chapter 4: personal identity and free will

Here is a story of a sort that you might find yourself hearing about on the news or reading about one day in the morning paper: A middle-aged housewife whose married name is Susanna White has been arrested and allegedly identified as the notorious Susan Black: a young student from a wealthy family, who was kidnapped in 1968 by a radical group and persuaded by her captors to take an active part in a bombing, in which several people were killed. Further investigation reveals that Susan Black escaped by altering her appearance and obtaining a forged driver’s license under another name. She has since married, had three children, and has apparently been living for many, many years as a respected member of a suburban community, one who actively supports many worthy causes and has never again engaged in or had anything to do with radical or terrorist activities. Federal and state authorities nonetheless regard Susanna as still responsible, both morally and legally, for the crimes in which Susan was involved, and they charge her with murder, along with a number of lesser crimes.

Along with many other issues, both moral and legal, this example raises two important metaphysical issues having to do with human persons. The first and more obvious pertains to identity across time. The claim involved in the story is that Susanna (the housewife) is identical with—the very same person as—Susan (the radicalized student). But what is required for this claim of identity to in fact be correct—in a way that, among other things, makes Susanna genuinely morally responsible for the things that Susan did? Does it matter whether or not Susanna has any memory of the earlier events? Does it matter whether her basic character traits are the same as Susan’s or perhaps partially or wholly different? A simple answer might be that all that matters is that Susanna’s body is the same ongoing human body that belonged to Susan, as might be established by things like fingerprints and DNA. But we will see that this simple answer is by no means obviously correct. The first part of this chapter will investigate this issue: the problem of personal identity.

A second issue that is raised by the Susan-Susanna case—one which is also relevant, though in a quite different way, to the issue of moral responsibility—is whether Susan acted freely when she participated in the bombing: whether she did so as a result of a genuinely free choice. In this case, there are special reasons for doubting whether Susan’s action was free. Perhaps she was coerced in some way that was not immediately apparent. Or perhaps she was psychologically conditioned or “brainwashed” in a way that made her action not genuinely free. Both of these possibilities represent ways in which Susan’s action might have been causally determined by forces outside her control in a way that seems, initially at least, to be incompatible with genuine freedom. But apart from the special circumstances in this case, there are also much more general reasons for doubting where Susan’s choice was genuinely free, reasons that apply to all human actions. The second part of this chapter will be concerned with the general issue of whether human actions or choices are ever free and of what genuine freedom would require: the problem of free will.

The problem of free will

The second metaphysical issue raised by the Susan-Susanna case is whether Susan acted freely in participating in the bombings, as seems required for her—or Susanna—to be morally responsible for the results. As we have already seen, it would be easy to devise more specific versions of the story that would make it very doubtful whether free choice and moral responsibility should be ascribed to Susan. Perhaps the other radicals were armed and she was not, and she was threatened with being instantly killed if she did not play her assigned role accurately and with apparent zeal. Or perhaps she had been “brainwashed”: psychologically conditioned through such things as sleep deprivation, mild torture, threats, misinformation, and so on, so as to cause her to act in ways that did not represent her true values and beliefs.

But suppose that what actually happened wasn’t like either of those scenarios. Suppose that although Susan was indeed held briefly against her will, she was only given some statements of the group’s views to read, statements that she found to be initially quite persuasive, and that after discussing them further with the members of the group, she found herself fully in agreement with them. Suppose further that when the leader of the group offered at this point to release her, she expressed a desire to stay, join the group, and actively participate in its planned actions—including the bombings, whose purpose she now understood and endorsed. Given the overall set of circumstances, it would still be at best an open question whether Susan’s choice was really free. Perhaps she was still suffering from a kind of psychological shock as a reaction to the kidnapping, which caused her to do things that she would not otherwise have done. But further details might also make it quite plausible that her choice was in fact an instance of genuine free will, based on information and values that she had adopted in a considered and rational way. And then it might seem correct to hold her morally responsible for her subsequent actions and for their results.

But even given further details of the right sort, there are still questions that can be raised about all this. Many people would not have been so easily persuaded. And if Susan was, then this must be a result of her previous moral and political opinions and her general beliefs about the world, together with various aspects of her formed character and values. And where do all of these things come from? Clearly they are in turn largely a result of Susan’s whole history, of the ongoing environment in which she was raised and grew up. This environment includes all of the various external influences on her, including those exerted by family and friends, by all of the various things summed up as the media, by other people and situations that she encountered more causally, and of course by all of her formal education at various levels. Obviously a person for whom some of these factors were different, might have been less (or even more) receptive to the radical group’s views and plans. And in addition, the kind of person Susan was and the various sorts of capacities she possessed was strongly influenced by her basic genetic endowment, her heredity.

Sorting out the contributions of environment and heredity to Susan’s overall makeup is far from easy. But if we reflect on the multifarious influence of each on Susan’s basic nature (as on that of any person), the following question suggests itself: might it be that Susan’s response to her situation after being kidnapped and all of her subsequent actions were completely caused, in every last detail, by the combination of her environment and heredity, together with her immediate circumstances? Might it be that given all of those influences, there was only one series of actions that was genuinely possible for Susan, in exactly the same way that when a rock rolls down a hill, everything about the way it moves is entirely caused by its own nature (its size and shape and distribution of weight, and so on), together with the external influences that are exerted by it at various points? Might it be that Susan—or any person—is merely much more complicated than the rock in her nature and in the sorts of things she is influenced by, but no more free? Or is there a way of understanding Susan’s situation that makes room for genuine freedom, despite the obvious and pervasive causal influences on her? Different positions on the free will problem give radically different answers to this difficult question.

Hard determinism

Many scientists and philosophers in earlier times advocated a general thesis about the way that the universe works that is highly relevant to the question just asked. According to this thesis, the thesis of causal determinism, every event that takes place in the universe is completely determined by the combination of antecedent events and laws of nature. This means that at any point, given those determining factors, only one thing can happen. It also means that given a complete knowledge of the state of the universe at some point in the past, however far back, together with a complete knowledge of the laws of nature, it would be possible in principle to predict the entire future course of the world, including all human actions, in complete detail. (Indeed, assuming that the laws of nature are of the sort standardly assumed, ones that work both backwards and forwards in time, the same thing would work in reverse: from the same sort of starting point, one could also deduce—retrodict—the entire previous history of the universe.)

The thesis of causal determinism seems to be strongly supported, at least initially, by both common sense and science. The natural reaction of common sense to an unexpected or unusual occurrence is to ask what caused it. In most or perhaps all cases, the suggestion that perhaps it wasn’t caused at all would be very difficult to take seriously. Moreover, it can be plausibly said that the fundamental task of science is to determine the causes of various sorts of occurrences, along with the general laws according to which these causes operate. And this endeavor has been enormously successful, as reflected in the scientific and technological marvels that surround us. Thus the whole scientific approach seems to presuppose that there are always determining causes to be found, and the success of science seems to show that this presupposition is correct.

But if the thesis of causal determinism is true, then Susan’s actions, like everything else that happens, were completely determined by antecedent causes—presumably at least principally by her environment and heredity, though adding other sorts of causes makes no real difference to the general picture. This means that someone existing, say, 10,000 years before Susan was born, would with the right sort of knowledge have been able to predict everything about her—including, of course, how she would react to the situation where she was kidnapped by the radical group. And this in turn makes it hard to understand how Susan’s actions could have been free in any meaningful sense. To be free, we are inclined to think, Susan must have been able to do something other than what she actually did. And if causal determinism is true and there was only one thing that was actually possible, given her previous history, then Susan could not in fact have done otherwise. Nor, of course, could any other person in any other situation. Thus neither Susan nor any other person is genuinely free.

The view that accepts the general thesis of causal determinism and then reasons in the way just indicated to the conclusion that genuine freedom does not exist is called hard determinism. Unlike the thesis of causal determinism, which is a general thesis about the universe and says nothing specifically about freedom, hard determinism is a thesis specifically about freedom or free will—one of the three main general views that have been advocated on this issue (though, as we will see, there are importantly different variants of each of these views). If hard determinism is the right view, then freedom is a kind of illusion, and a further consequence, also advocated by the hard determinist, is that there is no genuine moral responsibility. How, the hard determinist will ask, can we hold Susan responsible for her participation in the bombings or for anything else, when there was in each case no other action that was genuinely possible for her? Wouldn’t that make as little sense as holding the rock responsible for the way that it rolls down the hill? No one would dream of trying to blame or punish the rock for what it does, but from this perspective it is just as absurd to blame or punish Susan—or, of course, Susanna. Â

But is hard determinism correct? The argument for it so far depends crucially on the thesis of causal determinism. And despite the initial plausibility of this thesis, there are in fact reasons for thinking that it is false, reasons that derive primarily from fairly recent results in the area of physical science called quantum theory. Though the details are complicated and are beyond the scope of this discussion, the general point is that according to the very highly confirmed results of quantum theory, the small-scale events out of which everything else is constituted are not in fact causally determined, but instead are in certain respects essentially random or chance in character.

Consider a simple example. Suppose that we have isolated a sample of the element radium, specifically the most common isotope, which is radium 226. Radium 226 is radioactive, which means that it eventually undergoes radioactive decay, in this case by emitting an alpha particle (that is, a helium nucleus) and being thereby transformed into a quite different element, the radioactive gas radon 222. The half-life of radium 226 is approximately 1620 years. This means that if we start with a pure sample, after 1620 years, half of the atoms in the sample will have undergone decay and turned into radon (while the other half will still be radium 226); that after another 1620 years, half of the remaining atoms of radium will have done this; and so on. But while predictions of this sort can be made in relation to a group of radium atoms, it is a consequence of quantum theory that for an individual atom of radium, no specific prediction can be made as to when it will undergo radioactive decay. We can say that there is a 50 per cent chance that a given atom in our sample will decay in the first 1620 years, a twenty-five percent chance that it will do so in the next 1620 years, and so on. But no more definite prediction is possible—not, according to quantum theory, because we do not know enough about the underlying mechanism, but simply because there is no underlying deterministic mechanism to be known. Instead, the decay of a particular atom is simply a random or chance event, one that is unpredictable even in principle.

And this is not just an isolated example. In fact, the entire material world is made up of materials and processes for which causal determinism fails to hold in ways like these. Many things may still be predictable at the macroscopic level of ordinary objects in at least approximately the way in which we can make statistical predictions about a group of radium atoms. But if we accept the scientific picture at face value, it follows that causal determinism is not only false but pervasively and massively false.

It might seem to follow that the threat to free will evaporates along with causal determinism, but this would in fact be much too hasty a conclusion. The results of quantum theory show that human actions, instead of being causally determined, are likely to have aspects that are instead random or chance in character. But, somewhat surprisingly perhaps, this possibility does little or nothing to enhance the chances for genuine freedom.

One reason for this is the possibility that human actions are still completely determined and predictable at the level of description that matters for the issues we are really concerned with. Perhaps while certain aspects of Susan’s actions were a result of random occurrences that could have been different, the randomness in question was so small-scale that all of the subtly different actions that were causally possible for her still were ones where she joined the group and participated in the bombings, even though she did so by moving and speaking in slightly different ways. If this were so, then even if the degree of randomness affecting her actions amounted to a small degree of freedom for Susan, it would not be sufficient to allow her to do otherwise with respect to participating in the bombings: that she did so in some way would still be determined, despite the possible variation in small details. (In the same way, the radium 226 atom is determined to eventually decay into radon, even though exactly when it does this is not determined—so that a degree of randomness operates within limits that are themselves determined.)

But the more important reason why there being a degree of randomness or chance in Susan’s actions does not yield genuine freedom is that even if the degree of randomness was large enough to yield significantly different possible outcomes, Susan still seems to have no real control over which of these outcomes occurred. Suppose that the element of randomness operated in such a way as to make it possible that Susan might have (a) joined the group and participated in the bombings (as she in fact did), (b) stubbornly refused to have anything to do with them, or (c) pretended to join the group and then informed the authorities at a crucial moment. If one rather than the others of these alternatives simply occurred by chance or at random, and if there is nothing further to be said (of the sort that some other, libertarian views attempt to add—see below), then Susan seems to be just as much a helpless victim as she was in the case where one outcome was determined. While it is true on this view that she might have been a hero (as in alternative (c) ), rather than an accomplice to murder (as she actually was), she had no way to bring it about that (c) rather than (a) actually occurred—any more than the radium atom has any way to bring it about that it decays at one moment rather than another. Instead, one of these things just happened at random, with Susan being no more than a helpless and no doubt frightened observer. (Try to imagine what it would be for your own actions to be in this way random between some significantly different set of alternatives: suppose that as you walk down the street, it is random whether you walk by a person coming from the other direction in the normal way, deliberately bump into him, or run screaming in the other direction. Clearly this would be very different from choosing freely which of these things to do. Â)

Thus even if causal determinism is in fact false and there is a degree of randomness or chance in human actions, the hard determinist claim that there is no room for genuine freedom—and so also no basis for genuine moral responsibility—seems to survive unscathed, indeed even strengthened. For someone who originally advocated hard determinism can now argue that human actions are either causally determined or random or (most likely) some combination of the two, and that on none of these alternatives is there any clear room for genuine freedom. Such a view is sometimes still called hard determinism, but it might perhaps better be labeled hard incompatibilism: “hard” because it denies the existence of freedom; incompatibilism because it denies that freedom is compatible with determinism (thereby disagreeing with the compatibilist view to be considered next).

Thus we see that no matter what the specific details of Susan’s situation may have been, there are serious reasons for denying that she was free in what she did or morally responsible for it. And reflecting on the basic hard determinist or hard incompatibilist argument that incorporates the possibility of chance or randomness seems to show that there are only two possible ways to resist this conclusion. Since it is quite clear that randomness is not to be identified with freedom, a view that advocates the possibility of genuine freedom must either: (1) deny that freedom is incompatible with causal determination; or (2) hold that there is some third alternative to an action being causally determined or merely random (where simply combining these two possibilities in various degrees does not help). These alternatives correspond to the other main alternatives on the free will problem, to which most of the rest of the chapter will be devoted.

Compatibilism

We have so far accepted without much question the hard determinist’s claim that freedom is incompatible with causal determination. But is this claim really so obviously correct? Here the place to start is with common-sense judgments about the actions of people in ordinary circumstances. It is clear that we ordinarily assume with no real doubt that people are acting freely in much of what they do: that, for example, they freely choose such things as when to go to work, what route to take, whether or not to stop at stores along the way, what to buy if they do, whether or not to call a particular friend at a given time, what to have for dinner, and so on. A hard determinist will say that we assume that such actions are free only because we are unaware of the ways in which they are caused, but in fact this is seriously open to question. For in fact if the issue was raised as to why a person did one of these ordinary things, we would normally expect there to be some cause for the behavior and would be very puzzled by the suggestion that it was simply uncaused. Why did Susan stop at a particular store on a particular day—perhaps the place where she was then kidnapped? Maybe because she wanted to buy some particular thing, maybe because she was curious about what the store sold, maybe because she expected to meet someone there—but surely not for no reason at all. And none of those possible causes for her action seem, from a common-sense standpoint, to be incompatible with the claim that her choice to act in that way was free. Perhaps freedom and causal determination of the right sort aren’t incompatible with each other after all?

The view that holds that an action can be both free and causally determined is called compatibilism. On the compatibilist view, freedom, rather than being incompatible with causal determination, is just a certain specific kind of determination: as a first stab, an action being determined by the agent’s own beliefs, desires, and other psychological states, rather than by forces that are external to the agent. If Susan stops at the store because she remembers that she needs some flour for a recipe she wants to try, her choice to stop and the resulting action are free, even though they are also (we may suppose, setting aside worries about randomness) causally determined by the combination of her states of mind and the relevant circumstances. She freely does what she wants to do, even though what she wants to do is causally determined. Whereas when she, having been kidnapped, climbs at gunpoint into the radical group’s car, she is not acting freely, because her action is not determined by her own desires, but instead by external constraint or compulsion. Perhaps she still has some small degree of freedom left, since she could choose to resist and presumably be killed. But she is not free to escape, which is what she strongly wants to do: her desire to escape is causally ineffective in a way that her desire to stop at the store was not. And if the radicals simply lift her and carry her into the car, she has, assuming that they are strong enough to control her, no free choice at all, because her own desires have no causal influence. Thus, according to the compatibilist, while freedom is certainly incompatible with some kinds of causal determination, it is entirely compatible with determinism of the right kind. (Compatibilism is also sometimes referred to as soft determinism, though this label is potentially misleading, since it suggests a view according to which the causal determination is less rigid, more flexible; in fact what is “soft” about “soft determinism” is only the conclusion that it draws from causal determination, namely that a causally determined action can still be free.)

We saw earlier that it is a plausible requirement for freedom that a person who acts freely could have done otherwise: could have performed some action that is relevantly different from the one that was actually performed. The main compatibilist view accepts this requirement, but insists that it must be understood in the right way. A person who acts freely, that is, whose actions are a result of his own beliefs, desires, and other psychological states, could have done otherwise in the sense that if he had wanted to do otherwise as a result of different beliefs and desires, he would have done otherwise. When Susan stopped at the store to buy flour, she could have done otherwise in this sense: if she had not wanted to buy the flour or if some other desire had been more important to her, then she would have driven on rather than stopping. Whereas when she was forced into the radical group’s choice, she could not, in this sense, have done otherwise. It is still true that if causal determinism is true in relation to Susan, then both her stopping at the store and her getting into the car were determined and predictable by causes that existed thousands or millions of years before she was born. But, according to the compatibilist, the specific ways in which these two occurrences were determined makes one of them a free action and the other not. (Though compatibilists acknowledge the possibility of randomness, they regard any random element as being just as incompatible with freedom as the wrong kind of causation.)

Compatibilists often suggest that the idea that free will or free choice is incompatible with any sort of causal determination results from a kind of semantic confusion. People confuse causation, which is quite compatible with freedom, with things like compulsion, coercion, and constraint, which are not. The radical group’s actions may compel or coerce Susan to get into the car. Their presence and behavior may impose constraints that limit her range of free choice or eliminate it entirely in important respects. But the fact that causation of these specific sorts is incompatible with freedom does not mean that causation of any sort is. And, after all, what more could we want for genuine freedom than being able to do whatever we want to do—having our actions determined by our own beliefs, desires, and other psychological states, rather than by external factors?

There can be little doubt that the compatibilist view draws the line between actions that are free and actions that are not in at least approximately the place where common sense draws it. But whether the compatibilist view really offers an adequate account of freedom and an adequate basis for moral responsibility can still be questioned. One problem for the compatibilist is the possibility of compulsive psychological states that result in actions that do not seem intuitively to be free. Consider a compulsive hand-washer: Due to a deep-seated and irrational belief in his own uncleanliness, he washes his hands a hundred times a day or more, causing very serious difficulties for his work, his family relationships, and so on. Despite fully understanding the consequences and wanting at one level to quit doing this, he is unable to stop himself, and eventually his life is ruined as a result: he loses his job, his wife divorces him, and so on. Is he acting freely when he goes to wash his hands for the hundredth time in a given day? His action is indeed caused by his own psychological states rather than by anything external, and it is true that he could do otherwise if his dominant desire were to do otherwise. But it is still very difficult to regard him as free—or morally responsible—in his handwashing activities. Â

How might a compatibilist respond to this case? One possible response is to appeal to the idea of second-order desires: desires whose subject matter is not particular actions and results in the world, but rather one’s own desires. The compulsive hand-washer’s strongest first-order desire is to wash his hands, even though fulfilling this desire interferes seriously with other things that he also desires. But he also has, we may suppose, a strong second-order desire not to have this compulsive first-order desire to wash his hands, but instead to have only more normal first-order desires. Unfortunately for him, this second-order desire is ineffective because the first-order desire is too strong and too compulsive. And this, it might be suggested, is why the compulsive hand-washer is not free: he can do what he wants, at the first-level, to do, but he cannot want what he wants, at the second level, to want. Thus the suggestion might be that freedom requires more than being able to do what you want to do: it also requires being able to want what you want to want. (Is this an adequate solution? What if the compulsive hand-washer is too confused or unreflective to have any second-order desires? Are second-order desires always required for freedom? How common is it for ordinary people to have such second-order desires? Might a second-order desire itself be compulsive? Â)

But the deepest problem for any version of compatibilism is whether a person who satisfies the compatibilist account of freedom is really free in any meaningful sense, even apart from worries about psychological compulsion. If we suppose that Susan’s actions in joining the group and participating in the bombing were free in the compatibilist sense, it still remains true that these were the only things that she could have done, given the total set of causes operating on her. (Again we set aside the possibility that there were elements of randomness as simply irrelevant to the possibility of freedom.) It is true that Susan could have done otherwise if she had wanted to do otherwise, but could she have wanted to do otherwise? She could have if some of her desires, beliefs, and other psychological states had been different, but (again setting aside randomness) they could have been different only if some of their antecedent causes had been different, and so on back to causes that were already in place long before Susan was born. Thus the sense in which Susan could have done otherwise according to the compatibilist seems to amount to saying that another person, one very much like Susan but different in significant respects, would have done otherwise. And it is hard to see why that fact makes the actual Susan either free or morally responsible.

When faced with this last sort of objection, compatibilists sometimes concede that freedom in the sense that they have defined is less than fully satisfactory from a sheerly intuitive standpoint. Intuitively, it seems that for Susan’s action to be free, more than one alternative must be genuinely open to her, and she must genuinely control which of those open alternatives actually occurs. But, a compatibilist will argue, once it is fully appreciated that randomness is of no help, freedom of this more intuitively satisfying sort turns out to be simply impossible. There is, he claims, simply no way for a choice between genuinely open alternatives to be other than simply a random event. So the choice is really between abandoning all talk of freedom (and moral responsibility) or accepting a conception of freedom that is intuitively unsatisfying in certain respects, but still draws the line between free and unfree actions approximately where common sense would draw it and thereby accords with at least many of our beliefs about freedom and free actions. His suggestion is that the latter of these alternatives is clearly preferable, even though it does not give us all that we might think we want. (Is he right about this? In particular, is this conception of freedom really enough to support any serious ascription of moral responsibility? Would it be fair to blame or punish someone who could have done otherwise only in the compatibilist’s hypothetical and seemingly irrelevant sense? Â)

Libertarianism

The first two alternatives we have considered offer us a rather dismal choice: One alternative is to agree with the hard determinist (or hard incompatibilist) that there is no genuine freedom and so also no genuine moral responsibility. In addition to being extremely implausible from an intuitive standpoint, this view would make nonsense of many of our most basic practices and institutions. Indeed, it is highly doubtful that anyone could ever be a practicing hard determinist: what would it be like to go through the world really believing that neither you nor anyone else could ever do anything other than what you actually do (except, possibly, in an entirely random way)? Â But the second, compatibilist alternative does not really alter this basic result, but only seems to disguise it a bit by describing some of the cases where only one action is genuinely possible as nonetheless free in a sense that does not really speak to the basic issue of whether the person could really do anything else.

Is there any further alternative? Hard determinists and compatibilists share the view that sheer randomness or chance is not the same thing as freedom and is indeed incompatible with it, but this is something that no one really questions. Beyond that, the common ground between these first two views is the claim that causal determination or randomness or some combination of the two are the only possibilities available. Thus a view that rejects both hard determinism and compatibilism must apparently deny this last claim and hold that there is a further possible way in which an action might occur, one that is different from both causal determination and randomness (and is not merely some mixture of these two). Libertarian views attempt to specify and defend such a further possibility, with the main issue about the libertarian view being whether they really succeed in doing this in an intelligible way. (Almost everyone would agree that a fully intelligible version of libertarianism would be preferable to the other two main views.)

What would a libertarian action or choice have to look like? What would it be for Susan’s choice to join the radical group to be free in a libertarian sense? Part of the answer is obvious: Susan’s action must not be completely causally determined, and it must also not be random, either entirely random or random within some causally determined set of alternatives. Various causally determined limits on Susan’s action are compatible with a significant degree of freedom, but within those limits the choice must genuinely be up to Susan in a way that means that she could have done more than one thing—and that she, rather than chance, controlled the outcome. Intuitively, this is just what a genuine free choice, one for which it is fair and reasonable to hold Susan morally responsible, should look like.

But the crucial question is how Susan brings about one of these alternatives, such as agreeing to join the group, in a way that is neither determined nor random. What does her control of the outcome really amount to? Different libertarian views give different answers to this question, and we will look at some of the leading ones. The question throughout will be whether any of them succeeds in making adequately clear how their alternative is distinct from both determinism and randomness.

Perhaps the most commonly held libertarian view appeals to the idea of agent causation. The suggestion is that Susan’s action, if it is genuinely free, is still caused, but not in a deterministic way by any set of antecedent factors, including any of Susan’s psychological states. Instead, the cause of Susan’s choice and resulting action is simply Susan herself, an intelligent and rational agent. Unlike the mechanistic causation involved in natural processes, which conforms to laws of nature, agent causation does not involve any laws and so no basis for prediction. If Susan’s choice to join the group was free, then she could have made a different choice even though everything about her, including everything about her psychological processes up to the moment of choice, remained exactly the same. But the choice was still not random or chance in character, because Susan controlled it.

The foregoing account has some of the right intuitive flavor, but is it really intelligible? If everything about Susan, including everything that went on in her mind, could have been the same and the choice have been different, why isn’t it simply random which of these alternatives occurred? What is the difference between a libertarian free choice and a random event? Assuming that a random selection of one of the alternatives is at least possible (and why wouldn’t it be), how could we tell whether it was random selection or agent causation that actually occurred? Could even Susan herself tell the difference? It’s not enough that she thinks afterwards that she caused the result, because that thought could itself just be part of the random result. And what does Susan’s alleged control of the outcome really amount to if she could have been exactly the same and still have done something different? Â

Some libertarians have tried to further elucidate the idea of agent causation by saying more about what it involves. According to one such view, agent causation occurs when the agent assigns weights to the reasons for and against different possible actions in a way that is not causally determined by previous factors. Again, this has some of the right intuitive flavor. Susan, we may suppose, has reasons both for and against joining the group, reasons that appeal to such things as her basic political values, her fear for her own safety, her concern for her family and friends, and so on. If the relative strength or weight of these reasons is not antecedently determined, then it is easy to see how assigning various weights to them could lead to one choice or another.

But the problem—one that you should and probably have already seen—is to understand how the choice of different weights isn’t itself merely random (perhaps random within determined limits). To say simply that Susan herself causes the assignment of weights that she actually adopts seems no more helpful than saying simply that she causes the resulting choice. Again the problem is to understand how her assignments of weights, given that it is not causally determined by anything about her, including anything about her antecedent thoughts and other psychological states and processes, differs from a merely random event. Â

Thus it is doubtful at best that the agent causation version of libertarianism really offers an intelligible third alternative to causal determination and randomness (one that is not merely a combination of the two). Is there any further possibility open to the libertarian? Perhaps, contrary to what we have been assuming so far, it is a mistake to think that the libertarian really needs such a third alternative. Might if be that even though pure causal determination or pure randomness are incompatible with freedom, freedom can after all be understood as some carefully constructed combination of the two? Let’s think about this possibility in more detail, again using Susan as our leading example.

Think again of Susan’s situation in the version where her captors explain their political views and purposes to her and then offer to release her. Susan is attracted by the views and purposes of the group. But she also feels some fear for her own safety, at least some aversion for the violent means that they are proposing to use, some concern for her family and friends, and some attraction to her previous life and purposes. She carefully deliberates about all this, constructing detailed alternatives as to what she might do and assembling and weighing her various reasons for them. Suppose that eventually she arrives at two alternative choices that seem more strongly supported and appealing than any of the others: one involves joining the group, engaging in their activities, and renouncing much of her former life; the other involves leaving the group, returning to her previous life, but actively attempting to advance the causes in which she has come to believe in legal and non-violent ways. All this, we are now supposing, is entirely causally determined by the various factors operating on her, including her own psychological states and processes.

But suppose now that the choice between these alternatives is not causally determined—that the reasons for each are strong enough to put her in a kind of balanced state in which there is no deterministic outcome. But Susan is not frozen, unable to choose. Instead, at some point, after much back-and-forth deliberation, a choice simply occurs. That she makes a choice is not random, but which choice she makes is. Her mental scales weighing the two sets of competing reasons simply tip, for no deterministic reason, in one direction rather than the other. Afterwards, we may suppose, she is able to identify to a substantial degree with her choice, for she has weighty reasons that for it. But at the same time, she also significantly regrets having not made the other choice, for which there were also weighty reasons. And she simply goes on from there, in a way that is causally determined much of the time but also involved further choices of this same basic sort.

Again, we have an account of free choice that possesses a significant degree of intuitive plausibility, and this time there is no real mystery about it, once the possibility of genuinely random events is accepted. But is Susan’s choice, as described by this account, really free in an intuitively satisfying sense, one that would make it fair and reasonable to hold her—and Susanna—morally responsible for the results? As we contemplate prosecuting Susanna for murder, we have to admit that there was an alternative possibility in which she led an entirely blameless life, and that which of these occurred was not in her control in any meaningful sense. Since the eventual choice was a random event, since the opposite choice could just as well have occurred given everything about Susan up to that point, how can she be responsible for the outcome that actually occurred? (Nor, of course—as the libertarian would agree—can she be held responsible for those being the two choices that most appealed to her, for all that was causally determined.) Â

Thus libertarianism, despite its undeniable intuitive appeal, seems either to result in a view whose very intelligibility is seriously in doubt or else to after all end up equating freedom with randomness in a way that seems indefensible. Is there a better libertarian view still to be found? Does the sheer logic of the issue leave any room for one?

Conclusions on free will

It is often said that the free will problem is the most difficult philosophical issue of all, and you are now in a position to appreciate why. We seem to be forced to choose between (a) a view that is virtually impossible to believe or to apply in practice, (b) a view that seems to merely evade the main issues and offer a view of freedom that gives us much less than we intuitively want, and (c) a view that seems intuitively more satisfying but that turns out either to be unintelligible or to collapse into the unsatisfactory equation of freedom with randomness. And yet the structure of the dialectic makes it seem as though one of these thoroughly unsatisfactory views must be correct. Is there some deeper insight into the nature of freedom itself that we are missing, perhaps one that would either avoid both determinism and randomness or make the equation of freedom with a kind of determinism more palatable than it presently seems? That is the main thing to focus on as you attempt to think further about this issue.