Sleep loss causes social withdrawal and loneliness

Abstract

Loneliness and social isolation markedly increase mortality risk, and are linked to numerous mental and physical comorbidities, including sleep disruption. But does sleep loss causally trigger loneliness? Here, we demonstrate that a lack of sleep leads to a neural and behavioral phenotype of social withdrawal and loneliness; one that can be perceived by other members of society, and reciprocally, makes those societal members lonelier in return. We propose a model in which sleep loss instigates a propagating, self-reinforcing cycle of social separation and withdrawal.

Conflict of interest statement

The authors declare no competing interests.

Figures

Fig. 1

Experimental design. a In-laboratory repeated measures counterbalanced design. From left to right, the degree of social separation participants wished to keep from approaching others was assessed using a real, in-person human approach task, followed by a computerized version, used during a functional magnetic resonance imaging (fMRI) scan. At the end of each session, participants performed a recorded interview with open-ended questions, which was subsequently rated by independent judges, blind to study goal or sleep manipulation. b Online phase study design. Participants were asked to complete daily sleep logs for two consecutive nights tracking their habitual variations in sleep time. Following each sleep survey, participants completed a next-day questionnaire assessing social behavior and loneliness

Fig. 2

Behavioral results. a Significant increases in social distancing following sleep deprivation, relative to the sleep rested condition, for both the in-person (left) and computerized (right) social distance tasks (13.2% and 17.7% differences, respectively). b Night-to-night increases or decreases in sleep efficiency were associated with significant day-to-day increases and decreases in loneliness, respectively. *P ≤ 0.05; error bars reflect standard error of the mean

Fig. 3

fMRI results of the in-lab study. a and b Significant clusters of activation within a priori regions of interest (ROIs) during human-relative-to-object approach, showing sleep-deprivation-related increases in activity in regions of the Near Space network (red), yet sleep-loss-related decreases in activity in regions of the Theory-of-Mind network (blue; all P < 0.05, FDR corrected). c Increased reactivity (human > object approach) within the Near Space network was positively associated with greater social distance separation following sleep deprivation. This relationship was not significant for activity in regions of the Theory-of-Mind network. **P < 0.005; error bars reflect standard error of the mean; gray lines reflect zero crossings. DIPS—dorsal intra-parietal sulcus, vPM—ventral premotor region, TPJ—temporal–parietal junction, SD—sleep deprivation, SR—sleep rested

Fig. 4

Ratings of independent judges, blind to the experimental conditions. a Change in social ratings of sleep-deprived participants (relative to their ratings in the sleep-rested condition) by independent judges (see Table S1). Following sleep deprivation, participants were rated as being significantly lonelier and less desirable to interact with. Judges also indicated feeling significantly lonelier themselves after watching videos of sleep-deprived participants. b Lonely ratings were positively associated with judges’ corresponding feeling of loneliness: the lonelier a judge rated an individual in a video, the lonelier they themselves sequentially felt. c Judges were significantly more likely to choose a sleep-rested participant to collaborate with compared to chance and significantly more likely to choose a sleep-deprived participant as their least favorite option for collaboration. d The sleep-deprivation increase in reactivity within the Near Space Network (human-only approach) of the experimental participants, correlated with the decreasing desire of judges to socially interact with these sleep-deprived individuals. This relationship was not significant for judges’ rating of perceived loneliness. *P < 0.05, **P < 0.005; correlations remain significant without each bottom outlier data point (Fig. 4b: R = 0.5, P < 0.05, Fig. 4d: R = −0.52, P < 0.05); error bars reflect standard error of the mean. SD sleep deprivation, SR sleep rested