Abstract
The free energy principle in the neurosciences stipulates that all viable agents embody and minimize informational free energy to fit their environmental niche. We implement free energy minimization in a more physically principled manner by effectively casting it to Bayesian control dynamics. Specifically, we build a continuous-state formulation that prescribes the brain's computation for actively inferring the causes of sensory inputs, based on the principle of least action. For this, we consider that the motor signal arises from the residual errors in the cognitive expectation about the nonstationary sensory inputs at the proprioceptive level. Consequently, we derive the effective Hamilton equations that determine the optimal trajectories in the perceptual phase space, which minimize the sensory uncertainty. We also demonstrate their utility using a simple model with biological relevance to sensorimotor regulation processes such as motor reflex arcs or saccadic eye movement.

This publication has 43 references indexed in Scilit: