*My* Compatibilism

A primer on free will and compatibilism that includes a fun (short) time travel argument

How to be an incompatibilist

Incompatibilism is the view that the truth of determinism is incompatible with free will, that is to say, it is the claim that if the natural laws are deterministic then no person has free will.¹

Two questions: what is determinism? Determinism is a thesis about the laws of nature. Determinism is true just in case the laws of nature are deterministic. The laws are deterministic if the description of the universe at a point in time, conjoined with a description of those deterministic laws, jointly entail the state of the universe at all other points in time. We might put the point epistemically. Supposing the laws are deterministic, a sufficiently sophisticated inference machine, that knew the state of the universe and the laws, could figure out what has happened, is happening, and will happen, with perfect accuracy.

What is free will? Free will is a sort of control that we have over our actions. I am, in one sense, in control of my involuntary twitches. I am the one that performs them. But I’m not morally responsible for my involuntary twitches. I wouldn’t be blamed, for instance, if I had a spasm in my sleep and knocked over my partners warm glass of milk. Contrast this with premeditated murder. My neighbor plays the drums at 6 am every day, so I plot and scheme to bring about his demise. I stay up late tracking his schedule, learn where he keeps his spare key, and deliberate about when he will be most vulnerable. Whether or not I succeed, I am morally responsible for any harm that might come to my neighbor as a result of my plans; and I am responsible because I enact my plans of my own free will. With premeditated murder, I’m responsible partly due to a certain sort of control over my actions, and this is exactly the sort of control we are interested in when we talk about free will.

Returning to our incompatibilist. The incompatibilist says that if the laws of nature are deterministic, we lack free will. Why should we be incompatibilists? Here’s a reason: To have free will, we have to be able to do things we don’t actually do. If I chose to kill someone, but that choice was unavoidable, then I didn’t perform the act freely. Yet if determinism is true, anything I won’t do, I cannot do. Won’t implies can’t. Why? If the laws are deterministic, then the world cannot differ at some point in time without differing at every point in time. In other words, if someone were to do something they didn’t actually do, the universe would not be different just then, but throughout all of time. So, if someone were to do something other than what they actually did, the past would have to be different. They would have to change the past. They could, of course, also change the laws of nature, but that isn’t much better. It’s in no one’s power to change the past, or the laws of nature. If determinism were true, we could only act differently if we could change the past or the laws of nature, but since we cannot, the truth of determinism entails that we lack the ability to do anything other than we actually do. In other words, if we don’t, or won’t, do something, then we can’t. Won’t implies Can’t. This argument (or something close enough to it) is often called the Consequence Argument, commonly attributed to Peter Van Inwagen, and it is the strongest reason to not be a compatibilist.²

And how not to be

Time for a detour. Fatalists are rare. What do fatalists believe? They think that the fact that I won’t do something implies that I cannot do it. Yesterday, it was true that I would eat cereal today. But I cannot change the past! Thus, I cannot, could not, avoid eating cereal today. The fatalist differs from the determinist in that they think this is a necessary truth. Whatever the laws of nature, the fact I won’t do something implies I can’t. Won’t implies Can’t. So I am not free, even if determinism is false.

What is wrong with the fatalists reasoning? When we evaluate counterfactuals, we cannot hold all facts fixed. If we do, then every counterfactual will turn out false. If we ask whether I would have cereal if I chose not too, and evaluate this counterfactual by looking only at scenarios where I do have cereal, then of course it will be true that I would not succeed if I tried. But we should not hold my failure fixed. We are wondering whether or not I would fail. So there is no reason to say I could not have acted such that, yesterday it was true that I would not eat cereal today. I could act in such a way, namely, by not eating cereal today. The fatalist has given us no reason for thinking that I could not do this if I tried, except by stipulating that we hold my failure fixed, which they have given us no reason to do.

Time to tip my hand (although the title of this post may have already done this). I am no incompatibilist. I am a compatibilist of the commonsensical sort. I think we have free will, and I think determinism is no threat to this fact. As such, I think the Consequence Argument fails. Why? What was the lesson of our detour? The Consequence Argument has the same flaw as the argument of our mock fatalist. It is of course true that, given determinism, the laws or the past would differ if we acted differently. What the Consequence Argument fails to provide is a reason for holding the laws and the past counterfactually fixed. It is natural, in fact, to allow the past to range. Suppose I had chosen to have pancakes instead of cereal this morning. I would have made this decision for a reason. I might have chosen to have pancakes because they were closer on the shelf, or because I woke up with a sweet tooth. It is natural to think that a difference in our choices is precipitated by a difference in the surrounding circumstances. If that’s right, then we shouldn’t hold the past counterfactually fixed. If we shouldn’t hold the past counterfactually fixed, then free will is not threatened by the Consequence Argument. It is true that determinism implies that, if I were to do something I didn’t actually do, the past or the laws would have been different. But that’s fine. If I had made a different choice, it would have been because different reasons were presented to me. It is still true that I would have eaten pancakes if I so chose. It just also happens to be true that, if I had eaten pancakes for breakfast this morning (or chosen to do so) the past (or the laws) would have been different.

The trick with determinism is that it leaves the distant past counterfactually dependent on our present actions. Most people agree that the recent past would have been different if the present was (although it may be difficult to specify how it would have differed). But we do not naturally think of the distant past as depending on the present in the same way. It is odd to think that the location, spin, charge, of a particle at the moment of the big bang might be intimately relevant to what I do now. We can easily think of the sorts of things in the recent past that would need to change to enable differences in the present — see the pancake example — but it is often entirely opaque how the distant past would need to differ for the present to be different. But just because our laws permit counterfactual dependence of the present on the distant past does not mean that our freedom is broadly undermined.

Consider worlds with laws of nature amenable to time travel, but that are not deterministic. These are worlds with backwards counterfactual dependence. If John shoots John F. Kennedy on the grassy knoll after hopping in his time machine in the year 2150, then the past counterfactually depends on the future. Whether the 35th president of the United States was assassinated depends on John using his time machine in the mid 22nd century. But should this counterfactual dependence alone give us any reason to think John could not, if he so chose, have refrained from killing JFK? Could John have gone back in time, sat on the grassy knoll, and reconsidered? It seems that he could have. The world would have been different, indeed the period between 1963 and 2150 would differ greatly, but so what? If John had acted differently, the past would have been very different, and this is true even though his world is not deterministic; yet we should not think John suffers from any sort of ineptitude. Notice further that we might extend the duration of John’s temporal jaunt. He may have traveled from the 24th century, or the 30th, or whichever. So even if John’s temporal escapade results in counterfactual dependence of the past on the very distant future, our judgement regarding his abilities should not change. We can see that John does not lose his free will as a result of stepping into his time machine, or as the result of occupying a world with time-travel-friendly laws, and this does not depend on the duration of John’s travel..

The moral is that deterministic worlds are worlds where the past counterfactually depends on the present. But this should only worry us, with respect to our freedom, if we are unable to do otherwise when there is backwards counterfactual dependence. We should not think this dependence is so problematic, so we should not conclude that worlds with deterministic laws are worlds with occupants who lack free will. Time travel worlds provide the case in point. The fact that someone occupies a time travel world gives us no more reason to think they are not free than finding out they occupy a non-time travel world. This is true even if the time travel world in question contains counterfactual dependence of past on the distant future. Much the same with determinism. Deterministic worlds are worlds with counterfactual dependence of distant past on the present and future. This may make us queasy but it should not make us think that we lack the ability to do things we don’t or won’t do, and it should not make us think we lack free will.

An important aside

I’ve left a lacuna unaddressed thus far, and unfortunately I will not be able to address it adequately here. Time is asymmetric. The past is different from the future. The future is ‘up to us’ in a way the past is not. The past is given, fixed, unchangeable. The future is open, undecided, under our power. These are mere suggestions at the idea of this asymmetry, but the intuitive idea is familiar. There is a way in which the future is within our control now which is not true of the past.

One widely accepted explanation of this asymmetry is credited to David Lewis. In “Counterfactual Dependence and Time’s Arrow” Lewis’ suggestion is that the past is counterfactually fixed, in ordinary contexts, and that the future is not. The idea is that the past would be the same, no matter what we do, while the future might differ. If I had cereal for breakfast, JFK would still have died in 1963, but I might have died earlier than I actually did because I choked on the cereal. The past is different from the future because we tend to think that the past would be almost exactly the same, even if we acted differently, but the future would not. I cannot appeal to Lewis’ answer. I have argued we should not hold the past counterfactually fixed and, therefore, lose access to Lewis’ explanation. “So explain the difference!” cries my objector.

Two things to say in response. Firstly, it is worth noting an obvious deficiency in Lewis’ answer: we all agree that the past counterfactually depends on the present, at least the near past. Lewis himself accepted this given his theory of counterfactuals.³ But the apparent asymmetry between past and future does not discriminate between the near and distant past. The near past is just as ‘fixed’ as the distance past, in the sense relevant to the asymmetry of time. So Lewis’ answer fails to solve the puzzle posed.

The second thing to say, my suggestion which I cannot sufficiently develop, is that although deterministic worlds are worlds where the past counterfactually depends on the present, they are not worlds where facts about the present cause facts about the past. This is dissimilar from time travel worlds. John’s time travel causally contributes to the death of JFK. We might modify Johns’s scenario so that the cause is even more straightforward (suppose John made his time machine appear inside of JFK’s car, exploding it. This is getting rather grim.). So time travel worlds violate the asymmetry of time in a way determinism does not. I find this rather intuitive. John’s actions seem to violate my ordinary notion of the asymmetry between past and future in a way that the actions of agents in deterministic worlds does not. Any good theory of the asymmetry of time should say that it is a contingent fact about our world that time is asymmetric and (I think) that time travel worlds violate this fact in a way that deterministic worlds do not. Thus, on my view, the asymmetry between past and future should be explained in terms of causal dependence, and we should. This may turn out to be a special case of counterfactual dependence (supposing a counterfactual analysis of causation is correct) but I cannot expand on this point without a theory of causation, and I’m not ambitious enough to structure that into this post.

Conclusion

To sum up: fatalists and incompatibilists are alike in two ways. Both license the inference from ‘Will not’ to ‘Cannot’, and both are mistaken in licensing this inference due to a failure to appreciate what is (and isn’t) held fixed in counterfactuals. It is important to see how far the needle has moved. I deliberate, I reason, I choose to eat pasta for dinner. Supposing the incompatibillists objection succeeded, if determinism were true, I was unable to deliberate, reason, or choose in any way other than I did. I couldn’t have deliberated, reasoned, and chosen to have a burger instead. But the incompatibilist is wrong. Even if determinism is true, I can deliberate in a whole host of ways, reason in just as many, and choose between as many options as there Tupperware in my fridge. The incompatibilist must ask us to believe that even though determinism poses no threat to such abilities, our free will remains threatened. In such a dialectical millieu I find accusations of compatibilism’s implausibility rather overblown. Thus why I am a compatibilist. I think having free will is a matter of having a variety of abilities. The ability to respond to reasons, to deliberate, to decide, to choose. If determinism threatened those abilities, I would be concerned for my status as a free agent. But I determinism does not threaten those abilities, and so I am not concerned. Thus, my compatibilism.⁴

1. The title of this post is, of course, a reference to John Martin Fisher’s “My Compatibilism.”

2. See Van Inwagen’s An Essay on Free Will or his earlier “On the Incompatibility of Free Will and Determinism.” Although I say I find this argument the most convincing, and I don’t address them here, I do find manipulation arguments highly compelling on odd days of the week.

3. Again, see “Counterfactual Dependence and Time’s Arrow,” particularly where Lewis discusses a transition period in the worlds used for evaluating counterfactuals.

4. Much of what I say here is heavily influenced by Kadri Vihvelin’s work. See Causes, Laws, and Free Will: Why Determinism Doesn’t Matter, and “Determinism, Counterfactuals, and the Possibility of Time Travel.” I also credit Cian Dorr’s “Against Counterfactual Miracles” for my sympathy to fixed law theories of counterfactuals.