Sapolsky vs Mitchell on Free Will

by Oliver Waters

Two popular books released this year have breathed new life into the ancient debate over whether we have free will.

In Free Agents, Kevin Mitchell argues that we do, and in Determined, Robert Sapolsky argues that we don’t. To be blunt, on the big issue at hand – Mitchell is right and Sapolsky is wrong.

Despite this, Sapolsky does make a compelling case however for not excessively condemning or praising people for their decisions. The sections of his book that focus on well-evidenced pathologies that drive criminal behaviour and why policymakers should take these into account are insightful and engagingly written.

The problem is that Sapolsky extrapolates from this to an outrageously untenable position:

“We are nothing more or less than the cumulative biological and environmental luck, over which we had no control, that has brought us to any moment.”

Sapolsky mistakenly infers that because there are many factors beyond your control that influence your decision-making, those factors entirely caused your decision. But all that is needed for a reasonable concept of ‘free will’ to get off the ground is that part of the decision was within your control. This of course begs the question whether a meaningful entity called you actually exists, which has causal power beyond the sum its parts.

Sapolsky will have none of this talk of a causally meaningful self, mainly due to his failure to distinguish between substance reductionism and causal reductionism. He thinks that just because we’re made of the same stuff as inanimate objects, we must operate according to the same causal principles. He takes believers in free will to be claiming that ‘emergent properties’ like conscious intentions and ‘selfhood’ somehow magically reach back down into fundamental particles and change their physical natures.

“It’s like believing that when you put lots of water molecules together, the resulting wetness causes each molecule to switch from being made of two hydrogens and one oxygen to two oxygens and one hydrogen.”

But no one is claiming this. What coherent defenders of ‘emergent properties’ do claim, to stick with Sapolsky’s aquatic analogy, is that when H2O molecules come together at a certain temperature and pressure, they start acting like ‘water’, and not like ice or water vapour. The resulting ‘causal properties’ of these different states of the same matter can be very different. You can craft a sculpture out of ice, for instance, but not water. Arranging simple neurons together in the right way similarly produces emergent causal properties (like free will) that those individual neurons do not possess.

Throughout the book, Sapolsky is like the proverbial man with a hammer looking only for nails. His deep expertise in biology leads him to focus on all things neuronal and hormonal. But he doesn’t draw enough upon information theory or computer science, making his approach to understanding the brain akin to someone seeking to understand the behaviour of a computer armed only with knowledge of hardware and none of software.

When something goes wrong with your computer, you know it may be a problem with a component of the machine, but it may also be a ‘bug’ in the programming. The former is a physical cause, and the latter is an informational cause. Prior to computers being a pervasive part of modern life, we lacked a demonstration proof of abstractions pushing around matter and energy, apart from in our own minds.

This is partly why Descartes and others were led to a supernatural dualism wherein our souls are somehow fundamentally separate to the material world. But nowadays we don’t just have a more rigorous conception of information, we have actual machines that can perform information processing. We fully understand how and why their specific outputs are caused by their programming, not by their material constitution.

To be fair, Sapolsky doesn’t reject abstract causes altogether. He does accept that your cultural beliefs and attitudes are prominent influences upon how your life turn out. But where he does acknowledge such informational causes exist, he still contends that they merely push us around, just like physical causes do.

This is the challenge that Kevin Mitchell tackles head on in his book: how can the information in our brains come together to form a coherent and causally potent self? Mitchell offers a strikingly lucid evolutionary story of how such a self emerged. I recommend reading his entire account – for now I’ll focus on the basic architecture of the self that he proposes.

The first aspect required is ‘autonomy’, which entails being causally insulated from one’s environment. Creatures with autonomy do not always react the same way to the same stimulus. They consist of a complex inner set of processes that are constantly active and formative, not merely reactive.

The second key quality is ‘wholeness’, which means that the brain is not just a set of complicated mechanisms. Every part is reciprocally related to, and constraining of, every other, and the activity of all parts is further constrained to be goal-directed. This integration of information in the brain means that its activity cannot be reduced into constituent components. Rather, decisions made by the system derive from its interpretation of the meaning and value of information relative to its overall goals. This property of ‘wholeness’ is what makes the ‘self’ more than the mere sum of its parts.

At its zenith in humans, a self can imagine alternative possibilities, and reason about which of those possibilities would be best. It can reason about its own goals and beliefs and decide to revise them. As Mitchell points out, this means that:

“Each of us has been very actively involved in the process of shaping the person we’ve become.”

Sapolsky spends much of his book describing examples of how factors beyond our control can indeed influence or dictate our life outcomes. If you are unlucky enough to inherit a deadly gene mutation, for instance, your life – or lack thereof – will not be up to you. But these are the exceptions. Most of the time, for most people, the dominant cause of their actions and life outcomes is their character, which is continually shaped by their active decision-making.

It certainly often seems like we’re not in control of our thoughts or desires. Sapolsky appeals to the common experience of not knowing what you’re going to think or do next:

‘…no matter how much you try, you can’t intend to intend something. You can’t will yourself to have willpower. You can’t think of what you’re going to think of next. It’s simply not possible.”

We certainly don’t have conscious control over the details of our next thoughts. But this is just like we don’t have conscious control over the detailed fine motor movements when we initiate actions. I don’t know ahead of time the exact order of letters that will come out as I write this, let alone the movements of my fingertips on the keyboard. In the same way, I can’t predict the exact words and images that will next pop into my mind. But it would be silly to conclude from this that I have no idea of what my next actions or thoughts will be. For instance, I have a very clear intention right now to think about free will over the next few hours.

Intending to want to do something is of course a little tricker. After all, I can’t just decide not to be thirsty right now. But while true in certain moments, this is not generally true. We intentionally control what we want all the time. For instance, I intend that tomorrow morning when I wake up that I will want to go to the gym. I plan on this by going to bed early, so that my future self is well-rested and therefore feels motivated to exercise.

Humans are unique in being able to shape not just the external world, but their own future desires. We can do this more prosaically by avoiding situations that trigger unwanted desires (not walking past the casino after a few drinks) or more fundamentally by building technology to directly modify our natures (an addiction-curing drug).

This makes us the only known creatures who are not ultimately motivated by survival in the moment, but rather by our moral ideas. We can ask whether the desires we hold right now are the right or best desires to hold. Indeed, there are endless examples of people choosing their own death because it was the right thing to do, perhaps in defence of a just political system, or in order to save many other lives.

Of course, this generic human faculty to make rational choices can often be inhibited or disabled altogether. Sapolsky does a great job at highlighting such cases and their implications for handing down moral judgments. Just as we used to punish people with epilepsy for ‘choosing’ to ‘let the demons in’, we no doubt still punish many people today for actions that are the result of biological causes we don’t yet understand.

But Sapolsky ventures too far by arguing that in principle, a mature neuroscience could attribute a strictly biological cause to any good or bad behaviour, just by detailing the relevant causal pathways. He fails to appreciate that above a certain (low) threshold of biological well-being, the dominant causes of our behaviour become our reasons. Sapolsky raises lots of studies that show our decisions are affected by changes in our hormones or blood sugar levels, but you won’t suddenly start believing in fascism or that the death penalty is a suitable punishment for shoplifting because you skipped lunch.

The deep grooves of our moral beliefs are not washed away by momentary physiological changes. The relative stability and robustness of our thoughts and desires over time is only possible because the rapidly changing physics of our bodies and brains do not determine them.

Sapolsky’s insistence that we are merely pushed around by our biology has important implications for how he thinks about moral and legal responsibility. He advocates for a ‘quarantine model’ that would treat criminals similarly to patients who have an infectious disease. The idea being that merely confining them suffices to protect others, and the threat of such confinement (if long enough) would act as a deterrent. The trickiness comes with the third usual justification for incarceration: moral rehabilitation or reform of the criminal.

The issue is that, contra Sapolsky, ‘criminality’ is not always synonymous with ‘biological illness’. Simplistically speaking, those who do the wrong thing are either neurologically or conceptually flawed. If they are the former, then it is indeed silly to morally condemn them for their actions, since they may have had no conscious control over them. But if they are conceptually mistaken, then they acted according to moral beliefs that were flawed or outright false.

Immoral and false beliefs cannot be surgically removed like tumours. Instead, we have no choice but to treat such a person as an agent who is voluntarily engaging with the world and can reflect upon their own ideas and values. This is what we mean when we say such criminals ‘could have chosen otherwise’. They could have chosen the correct course of action if they had known better. This class of criminal has the potential for moral reform, by learning why what they did was wrong, and why they ought to avoid doing similar wrong things in the future.

They are ‘responsible’ for what they did, in a way that a cat or emu is not, because they possessed at the time of the crime the potential to understand the wrongness of their action. Of course, if you wound back the clock to the exact physical conditions the microsecond before they committed the crime, then they would have done exactly the same thing. But this is irrelevant. What is relevant is that the criminal is a subjective self whose identity is extended over time. He is the kind of being who can think about and explain for himself what he ought to do and not do.

What we are blaming him for is his failure to cultivate the kind of character that does not commit crimes. In this sense it is not really the ultimate choice to commit the crime that we hold him responsible for, but the thousands of choices that preceded that. Holding him accountable today is sending a message that he could have sought a better understanding of right and wrong: he was not a mere passive receiver of experiences.

The very same propensity to do the crime likely still exists within his mind, at the time of sentencing. This means that he is likely to do that crime again, unless he changes. This change can only happen if he accepts and guides it himself, given the kind of autonomy and wholeness that characterises human minds. That is why he is ultimately responsible for becoming the kind of person who will not do the crime again.

We should have much more compassion for criminals, and should not be seeking to make them suffer, just for the sake of suffering. In this sense, Sapolsky and others are correct that the emotional drives we have for ‘retribution’ or ‘revenge’ are indefensible remnants from our evolutionary past that we should try to do away with.

But feelings of guilt and shame are essential for taking responsibility and undergoing moral reform. And unless we are seeking to completely erase a criminal’s personality and install a new one, they must undergo a necessarily painful, self-determined transition from their old self to a better self.

Consistent to a fault, Sapolsky also rejects praising people for their good behaviours. He thinks it’s perverse for a neurosurgeon, for example, to be considered ‘a better human’ than a criminal because ‘circumstances produced someone with the capacity to be a competent neurosurgeon.’ It’s wrong, in Sapolsky’s eyes, for the neurosurgeon to feel morally entitled in any way to his superior status and pay packet.

But for the same reasons discussed above, the neurosurgeon does in fact deserve some praise for becoming the person that they are today. Only by a mistaken denial of the extended self can we consider her to have not contributed at all to her own success.

Here Sapolsky also fails to distinguish between two meanings of being ‘a better human’:

1) The value contributed by a person

2) The moral status/rights of a person

The neurosurgeon is indeed probably better than the criminal in the first sense, but not in the second. No person is better in the second sense than any other, because their rights derive from their basic experiential capacities, not what they actually do with those capacities. Humans enjoy equal rights and dignity by virtue of what they are, not what they do.

In other words, attributing blame or praise to a person doesn’t imply degrading or enhancing their moral status relative to someone else. It also doesn’t imply attributing 100% responsibility for their actions to their character. Sapolsky’s fear that our belief in free will is causing unfair treatment is unfounded.

To sum up, Sapolsky’s denial of the existence of free will is really a denial of the existence of the self, as a persisting, informational entity.

This derives from a false, reductionist conception of causation, leading to a desolate, depressing landscape in which we are merely pushed about by the whims of our genes and upbringings. There is no pilot on the deck and no captain of our soul in this vision of our passive place in the world.

Mitchell’s book provides a good antidote – I recommend taking it.