The death of free will has been announced many times. Often neuroscientists get the credit for killing it. Over the past fifteen years or so, I have devoted a lot of time and energy to explaining why the news is premature at best. Despite my efforts, the obituaries continue to emerge. Here I will content myself with tracing an interesting strand in the ongoing debate about whether neuroscientists have killed free will and commenting on a recent development that takes us beyond neuroscience to some of you, dear readers.
Philosophers love arguments. We love to name them too. Here’s an argument to consider. I call it Psycho Killer:
1. In neuroscience experiments of a certain kind, all the decisions on which data are gathered are made unconsciously.
2. So probably all decisions are made unconsciously.
3. A decision is made of one’s own free will only if it is made consciously.
4. So probably people never decide of their own free will.
I have argued against claims 1 and 2 in print, online, on stage, and in TV and radio interviews. To readers interested in a detailed but accessible presentation of these arguments of mine, I recommend a very short book I wrote for a general audience, Free: Why Science Hasn’t Disproved Free Will. There I argue that the data do not support the claim that decisions made in the experiments at issue are made unconsciously, and I argue that even if these decisions were made unconsciously, they are so different from some other kinds of decision that generalising from the claim that the former decisions are made unconsciously to the claim that all decisions are made unconsciously would be a huge stretch.
I will explain these points briefly before forging ahead. Let’s start with the experiment by neuroscientist Benjamin Libet that got the ball rolling.
Participants are instructed to flex a wrist whenever they wish and then report a bit later on the first moment at which they were conscious of their decision (or intention or urge) to flex. The reported time is called “W-time”. (The “W” stands for “will,” which is short for “conscious will”.) The decisions at issue are not about flexing sooner or later; the participants have already agreed to do that – and to do it many times. Instead, they are decisions to flex right then. I call such decisions – decisions for immediate action – proximal decisions.
Participants watch a Libet clock. It’s fast: a spot or a hand revolves around the clock face in about two and a half seconds. A bit after they flex, the spot or hand stops moving and it’s time to make a W-time report. During my stint as a subject in a Libet-style experiment, I made my reports by moving a cursor to mark the place on a computerised clock where I estimated the hand was when I was first conscious of something like a proximal decision.
Readings of electrical conductivity are taken from the scalp, using EEG technology. (“EEG” is short for “electroencephalogram”.) In order to produce readable EEGs, Libet’s participants flex at least forty times during each session. Readings are also taken from the wrist so that the experimenter can tell when muscle motion began during wrist flexes. Libet discovered that when he repeatedly reminded his participants to be spontaneous and not to plan their flexes in advance, he got EEG results that started ramping up about 550 milliseconds – a bit more than half a second – before the wrist muscle started moving.
On average, W time was about 200 milliseconds – one-fifth of a second – before muscle motion began. Libet’s take on these results was that a proximal decision to flex was made unconsciously about half a second before the wrist muscle moved and about a third of a second before the person became conscious of the decision. If he is right about this, then claim 3 in Psycho Killer tells us that these decisions were not made of the participants’ own free will.
It turns out that there is no good reason to believe that proximal decisions are made when the EEG starts ramping up and some reason to believe that they are made around 200 milliseconds before muscle motion begins (see Free). Also, any proximal decisions made in Libet’s experiment are made arbitrarily; and if participants are following their instructions, they are not consciously reasoning about what to do. So these decisions occur in very different circumstances than do nonarbitrary decisions, made after painstaking evidence gathering and reflection, about things that are very important to us. And, given this difference, the generalisation from the alleged finding that the decisions Libet studied are made unconsciously to the claim that all decisions are made unconsciously rests on very shaky ground. Compare it to the generalisation from the finding that all Americans who voted in the 2016 presidential election and proudly wear MAGA hats voted for Donald Trump to the claim that all Americans who voted in that election voted for Trump.
The points I have just summarised may also be made about new-wave Libet-style experiments. I have made them myself about experiments of this kind that use such technology as functional magnetic resonance imaging and electrode grids placed on the brain (see Free). Instead of doing that again here, I will take up a response I occasionally hear. Here it is: Even if I am right that Psycho Killer is flawed in the ways I sketched, studies like Libet’s prove that there is no free will because free will depends on something nonphysical – specifically, a soul – making or causing decisions, and the experiments prove that our decisions are caused by brain events.
Once we get to this stage of the debate, a question about meaning looms large. What does “free will” mean? Is a soul being at work built right into the meaning of “free will”? The overwhelming majority of living philosophers who write about free will definitely answer no to this last question. But people who answer yes to it may not be impressed by this fact. If a group of philosophers who answer no and a group of neuroscientists who answer yes were to seek a way to resolve the dispute, how might they proceed? Neuroscientific studies won’t reveal what the expression “free will” means, and I have already mentioned that agreement among philosophical specialists might not impress the opposition. What is to be done?
Is there a third party the disputants can appeal to for some guidance? What about ordinary speakers of English who are neither scientists nor philosophers? We can gather evidence about what nonspecialists mean by “free will”, including evidence about whether, in ordinary usage, free will is essentially bound up with souls. In fact, this project is already under way.
Andrew Monroe and Bertram Malle conducted a study (published in 2010) in which participants responded to the following request: “Please explain in a few lines what you think it means to have free will.” The 180 participants were undergraduates at the University of Oregon. Monroe and Malle report that none of them mentioned a soul. Of course, even if souls don’t quickly come to mind when nonspecialists are thinking about what “free will” means, a little prodding might conjure up souls. In a study by Eddy Nahmias (mentioned in a 2011 publication), participants were presented with the statement “Humans have free will only because they have nonphysical souls”, and only 15% to 25% agreed. Also, only 18% agreed with the statement “Our power of free will is something that is not part of our brain”.
As part of a larger study (published in 2018), Andy Vonasch, Roy Baumeister, and I ran the following story by online participants: “Imagine that scientists finally discover that there are no souls. So, for example, they discover that you don’t have a soul, that your friends and neighbours don’t have souls, and so on. Like everyone else, John doesn’t have a soul. One day, he sees a twenty-dollar bill fall out of the pocket of the person in front of him. He picks it up and keeps it for himself.” The results: 90% agreed with the statement that John has free will in this situation, and only 4% disagreed. We got very similar results with a much larger group of participants when we had John pick the person’s pocket (from which a $20 was dangling): 87% said John had free will, and 10% disagreed. About 90% of respondents are telling us that people don’t need a soul to have free will.
Recall how we got here. In various places, I have argued against claims 1 and 2 in Psycho Killer; and I have heard it said in reply that what actually kills free will is that it depends on a soul making or causing decisions and studies like Libet’s show that the brain is doing all the work. That reply takes the debate to another level. The claim that free will depends on souls is, at bottom, a claim about what “free will” means. What “free will” means can’t be settled by neuroscience, and it may never be settled by philosophy. But we can at least get evidence about whether or not ordinary usage of “free will” is in line with the idea that free will depends on souls. And the evidence I have seen speaks very strongly in favour of a resounding no.
I have written a lot about free will, at least as I conceive of free will. If I were to learn that 90% of nonspecialists build hard-working souls into the meaning of “free will”, I would probably conclude that what I have written so much about, however interesting and important it may be, isn’t what goes by the name “free will” outside my ivory tower. Perhaps those who have given the soul reply to my critique of Psycho Killer will be moved by results of the kind presented here to question their assertion that free will depends on souls. After all, there is no neuroscientific evidence that you need to have a soul in order to have free will.
The story does not end here. In the literature on free will, there is a position called restrictivism. According to some restrictivists, free will can come into play only when we are tempted to act contrary to what we believe we ought to do, morally speaking. (The claim is that free will is restricted to scenarios of that kind; hence the name.) Other restrictivists see free will as extending more broadly, but none of them counts arbitrary picking – for example, arbitrarily picking a moment to begin flexing a wrist – as an exercise of free will. So restrictivism implies that Libet-style studies definitely are not about free will. It implies that Libet is looking for free will in the wrong place.
Is restrictivism in line with ordinary usage of “free will”? In an interesting study (published in 2017), Robert Deutschländer, Michael Pauen, and John-Dylan Haynes put this question to the test as part of an effort to ascertain whether conceptions of free will that seem to be guiding Libet-style experiments are in line with lay thinking about free will. One of their eight vignettes features a man who deliberates about which of two very different job offers to accept, decides on one of them, and signs a contract for it. In another vignette, a man who receives two job offers with identical “job conditions” does not deliberate and spontaneously signs a contract for one of them. Participants answered the following question on a seven-point scale ranging from “not free” (1) to “free” (7): “How free was the presented action?” The mean score on the former vignette (thoughtful choosing) was 5.03; and the mean for the latter (arbitrary picking) was 5.51. On the whole, then, lay respondents regard the action as free in both cases – and perhaps more free, or more clearly free, or something of the sort, in the latter case. One implication is that nonspecialists, for the most part, seem not to be restrictivists.
I myself have never endorsed restrictivism. I am happy to suppose that both thoughtful choosing and arbitrary picking can be free. I am also happy to say that both of the men in the vignettes under consideration decide to sign the contract they sign: one does so on the basis of deliberation, and the other does so spontaneously, with no deliberation. The latter decision is much more like the decisions participants in Libet’s experiment are supposed to make than the former decision is.
Does the anti-restrictivist finding at issue have any special bearing on claim 2 in Psycho Killer – the generalisation from the assertion that all the decisions on which data are gathered in Libet-style experiments are made unconsciously to the further assertion that all decisions are made unconsciously? Does it give us grounds for believing that this generalisation is warranted? No and no. The problem with the generalisation is that making arbitrary trivial decisions while complying with instructions to refrain from thinking about what to do is different in some obvious ways from making nonarbitrary decisions about things that matter to one after careful deliberation, and these differences are obstacles to the inference made in claim 2. For one thing, conscious reasoning about what to do might raise the probability of conscious deciding. (Consciously deciding to do something is a matter of being conscious of making the decision one makes when one makes it – as opposed, for example, to becoming conscious of the decision after one has made it.)
Suppose you discover that every tweet ever sent by Donald is false. Would you infer that all tweets – no matter who sent them and no matter what they say – are false? Of course not. The circumstances surrounding Donald’s tweets may differ from the circumstances surrounding some other tweets in such a way as to explain why although his tweets are all false, many other tweets are true. Something similar may be true of decisions made in Libet-style experiments as compared to many other decisions: even if the former decisions are made unconsciously, many others may be made consciously. Because I have carried on about this, I should add a reminder to forestall misunderstanding: As I mentioned, I am not convinced that proximal decisions are made unconsciously in Libet-style experiments. My point is that, even if they are, that leaves plenty of room for consciously made decisions. And bear in mind that, according to Psycho Killer, what proves that free will is probably dead is the fact that all decisions are probably made unconsciously.
When Mark Twain spotted a newspaper account of his death, he drolly announced the truth: “The report of my death was greatly exaggerated.” Reports of the death of free will – at least as I and perhaps a majority of nonspecialists conceive of free will – are in the same boat.