The Changing Weight of Expectation: How Shifting Priors Underpin Variability in Hallucination Frequency

Abstract

The predictive processing framework encourages us to consider perception as arising from the integration of preexisting knowledge with current sensory input (1). It formulates hallucinations—percepts without accompanying stimuli—in terms of increased weighting of predicted over actual input and thereby suggests an explanation of the computational mechanisms underlying such experiences. Several studies of individuals with or prone to hallucinations have exploited this framework and have suggested ways in which the balance between prediction and input may be altered.