Do we have free will, or fate, or both? (Sam Harris is Wrong, Part 2)

[I’m going to re-write most of this. Please be patient for now.]

This series will be written with the assumption that you are up to speed with the nature of Harris’s positions, or that you understand the state of the debates in question. If that assumption does not apply to you, that’s ok; just please know that many counter-arguments you might come up with have probably already been answered by Harris.

I would rather ask people to do their own research than rehash Harris’s entire arguments for him. I will be skipping some beginning and intermediate steps – mainly comprised of addressing the most common (and bad) objections to Harris’s positions. This is necessary in order to skip to the more interesting (and better) objections.

With that context, determinism is true, as far as everyday existence is concerned, notwithstanding gaps in our knowledge. I won’t belabor that point, and simply take for granted the assumption that Harris is right about determinism.

Determinism would theoretically allow us to predict the future perfectly, given perfect knowledge of starting conditions, perfect knowledge of physics, and infinite information processing power. In other words, determinism holds that what happens is an exact and inevitable consequence of the preconditions playing out according to rules of cause and effect.

Determinism reminds us of fate, except for the fact that fate is a literary rather than physical concept, and the fact that fate connotes the absence of free will, which brings us to our topic. 

Sam Harris believes that free will does not exist. I on the other hand am a compatibilist; I believe that free will is logically compatible with determinism. They can both exist without contradiction. 

Neither of us believe that decisions are without consequence. We agree that human decisions are physically predetermined. And if you think, “what’s the point of trying to do anything, if it’s all predetermined?” you’re missing the point; the fact that your decisions are predetermined does not reduce their importance. More on this later, but I believe people like Sam Harris are mainly responsible for the confusion here.

There is dumb conception about free will, with an equally dumb name, “libertarian free will.” Libertarian free will rejects determinism in favor of the notion that ideas spring independent somehow from the world of matter. Neither Harris or I believe this.

I’m not content to leave it at that. I contend that, despite the fact that thoughts are technically just the manifestation of a materialistic causal chain, there is more to the story. The activities of the brain are more interesting than ordinary materialistic movements.

So the question is: if our thoughts are just the firing of neurons, and the firing of neurons is just the movement of atoms, in what sense do we have free will? To answer this, I fill begin with an analogous question. Is a dice roll random?

The deterministic view is that a dice roll is not truly random. If you know the starting positions + momentum + air resistance + friction + weights + shapes + etc., you could theoretically calculate the exact result of a given dice roll.

Despite this, the purpose of a dice roll is to intentionally obscure these variables to the point where no one could realistically predict the outcome of a given roll. This is done by inducing chaos through complexity.

Here we see a divergence between what determinist physicist mean by “random,” and what casual people mean by “random.” A dice roll may not be truly random, deterministically. But it is random enough.

We should preserve what matters to a layperson. In keeping with that, if we were to imagine a “compatibilist” view of randomness, it would be the following. A deterministic universe is compatible with the idea that a dice roll could be justifiably, for all intents and purposes, be called random.

It’s the same with free will. We may not have free will, in the world of theoretical physics, but it doesn’t matter; we have free will “enough,” and that “enough” is what I mean by “free will.”

What convinced me to become a compatibilist was actually a conversation between Sam Harris and another prominent atheist, on Harris’s podcast no less, in the episode Free Will Revisited.

You would be forgiven for thinking that Harris comes off better in that exchange; he has a gift with words. Dennet convinced me, though, with an argument about chess AI programs:

We have a decision maker [a chess playing program] that is in a demonstrably deterministic world. It’s playing chess. And it looses the game, and it’s designer says, “well, it could have castled.” What do you mean, it “could have castled?”

What the designer means is it was just the luck of the draw: the chess program, like any program, is going to consult a random number generator (or pseudo random number generator) at various points, and this time it chose wrong. However, it chose wrong because when it got a number from the pseudo-random number generator, it got a 1 rather than a 0. Flip a single bit, and it would have made the other choice. In other words, it’s not a design flaw.

What’s great about this analogy is that the decision-maker is an obviously and fully deterministic system, even in a way that humans can design and understand. Computers are never truly random. They only pseudo-random, meaning they have processes which mimic randomness.

Suppose you have two copies of the same chess program. Give them the same keys for generating randomness. Then play the exact same moves against both. Both programs will play exactly the same moves as each other back.

Despite this, we still talk about programs as if they have agency to some extent. “It chose the wrong move.” It didn’t really get a choice. What we mean is that its domain of permissible choices came down to rng. I’ll come back to unpredictability shortly.

We are starting to see a glimpse of deterministic “agency.” But humans brains are, of course, even more interesting than chess programs.

Have you ever given much though to what makes something “conscious?” This is an extremely complicated topic, and outside the post of this post. Suffice to say that information processing of a complicated and integrated enough form can assume such a character that what we call consciousness becomes an emergent property (or byproduct) of that system.

Here is my running definition of free will. Someone or something has free will to the extent that its consciousness has the freedom to express itself as a conscious agent. To roughly elaborate on “express itself,” a “will” is free if it could only have emerged as a consequence of the nature of consciousness.

That is aloof. But we all intuitively understand the difference between giving away $100,000 at a charity fundraiser vs at gunpoint. Both are actions in our deterministic universe, but one is the product of careful consideration, while the other is hardly a choice.

The hardest decisions are the ones that most seem to emerge inexplicably “out of the ether” of your mind.  Harris would only use this fact to denigrate the concept of free will. Your decisions are ultimately a mystery to you, he argues. You have no idea why you picked what you picked.

It would terrify most people, he states, to learn that their most important decisions come down, proverbially, to a single neuron. Are we really making decisions, if they’re like a flip of the coin?

I argue that that characterization undersells the beauty of what’s going on.

Dennett uses the term “degrees of freedom.” Dennet gives the example of joints of the body. The more joints, the more possibilities of expression in, say, dance. This concept is crucial for understanding the degree to which something has free will.

The more degrees of freedom you have, the harder it is to explain the system as a whole. Counter-intuitively, the decisions that are the easiest to explain are the most forced, and therefore the least attributable to free will. You do not have much free will when you sign a document at gunpoint, or when you move your hand away from a hot stove. The free will choices arise mysteriously.

In some complex systems, complexity does not scale linearly. The addition of 10% more subcomponents does not make it 10% more complex. Instead, 10% more subcomponents may double the complexity, because of how those new components interact with existing components. Imagine an insane version of this, and that’s the human brain.

This is a picture of a NEAT-style neural network I coded up:

Screen Shot 2019-08-15 at 12.27.30 AM.png
Circles are neurons (nodes), arrows are synapses (edges). Colored nodes are input and output nodes

And this is a connectome mapping of an insect brain:

fly brain
And this is only a fly brain!

The key word here is integration. I can only speak to my neural network, but, every neuron:

  • Matters. The addition or subtraction of a single neuron can substantially alter the choices made by the brain. Notice how interconnected the network is. One tiny change, the firing of a single neuron, can cause ripple effects that radiate outwards and effect the entire brain. A single switch in the mind can spiral and compound into an overall decision.
  • Only matters in the context of the network. A neuron does not work in isolation. A so-called single point of change only becomes such, is only paid deference, in the context of a complex system for arriving on that point.

To build off that second point, your freest decisions are irreducible. When an idea will seem to “pop” into existence, it’s hard to identify  specific “cause.” If we had to name a “cause,” it would be, altogether, the nature of your consciousness, in all of its ambiguity.

At this moment, ideas could be competing with each other in a darwinian struggle. Whatever consciousness constitutes, that’s the very thing we care about, whether it is a mystery to us or not. That is synonymous with the source of your freest decisions.

Harris gives the example of a man who committed murders because of a brain tumor. It is self evident that that man did not have free will. Fair so far. Harris then [equivocates] all brains as the tumor case with more steps. This is an error. Free will is a property that comes in degrees, and those extra steps are exactly what we care about.

Of course the man with a brain tumor did not have free will. A tumor is not conscious. His decisions were driven by a single factor, (the brain tumor), rather than from the nature of consciousness. His decisions were dependent, not integrated. The billions of neurons, and trillions of synapses, were not allowed to express themselves.

Now I’ll address and accusation that is put to Dennett, and will be put to me. The claim is that what we are calling free will is not free will. This is the great smear against compatibilists: that we are trying to “redefine” the term to suit our own interests.

I believe the opposite is true, and that “libertarian” free will is merely a straw-man that ends up confusing fence-sitters into rejecting determinism. My evidence for this is the ubiquity of the following inquiry:

If we don’t have free will, then why should anyone be judged with moral accountability? Are you culpable for your actions, or can you be held responsible?

To be blunt, these questions obviously come from the type of person who is not very familiar with the free will debate. But that is most people. You would be shocked at how common are variants of these questions.

We have to identify the source of the confusion. The questions are incoherent, unless we allow compatibilists to define free will. This question conflates free will the ability of conscious people to make decisions which have consequences.

That’s a compatibilist notion of free will! The prevalence of those questions proves that the compatibilist definition is held/assumed by the general public! The compatibilists aren’t trying to redefine anything. It’s the “free will doesn’t exist” crowd that is actually trying to change the definition to suit their interests!

What am I calling their interests? Well, I think the anti-free-will crowd does have something interesting to say… about libertarian free will. Perhaps it’s easier for them to make their point if they call that simply “free will,” so they can avoid getting sidetracked by ethical questions, or worse, artificially expand the scope of their argument.

Let’s step back and consider how people think about these topics outside of philosophy circles. People like Sam Harris, and even me, think about things on an abstract yet granular level, more so than the general public.

To again be blunt, I believe most people are indifferent to determinism. They don’t wonder if, from a physical perspective, everything is technically part of a stringent chain of cause and effect. They may find that interesting, but when most people ask about free will, they’re probably not asking about it on such a granular level. They more likely mean, “can my brain make decisions that have a real effect on the world?” to which the answer is yes. They’re asking about the real-life implications of this topic.

Speaking of real-life implications, this is not just a semantic disagreement. This topic is more than theoretical, and more than philosophical waxing. Harris confesses that his beliefs on free will do influence the attitudes and behaviors he endorses. I happen to disagree with those as well, and I think they get to the heart of our disagreement.

I’ll start with but one example. Harris explains that, because no one acts of their own free will, there is no room for hating people (in the most literal sense). In his view, you should not hate the gunman who killed your family any more than an equivalently bad tornado. It’s nonsensical to hate someone for what they didn’t choose, after all!

That seems uncontroversial enough. Hate is the emotion it’s most popular to demonize. So I will do the unpopular thing and challenge that position. A tornado is not a human. A hurricane is not conscious, and does not know it is doing harm. A human can act based on imagined futures. A human can understand retribution and justice.

The desires for hate and revenge do not exist randomly. They are the product of evolution. They must have helped our ancestors survive somehow, and the question is how. This may not be a relevant in our age of advanced legal systems, but in ancient and prehistoric times, revenge upheld your status as not someone to be messed with.

We are starting to see that this issue informs our views of morality, retribution, and criminal justice. Before I give my own take on these issues, I will first address the anti-free-will take. The likes of Sam Harris have their own answer to the pivotal question (repeated below):

If we don’t have free will, then why should anyone be judged with moral accountability?

They would us an analogy of a computer program. If a computer program is malfunctioning, then we will fix it, or even decommission it. For this utilitarian reason, there are consequences for misbehavior, regardless of whether or not the misbehavior emanates from an entity that has free will.

And thus, you have a purely practical, anti-culpability, anti-blame version of justice. How convenient.

…And also insufficient. I have noticed, among opponents of free will, an effort to downplay the importance of deterrence. It’s as if they forget about it.

Deterrence is not just another variable in the equation. The way most people instinctively conceptualize culpability is directly and primarily informed by the role of deterrence, whether they know it or not.

Why, for instance, do we distinguish between the moral culpability of someone who commits murder because of mental illness, and someone who commits murder out of calculated spite?

The mentally ill person cannot understand the concept of legal consequences, and if they can, they cannot make it prevent their actions. For these reasons, deterrence does not work on them.

When we catch the mentally ill murderer, we will try to remove them from society for the good of themselves and others, but we do not harbor the desire for revenge against them like we would a sane person. We say they are “not responsible for their actions,” even though when actions emanated exclusively from them.

What does this mean? It means that they “get a pass” from the deterrence-based aspect of our justice system. There’s no point in trying to deter people who can’t understand deterrence. We need only “send a message” to the people who could potentially be receptive to it, that if they do wrong, they will be punished.

The double standard is perfectly explicable. We do not hate something without free will. Hate fuels the drive for vengeance. The desire for vengeance creates deterrence. A person without free will does not respond to deterrence. It’s all connected.

My point is this. Justice should care that an individual “has free will” to the extent that an agent with free will should be able to expect, envision, and comprehend the consequences of their actions.

Harris’s example of the man with the brain tumor only highlights the usefulness of the concept of free will. On one hand, a sane man might commit heinous violence in full knowledge of what the the legal consequences would be if they got caught, but thinking they could get away with it. Part of the point of the punishment is to disabuse them of that notion. By contrast, a murder induced by a brain tumor was not factoring in the legal consequences at all.

I’m not trying to “pretend” free will exists out of some worry about the moral consequences of its nonexistence. Instead, I am affirming that the kind of free will that has implications for moral systems, the kind that informs accountability, the kind that most people actually care about, does exist.

Sam Harris is wrong for thinking that nothing real is worthy of the name “free will.” Free will, of the type that matters, insofar as society is concerned, is real.