Immediate Recall of Faces with Positive Emotions Correlates with Density of Eye Movements During REM Sleep In Children with Autism

Background: Among multiple variants in their sleep phenotype (Limoges at al 2005) autistics show original REM sleep measures. Specifically, autistics display less rapid eye movements than typical individuals. Rapid eye movements of REM sleep are generated by a limbic system-based neural network that includes the amygdala. This structure is implicated in the socio-cognitive function in which autistics display atypicalities, in the form of diminished overt and neural responses to emotional loaded stimuli

Objectives: Investigate the relationship between rapid eye movements density during REM sleep, a marker of limbic system functioning, and performance on a recognition performance task involving emotional stimuli. A positive correlation between performance in this task and amounts of rapid eye movements during REM sleep in autism is predicted, on the basis of the typical relation between this structure and this behavior.

Methods: Thirteen male autistic children (AUT; 10.3 ± 1.9 years) and thirteen comparison children (COM; 9.6 ± 2.2 years) were recorded for two consecutive nights. Before and after night 2, participants were administered a recognition task of unfamiliar face showing positive, negative or neutral emotions; immediate recall (in the evening) and delayed recall (in the morning) were also tested. The total number of correct answers on immediate and on delayed recall were computed; the number of omission errors were computed separately for positive, negative and neutral faces for immediate and for delayed recall. The duration of REM sleep and the number of rapid eye movements during REM sleep were computed for night 2. Results from the two groups were compared using Mann-Whitney U-tests. The correlation between REM sleep measures and face recognition was tested using Spearman’s rho. Significance level was set at .05.

Results:  REM sleep time (AUT = 97.2 ± 23.5, COM = 87.9 ± 17.0) and number of rapid eye movements per hour of REM sleep (AUT = 514.5 ± 133.5, COM = 462.2 ± 116.4) were comparable in the two groups (see also Lambert A. et al., this meeting). However, the correlation pattern differs between groups. Whereas immediate recall of faces with positive emotions was positively correlated with the number of rapid eye movements per hour of REM sleep in autistic children (rho = 0.62), it was not the case in COM children (rho = 0.10). In contrast, only COM children showed a significant positive correlation between the duration of REM sleep and the number of errors on neutral faces on immediate recall (rho = 0.57), a correlation not found in AUT children (rho=0.28). This different pattern of correlation may be used to interpret the inferior performance of autistic children on immediate recall of faces with positive emotions (errors: AUT=3.8 ± 2.0, COM=2.1 ± 1.8).

Conclusions: Although correlations do not allow drawing causative relationships, these findings indicate that the neural networks dedicated to the recognition of faces with positive emotions relate differently with sleep variables in typically developing and in HFA children.