Like Goldilocks’ porridge, sleep needs to be “just right” for a person's brain to stay healthy. In the April 28 Nature Aging, researchers led by Wei Cheng and Jianfeng Feng at Fudan University in Shanghai, China, linked getting too little or too much sleep in middle age to worse cognitive scores and mental health, as well as to a having a smaller brain, about eight years later. Analyzing data from about 150,000 older adults in the U.K., the study confirmed previous findings from smaller studies that the optimum amount of sleep is about seven hours a night. Because the data are correlative, they reveal nothing about causality, i.e., whether an abnormal amount of sleep causes cognitive decline and brain atrophy, or whether incipient Alzheimer’s or some other factor causes poor sleep.

  • Too little or too much sleep in middle age tracks with worse cognition.
  • Abnormal sleep also predicts worse cognition and mental health a decade later.
  • And it predicts a smaller hippocampus and cortex.

Other researchers called the findings valuable. “The contribution of this paper is the incredible number of individuals included, longitudinal follow-up, and multiple types of data,” Lea Grinberg at the University of California, San Francisco, wrote to Alzforum. Brendan Lucey at Washington University in St. Louis noted that this research extends previous findings on sleep and cognition to mood disorders (full comments below).

Many previous studies have found a relationship between disrupted sleep and Alzheimer’s disease, but cannot pin down which comes first (Aug 2017 conference news; Dec 2017 conference news; Mar 2018 news). Several papers reported a U-shaped association between the amount of sleep and cognitive health; that is, faster cognitive decline and more abnormal AD biomarkers in people at either extreme than in people in the “sweet spot” of seven to eight hours per night (Ma et al., 2020; Xu et al., 2020).

The Zzs U Need. People who self-reported seven hours of sleep had higher cognitive scores (left) at baseline and less depression (right) at follow-up than those at either extreme, forming typical U- and inverted U-shaped curves. [Courtesy of Li et al., Nature Aging.]

To explore the mechanisms behind this, Feng and colleagues turned to the UK Biobank cohort, a huge observational study. Joint first authors Yuzhu Li at Fudan and Barbara Sahakian at the University of Cambridge, U.K., analyzed baseline data from 498,277 mostly white participants whose ages ranged from 38 to 73, with the average being 56. Participants self-reported how much they slept; the average was 7.15 hours per night. The cohort took a variety of executive function and cognitive tests at baseline, though not everyone took all the tests. At the high end, 224,055 took the pairs-matching test, and at the low end, 33,326 took a trail-making test.

Overall, those who slept either much less or much more than the average scored worse on measures of memory, reaction time, executive function, and fluid intelligence. The findings did not change when adjusted for age, sex, and education, as well as dementia risk factors like obesity, smoking, and alcohol use.

How did abnormal length of sleep influence brain aging? The researchers examined structural MRI data from 39,692 participants scanned about eight years later. Again, they saw a U-shaped relationship with sleep. People who reported average amounts of sleep at baseline had the most gray matter at follow-up, while those at either extreme had the least. The effect was most pronounced in the precentral cortex, lateral orbitofrontal cortex, and hippocampus.

Optimum sleep at baseline came with better cognitive and mental health scores at follow-up, again forming U-shaped curves. About 40,000 people retook cognitive tests during their neuroimaging visit, while 156,884 filled out a comprehensive online mental health questionnaire during the same time frame. The more abnormal a person’s amount of sleep was at baseline, the worse they scored at follow-up on cognitive tests as well as measures of mental health such as depression, anxiety, and mania.

The consistency of a person's sleep length was also important, in that those in whom it changed the least between baseline and follow-up maintained the best cognitive and mental health.

Strikingly, the pattern of these associations changed with age. The U-shaped curves between sleep duration and brain volume, cognition, and mental health were most pronounced in younger participants, and flattened in older people. This may indicate that the effect of sleep is most crucial in middle age. Other studies have found an association between the presence of risk factors such as hypertension, diabetes, and systemic inflammation in middle age, and later dementia risk (Jun 2018 news; Feb 2019 news; Apr 2021 news).

How Things Connect. Statistical analysis revealed how different factors interacted. In the main pathway, polygenic risk scores (PRS) influenced sleep duration, which in turn affected brain volume, which acted on cognition (blue arrows). Separately, genetic variants also affected cognition via sleep duration (orange arrows) or brain volume (green arrows). [Courtesy of Li et al., Nature Aging.]

The authors ran a GWAS on the cohort to identify genetic factors associated with too little or too much sleep, and combined these into a polygenic risk score. The paper does not describe how many variants turned up, what they were, or how many were used in the PRS. Nevertheless, this genetic risk score strongly associated with sleep duration, brain volume, and cognition and mental health. In addition, sleep duration at baseline predicted brain volume at follow-up, and brain volume associated strongly with cognitive scores and emotional well-being. These relationships hint at a model where genetic factors influence sleep, which in turn affects brain atrophy, and that modulates cognition and mental health, the authors said (see image above).

Among these factors, sleep duration correlated most closely with mental health. Previous work has found that people process emotions during sleep (for review see Walker and van der Helm, 2009). 

Anders Fjell at the University of Oslo, Norway, noted that the effect of sleep on cognition is quite modest in observational studies. “It is impossible to know what causes what, and it is likely that both are caused by other variables. For instance, many types of brain pathology and changes in normal aging affect both cognitive function and sleep,” he wrote to Alzforum (full comment below). Grinberg agreed that underlying factors such as pathological changes in early AD are probably responsible for the association between cognition and sleep. “Studies in humans suggest that AD pathology affects sleep-modulating areas before affecting memory-related areas,” she noted.

Overall, this paper points toward mechanisms that should be investigated further, Lucey said. The authors wrote to Alzforum that they plan to do this, and will track how the interaction between sleep, cognition, and mental health changes across the lifespan.—Madolyn Bowman Rogers


  1. Previous studies had shown that seven hours is a kind of magic number when considering ideal hours of sleep in the population. The contribution of this paper is the incredible number of individuals included, longitudinal follow-up, and multiple types of data. This allowed the authors to subdivide the group (by age, for instance) and also minimize artifactual measurement variations that always happen when individuals are tested by different professionals and using various machines. As others have done on a smaller scale, this work shows that poor sleep tends to precede cognitive decline.

    That said, this study is not suitable to answer the question of whether [poor] sleep causes cognitive decline or if underlying factors cause both. Studies in humans suggest that AD pathology affects sleep-modulating areas before affecting memory-related areas, supporting the latter hypothesis. Experimental studies suggest a bi-directional feeding mechanism between worse sleep and worse cognition.

    The bottom line, in my opinion, is that some therapies under development for neurodegenerative diseases, especially AD and LBD, will also likely help treat sleep disturbances in adults, because these symptoms have overlapping pathological mechanisms. Because the glymphatic system is more active during sleeping, some argue that improving sleep, even if artificially, may prevent or delay the spreading of AD pathology. Further studies on humans are needed to answer this question.

  2. This paper builds on previous work showing that short and long self-reported sleep duration is associated with poorer cognitive performance and worse measures of mental health such as anxiety and depressive symptoms. Similar non-linear relationships between sleep and cognitive performance have been shown in prior work, including our recent paper in Brain (Lucey et al., 2021), using objective sleep measures such as sleep duration, time in different sleep stages, and average non-rapid eye movement slow wave activity.

    Li et al. leverage the richly characterized participants from the UK Biobank study to extend these findings to include mental health symptoms and how these relationships may be mediated by changes in different brain regions and genetic factors. Many of the conclusions in this paper need to be replicated, but the findings have implications for using sleep as a marker and/or potential therapeutic intervention in AD and other brain disorders. 

    Sleep duration likely has a “sweet spot” that has the greatest benefit for brain health, and this paper points to potential mechanisms that warrant further investigation. Future research is needed to establish how these mechanisms mediate the relationship between sleep duration, cognition, and mental health, as well as if they change with age.


    . Sleep and longitudinal cognitive performance in preclinical and early symptomatic Alzheimer's disease. Brain. 2021 Oct 22;144(9):2852-2862. PubMed.

  3. It is well known that short and long sleep are both associated with lower-than-average cognitive function, although it is important to stress that effect sizes are quite modest in observational studies. It is impossible to know what causes what, and it is likely that both are caused by other variables. For instance, many types of brain pathology and changes in normal aging affect both cognitive function and sleep, but it is not clear whether the two are directly and causally related.

    Still, this study confirms previous studies showing that shorter sleep than the often-recommended seven to eight hours does not necessarily come with lower cognitive function or smaller brain volumes.”

Make a Comment

To make a comment you must login or register.


News Citations

  1. New Ties between AD and the Stages, Waves, and Molecules of Sleep
  2. Disturbed Sleep Exerts Toll on Memory and Neurodegeneration
  3. Does Daytime Drowsiness Foreshadow Aβ Accumulation?
  4. Blood Pressure Affects Dementia Risk Only in People Under 60
  5. Midlife Peripheral Inflammation May Drive Later Cognitive Decline
  6. Diabetes in Mid-Life Drives Up Dementia Risk

Paper Citations

  1. . Association Between Sleep Duration and Cognitive Decline. JAMA Netw Open. 2020 Sep 1;3(9):e2013573. PubMed.
  2. . Sleep characteristics and cerebrospinal fluid biomarkers of Alzheimer's disease pathology in cognitively intact older adults: The CABLE study. Alzheimers Dement. 2020 Aug;16(8):1146-1152. Epub 2020 Jul 13 PubMed.
  3. . Overnight therapy? The role of sleep in emotional brain processing. Psychol Bull. 2009 Sep;135(5):731-48. PubMed.

Further Reading

Primary Papers

  1. . The brain structure and genetic mechanisms underlying the nonlinear association between sleep duration, cognition, and mental health. Nat Aging. 2022 Apr 28. Nature Aging