Book review: Surfing Uncertainty

Abstract

The predictive processing (PP) model proposes that the brain functions as a hierarchical prediction engine, continuously matching incoming sensory data (bottom-up) with internally generated predictions (top-down). At each level of neural processing, these streams are integrated via Bayesian inference, with the overarching goal of minimizing prediction error, or “surprisal.” Perception is thus a form of “controlled hallucination,” where top-down expectations actively shape sensory experience. This framework offers explanations for a wide range of cognitive phenomena, including attention, imagination, dreaming, priming, learning, and motor control, where actions are initiated by strong proprioceptive predictions. Furthermore, the PP model provides insights into the mechanisms underlying the inability to tickle oneself, the placebo effect, social conformity, and the roles of specific neurotransmitters. It also offers potential explanations for conditions such as autism, characterized by an over-reliance on bottom-up processing and overly precise predictions, and schizophrenia, involving weak priors and aberrant prediction error signals, suggesting a mathematical basis for how expectations shape reality. – AI-generated abstract.