DEED: A Dataset for Dream-related Emotion Research

Abstract

AbstractEmotion is closely related to human cognition and behaviour. In recent years, scholars have conducted extensive research on emotion in waking state based on electroencephalography (EEG) and achieved certain results. However, Emotional activity continues after sleep, with a concentrated response of sleep emotions in dreams. Sleep emotions are concentrated in dreams, which can better reflect a series of real physical and psychological states of the human body. Currently, there is no publicly available dataset for the assessment of dream mood. Therefore, we present a physiological dataset Dream Emotion Evaluation Dataset (DEED) for the assessment of dream mood, which recorded EEG signals from 38 participants over 89 whole sleep nights and 533 dream segments(after exclusion of unqualified nights, those dream segments are extracted from 82 whole sleep nights). We studied the correlations between the subjective ratings and the EEG signals and brain network patterns for dream emotions. In addition, the relationship between the asymmetry of left and right brain bands and positive and negative dream emotions was studied. The machine learning algorithm was also used to classify different emotional EEG, which confirmed the validity of the dataset. In the meantime, we encourage other researchers to explore the underlying neural mechanisms involved in sleep.