Pre-sleep affect predicts subsequent REM frontal theta in nonlinear fashion

Pre-sleep affect is thought to influence sleep, but associations with both sleep architecture and the electroencephalographic (EEG) power spectrum are mixed. In this pre-registered study, we assessed negative valence and arousal 1 h pre-sleep in 52 adults drawn from the community, then recorded one night of polysomnography (PSG) in participants' own homes. Pre-sleep affect was not associated with nonrapid eye movement (NREM) or rapid eye movement (REM) sleep architecture parameters, but we did observe inverted U-shaped relationships between both negative valence and arousal and REM frontal theta power, such that theta power was highest at moderate negative valence and arousal, and lowest at either affective extreme. When entered into a model together, both valence and arousal accounted for independent variance. Secondary analyses revealed a similar quadratic association with pre-sleep positive valence, suggesting a nonspecific effect of pre-sleep valence on REM frontal theta. Robustness checks confirmed that effects were not explained by homeostatic sleep pressure or sleep timing. Our results suggest that mixed findings in the literature may reflect different ends of a quadratic function, underscoring the importance of assessing how different components of pre-sleep affect relate to sleep.