And it is prediction error minimization that provides the connection between controlled hallucinations and Bayesian inference. It takes a Bayesian claim about what the brain should do and turns it into a proposal about what it actually does do. By minimizing prediction errors everywhere and all the time, it turns out that the brain is actually implementing Bayes’ rule. More precisely, it is approximating Bayes’ rule. It is this connection that licenses the idea that perceptual content is a top-down controlled hallucination, rather than a bottom-up “readout” of sensory data.
