Slime mould intelligence sparks scientific battle over cognition's definition

Forty years ago, a Japanese undergraduate named Toshiyuki Nakagaki noticed something moving with unusual purpose inside a petri dish. It was a lemon-yellow organism called Physarum polycephalum—a slime mould, a single cell containing many nuclei, capable of traveling at a centimetre per hour. That observation would eventually force the scientific world to confront a question it had considered settled: what, exactly, is intelligence?

For centuries, the answer seemed obvious. Intelligence required a brain. It required neurons, the cells that form connections to create thought and memory. This understanding placed animals in a special category within the living world, and humans at the apex. Slime moulds, by contrast, were dismissed as simple masses of protoplasm, organisms that fed and reproduced and little else. But Nakagaki began to watch Physarum more carefully. The mould sensed its food—bacteria, microbes, oat flakes in laboratory conditions—and reached toward it with thin fingers of protoplasm. It appeared to deliberate. It moved with what looked like intention.

In 2000, Nakagaki published an experiment showing that Physarum could solve a maze, finding the shortest route to food. The finding made headlines. A decade later, his most famous work appeared: he arranged oat flakes to match the map of Tokyo and its 35 satellite towns, then released a Physarum from the centre. The mould spread outward, linking the food sources with tubules, and when it finished, the network it had created nearly matched the design of Tokyo's actual railway system. The organism had optimized a transportation network without ever seeing a map, without a brain, without anything resembling conscious thought. Nakagaki won the Ig Nobel Prize twice—once for cognitive science, once for transportation planning. He began to speak of rewriting the dictionary.

But this is where the scientific world fractured. Over the past four decades, a growing number of researchers have argued that cognition and intelligence are not unique to animals with brains. They point to plants that appear to communicate through underground fungal networks. They cite experiments suggesting that individual cells can form memories. They argue that slime moulds, bacteria, and even molecular networks display genuine intelligence. This camp—which some call the "biogenic" school—has challenged the foundational assumption that separated humans and animals from the rest of life. The response from the old guard has been fierce. Lincoln Taiz, a plant physiologist, dismissed these theorists as suffering from "brain envy." In 2007, dozens of scientists signed a declaration in a scientific journal insisting that plants were not intelligent and that "plant neurobiology" was pseudoscience. One signatory called the letter "the last serious confrontation between the scientific community and the nuthouse."

The battle has grown personal. Michael Levin, a Tufts University professor and perhaps the most polarizing figure in the biogenic camp, claims that every individual cell possesses a form of cognition. For this, he receives hate mail calling him "dangerously insane." When Pamela Lyon, a philosopher at Adelaide University, co-edited a special issue on cognition for the Royal Society's Philosophical Transactions in 2021, one reviewer called it "the greatest thing since sliced bread." Another said it was "dangerous" and that the journal should not dignify such ideas. The disagreement runs deeper than scientific methodology. It touches on how humans understand their place in nature, whether intelligence is something humans possess uniquely or something woven through all living things.

The core dispute hinges on language itself. Alex Kacelnik, a zoologist at Oxford who has studied tool-using crows, argues that stretching the word "cognition" to describe brainless organisms "hollows out the word itself." When people speak of cognitive therapy, they mean something fundamentally different from what an oak tree does when it decides how many acorns to produce. To call both cognition, he contends, is to trivialize the term and obscure rather than illuminate. Yet Michael Levin counters that refusing to apply the same language across life forms prevents scientists from recognizing genuine similarities. He points out that physics and chemistry underlie all intelligent behaviour, including that of human brains. The question becomes not whether slime moulds are intelligent in the way humans are, but whether intelligence itself exists on a spectrum rather than as a binary property that some organisms possess and others lack.

The experiments that fuel this debate are genuinely strange. David Glanzman, a UCLA professor, has shown that memories in sea slugs may not reside in the connections between neurons as long believed, but in the cells themselves. When he injected RNA from trained sea slugs into untrained ones, the untrained slugs displayed the same learned behaviour—as though memory could be transferred like data. The finding was so controversial that it cost him his NIH funding. His colleagues simply would not accept it. Yet if Glanzman is right, if memory can exist in individual cells without synapses, then the boundary between cognitive and non-cognitive life dissolves. Slime moulds, bacteria, plants—all might possess forms of learning and memory that we have simply failed to recognize because we were looking for brains.

What makes this debate so intractable is that humans have repeatedly redefined intelligence to suit their current understanding of the world. Descartes believed all non-human animals were automata. Darwin saw intelligence in earthworms. The rise of computers led scientists to model minds as information processors. Then robots appeared that could navigate without building internal models at all. Each era has reshaped the definition, and each time, humans have placed themselves at the centre. Now, as artificial intelligence grows more sophisticated, the pressure intensifies. Silicon Valley embraces the biogenic view because it validates the idea that intelligence is fundamentally a problem of computation and data—which would make machines intelligent too. But this creates a strange loop: treating machines as intelligent because they process information like brains, while simultaneously arguing that brains are just machines. The question of what intelligence actually is remains unanswered.

In the end, the slime mould forces us to confront the inadequacy of human language itself. When 174 scientists were asked to define "behaviour," they produced more than 25 distinct definitions and 100 edge cases. We have three crayons and are trying to recreate the Pantone catalogue. Levin argues that scientists have a responsibility to lead the use of language, to offer better terms even as they challenge old ones. But until we can agree on what we mean by cognition, intelligence, learning, and memory, we cannot settle whether a brainless yellow blob moving through a maze is solving a problem or simply following the laws of physics and chemistry. What we can say is this: the question itself has changed us. We can no longer assume that intelligence is ours alone, or that the living world divides neatly into the thinking and the thoughtless. Whether that realization leads to genuine insight or merely to confusion remains to be seen.