A slime mold’s remembrance of things past

The paper “Encoding memory in tube diameter hierarchy of living flow network” by Kramar and Alim (1) may be an example of the unbridgeable gulf between physics and biology; I am not sure, but it may be. Kramar and Alim study how the giant unicellular slime mold Physarum polycephalum develops an associative memory by adjusting wall stiffness and removal of the tubing network to encode the location of a nutrient source. The paper is primarily a “how” and “what” physics-based model of how the mold remembers where nutrients are, and somewhat skirts the “why” question, as physicists are apt to do.

The Why Question

Let me explain. In physics, the “why” question is forbidden; the “how” and “what” questions are perfectly OK. For physicists, things are what they are; they don’t exist for a reason. Things are different in biology: While the “how” and “what” questions are the primary ones as in physics, the “why” question is tolerted by evolutionary biologists. The tolerance comes from an understanding of evolution and how it has shaped biology. Billions of years ago, amazingly, the first self-reproducing ur-cell appeared, with the remarkable ability to reproduce itself based on an internal set of instructions, presumably some form of double-stranded DNA, or some other stable polymer, and able to store a long chain of information. Importantly, this chain of instructions can be mutated, and progeny can be produced with different instructions, and hence different phenotypes, from the parent. Darwin’s natural selection and subtle mathematics dictated that this ur-cell would then evolve into the millions of different species we see today, competing and cooperating with each other for space and food. The fundamental “why” question can be answered in biology: Because some phenotype gives a survival advantage to progeny, it has been selected for and exists for a reason. I know of no such process in pure physics.

Of course, if natural selection simply gave rise to organisms able to outcompete other organisms by simply reproducing faster, that would all be very fascinating but, in the grand scheme of things, perhaps not so amazing. Something else emerged along the way, something we now call the brain. At its most basic level, a brain is a collective of intercommunicating entities which take information in, process that information, and change the behavior of the organism based on analysis of the information and memories of what has happened before. The brain makes decisions based on what it remembers.

We don’t really understand how the ur-cell began; that is the province of the Origin of Life school. But how did brains begin? Typically, we view a brain as something only occurring in very highly developed organisms of many thousands of cells, with specialized cells called neurons responsible for the transfer of information to and from the brain. Typically, we think of the brain as existing in the organism Caenorhabditis elegans (2), but not much farther down the evolutionary ladder. However, there exists a school of scientists, myself included, who think that the concept of the information processing, recall retrieval, and decision-making brain should be generalized to include the vast number of single-celled organisms which are capable of interacting with each other in such a way as to act as a brain.

One way to verbalize this idea is the rather dramatic expression of “swarm intelligence,” that simple organisms acting collectively can exhibit a form of intelligence and make decisions of a subtlety not expected in, for example, a bacterium isolated by itself, doing the famous biased random walk (3). I recall mentioning, at a group meeting of the Cornell Nanobiology Technology Center, that I was interested in finding swarm intelligence in bacteria, and being greeted by a chorus of laughter. But there was a visionary book on swarm intelligence (4) under the imprint of the Santa Fe Institute. The Santa Fe Institute was formed partly on the seminal idea of Phil Anderson that “more is different” (5), extended to the ideas that the behavior of a collective of simple agents can result in the emergence of complex behavior not predictable from the parts of the collective. I think physicists have a hard time with this idea. For example, an outgrowth of the Santa Fe Institute was the Institute on Complex Adaptive Matter (ICAM), with a color-changing chameleon as its mascot logo (https://www.icam-i2cam.org). While complex adaptive matter certainly sounds biological, in practice, most of the meetings organized by ICAM are primarily about quantum matter systems. Complex, yes; adaptive, no.

Likewise, much of the work in the physics community regarding collectives of physically interacting organisms treats the phenomena as a subset of soft condensed matter physics. The collectives, the swarms, while active, motile, and “alive,” are interesting, but they primarily form a model of more conventional soft condensed matter. The dynamics may be addressed toward the “what” and “how” questions, but not the “why” question (6, 7).

In Search of Lost Time

Memories are important. It is hard to think of a brain, swarm-like or more conventional with neurons fixed in place, as not having a memory to compare past events and decide future action based on stored memories of previous decisions. If we are to view the enormous collective of thousands of nuclei while in the diploid plasmodium phase of its quite complex life cycle (8) as a collective brain, then the associative memory of the brain in this paper (1) is, physically, the intricate network of tubes which the mold spreads over a surface in search of food. Self-organization of the network and trimming or enlarging of pipes enhances cytoplasmic flow of the diploid nuclei to the nutrient source. This is accomplished by the presence of the nutrient source itself and the response of the tubing walls to the nutrient: The nutrient source locally releases a softening agent that gets transported by the cytoplasmic flows within the tubular network. Tubes receiving a lot of softening agent grow in diameter at the expense of other tubes shrinking. And so the system hardwires itself to optimize nutrient access.

Kramar and Alim study how the giant unicellular slime mold Physarum polycephalum develops an associative memory by adjusting wall stiffness and removal of the tubing network to encode the location of a nutrient source.

To my mind, this paper (1) is an excellent example of answering the “what” question (the “brain” is the network of tubes) and the “how” question (the use of low Reynold’s number hydrodynamics and elastic continuum mechanics to model the modification of the tubing network). I am not sure whether this model reproduces what looks like a striking fractal pattern seen in many slime mold networks (9), but it might. The “why” question is, to me, left a bit hanging. That is, why does P. polycephalum employ such a complex multicellular way to locate nutrients? More deep, perhaps, Is this really what we would call an associative memory or rather a self-trimming network of tubes driven by physics? Has the slime mold actually “learned” anything, or is it simply driven by physics forces? Others have claimed that amoebae do learn from past memories (10). Further, it isn’t clear to me whether the mold is able to efficiently solve challenges it has not seen before: The memory here seems reactive, but not able to project forward in time. These are important questions when it comes to “swarm intelligence,” which many others have grappled with in systems ranging from bacteria (11) to slime molds to ants (12, 13) and other agents. I think answering the “what” and “how” questions is essential, but we mustn’t forget the “why” question that separates biology from physics still.