I don’t know if Physarum polycephalum is more interesting in any way than other slime molds, but it’s the one that scientists have studied the most, and so it’s the one I’m going to write about. It’s a very strange creature, neither bacterium nor fungus but rather a protist, a single-celled amoeba-like creature that can grow large enough to cover the decomposing logs that it prefers to live on. It begins life as a spore ejected from its parent, remaining dormant as long as it takes for its environment to become hospitable. Once it does, it morphs into its main form, which is a flowing mess of gooey tendrils called a plasmodium. Even in this state, the organism remains single-celled, but interestingly it has thousands of nuclei distributed throughout its body. Here’s a time-lapse video of the plasmodium’s strange ebbs and flows (the action starts at 0:45):
As you might be able to tell from the clip, the slime mold is pretty good at hunting its prey, the white fungus on the log. It uses an interesting mix of strategies, with some of its branches greedily focusing on the fungus at hand and others exploring off to the sides. As Steven Shaviro notes in a blog post about slime molds, the mold is unpredictable and indeterminate, and that should be read not as noise, but as an adaptive trait: indeterminacy doesn’t just arise from faults in a would-be-deterministic machine, but might also as a strategy. Shaviro notes, “Any organism that reacted to stimuli in a completely predictable manner could all too easily be wiped out by predators who were able to anticipate these responses.”
This sort of unpredictability provides an interesting counterpoint to the more formal, symbolic methods of decision, ones which non-zoopoetics humans are more likely to describe as cognition. From a computational point of view, the slime molds are able to solve problems which are known to be very hard to solve algorithmically. One example is a problem called the Steiner tree problem, which has been shown by computational complexity theorists to be as hard to solve as the hardest problems out there, in a well-defined sense. This problem arises in the design of road or rail networks, or sometimes when solving mazes, and it’s hard enough that it’s not worth running any deterministic algorithm on a computer to try to get it perfect. But Physarum‘s heuristic and unpredictable methods allow it to find near-optimal solutions very quickly. Physarum solves mazes and even designs rail networks. Here’s a video from that last paper of the mold redesigning Tokyo’s rail network as it arranges its body to best digest a bunch of oats laid out according to the locations of Tokyo’s train stations.
I would argue that what makes the slime mold so effective, apart from its unpredictable exploration, is the materiality of its cognition. The mold is not solving problems abstractly, but rather trying to route nutrients through its own body, storing intermediate results from prior computations in physical space, and feeling the differences which result from slight motions in one way or another. A relevant quote from Shaviro’s book Discognition appears in an interesting blog post about slime molds by Elvia Wilk: “Physarum’s spatial memory works not by internal representation, but rather by a physical marking of the very space that is being remembered. In this case, the map actually coincides with the territory.” I would argue that the same can be said of humans, and not in a metaphorical sense, if you think of how our road networks and cities form. It cannot be said that the huge, disparate groups of people that collectively construct these things have some common “internal representation” of the task at hand, or even that they are aware of what is happening. Rather, their memory is on the ground, material, in the old roads that fall out of use or are repaved, gradually and unpredictably shifting until the network is good enough.