The Slime Mold That Can Solve Complex Labyrinths Without a Brain

Seeing a blob of yellow goo act as though it knows what it’s doing has a slightly ridiculous quality. When you first see it, you think there must be a trick—some secret system, a researcher pushing things along. It doesn’t exist. When given a tiny glass maze with a piece of oatmeal at one end, the slime mold just works it out. Within a day or two, it finds the shortest route between two food sources after oozing into every dead end, hesitating, and retracting. No neurons. Not a nervous system. As far as anyone can tell, there are no instructions.

Physarum polycephalum has long piqued the interest of biologists, but it wasn’t until a Japanese team demonstrated that it could replicate the layout of the Tokyo rail system in the early 2000s that the general public began to take notice. If you put oat flakes on a map of the locations of major stations, the slime mold will, given enough time, create something uncannily similar to what human engineers spent years creating. This is the kind of outcome that seems manufactured. The part that unnerves people is, of course, the time-lapse footage, which gives the impression that you are witnessing intention without a thinker.

In actuality, it grows everywhere before pruning. Once it finds food, the redundant branches of the mold collapse back into themselves, leaving a trail of slime to indicate areas it has already explored. The mold grows outward in a sprawling, disorganized network. Don’t bother returning here; the slime functions almost like a chemical memo. Wallace Marshall, a researcher at the Marine Biological Laboratory in Woods Hole, believes the organism senses flow resistance through its tube-like body to determine the length of its own branches. More resistance with a longer tube. It’s a form of cytoplasmic arithmetic.

Even stranger is the more recent work, most of which comes from Tufts and the Wyss Institute. Researchers demonstrated in 2021 that Physarum could decide where to grow based only on mechanical cues, such as the slight strain patterns created by glass discs resting on a flexible gel, without the need for food or chemical signals. The mold moved steadily in the direction of heavier objects by pulling on its surroundings and somehow sensing the deformations. However, it was unable to distinguish between the discs when they were stacked. The variable was not mass. There was a pattern. It’s difficult to ignore how much that sounds like perception.

More than 64 million people have watched slime molds on TikTok, which is a cultural phenomenon in and of itself. They are kept as pets by the French. One received an exhibit from a zoo in Paris. At least one of them has been sent to the International Space Station, and researchers have used them to propose more effective road systems for the United States. They are adored by the biomimicry community, in part because they have withstood multiple mass extinctions while more ostentatious organisms did not, and in part because they continue to yield results that no existing model can adequately account for.

It’s still unclear if any of this qualifies as intelligence or if intelligence is just the incorrect term, taken from animals with brains and awkwardly applied to something that existed before brains at all. According to what I’ve read, the majority of scientists are at ease with ambiguity. They refer to it as simply computation or proto-cognition. What we call it doesn’t matter to the mold. It continues to solve its puzzles, get rid of its slime, and subtly argue that thinking—whatever it is—might not need what we always thought it did.