This lecture provides an overview of the principles of active inference, with a focus on neurobiological process theories. Specifically, we will consider the brain’s implicit use of internal (generative) models to explain sensory data. The form of such models has implications for the form of the message passing between neuronal populations in forming (perceptual) inferences and generating effector responses. Key themes are the microcircuits that emerge from models formulated in discrete and continuous time, the formulation of motor commands as predictions, synaptic gain as precision control, and the potential role of subcortical structures in adjudicating between behavioural policies. Throughout, we will consider examples of neurobiological systems to illustrate the important concepts.

Recording: https://youtu.be/iPMGLWM5le0