UPDATE: Scott Aaronson has emailed me and pointed out that his views on this matter are set out in a clearer and longer way in a publicly available paper he wrote called “The ghost in the quantum Turing machine.” It’s 85 pages long, I wasn’t aware of its existence, and it is probably above my pay grade, but perhaps not for some readers who are physics-savvy and also willing to read the paper. If you do, weigh in below. Please consider this post a response to Aaronson’s 8.5-minute explication of free will on the “Closer to Truth” interview and not to the totality of his published views.
I’ll add, with Scott’s permission, a clarification that he emailed me along with the link to his paper:
“Briefly, you can make any theory “deterministic” by the addition of hidden variables, which is exactly what de Broglie and Bohm did for quantum mechanics. In that sense, to ask whether the world is deterministic is not even to ask an interesting question about it; it’s only to ask about a particular description of the world. But whether you can predict someone’s behavior without destroying them IS a question about the world. So focusing on that means you’re led to actual meaty empirical questions about the world, rather than endless and boring semantic debates about the exact meanings of terms like “free will” and “compatibilism.” I’d hope anyone with a scientific mindset would find that to be a feature.”
I may be remiss in not knowing who Scott Aaronson is, as I gather he’s quite well known. He’s the David J. Bruton Centennial Professor of Computer Science at The University of Texas at Austin, and works on quantum computing. He also has a popular website called Shtetl-Optimized (“Shtetl” refers to small Jewish towns in eastern Europe), which deals largely with computation but also personal revelations and amusing tales (see here about his quirks and then here for a story in which airport police arrested him because he absent-mindedly took change from his purchase of a smoothie—on a debit card!—from the tip jar in an airport food court.
In this 8½-minute video, Aaronson is quizzed about free will by Robert Lawrence Kuhn for the “Closer to Truth” public television series. (I’m weighing an invitation to participate in this series.) Do watch the video if you have time.
Aaronson thinks there’s a real and important question in the free-will debates, but argues that that question is not whether physical determinism of our thoughts and actions be true, but whether they are predictable. He thinks that the question of whether there is free will could in principle be solved by answering the following question: “
Would one be able to construct a machine that, if it was programmed with the immense knowledge about your brain and your environment that would be required to do such a job, would be able to predict your actions?”
If the answer is “yes,” he says, then the question is solved: we have no free will and are “automatons”.
(By the way, Aaronson doesn’t appear to be a compatibilist, as he says that if such a machine were possible, and all our choices were predictable, we would certainly not have free will, even though some people would say we would still have free will. Those “some people” are surely compatibilists, who find the notion of free will compatible with physical determinism.)
Because, Aaronson says, the answer to the Big Machine question is not known (I would argue that in principle, because behavior and thoughts are the results of physical processes, the answer is “yes”), he finds the question of free will to be moot.
As he says at 4:15,
“My view is that I don’t care about determinism if it can’t be cashed out into actual predictability.”
This seems to me misguided, conflating predictability with the question of determinism. Surely it will be impossible, at least in our time, to gather the requisite information to accurately predict someone’s behavior, for such a computer would have to model not just a person’s brain, but also the entire universe, for the universe impinges on a person’s brain in ways that affect their behavior.
Further, insofar as fundamentally unpredictable quantum events may determine behavior, no machine could ever model those: at best it could give probabilities of different behaviors. But those quantum effects do not violate physical determinism, and cannot give us free will in the sense that most people think of it. (Surveys show that most people think of free will as “contracausal”: the you-could-have-done-otherwise form of free will.) In fact, insofar as Aaronson espouses physical causation of behavior in his answer—he says he sees the brain as a kind of computer running a program—he’s already admitted physical determinism.
Do Aaronson’s lucubrations, then, make the question of free will uninteresting?
Not to me; I find the issue not only philosophically interesting but a question that has practical ramifications, even if we can’t build the Aaronson Machine. If we are truly biological automatons, which I think is true on first principles (viz., we are made of molecules), then that has huge implications for religious thought and dogma, which of course depend on assuming contracausal free will. You are free to choose your saviour, your faith, your actions, and, for gay Catholics, whether to commit homosexual acts. Because you make free choices, making the wrong choice will send you to perdition, and making the right one to God, Yahweh, or Allah.
There are ramifications for the justice system. I firmly believe that if we grasped that nobody, including criminals, has a “choice” in whether or not to do something, like mugging someone, we would structure the justice system differently, concentrating less on retribution and more on keeping baddies out of society, trying to reform them, and using punishment as a deterrent to improve society.
There are ramifications for politics. Once you realize that people’s acts solely reflect the physical consequences of their genetic endowment and environment, you (or at least I!) become more sympathetic to the plight of those who drew a bad hand in the poker game of life. The notion of the “Just World”, in which people get what they deserve, depends on accepting contracausal free will. But that view must be tempered by realizing that neither the successful nor the downtrodden freely chose their paths.
I’ve always said that I don’t really care if you say people have “free will” if you define that term in some compatibilist way like “the inputs to creating a human behavior (output) are complex”.
What I care about is whether determinism be true. And I think it is, though of course I can’t prove it. All I can say is that the laws of physics don’t ever seem to be violated, and, as Sean Carroll emphasizes, the physics of everyday life is completely known.
What we see above are the ruminations of a man whose life is devoted to computing, and his profession shows in the way he thinks about free will, turning it into the question of whether machines could predict behavior.