Brain Foods for Memory: What the Trials Actually Show

Contents — 12 foods

There is no single brain food. Every ranking page says so in its first paragraph and then spends the next thousand words ignoring it. Here is the version that does not.

Most brain-food advice rests on observational studies. Researchers track what thousands of people eat, wait a decade, and see who declines. That design is genuinely useful and it is not proof. When these diets have been put into randomized trials — where you assign the diet and wait — some held up and some did not.

The most-studied brain diet there is, the MIND diet, did not beat its control diet over three years. That trial was published in the New England Journal of Medicine in 2023, and almost no ranking page mentions it.

So this list grades every food by what kind of evidence stands behind it, gives you the dose, and tells you where the evidence runs out.

How this list is graded

Every food below carries an evidence badge, and the badge describes the research, not the food. A vegetable can be excellent for you and still sit in the bottom tier here, because the badge answers one question only: how well has this been tested for memory?

Strong

Randomized trial evidence, or a meta-analysis of several large cohorts pointing the same way. Worth building meals around.

Moderate

Real cohort evidence, but from one population, one sex or one study, and never tested in a trial. Reasonable; keep expectations modest.

Weak

Popular, but the memory claim rests on one small study, a null trial or a failed replication. Eat it if you like it, not for your memory.

Avoid

Tested properly and it did not work, or the evidence does not exist at all. Do not spend money on it.

Say this part out loud: almost everything in the Strong and Moderate rows is associational. People who eat leafy greens differ from people who do not in a hundred ways that researchers can only partly adjust for. The grades describe the strength of the evidence, not the size of the guarantee.

Four claims that circulate widely are missing from this article on purpose: a hazard ratio for the ginkgo trial, the industry-funding story attached to one walnut study, the effect sizes from the 2023 cocoa subgroup paper, and the “two servings of berries a week” threshold. None of them survived a check against the primary source.

All 12 at a glance

If you read nothing else, read the chart and the table below. Between them they are the whole article: the best measured effect for each food, the design that produced it, and the dose that goes with it.

Chart A

The MIND diet did not beat the control diet

Difference between the arms: 0.035 standardized units (95% CI −0.022 to 0.092), P = 0.23.

How to read this: Both groups got slightly better over three years, and the diet group’s edge is small enough that it could easily be chance. The finding here is the absence of a gap, which is why two near-parallel lines show it better than bars would.

Barnes et al. (2023), New England Journal of Medicine 389(7):602–611 — 604 adults, 3 years, global cognition composite. Brain MRI outcomes did not differ either.

Food or change	Evidence	Best measured effect	Design	Dose	Watch out for
Mediterranean diet + olive oil	Strong	Frontal cognition +0.23 vs −0.33 in controls	RCT, n = 334	~1 L olive oil a week	It means changing the whole diet
Fatty fish	Strong	Dementia risk RR 0.80 (0.74–0.87)	Meta-analysis, 7 cohorts	2 × ~100 g a week	Pregnancy (mercury); vegetarians
Leafy greens	Strong	Decline rate equal to being 11 years younger	Cohort, n = 960	~1.3 servings a day	Warfarin users (vitamin K)
Extra virgin olive oil	Strong	28% lower dementia-related death	Cohort, n > 92,000	>7 g a day (½ tbsp)	The endpoint is death, not memory
Berries	Moderate	Cognitive ageing delayed up to 2.5 years	Cohort, n = 16,010	Not specified in the paper	Women only, nearly all nurses
Whole grains	Moderate	Memory decline equal to >8 years younger	Cohort, n ≈ 3,300	Up to ~2.7 servings a day	No effect found in white participants
Walnuts and nuts	Weak	No effect: SMD 0.27 (−0.65 to 1.19), P = 0.57	Meta-analysis, 5 RCTs	30–60 g a day	Memory. The 2-year trial was null
Coffee, tea, caffeine	Weak	The 200 mg finding did not replicate	RCT + failed follow-up	—	Memory consolidation; sleep
Dark chocolate and cocoa	Weak	Global cognition −0.01 SD (−0.08 to 0.05)	RCT, n = 573	500 mg flavanols a day	Everyone. The trial was null
Turmeric and curcumin	Weak	28% memory gain in 40 people	RCT, n = 40	90 mg twice a day	Supplement supplied by its maker
Eggs and choline	Weak	Alzheimer’s HR 0.53 (0.34–0.83)	Cohort, n = 1,024	>1 egg a week	Egg-industry funded
Omega-3 supplements	Avoid	Two meta-analyses disagree sharply	Meta-analyses of RCTs	—	Replacing fish with pills
Ginkgo biloba	Avoid	No difference: 277 vs 246 dementia cases	RCT, n = 3,069	120 mg twice a day	Everyone
Coconut oil and MCT	Avoid	No adequate trial evidence	—	—	Everyone
Cutting ultra-processed food	Strong	Swapping 10% out: 19% lower dementia risk	Cohort, n = 72,083	—	Nothing. It also saves money

How to read this table: The grade describes how well the food has been tested, not how nutritious it is. Walnuts are good food; they sit in the Weak tier because the two-year randomized trial testing them for memory found nothing. A Strong badge does not promise you a benefit — it means the evidence behind the claim is the best available.

Before the listStart with the pattern, not the food

The two best-tested things on this page are not foods. They are whole diets. One of them worked in a trial and one of them did not, and the difference between those two results is the most useful thing in this article.

Mediterranean diet with extra virgin olive oil

Strong

The only dietary pattern here with a positive randomized trial behind it. Start with this.

Valls-Pedret and colleagues reported the cognitive arm of PREDIMED in JAMA Internal Medicine in 2015. They analysed 334 participants, mean age 66.8, over a median 4.1 years, across three arms: a Mediterranean diet with a litre of extra virgin olive oil a week, the same diet with 30 g of mixed nuts a day, and a control arm given low-fat advice.

The control group declined on all three cognitive composites. The olive oil group roughly held. That is the honest framing — decline avoided, not improvement gained.

Chart B

Only the olive oil arm held its ground

How to read this: Bars to the left of the line mean people got worse over the four years. The control group moved left on every measure. The olive oil group did not. The nuts arm landed in between and never separated from control.

Valls-Pedret et al. (2015), JAMA Internal Medicine 175(7):1094–1103 — 334 adults analysed, median 4.1 years, adjusted change in composite z-scores.

Note what the nuts arm did, or rather did not do. It never separated from control on frontal or global cognition. That is the first hint that the walnut story further down this page is weaker than everyone writes.

The MIND diet

Moderate

Brilliant on paper. It did not beat a control diet when someone finally ran the trial. Both facts belong here.

Martha Clare Morris and colleagues built the MIND diet by combining the Mediterranean and DASH diets and weighting it towards the foods with the best cognitive evidence. In the 2015 Rush cohort it was associated with a substantially lower incidence of Alzheimer’s disease. It is the most cited brain diet there is.

Barnes, Dhana, Liu and colleagues ran the trial. 604 participants, 301 on MIND and 303 on control, 93.4% completed, three years. All of them were older adults without cognitive impairment, with a family history of dementia, a BMI over 25 and a poor baseline diet — exactly the people you would expect to benefit.

The MIND arm improved by 0.205 standardized units. The control arm improved by 0.170. The difference was 0.035 (95% CI −0.022 to 0.092), P = 0.23. Brain MRI outcomes did not differ either. Chart A at the top of this page is that result.

Group 1The strongest evidence: fish, greens, olive oil

These three have either a randomized trial or a meta-analysis of several large cohorts behind them. They are also, conveniently, the three cheapest things to actually do.

Fatty fish and memory

Strong

The strongest single-food evidence on this page — and the clearest example of why the food is not the same as the pill.

Bakre and colleagues pooled seven prospective cohorts in Public Health Nutrition in 2018: 40,668 participants and 3,139 dementia cases. Fish eaters had a relative risk of 0.80 (95% CI 0.74–0.87) against non-eaters. That is roughly 20% lower dementia risk.

The dose-response is the part nobody reports. Going from no fish to some fish is where almost all of the signal sits. Going from some to a lot adds very little.

Chart C

More fish helps, but the curve flattens fast

How to read this: Anything left of the 1.0 line means lower risk than eating no fish. The low, medium and high bars barely move as consumption rises — most of the benefit arrives with the first servings. The amber bar is a different outcome, Alzheimer’s specifically, so do not read it as the next step up the same ladder.

Bakre et al. (2018), Public Health Nutrition 21(10):1921–1932 — meta-analysis of 7 prospective cohorts, 40,668 participants, 3,139 dementia cases. Observational: association, not proof.

Dose: two servings a week, about 100 g each. Stick to low-mercury species — salmon, sardines, cod, pollock, canned light tuna. The mechanism usually offered is EPA and DHA being built into neuronal membranes, which is plausible and is a separate claim from the cohort data above.

Not great for: pregnancy, where mercury limits govern which species you choose; vegetarians and vegans; anyone with a fish allergy. Plant sources give you ALA rather than EPA and DHA, and the evidence for ALA is weaker.

Leafy greens and cognitive decline

Strong

The cheapest high-evidence food here. The famous “11 years younger” figure is real, and it is the authors’ own framing.

Morris and colleagues followed 960 older adults, mean age 81, for about five years and published the result in Neurology in 2018. The group eating the most greens managed a median 1.3 servings a day. Their rate of cognitive decline was slower by an amount the authors describe as equivalent to being 11 years younger.

The association survived adjustment for cardiovascular conditions, depressive symptoms, low weight and obesity. That is a more robust finding than most things on this page.

11 years

Slower decline, expressed as a cognitive age equivalence

960 adults, mean age 81, about five years of follow-up. The highest-intake group ate a median 1.3 servings of greens a day — a large handful raw, or half a cup cooked.

Morris et al. (2018), Neurology 90(3):e214–e222

Not great for: anyone taking warfarin. Leafy greens are high in vitamin K, which interacts with it, and a sudden large increase needs a conversation with the prescriber first. None of the pages currently ranking for this term mentions that.

Extra virgin olive oil as a food

Strong

A very small dose with a very large association. The endpoint is the catch.

Tessier and colleagues pooled the Nurses’ Health Study and the Health Professionals Follow-up Study in JAMA Network Open in 2024: more than 92,000 adults, mean age around 56, followed for 28 years. More than 7 g a day — about half a tablespoon — was associated with 28% lower risk of dementia-related death, and the association held regardless of overall diet quality.

This is the easiest change on the page. It requires no new habit, only a different bottle.

Group 2Cohort evidence only: berries and whole grains

Both of these have real, large, well-conducted cohort studies behind them. Neither has been tested in a trial, and both come with a population caveat that changes what the number means for you.

Berries and cognitive ageing

Moderate

Pleasant, cheap enough, decent cohort evidence — from a study of women only.

Devore, Kang, Breteler and Grodstein tracked 16,010 women over 70 from the Nurses’ Health Study, measuring cognitive function every two years between 1995 and 2001. Greater berry intake was associated with cognitive ageing delayed by up to 2.5 years.

Every participant was a woman, nearly all were registered nurses, and the cohort was predominantly white. Most pages that cite this study mention “women” in passing and move on. It is a limitation, not a detail.

Whole grains and memory

Moderate

Real evidence, and the most interesting caveat in the whole article.

Liu and colleagues published a biracial community cohort in Neurology in November 2023: more than 3,300 participants, all over 65, roughly 60% Black, followed about six years. Intake ranged from under 0.2 to 2.7 servings a day — every group sat below the federal three-serving guideline.

Higher intake in Black participants tracked with memory preservation the authors describe as equivalent to being more than eight years younger. No similar trend appeared in white participants. The study was funded by the Alzheimer’s Association and the NIH.

Every competing page states a flat whole-grain benefit. The finding was subgroup-specific, nobody yet knows why, and saying both of those things is the honest version.

Chart D

“Years younger” claims, and where no such number exists

Leafy greens11 years

Morris et al., Neurology 2018 (n = 960).

Whole grains>8 years

Liu et al., Neurology 2023 (n ≈ 3,300). In Black participants only; no association found in white participants.

Berries2.5 years

Devore et al., Ann Neurol 2012 (n = 16,010). Women only.

Fatty fishnot measured this way

Extra virgin olive oilnot measured this way

Walnutsnot measured this way

Coffee and caffeinenot measured this way

Dark chocolatenot measured this way

036912

Years younger, as reported by each study

Foods marked “not measured this way” have never been reported in years-younger terms. That is not the same as having no effect.

How to read this: A dash means nobody has measured that food in these units. It does not mean zero, and it does not mean the food failed — it means the question was never asked this way. The hatched bar marks a finding that applied to one subgroup only.

Each figure comes from a different observational cohort and is the study authors’ own equivalence framing, not a biological age. The three studies used different designs, populations and follow-up lengths, so the bars are not strictly comparable with each other.

Group 3Popular, and barely tested

Five foods that appear on every brain-food list, where the specific memory claim rests on one small study, a null trial, or a replication that failed. None of them is bad for you. All of them are oversold.

Walnuts and nuts for memory

Weak

Good food, weak memory evidence. Three independent lines of evidence tested it and all three came back null.

The claim traces to Arab and Ang, who looked at walnut consumption and cognitive function in NHANES data from 1988–1994 and 1999–2002. In adults over 60, story recall was 7.1 percentile points higher (P = 0.03) and digit-symbol substitution 7.3 points higher (P = 0.05). Intake averaged 13.1 g a day in that group.

That study is cross-sectional. It measured diet and cognition at the same moment, so it cannot show which came first, let alone cause. Then people tested it properly.

WAHA (Sala-Vila et al., 2020): 708 enrolled, 636 completed, ages 63–79, two years, walnuts at about 15% of daily energy. No between-group difference in the global cognitive composite.
Moabedi et al. (2024) pooled 5 randomized trials and 928 adults: SMD 0.27, 95% CI −0.65 to 1.19, P = 0.57, with heterogeneity I² = 94%. Their own conclusion: no significant effect of nuts on cognitive function.
PREDIMED’s nuts arm did not separate from control on frontal or global cognition, while the olive oil arm did.

A post-hoc subgroup in WAHA’s Barcelona cohort suggested possible benefit in higher-risk participants. Post-hoc means hypothesis-generating, so it belongs in the next trial’s design rather than in your shopping basket.

To be precise about what this means: walnuts are good for blood lipids and for satiety, and unproven for memory.

Coffee, tea and caffeine

Weak

Caffeine works for alertness. The memory-consolidation claim is the weakest well-known finding on this page.

The original is Borota and colleagues in Nature Neuroscience, 2014. 160 participants, double-blind and placebo-controlled, 200 mg of caffeine given after studying a series of images, with better image discrimination a day later. The paper also carries a corrigendum issued in August 2014 correcting P-values and degrees of freedom.

Aust and Stahl ran the follow-up in Memory in 2020. They did not restrict participants’ normal caffeine intake, and they compared average consumers, around 1,200 mg a week, against low consumers at around 50 mg. They found no significant caffeine effect, and their non-superiority testing ruled out medium and large enhancements in both groups. The title is “The enhancing effect of 200 mg caffeine on mnemonic discrimination is at best small.”

Dark chocolate and cocoa flavanols

Weak

The big trial was null. A subgroup finding is being reported as though it were the headline.

Vyas and colleagues reported the clinic subcohort of COSMOS in the American Journal of Clinical Nutrition in 2024: 573 participants, 285 on cocoa extract and 288 on placebo, two years, 500 mg of flavanols a day including 80 mg of (−)-epicatechin. Global cognition differed by −0.01 SD (95% CI −0.08 to 0.05). That is null.

The study behind most “chocolate is brain food” headlines is a different one: Brickman and colleagues in PNAS, 2023, which restored hippocampal-dependent memory in participants who started with lower diet quality and lower habitual flavanol intake. We could not open the primary paper, so this article reports the direction and not the effect sizes.

One more thing, and it matters: the trials used a 500 mg flavanol extract. Flavanol content in commercial chocolate varies enormously and is not on the label. Eating dark chocolate does not deliver the trial dose.

Turmeric and curcumin

Weak

A promising pilot. A pilot is a reason to run a bigger trial, not a reason to buy the supplement.

Small and colleagues ran an 18-month double-blind placebo-controlled trial of a bioavailable curcumin, 90 mg twice daily, in non-demented adults aged 50–90. Memory improved 28% in the treatment group, and PET imaging showed less amyloid and tau signal.

There were 40 people in it. And the supplement and placebos were supplied by Theravalus Corp., which also funded laboratory testing and conference travel.

Eggs and choline

Weak

A real association from a real cohort, with a funding disclosure that belongs next to it.

Pan and colleagues analysed 1,024 older adults from the Rush Memory and Aging Project, mean age 81.4, mean 6.7 years of follow-up. More than one egg a week was associated with an Alzheimer’s hazard ratio of 0.53 (95% CI 0.34–0.83); two or more a week gave 0.53 (0.35–0.81). About 39% of the effect was mediated through dietary choline.

The study was funded by “an unrestricted investigator-initiated grant from the Egg Nutrition Center to Think Healthy Group, LLC,” and one author has received past research support from the Egg Nutrition Center.

Group 4Tested properly, and they failed

This is where the food-versus-pill distinction earns the article its keep. Fish is associated with 20% lower dementia risk. Fish oil capsules, put into randomized trials, do not cleanly reproduce it.

Omega-3 supplements. Two meta-analyses disagree, and honest reporting means showing both. Brainard and colleagues reviewed randomized trials in 2020 and concluded there is little to no effect of long-chain omega-3 supplementation on cognition. Shahinfar and colleagues pooled 58 trials in Scientific Reports in 2025 and reported, per 2,000 mg a day, a global cognition SMD of 1.08 (95% CI 0.73–1.44).

Ginkgo biloba

Avoid

The largest prevention trial on this page, and it found nothing.

DeKosky and colleagues ran the Ginkgo Evaluation of Memory study and published it in JAMA in 2008. 3,069 participants aged 75 and over took EGb 761 at 120 mg twice daily for a mean 6.1 years. There were 277 dementia cases on ginkgo against 246 on placebo. No difference.

Two sources reported only the raw case counts, so this article does not print a hazard ratio for that trial. The counts say enough.

Coconut oil and MCT

Avoid

No adequate randomized evidence for cognition, and it is high in saturated fat.

The internet believes otherwise largely because of a widely shared anecdote about one physician’s husband and a small set of MCT studies in people who already had Alzheimer’s disease. There is no adequate trial evidence for cognition in healthy adults. One paragraph is all it earns.

Chart E

Food versus pill: what happened in randomized trials

Mediterranean diet + olive oilValls-Pedret 2015n = 334Positive result
MIND dietBarnes 2023n = 604Null result
WalnutsSala-Vila 2020n = 636Null result
Nuts, five trials pooledMoabedi 2024n = 928Null result
Cocoa extractVyas 2024n = 573Null result
Ginkgo bilobaDeKosky 2008n = 3,069Null result
CurcuminSmall 2018 — industry-supplied supplementn = 40Positive result
Omega-3 supplementsBrainard 2020 vs Shahinfar 202558 trialsMixed — see text
Fatty fish, the foodCohorts only—Never tested
Leafy greensCohorts only—Never tested
Coconut oil and MCTNo adequate trial—Never tested

Curcumin’s positive result comes from 40 people, using a supplement supplied by its manufacturer. Omega-3 is marked mixed because two meta-analyses of the same literature reach opposite conclusions.

How to read this: The right-hand column is what happened when someone actually assigned the intervention and waited. A dash means no adequate trial exists — that is a gap in the science, not a negative result. Note which rows are foods and which are extracts.

Trial outcomes for each intervention, with the number of participants. Every cell traces to a study in the sources list below.

The one change with the best evidence is a subtraction

The highest-leverage item on this page is not a food you add. Li and colleagues followed 72,083 UK Biobank participants aged 55 and over, dementia-free at baseline, for a median ten years — 717,333 person-years and 518 incident dementia cases.

Replacing 10% of ultra-processed food by weight with unprocessed or minimally processed food was associated with a hazard ratio of 0.81 (95% CI 0.74–0.89). A 19% lower dementia risk, for a swap rather than a purchase.

19%

Lower dementia risk, for swapping 10% of ultra-processed food for unprocessed food

The same 10% of the diet moved the other way was associated with 25% higher risk. Alzheimer’s specifically: HR 1.14 (1.00–1.30). Vascular dementia: HR 1.28 (1.06–1.55).

Li et al. (2022), Neurology 99(10):e1056–e1066

Chart F

The same 10% of your diet, moved in opposite directions

How to read this: The line is no change in risk. Right of it is worse, left of it is better. Both bars describe the same tenth of your shopping — one version of it added, one version of it swapped out.

Li et al. (2022), Neurology 99(10):e1056–e1066 — 72,083 UK Biobank adults, median 10 years, 518 incident dementia cases. Observational. The authors note regression dilution over long follow-up and cohort representativeness as limitations.

Ultra-processed here means the NOVA group 4 definition: industrial formulations built from substances you would not cook with, rather than simply “packaged.” In practice the swap is the ready meal for the same meal cooked, the flavoured yoghurt for plain yoghurt and fruit, the crisps for the nuts you are eating for their lipids anyway.

Who should be careful

Four situations where the advice above needs adjusting before you act on it. No other page in the current top five has this section, which is a strange thing to be true of health content.

If you	Because	Do this
Take warfarin	Vitamin K in leafy greens interacts with it, and a sudden large increase can move your INR.	Change intake gradually and tell your prescriber first
Are pregnant	Mercury limits govern which fish species you choose, and caffeine intake is usually capped.	Check current national guidance on species and limits
Are sensitive to caffeine, anxious or sleeping badly	The alertness benefit is real; so is the cost to your sleep.	Protect the sleep — it has clearer evidence than any food here
Take any medication	Turmeric and curcumin supplements at pharmacological doses can interact with anticoagulants.	Ask a pharmacist before starting a supplement

A week, assembled

Twelve foods is a list. One week is a plan. Everything below comes from the Strong and Moderate tiers only, at the doses in the master table, which means nothing here is on the strength of a single small study.

Original graphic

Seven days, built only from the top two tiers

Frozen berries and tinned sardines count. So does the supermarket’s own-brand extra virgin olive oil. Nothing on this page requires an expensive version of itself.

How to read this: This is a schedule, not a measurement — every dot is a dose from the master table placed on a day. The three daily rows are the ones doing most of the work; the two fish servings are the single highest-confidence food on the page.

Doses: greens ~1.3 servings a day (Morris 2018); olive oil >7 g a day (Tessier 2024); fish 2 × ~100 g a week (Bakre 2018); whole grains up to ~2.7 servings a day (Liu 2023). Berry frequency is our own suggestion — the cohort paper does not specify one.

Then do the subtraction, because it has the largest number attached to it. Pick the one ultra-processed item you eat most often and swap it for the unprocessed version of the same meal. That is the 10% in Chart F.

And keep the comparison honest. Diet is one lever among several, and the ones with the clearest evidence for memory are not foods at all — which is the subject of our ranking of study methods by effect size, where sleep alone explained about a quarter of the variation in one cohort of students’ grades.

Questions people actually ask

What is the best brain food for memory?

If you want one answer: oily fish. Across seven prospective cohorts covering 40,668 people, fish eaters had about 20% lower dementia risk than non-eaters (RR 0.80, 95% CI 0.74–0.87), and the association was stronger for Alzheimer’s specifically. The honest answer is that no single food has been shown in a randomized trial to improve memory. The intervention with the best trial evidence is a whole pattern — a Mediterranean diet with extra virgin olive oil — not any one ingredient.

What are the 10 genius foods?

“Genius foods” comes from a popular book, not from the research literature, and there is no scientific list by that name. If you want the foods with the strongest published evidence, this article grades twelve of them: fatty fish, leafy greens, extra virgin olive oil, berries and whole grains have the best support; walnuts, coffee, dark chocolate, turmeric and eggs have weaker or conflicted evidence; ginkgo and coconut oil have been tested or examined and do not hold up.

How do I boost my memory naturally?

The dietary change with the largest measured association is not adding anything — it is subtracting ultra-processed food. In a 72,083-person UK Biobank cohort, each 10% increase in ultra-processed food intake was associated with 25% higher dementia risk, and replacing 10% of it with unprocessed food was associated with 19% lower risk. Beyond diet, sleep has clearer evidence behind it than any single food on this page, and it costs nothing.

What foods eliminate brain fog?

No food has been shown in a trial to eliminate brain fog, and “brain fog” is a description rather than a diagnosis — it can come from poor sleep, dehydration, medication, thyroid problems, anaemia, depression, long COVID or perimenopause. Caffeine will make you feel sharper temporarily; when researchers tested whether it actually improves memory, the effect did not replicate. If the fog is persistent or new, that is a conversation with a clinician, not a shopping list.

Sources

Every link below goes to the paper, its publisher record, or a government summary of it — never to a blog summarising it.

01Valls-Pedret, C., et al. (2015). Mediterranean diet and age-related cognitive decline: a randomized clinical trial. JAMA Internal Medicine, 175(7), 1094–1103. doi
02Morris, M. C., et al. (2015). MIND diet associated with reduced incidence of Alzheimer’s disease. Alzheimer’s & Dementia, 11(9), 1007–1014. PubMed · doi
03Barnes, L. L., et al. (2023). Trial of the MIND diet for prevention of cognitive decline in older persons. New England Journal of Medicine, 389(7), 602–611. doi · PubMed
04Bakre, A. T., et al. (2018). Association between fish consumption and risk of dementia. Public Health Nutrition, 21(10), 1921–1932. doi
05Morris, M. C., et al. (2018). Nutrients and bioactives in green leafy vegetables and cognitive decline. Neurology, 90(3), e214–e222. doi · NIA summary
06Tessier, A.-J., et al. (2024). Consumption of olive oil and diet quality and risk of dementia-related death. JAMA Network Open, 7(5), e2410021. NIA summary · Harvard Health
07Devore, E. E., Kang, J. H., Breteler, M. M. B., & Grodstein, F. (2012). Dietary intakes of berries and flavonoids in relation to cognitive decline. Annals of Neurology, 72(1), 135–143. doi
08Liu, X., et al. (2023). Association of whole grain consumption and cognitive decline: a community-based biracial cohort. Neurology. Journal · PubMed
09Arab, L., & Ang, A. (2015). A cross-sectional study of the association between walnut consumption and cognitive function. Journal of Nutrition, Health & Aging, 19(3), 284–290. doi
10Sala-Vila, A., et al. (2020). Effect of a 2-year diet intervention with walnuts on cognitive decline: the WAHA randomized controlled trial. American Journal of Clinical Nutrition, 111(3), 590–600. PubMed
11Moabedi, M., et al. (2024). The effect of consuming nuts on cognitive function: a systematic review and meta-analysis of randomized clinical trials. Frontiers in Nutrition, 11, 1463801. doi
12Borota, D., et al. (2014). Post-study caffeine administration enhances memory consolidation in humans. Nature Neuroscience, 17(2), 201–203. doi — see the corrigendum issued 26 August 2014.
13Aust, F., & Stahl, C. (2020). The enhancing effect of 200 mg caffeine on mnemonic discrimination is at best small. Memory, 28(7). PubMed
14Vyas, C. M., et al. (2024). Effect of cocoa extract supplementation on cognitive function: the COSMOS clinic subcohort. American Journal of Clinical Nutrition, 119(1), 39–48. doi
15Brickman, A. M., et al. (2023). Dietary flavanols restore hippocampal-dependent memory in older adults with lower diet quality and lower habitual flavanol consumption. PNAS, 120(23), e2216932120. doi
16Small, G. W., et al. (2018). Memory and brain amyloid and tau effects of a bioavailable form of curcumin: an 18-month trial. American Journal of Geriatric Psychiatry, 26(3), 266–277. PubMed · UCLA newsroom
17Pan, Y., et al. (2024). Association of egg intake with Alzheimer’s dementia risk in older adults: the Rush Memory and Aging Project. Journal of Nutrition. doi · funding disclosure noted by Marion Nestle, Food Politics
18Brainard, J. S., et al. (2020). Omega-3, omega-6 and polyunsaturated fat for cognition: systematic review and meta-analysis of randomized trials. JAMDA, 21(10), 1439–1450.e21. PubMed
19Shahinfar, H., et al. (2025). A systematic review and dose-response meta-analysis of omega-3 supplementation on cognitive function. Scientific Reports, 15, 30610. doi
20DeKosky, S. T., et al. (2008). Ginkgo biloba for prevention of dementia: the Ginkgo Evaluation of Memory study. JAMA, 300(19), 2253–2262. PubMed · ALZFORUM
21Li, H., et al. (2022). Association of ultraprocessed food consumption with risk of dementia: a prospective cohort study. Neurology, 99(10), e1056–e1066. doi · PubMed

Reviewed every twelve months, and sooner if a new randomized trial lands on cocoa, omega-3 or the MIND diet. Last checked 13 September 2026.

← All topics