On 10 September 2019, a headline in the Atlantic announced, “A Famous Argument Against Free Will Has Been Debunked.” The argument in question was based on the findings of what are usually called the Libet experiments.
Here’s how Theodore Schick and Lewis Vaughn describe the original experiments in their book Doing Philosophy:
By means of an electroencephalogram (EEG), which measures electrical activity in the brain, it’s possible to detect the nerve impulses that result in bodily movement (known as the readiness potential). By means of the electromyograph (EMG), which measures voltage changes in muscles, it’s possible to detect bodily movement. So Libet seated his subjects at a table, attached an EEG to their scalps, and attached an EMG to one of their forearms. In front of them was a rapidly moving clock and a button. They were given the following instructions: “Flex your finger to push the button when you feel like it, and tell us where the hand on the clock is when you decide to do that.”
These experiments, some of which were conducted by the American neuroscientist Benjamin Libet, were widely taken to show that the way we experience decision-making is misleading. As Bahar Gholipour puts it in the Atlantic, Libet used his findings to “make the case not only that the brain shows signs of a decision before a person acts, but that, incredibly, the brain’s wheels start turning before the person even consciously intends to do something.”
The story Gholipour tells of how later neuroscientists overthrew Libet’s results is a fascinating one. What interests me, though, is the way this episode intersects with the larger philosophical debate about free will and moral responsibility.
What, if anything, did either the Libet experiments or their subsequent debunking tell us about whether we are in control of our decisions in the right way for anything we do to truly be our fault? And why would it matter if this kind of autonomous decision-making were an illusion? Is this purely about satisfying our intellectual curiosity or does the debate have real-world consequences?
Libet and Free Will Skepticism
Some of the early reactions to Gholipour’s piece leave the impression that the original Libet results were a serious problem for belief in free will. Here, for example, is Daniel Dennett, a philosopher who does believe in free will and moral responsibility, crowing about the debunking:
The Libet results on free will and their many descendants are crumbling now, and there is more to come. A nice case of science exposing hidden dualist assumptions in neuroscience. https://t.co/Tc5hhVNMpN
— Daniel Dennett (@danieldennett) September 12, 2019
And here’s free will skeptic Sam Harris, sounding a bit defensive:
I have always regretted mentioning the Libet work in my book "Free Will" because it was never integral to the argument. When/if it is fully debunked, the case against free will remains unchanged. Free will makes no sense even if our actions arise exactly when we feel they do. https://t.co/S2ZhD0ABsy
— Sam Harris (@SamHarrisOrg) September 12, 2019
I’ve been critical of Harris’ past work on free will but, in this case, he’s half-right. The debunking of the Libet results doesn’t tell us much about the larger issue. Harris was wrong to think that those results gave us a reason to doubt the existence of free will in the first place.
Compatibilism and the Libet Results
Most of the debate about free will among academic philosophers is about whether free will is compatible with determinism. Compatibilists say yes. Incompatibilists say no. One-way incompatibilists say that free will is only possible in an undetermined or not fully determined universe. Two-way incompatibilists do not believe we have free will, independently of whether or not the universe is completely determined.
When I think about determinism, I find it helpful to think about the YouTube domino artist Hevesh 5, who spends days—sometimes weeks—creating elaborate structures made of thousands of dominos. When it comes time to record the video, she pushes the first domino onto the second. Once that’s done, she can sit back and watch with the rest of us.
We’re all determinists about falling dominos. There’s no question in our minds about whether domino #1524 in one of Hevesh 5’s structures will fall down when domino #1523 falls onto it. We don’t hold out any hope that #1524 will spontaneously bounce back and set #1523 back upright. Given the initial conditions and the relevant physical laws, #1524 falling down when #1523 hits it is the only thing that could happen—in other words, it’s determined.
Whether the physical world as a whole is like that is an open scientific question. According to the hidden variables interpretation of quantum physics, the laws governing the most basic particles that make up everything in our universe are deterministic. According to the Copenhagen interpretation, they’re probabilistic. (Let’s put aside exotic possibilities about splitting universes and the like.) If one-way incompatibilists are right, this debate among theoretical physics has a vital moral significance. If everything that ever happens, including every human decision, is determined, then these thinkers believe that no one is ever truly in control of their actions.
When we get up to the level of tables, chairs, dogs, cats, people and so on—what the late great philosopher W. V. O. Quine liked to call “mid-sized dry goods”—no one denies that most physical processes are deterministic, or close enough to deterministic as to make no difference. When my car breaks down, I expect my mechanic to be able to give me a cause-and-effect explanation. If he tells me that sometimes cars just stop working for no particular reason, I won’t conclude that reality is more mysterious than I thought. I’ll conclude that I need a better mechanic.
If one-way incompatibilism is correct, however, there must be at least some exceptions to this general rule—patches of indeterminism in a mostly deterministic world—and some of these patches must involve decision-making in the human brain. According to compatibilists (and two-way incompatibilists) however, the debate about quantum physics is irrelevant to human freedom—or at least the kind of human freedom that matters for free will.
Many people find compatibilism counterintuitive when they first encounter the position. I try to explicate it here and especially here. The important part is that many contemporary compatibilists grant that there’s a kind of free will that we wouldn’t have if determinism turned out to be true. We wouldn’t be free to do otherwise. However, they think that we can exercise the kind of control over our own actions that matters for moral responsibility—the kind of free will that matters for sentences like the defendant killed all those people of his own free will—even if the world turns out to be completely deterministic. Philosophers such as John Martin Fischer and Ryan Lake typically believe that what matters for this kind of autonomy is our ability to reflect, consider, understand and be moved by reasons for and against various courses of action. Whether or not every stage in this process is mechanistic is beside the point. What matters is that, in order to be the kinds of agents that can be truly blameworthy or praiseworthy for our actions, we have to be governed by our ability to reason.
If this compatibilist reasons-responsiveness account of free will is the right one, it’s hard to see how the original Libet results could have been relevant. Let’s say I decide to donate another $27 to the presidential campaign of Bernie Sanders. Presumably, my doing so is a result of my general awareness of the reasons why I think a Sanders administration would be more in line with my political values than, for example, a Biden administration. If the gears started turning a fraction of a second before I was aware of my decision to make the donation, what of it? If I’m a convinced compatibilist, I think I could be making a free decision for which I’m fully responsible even if there were an unbreakable chain of cause and effect.
Incompatibilism and the Libet Results
But what if compatibilism is wrong? Perhaps a truly free action must be one that could go either way right up until the moment of conscious decision. Should the Libet results (as originally interpreted) threaten our belief in this kind of incompatibilist free will?
Libet himself thought that his results were compatible with, if not free will as we usually understand it, at least free won’t—the power to consciously veto decisions we’ve already started to make.
Let’s assume that a Libet subject, with an EEG attached to his scalp and an EMG on his forehead, given the confusing instruction to press a button when the feeling arises, is making decisions in the same way that most of us make most decisions—including far more morally important ones—in our day-to-day lives: in other words, on a kind of autopilot.
The incompatibilist account of free will developed by philosopher Robert Kane starts from precisely this assumption. While Kane believes that the human brain really does sometimes operate in an indeterministic way, he thinks it’s likely that the vast majority of our day-to-day decisions are causally determined by our preferences, values, and so on. However, he thinks that we can be responsible for all those decisions as long as those preferences, values, etc.—those autopilot settings of our character that govern our day-to-day behavior—are themselves set during what he thinks are relatively rare moments in which we make truly indeterministic self-forming decisions. The idea is that, from time to time, we experience conflicts between approximately equally strong preferences, and that these are moments in which our characters are formed. Since this mental arm-wrestling happens against the background of quantum uncertainty in our brains, the outcome isn’t predetermined.
To see how Kane’s view works, think about drunk driving. If I get behind the wheel of a car after far too many glasses of Laphroaig, I’m morally responsible for any reckless driving that might result—even if, as is probably the case, I actually can’t drive any more carefully than I am in that moment, given the road conditions and the conditions in my scotch-addled brain. The reason we normally think I’m responsible anyway is that we think that I could have done otherwise at some point earlier in the evening. I’m endangering people now because of my free decision not to hand my keys to a designated driver hours ago. Similarly, Kane thinks that, no matter how rare indeterminsitic self-forming decisions are, if my day-to-day behavior can be traced back to them, moral responsibility is intact.
Free Will and Punishment
Let’s say some theory of free will—whether an incompatibilist theory like Kane’s or a compatibilist one like the reasons-responsiveness account—is correct and we really can make decisions for which we are responsible. Why would that matter?
A common claim, made for example by the biologist Jerry Coyne (who responds to one of my previous articles on this topic here), is that if free will turns out to be an illusion, this is good news for those pushing for a more humane and enlightened criminal justice system. This claim is based on the assumption that there are only two ways of making sense of why we should legally punish dangerous criminals—the theory that criminals should be punished for having done bad things for which they’re morally responsible and the theory that criminals should be punished for the same reason that we quarantine plague victims—not because we bear any ill will towards them, but because it’s necessary to keep everyone else safe. The first is a retributivist theory and the second is a consequentialist theory of punishment.
These are not, however, the only options. A third theory, popular among some contemporary philosophers, is a two-level rights forfeiture view. According to this view, innocent people have a categorical right not to be punished. Guilty people forfeit this right by committing sufficiently bad acts for which they are morally responsible. However, this by itself doesn’t give us a reason to punish them. Punishment itself can only be justified by consequentialist reasons.
If free will is an illusion, the pure consequentialist view of punishment is the only option that makes sense. If we really are sometimes in control of (and hence responsible for) our actions, however, the purely retributive theory of punishment and the two-level rights forfeiture view are both on the table.
It’s obvious why the difference between pure retributivism and pure consequentialism matters. It might be a lot less obvious, however, why the difference between pure consequentialism and the two-level view is important. After all, both views suggest that we move toward a much more humane and minimalistic prison system, focused on rehabilitation. Both invite us to reject the common refrain made in response to anything perceived as overly lax treatment of prisoners—it’s supposed to be punishment!
So what’s the difference?
Free Will and the War on Terror
The day after Gholipour’s free will article was published in the Atlantic marked a grim anniversary. On 11 September 2019, US citizens who hadn’t yet been born when Al Qaeda attacked the World Trade Center and the Pentagon turned eighteen and hence became eligible to serve in the ongoing war in Afghanistan. President Trump called off peace talks between the United States and the Taliban just before the anniversary.
Take a moment to think about everything that’s been done in the last eighteen years in the name of fighting terrorism. Two countries were first bombed and then invaded and occupied, resulting in vast numbers of deaths and mutilations and even larger numbers of refugees. The overwhelming majority of these victims weren’t terrorists—or criminals of any kind. America has also carried out drone strikes in Pakistan, Somalia, Libya and Yemen. Even though the United States is not at war with any of these countries, these strikes were used to kill people based on unproven suspicion. In the case of signature strikes, the people planning them didn’t even know the intended victims’ names. CIA analysts watching surveillance footage in Virginia would make a determination, based on a target’s pattern of life, that he was probably involved in terrorist activity, and then he would be extra-judicially executed. It’s impossible to know how many of these intended targets were guilty. The number of non-intended targets who have been killed is also hotly contested—as is the underlying question of whether drone strikes are more precise than strikes carried out by manned planes—but no one seriously denies that intended targets have been a minority of the victims. Finally, between Guantanamo Bay, Bagram Air Base in Afghanistan, various Bush-era black sites in Eastern Europe and other cases, a disturbingly large number of terrorism suspects have been detained and often tortured, without any of the usual legal protections granted to accused criminals.
All of this was justified based on consequentialist calculations—that it was better to kill, maim, torture, dislocate or indefinitely detain innocents than to let a larger number of innocents die in the terrorist attacks that such measures might prevent. One way of pushing back is to argue that the calculation is flawed—that the cumulative effect of all this is to inspire more terrorism than it prevents.
If this is our only reason for opposing these practices, however, our opposition to them hangs on a thread. What if the maths of the war on terror doves is wrong, and more lives are spared than are taken or ruined? Does that really make all of this morally acceptable?
Would such a discovery be enough to prompt you to drop all your objections to drones and indefinite detention and the rest? If not, rights-forfeiture theory could make sense of why not. At the very least, all of this should shake your confidence that the Sam Harrises of the world are right when they tell you that, if free will and moral responsibility turn out to be illusions, we won’t lose anything that matters.