Hawkins: A Thousand Brains
Summary: Jeff Hawkins’s theory that the neocortex works through thousands of reference-frame-based cortical columns — each learning a model of the world — and its implications for machine intelligence, consciousness, and humanity’s choice between gene-driven and knowledge-driven futures.
Sources: Clippings/Hawkins-A-Thousand-Brains.md
Source pages: A Thousand Brains: A New Theory of Intelligence — Quotes
Quote pages: Q01, Q02, Q03, Q04, Q05, Q06, Q07, Q08, Q09, Q10, Q11, Q12, Q13, Q14, Q15, Q16, Q17, Q18, Q19
Last updated: 2026-05-07
The Thousand Brains Theory
The neocortex contains roughly 150,000 cortical columns. Each column independently builds a predictive model of the world using reference frames — a mechanism borrowed from the old brain’s place cells. “Knowledge of something is distributed in thousands of columns, but these are a small subset of all the columns.” Intelligence emerges from the voting and consensus across these models, not from any single place.
We learn by moving
“we learn by moving. In order to learn a model of a building, we must walk through it, going from room to room.”
Knowledge isn’t passively received; it requires active engagement with the world. The motor and sensory systems are inseparable. This has direct implications for AI design — systems that can only observe, not interact, are limited in what models they can build.
Reality is a brain construction
“Everything we do perceive must be fabricated in the brain. Even the most basic feelings of light, sound, and touch are creations of the brain; they only exist in its model of the world.” “Our reality is similar to the brain-in-a-vat hypothesis; we live in a simulated world, but it is not in a computer — it is in our head.”
The brain doesn’t receive reality; it constructs it. The nerves send only spikes; the brain interprets them as color, warmth, pain. See narrative-bias: the brain’s tendency to construct coherent stories from incomplete data is not a bug — it is the fundamental architecture of perception.
Old brain vs. neocortex
“In the battle between the old brain and the neocortex, the old brain usually wins. We eat the cake.”
The neocortex can model long-term consequences; the limbic system and other ancient structures evolved to satisfy immediate needs. The result: a species capable of reasoning about the heat death of the universe but unable to reliably resist a dessert trolley. Compare sapolsky-behave: the frontal cortex can override impulses but often loses to subcortical systems.
False beliefs as viral
“how all of us are susceptible to false beliefs and how some beliefs are viral.”
Beliefs that spread easily are selected by social transmission, not by truth value. Viral beliefs exploit the same old-brain mechanisms that make us susceptible to social pressure and in-group conformity. See in-group-out-group and narrative-bias.
Intelligence vs. self-replication as existential risks
“The point is that anything capable of self-replication, especially viruses and bacteria, is a potential existential threat. Intelligence, on its own, is not.”
Hawkins inverts the AI-doom narrative: intelligence doesn’t create the drive to harm — goals do, and goals come from the old brain, not the neocortex. A purely intelligent system without the goal-generating apparatus of the old brain wouldn’t “want” anything.
Genes vs. knowledge as civilizational drivers
“The choice we face as we think about the future is, should we continue to be driven by our biological past or choose instead to embrace our newly emerged intelligence?”
Hawkins frames the central question of civilization: gene-driven behavior is short-sighted and zero-sum (survival, replication); knowledge-driven behavior can be long-sighted and positive-sum. Compare free-will-and-moral-responsibility: whether we can override biological programming is the same question in a different register.