Deriving Time-Averaged Active Inference from Control Principles

Active inference offers a principled account of behavior as minimizing average sensory surprise over time. Applications active to control problems have heretofore tended focus on finite-horizon or discounted-surprise problems, despite deriving from the infinite-horizon, average-surprise imperative free-energy principle. Here we derive an formulation optimal principles. Our returns roots in neuroanatomy and neurophysiology, formally reconnecting feedback control. provides unified objective functional for sensorimotor allows reference states vary