Biotech 4 November 2026 4 MIN READ
The Slime Mould and the Tokyo Rail Map
A single cell with no brain grew a network like Tokyo's railways. What the 2010 experiment showed, what press releases added and the algorithm behind it.
THE SHORT ANSWER
A slime mould with no brain grew a network between 36 food sources arranged like the cities around Tokyo. In a 2010 Science paper, its cost and efficiency were comparable to the rail system, though the railway was more fault-tolerant. The lasting result is an algorithm, not a design.
On this page
A single cell with a sense of direction
Physarum polycephalum is a slime mould, and despite the name it is not a fungus. It is a single-celled organism: one huge amoeba-like cell that spreads across a surface as a network of tubes, looking for patches of food. It has no brain and no neurons. Researchers have used it for years to study how simple systems solve complicated problems.
The maze
In 2000 Nakagaki, Yamada and Tóth published a paper in Nature showing that the slime mould could find the shortest route through a maze. According to reports in The Guardian and New Scientist at the time, the cell filled the maze, then withdrew from dead ends and shrank to a single tube along the shortest of four routes between two food sources, after about eight hours. We could not read the paywalled paper, so these details come from contemporary reporting and from the way a later paper cites it.
The Tokyo experiment
The best-known experiment came ten years later. In a paper in Science on 22 January 2010, a team from Hokkaido University, Hiroshima University and Oxford placed oat flakes at 36 positions representing the cities around Tokyo, on a wet surface inside a map-shaped arena. They put a small slime mould on the spot for Tokyo itself and let it grow.
To imitate mountains and the sea, the researchers used light, which the slime mould avoids. So the geography was imposed by the experimenters, not discovered by the organism. After the cell had built its network, they compared it with the real Tokyo rail system on three measures.
- Cost. The rail network’s total length is about 1.8 times that of the shortest possible network that connects every city. The slime mould’s average over 21 runs was 1.75, with a spread of 0.30.
- Efficiency. Travel between points was similarly efficient.
- Fault tolerance. Here the railway was better. Only 4% of single failures would cut off part of the rail network. For the slime mould, the figure was 14% with the light constraint and 20% without.
What the paper does not say
The authors’ own conclusion is careful: the slime mould’s networks had “similar” characteristics in cost, efficiency and fault tolerance. Results varied from run to run, and the match was in aggregate measures, not in an identical map. The university press release went further and used the phrase “nearly identical”, which the paper does not.
Nor did the slime mould design a railway. The real value of the paper is in its second half: a mathematical model of how the tubes thicken where more fluid flows and fade where it does not. This feedback rule produces networks comparable to those of real infrastructure, and it is a rule an engineer can use. The authors point to self-organising networks without a central controller, such as sensor arrays and wireless mesh networks, as the application, not railways.
Another trick: anticipating time
In 2008 the same group published a different result in Physical Review Letters. Slime moulds that were exposed to three dry, cool pulses at regular intervals slowed down at the time the next pulse would have arrived, even when it did not come. The authors interpret this as a kind of anticipation, and they describe their results modestly: they “hint” at cellular origins of primitive intelligence and suggest that simple dynamics may be enough. It is a single-laboratory experiment, and it does not show that slime moulds learn or remember in a general sense.
Why it matters
The slime mould is useful as a reminder. Problem-solving does not need a brain if the physics of the system already contains a good rule, here flow that strengthens the tubes it uses. Engineers have taken that idea and turned it into an algorithm.
Our take: a clever cell, a more useful equation. The Tokyo map made the headlines. The lasting result is a feedback rule that builds cheap, efficient and reasonably robust networks without a plan.
Do you agree?
Take it to Chevlume: write your view, add your sources and let others answer. The closed alpha is by invitation.
Keep reading
Therac-25: When Software Was the Only Safety Net
Six overdoses between 1985 and 1987 from a radiation machine whose safety rested on software. What the Leveson and Turner investigation found.
FUTURE / DISCOVERYKnight Capital: 45 Minutes, $460 Million
A forgotten code path on one of eight servers cost a trading firm $460 million in 45 minutes. What the SEC found about deployment and controls.
SPACE / DISCOVERYThe Voyager Golden Record
What is on the disc launched into interstellar space in 1977, how an alien is meant to play it, how long it may last and the arguments around it.
Sources
- Rules for biologically inspired adaptive network design · Science (via PubMed / NIH National Library of Medicine) · Accessed
- Rules for Biologically Inspired Adaptive Network Design (Science 327:439) · Science / AAAS (PDF copy hosted at University of Vermont) · Accessed
- Slime design mimics Tokyo's rail system · EurekAlert! / AAAS · Accessed
- Amoebae Anticipate Periodic Events (Saigusa, Tero, Nakagaki, Kuramoto), Phys. Rev. Lett. 100, 018101 · American Physical Society (author copy in Hokkaido University repository HUSCAP) · Accessed
- Amoebae anticipate periodic events · Physical Review Letters (via PubMed / NIH National Library of Medicine) · Accessed
- Slime slips the maze and jolts brain researchers · The Guardian · Accessed
- Primitive intelligence · New Scientist · Accessed
Published
Discuss on X