A naturalistic database for understanding individual differences in affective responses to movie clips

Given the multidimensionality and dynamicity of movies, and open-ended nature of movie viewing, movie-based paradigms may hold potential for uncovering a richer array of human responses than possible with controlled experimentation. We created a database of movie clips, emotion ratings and derived features, and stimulus annotations, as a reference for studying normal and pathological emotional reactivity. We estimated the normative distribution of features derived from holistic ratings of negative and positive emotions, and real-time valence ratings (n = 108 healthy adults), of 22 short clips exhibiting unfolding social behaviors in diverse real-life situations. To capture dynamic and integrative aspects of emotional reactivity, we computed the intensity, complexity and intersubject synchronization of the real-time ratings, and intensity of the holistic ratings. We include biological (sex), individual (familiarity with the clips), and stimulus (gender representation in the clips) variables to facilitate investigation of their influence on the viewer's emotional experience. We illustrate the potential of emotional reactivity profiling, and provide guidelines for stimulus selection for different applications, based on the real-time and holistic ratings.