedom as a form of causal power. For freedom really is very different from causation.

As I have emphasized throughout this book, freedom, by its very nature, is a power that can be exercised in more than one way – which way being under our control. Freedom, by its very nature, leaves it up to us which actions we perform. But ordinary causal powers, powers to cause things, are not like this at all.

True, ordinary causal power can sometimes be exercised in more than one way. An ordinary cause may be fundamentally probabilistic. It may in fact be fundamentally chancy, not just whether it produces an effect at all, but which effect it produces. Perhaps, on the one hand, pushing that button will cause a light to turn green; or alternatively perhaps it will cause the same light to turn red. But even if that is true, that of course does not leave an ordinary cause in control of which effect it produces.

There is more. Such a merely probabilistic or chancy cause does not of course count as determining which of its two possible effects it produces, in the sense of removing any dependence of the outcome on simple chance. There is only one way for an ordinary cause so to determine the occurrence of one effect rather than another. It must be true that, given all the relevant circumstances and the cause, the occurrence of any alternative effect is left impossible. Should more than one effect be left possible, which effect the cause produces will be left random and undetermined – a matter of pure chance.

But when an agent determines what will happen by exercising p. 115freedom, things are quite different. Suppose the agent has to choose whether to do A or not-A. And suppose that, given the circumstances and his presence as a free agent, either of A or not-A remains equally possible. Which action the agent performs need not be left random and undetermined. The power of freedom – his being in control – still allows it to be the agent who determines that he does A rather than not-A, so that which he finally does is not a matter of pure chance. But that surely implies that freedom is a quite different kind of power from causation. An ordinary cause determines its effects and excludes randomness only when its very presence leaves alternatives impossible. But a free agent determines his actions and excludes randomness in a quite different way. Why then suppose that way is causal too?

The consequence is obvious. Because the two powers, freedom and ordinary causation, work to determine outcomes and exclude randomness in very different ways, the fact that ordinary causal determination is randomness-excluding does nothing to explain how the exercise of freedom might also be randomness-excluding. You may label the power of freedom as ‘causal’ too, if you like, but attaching that label to it does nothing whatsoever to solve the randomness problem. For the labelling is merely that. It does not increase our understanding of how the exercise of freedom prevents the final outcome being a matter of pure chance.

Calling freedom ‘causal power’ does nothing to explain how freedom excludes randomness, or to assimilate freedom reassuringly to the familiar power had by sticks and stones. In fact, applying a common term to each, to both freedom and ordinary causation, only reminds us of the profound difference between these two kinds of power. So why insist on thinking of freedom as a causal power at all? Until and unless we are given compelling reason for thinking otherwise, we should suppose that freedom is not another case of causal power, but recognize it for what it is: a power of a quite new and different kind.

Libertarian freedom without reduction

We should conclude then that freedom is not a causal power. My initial exercise of control over my action, such as, for example, over a decision that I take, does not involve that action occurring as an effect, whether of my motivations or of myself as agent-cause.

But if freedom is not a causal power, what is the relation of a free agent to his free actions, the actions that he controls? It is clear what the relation must be. When I freely perform an action of the will – when I decide to go out rather than stay in – my decision is not an effect of me. And so it is not an effect of any power that I possess. My decision stands to my control or freedom, not as its effect, but as its medium or vehicle. I do not exercise my freedom to cause my decision. Rather my freedom is exercised in the taking of the decision itself. That decision is what immediately constitutes my exercise of my control.

Here is an analogy. Consider, for instance, my power as your creditor to release you from your debt to me. When I declare and say ‘I release you’, I do not exercise my power of debt-release to cause that declaration. For that declaration does not occur as an effect produced by my power to release you, but constitutes that power’s very exercise. The declaration is the very medium in and through which the power is exercised. So too it is with action and the exercise of freedom. Action is the vehicle or medium for the exercise of freedom, not its effect.

Then it follows that libertarianly free actions really are causally undetermined. They may even lack causes altogether. But does this not land us back in the randomness problem? If the actions that the libertarian claims to be free are actions that lack any determining cause whatsoever, does that not leave them random? Are they not occurring by pure chance? No. Libertarianism is only in danger of turning freedom into randomness if – unwisely – it says that the exercise of freedom is nothing more than the occurrence of p. 117causally undetermined action. For randomness, as we have already noted, is what you get with causal undeterminedness and nothing else.

Libertarianly free action must, by its very nature, be action that is causally undetermined. And to that extent then, of course, such free action does involve chance. Prior to the agent’s choice it must be causally undetermined, and so in this respect chancy, how the agent will exercise his freedom. But this is not inconsistent with freedom still being exercised. The exercise of genuine freedom is ruled out only if chance is all that there is – if an event is causally undetermined and nothing more, so that no power at all is being exercised to determine whether the event occurs.

But libertarianism has no business claiming that the exercise of freedom comes to nothing more than chance – to no more than the occurrence of causally undetermined action. For that is obviously false. All the libertarian should be claiming is that causal undeterminedness is a condition on the exercise of freedom. To read this incompatibilist condition on freedom as an exhaustive specification of all that freedom involves is quite gratuitous. It is to impose a reductionist interpretation on Libertarianism that is quite unwarranted. It is to read the libertarian as trying to do what Hobbesian compatibilists try to do – which is reductively to explain all that freedom is in other terms. There is no reason why the libertarian should be committed to doing any such thing.

The libertarian would be foolish to attempt such a reduction in any case. For it is inconsistent with a fundamental feature of common-sense morality. It is inconsistent with the moral significance that common sense gives to freedom in relation to moral responsibility. As we have seen, common sense appeals to freedom – to our control over how we act – to explain why it is for how we act that we are directly responsible, and not for our passive beliefs and desires. Common sense’s explanation is contentful and intelligible. It is a substantial story why. The explanation is that we are directly p. 118responsible for our actions and not for our beliefs and desires because it is our actions that we control.

Any account of freedom that is to be consistent with our ordinary concept must permit this explanatory story to survive as, at least, a contentful explanatory story. Whether we believe the explanatory story given really is true or not, it must remain possible to give it. Why can we be directly responsible for our causally undetermined actions, and not for, say, causally undetermined desires? The obvious and immediate explanation is that we can directly control our actions, as we cannot directly control our desires, even our causally undetermined ones. But then for there to be anything to this explanation, there had better be more to freedom than the mere lack of causal determination in action. For otherwise all the explanation comes to is that we can be directly responsible for our causally undetermined actions and not for our undetermined desires, because our undetermined actions are actions. By reducing freedom to the mere lack of causal determination in action, what was an intelligible explanation of the restriction of moral responsibility to action would vanish into a clear non-explanation.

The same argument could be used against another familiar compatibilist reduction of freedom – this time not the reduction favoured by naturalist Compatibilism but the rather different one favoured by rationalist Compatibilism. This is the identification of freedom with practical reason. The rationalist tries to reduce freedom to nothing more than a capacity to act rationally. It should by now be obvious what is wrong with this reduction of freedom to something else. For we exercise reason, not only practically in how we act, but also non-practically in the formation, prior to our action, of passive beliefs and wants. And we can ask why, as rational beings, we can be directly responsible for actions and not for these passive beliefs and wants. The common-sense answer again appeals to freedom. We directly control those exercises of our rationality that constitute our actions, and not those that constitute our formation of beliefs and desires. But again, for this explanation to p. 119work, there had better be more to freedom than our capacity for rationality taking the form of action. Otherwise all we are left with is the thought that we are responsible for our actions because they are actions – which is no explanation at all.

And of course there is more to freedom than mere practical rationality, and we all know what it is: it is the distinctive up-to-us-ness of our actions, our being in control of them. And like so much else in our view of the mind, like being conscious, understanding something, being reasonable itself, we simply cannot adequately specify all that this control, all that this up-to-us-ness involves in other terms. Freedom is not simply a capacity to act undeterminedly. And it is not simply a capacity to act rationally. And freedom is not simply a kind of causal power. In fact freedom is not a causal power at all. As a power, freedom is simply what it is – and not another thing.

In defence of libertarian freedom

The sceptical case against the very coherence and possibility of libertarian freedom is far less formidable than it initially appeared. In fact, it now appears to be profoundly question-begging. And this becomes clear now we have uncovered where the roots of this scepticism really lie. They lie partly in a Hobbesian caricature of human action – a caricature that reduces action to nothing more than an effect imposed on us by our desires, and which brazenly excludes from human self-determination the very decision-making in and through which our self-determination is principally and initially exercised. And otherwise they lie in a dogmatic exclusion from the outset, both from the world and even from our very experience and understanding of the world, of the very freedom that is being denied.

The sceptical attack amounts simply to a dogmatic determination to describe the world only in terms that already exclude freedom as a distinctive feature of human life. The sceptic assumes that the world p. 120can contain no power other than causation; and that any event that is not causally determined by prior events must just be random. But if we insist on describing the world only in these terms, then of course it may well appear that libertarian freedom is not possible and cannot exist. But by what right do we so exclude such freedom from the very outset?

There is no convincing sceptical argument to show that libertarian freedom is impossible in principle – no argument to this conclusion that does not simply beg the question. But is there instead a convincing sceptical argument to show that even if libertarian freedom is possible, its actual possession and enjoyment is still unlikely? No. For example, no one has actually established the truth of causal determinism. Any causal influences on our action that can be traced back to our desires, and perhaps even further to environmental or genetic factors, seem in general to be merely that: influences. No one has shown that the generality of human decision and action is outright determined by such causes. Provided such causes do merely influence us without actually determining how we act, they leave room enough for freedom. Such influences may sometimes reduce or constrain our control over what we do. But they need not remove it entirely.

Do we have any direct evidence that we actually are libertarianly free? The sceptic will allege that we do not. The sceptic will allege that the only power that we ever directly experience is ordinary causal power – or its lack. And very plausibly, neither an experience of ordinary causation nor of its mere lack is enough for an experience of what libertarians understand as freedom. Of course, if the sceptic were right about this, we might well be left with, at best, the Kantian option. Our freedom would not be an object of experience at all, but something that we somehow just assume on other, perhaps more dubious grounds.

Yet by what right is it supposed that we do not have any direct experience or awareness of our own freedom? In fact, by a curious p. 121irony, just as the sceptic about freedom seeks to exclude any representation of it from our experience, so others have tried to do the same with regard to causation. Causation, of course, is the power that was earlier supposed by some, by compatibilists and agent-causationists, to explain what freedom is. Causation is the power that some philosophers hoped to use as a means to naturalizing freedom. If freedom could be made out to be a kind of causation, they hoped, then that would leave freedom a perfectly familiar part of wider nature. But one philosopher in particular – David Hume – tried to exclude causation, too, from our experience, and in just the same way that our sceptic wants to exclude freedom.

Hume thought causal power was just as hidden from us as the sceptic supposes libertarian freedom to be. He thought that our experience never actually represents causation to us. All we have direct experience or awareness of, Hume argued, is regularities in nature – one kind of thing, such as a fire being lit, regularly being followed by another kind of thing, water above it boiling. We never have direct awareness of something else, causal force, as a further feature in the world connecting these.

Such attempts to exclude from experience the representation of things such as freedom and causation are notoriously problematic. How far along this route do we want to go? We can soon reduce the content of visual experience, for example, to nothing more than the presentation of an array of differentially coloured surfaces. Surely it is obvious, the argument will go, that all we directly see are areas of colour. The rest, a world of solid material objects, is something which visual experience does not directly represent, but which we have to infer. Belief in a solid physical world is something going far beyond what experience itself ever reveals.

This is not an attractive view of experience. But how can it be resisted? I suspect that the only way is to appeal to the guidance that experience actually gives to our belief. If experience regularly and normally guides our beliefs about whether and when a given p. 122kind of thing is to be found, then experience must be representing that kind of thing and its condition to us. How else to determine what experience represents about the world, than by referring to its normal impact on our beliefs about what the world contains?

So experience regularly and normally leads us to believe in a world of solid objects, objects that exert and are subjected to various kinds of causal force. Experience leads us to believe in a world of objects such as sticks and stones – objects with the power to strike other objects and damage and destroy them. This then is the world that experience represents – not just a world of mere regularities unconnected by causation, still less just a world of coloured shapes.

Equally, experience regularly and normally guides us in forming beliefs about our freedom. It leads us to believe that we possess varying degrees of control over how we act – sometimes that this control is present, sometimes that it is diminished or even absent. Experience guides us to form beliefs not only about the causal powers of objects but about the non-causal power that is our own freedom. So freedom, as much as causation, is something that experience represents. And if experience is not an infallible guide in the case of freedom, nor is it infallible in its representation of causal power. But that does not show that freedom is unrepresented by experience, any more than it shows that causation is unrepresented by experience. If we still can acquire knowledge of causal power by relying on its fallible representation by experience, then we can just as well acquire knowledge of non-causal freedom too, and in the same way.

Arguing over the telephone with an awkward and deeply exasperating colleague, I raise my voice, deliberately speak ever more woundingly – and then, as my temper mounts, finish by quite intentionally delivering a gross insult and smashing down the phone. I feel myself doing all this – and I feel my control over what I do lessening progressively as I do it. I can feel myself just losing it. As I experience my action, I feel it is increasingly my anger that is p. 123determining how I am acting, not I. Who is to say that my experience of my agency is not representing all this to me? My experience is just the kind that leads those having it to believe that they are losing control. It is just the kind of experience that we would report as the ‘feeling that one was losing it’.

We have a widely shared idea of freedom – a freedom or control of what we do that we naturally conceive in libertarian terms. It is an idea that is as much and vivid an element in our experience of ourselves and of the world as is the very different idea of causal power. So why try to turn one power into the other? And why be selectively sceptical of one power and not the other?

By contrast to causation, freedom seems limited to humans, or to at most humans and the higher animals. Freedom is unlike anything outside the mind in wider nature. But then the same is also true of many other features of the mind, such as our consciousness, our rationality and our very capacity to understand. Yet all these, having control of what we do, being conscious, understanding things, are aspects of ourselves of which we are directly aware – as aware as we ever are of anything. Human freedom is certainly as puzzling and distinctive a phenomenon as any other of these features of our mentality. But it seems no less worthy of our belief than any of these others – a belief that we in any case seem perfectly incapable of abandoning. We can as little lose our everyday conviction that much of our action is up to us to perform or not than we can abandon belief in our own capacity for reason or for understanding. And there is nothing yet to prove this conviction or the other beliefs accompanying it improbable or wrong.