Skip to content
NatureHQ

Claim check

Are slime moulds intelligent?

Exaggerated

Something true sits underneath, stretched well past what was shown.

The experiments are real and the word is doing too much. Physarum finds the short path through a maze because tubes carrying more flow thicken and the rest are reabsorbed — no representation of the maze, no comparison, no choice. It genuinely does habituate, which is real learning of the simplest kind. Neither result needs, or supports, intelligence.

The claim as it circulates

“A slime mould is a single cell with no brain that can solve mazes, remember things and design efficient rail networks — a kind of intelligence without neurons.”

Where you may have met it: Science journalism, where the maze result is a perennial favourite; Talks and books about non-neural cognition and "basal cognition"; Design and complexity writing, where Physarum is used as a network-optimisation metaphor

Taking it apart

What was claimed
That a slime mould exhibits intelligence — solving problems, remembering, and optimising networks — despite having no nervous system.
What was actually observed
A plasmodium spread through a maze and offered food at two exits withdraws from dead ends and retains a tube along the shortest connecting path within hours. Separately, plasmodia required to cross a bridge containing quinine or caffeine cross progressively faster over several days; the change is specific to the substance experienced, does not transfer to the other one, and is lost after two days without exposure — the formal criteria for habituation.
What the evidence supports
That a minimum-length network can be produced by local flow reinforcement with no central control. That habituation, the simplest form of learning, occurs in an organism with no neurons. Both are genuinely surprising results.
What it does not support
Problem-solving in any general sense: the organism starts by occupying the whole maze, so it prunes rather than searches, and a two-exit maze is one task rather than a capability. Memory in the everyday sense of recalling a particular event. Design or optimisation as goals — the same short-path outcome arises in purely physical flow networks with no organism involved.

The strongest evidence here is the one that gets the least attention. The maze is photogenic; the habituation experiment is the one that tests its own conclusion, with a stimulus-switch control and a recovery test built in specifically to rule out the boring explanations. It is the better result and it is almost never the one in the headline.

None of this makes slime moulds less interesting. An organism with no nervous system that reliably arrives at short paths tells you something about how little machinery a good solution requires — which is a more useful thing to know than that something small is clever. "Intelligence" ends the enquiry by supplying a familiar answer; the mechanism keeps it open.

Go deeper

The claims underneath

Each one carries its own evidence, scope and caveats. Expand any of them to reach the studies.

A slime mould solves a maze by pruning, not by working out a route

Established

Specialists would state this without hedging. Multiple independent lines of evidence agree.

A plasmodium of Physarum polycephalum spread through a maze and presented with food at two exits retains a thick connecting tube along the shortest path and withdraws from dead ends. The mechanism is local reinforcement: tubes carrying greater cytoplasmic flow thicken while those carrying less are reabsorbed, converging on short connections without any representation of the maze.

Who this applies to
one species in laboratory mazes with two food sources
Studied in
Physarum polycephalum

You may have heard

“Slime moulds are intelligent and can solve mazes”

The maze result is real and the explanation is more interesting than intelligence. Nothing represents the maze and nothing decides; thick tubes get thicker because more is flowing through them. It shows how little machinery a good solution needs, which is a stronger finding than a small brain would be.

Why we rate it this way, and what the caveats are
EstablishedHigh confidence

The result is straightforwardly reproducible and the flow-reinforcement mechanism reproduces it in physical and computational models.

Caveats

  • The organism begins occupying the whole maze, so it prunes rather than searches.
  • A two-exit maze is not a general problem-solving assay.
  • Comparable short-path solutions arise in purely physical flow networks.

Still unanswered

  • How closely does tube reinforcement match the optimisation rules used to model it?

Last reviewed 2026-08-10

The evidence (1 study)

A single cell with no nervous system can learn to ignore something unpleasant

Well supported

Good evidence backs this, though some details remain open.

Plasmodia of Physarum polycephalum required to cross a bridge containing quinine or caffeine crossed progressively faster over several days until reaching control speed. The change was specific to the substance experienced, did not transfer to a different bitter compound, and was lost after two days without exposure — satisfying the diagnostic criteria for habituation rather than sensory adaptation or fatigue.

Who this applies to
one species, two bitter compounds, laboratory conditions
Studied in
Physarum polycephalum

You may have heard

“Slime moulds have memory”

They meet the behavioural definition of habituation, which is real learning and the most minimal kind there is. Memory in the sense people mean it — recalling a particular thing that happened — is not what was shown, and the experiment is careful about exactly this.

Why we rate it this way, and what the caveats are
Well supportedModerate confidence

A well-designed experiment testing the formal criteria for habituation, including stimulus specificity and spontaneous recovery. Confidence is moderate pending independent replication in other species.

Caveats

  • Habituation is the simplest form of learning and implies nothing about higher cognition.
  • One species and a single class of stimulus.
  • The cellular mechanism is unidentified.

Still unanswered

  • What inside the cell stores the change, given there is no nervous system to hold it?

Last reviewed 2026-08-10

The evidence (1 study)

Go deeper

Slime moulds

One cell. Millions of nuclei. No nervous system. It still finds the short way through a maze.

Last reviewed 2026-08-10