Friday, September 07, 2007

Finding Free Will in the Brain

Every now and then incautious neuroscientists (or more likely science journalists) claim to have found the site in the brain responsible for some universally shared bit of behavior or psychology. It's much like the tendency to tout discoveries of a "gene" for obesity, addiction or extroversion: often it turns out on closer inspection that there's no single gene, or site in the brain, responsible for the trait in question. Rather it's a more complex story of how genes and environment and multiple brain systems interact. But the fascination with such findings, oversimplified or not, reflects the attractiveness of physicalism as an explanatory story. Once we've identified the physical embodiment or correlate of something - the mechanism constituting it or causing it - it becomes categorically real. This is probably because mechanistic explanations are the most transparent we've got. Everything's out in the open, observable and pinned down, even if it's fiendishly complex, such as the neural reward system that regulates our cravings for sex, drugs and Monteverdi. Everyone now agrees the placebo effect is real since it's been shown that believing a sugar pill is a drug actually changes the brain!

So when neuroscientists announce they've found the place in the brain where free will resides, it might generate some puzzlement. Isn't free will supposed to be that which escapes mechanism, such that we choose freely, independently of deterministic causation? If it's the brain that gives us free will, then it must boil down to the physical goings-on of neurons and neurotransmitters, in which case how does that make us truly free and responsible? Finding free will in the brain doesn't make it real, it destroys it.

This question points up the fact that there are different meanings of free will floating around. Looking at the various articles about this research (here and here for instance), it's clear that what these scientists mean by free will is a specific capacity, namely to consciously reconsider an intended action. Why call that free will? Well, the ability to control impulses by means of higher level cognitive processes arguably gives us an important kind of freedom, namely freedom from being simple slaves to our appetites. We gain tremendously in flexibility by virtue of all the brain-instantiated firmware that decides what actions are appropriate given one's situation, long-term goals, and present urges.

This is clearly quite different from the folk-metaphysical concept of free will: a conscious capacity that transcends the mechanistic workings of the brain. The folk concur that free will is about self-control, but real self-control comes from a self beyond mechanism; if not, it's obvious that people are just the working out of mindless physical parts, and what's free about that? Doesn't that make us mere robots? If not having an immaterial soul makes us robots, then yes. Some, such as Keith Stanovich, have argued that being a sufficiently complex organic robot is all we need to be. But of course many balk at this prospect.

Many scientists (and most philosophers) know full well that the physicalist/functionalist downsizing of free will doesn't cut it for laypeople, so they tend to soft-pedal the implications for our self-image. But a few, such as neurophilosophers Patricia Churchland and Joshua Greene, are pretty forthright in laying out the implications. If free will isn't a magical quality that people can just choose to exercise, but rather a physically-based capacity for self-control, then we can't blame someone who lacks this capacity, such as an addict, for not having exercised their free will. This is one aspect of the moral significance of moving from dualism to physicalism: we'll be more likely to cure or rehabilitate, rather than punish, those who, for one reason or another, lack normal capacities for self-control. Another aspect is that we'll see that those who do have normal capacities but misuse them in anti-social, damaging ways are products of specific conditions that shaped their values; they are not self-made. This suggests that originative responsibility for behavior doesn't inhere in the person alone, but also in the culture, community and family that produced him. So our responsibility practices should take this into account, which again means not merely punishing, but preventing and rehabilitating.

Some will say we'll never find free will in the brain since free will by definition is that which transcends mechanism. Others will say that free will, mechanistic or not, is just that which makes us morally responsible. Or (my preference), we can clarify the debate by dropping talk about free will, which will always be ambiguous since it has multiple meanings, and argue about what sorts of freedom and responsibility we plausibly have on a naturalistic understanding of ourselves. Getting a fix on these will then inform our responsibility practices, making them more effective, and more humane.

Science and the Supernatural

Creationists often argue that science, by refusing to consider supernatural explanations such as intelligent design (ID), presumes naturalism and thus illegitimately promotes a sectarian worldview. An unbiased scientific approach to explaining phenomena would consider all hypotheses, not exclude some simply because they invoke a creator. In an excellent article for Science and Education (available here) Yonatan Fishman of the Department of Neurology at the Albert Einstein College of Medicine gives the lie to this charge, showing that science can and does consider supernatural hypotheses.

Fishman first points out that some very reputable science organizations, such as the National Academy of Sciences and the American Association for the Advancement of Science, seek to defuse the science-religion conflict by saying that science establishes nothing about god since it only considers explanations involving natural processes. But this plays right into the creationist argument. Likewise, Judge John E. Jones gave creationists ammunition in his recent Kitzmiller vs. Dover decision against teaching ID by saying "rigorous attachment to ‘natural’ explanations is an essential attribute to science by definition and by convention." Again, this seems to confirm the suspicion that by ruling out supernatural explanations in advance, science enshrines naturalism as its worldview.

But Fishman mounts a detailed, persuasive argument that science can indeed evaluate supernatural hypotheses, assuming they have a shred of content. Using a Bayesian framework as a model for scientific inference, he shows that the probability of supernatural claims can be estimated by comparing them to what we already know about the world, by looking for evidence for and against the claim, and by seeing if there are plausible naturalistic alternatives. The probability of the existence of god can be evaluated according to these methods, and Fishman proceeds to do just that with many examples. For instance, if a benevolent, concerned god were at work in the world, we'd expect intercessory prayer to have positive effects on the health of those prayed for. But experimental studies of prayer, including the largest and most carefully controlled study recently conducted by Herbert Benson at Harvard, have turned up no such effects. So, as Fishman puts it

The essential point is that methodologically sound studies published in reputable scientific journals have been conducted to directly test the consequences of a supernatural hypothesis....In general, as reflected by the likelihoods in Bayes’ theorem, whenever a supernatural claim predicts with a specified degree of probability some state of the world, that claim can be tested simply by inspecting the world to see whether or not the world displays that state.
Science's verdict on god as Fishman presents it is unsurprising: the probability that a benevolent, concerned, intelligent designer exists, given background knowledge, current evidence, and alternative hypotheses, is very, very low. We can't of course disprove god's existence, but as Fishman points out, that's not a good reason to believe in god; instead, the very low probability of god's existence is good reason to believe in naturalism.

Of course it's only reason to believe in naturalism on the assumption that it's best to go with the probabilities as established by scientific investigation when deciding what's real. If you're not an empiricist, then never mind about science, just believe on faith, tradition, authority, revelation or intuition. But if you want to play the science game, be prepared to have your supernatural hypotheses tested, and very likely rejected. Science rules out intelligent design and other supernatural hypotheses not because it assumes naturalism, but because these hypotheses have failed all the scientific tests put to them thus far.

If Fishman is right, and I think he is (I've sketched similar arguments here, here and here), it makes things more difficult for those wanting to reconcile science and religion. But there is a fallback position for science organizations not wanting to offend religious sensibilities. This is simply to say that nobody's forced to take science as their way of deciding what's real. It's a free country, after all, so having faith in god is permitted. Indeed, a recent Pew Research poll shows that's exactly what people continue to do when apprised of scientific findings that contradict their beliefs. But what science organizations can't do, of course, is to say faith is equal to science as an epistemology, since that would betray their very mission. If that counts as a bias, it's one that creationists and IDers will have to live with. In an open society, not only are we free to believe on the basis of faith, we're free to disbelieve on the basis of evidence.