The memory supplement guide: what is on the shelf and what the trials found

Two things make it difficult. The first is that the outcome is invisible. Nobody can compare their memory this month against their memory in March, which means the category runs almost entirely on expectation, and expectation is exactly what a marketing page is built to create.

The second is that the labels are frequently unhelpful. An analysis of products promoted for brain health found stated contents that did not survive laboratory comparison, and the authors treated the situation as a public health problem rather than a matter of consumer preference (a 2020 analysis of brain-health supplements).

The consequence is that the useful skills here are not pharmacological. They are reading a label, recognising a dose, and knowing which of the big trials in each family went which way.

Ginkgo biloba, bacopa monnieri, lion's mane, huperzine A, panax ginseng. This is the oldest part of the shelf and it carries the largest trials, which is both its strength and its problem.

The largest is the Ginkgo Evaluation of Memory study: more than three thousand older adults, a median of six years of follow-up, and no reduction in the incidence of dementia (the Ginkgo Evaluation of Memory trial). That is about as definitive as this field gets, and it is worth knowing before paying for ginkgo on the strength of a marketing page.

Botanicals also carry the interaction problems. Ginkgo and anticoagulants is the classic pairing, and the general advice from the National Center for Complementary and Integrative Health is to tell a clinician what you take (NCCIH on memory and cognition products). A botanical is a mixture of compounds at variable strengths, which is the opposite of a defined molecule.

Niacin, B6, B12 and folate. This family has the clearest biological story on the shelf and the most commonly overstated one.

The story is real: severe deficiency of several B vitamins produces cognitive symptoms that reverse when the vitamin is replaced. Niacin deficiency in its full form includes a dementia among its classic features. That is not a marketing claim, it is a described disease.

The overstatement is the jump from deficiency to supplementation in people who are not deficient. The evidence for that jump is mostly observational. Higher dietary niacin tracked with less incident Alzheimer's disease over six years in a large cohort (the Chicago Health and Aging Project analysis), and a separate study followed intake from young adulthood into midlife (the CARDIA midlife cognition study). Both describe what people ate, not what happened when they were given a capsule.

The one randomised result worth knowing is COSMOS-Mind, which gave older adults a daily multivitamin for three years and found a modest benefit on global cognition (the COSMOS-Mind trial). That is the strongest trial-level finding anywhere on this shelf, and it is for a cheap, unglamorous product.

Figures for adults. The niacin row is the one that matters on this shelf, because niacin appears in a great many products at once (the NIH Office of Dietary Supplements niacin fact sheet).

DHA, EPA and phosphatidylserine. A structural argument rather than a functional one: these are materials the brain is built from rather than substances that act on it.

The argument is coherent and the trial record is mixed, with most of the positive signal in people whose intake was low to begin with. That is a recurring shape across this whole shelf and it is worth noticing: almost every nutrient intervention that works, works best in the people who were short of it.

For a buyer the practical points are dull and useful. Dose is measured in the DHA and EPA content rather than in the weight of the oil. Oxidation is a real problem and a fishy aftertaste is a signal. And a person eating oily fish twice a week is already at the intake most of the trials used.

L-arginine, L-citrulline and their salt forms. This is the family the product sold on this website belongs to, and the one whose evidence base sits almost entirely outside cognition.

The chemistry is settled: arginine is the substrate for nitric oxide, nitric oxide relaxes blood vessels, and citrulline is the better oral route because arginine is destroyed before it arrives (the 2008 pharmacokinetic study). The doses are large. Arginine moved blood pressure at four grams a day and above (the 2022 dose-response meta-analysis); citrulline needed three grams twice daily to lift plasma arginine (the 2008 pharmacokinetic study).

The awkward result is the one this family least likes to quote. Thirteen weeks of escalating dietary nitrate, which is a more reliable way to raise nitric oxide than any amino acid, changed neither cognitive function nor cerebral blood flow in overweight and obese older adults (the thirteen-week cerebral blood flow trial). One pilot trial does not close a question, and it is the most direct test anybody has run.

There is one positive acute cognition result in this family, and precision matters: healthy young adults given 1.5 g of inositol-stabilised arginine silicate performed better on several measures than on placebo (the one acute arginine cognition trial). That is a silicate complex rather than free arginine or citrulline.

Caffeine, L-theanine, and the various combinations sold as focus products. This is the one part of the shelf where something reliably happens within the hour, and it is the part least often described as a memory supplement.

That is an honest distinction rather than a pedantic one. Caffeine improves alertness and sustained attention; it does not improve memory formation, and most of what people call forgetfulness is an attention problem wearing a memory costume. Walking into a room and forgetting the errand is usually a failure to encode rather than a failure to recall.

The trade-offs are well known and worth stating: tolerance develops, sleep is affected at doses and times people underestimate, and sleep is where memory consolidation actually happens. A product that improves your afternoon at the cost of your night is not a memory product.

One strongest and one most awkward result per family. Both columns are cited in the source list at the foot of this page.

The conclusion this guide reaches is not that nothing works. It is that the biggest levers in this whole area are not on this shelf at all, and the buying guide beside this one starts there.