The Brain That Undoes Everything You Do to It

— With that — nothing. You probably really do feel it. It's a completely different word in your sentence that bothers me.

— "15 billion." Where did you get it?

— Everywhere isn't a source. I went looking for something to anchor that number to and found nothing: "biohacking" isn't a market category at all, nobody counts it as an industry. What has actually been counted is the market for "brain health" supplements: on the order of $3 billion in global sales in 2016, with a forecast of $5.8 billion by 2023. Everything else is forecasts from firms that make their money selling forecasts. I'm starting here not to catch you out on nonsense. I'm starting here because you're about to hear something like 50 numbers from me, and I want you to treat every one of them exactly the way I just treated yours.

— Not with the idea. With the direction. Tell me, what does an improved brain look like to you — physically, if you picture it?

— There. And that's exactly what the whole industry says: grow it, build it up, add. Now look at what your own brain was doing while it was getting better. Synaptic density in the human cortex peaks at around 1–2 years and exceeds the adult level by something like 50 percent. After that — downhill. All development after age 2 is controlled destruction of the surplus.

— Denser. Not simpler, not half-built — denser. And in the frontal cortex, dendritic spine counts in children exceed adult values by a factor of 2–3. An infant's brain isn't a rough draft of the adult one. It's an overloaded system that hasn't yet decided what in it is surplus.

— It was measured by hand. Peter Huttenlocher, a child neurologist in Chicago, took post-mortem human cortex in the late 1970s, made electron micrographs and counted synapses one at a time. One by one. No elegant technology — counting. And it turned out that brain development runs downward, not upward.

— It doesn't know. It builds a deliberately excessive network and then watches what works. Imagine you're not designing a city's road network — instead you first pave the whole city in asphalt, one solid field, anywhere to anywhere — and only then look at where cars actually drive, and break everything else back up into gravel. The speed on the final arteries doesn't come from there being a lot of asphalt. It comes from having removed everything that led the wrong way.

— Literal. And it's the least comfortable part of the story. Synapses aren't destroyed by neurons. They're destroyed by the brain's immune system. Complement proteins, C1q and C3, settle on a weakly working synapse like a tag. And microglia — the resident macrophages inside the skull — follow that tag to the marked synapse and physically engulf it.

— The very same thing. The brain tags its own connections the same way the body tags a bacterium before eating it. Brain development is literally set up as controlled inflammation.

— By activity. When people looked at what microglia had actually swallowed, the engulfed presynaptic terminals contained no mitochondria — a sign the terminal had barely been working. The rule isn't "remove the random one." The rule is "remove the unused one." And there's more than one cleaner: astrocytes do the same thing by their own pathways, and in the adult brain they go on eating synapses continuously. Pruning isn't an episode of childhood that ended. It's the background.

— In the best-known study on this — 307 children and adolescents, scanned over years — the group with the highest intelligence, 121–149 on the tests, had a cortex in early childhood that was precisely thinner than everyone else's.

— So the question was the wrong one. What mattered wasn't thickness but trajectory: their cortex thickened longer and more aggressively, to about age 11, and then thinned faster than anyone's. By late adolescence the thickness in all groups converged. The difference was in the tempo, not the size.

— And you're right to be wary. What comes next is the part that gets cut out of the retellings. That finding still hasn't been reliably replicated. The ages at which the thickness peaks fall are exactly the ages at which children move the most in the scanner, and motion systematically eats cortical thickness in the reconstruction. When in 2016 a similar sample was run through three levels of image quality-control strictness, at the strictest level the quadratic thickness trajectories across ages 5–22 simply stopped showing up.

— It's influential. That isn't the same thing, and you're going to have to live with that distinction for the rest of this conversation — it'll come up in every other topic.

— No. And this is where the whole "fewer connections, sharper brain" story stops being a slogan. There's a variant of a complement gene that raises C4A expression, and it's associated with increased risk of schizophrenia. The proposed mechanism is precisely excessive pruning in adolescence.

— The very same. The difference is the volume. And it's a hypothesis the field is working around, not a proven cause in humans — but it's serious enough to close the conversation about "prune more." This isn't a scale where higher means better. It's a setting where both sides of the optimum are disease, and we don't know where the optimum is.

— An antibody that blocks it exists. A humanized anti-C1q antibody — an off switch for the "eat this synapse" tag. It's being developed as a drug: in the pivotal phase 3 trial in Guillain-Barré syndrome, a dose of 30 mg per kilogram produced a significant improvement on the disability scale by week 8.

— The other way. And there's no accelerating one, and I'm afraid there won't be. To speed it up you'd have to know which connections specifically are surplus in you. Your own brain worked that out by brute force — and not in a couple of years: elimination in the frontal cortex continues throughout the third decade of life and only plateaus closer to 30. A quarter-century of brute force can't be packaged into capsules. Only addition can be sold — because subtraction requires a criterion that no pill has.