Abstract:
Maintaining physiological stability requires resolving conflicts between multiple needs, such as energy, hydration, and temperature, while also accounting for the costs and uncertainties of action. In this talk, I will describe ongoing hybrid theory-experiment work for understanding how such physiology-driven behaviors are organized. First, I will present a control-theoretic model in which physiological variables evolve with distinct timescales and are regulated by behavioral actions that produce delayed and uncertain gains. The model shows how optimal policies partition physiological space into regions associated with different behavioral priorities, the ideal environmental size constrained by action costs and timescales, and the conditions under which a behavioral hierarchy emerges as an optimal and compressible solution to homeostatic control. Second, I will turn to the experimental challenge of continuously measuring the relevant physiological state in freely moving animals. Building on our previous work on chronic jugular microdialysis in freely moving mice, I will describe current efforts to combine microdialysis with continuous measurements of behavior, CO₂/O₂ exchange, and food and water intake over 72 hours. These datasets create an opportunity to reconstruct the latent circadian dynamics of physiology during natural behavior. Third, I will discuss how computational methods such as dynamic mode decomposition, Gaussian-process factor analysis, and reduced-rank regression can be used to identify low-dimensional physiological trajectories, characterize their dominant timescales, and relate them to behavioral structure. The overarching goal is to close the loop between theory and experiment by using high-dimensional physiological measurements to test whether behavioral hierarchies reflect the dynamics of internal state space.