Sleep deprivation impairs the accurate recognition of human emotions

Abstract

Study objectives: Investigate the impact of sleep deprivation on the ability to recognize the intensity of human facial emotions.

Design: Randomized total sleep-deprivation or sleep-rested conditions, involving between-group and within-group repeated measures analysis.

Setting: Experimental laboratory study.

Participants: Thirty-seven healthy participants, (21 females) aged 18-25 y, were randomly assigned to the sleep control (SC: n = 17) or total sleep deprivation group (TSD: n = 20).

Interventions: Participants performed an emotional face recognition task, in which they evaluated 3 different affective face categories: Sad, Happy, and Angry, each ranging in a gradient from neutral to increasingly emotional. In the TSD group, the task was performed once under conditions of sleep deprivation, and twice under sleep-rested conditions following different durations of sleep recovery. In the SC group, the task was performed twice under sleep-rested conditions, controlling for repeatability.

Measurements and results: In the TSD group, when sleep-deprived, there was a marked and significant blunting in the recognition of Angry and Happy affective expressions in the moderate (but not extreme) emotional intensity range; differences that were most reliable and significant in female participants. No change in the recognition of Sad expressions was observed. These recognition deficits were, however, ameliorated following one night of recovery sleep. No changes in task performance were observed in the SC group.

Conclusions: Sleep deprivation selectively impairs the accurate judgment of human facial emotions, especially threat relevant (Anger) and reward relevant (Happy) categories, an effect observed most significantly in females. Such findings suggest that sleep loss impairs discrete affective neural systems, disrupting the identification of salient affective social cues.

Figure 1

(A) Experimental design: Participants in each group performed the emotional face recognition task twice; once on Day 2 (16:00) and once on Day 3 (16:00). The experimental manipulation occurred across the night before the first test session (Test1). Participants in the sleep control group were awake across Day 1 and slept normally across Night 1 before Test1 on Day 2. Participants in the total sleep deprivation group were similarly awake across Day 1, but were subsequently kept awake across Night 1, and across Day 2 accumulating 30 h of total sleep deprivation prior to Test1. Following Test1, both groups had a normal night of sleep (Night 2), and repeated the same task a second time on Day 3 (Test2). (B) Example of the emotional face stimuli gradient used for the affective category HAPPY.

Figure 2

Emotional face recognition performance in the total sleep deprivation group. Upper left panel: Average rating of face stimuli across all three emotional categories (ANGRY, HAPPY, SAD) combined at Test1 (clear bar, when sleep deprived) and Test2 (filled bar, when sleep recovered), together with response curves at Test1 (clear squares, when sleep deprived) and Test2 (filled squares, when sleep recovered) for the rating of the faces stimuli when plotted across the emotional intensity gradient, from low to high (left to right). Remaining panels are mirror data for each of the emotions separately, labeled respectively on each panel. †P < 0.08; *P < 0.05; **P < 0.01; n.s. Nonsignificant. Error bars represent standard error of the mean.

Figure 3

Emotional face recognition performance in the sleep control group. Upper left panel: Average rating of face stimuli across all three emotional categories (ANGRY, HAPPY, SAD) combined at Test1 (clear bar) and Test2 (filled bar) and response curves at Test1 (clear squares) and Test2 (filled squares) for the rating of the faces stimuli when plotted across the emotional intensity gradient, from low to high (left to right). Remaining panels are mirror data for each of the emotions separately, labeled respectively on each panel. †P < 0.09; n.s. Nonsignificant. Error bars represent standard error of the mean.