An example: Free will and determinism

I have been arguing that philosophical questions run into other areas—religion and science, in particular—so that they cross bound­aries between subjects.

The basic problem of freedom of the will, which I will be spelling out more fully in a moment, can be simply stated: If everything we do is caused by earlier things that we didn't do, how can we be morally responsible for our actions? This isn't just a question in the philosophy of mind, because it involves morality, and it isn't just a question in ethics, because it involves problems in the philosophy of mind. It also obviously involves the truth of determinism—whether everything that happens was in fact caused by other things that hap­pened earlier—which came up in the chapter on philosophy of sci­ence. But questions about freedom of action raise issues about what it is possible for us to do, and questions about possibility and neces­sity lead, as we saw in Chapter 8, straight into metaphysics. Furthermore, since there are versions of the free-will problem that arise from thinking about the compatibility of human freedom with God's knowing in advance what we are going to do, the issues raised by freedom of the will can lead us into epistemology and philosophy of religion as well. Finally, since it's presumably wrong to punish someone who isn't responsible for their acts, free will is a central issue in the philosophy of law, and thus of political philosophy.

So, as I say, the cluster of problems about freedom of the will is an important example of the way in which some philosophical ques­tions cross boundaries within the subject, just as—in the ways we have just seen—many questions cross the boundary between phi­losophy and other disciplines. Many of the questions that we have discussed in this book start in one broad area—philosophy of mind, say—and end up drawing on others—logic or epistemology, for example. But freedom of the will is a question that begins at an intersection: it starts at the junction of metaphysics, philosophy of mind, and ethics.

Because the problem of free will has so many ramifications, you can start to explore it in many places. One possible point of entry is with the functionalist picture of the mind from Chapter 1. The basic idea of functionalism, you'll remember, is that the mind is a system of causal relationships: to have a mind is to have states that have characteristic causes and effects. “Pain is caused by pinpricks and causes brow wrinkling,” said the simple-minded theory of pain in section 1.7. Among the most important characteristic effects of a mental event are the things we do, our actions. Now, ordinarily, our actions are caused by our beliefs, our desires, and our intentions. I go to the kitchen because I want to make myself a cup of coffee. So my desire for coffee, my belief that there is some coffee in the kitchen, and so on produce in me a desire to go the kitchen. I then form an intention to go there, and that's what sets me walking. The fact that my mental states caused me to move by this sort of process makes this movement an action of mine.

Sometimes, by contrast, a person's body moves, and this isn't an action.

Notice, however, that on the functionalist view, even when I do something as a result of my own mental states, those states them­selves will have causes. That's part of the picture, too. And if we trace the causes back far enough, eventually we'll get to something outside me. My desires are shaped by my genes and my environ­ment, and so (though in different ways) are my beliefs. As a result, there's a sense in which every action of mine, though caused by mental states and events of mine, is ultimately caused by things out­side me as well.

So far, of course, no problem. But now let's add some moral ideas to this set of ideas in the philosophy of mind. Consider, once more, the two ways that my hand might close: as the result of my inten­tions and as the result of an electric current turned on by you. Suppose that my hand is holding a detonator switch, and if it closes, some dynamite will go off, blowing up a dam, which will flood a val­ley. Now, for moral purposes, it matters very much which way my hand closes. Suppose we both know that this is the setup: know, that is, about the dynamite, the dam, and the valley. If my hand closes because I choose to close it, because it's my intention to close it and so to set off the explosion, then the flooding of the valley is my responsibility. If my hand closes because you turn on the current that contracts my arm muscles, then it's your responsibility. Generally speaking, we seem to accept something like the following principle of moral responsibility:

MR: If you are to be morally responsible for something that happens, that happening must be (or be the result of) an action of yours.

So, on one hand, all of our mental states and events have earlier ultimate causes outside our minds; on the other, we're responsible only for what is caused by our own mental states or events.

But if that is so, we can reason as follows. True, when I clenched my hand and, in effect, flooded the valley, that was the result of my intention. My intention, however, was itself the result of earlier mental states of mine whose ultimate causes were outside me. So while I was responsible for my act, according to the principle of moral responsibility, MR, I wasn't responsible for my intention. But if I wasn't responsible for the intention, why am I responsible for the act? It seems very odd indeed that I can be responsible for the con­sequences of my intention even though, by MR, I am not responsi­ble for the intention itself.

There is a further problem, which comes out if we consider another fundamental principle governing moral responsibility.

MR₁: You are morally responsible for an outcome only if you do something that caused that outcome and you could have done otherwise.

But I could have done otherwise only if I could have had a different intention, and (if determinism is true) the intention was, in fact, the result of earlier events—both inside my mind and, ultimately, out­side it—over which I had no control. So I am not morally responsi­ble for my acts.

The view that the fact that our mental states are causally deter­mined means that we do not have free will is called “incompatibil- ism.” Incompatibilists say that free will is incompatible with deter­minism. If we don't have free will, we aren't responsible for our acts, and so incompatibilism leads to the view that our conventional ways of assigning moral responsibility are misguided. If that is right, then, of course, it will have far-reaching consequences for such practices as punishment.

Many philosophers have held, however, as we shall see later, that we can still be held responsible for our acts, even if we are causal systems whose ultimate causes lie outside us. This position is called “compatibilism,” and there are many steps in the argument I just gave for incompatibilism that a compatibilist might want to ques­tion.

MR, for example, seems much too strong. People can be respon­sible for things they didn't do as well as for things they did. And not doing something isn't an action. Thus, for example, people can be responsible for something that happens because they had a respon­sibility to stop it happening. If I was in charge of the dam and it flooded because I failed to open a sluice to run off some excess water, then I could be responsible for the flooding even if I was fast asleep at the time. So, perhaps we should modify MR to read:

MR₂: If you are to be morally responsible for something that happens, that happening must either

a)        be (or be the result of) an action of yours, or

b)     be the result of your failure to act in circumstances where you ought to have acted.

We could summarize the ideas here by saying that you are respon­sible for what happens only if either—this is (a)—it was (or was the result of) something you did or—this is (b)—it was the result of your negligence. In a slogan, we are responsible only for our acts and for our negligence. And we can now modify MR₁ as well to read:

MR₃: You are responsible for an action only if you could have done otherwise, and you are responsible for a failure to act only if you could have acted.

But these modifications of the principle of moral responsibility don't really help get us out of the difficulty we are in.

There's another point at which this argument might seem vulner­able, however. For it depends explicitly on the assumption of deter­minism. But, as I suggested in Chapter 4, contemporary science suggests that determinism isn't true. So, does the falsity of deter­minism—the truth of indeterminism—offer a way to escape the conclusion that I am not responsible for flooding the valley?

Contemporary physics in fact offers two reasons for thinking that, even given a full specification of the past states of the universe, you cannot predict everything that will happen in the future. One of these is what is called technically (and entirely appropriately!) “chaos.” There are many processes in the world—the weather among them—that are governed by laws that have the following property: given a finite difference in initial conditions, however small, you can get very large differences in outcome. Such systems are said to be “chaotic.” So, to use an example that has often been invoked in discussing the weather, the difference between a hurri­cane and a lovely sunny day in Jamaica could be the result of the fact that a butterfly flapped its wings in West Africa some days ago. Chaos is an extremely important phenomenon, but it isn't relevant to the truth of determinism. For you can have chaos in systems that are entirely deterministic. What chaos shows is that it is wrong to assume that because a system is deterministic you can know how it will develop in advance. For, given chaos, that would require the ability to know every relevant fact to an arbitrary degree of preci­sion, which isn't possible.

The claim of modern physics that is relevant to the truth of deter­minism is not that the world is chaotic—in the technical sense—but that the fundamental laws of nature, the laws embodied in the quan­tum theory, are irreducibly probabilistic. The fundamental laws, that is, state not that E (an effect) will happen if C (its cause) does, but that E has a certain probability of happening if C does. Suppose, for example, that the quantum theory says that the probability that a certain α-particle will be emitted from a radioactive sample in a cer­tain interval of time is 50 percent. That will mean that in a sufficiently large sample of emissions, we will find α-particles emitted in that interval roughly half the time. But quantum theory says that there is no physical difference between the cases where a particle is emitted in that time interval and those where it isn't. The difference, then, is not the result of some so-called hidden variable; it's just a fact that the world contains events that have a certain probability of hap­pening in a certain time interval, and that's all that can be said about it. Fundamental physical processes are thus sometimes random— what happens is not determined by earlier events.

Now, whether this claim is true or not is a question for physics (though there is a good deal for philosophers of physics to say about how the physics should be interpreted). So let's suppose that the physicists are right and the world really is indeterministic. Does this help with the problem of free will?

Unfortunately, I think the answer must be no. To see why, remember that indeterminism means that there can be two kinds of events. Some are fully determined by earlier events: the quantum laws say their probability, given the earlier states of the world, is 1. It may be that there are, in fact, none of these in this possible world, but if there are any, they are the determined events. Other events have probabilities less than 1. These events, we can say, are “par­tially random”: they are not fully determined by earlier events. So, according to modern physics, when I form my intentions (if they are the results of physical events, as functionalism supposes) that process is either determined or partially random. If they are deter­mined, we are still in trouble. But if they are partially random, we are left with a new problem.

For if my intentions are partially random, then whatever made me form the intention, it wasn't something that was under my con­trol. According to the indeterminist, it just happened—and it could just not have happened. Talk of what “could have happened” should remind us that we can express the matter here in terms of possible worlds. If my forming my intention was partially random, there are some physically possible worlds that are exactly the same until the moment where I formed the intention, in which I didn't form the intention. Those are the worlds that make it true that I could have failed to form the intention.

Philosophers going as far back as Epicurus (who died in 270 B.C.) have thought that there is a place here for free will. Lucretius, in Book II of his De Rerum Natura (On the nature of things) asked, fol­lowing Epicurus:

If atoms never swerve and make beginning

Of motions that can break the bonds of fate

And foil the infinite chain of cause and effect

What is the origin of this free will... ?

We could imagine a modern Lucretius supposing that a person's mind could step in and provide the explanation for the difference between the possible worlds where a particular action happens and the ones where it doesn't. But the intervention of the mind in this way would raise at least two problems.

First, this proposal requires mental events that intervene in phys­ical processes from, as it were, “outside” the physical realm: and this raises all the difficulties of interactionism that we identified when discussing Descartes' views in 1.2. Do we really want to be driven to take up the difficulties of dualism in order to avoid the problem of free will?

Second, because the quantum laws say that the probability of events is fixed, this sort of mental intervention could produce events that were more and more unlikely. If the mind can intervene in the process, then it would be possible in principle for a person to inter­vene repeatedly in a way that ended up producing a sequence of events that the laws of physics said were fantastically improbable. There is in fact a device that was designed by people interested in investigating extrasensory perception that is meant to test this pos­sibility. It is called the “Schmidt machine,” because it was invented by an engineer of that name. The basic idea is simple. You set up a device with four lights; which light goes on depends on when a radioactive sample emits particles. When the machine is left run­ning alone, each of the lights is on one-quarter of the time. Because a radioactive sample emits radiation in a way that quantum theory says is irreducibly probabilistic, quantum theory says that which light is on at any time is not determined in advance. Now you give a person the chance to press one of four buttons, depending on which light she wants to go on. If there is a statistically significant correla-

tion between the button pressed and the light that goes on, then we have evidence that this process, which is random when there is no one around, can be affected by thought. (Of course, there are many other possible explanations: remember the discussion of theory- ladenness in 4.8.).

Now suppose that someone were to postulate a sort of inner Schmidt machine, where mental events directed physical events that were otherwise random. Then minds would be sites of devia­tions from the basic laws of nature. And, in fact, it would follow that these supposedly basic laws were not basic, since these events would no longer in fact be random. So this possibility is just inconsistent with the idea that the world is fundamentally indeterministic.

The idea that the mind can intervene to opt between otherwise random processes is no help. And that means we are left with only two options. If my intentions are causally determined, they’re not my responsibility. But if they’re not causally determined, then they aren’t determined, in particular, by me; and so they’re not my responsibility either. In what follows, I shall conduct the argument as if determinism were true, since, as we have just seen, it wouldn’t help if it weren’t.

9.11