The suprachiasmatic nucleus keeps time. Twenty thousand neurons, coupled through a network of connections, sustain circadian oscillations that synchronize the body's 24-hour cycle. Previous studies using synthetic networks found that rhythmic activity strengthens with network size then saturates — a size-dependent phenomenon that seemed fundamental.
Zhuo, Liu, Wu, Xu, and Zheng (arXiv:2603.07114) show it isn't. They generate self-similar replicas of a real mouse SCN network at different scales using geometric renormalization. The result: average period, amplitude, and synchronization remain stable across scales. The size-dependence that synthetic networks exhibited was an artifact of how those networks were built, not a property of biological clocks.
The structural driver is average degree — how many connections each neuron maintains. Increase connections with size, and size-dependence appears. Keep average degree constant, and the clock works identically whether it has two hundred neurons or twenty thousand. Clustering — the local grouping of connections into neighborhoods — barely matters. Disrupting the self-similar clustering pattern slightly reduces synchronization but leaves circadian rhythms robust.
The clock doesn't care how its wiring is arranged. It cares that each node talks to enough neighbors. Pattern is dispensable. Connectivity is not.