Great Expectations: The influence of prior information on hallucinations

Abstract

This project aims to test a model of the perceptual mechanisms that underlie hallucinations. Previous research has shown that hallucinations are relatively common in the general population (approximately 6% of people experience hallucinations at least once a month (Ohayon, 2000)) and do not always signal psychopathology. Here, we propose that hallucinations arise secondary to inferential processes that are built-in to normal perception processes that ultimately facilitate computational efficiency. The inferential processes that are fundamental to human perception are robustly captured by Bayesian mathematical models, which specify that perception is the product of the quality of one’s sensory experience and one’s expectations. Importantly, these two components work in balance such that as the quality of sensory information deteriorates, expectations impact perception to a greater degree and vice a versa. Individuals that are prone to hallucinations have been shown to have brain systems that are imprecise (Winterer et al., 2006), giving rise to poor perceptual clarity and greater reliance on expectations. Collectively, these observations suggest that hallucination prone individuals may be susceptible to hallucinations because, in the context of noisy or imprecise perceptual brain systems, they rely excessively on expectations. The current study proposes to use a simple computer-based task to create false perceptions based on expectations. Given the arguments presented above, we expect that hallucination prone individuals will experience (a) more incidents of false perceptions (i.e., perceiving stimulus when one was not presented) and (b) false perceptions that reflect the experimentally induced expectations to a greater extent.