Abstract
Although both physical activity (PA) and sleep have been linked to cognitive functioning, many questions remain with regard to the nature of their relationships. Pathways that link sleep and PA to cognition have been studied separately, but recent work suggests that PA and sleep are also related. Studies have found that more active people sleep better and longer than those who are inactive. Within persons, sleep quantity and quality are better on days when one is more active. Further investigation is needed to understand the nature of the relationship between PA and sleep and their contribution to cognitive functioning. Findings have differed as a function of whether the measures used were based on self-reports or objective measures, suggesting the need to compare results from different modes of assessments. Although both PA and sleep are related to cognition, little is known about how PA and sleep work in concert to affect cognition. These relationships were tested in 3 studies using different assessment methods and study designs.
The first study was a four-week walking intervention in a sample of middle-aged and older adults, with a newly developed smartphone app, StepMATE, that included setting daily walking goals with behavioral supports. Average daily steps during the four-week intervention were significantly higher than the week before participants downloaded the app. Cognition improved from pre- to post-test, however no changes in sleep were found. Within-person analyses revealed that on days in which females were more active than average, they reported better mood and higher energy. These daily relationships were not significant for males. Results suggest that self-monitoring and goal setting may be sufficient to increase daily walking but understanding how activity is related to sleep and cognition requires further study.
The second study more closely examined daily relationships between PA and sleep, including both objective and subjective measures over the course of one week. Self-reports of sleep were significantly correlated with objective measures, although there were significant mean differences for total sleep time (TST), sleep latency, and waking after sleep onset (WASO), with greater differences for males and non-white participants. Self-reported, but not objective TST was significantly correlated with executive functioning (EF). Both objective and subjective latency were related to cognition; those with shorter objective latencies had better concurrent episodic memory (EM), and those with shorter self-reported latencies had better EF 10 years later. Objective WASO was related to EM; those who woke up more often had worse concurrent EM. Those with better episodic memory had smaller differences between self-reported and objective sleep. Daily analyses revealed concurrent and lagged relationships between PA and both subjective and objective sleep. On days in which one is more active, they have shorter TST and wake up less. On nights in which one sleeps less or wakes up less, they engage in more PA the following day. Thus, both subjective and objective sleep seem important to PA and cognition, and results support prior work that found bidirectional effects of daily PA and sleep.
Finally, the third study examined concurrent and longer-term longitudinal relationships between PA, sleep, and cognitive functioning over 10 years. As predicted, those who were more physically active had better concurrent cognitive functioning (EM and EF). Contrary to hypotheses, neither sleep quality nor duration predicted concurrent EM. Sleep quality and duration both significantly predicted EF, but in opposite directions. Those who reported better sleep quality had worse EF, and those with healthier sleep durations (closer to 7.5 hours) had better EF. Change in sleep duration significantly predicted change in EF, such that those whose sleep duration declined less over the course of 10 years also experienced less EF decline. PA did not significantly predict concurrent or change in sleep quality or duration. To address questions of directionality, a reverse model was also tested. Sleep significantly predicted PA, such that those with better sleep quality engaged in less PA, and those with healthier sleep durations engaged in more PA. Sleep did not significantly mediate the relationship between PA and cognition, but PA was a significant mediator in the sleep and cognition relationship. Sleep duration appears to be a better predictor of EF than EM, both in terms of level of EF and subsequent change. Again, results from the bidirectional models suggest that sleep predicts PA engagement.
Together, these three studies add to our understanding of the relationships between PA, sleep, and cognitive functioning. PA and sleep are related at both the between- and within-person levels. More active people sleep less and awaken less frequently, on average, and days in which people are more active they sleep less and wake up fewer times. Subjective reports of sleep were correlated with objective measures, although men, non-white participants, and those with poorer episodic memory showed greater differences between assessment methods. Significant correlations with cognition emerged for both subjective and objective reports of sleep; both measurement types seem important to consider. Results from testing alternative longitudinal models provide further evidence that relationships between PA and sleep are bidirectional. Findings from the mediation analyses suggest that PA is a mechanism linking sleep and cognitive functioning. The results echo prior findings that sleep durations that are either too short or too long are not ideal, and future studies should continue to consider the curvilinear relationships between sleep duration and outcome variables such as cognition and health.