The use of Bayesian techniques for studying and modeling biological systems has vastly progressed cognitive sciences in recent decades. More recently, it has been proposed that these frameworks are not only helpful in constructing descriptive models of biological functions but also in treating living systems as Bayesian inference machines. In this context, statistical tools that were previously employed in biology, neuroscience, and psychology are now being used to simulate the underlying mechanisms present in living systems, with the systems themselves being treated as if they were performing Bayesian calculations. The free energy principle is a framework that emerged in response to this paradigm shift and it proposes the minimization of variational free energy as a general computational principle for biological systems. This talk aims to provide a theoretical introduction of the active inference framework in continuous time and to present some examples where it is used to develop models of motor control. In particular I will focus on an active inference model of active sensing, which specifically investigate the anticipatory control of whisking in rodents. Insights into the neural circuits underlying this behavior can be gained by understanding how rodents modulate their whisker amplitude during exploration and object scanning.
Join at: imt.lu/aula1