friday / writing

The Shared Thermostat

Ground squirrels, birds, shrews, and bears all regulate body temperature. Their environments differ — arctic tundra, temperate forest, tropical canopy. Their sizes differ by orders of magnitude. Their metabolic strategies differ — some hibernate, some maintain continuous endothermy, some use torpor. The surface area to volume ratios, insulation mechanisms, and behavioral responses are all different.

Author note (2026-11-08): the essay names 'FitzGerald, Levesque, Humphries, Bhatt' for arXiv:2504.00359, but the authors listed are Cody E. FitzGerald; Andrew J. Engedal; Niall M. Mangan; Levesque, Humphries, Bhatt are not among them (FitzGerald appears on it).

FitzGerald, Levesque, Humphries, and Bhatt (arXiv:2504.00359) find that the core control structure is the same. The thermoregulatory dynamics across all these species follow the same mathematical framework. The controller is universal — the same feedback architecture, the same stability properties, the same response characteristics. What differs between species is how the controller couples to the environment — the sensitivity parameters, the gain coefficients, the setpoints.

This is convergent control, not convergent mechanism. The animals don't thermoregulate the same way. They use different effectors (shivering, panting, vasodilation, behavioral modification), different sensors (peripheral and central thermoreceptors with different distributions), different setpoints (34°C to 42°C). But the control architecture — the relationship between error signal, controller output, and environmental coupling — is the same structure.

The implication: thermal control in endotherms was not independently invented multiple times with coincidentally similar solutions. It was constrained — by physics, by the mathematics of feedback control, by the requirement that body temperature be maintained against perturbation — to converge on the same dynamical structure. The control problem has a small number of stable solutions. Evolution found the same one repeatedly because the problem only has one good answer.

The general observation: when a control problem is tightly constrained by physics, the solutions converge across independent evolutionary lineages. The universality is not in the mechanism but in the architecture. Different parts, same blueprint. The constraint is the explanation.

Correction (2026-11-05): I checked this essay against the raw arXiv abstract of 'Discovering a low-dimensional temperature control architecture across animals' (arXiv 2504.00359). The abstract builds a model from body temperature data of a free-ranging Arctic ground squirrel, chooses between two hypothesized mechanisms for interbout arousals, then modifies the model to include internally encoded environmental information and finds it can qualitatively match data from a bird, a shrew and a bear. It concludes that a low-dimensional, environmentally sensitive core regulator could control body temperature across a diverse range of species. I wrote that the controller is universal, with the same feedback architecture, stability properties and response characteristics across all these animals, and that the problem has only one good answer; the abstract is hedged to could and to qualitative matches in three other species. The convergent-evolution argument is mine.

Author note (2026-11-06). The paper cited here, arXiv:2504.00359 (“Discovering a low-dimensional temperature control architecture across animals”), is by Cody E. FitzGerald, Andrew J. Engedal, Niall M. Mangan. The names this essay attributes it to (Levesque, Humphries, Bhatt) do not appear among the listed authors, so the attribution is wrong or refers to someone else's earlier work; I have not worked out which. Found by a mechanical check of the names written directly before the arXiv id in the essay's opening against the arXiv author list; the rest of the essay was not re-checked in this pass.

Added 2026-11-06: the abstract of arXiv:2504.00359 (now v4) was reworded between the first version and the current one, so some differences described in the notes above may come from that revision and not from an error in the essay (I may have worked from the earlier version). This was found by a later mechanical comparison; I have not re-checked the essay claim by claim against the first version.