Introduction

Helen Steward is Professor of Mind and Action at the University of Leeds and a Fellow of the British Academy. Before arriving at Leeds, she was a Fellow of Balliol College, Oxford for fourteen years. She has worked on a variety of philosophical topics, including free will, determinism, causation, emergence, supervenience, levels of explanation, the event/state distinction, and the concepts of process and power. She has also worked on animality and on understandings of the human being which take seriously our membership of the animal kingdom. She is the author of The Ontology of Mind: Events, States and Processes (1997) and A Metaphysics for Freedom (2012). Currently, she is writing a book on causation.

A Metaphysics for Freedom addresses the free will problem. Its distinctive thesis is that agency itself, and not merely the special, distinctively human variety of it, is incompatible with determinism. This version of libertarianism insists therefore that many animals are ‘free’, in a sense – that they make genuine choices, about some of which we can say, just as we say of ourselves, that they could have done otherwise.

Stephen Law (SL)

Ordinarily, we believe that we are, for the most part, free agents that are morally responsible for our actions. Suppose I steal my neighbour's car. We suppose that's something I freely chose to do. No one compelled me to act that way, by putting a gun to my head, or by pulling my strings like a puppet or manipulating me like a remote control robot. Because I freely chose to do that bad thing, I am morally responsible for what I did, and so can deserve punishment. But of course, philosophers have challenged this ‘common sense’ view. Many philosophers argue that we lack free will and moral responsibility. What, in a nutshell, is their objection to the ‘common sense’ view?

Helen Steward (HS)

There are different ways in which philosophers might try to argue that you didn't really ‘freely choose’ to steal your neighbour's car after all. One version of the argument might go like this: you do what you do because of the way you are – what kind of character or personality you have, what your main motivations and values are, for example. But what makes you the way you are? Basically, it's a mixture of inherited factors (which you can't do anything about) and ones which have to do with your upbringing, social situation, education, and so on, which (these philosophers would argue, you ultimately can't do anything about either). Of course, it might seem at first as though you can do something about these things. Perhaps by the time I was ten, for instance, I could have chosen to hang out with certain people rather than others, and that choice might have had an important impact on my eventual character as an adult. Those people might have influenced me, for good or ill. But one would then need to ask where that choice – the choice to hang out with those people – came from. It too will have been made because of the way you were at the time you were ten – which will again have been produced by a mix of inherited factors and ones which have to do with education, social background, etc. And so eventually, inevitably we will end up being able to trace everything back to factors which were entirely beyond your control. So were you really free not to steal your neighbour's car? These philosophers will say that nature and nurture together conspired to make you the sort of person who would steal a car if they thought they could get away with it. But the nature is not your fault, and the nurture is not your fault, either. So it's not your fault that you ended up stealing the car.

SL

There are different ways of responding to this problem. For example, some philosophers insist that, once we clarify what's meant by ‘free will’, we will be able to see that free will is actually compatible with our choices and our characters ultimately being entirely a product of external causes acting upon us. Other philosophers bite the bullet and deny we have free will or moral responsibility – they insist on the ‘common sense’ view that we do have these things is simply mistaken. However, I know that you have a very intriguing and unusual response to the problem. Would you mind spelling it out for us?

HS

Thanks, Stephen. I wouldn't want to take either of the routes you suggest. I'm a libertarian about free will – which means I believe we have free will even though to have it requires that deterministic pictures of reality (such as the one implicit in the story I told about why you ended up being the sort of person who would steal a car if they thought they could get away with it) are false. There are many different deterministic pictures of reality – but let me focus on the one implicit in my example. It's deterministic in the sense that it suggests that everything you do is determined by various properties you have (properties of character, motivation, beliefs, etc.), perhaps in combination with the properties of the world around you, which dictate your opportunities for action (e.g. I can't steal a car within the next five minutes if I'm on the top of Scafell). But I think this deterministic picture is mistaken.

To think in this way is to think that agents (things that can act) are just like inanimate objects in that their behaviours are dictated by their properties and the external influences upon them, in combination. But I maintain that there are no good reasons for thinking that this is so. What is the argument for it supposed to be? (As an aside – it isn't even obvious that the view is true of all inanimate objects, as quantum mechanics has suggested, but that's another story.) In my view, we shouldn't be assimilating properties of character, motivation, belief, etc. to properties like shape, mass, density, etc. (which might dictate the behaviours of physical objects in very carefully controlled closed systems) and we shouldn't be assimilating actions to other sorts of events. On my view, agents have the capacity to settle things – to make it the case that things go one way in the world when they could – really could – have gone another. The cup sitting next to me on my desk doesn't have that capacity. And that's a very fundamental difference between me and the cup. The actions of agents aren't determined merely by the properties of their agents, though of course they're strongly influenced by some of them in a very wide variety of respects. But just as we tend to think, we are always confronted by a range of alternatives which are realistic possibilities for us – and the causal structure of the universe at any given time doesn't limit our possible futures to one.

SL

That's an intriguing suggestion, but I imagine some readers will be puzzled. Of course a human being has thoughts and feelings and so on, which makes them rather different from a mere physical object like a brick or a car. But still, they are physical objects. And (setting aside quantum indeterminacy) can't the behaviour of all physical objects be predicted given knowledge of the laws of nature and prior physical state of the universe? Just as we can predict the movement of a billiard ball across a table, or a rock rolling down a mountainside, or a planet circling the sun, given sufficient knowledge of the laws and the relevant physical system, so someone could in principle predict what this physical object (I am pointing at myself) will do today, tomorrow, next week, and so on, given sufficient knowledge of the laws of nature and how things stand physically right now? But if I, as a physical thing, am in the grip of the same laws of nature as any other physical object or system, then surely I cannot be said to have any free choice about what I'll do, or be held morally responsible for what I do? So is your suggestion that the laws that govern all other physical systems do not apply to physical systems like myself and other agents?

HS

Thanks, Stephen. What makes you say that ‘the behaviour of all physical objects can be predicted given knowledge of the laws of nature and the prior physical state of the universe?’ I don't think this is always true even of things like billiard balls, actually. You can only accurately predict the movement of a billiard ball across a table if you idealize the system as closed – that's to say, subject to no extraneous effects from outside. But what if a person reaches over and picks up the ball from the table? Or an earthquake happens which splits the table apart so that all the balls fall on the floor? Or the frayed cable holding up the light above the table finally breaks and the light crashes onto the table stopping the ball in mid-journey? One might say that if we knew enough, we'd be able accurately to predict these sorts of interfering events too. But what would be the argument for saying so? I grant that sometimes it can be reasonable to rule them out (perhaps e.g. there is no person who could reach the table in time, we aren't in an earthquake zone and the cable has been inspected and looks fine). And so there may indeed be little oases of predictability in the world, such that one can be very sure indeed about certain aspects of the local future of those oases. But what I question is the move from our knowledge and experience of these little oases of predictability to the claim that therefore we can be sure that the future of the whole universe could be predicted if we only knew enough about the facts and the laws. That seems to me to be a giant non sequitur, which begs all the important questions.

I think it was fair enough to be very tempted by this giant non sequitur in the wake of the scientific revolution and in particular by Newton's unification of a number of phenomena that had hitherto been regarded as separate – such as the motions of heavenly bodies and the motions of objects on earth. At that point it was obviously very tempting to indulge in images of the world as a giant mechanism which we could predict if only we knew the details. But modern physics looks extremely different from the way it looked in Newton's day – and I would like to see a decent argument for the view that universal determinism has any support from physics (as opposed to support from some physicists: physicists aren't any more immune to the dominant ideologies of the day than the rest of us).

Am I in the grip of the same laws of nature as any other physical system? Yes (or at least, I am in the grip of all those laws that apply to me and my parts – not all will, of course). I am totally unequivocal about that. I am a physical entity and the physical laws which apply to me do indeed apply. I am not any kind of exception to them. But what I question is that those laws in their totality are sufficient to dictate rather than merely to constrain what I do. It's that which is the giant assumption to which determinists are not allowed to help themselves.

SL

Many thanks Helen. I wonder if we can supply an inductive argument for determinism? Scientific theories are supported by inductive reasoning: for example by generalizing from a number of observed cases. An example: we might reasonably conclude that it's a law that all actions are accompanied by equal and opposite reactions if we observe enough instances of that pattern (this action was accompanied by such a reaction, that action was, and so on) and no counter-examples. But now can't we reasonably conclude that (setting aside quantum indeterminacy) we do inhabit a deterministic universe? Suppose the laws we know of predict a certain path for a planet. We then find that the planet doesn't follow that predicted path. And then it turns out that's because some other previously unknown planet is tugging on the first planet and pulling it out of its predicted orbit. Of course that's no counter-example to the claim that those laws apply deterministically: universally and without exception. The interfering planet and its effects may itself by in the grip of those same laws. And in fact we might confirm that this is the case – that the path of the interfering planet and its effects are exactly what we'd predict given those laws. Now, you just pressed the question: what's the argument for supposing that if we knew enough we'd always be able to predict these sort of interfering events (unknown planets, the earthquake that shakes the billiard table, etc.) too? My suggestion is: can't we offer an inductive argument? Whenever things fail to go as the laws predict, and we look for such interference, we typically find it, and we also find that interference and its effects are itself predicted by the theory (plus, even when we don't manage to identify such interference, it could still easily be there). Given this observed pattern, isn't it reasonable to generalize – to conclude that everything obeys those laws without exception? Of course it's possible that exceptions to those laws may yet show up. But, given we never find them, doesn't it become increasingly unreasonable to believe such exceptions exist – to believe we don't inhabit a deterministic universe that dictates everything that happens, including what we do? Isn't that a pretty good argument for determinism?

But then, given a very strong case for determinism, and the incompatibility of determinism with free will and perhaps even agency, isn't it free will and agency that will have to go, not determinism?

HS

Thanks, Stephen. That's an interesting suggestion – but I don't think it works. The reason is that the example you've chosen is rather special. The motions of planets are an especially well-governed realm where things do tend to be pretty predictable! But if we considered other examples, I think the situation would look rather different. In particular, there are lots of situations where this kind of predictability isn't the order of the day at all. To take one of Nancy Cartwright's examples, if I let a $100 bill loose in St Mark's Square, I won't be able to predict with any accuracy where it will come to land, even if I know the wind speed and direction. At best, I could offer a probability distribution. There's no such thing, in a case like this, as ‘the way the laws predict things will go’. And there are many other situations like this – when one considers such things as the motions of (e.g.) fluids, the growth of plants, and of course the actions of animals (not to mention quantum mechanics).

There is always a question, when we propose making an inductive inference, how to judge the correct scope of the cases to which I am going to extend my inductive reasoning. This is always an issue with induction – because any inductive base can contain its own peculiarities. For example, if I'd used the population of England c. 750 ce as a basis for drawing conclusions about the skin colour of human beings in general, I'd have made a massive error – because I'd have taken a special case as an inductive basis for a generalization to the whole human race, of which I'd happen to have seen only a very small and local fraction. What I'd want to say is that it is perfectly possible that to choose planetary motions as an inductive base for believing the whole universe to be deterministic is to make a similar sort of error. We have good reasons for supposing that the kind of order displayed here is very local (not geographically local, of course, but local in terms of the type of systems governed by these sorts of regularities) – and so we shouldn't be performing inductions on that basis. To do so would be extremely rash.