Reading Journal: Sapolsky on Free Will

Plus: Is Free Will Actually a Quale?

Discussed in this post:

Determined: A Science of Life Without Free Will by Robert Sapolsky

The Case Against Reality: Why Evolution Hid the Truth from Our Eyes by Donald Hoffman

Breaking the Spell: Religion as a Natural Phenomenon by Daniel Dennett

Robert Sapolsky’s New Book Has Some Problems

I've been reading Robert Sapolsky's new book about free will, and I'm finding it intriguing but immensely frustrating. Not because I disagree with Sapolsky's main thesis, but because I don't think it supports the conclusions he wants to draw from it.

Sapolsky's a neuroscientist. He doesn't believe free will exists, and in arguing against it, he mostly takes the standard determinist line. If our actions are caused by our brain chemistry, and our brain chemistry is caused by a host of environmental and biological antecedents, where can will be said to enter the picture? At best, will is simply superfluous, an unnecessary explanation for events that can be accounted for in a strictly materialist fashion. At worst, the concept of free will violates the principles of scientific materialism by being, in effect, a "causeless cause"--something that explains events in the world while having, itself, no further explanation.

So far, so good. The problem is that Sapolsky goes on to build a moral case for the abolition of our belief in free will. If free will doesn't really exist, he argues--if no one, properly speaking, decides whether or not to do anything at all--then there's no justification for blaming people for bad decisions. We ought simply to withhold judgment and treat human actions as natural occurrences, like weather events or earthquakes: things to be prevented, even lamented, but not subjected to moral censure.

Of course, in Sapolsky's view, this doesn't mean we should never take steps to discourage or prevent unwanted behavior. It just means our efforts ought to be strictly rational, wholly focused on producing desired outcomes by intervening at key points in the causal chain. If we don’t want people to steal stereos, we should take measures to learn about the underlying causes of stereo theft--poverty, opportunity, poor incentives, certain neurological conditions--and change those causes to make stereo theft less likely, just as we might take prudent measures to control insect populations or prevent storm damage. There's no point judging criminals for their behavior, in Sapolsky's view, and what's more, it's deeply immoral to do so. After all, criminals have no free will. They literally can't help themselves.

There's an obvious problem here that I assume Sapolsky will address at some point.¹ If stealing stereos is something people can't help doing, given certain physical and environmental conditions, isn’t judging stereo thieves for their behavior also something people can't help doing, given other underlying conditions? Believing that people have free will, and consequently holding people accountable for their actions, is the most natural human behavior in the world. If we're going to refrain from judging people for bad behaviors, shouldn't we also refrain from judging them for acting as if free will is real? It's important to emphasize that Sapolsky isn't only mounting a philosophical or empirical case against free will; he specifically argues at many points in his book that it's wrong to judge people for actions they can't control. But if we take his thesis to heart, this particular moral judgment is as worthless as any other. We end up in an infinite regression, in which people can't be blamed for committing crimes, but also can't be blamed for blaming people for committing crimes, but also can't be blamed for blaming people for blaming others … and so on.

Now, there are ways to salvage aspects of Sapolsky's argument. We could take him at his word, for instance, suspend all moral judgment, and simply focus on changing human behavior by dickering around with the systems that generate it. If we conclude, as Sapolsky evidently has, that the human tendency to believe in free will produces bad outcomes, and we want to find some way to change those outcomes, we might look for interventions we can undertake to make belief in free will less popular. And we might conclude, as Sapolsky clearly has, that writing books arguing against free will is one such effective intervention. We might further note that humans are highly susceptible to being swayed by moral arguments, and might therefore decide that casting belief in free will as a moral offense is a shrewd way to disincentivize it. In other words, we might harness belief in free will against itself, by arguing--in a purely rational fashion, of course--that people should exercise their free will by choosing to stop believing in free will.

But now we're right back where we started, arguing that people should be blamed for some behaviors and not others. This roundabout reasoning highlights the whole problem with Sapolsky's argument. If we want to get people to stop believing in free will--or to stop acting as if they believe in free will--we have to figure out why people hold that belief in the first place. Why do so many people, perhaps all people, instinctively believe in something that, as Sapolsky and others have persuasively argued, pretty clearly doesn't exist? Where did the belief in free will come from? Why did it evolve? How is it related to other human beliefs? Is believing in free will like believing in goblins or Atlantis--a fairly simple epistemological error that can be corrected by showing people enough compelling evidence? Or is it like believing that your girlfriend is hot--a deeply held conviction that persists in the face of every intellectual attack?

We've spent centuries debating free will itself—what it is, whether it exists. What's needed now, I think, is exploration of this secondary phenomenon, the belief in free will. And a clear-eyed approach to he question, in my view, brings up serious issues with Sapolsky's approach. We shouldn't be trying to eradicate belief in free will. We should embrace it, and nurture it, for what it is--while recognizing, in a scientific and clear-eyed fashion, what it isn't.

What if Free Will Isn’t What We Think It Is?

So what we can say about belief in free will?

I'm drawing here primarily on popular books I've read on the subject, all of which (including Sapolksy's) agree on a few key points:

1. From a strictly scientific perspective, free will not only doesn't exist, it *can't* exist. Either our actions have natural causes (in which case they're determined by physical processes, like anything else in the universe), or they have supernatural causes (in which case we've left the realm of scientific analysis), or they're random. Those are the only three options, and not one of them allows for a scientific theory of free will. Free will is an inherently unscientific idea.

2. Nevertheless, the feeling that free will exists is near universal, and even people who recognize, intellectually, that the concept of free will is nonsensical can't help feeling and behaving as if free will is real.

3. This compulsion to believe in free will must be compatible with Darwinian evolution and other established scientific theories; it must have evolved in humans like any other trait and be connected in some way to human behaviors that serve an adaptive purpose.

My own view, then, is that we should take all these points as given and think of free will as a kind of quale, an emergent property of the human sensory interface. We don't believe in free will, I would argue, so much as we perceive it, in much the same way we perceive sweetness, cuteness, grossness, stinkiness, or beauty. Once you make the shift to thinking about free will this way, it's clear that arguing about whether or not it exists is somewhat beside the point. Does beauty exist? It's in the eye of the beholder.

More narrowly, I'd hypothesize that our perception of free will has a clear adaptive purpose: it functions as a rough-and-ready unpredictability detector, a module in our minds that's programmed to identify, and pay close attention to, entities exhibiting highly complex behavior, at least within certain time bounds. The philosopher Daniel Dennett (who comes in for quite a drubbing in Sapolsky's book) theorizes that humans have a kind of built-in "agent detector" that attributes intentionality to entities in the external world (and when this "agent detector" goes haywire, we might conclude, it starts detecting agency all over the place, which explains religious beliefs and many kinds of mental illness). It’s been a while since I read Dennett and I wouldn’t dare to try and summarize the philosophical debates around his work, but my recollection is that someone neatly skirts the question whether agency properly exists. I myself would suggest that words like "agency" or "will" or "intentionality" are nothing more or less than our labels for the quality we perceive when we observe unpredictable behavior, much as "cuteness," "sexiness," and "beauty" serve as labels for complex aggregations of traits that underpin our perception of reproductive fitness.

So let's run with this idea, entertaining the thesis that "free will" is the quality we perceive when we observe unpredictable behavior. I've found that people have some common objections to this notion, which I'll address below:

1. Why would we be subject to this bizarre superstition? Why not just perceive complex behavior as complex behavior and leave it at that?

Here I'll draw on the work of cognitive scientist Donald Hoffman, who argues that all our perceptual faculties are essentially highly evolved mental shortcuts--that we never see the world as it truly is but instead experience a suite of illusions designed to increase our fitness at minimal cost. Instead of detecting blood sugar levels, we feel faint or hungry; instead of perceiving photon streams, we perceive light and color; instead of rationally assessing the variable risks and rewards of potential future events, we feel hope and trepidation. I'd wager it's the same with our perception of free will: our neural systems evolved this shortcut to make our allocation of attention more efficient. If you drop a stone and it falls in a straight line to the ground, you think to yourself, "Ho-hum, ordinary stone acting in accord with natural processes, nothing to see here" and go about your business. If you drop a stone and it pauses in mid-air, does a few loop-the-loops, zips across the yard to tap three times on the garden shed, then flies back into your hand and vibrates for three seconds before jumping free and landing on the ground, you'll probably think, "Whoa, something weird is going on here, an agent with will and intent appears to be acting upon this stone!--I'd better pay lots of attention to it and try to figure out what's going on." And I would further wager that most entities complex enough to excite our free-will-perceiving module are, in fact, entities that tend to be, over a long time horizon, predictably unpredictable—i.e. entities so dynamic they exhibit highly chaotic behavior and therefore demand constant vigilance.

2. But why don't we attribute free will to all complex systems, like weather and computer programs and ant colonies? Why do we only ascribe free will to humans and maybe a few kinds of animals?

Well, I would argue that we do perceive something like free will in those systems—we've just learned to override that instinct thanks to our advanced scientific knowledge. Remember, I'm arguing that we don't believe in free will so much as we perceive it—or, really, that our belief in free will follows on this form of perception. Just as I can learn not to believe in magic tricks or optical illusions, despite what my eyes might be telling me, I can learn to ignore my free-will detector when it suits me to do so. Left unmolested, people generally do perceive will in all kinds of complex systems, including ones we've constructed ourselves, like computer systems and advanced machines. What's more, it's precisely when such systems behave unpredictably that we're most likely to attribute free will to them; so long as the toaster functions properly, it's an ordinary kitchen appliance, but as soon as it starts popping toast out at random and malfunctioning in unexpected ways, you'll find yourself cursing and yelling at the "stupid" thing, asking "what's gotten into it," and fighting an urge to punish it for its refractory behavior. I think it's pretty clear that people in Ye Olden Times used to "see" free will in everything from bodies of water to weather systems to the universe at large, which is where we get a lot of our old-time religious beliefs. And the tendency to see free will in advanced computer systems, even among users who know exactly how they function, is so overfamiliar by now as to be banal.

3. But why do we believe that we ourselves have free will? After all, our own behavior isn't unpredictable. I can predict what I'm going to type the word "and"--and there, I just did it! See. I can predict my own behavior very well. So, using your model, I can see why other people would attribute free will to me—but not why I would attribute it to myself.

Are you kidding? I would guess that roughly 60% of human thought consists of people wondering either:

A) Why did I do X?

or

B) How can I get myself to do Y?

Why did I eat that Twinkie? Why did I skip the gym yesterday? Why did I stay up so late? Why did I just have that inappropriate thought? Why can't I stop thinking about that cute guy in my chem class? Why do I always fall for losers? Why can't I bring myself to finish that novel? Why can't I seem to concentrate lately? Why do I always sabotage my own best-laid plans? Why do I feel so nervous? Why do I like that dumb movie so much? Why am I always so stupid?

How can I get myself to exercise more? How can I get myself to finish what I start? How can I improve my concentration? How can I stop feeling so sad? How can I kick this awful habit? How can I become more devout? How can I get rid of my allergies? What's causing the weird pain in my side, and how can I make it stop? How can I get over that guy in my chem class? How can I resist eating one more potato chip? How can I make myself jump off this high diving board? What if I just do it—now. Nope, didn't do it. OK … now. Nope. OK …

We spend our entire lives ruminating on the infinitely rich mystery of our own unpredictable behaviors. Even when our actions appear outwardly predictable (your wife knew you were definitely going to eat that last potato chip) we ourselves remain painfully aware of the vast, roiling ocean of chaotic thoughts and emotions behind them, such that no one ever quite knows what he or she is going to think or feel from moment to moment to moment. We have more opportunities to observe our own behavior than we do with any other complex system, and therefore more opportunities to invent and test explanations for those behaviors, some of which prove accurate (I predict I'll type the word "and" now) and some of which prove false (I'm definitely going to make myself exercise before bed tonight). This lifelong observation of our own complex behavior is a perfect Petri dish in which to incubate the perception of free will.

4. But wait a sec. You're arguing that we falsely perceive free will in ourselves—even though we have none—and that we falsely attribute free will to things like storms and broken toasters and chatbots. But surely some of those attributions are more correct than others? Surely humans and tigers have more agency than stormclouds and toasters? If you believe that types of free will are equally illusory, wouldn't believing in gods and spirits and obstreperous toasters be just as rational as believing in human agency?

Well, yes and no. I'm arguing that free will is an illusion, yes. But it doesn't follow that the reality behind the illusion is the same in all instances. The illusion can be more or less informative depending on the circumstances. And, like most of our instincts, our instinct to perceive free will in complex systems is slapdash and somewhat flexible.

It's the same with any form of perception, after all. Our perception of taste can be thought of as a proposition strongly biasing behavior: e.g. this substance will be good (or bad) to eat. In our natural environment, this instinct served as a useful guide; in our modern environment, it's less useful, since we have technology that can make junk food, non-food, and even poison taste great. Similarly, our perception of free will biases us toward certain conclusions—this entity exhibits highly complex behavior, ergo it will respond to social incentives—that aren't always useful. The important thing is to be able to look past the instinct when needed.

Bringing the Argument Back Home

And that takes us back (finally) to Sapolsky's book. I think Sapolsky makes a convincing case that:

A) Free will is an illusion.

B) This illusion may have served us well in the distant past, but it often serves us poorly in modern societies.

C) Our susceptibility to the illusion is somewhat flexible, and we can and should learn to look past it in more instances.

I'm on board with that argument. But it doesn't necessarily follow that we can, or should, try to scrap the illusion entirely, and it certainly doesn't follow that we should hold people morally culpable for the sin of believing in moral culpability! In many ways, our perception of free seems analogous, to me, to our perception of beauty, in that:

A) It most likely evolved naturally under primordial conditions that no longer obtain,

B) It brings us great satisfaction but also causes considerable distress,

C) It results in social outcomes that are profoundly unfair from a rational perspective,

D) It's largely unconscious and automatic, and

E) It's highly variable and somewhat pliable.

So what do we do? Is it wrong that some people are perceived as beautiful while others are perceived as ugly? If so, how do we correct that wrong? The philosopher Kate Manne argues that we should train ourselves not to perceive beauty at all—any beauty, including the beauty of pleasing landscapes and natural objects. Should we similarly seek to abolish our perceptions of cuteness? Of charisma? Of harmony? Of grossness? All of these atavistic perceptual shortcuts get us into trouble in modern times. But what's the alternative? Eradicating the human sensorium? Forcibly rewiring the human brain so as to eliminate all heuristics and mental shortcuts? Reprogramming human cognition to align with utopian ideals? We can't even build computers that are free of such biases; how the heck can we justify a wholesale revision of the entire human species? And if we did undertake such a project, wouldn't we have to rewire animals, too, since they exhibit many of the same instincts? After all, Sapolsky has already concluded that neither humans *nor* animals have agency, so there's no sense in which either humans or animals could meaningfully consent to—or object to—such a procedure.

The only approach that makes sense to me is one that balances indulgence with moderation. Most of us learn, without too much trouble, to revel in the spectacle of impossibly beautiful movie stars without demanding that our life partners conform to those ideals; we enjoy the illusion of beauty without letting it rule our lives. What might it mean to develop a similarly balanced approach to our perception of free will? In some ways, it might mean moving forwards and backwards at once, indulging in a heightened moral sense in our recreational lives while behaving more coolheadedly as a society. It might even mean reveling in superstitions within safe contexts while understanding human behaviors in the real world in increasingly mechanistic terms. The vlogger Tom Vanderlinden argues that video games exert a forceful hold on people precisely because they present worlds in which religious feelings are justified—in which everything, in the narrowest sense of the phrase, has a purpose. It may be that in the not-too-distant future, we'll live dual lives: instinct-driven free-will-believers online and by night, pragmatic incompatibilists IRL and by day. Is such a world possible? Well, you decide.

1. Update: Finished the book. Turns out he doesn’t address it, actually. Not to my satisfaction, anyway. Weird.