Tokyo's rail network grows like slime mold

This Science magazine article explains us a very interesting new discovery by a team of researchers at Hokkaido University.

The researchers put some oat flakes as if they were places in the Tokyo metropolitan area where most of the people are accumulated. Then they put a slime mold cell in the middle of the oat flakes and let it grow. As slime mold loves oat flakes, it soon started to spread and in some hours it had reached all of the flakes. As the time went by the slime mold erased “connections” between some small oat flakes and strengthen the connections between bigger and central flakes. The most impressive thing is that it was “intelligent” (slime mold doesn’t have a brain) enough to maintain redundant connections between the most “important” flakes, something that allows the slime mold to keep the connection to an important flake even if an “accident” happens. The slime mold really created a really good communication network between the oat flakes, in fact the network is very similar to Tokyo’s rail network designed by engineers. For example, in Tokyo there are many redundant lines to make sure that if something unexpected happens in some part of the city, like an earthquake, it will not affect the rest of the network.

After analyzing the slime mold growth pattern, the researchers created a mathematical model that describes the expansion of the network and when simulating it on a computer the algorithm ends up finding a quite optimum solution in which the most robust connections are established between the most important/central nodes and the redundant connections slowly disappear. Moreover, the slime mold tries to find the shortest path to each location (a classical problem of computing science). We can simulate this using software, in fact there are many algorithms to design networks, but… how did slime molds learn how to grow optimally to reach the maximum amount of food/energy? how does the slime mold “know” how to solve the problem? where is that behavior programmed if the slime mold doesn’t have a brain? in the genes? small rules that create a complex behavior? cellular automatons?

Why would we want to make very complex calculations and build algorithms to design networks if we can “ask” a slime mold to do it for us? Why spend millions hiring engineers if slime mold can design transport networks? In fact, why spend time solving problems that nature has solved during millions of years of evolution? This is one of the perspectives that Biomimicry / Biomimesis uses to solve problems; if you are interested in this topic and want to learn more, I recommend you to watch these talks by Janine Benyus and Robert Full.