Kyuhyung Kim
Traditional dogmas assume that executing a stable behavior requires neural circuits to remain in a fixed, steady state. However, recent population imaging reveals "representational drift", a continuous, unprompted shifting of neural activity over time despite invariant behavioral outputs. This introduces a profound paradox where behavioral permanence must emerge from a volatile physical substrate. Here, I review how nervous systems maintain dynamic stability across diverse biological scales. In Caenorhabditis elegans, a structurally fixed connectome achieves functional invariance amid intrinsic whole-brain drift by dynamically reweighting sensory-to-motor paths via acute neuromodulation, metabolic gates, and intergenerational RNA cascades. In mammals, behavioral constancy emerges through collective population dynamics constrained within low-dimensional neural manifolds and stabilized by homeostatic compensation. Far from a biological flaw, representational drift acts as an active computational feature to explore degenerate solution spaces. Together, these multiscale insights demonstrate that behavioral permanence is anchored not in structural rigidity, but in the regulated preservation of network-level dynamics.