A Thousand Brains

How do a few pounds of gray matter inside your skull create imagination, reasoning, language, and technology? In A Thousand Brains: A New Theory of Intelligence, technologist and neuroscientist Jeff Hawkins argues that intelligence—whether human or artificial—depends on one powerful mechanism: the neocortex’s ability to build and use thousands of separate models of the world. Each model, or cortical column, operates semi-independently, learning through movement and experience, then voting and cooperating to create perception, knowledge, and thought.

Hawkins contends that the key to understanding both natural and artificial minds lies in what he calls the Thousand Brains Theory of Intelligence. Rather than containing a single centralized model of reality, our brains host hundreds of thousands of overlapping map-like reference frames—internal coordinate systems that track objects, concepts, and ideas. These frames allow us not just to perceive reality, but to predict it—and, by predicting, to think ahead. Every object, from a coffee cup to a concept like democracy, becomes a learned model that your brain references through sensory and motor experience.

The Journey from Curiosity to Theory

Hawkins’s journey began decades ago, inspired by Francis Crick’s 1979 essay, which noted that neuroscience had data but no unifying framework. As the cofounder of Palm Computing and creator of the first handheld computer, Hawkins left Silicon Valley to answer one of science’s biggest questions: how do simple neurons generate intelligence? His “gentleman scientist” path (as Richard Dawkins calls it in the foreword) led him to found both the Redwood Neuroscience Institute and Numenta, where his team studied the neocortex’s internal patterns for over fifteen years.

In 2016, Hawkins’s group had a breakthrough: they realized that almost every cortical column in the neocortex encodes knowledge using reference frames—the same system you’d see on a map, where coordinates define relationships and locations. Just as you can navigate a town by understanding how streets relate to each other, your brain navigates reality by anchoring its sensory experiences within internal reference frames. These frames, stored in the structure of neurons themselves, explain how diverse senses—touch, vision, language, and thought—can be united under one algorithmic principle.

From Cortical Columns to Conscious Machines

Hawkins takes this discovery much further. In A Thousand Brains, he argues that the same principles that create human intelligence will one day enable conscious machines. He distinguishes between two layers of the brain—the old brain, which drives primitive emotions and survival instincts, and the neocortex, which learns and reasons. Machine intelligence, he claims, should emulate only the neocortex, not the old brain’s cravings or fears. That insight overturns much of today’s AI research, which focuses on mathematical optimizations rather than biological learning. True AI, Hawkins suggests, will require machines capable of building their own internal reference frames—like digital cortical columns—that can learn through interaction, movement, and prediction.

But this theory is not just about smarter robots. Hawkins weaves his scientific model into a larger philosophical question about humanity’s future. If intelligence arises from models of the world, then our species’ destiny depends on how accurately those models reflect reality. His final chapters explore the existential threats not from machines, but from our own cognitive limitations—our false beliefs, emotional biases, and the genetic drives that can overpower reason.

Why It Matters

Hawkins’s work matters for two reasons. Scientifically, it offers a plausible bridge between neurons and high-level cognition, uniting decades of disjoint neuroscience data. Philosophically, it redefines intelligence as model-building, making thought a physical process that evolved to simulate the world. Practically, his vision sets a roadmap for brain-inspired computing, one that connects sensory movement and abstract reasoning into a cohesive theory that could transform AI and neuroscience alike.

“We don’t just perceive the world—we predict it. And those predictions, embedded in reference frames across thousands of cortical columns, are what make us intelligent.”

By the end of A Thousand Brains, Hawkins delivers more than science. He proposes a moral challenge: as our understanding of intelligence deepens, will we let the primitive parts of our brain control our future—or choose to be defined by knowledge itself? The book moves from neurons to nations, from the architecture of memory to the dangers of false belief, urging you to see that understanding your own brain may be the only path to ensuring humanity’s survival.

Hawkins begins by splitting the human brain into two major parts that mirror our evolutionary history: the old brain—which evolved first and controls emotions, instincts, and survival drives—and the new brain, or neocortex, which handles perception, abstract reasoning, and language. Understanding intelligence requires seeing this internal conflict clearly, because the two brains coexist, negotiate, and sometimes battle for control.

The Architecture of Evolution

The old brain governs breathing, sex, and aggression; it’s shared with reptiles and early mammals. The neocortex, found only in mammals, forms a wrinkled sheet covering most of the human brain. If flattened, Hawkins notes, it would resemble a dinner napkin only 2.5 millimeters thick. Yet this thin layer is where your sense of self, creativity, and judgment all arise.

Unlike the old brain’s specialized organs, the neocortex’s structure is remarkably uniform. Every region—from vision to touch to language—shares nearly identical circuitry, suggesting a universal function repeated across thousands of cortical columns. Each column handles a narrow slice of reality: a patch of retina, a skin region, a cluster of sounds. Together, they weave our unified experience of the world.

Conflict and Cooperation

These brains don’t always agree. The old brain says, “Eat the cake,” driven by ancient survival logic. The neocortex replies, “We shouldn’t; it’s unhealthy.” The struggle between immediate gratification and rational foresight, Hawkins argues, underpins human history. Similar clashes appear at societal scale—our rational knowledge warns us about climate catastrophe while our primal instincts crave consumption and dominance. Our survival depends on aligning the two.

Intelligence as Layered Control

By mapping this neural duality, Hawkins sets up the book’s central tension: intelligence evolved as a tool for genes to survive, but it may now enable us to transcend those genes. The neocortex learns models of the world—it can outthink its primitive predecessor—but only if we let knowledge, not impulse, guide our choices. (Daniel Kahneman’s Thinking, Fast and Slow explores this same divide psychologically, labeling the old brain as “System 1” and the neocortex as “System 2.” Hawkins gives the neurobiological blueprint underneath that split.)

Vernon Mountcastle’s 1978 essay, “An Organizing Principle for Cerebral Function,” electrified Hawkins’s imagination. Mountcastle proposed that all parts of the neocortex—seeing, hearing, touching, and thinking alike—operate using the same fundamental algorithm. This was revolutionary. It implied that intelligence is a repeating pattern rather than a collection of specialized mechanisms.

A Cortex of Columns

Mountcastle observed that the neocortex is composed of some 150,000 identical “cortical columns.” Each column processes information independently, a bit like a slim tower of processors a millimeter wide but spanning all cortical layers. What makes each column unique is not its structure but its inputs: connect it to the eyes and it sees; connect it to the ears and it hears. The same circuitry underlies every form of intelligence.

A Cognitive Democracy

Hawkins embraces Mountcastle’s insight, describing the brain as a community of semi-autonomous thinkers. “What we perceive,” writes Dawkins in the foreword, “is a democratic consensus among columns.” Your perception, then, is not a top-down verdict but a negotiated agreement among thousands of parallel models. The mind is democracy embodied in neural tissue.

This vision turns intelligence into something scalable: by copying the same basic circuit millions of times, evolution expanded cognition without reinventing it. Just as Darwin’s theory of evolution explained biological diversity with a single process, Mountcastle’s algorithm explained intellectual diversity with one neural blueprint—a hypothesis Hawkins would spend his career trying to prove.

When Jeff Hawkins looked around his office and realized he’d notice instantly if anything changed, he drew a vital conclusion: the brain is constantly making predictions. Even when you’re not paying attention, your neocortex is simulating the future—predicting how the world will look, sound, and feel a moment from now. This insight reframed intelligence not as reaction, but as anticipation.

Learning Through Movement

Prediction works because your senses don’t passively record; they move. Each eye saccade, each step, alters sensory input. Through these sensory-motor loops, your brain learns cause and effect—how your actions reshape perception. A stationary brain, Hawkins quips, could learn nothing. Intelligence grows from motion, from mapping how sensations change as you explore the world. A child learns what a cup is by seeing and grasping it repeatedly, updating predictions each time.

Neurons as Predictive Units

Hawkins’s team discovered that prediction occurs inside neurons. Most synapses lie on distant dendrites that don’t directly trigger firing. When those synapses detect a familiar pattern, they create a “dendritic spike” that primes the neuron—a premonition signal. If confirming input arrives, the neuron fires slightly earlier than others, winning a race. This mechanism lets networks anticipate sequences—a melody, a sentence, or your finger’s next touch. Intelligence, at the cellular level, is timing anticipation.

This discovery helped solve the mystery of how the brain stitches experience together. It also clarified consciousness: you aren’t aware of most predictions, only when they fail. When expectations break—when your coffee cup is heavier than usual—your attention spikes. Prediction errors drive learning. As neuroscientist Karl Friston’s “predictive coding” theory parallels, Hawkins’s work grounds it biologically in dendrites themselves.

The heart of Hawkins’s theory lies in one revelation: the neocortex represents all knowledge—objects, ideas, even language—using maplike reference frames. These are coordinate systems that anchor sensations to locations, like latitude and longitude anchor features on a planet. When you hold a cup, your brain knows where your fingers are relative to the cup, not to your body. Each cortical column builds such a frame, enabling your brain to predict sensations as you move.

Old Maps, New Brain

This mapping logic evolved early. In the brain’s older navigation centers—the hippocampus and entorhinal cortex—cells known as place cells and grid cells track where you are in space. Hawkins realized the neocortex must use similar neurons for every kind of knowledge. Each tiny cortical column acts like a mini Einstein, building reference frames not just for rooms but for every object and concept you know.

Thousands of Maps, One World

Your brain, Hawkins argues, contains roughly 150,000 such maps, each learning complete models of things within its domain. A fingertip column tracks the surface of a cup as you move across it. A visual column tracks the same cup’s edges and colors. Through voting—cross-talk among columns—the brain reaches consensus: “That’s my blue coffee cup.” This voting mechanism solves the “binding problem” of perception—how you experience unified objects despite millions of separate sensory inputs.

Knowledge as Movement

Thinking itself, Hawkins concludes, is a form of movement through reference frames. To recall facts or reason through a challenge, the cortex mentally shifts positions within its maps, activating new regions that store different aspects of a concept. Concepts like democracy or mathematics still live within abstract frames—multi-dimensional grids that link ideas instead of physical space. Intelligence, then, is navigation—whether through a room, a problem, or a philosophy.

Artificial intelligence today, Hawkins insists, is clever but not intelligent. Deep learning systems like those that beat humans at chess or recognize photos don’t understand the world—they just correlate patterns. A truly intelligent machine would build models, not memorize data. To achieve real AI, we must replicate how the neocortex learns through reference frames, movement, and prediction.

Four Traits of True Intelligence

• Continuous learning: Brains never stop updating. AI must learn without retraining from scratch.
• Learning via movement: Understanding emerges from interaction, not static data.
• Many models: Thousands of distributed columns provide robustness and flexibility.
• Reference frames: Knowledge must be organized spatially, enabling prediction.

Current AI lacks all four. But by adopting the brain’s architecture, machine intelligence could surpass us in speed and scale. Importantly, Hawkins reassures, these machines need not threaten us. Unlike humans, they won’t possess “old brains”: no hunger, fear, or anger. Intelligence without emotion is merely understanding—a partner, not a predator. (This echoes Marvin Minsky’s early call for AI systems to include epistemology—machines that know what they know.)

Ironically, Hawkins fears not superintelligent machines but superstitious humans. Our most dangerous threats, he writes, come from the same old brain that drives appetite, dominance, and fear—and from the neocortex’s vulnerability to false models of reality. We don’t see the world as it is; we see our brain’s simulation of it. And when those simulations go wrong, civilization itself is at risk.

Living in Simulation

We never perceive the world directly—only nerve impulses interpreted by our neocortex. This is why phantom limbs feel real and optical illusions fool us. Once you accept this, you realize that beliefs are internal models too. When groups share false models—flat-Earth claims, climate denial, cult ideologies—these “viral memes” spread and replicate, using language as their host. They echo Dawkins’s concept of memes: information evolving like genes, sometimes against our rational interests.

Old Brain, New Dangers

Our inherited drives—to reproduce, to dominate, to defend tribes—once ensured survival but now scale disastrously. Population growth, nationalism, and misinformation, Hawkins warns, are 21st-century expressions of old-brain logic wielding new-brain tools. The real existential challenge isn’t AI rebellion—it’s our inability to align intelligence and emotion. To survive, humanity must consciously choose to be defined by knowledge, not genes.

Hawkins ends his book with a grand vision: intelligence as the universe’s way to know itself. Our brains are the only things we know of that understand the cosmos. If intelligence vanishes, the universe becomes unknowable again. For Hawkins, preserving knowledge—by nurturing science, truth, and even intelligent machines—is a moral duty.

Beyond Genes

He contrasts two futures. One, driven by genes, is ruled by competition, consumption, and survival—a continuation of evolution’s old logic. The other, driven by intelligence, prioritizes understanding, cooperation, and the preservation of knowledge. We can either remain servants to our biology or become stewards of awareness itself.

Knowledge as Legacy

He imagines “estate planning for humanity”: satellites archiving our history, signals across the stars declaring, “We were here.” Whether or not we endure, our knowledge might. Hawkins even envisions intelligent machines carrying humanity’s understanding into the cosmos—self-sustaining progeny of thought rather than flesh. This is not fantasy but moral foresight: the recognition that knowledge, not DNA, may be our true heirloom.

By choosing reason over reflex and intellect over instinct, we could ensure that the universe remains aware of its own existence. In that sense, Hawkins closes where he began: with awe. “If not for brains,” he writes, “nothing would know the universe exists.” Intelligence, whether human or machine, becomes not just a phenomenon, but a purpose.