Cognitive aging

Cognitive abilities do not change at the same rate. Abilities that depend on processing new information quickly, such as reasoning, short-term memory and verbal fluency, decline from midlife, whereas vocabulary and accumulated knowledge are maintained [1]. In Whitehall II, 5,198 men and 2,192 women aged 45 to 70 took the same tests three times over ten years. All scores except vocabulary declined in every age group, including those aged 45 to 49 at the start. Among men, the ten-year fall in reasoning was 3.6% of the test range in those aged 45 to 49 and 9.6% in those aged 65 to 70; among women the corresponding figures were 3.6% and 7.4% [1].

The same study showed why comparing age groups at a single point in time can mislead. Among women aged 45 to 49, following the same people over time showed a 3.6% decline in reasoning, whereas comparing younger and older women at one time suggested an 11.4% decline, because older generations had received less education [1]. Differences between generations therefore inflate cross-sectional estimates of cognitive aging.

Two main methods express cognitive aging in years. Cognitive test ages compare a person's scores on a battery of tests with the average scores of people of different ages. Brain-age models are trained on brain scans of many healthy people to predict age from brain structure; the difference between predicted and actual age, called the brain age gap, is positive when a brain appears older than expected. One such model, trained on 1,537 healthy adults, predicted age with a correlation of 0.92 [8]. The two methods measure different things: a test age reflects performance on the day of testing, while a brain age reflects long-term changes in brain tissue.

Sleep has one of the best-described relationships with test performance. In over 10,000 adults who completed 12 cognitive tests online, reasoning and verbal ability were best at 7 to 8 hours of habitual sleep and worse with both less and more; short-term memory was unaffected, and a typical 4-hour night was equivalent to 8 years of aging in overall cognition [2]. The relationship was the same at all adult ages. The physiology of sleep is described under sleep regulation and sleep stages.

Hearing loss is associated with lower scores on memory and executive function tests (tests of planning and mental flexibility). In the Baltimore Longitudinal Study of Aging, the reduction in scores associated with 25 decibels of hearing loss equalled that associated with 6.8 years of age [4]; hearing and vision loss are covered in a separate article.

Conditions that damage blood vessels are associated with older brain age. In 185 older adults, those with type 2 diabetes had brains estimated 4.6 years older than their actual age, and the gap grew by 0.2 years for each year of follow-up [3]. In 686 adults scanned repeatedly over 12 years, every 10 mm Hg of blood pressure above optimal levels was associated with a brain about 51 to 66 days older, and a person with ideal blood pressure was predicted to have a brain more than 6 months younger at midlife than a person who became hypertensive [10].

Other conditions carry smaller or less precise estimates. In an analysis of 19 samples from the ENIGMA consortium, people with major depression had brains 1.08 years older than controls, a difference not explained by medication, remission or severity [11]. After traumatic brain injury, grey matter appeared 4.66 years older and white matter 5.97 years older than actual age, and the gap increased with time since injury [8]. Daily or near-daily tobacco and alcohol use were both associated with older relative brain age in 12,115 UK Biobank adults [12], and in 36,678 UK Biobank adults lower brain volumes were already apparent at an average of one to two alcohol units a day [13]. In a New Zealand birth cohort, social isolation in adulthood was associated with older brain age at 45 [14].

Education and everyday physical activity are associated with younger brain age. In 331 healthy adults aged 19 to 79, each year of education was associated with a brain 0.95 years younger and each flight of stairs climbed daily with a brain 0.58 years younger [5]. In 2,354 Framingham Heart Study participants wearing activity monitors, each additional hour of light activity was associated with about 1.1 years less brain aging, and among people below the activity guidelines, taking 7,500 or more steps a day was associated with about 1.4 to 2.2 years less [9]. The wider evidence on activity is covered under exercise and longevity.

Mentally demanding activities show similar associations. At age 50, 50 long-term meditators had brains estimated 7.5 years younger than 50 matched controls [15]. Amateur musicians had lower brain-age scores than both professional musicians and non-musicians [16]. Among 19,078 adults aged 50 to 93, people who did word puzzles more often scored better on all 14 cognitive measures studied [17]. In a study of over 1,000 adults, more time spent playing video games was associated with better cognitive scores, while physical activity was associated with better mental health but not with cognitive scores [18]. Among 184 memory-clinic patients with dementia, lifelong bilinguals developed symptoms 4 years later than monolinguals [19].

Diet shows associations with brain structure. In 674 older adults (mean age 80), higher adherence to a Mediterranean diet was associated with 13.1 ml more total brain volume, an effect similar to 5 years of aging, with higher fish and lower meat intake contributing most [20]. In 3,846 Framingham participants, drinking one to two sugary drinks a day was associated with lower total brain volume, and daily fruit juice with lower total and hippocampal volumes [21]. Taken together, lifestyle explains only a small part of the differences between people: in almost 10,000 adults, 20 lifestyle choices together explained 7 to 9% of the variation in cognitive scores [22].

Full deep dive: Cognitive age and lifestyle. The combined estimates are shown on the cognitive age chart.

Exercise has the most trial evidence. In a trial of 120 older adults, a year of aerobic exercise increased the volume of the hippocampus (the brain's memory centre) by 2%, reversing about 1 to 2 years of age-related loss [23]. In a trial of 130 adults aged 26 to 58, twelve months of moderate-to-vigorous aerobic exercise (about 150 minutes a week) reduced brain age by 0.60 years while it rose by 0.35 years in controls, a difference of about 0.95 years [6]. In the LISA trial of 309 older adults, heavy or moderate resistance training reduced brain age, estimated from brain activity scans at one and two years, by 1.4 to 2.3 years compared with no exercise [24].

COSMOS randomly assigned 21,442 US adults aged 60 and over to a daily multivitamin-mineral supplement or placebo. In a combined analysis of three cognitive substudies, the multivitamin improved global cognition and episodic memory (memory for events), by an amount the authors estimated as equivalent to 2 years less cognitive aging [7]. Blood pressure treatment has trial evidence for a related outcome: in SPRINT MIND, a systolic target below 120 mm Hg reduced mild cognitive impairment by 19% compared with a target below 140 mm Hg [25].

Normal cognitive aging, mild cognitive impairment and dementia lie on a continuum. Mild cognitive impairment is decline greater than expected for age that does not yet prevent independent daily life; dementia is decline that does. Many factors associated with faster cognitive aging, including high blood pressure, diabetes, hearing loss, smoking, physical inactivity and social isolation, are also dementia risk factors. Smoking illustrates the overlap: in Whitehall II, men who smoked declined faster in global cognition and executive function than never-smokers over ten years, while men who had stopped at least ten years earlier declined at the same rate as never-smokers; no difference was seen in women [26].

A younger cognitive or brain age is not a measure of dementia risk. Brain-age models capture average patterns of tissue loss, and the diseases that cause dementia, such as Alzheimer's disease, have their own biology that may progress in people whose test scores are still normal.

Most estimates in this field come from cross-sectional studies, which cannot show whether a habit slows cognitive aging or whether people with sharper minds are more likely to adopt it. Reverse causation is likely for activities such as puzzles, music and gaming. Several brain-age studies are small, including those of meditators (100 people) and diabetes (185 people), and different studies use different brain-age models, so their year figures are not directly comparable. Online cognitive studies recruit volunteers who may differ from the general population, and self-reported sleep, diet and activity are measured with error. Trials are few, short and measure intermediate outcomes such as brain volume or test scores rather than long-term cognitive health.