In 2000, a biologist named Toshiyuki Nakagaki at Hokkaido University in Japan placed a slime mould — a single-celled organism called Physarum polycephalum — at the entrance of a maze. He put food at the exit. Within hours, the organism had extended itself through the maze, filling every available path. Then it did something unexpected: it retracted from the dead ends and redundant routes, leaving behind a single tube connecting the entrance to the exit along the shortest possible path.
The slime mould had solved the maze. It had no brain. It had no nervous system. It had no neurons at all. It was one cell.
Nakagaki's experiment was published in Nature and became one of the most discussed biology papers of the decade. But the team wasn't finished. In a follow-up study, they scattered food sources across a surface in a pattern that matched the major cities of greater Tokyo, then introduced the slime mould. Over the course of several days, the organism built a network connecting all the food sources. When the researchers compared the result to the actual Tokyo rail system — a network designed over decades by professional engineers — the slime mould's solution was comparable in efficiency, and in some respects superior.
A single cell, with no central processing unit of any kind, had independently produced a transportation network that rivalled one designed by human intelligence.
What the slime mould knows
The initial response to Nakagaki's findings was fascination mixed with unease. A slime mould solving a maze was a good story. A slime mould outperforming human engineers was harder to process.
Further research made it harder still. In 2021, a team at the Max Planck Institute discovered that Physarum doesn't just solve problems — it remembers. The organism weaves memories of food encounters directly into the architecture of its physical body, softening certain tubes and strengthening others to create a structural record of where food has been found. When faced with a future decision, it consults this record — choosing paths that lead toward previously successful locations and avoiding areas associated with danger.
This is memory without a brain. Decision-making without neurons. Learning without a nervous system. And it's not confined to slime moulds.
Intelligence before neurons
The first neurons appeared on Earth roughly 600 million years ago, in the earliest animals with nervous systems — simple creatures like jellyfish and hydra. Before that, for nearly three billion years, life existed and thrived without a single nerve cell.
And it wasn't passive. Bacteria, the oldest life forms, coordinate behaviour across colonies using electrical signals — potassium ion waves that propagate between cells in patterns that, as a 2015 study in Nature demonstrated, bear a striking resemblance to the neural signalling in human brains. The bacterium Bacillus subtilis doesn't just survive. It communicates. It cooperates. It makes collective decisions about resource allocation, and it does so using a signalling mechanism that would not look entirely unfamiliar to a neuroscientist.
Plants, despite having no nervous system, make decisions that meet any functional definition of intelligence. The root tips of a growing plant evaluate soil conditions — moisture, nutrients, toxicity, obstacles — and adjust the direction of growth accordingly. Charles Darwin, in a less-celebrated but remarkable passage in his 1880 work The Power of Movement in Plants, proposed that the root tip acts as "the brain of one of the lower animals," directing movement and responding to stimuli with what he considered a rudimentary form of intelligence.
Darwin's suggestion was largely ignored for over a century. Recent research in plant neurobiology — a field that didn't exist twenty years ago — has vindicated him. Plants use electrical signalling, chemical communication, and hydraulic pressure to process information and coordinate responses across their bodies. They respond to sounds. They recognise kin. They adjust their behaviour based on experience. They do all of this without a single neuron.
What intelligence actually is
The standard story goes like this: intelligence is a product of brains. Brains evolved to help organisms navigate complex environments. The more complex the brain, the more intelligent the organism. Humans have the most complex brains, so humans are the most intelligent. And artificial intelligence, modelled on the brain's computational architecture, is our attempt to build intelligence from scratch.
This story is clean, intuitive, and almost certainly wrong — or at the very least, incomplete.
If intelligence is defined by what it does — solving problems, making decisions, learning from experience, adapting to conditions — then intelligence didn't begin with brains. It began with cells. Single-celled organisms have been solving problems and adapting to environments for 3.5 billion years. They were intelligent long before the first neuron fired.
Antonio Damasio, the neuroscientist at the University of Southern California whose work has reshaped the understanding of the relationship between body and mind, has been making this case for decades. In Descartes' Error, he wrote: "Nature appears to have built the apparatus of rationality not just on top of the apparatus of biological regulation, but also from it and with it." Rationality — the thing we consider the pinnacle of intelligence — isn't a separate system bolted onto the body. It grew out of the body's own regulatory processes. It's an extension of the same intelligence that keeps a cell alive.
In his most recent book, Natural Intelligence and the Logic of Consciousness, published in 2026, Damasio goes further. He argues that consciousness itself is a natural solution to the problem of homeostatic regulation — the body's need to maintain its internal conditions within liveable parameters. Feelings — hunger, thirst, pain, pleasure — are not decorations on top of cognition. They are the foundation. The first glimmers of subjectivity arose not in complex brains but in the basic act of a living system monitoring its own state and caring about the result.
The copy and the original
This is where the AI conversation becomes interesting — and where it takes a turn that the AI industry would rather not dwell on.
We built artificial intelligence by studying the brain. The architecture of neural networks — the layers, the weights, the activation functions — is modelled on a simplified version of how neurons connect and fire. The entire field of machine learning is, at root, an attempt to replicate what the brain does.
But the brain is not where intelligence began. The brain is the most recent, most visible, most complex expression of an intelligence that has been operating in living systems for billions of years. Neurons are the latest chapter in a very long book.
By modelling AI on the brain, we copied the most recent expression of intelligence and treated it as the source. We looked at the crown of the tree and assumed we'd found the root.
This doesn't mean AI is useless or unimpressive. It is, by any measure, an extraordinary achievement. But it does mean that the kind of intelligence AI embodies — pattern recognition, data processing, optimisation — is a narrow slice of what intelligence actually is. The broader phenomenon — the capacity of living systems to sense their environment, respond adaptively, learn from experience, and maintain themselves in the face of change — predates computation by billions of years and operates by entirely different principles.
A slime mould doesn't compute the shortest path through a maze. It grows into the solution. A plant root doesn't calculate the optimal direction. It feels its way. A bacterial colony doesn't process data about resource scarcity. It communicates through chemical and electrical signals that emerge from the physical interactions of its members.
These are not lesser forms of intelligence. They are older forms. And they operate without anything resembling a computer.
What we overlooked
There's a reason this matters beyond academic interest.
The dominant metaphor for intelligence in the modern world is the computer. We speak of the brain as "processing" information, of memory as "storage," of learning as "programming." This metaphor is so pervasive that it feels like description rather than metaphor. Of course the brain is a computer. What else would it be?
But the metaphor reverses the actual history. Computers were modelled on brains, not the other way around. And brains were built on cellular intelligence, not the other way around. The computational model of intelligence is a metaphor that has been mistaken for a fact — and that mistake has shaped not only how we build machines, but how we understand ourselves.
If you believe intelligence is computation, then the logical response to AI is anxiety. A machine that computes faster than you is more intelligent than you. Your relevance depends on your computational advantage, and that advantage is shrinking.
But if intelligence is something older, broader, and more deeply rooted in the body and in life itself — if it includes the felt sense of being alive, the capacity to care about one's own state, the ability to grow into a solution rather than calculate it — then the picture changes. Not because AI becomes less impressive, but because the human capacity it's being compared to turns out to be far larger than the part of it that machines can replicate.
The slime mould doesn't worry about being replaced by a computer. It has a different kind of intelligence — one that doesn't compute but grows, doesn't process but feels its way, doesn't optimise but adapts. It's been doing this for longer than brains have existed.
The intelligence you carry in your body is not a machine. It was never a machine. It's something older than machines, older than brains, older than any computational metaphor we've imposed on it. And it's still running — in your cells, in your gut, in the way your hand reaches for something before your mind decides to — while the part of you that worries about AI is, ironically, the newest and most fragile layer of a very ancient system.
Sources and further reading:
- Nakagaki, Yamada & Tóth, "Maze-solving by an amoeboid organism," Nature (2000) — the original slime mould maze experiment
- Tero et al., "Rules for Biologically Inspired Adaptive Network Design," Science (2010) — slime mould recreating the Tokyo rail system
- Kramar & Alim, "Encoding memory in tube diameter hierarchy of living flow network," Proceedings of the National Academy of Sciences (2021) — memory without neurons in Physarum
- Prindle et al., "Ion channels enable electrical communication in bacterial communities," Nature (2015) — bacterial signalling resembling neural dynamics
- Charles Darwin, "The Power of Movement in Plants" (1880) — on root tips as rudimentary brains
- Antonio Damasio, "Descartes' Error" (1994) — on rationality built from biological regulation
- Antonio Damasio, "Natural Intelligence and the Logic of Consciousness" (2026) — on consciousness as homeostatic regulation