A Brainless Blob Drew Tokyo's Metro

Bölüm metni

On a tray lies a miniature Japan made of oatmeal: every flake a city. A yellow, single-celled smear placed at Tokyo connects them within hours, and the network it draws is nearly as efficient as the rail system engineers spent decades building. This thing has no brain, no nerves — it is not even more than one cell.

In a laboratory at Hokkaido University, a thin layer of agar had been poured across a flat tray. Agar is a jelly-like growth medium used in labs; it is made from seaweed, it is transparent, and it gives whatever is placed on it both a floor and moisture. This tray was no ordinary petri dish. The edges of the jelly had been cut with great care, and the shape they were cut into was a map of Japan's Kanto plain. The notch of Tokyo Bay, the line where the sea meets the shore, the places where the mountains rise; all of it was there, to scale. The sea and the mountains were left without medium, or else lit with a strong light, because the star of this experiment flees from bright light. In other words, the researchers had written geography itself into the dish as an obstacle.

Then came the oat flakes. 36 small pieces of oat were placed one by one, each at the true position of a city around Tokyo. One where Yokohama is, one where Chiba is, one for Utsunomiya in the north, one for Hachioji in the west. The map was ready, the cities were in place, and between them there was nothing at all. At the point where Tokyo lies, however, what was placed was not an oat flake but a yellowish, veined, moist piece of tissue.

The name of that piece of tissue is Physarum polycephalum, the many-headed slime mould. It has mould in its name, but it is not a fungus; it is not a plant either, nor an animal. It belongs to a separate branch known as the amoebozoans. And here is the most astonishing thing about it: that yellow patch, the size of a hand, sometimes the size of a plate, exceeding 30 centimetres under ideal conditions, is a single cell. One undivided cell, not partitioned by any walls. Millions of nuclei float inside it, but all of them lie beneath the same membrane. It has no nerve cells. It has no brain. It has no eyes, no ears, no centre. It lives under rotting logs in forests, eats bacteria and fungal spores, and travels a few centimetres an hour.

There was a history behind seating this creature at such a table. In 2000, Toshiyuki Nakagaki, who headed the same group, had published a half-page note in the British science journal Nature. The claim in its title was blunt: an amoeba was solving a maze. The experiment was simple. The slime mould had been spread through a small maze partitioned by plastic walls, with an oat flake placed at the entrance and another at the exit. There were four possible routes from entrance to exit; the shortest was about 33 millimetres, the longest 45. At the start, the creature was everywhere in the maze, filling every corridor. After roughly four hours it had withdrawn from the dead ends. Four hours later, only one thing remained: a single thick tube joining the two oat flakes by the shortest route. The other three routes had been abandoned, absorbed, erased. A cell without a brain had solved a problem that humans solve by drawing on paper, and it had solved it by reshaping its own body.

The Tokyo tray was that idea, scaled up. In the maze there had been two points; here there were 36, and it was not obvious what the right answer even was. The camera was set to time-lapse, and the waiting began.

In the first hours the patch spread with an appetite that would not stop. It widened outward from Tokyo like a wave, engulfed every oat flake in its path, and covered almost the whole plain in a yellow sheet. At this stage there was no network; there was an occupation. Then the second phase began. The creature started to abandon most of the area it had taken, to pull its own mass back in. The blindness of spreading gave way to a choosing. Some arms thickened from one day to the next; others thinned and snapped. The sheet grew sparse, and as it thinned, a skeleton emerged beneath it.

At the end of 26 hours, the image on the tray threw everyone who looked at it to the same place. The yellow tubes linking the oat flakes overlapped, uncannily, with the lines of Tokyo's rail network. Thick trunk corridors opening outward from the centre, ring links joining the outer cities to one another, curves bending around the mountain. The rough plan of a network that was the product of decades of engineers, planners, land surveys, budget arguments and political bargaining had been drawn by a single cell in a day and two hours, while chasing oats.

The findings were published in the 22 January 2010 issue of the American science journal Science. The paper's first author was Atsushi Tero, then working at Hokkaido; Nakagaki, the mathematician Ryo Kobayashi and Mark Fricker of the University of Oxford were among the others. That same year the team received the Ig Nobel Prize, given for research that first makes you laugh and then makes you think, in the category of transportation planning. For Nakagaki it was the second; the maze experiment had won the same prize in 2008.

And once the smile of the award had faded, a single question remained. How does this thing happen?

The search for an answer has to begin by accepting what the creature is not. There is no decision centre. There is no manager cell sitting somewhere on the tray, looking at the map and saying, "let's cancel that line." Nor is there a nervous system gathering and processing signals. All there is is a plumbing system of fluid-filled tubes, and the cytoplasm flowing inside that plumbing.

The way this plumbing works is the slime mould's most distinctive feature. In the walls of the tubes there are proteins much like those in our muscle cells: actin and myosin. Those walls contract and relax rhythmically; a wave of contraction travels along the tubes, raising the pressure in one place and lowering it in another. The result is a flow that runs back and forth like the shuttle on a weaving loom. The fluid moves in one direction at speeds reaching a millimetre per second, then, after about one to two minutes, reverses and comes back. The entire body stirs its own interior without pause. Nutrients, signalling molecules and chemical messages circulate on this coming and going current.

And this is exactly where the network's self-shaping comes from. The rule is simple enough to write in a single line: a tube through which plenty of fluid passes grows thicker, and a tube through which little passes grows thinner and is eventually absorbed.

To see how powerful that rule is, it is enough to recall how fluids behave. The resistance met by fluid passing through a tube is extremely sensitive to the tube's width; widening the tube a little lowers the resistance by a far greater proportion. Which means the thick tube pulls flow towards itself, at the expense of the thin one. As flow increases the wall relaxes more, the tube widens further, and it draws still more flow. This is a snowball effect. There is a pressure in the opposite direction too: every unused tube steadily weakens, thins and breaks. The creature takes the material it has committed to an unused line and invests it in a used one.

Now think about the maze with that in mind. At the start the creature fills every corridor, there are four separate lines between the two oat flakes, and all four connect the same two points. But the short line carries the same quantity of fluid with less resistance. The flow through the short line is a little greater. A little more flow means a slightly thicker tube; the thickening tube draws still more flow on the next pass, and the long lines quietly dry out. In the dead ends there is no flow at all; since no fluid can enter at one end and leave at the same end, those stretches never even enter the race. Nobody calculates anything. Physics alone, by itself, accumulates the right answer.

This is why the slime mould's memory bears no resemblance to what we call memory. The route it has learned is not a record stored somewhere; it is the very geometry of its body. A thick tube is the sentence "this way is worth travelling," made flesh. A study published in 2021 in the journal of the National Academy of Sciences of the United States pushed the idea one step further: Mirna Kramar and Karen Alim showed that when the creature touches a food source, a softening substance is released into the tubes from that point, that this substance spreads through the network with the flow, and that it widens the tubes it passes through. Which means the information about where food is found gets engraved into the hierarchy of tube thicknesses. The body is its own map.

Nakagaki and Kobayashi turned this logic into equations. In the model each tube carries a conductivity value, flow is distributed according to that conductivity, and conductivity in turn grows or shrinks according to the flow it carries. The result was both surprising and reassuring: without watching a real creature on a tray, simply by turning these two rules through numbers, they could produce networks resembling the one on the Tokyo tray. What is more, adjusting a single coefficient in the equation changed the character of the network; sometimes out came the cheapest, barest possible skeleton, sometimes a weave with spare routes, more expensive but more resistant to breaks.

Then the mathematicians took over. In 2012, Vincenzo Bonifaci, Kurt Mehlhorn and Girish Varma proved that this system of equations really does converge on the shortest path between two points, independently of the shape of the network and of the initial distribution of mass. So the slime mould's solution is not a coincidence or a piece of laboratory luck. It is hidden in the rule itself.

But is a thickening tube thinking?

There is a debate around that question that continues to this day and sharpens from time to time, and both sides have serious grounds.

The sceptics' objection is direct: what we are seeing here, they say, is not intelligence but a minimisation process. Nature is full of such things. Dip a wire frame into soapy water and pull it out, and the film builds the surface that joins several points with the least possible surface area; nobody calculated it, surface tension did. Rivers on their way to the sea find the bed of least resistance. Lightning picks the corridor in the air that ionises most easily. The thickening of the slime mould's tubes in proportion to flow belongs to the same family: a flow network reorganises itself so as to reduce its loss of energy. On this view, saying "the blob designed the Tokyo metro" is as figurative as saying "the soap film solved a geometry problem."

The other side says that this analogy jams up beyond a certain point. A soap film does not go hungry, does not keep its past experience, does not acquire habits. The slime mould does. In 2016, Romain Boisseau, David Vogel and Audrey Dussutour, working in Toulouse, built a gel bridge that the creature had to cross in order to reach an oat flake, and laced the bridge with bitter substances, quinine or caffeine. In the first days the creature entered the bridge reluctantly, slowly. As the days passed the hesitation diminished; by the end of the sixth day it was advancing as fast as the control group crossing a harmless bridge. Then, when the substance was removed for a while, the wariness came back. This is a demonstration of habituation in an organism with no nervous system; one of the most basic forms of learning. The same team also points back to older findings showing that the creature can anticipate a periodically repeated drying or cooling event in its environment and begin to slow down before it arrives.

Even so, to stay honest about the real subject at hand, a few things that have been exaggerated about the Tokyo experiment need correcting. First, the slime mould did not design the railway and did not "beat" the engineers. The comparison the researchers made was far more measured: the network the creature built was comparable to the real rail network in total tube length, that is cost, in average distance between cities, that is efficiency, and in tolerance to the failure of a random link, that is robustness. On some measures it was a shade ahead, and on resistance to breaks somewhat behind. What there was, was not a superiority but an equivalence, and that was precisely the striking part.

Second, the tray was not Tokyo. Most of the constraints that draw a real railway network were not present on that jelly: land ownership, the cost of expropriation, the geology in which a tunnel can be dug, the distribution of population over time, which city carried political weight in which period. The slime mould did not know the cities' populations either; every oat flake was a reward of equal size to it. The answer to the experiment was already known, and the obstacles were built around the geography that had produced that answer. This does not make the result worthless; but it does leave the sentence "a brainless creature planned better than Japanese engineers" without support.

Third, and most importantly: no mystical conclusion follows from any of this. There is no question of the slime mould knowing Tokyo, recognising cities, pursuing a purpose or carrying some enigmatic consciousness. What is there is a well-understood mechanism; chemical signals, rhythms of contraction, flow, and feedback accumulating in tubes. Claims that go beyond that are products of imagination, not of the present scientific consensus.

So what is left? I think the real legacy of the experiment is not the answer itself but where the answer came from. There is a family of algorithms in the engineering literature today inspired by the slime mould; they are used on problems like network design, route planning and the placement of communication lines. But the appeal of those algorithms comes less from their speed than from their architecture. Because our way of building networks is almost always central: a desk, a map, a planner, an approval. The slime mould's is not. No point in its network has information about the whole. Each tube only "knows" how much passes through itself, and thickens or thins accordingly. Out of a local, stupid, one-line rule repeated millions of times, a solution that sees the whole is born.

What this tells us is as illuminating as it is unsettling: for a system to become optimised, it does not need anyone optimising it. The same logic meets us elsewhere in the world. Our blood vessels also mature by pruning the little-used branches and thickening the ones through which much flows. The veins of a leaf obey the same economy. The synapses in our brains also strengthen and weaken according to use. The slime mould is the barest, most visible member of this family; because, having no skull to hide it in, it does all its reckoning out in the open, on the tray, as a yellow line.

And perhaps that was the genuinely disquieting thing about that time-lapse footage. Watching 26 hours compressed into a minute on the screen, nobody looking at that yellow patch spreading, then withdrawing, then appearing to have made up its mind, thought that what they were seeing was merely a flowing fluid. The question we are forced to sit with is this: if we look at what it does and say "that is just physics," why do we not say the same about what goes on inside our own heads?

Nöron Science'de bugünkü bölümün sonuna geldik. Yeni bir hikâyede yeniden buluşmak üzere.