Item

Abstract

Magic is a helpful tool for scientists to investigate human behavior and cognitive processes, such as attention, awareness and surprise. However, while there is a growing interest for magic in psychology, it has not been in the focus of neuroscientists yet. In this project, we attempt to shed light on the neural correlates of high-level prediction errors by means of functional magnetic resonance imaging (fMRI) using magic tricks. We created and validated a set of videos showing either magic tricks, similar control actions or surprising events. The magic videos contained one out of three magic effects and were shown to subjects either with or without prior knowledge about the underlying method of the magic tricks. By means of univariate analyses we could replicate previous results and show that high-level as well as low-level areas were activated during the perception of magic tricks in an effect-dependent manner. Moreover, we found areas differentially active in the perception of magic, when comparing neural responses before and after the revelation of the magic trick, suggesting that these areas are involved in higher level prediction. Multivariate analyses showed that information about the perceived magic effect can be read out of low-level visual areas, rising evidence for feedback prediction signals about high-level visual features.