The Brain Lies

In 1906, Ludwig Boltzmann hanged himself in a hotel room in Duino, on the Adriatic coast. He was sixty-two. He was one of the greatest physicists of his time — and one of the most rejected.

Boltzmann claimed that matter consisted of atoms. Today every schoolchild knows this, but in 1906 it was a contested hypothesis: Ernst Mach, one of the most influential physicists of the era, considered atoms a “convenient fiction,” the Nobel laureate Wilhelm Ostwald denied their existence altogether, and Boltzmann himself was mocked at conferences and his papers rejected. His idea that macroscopic properties — temperature, pressure — are the statistics of microscopic particles looked like pure metaphysics.

Five years after his death, Einstein explained Brownian motion — and atoms became fact.

On Boltzmann’s tombstone his formula is engraved: $S=klnW$. Entropy is the logarithm of the number of microstates. The link between information and physics. The one Shannon would lay at the foundation of information theory forty years later. The one from which Bekenstein, sixty years later, would derive the entropy of black holes.

Boltzmann didn’t live to see it, but his formula lives — on his tombstone, and in every bit of information.

I’m opening this chapter with Boltzmann because the next problem is built the same way. This time the person who posed it wasn’t ridiculed (the times are different, after all), but the problem hasn’t been solved either — yet.

---

In 1995, the Australian philosopher David Chalmers split the problem of consciousness in two. The “easy” problem and the “hard” one. Since then physics, neuroscience, and philosophy have not recovered.

The “easy” problems aren’t simple. They are “easy” only in the sense that we know how to approach them. How does the brain process visual information? Neuroscience. How do we recognize faces? Computational modeling. How does attention work? Cognitive psychology. Hard, slow, expensive — but solvable in principle. A method exists.

The “hard” problem is different. It sounds simple: why does subjective experience exist?

Before we get to it — one more story. About a man who, like Boltzmann, saw more than he was supposed to.

John Nash — a mathematician from Princeton, the one from A Beautiful Mind. At twenty-one — game theory, the Nash equilibrium, work that half a century later would win him the Nobel. At thirty — a diagnosis: paranoid schizophrenia, with the whole symptom set included — voices, patterns in newspaper ads, messages from outer space in license plate numbers — after which twenty years of his life go by in a fog.

The standard story: a genius broke down. A brain which too well found regularities began finding them where there were none — pareidolia, only worse: no longer faces in clouds, but messages in random numbers. Psychiatrists call this apophenia: the tendency to see connections between things that aren’t connected. The filter that ordinarily screens out meaningless coincidences in Nash’s case collapsed: the difference between a loosened filter and a collapsed one is the difference between a slightly open door and a wall through which a bulldozer has just driven.

But here is what won’t let me go. Nash saw patterns everywhere — in numbers, in people, in newspapers: ninety percent of them were hallucination, and the remaining ten were the Nash equilibrium, a theory that describes the behavior of any system of interacting agents from nuclear deterrence to evolution, and that works and turns out to be correct.

One and the same brain with one and the same capacity to see patterns: in ten percent of cases it produced genius, in ninety, madness. And nobody, including Nash himself, could tell one from the other from the inside, because from the inside both pattern and hallucination are equally real.

I sometimes think: what if the difference between Nash and me is just which ten percent turned out to be right? He saw patterns in numbers, and some of them were equilibria; I see patterns in reality, and some of them, maybe, are the structure of an information graph. Or perhaps all of them are hallucination, and my model is the same “schizophrenia” as messages from space in license plates.

How do you tell? By data. Nash told: his formulas worked — and if they worked, that meant not a hallucination. 171 galaxies, ${\chi }^2=0.77$ — so far my formulas also work, with a large reservation about that “so far”: tomorrow they may stop, and then this whole book moves in one step to the cupboard with the messages from license plate numbers.

Why is there something it’s like to see red, why does pain hurt, and why is there anyone at all to whom all this information is accessible?

You can build a robot that responds to red light, distinguishes it from green, and says: “I see red.” But what is it like for the robot to see that red — anything at all, or is there nothing in there, the way there’s nothing in a fridge that responds to temperature without feeling anything?

Chalmers showed: from a description of the mechanism, no description of the experience follows. You can fully describe which neurons switch on, which chemicals are passed between them, which brain regions are involved — and from that description it in no way follows that anyone is feeling anything. An objective description does not generate subjectivity. The map is not the territory. No matter how detailed the map.

And at this point there is a crack in science, right at its foundation. Science works with the objective; consciousness is subjective by definition; and between these two banks there is no bridge — only a chasm.