0139 Testing the Mood Regulatory Hypothesis of Dreaming

Abstract Introduction Dreaming has been theorized to facilitate fear extinction learning in a manner akin to desensitization therapy. A testable hypothesis from this theory is that the experience of fear in dreams leads to lower subsequent daytime negative affect. This study employed a novel natural language processing tool and multilevel modeling to test whether fear in dreams led to reductions in negative affective the next day.

Methods: Data from 553 participants (4986 observations) were analyzed from a larger online daily survey study of sleep and mental health from Boston College. Participants completed multiple days of dream reports and measures of daytime negative affect and sleep efficiency. Dream fear was estimated using a novel deep learning natural language processing model trained on fear-annotated tweets, applied to participants' dream report text. We fitted a Bayesian zero-inflated negative binomial multilevel model to estimate within- and between-person effects of dream fear on next-day negative affect while controlling for sleep efficiency, dream report word count, the difference between time of survey completion and time awake, total number of dreams per subject, and neuroticism.

Results: The analysis revealed that days with higher dream fear within individuals were associated with higher next-day negative affect, controlling for other variables. Specifically, for each standardized increase in dream fear, a 14% increase in next-day negative affect was predicted, surpassing the effect of poor sleep efficiency (6% increase). Moreover, individuals consistently reporting higher average dream fear tended to exhibit higher average negative affect (12%).

Conclusion: Contrary to the proposed hypothesis that dream fear regulates daytime mood, we found that fearful dreams may lead to lower next-day mood. Nights higher in dream fear were associated with increased next-day negative affect for that person, independent of sleep efficiency and neuroticism. Notably, dream fear appeared to contribute more than twice as much as poor sleep efficiency to heightened negative affect. Leveraging advanced natural language processing and multilevel modeling, this research increases our understanding of the ways dream experiences affect waking life by showing that fearful dreams may be an important day-to-day predictor of mood. Future work should explore dynamic relationships between dream emotions and mood. Support (if any)