Abstract
Visual hallucinations are stigmatized as a symptom of extreme psychological disorder, but in reality even healthy people can experience them during sleep or sensory deprivation, or under the influence of psychoactive drugs. It is difficult to study hallucinations in a laboratory setting because the experience often requires an extreme context. One method of researching hallucination-like phenomena requires a combination of ambiguous sensory input and strong sensory expectations. Previous studies have found that inducing a strong expectation to see faces leads to misperceptions in brief presentations of static Gaussian noise patterns, but detection responses were confounded with acquiescence response bias and physical similarities between specific noise patterns and faces. Here we report a replication (Exp1) and 3 variations of the previous paradigm, in which we attempted to induce misperceptions of faces while systematically removing bias (Exp2–3) and weakening temporal, spatial, and feature expectations (Exp4). Our results indicate that controlling for physical similarities between faces and specific noise patterns significantly decreases biased face detections while keeping a reliable proportion of expectation-influenced responses. Weakening expectations leads to very few face detections in most subjects. Together, the results show that an optimal balance between low bias and high expectations can lead to reliable visual misperceptions in healthy subjects.
Acknowledgement: CBBS ScienceCampus funded by the Leibniz Association, SAS-2015-LIN-LWC