No common factor for illusory percepts, but a link between pareidolia and delusion tendency: A test of predictive coding theory

• Published in: Frontiers in psychology Vol. 13; p. 1067985
• Main Authors: Lhotka, Magdalena, Ischebeck, Anja, Helmlinger, Birgit, Zaretskaya, Natalia
• Format: Journal Article
• Language: English
• Published: Switzerland Frontiers Media S.A 04.01.2023
• Subjects: bistable perception; delusions; hallucinations; illusions; predictive coding; Psychology; vision; vision; delusions; predictive coding; Bayesian prior; pareidolia; illusions; bistable perception; hallucinations
• ISSN: 1664-1078; 1664-1078
• DOI: 10.3389/fpsyg.2022.1067985

Predictive coding theory is an influential view of perception and cognition. It proposes that subjective experience of the sensory information results from a comparison between the sensory input and the top-down prediction about this input, the latter being critical for shaping the final perceptual outcome. The theory is able to explain a wide range of phenomena extending from sensory experiences such as visual illusions to complex pathological states such as hallucinations and psychosis. In the current study we aimed at testing the proposed connection between different phenomena explained by the predictive coding theory by measuring the manifestation of top-down predictions at progressing levels of complexity, starting from bistable visual illusions (alternating subjective experience of the same sensory input) and pareidolias (alternative meaningful interpretation of the sensory input) to self-reports of hallucinations and delusional ideations in everyday life. Examining the correlation structure of these measures in 82 adult healthy subjects revealed a positive association between pareidolia proneness and a tendency for delusional ideations, yet without any relationship to bistable illusions. These results show that only a subset of the phenomena that are explained by the predictive coding theory can be attributed to one common underlying factor. Our findings thus support the hierarchical view of predictive processing with independent top-down effects at the sensory and cognitive levels.