Fifty-three percent. That was the headline in 2015, when researchers in the journal Alzheimer’s & Dementia reported on 923 people aged 58 to 98 followed for about four and a half years. Those whose eating most closely matched a pattern called the MIND diet had a 53 percent lower rate of Alzheimer’s disease than those whose eating matched it least.
MIND stands for Mediterranean-DASH Intervention for Neurodegenerative Delay. It blends the Mediterranean diet with DASH, an eating plan built to lower blood pressure, and tilts toward foods earlier studies had tied to healthier brains: green leafy vegetables, nuts, berries, fish and olive oil.
Eight years later the diet got the test that matters most, a randomized trial, and did no better than a comparison diet that simply cut calories. That doesn’t make leafy greens bad for you. It means the most famous brain diet is still a hypothesis, and that the clearest links between what you consume and how you think lie elsewhere: in a vitamin deficiency a clinician can test for, and in alcohol.
How the MIND diet trial was built, and what it found
The trial, funded by the National Institute on Aging and published in The New England Journal of Medicine in 2023, enrolled 604 adults aged 65 and older in the Chicago and Boston areas. None had cognitive impairment, but all had three features that raise concern: a close relative with dementia, a body-mass index of 25 or higher, and a diet that scored poorly on a short questionnaire.
For three years, half followed the MIND diet and half kept their usual foods, with both groups asked to eat about 250 fewer calories a day and given equal telephone coaching from dietitians. The MIND group received regular deliveries of blueberries, mixed nuts and olive oil; the comparison group got $30 gift cards on the same schedule.
Ninety-three percent finished. On the investigators’ 14-point diet score, the MIND group rose from 7.7 to 11.0 and the comparison group from 7.8 to 8.5. Both groups lost similar amounts of weight, about 11 pounds.
Scores on a 12-test battery improved in both groups, by 0.205 standard units with MIND and 0.170 without it. The gap of 0.035 was about one-sixth of the improvement itself, and it was small enough to be chance. Brain scans in 200 participants showed similar changes in both groups.
Two caveats apply. The volunteers were mostly well educated and of European descent, about 10 percent Black, so the result may not hold for everyone. And the margin of error still allows a modest benefit. But the researchers had sized the study using estimates from long-term observational data, so it was built to catch an effect about as large as those studies implied, and it didn’t find one.
Graded on the library’s evidence scale, the MIND diet as protection for thinking in at-risk older adults has been tested once and not supported.
The Mediterranean trial with an asterisk
The other randomized evidence comes from Spain. PREDIMED was a large trial of 7,447 people aged 55 to 80 at high risk of heart disease. They were assigned to a Mediterranean diet with extra olive oil, the same diet with extra nuts, or advice to cut dietary fat.
At its Barcelona site, researchers tested thinking in 447 participants and retested 334 of them after a median of about four years. Their 2015 report in JAMA Internal Medicine found the olive oil group did better on composite scores for attention and overall thinking, and the nut group did better on a memory composite. Most individual tests showed no difference.
One detail changes the reading: the comparison group’s scores declined on all three composites. So the diet groups mostly held steady or edged up while the low-fat advice group slipped, and the authors’ own conclusion described the diets as associated with better thinking rather than claiming they caused it.
Then the asterisk. In 2018 the main PREDIMED report, published in 2013, was retracted and republished in The New England Journal of Medicine after the investigators found protocol problems. Some household members had been enrolled without randomization, some participants at one of 11 sites were assigned without it, and another site appeared to use its randomization tables inconsistently. The reanalysis still showed fewer major cardiovascular events, counting heart attacks, strokes and cardiovascular deaths together, with both Mediterranean diets.
The brain substudy itself has not been retracted, but it remains a secondary analysis of one site’s volunteers from a trial whose randomization proved imperfect. Verdict: positive trial evidence with a contested interpretation.
Why food studies tend to flatter the people who eat well
Why did observational studies look so much better than the trials? The MIND trial’s authors offered a list that doubles as a guide to any brain-food headline.
People who eat well differ in many other ways. Someone who eats leafy greens and fish most days may also have more education, more money and better medical care. Researchers adjust for the differences they measure, never for the ones they missed. The authors named bias and confounding, meaning those hidden third factors, as one likely reason the two kinds of evidence disagree.
Diet is measured by memory. The 2015 MIND study assessed diet with a food-frequency questionnaire, which asks people to recall how often they ate dozens of foods. That is a rough instrument, and it leans on the same faculty, memory, that the research is trying to protect.
Comparison groups don’t stand still. The MIND trial’s calorie-cutting group lost as much weight as the MIND group, and its diet score crept up too. Coaching both groups toward healthier eating shrinks the contrast a trial can measure.
Retaking tests raises scores. Both trial groups improved over three years. The authors wrote in the New England Journal of Medicine: “It is also plausible that practice effects of repeated cognitive testing could account for improvement in both of our trial groups…” In other words, people get better at a test simply by taking it again.
The authors also listed the plain possibility that the diet doesn’t improve thinking, or that three years is too short to tell. The National Institute on Aging’s consumer guidance agrees: no evidence yet shows that eating or avoiding any specific food prevents Alzheimer’s, and research on whether diet can delay it is still underway.
Vitamin B12: a nutrition problem a clinician can test for
The nutrition link to thinking with the firmest footing is not a superfood. It is a deficiency.
Vitamin B12 is needed to make red blood cells and DNA and for the nervous system to develop and work. According to the NIH Office of Dietary Supplements, a shortage can cause fatigue, a type of anemia with oversized red cells, and numbness and tingling in the hands and feet. The office also lists dementia among its possible effects. Two features make it easy to miss. Nerve-related symptoms can appear without any anemia, and because the body keeps a reserve, they can take several years to show up. The office stresses that early diagnosis and treatment matter to avoid irreversible damage.
The office’s risk figures explain why age matters so much. It describes atrophic gastritis as an autoimmune condition that cuts the stomach’s output of acid and of intrinsic factor, a protein needed to absorb B12. Its reach grows with age: roughly 1 person in 50 overall, but 8 or 9 in every 100 adults past 65. Pernicious anemia, an autoimmune disease of the stomach lining that stops intrinsic factor production altogether, accounts for roughly 15 to 25 percent of B12 deficiency in older adults. Medicines matter as well: the office reports that metformin and acid-reducing drugs can lower B12 absorption, and it says people taking such medicines regularly should discuss their B12 status with their health care providers.
Testing is less simple than one number. The office notes that the cutoff for deficiency differs by laboratory and method, and that a second marker, methylmalonic acid, is more sensitive but also rises with kidney problems and with age. Reading a borderline result is a clinician’s job.
A real deficiency is a medical problem. Extra B12 as a general brain booster is a different question: the office reports that randomized trials giving B12, alone or with other B vitamins, to older adults with or without dementia have not improved thinking. That supplement evidence belongs to the coming article on brain-health supplements. Here, a deficiency is a reason to ask about testing, not to stock up on pills.
Alcohol: the protective glass that wasn’t
Observational studies have often drawn a J-shaped curve, with light drinkers faring better than both heavy drinkers and nondrinkers. A daily glass of wine for the brain became folk wisdom. Newer evidence has taken most of that back.
What heavy drinking does
The 2024 report of the Lancet Commission on dementia counts excessive drinking among 14 modifiable risk factors, a list our earlier dementia-risk article walks through one by one. Its threshold is more than 21 UK units a week in midlife, or 168 grams of alcohol. By the National Institute on Alcohol Abuse and Alcoholism’s definition of a US standard drink as about 14 grams, that is roughly 12 drinks a week. An analysis of 131,415 people cited in the 2024 report found that heavier midlife drinking went with a 22 percent higher dementia risk.
The Commission notes that many people counted as nondrinkers once drank heavily, which makes abstainers look worse, and it reports a lack of clear evidence that not drinking raises dementia risk.
What lighter drinking shows
- Brain scans over 30 years. A 2017 BMJ study tracked 550 British adults for three decades before scanning their brains. Compared with not drinking, 14 to 21 UK units a week, about 8 to 12 US standard drinks, went with more than three times the odds of shrinkage in the right hippocampus, a memory structure. The authors stated: “There was no protective effect of light drinking (1-<7 units/week) over abstinence.” Drinking was not linked to most of the thinking tests.
- Scans of 36,678 adults. A 2022 UK Biobank study in Nature Communications found lower brain volume already at an average of one to two UK units a day, roughly half of one US standard drink to a little over one. Going from one to two daily units matched about two years of brain aging in a 50-year-old. But alcohol explained only about 1 percent of the differences in gray matter, the scans were a single snapshot, and volume is not memory.
- Genes as a natural experiment. A 2026 study in BMJ Evidence-Based Medicine followed 559,559 adults in the US Million Veteran Program and UK Biobank. The ordinary analysis reproduced the familiar curve, with higher dementia risk in nondrinkers, in people having more than 40 drinks a week and in people with alcohol use disorder. Mendelian randomization, a method that uses inherited variants nudging how much people drink, instead showed risk rising steadily with intake and no protection at low levels. People who went on to develop dementia had also cut back on drinking over time, a pattern that can make light drinking look protective. Genetic methods carry their own assumptions, so this is a strong pointer, not proof.
What the official reviews say
A 2024 review by the National Academies of Sciences, Engineering, and Medicine, requested by Congress, defined moderate drinking as up to two drinks a day for men and one for women. It could not draw any conclusion about moderate drinking and dementia, cognitive decline or Alzheimer’s. It did conclude, with moderate certainty, that compared with never drinking, moderate drinking is linked to lower overall mortality and to a higher risk of breast cancer.
The 2025-2030 Dietary Guidelines for Americans came out in January 2026 from the Department of Health and Human Services and the Department of Agriculture. They advise drinking less for better overall health but drop the numeric daily limits of earlier editions. They do not mention cognition or dementia. They say some people should avoid alcohol completely: pregnant women, people recovering from alcohol use disorder or unable to control their drinking, and people taking medicines or living with conditions that interact with alcohol. The American Association for the Study of Liver Diseases objected that the edition drops daily limits, ignores how differently men and women process alcohol, and says nothing about cancer.
The fair summary: light drinking as brain protection is not supported, harm at heavy intake is consistent across study types, and at light-to-moderate levels the scan findings lean negative but remain observational. Whether a medicine you take mixes badly with alcohol is a question for your prescriber or pharmacist, and the medicines people take for sleep get a closer look in our article on sleep and long-term brain health. If you can’t control how much you drink, the guidelines’ own advice is to avoid alcohol entirely, and a clinician can help with how.
Ultra-processed food: three countries, one signal, no trial
Ultra-processed foods are industrial products such as packaged snacks, sugary drinks and many ready-to-eat meals, usually classified with a system called NOVA. Three large cohorts have now linked them to thinking:
| Study | Who | What it found |
|---|---|---|
| JAMA Neurology, 2023 | 10,775 Brazilian public servants aged 35 to 74, followed a median of 8 years | People above the lowest quarter of intake showed 28 percent faster decline in overall thinking scores and 25 percent faster decline in executive function |
| Neurology, 2022 | 72,083 UK Biobank adults aged 55 and older, followed a median of 10 years | Each 10 percent increase in the diet’s share of ultra-processed food went with a 25 percent higher dementia risk |
| Neurology, 2024 | 14,175 Black and white US adults in the REGARDS study | Each 10 percent increase went with a 16 percent higher risk of cognitive impairment, regardless of how closely people followed the MIND, Mediterranean or DASH patterns |
Consistency across three countries is meaningful. But all three are observational, and ultra-processed eating tracks with income, education and overall diet, the same hidden factors that tripped up the MIND story. No trial has tested it. Verdict: positive evidence with a contested interpretation. The new federal guidelines push hard against what they call highly processed foods, but for general health, and never use the term ultra-processed.
Signs it’s worth raising B12 with a clinician
Any item below is a reason to ask whoever manages your care whether checking B12 makes sense for you. It is not a diagnosis, and it is not a reason to start a supplement on your own.
Symptoms worth mentioning together
– New numbness, tingling or pins-and-needles in your hands or feet
– Memory or thinking changes, or low mood, without an obvious cause
– Tiredness or paleness that doesn’t lift
Circumstances that raise the odds
– You take metformin, a common diabetes medicine, regularly
– You take an acid reducer, a proton pump inhibitor or H2 blocker, most days
– You have had stomach surgery or weight-loss surgery
– You eat no animal foods. If you are also breastfeeding, the Office of Dietary Supplements notes that an exclusively breastfed baby of a mother who eats no animal products may have very limited B12 reserves, a question for the pediatrician as well
– You have been told you have atrophic gastritis or pernicious anemia
Questions that keep the visit focused
– Given my medicines and diet, would a B12 test change anything we do?
– If the result is borderline, is a follow-up test such as methylmalonic acid useful?
– If it’s low, what is causing it, and does that affect which treatment fits?
One more decision belongs firmly with the prescriber: whether a medicine’s benefits outweigh any effect on B12. The National Institute on Aging’s advice is not to stop any prescribed medicine without talking to your health care provider first. And if numbness or memory changes are getting worse over weeks, that calls for a prompt appointment, not a wait for the next routine visit.
The point worth keeping
The MIND diet was tested once and not supported. The Mediterranean brain data are positive but contested. The ultra-processed food link is consistent and untested. Light drinking as protection is not supported, and heavy drinking is consistently linked to harm. B12 deficiency is established, testable and worth catching early, because some of its damage can become permanent.
The Mediterranean pattern has trial evidence for the heart from the republished PREDIMED analysis, and nothing in the brain trials argues against eating that way. Our guides to eating well after 60 and to memory-product advertising cover neighboring ground. If you keep one point from the brain-nutrition research, keep this. The diet with the best brand came up empty in its trial, while the clearest nutritional cause of thinking problems is a vitamin deficiency that a clinician can look for.
Sources
- Alzheimer's & Dementia. “Morris MC, Tangney CC, Wang Y, Sacks FM, Bennett DA, Aggarwal NT. "MIND diet associated with reduced incidence of Alzheimer's disease." Alzheimers Dement, 2015. PubMed. DOI: 10.1016/j.jalz.2014.11.009.” pubmed.ncbi.nlm.nih.gov
- The New England Journal of Medicine. “Barnes LL, Dhana K, Liu X, et al. "Trial of the MIND Diet for Prevention of Cognitive Decline in Older Persons." N Engl J Med, 2023. PubMed. DOI: 10.1056/NEJMoa2302368.” pubmed.ncbi.nlm.nih.gov
- National Institute on Aging. “What Do We Know About Diet and Prevention of Alzheimer's Disease? (content reviewed November 20, 2023).” nia.nih.gov
- JAMA Internal Medicine. “Valls-Pedret C, Sala-Vila A, Serra-Mir M, et al. "Mediterranean Diet and Age-Related Cognitive Decline: A Randomized Clinical Trial." JAMA Intern Med, 2015. PubMed. DOI: 10.1001/jamainternmed.2015.1668.” pubmed.ncbi.nlm.nih.gov
- The New England Journal of Medicine. “Estruch R, Ros E, Salas-Salvadó J, et al. "Primary Prevention of Cardiovascular Disease with a Mediterranean Diet Supplemented with Extra-Virgin Olive Oil or Nuts." N Engl J Med, 2018 (corrected and republished). PubMed. DOI: 10.1056/NEJMoa1800389.” pubmed.ncbi.nlm.nih.gov
- The New England Journal of Medicine. “Estruch R, Ros E, Salas-Salvadó J, et al. "Retraction and Republication: Primary Prevention of Cardiovascular Disease with a Mediterranean Diet. N Engl J Med 2013;368:1279-90." N Engl J Med, 2018. PubMed. DOI: 10.1056/NEJMc1806491.” pubmed.ncbi.nlm.nih.gov
- NIH Office of Dietary Supplements. “Vitamin B12: Fact Sheet for Health Professionals (updated July 2, 2025).” ods.od.nih.gov
- National Institute on Aging. “Cognitive Health and Older Adults (content reviewed June 11, 2024).” nia.nih.gov
- The Lancet (accepted manuscript, UCL Discovery). “Livingston G, Huntley J, Liu KY, et al. "Dementia prevention, intervention, and care: 2024 report of the Lancet standing Commission." Lancet, 2024. DOI: 10.1016/S0140-6736(24)01296-0.” discovery.ucl.ac.uk
- National Institute on Alcohol Abuse and Alcoholism. “What Is A Standard Drink? (updated December 2025).” niaaa.nih.gov
- The BMJ. “Topiwala A, Allan CL, Valkanova V, et al. "Moderate alcohol consumption as risk factor for adverse brain outcomes and cognitive decline: longitudinal cohort study." BMJ, 2017. PubMed. DOI: 10.1136/bmj.j2353.” pubmed.ncbi.nlm.nih.gov
- Nature Communications. “Daviet R, Aydogan G, Jagannathan K, et al. "Associations between alcohol consumption and gray and white matter volumes in the UK Biobank." Nat Commun, 2022. PubMed. DOI: 10.1038/s41467-022-28735-5.” pubmed.ncbi.nlm.nih.gov
- BMJ Evidence-Based Medicine. “Topiwala A, Levey DF, Zhou H, et al. "Alcohol use and risk of dementia in diverse populations: evidence from cohort, case-control and Mendelian randomisation approaches." BMJ Evid Based Med, 2026. PubMed. DOI: 10.1136/bmjebm-2025-113913.” pubmed.ncbi.nlm.nih.gov
- National Academies of Sciences, Engineering, and Medicine. “New Report Reviews Evidence on Moderate Alcohol Consumption and Health Impacts (news release, last updated December 17, 2024).” nationalacademies.org
- US Department of Health and Human Services and US Department of Agriculture. “Dietary Guidelines for Americans, 2025-2030 (January 2026).” cdn.realfood.gov
- American Association for the Study of Liver Diseases. “AASLD Raises Concern Over Removal of Evidence-Based Alcohol Guidance in 2025-2030 Dietary Guidelines for Americans (January 9, 2026).” aasld.org
- JAMA Neurology. “Gomes Gonçalves N, Vidal Ferreira N, Khandpur N, et al. "Association Between Consumption of Ultraprocessed Foods and Cognitive Decline." JAMA Neurol, 2023 (published online December 2022). PubMed. DOI: 10.1001/jamaneurol.2022.4397.” pubmed.ncbi.nlm.nih.gov
- Neurology. “Li H, Li S, Yang H, et al. "Association of Ultraprocessed Food Consumption With Risk of Dementia: A Prospective Cohort Study." Neurology, 2022. PubMed. DOI: 10.1212/WNL.0000000000200871.” pubmed.ncbi.nlm.nih.gov
- Neurology. “Bhave VM, Oladele CR, Ament Z, et al. "Associations Between Ultra-Processed Food Consumption and Adverse Brain Health Outcomes." Neurology, 2024. PubMed. DOI: 10.1212/WNL.0000000000209432.” pubmed.ncbi.nlm.nih.gov