MAGICONTROL: A Normed Video Database of Magic Effects and Matched Control Events for Experimental Research

Magic effects are staged events in which a performer uses concealed conjuring methods to produce an apparently impossible outcome, such as an object vanishing, appearing, or transforming. Because these events violate observers’ ordinary causal expectations, they provide useful stimuli for investigating epistemic emotions and expectation violation. Yet, existing stimulus databases often lack carefully matched control conditions and contain uncontrolled social variables. To address these methodological gaps, this study introduces MAGICONTROL, an open-access video database comprising 100 short clips divided into 50 genuine magic effects and 50 closely matched non-magical control events. All stimuli were recorded from a fixed frontal perspective by a single professional performer, intentionally eliminating speech and audience reactions to minimize social confounds. The database was normed via an online study with 112 participants, assessing trick recognition alongside subjective ratings for curiosity, interest, and surprise. Descriptive analyses showed comparable mean durations and file sizes across conditions, while pair-level analyses quantified the degree of temporal matching within each pair. Normative results demonstrate high overall trick classification accuracy alongside substantial, interpretable item-level variability across the 10-point Likert ratings for curiosity, interest, and surprise. Ultimately, MAGICONTROL offers a standardized, ecologically valid, and highly controlled paired stimulus resource that supports controlled comparisons intended to isolate the contribution of the magical outcome. This database supports reproducible experimental research in cognitive science, specifically concerning causal reasoning, attention, prediction error, and affective responses to impossible events.