Arousal, Information Density, and Temporal Distortion during Video Viewing

Abstract

Time perception research has predominantly relied on simple laboratory stimuli, leaving uncertain whether established timing mechanisms generalize to the complex, dynamic stimuli that characterize real-world experience. This study examined temporal distortion during naturalistic television viewing using a prospective time estimation paradigm. A total of 129 participants viewed 16 video clips (40 seconds each) while providing duration estimates and ratings of arousal, valence, information density, and attitude. At the level of the experimental manipulation, a significant main effect of arousal emerged: high-arousal content was perceived as lasting significantly longer than moderate-arousal content ( M = 34.52s vs. M = 32.73s), F (1, 128) = 12.09, p < .001, η p² = .086. However, mixed-effects models treating all predictors as continuous trial-level variables revealed that information density was the strongest predictor of duration estimates ( β = 0.31, p = .015), while continuous arousal ratings did not reach significance. Positive valence also predicted longer duration estimates at the trial level ( β = 0.20, p = .038). An unexpected finding showed that favorably rated clips were perceived as lasting longer, consistent with attentional engagement accounts. These findings extend time perception research to naturalistic viewing contexts and suggest that information density and attentional mechanisms play a central role in temporal distortion alongside, and possibly beyond, arousal-driven pacemaker effects.