The first two posts in this series have effectively treated free will as an unanalysable concept. This is not necessarily a bad thing; you can get away with unanalysable concepts if they’re simple enough, unambiguous and not the subject of the controversy at hand. But that’s unfortunately not the case here. So toward the end of part two, I began to move away from this, when I referred to free will as entailing that the actions of free agents must be irreducibly unpredictable: that is, unpredictable even with mathematical probabilities. Most arguments about the subject, even when they aren’t framed in quite those terms, rest on whether or not our actions are irreducibly unpredictable, so this is a good definition to use when analysing those arguments.
I believe, and will attempt to explain why below, that science does not disprove the existence of free will as usually understood. But as usually defined, as irreducible unpredictability, it does. The mind is a function of the brain, albeit a supremely complicated one, and the brain obeys the same (predictable or at least predictably random) laws of physics as the rest of the universe. Even if there were literally some form of soul or spirit of a separate substance to the brain, we can be pretty certain that it leaves no evidence of itself in the physical world — which is just another way of saying that it has no effect on the physical brain and its actions. If it has no effect on what we do, then it cannot be a mechanism to explain anything about the mind, free will or otherwise. There is, quite simply, no room for irreducible unpredictability in the brain.
Irreducible unpredictability makes for some fun arguments. But those arguments, for all their flair and rigour, are not actually about free will, in terms of what we actually want the concept to do once we’ve wrung it out. We don’t want to think we have free will so as to reassure ourselves that we (or other people) are fundamentally unpredictable. Rather, we want to think so in order to feel we have some legitimate basis on which we can take credit for our actions and blame others for theirs. Or, more rarely, blame ourselves and praise others. On an even more fundamental level, it seems necessary in order for us to have a sense of ownership over our thoughts, as well as our actions. And it’s not at all clear how defining free will as (entailing) irreducible unpredictability could help us do this.
We don’t think people are free to the extent to which they’re unpredictable. We don’t think a completely unpredictable person is normal; we think he’s a looney. We might consider a particularly predictable person to be dependable (if we like them) or boring (if we don’t), but we don’t think they’re less of a whole person for it.
We want our actions, and those of others, to be determined. We explain villains in terms of their tragic past, while acknowledging that this does nothing to excuse their actions. Our minds are determined entirely by what goes into them. You can’t imagine a colour that you haven’t seen in the same way as one you have. This can be very direct. If I tell you to not think about elephants, you immediately think of them. You can’t help it. But does that really mean you don’t have free will at all?
It only means that the freedom of your will is not unlimited. But of course it isn’t. You can’t calculate the value of π to a million digits in your head just by willing it. You can’t picture a colour you’ve never seen in the same way as you can picture the blue of the sky on a clear day. Most people can’t even consciously choose which gender they are, or which flavour of ice cream to prefer, or who they fall in love with. But we can choose what to do about those things, and to that extent, we have free will. Having murderous impulses doesn’t make you a murderer; acting upon them does. There is an extent to which we can deliberately affect our internal states. Consciously trying to become more generous of spirit, for example, or to genuinely appreciate a genre of art we find challenging. But such attempts too rely on a deterministic conception of the mind, as something that can be affected by things and affect things in its turn, including itself. This reflexivity is important, and I’ll come back to it later.
It is, to be sure, a more limiting conception of free will than the incompatibilist version used by both sides of the traditional argument. But it’s both more accurate in terms of describing the real world, and closer in its effects to what we actually want the concept of free will to do for us. We don’t need a less limiting conception of free will, any more than we need one that says we can fly.
And that point about what we want the concept to do for us is very important. The first question you should ask when you encounter an abstract concept like free will is what you want the concept to achieve. When you ask “are human actions irreducibly unpredictable?”, you’re not asking the same thing as “do people have free will?” But often people don’t realise that, because they load a preconceived definition of “free will” into the question. And their preconceived definition has nothing to do with the question they really want the answer to, that is, whether we can own our own thoughts and take credit or blame for our actions.
Some opponents prefer to describe this compatibilist definition of free will as autonomy, as it emphasises that we have free will to the extent we’re not entirely controlled by other wills. But that’s purely a semantic quibble. And as I’ve demonstrated, the concept in question — whatever you prefer to call it — does anything we could reasonably ask of the concept of free will, and does it better than the alternatives, which aren’t really free will at all.
There remain a couple of points I want to discuss. The first comes back to the predictability of the mind. There is a very good reason why people think about free will in terms of irreducible unpredictability, and it has nothing to do with being stupid or wanting to simplify the problem. Rather it has to do with the reflexive, or introspective, nature of consciousness. Consciousness involves having a privileged view of your self. You alone are aware of what your thoughts really are, in a qualitative way that nobody else could be. It’s difficult, if not impossible, to imagine what it would even be like to see inside someone else’s mind in the way you see inside your own. How could that happen without you effectively becoming that person? And conversely, how could anyone or anything outside your mind predict your thoughts without effectively becoming you?
If human minds are, even in theory, determinable, then it follows that everything that happens in your mind, including this sense of self-awareness, could be predicted from outside. You don’t yourself know what you’ll be thinking about in an hour from now, but someone with a perfectly accurate Prediction Machine™ could tell you. Surely, if it’s possible to know what you’ll do before you decide to do it, you’re not free in any meaningful sense — you don’t own your own thoughts, decisions, or actions after all.
In fact, we don’t need to go so far as to imagine a Prediction Machine™ so accurate it could never exist except in theory. A neurologist with a brain scanner can tell what decision you’re going to make — for example, whether to raise your arm — whole tenths of a second before you’re aware you’ve made it. Such experiments have been repeated many times, and there’s little doubt as to their veracity; they’ve been hailed by scientists and philosophers alike as the final nail in the coffin of free will.
Not so. A blip on a brain scanner is evidence of your decision to raise your arm in the same way that your arm ascending is — no more, no less. The blip is caused by your decision to raise your arm, not the other way around. The few tenths of a second it takes for you to be fully cognisant of that decision is analogous to the few tenths of a second it takes for you to process images from your eyes or sounds from your ears. The fact that a camera’s sensor registers an image faster than the vision centres of the brain hardly dealt a blow to the theory of human perception. The fact that the brain scanner can detect brain states quicker than the brain itself should be similarly untroubling.
Even if things were determinable on scales much longer than a tenth of a second, on a scale of minutes or days or even decades, all that this would mean would be that the results of any such determination would be the result, not the cause, of what was going to happen in your brain. But the catch here is that we’re back in the territory of a Prediction Machine™ again, and any Prediction Machine™ able to even remotely accurately predict (rather than merely observe, as the brain scanner does) a person’s thoughts must be powerful enough not just to perfectly simulate their brain, but also the brain of every person who causally interacted with them (which on a long enough timescale adds up to everyone in the world, due to the chaotic nature of causation).
And it also, and here’s the kicker, would have to be able to fully simulate its own operations, on top of all that, in order to know the effects of its predictions on the people it was simulating. In other words, while I couldn’t say exactly or even roughly how complex it would have to be, I can say with complete certainty that it would have to be more complex than itself. Such a machine is not just practically but logically impossible. And even then, its predictions would be determined by what was going to happen inside the minds it was simulating — not the other way around. We should be thankful that it’s impossible, not because it preserves our autonomy, but because the logic of time and causality would be violated if it weren’t.