Why do both too little and too much sleep look bad for the brain?
The pattern that keeps appearing is a U-shape: people sleeping around seven to eight hours tend to score best, while those at either end do worse. In a large group of older adults with normal cognition, short sleep of six hours or less was linked to lower memory scores, and long sleep of nine hours or more was linked to worse performance across several thinking domains [1]. A separate survey of Chinese adults aged 55 and over found the same two-sided pattern: sleeping six hours or less raised the odds of a low global cognitive score by about 23%, and sleeping ten hours or more raised them by about 47% [4].
The important difference is what sits behind each end. Short sleep in that older-adult study went hand in hand with higher amyloid-beta burden, the protein buildup associated with Alzheimer's disease, while long sleep did not [1]. Long sleepers instead tended to have higher body mass index, more depressive symptoms, and more daytime napping [1]. That fits a picture in which long sleep is often a marker of other health problems rather than a direct cause of cognitive decline.
Does poor sleep cause cognitive decline, or does cognitive decline cause poor sleep?
A seven-year study of nearly 8,000 Chinese adults aged 45 and older tracked both sleep and cognition over four waves and found the relationship runs both ways. Short or long sleep predicted lower global cognition in the next wave, and lower cognition also predicted short or long sleep later [2]. The second direction was stronger: the association from cognition to sleep was larger than the reverse, suggesting that declining cognition may be the bigger driver of unusual sleep patterns in this age group [2].
Genetic evidence adds a sharper contrast. Using naturally occurring gene variants as stand-ins for sleep habits, researchers found that short sleep was causally linked to lower cognitive performance, fluid intelligence, and memory, and to slower completion of a trail-making test of attention and processing speed [3]. For long sleep, there were too few genetic markers to draw a firm conclusion, so the causal status of long sleep remains unresolved [3].
Does the same pattern hold at every age?
In young children, the answer looks somewhat different. Among about 2,100 preschoolers followed from six months to four years, those on a persistently short total-sleep trajectory scored lower on visual-spatial reasoning and were more likely to have a low full-scale IQ than children on a medium trajectory [5]. Long total sleep did not show the same penalty; in fact, medium and long total-sleep trajectories were both associated with better cognitive performance than the short trajectory [5]. The researchers also found that adequate nighttime sleep combined with moderate daytime sleep was linked to higher IQ scores [5].
So the two-sided risk seen in older adults is not a universal rule. In early childhood, too little sleep appears to be the clearer concern, while in later life long sleep becomes a signal worth noticing, possibly because it reflects underlying health changes rather than a sleep problem in its own right [1][5].
What should a person take away from this?
The practical message is that seven to eight hours remains a reasonable target, and that regularly sleeping much less is the pattern with the strongest evidence of harm to thinking and memory [1][3][4]. Sleeping much more than usual is also worth paying attention to, but mainly as a possible sign of another issue, such as depression, weight-related health problems, or early cognitive change, rather than as a direct cause of decline [1][2].
One caveat matters for how much weight to put on any single number: most of these findings come from single cohorts rather than pooled analyses, and they rely largely on people's own reports of how long they sleep [1][2][4]. The genetic study offers the strongest hint of cause and effect for short sleep, but it could not settle the question for long sleep [3].
Sources used in this answer
Association of Short and Long Sleep Duration With Amyloid-β Burden and Cognition in Aging
In older adults with normal cognition, short sleep was linked to higher amyloid-beta burden and lower memory scores, while long sleep was linked to worse performance across multiple cognitive domains but not to amyloid burden.
Exploring the Bidirectional Associations Between Short or Long Sleep Duration and Lower Cognitive Function: A 7-Year Cohort Study in China
In a seven-year Chinese cohort, short or long sleep predicted later lower global cognition and lower cognition predicted later short or long sleep, with the cognition-to-sleep direction being stronger.
Mendelian randomization study of self-reported long sleep duration, short sleep duration, and insomnia and cognitive function
Genetic analysis suggested short sleep causally worsens cognitive performance, fluid intelligence, memory, and trail-making scores, but too few genetic markers existed to assess long sleep.
Both Short and Long Sleep Durations Are Associated with Poor Cognition and Memory in Chinese Adults Aged 55+ Years—Results from China Health and Nutrition Survey
In Chinese adults aged 55 and over, both short (six hours or less) and long (ten hours or more) sleep were associated with higher odds of low global cognitive scores and poor memory.
Early sleep duration trajectories and children’s cognitive development: a prospective cohort study
Preschoolers on a persistently short total-sleep trajectory had poorer visual-spatial reasoning and higher risk of low IQ than those on a medium trajectory, while medium and long trajectories were both linked to better performance than short.