ABSTRACT
This chapter describes hierarchical dynamic models and reviews a generic variational scheme for their inversion. It is then shown that the brain has evolved the necessary anatomical and physiological equipment to implement this inversion, given sensory data. Critically, the nature of the inversion lends itself to a relatively simple neural-network implementation that shares many formal similarities with real cortical hierarchies in the brain. The basic idea that the brain uses hierarchical inference has been described in a series of papers (Friston, 2003, 2005; Friston, Kilner, & Harrison, 2006). These papers entertain the notion that the brain uses empirical Bayes for inference about its sensory input, given the hierarchical organization of cortical systems. Here, we generalize this idea to cover dynamical models and consider how neural networks could be configured to invert these models and deconvolve sensory causes from sensory input.
