A colony of bacteria that keeps its own middle alive by stopping to let it eat.
Some bacteria live together in a thick mat called a biofilm. Food arrives from outside, so the cells at the edge eat first and the cells in the middle get what is left. Once the colony is large enough, the middle goes hungry. Then something remarkable happens: the edge stops growing for a while, the food passes inward, the middle recovers, and the edge starts again. The colony takes turns with itself.
The stop-and-start growth, and the size at which it begins: Liu and colleagues, Nature 523 (2015). That the hungry middle tells the edge with an electrical signal, a wave of potassium passed from cell to cell: Prindle and colleagues, Nature 527 (2015). The glow under the dye lamp stands for the dye those experiments used to see the signal. The two references were checked against PubMed (26200335 and 26503040); what is said of them here is a summary from memory, so read the papers before quoting it. The amber at the rim marks where cells are growing; it is drawn in by the model and is not a real dye.
The colony is a grid seen from above. Each square holds four things: how much colony is there, how much food, whether the cells are signalling, and how thick the mat has grown. Food spreads inward and is eaten. A cell starts signalling when it is short of food, sooner if its neighbours are already signalling, and stops only when it has been properly fed. Signal spreads and fades. A cell that receives signal eats less for growth, though it still eats to stay alive. A cell left signalling for a long time dies. Nothing else is put in: the turn-taking is not scripted, it comes out of those rules.
The dial sets how much a cell cuts its growth for a given signal. At nought the edge is deaf.
Four guesses were written down before the runs. Three held: a small colony does not take turns and a larger one does; a deaf edge loses more of its middle (about 83% alive against 95%); and the count of change in each turn varies less than the clock time the turn takes. One failed: I guessed the colony with the most living matter would be one that listens a moderate amount. It was the deaf one. In this model, listening costs growth and buys the middle its life. It is a trade, not a free gain.
Real biofilms are three-dimensional, are fed from below as well as from the side, and use a second chemical, ammonium, which the middle makes and the edge needs; that mutual need is in the experiments and is not in this model. The wrinkles are drawn for the eye and are not simulated. The experiments were done in a small flow chamber under a microscope, not on a dish this size. The time shown in hours is set so that a turn takes about as long as in the experiments; the model does not predict it. The numbers are the model's own and are not measurements.
The lines marked “one way of reading it” are analogies, offered for thinking, and are not findings. Holding and the good-enough carer: D. W. Winnicott (1960). Making-with, sympoiesis: D. Haraway, Staying with the Trouble (2016), after M. Beth Dempster. Counting change, the handover and the surplus follow a process reading (CRR), which is a way of reading and not a theorem. Bacteria do not intend, wait or care; the words are ours.
Renderer, lens and grain follow the studio’s Clay Sea piece. three.js r158.