Sunday, July 31, 2011

Science Phiction #2: Sean Carroll on Free Will

This month, physicist and popular science writer Sean Carroll weighs in on the everlasting debate about free will. He says it's "as real as baseball," which means that it is not the sort of thing that we would expect to find in a detailed physical description of the universe, but that we can't imagine trying to talk about humanity without accepting it as a real phenomenon.

I have to criticize Carroll for failing to explain what he means by the phrase "free will" (a phrase which, he explains, does not have an agreed upon meaning) and for failing to give us a reason to think it is as real as baseball. Carroll defends a pragmatic realism--a view that we should take as real whatever entities we benefit from postulating in a given language, regardless of whether or not we benefit from postulating them in our widest available language. So, we benefit from postulating the existence of baseball even though there is no need for it in the language of physics, and even though the language of physics has more predictive power than the language we use to talk about baseball. Similarly, he says, we can believe in free will even though it has no place (or need not have a place) in our most general and powerful languages. That's a fine point to make, if we had some reason to think that "free will" is emergent in the way that baseball is. Since Carroll hasn't told us what he means by "free will," how can we decide if it is as emergent (and thus as real) as baseball?


Sometimes philosophical essays offer food for thought without clearly defending their main theses. Carroll may not have helped us understand what free will is, or given us a reason to think it is as real as baseball, but he has made a number of interesting points on related topics. Unfortunately, he has also gotten some of the philosophy wrong. First, he claims that people who suppose that the laws of nature cannot account for free will are metaphysical libertarians. I don't think that's how "metaphysical libertarian" is generally applied. Second, talking about free will as an emergent property of physical systems, he mentions that that is what Dennett calls "a variety of free will worth having." That seems to misrepresent Dennett, though Dennett does interpret free will as an emergent property. When Dennett talks about varieties of free will worth having, however, he has something else in mind: namely, varieties in which our choices are not disconnected from our causal histories. According to Dennett (and others before him), the only variety of free will worth having is a variety in which we act in accordance with our needs and desires, and this is a variety which is perfectly compatible with determinism. Anything else would be a completely arbitrary freedom, one which would be as likely to kill us as it would get us what we want. Perhaps Carroll agrees with Dennett on this point, but he does not say so.

The next point I want to talk about is more about science, and not philosophy. Carroll talks about the possibility of having different levels of description, the microscopic (physics) and the macroscopic (emergent). Giving the example of time's arrow, he points out that the laws of physics are the same either forward or backward in time. The microscopic description of reality therefore does not distinguish between past and future--or, rather, the concepts of "past" and "future" are arbitrary when we are talking about physical laws. Yet, he points out, we clearly distinguish between the past and the future in a non-arbitrary way when we are talking about everyday life. The macroscopic world is irreversible. Carroll says the laws of physics don't account for this; so, to avoid contradiction when combining these two levels of description, we must add a new component to our discourse: the particular configuration of the universe. His claim is that, if we ignore the particular configuration of our universe, we end up with a time-reversible description of reality.

Maybe I should defer to the physicist here, but I have to pause and wonder: does this make sense? How does adding a description of the configuration of the universe make a difference? The second law of thermodynamics--the law which states that any isolated system will increase in overall entropy--is a fundamental law of physics and it is widely recognized as accounting for our common notion of time's irreversibility. So Carroll's discussion seems terribly wrong. What Carroll is saying is that the laws of physics are not consistent with the second law of thermodynamics, and that we must appeal to empirical facts about states of the universe in order to compensate for this discrepancy. It seems much more accurate to say that the laws of physics include the second law of thermodynamics, that the second law of thermodynamics cannot be deduced from the other laws of physics, and that it--like all the other fundamental laws--is supported by experimental evidence. So there is no conflict between the microscopic and macroscopic levels of description here.

True, quantum mechanics has suggested that the notion of temporal direction might not make sense on extremely small scales, but that is not what Carroll is talking about when he distinguishes the microscopic and the macroscopic. He's talking about two levels of describing the same reality: the level of emergent properties and the level of underlying physical laws. I think we can make such a distinction, but Carroll's discussion of the arrow of time seems to confuse the topic. Again, I'm no physicist, but I think he has said something quite wrong here, or at least misleading.

Now let's get back to philosophy. Carroll's curious discussion of levels of description leads him to a discussion of a well-known philosophical argument, called The Consequence Argument. The argument is as follows: If we do not have power over X, and X completely determines Y, then we do not have power over Y. Since we do not have power over the past or over the laws of nature, and (according to determinism) the past and the laws of nature together completely determine the future, then we do not have power over the future.

Carroll misrepresents the argument in a rather absurd way. He writes: "The consequence argument points out that the future . . . [is] determined by the present state just as surely as the past is." Does he really mean to say that the past and future are equally determined by the present? I doubt it, but it's not clear what he does mean to say.

In any case, Carroll says the consequence argument "mixes levels of description." The problem with the consequence argument is apparently just like the "problem" we have when we try to understand time's arrow by looking only at the laws of physics. Carroll continues: "If we know the exact quantum state of all of our atoms and forces, in principle Laplace’s Demon can predict our future. But we don’t know that, and we never will, and therefore who cares? What we are trying to do is to construct an effective understanding of human beings, not of electrons and nuclei."

It looks like Carroll has misunderstood the consequence argument. It does not depend on anybody being able to predict the future by looking at the present. It has nothing to do with electrons and nuclei, per se. There's no reason to think any levels of description have been mixed.

Carroll's argument aside, a "Who cares?" response to the consequence argument may be worth considering. Should we care if we are powerless to affect the future? Honestly, I don't see how we could not care. But there might be more here to consider. A fruitful discussion of this issue might focus on the notions of power and powerlessness. Perhaps we do have some power over the future, but not the sort that is implicated by the consequence argument. Maybe the consequence argument rejects a sort of power which is not worth having.

Update: I've just considered a different criticism of the consequence argument, and it can be found here. The upshot: Since what is determined by the past is part of what determines the future, and we are determined by the past, then we are part of what determines the future. So the consequence argument is not sound.