The memory supplement guide: what is on the shelf and what it rests on

Arginine, citrulline, and the salts of both. The reasoning is vascular: raise nitric oxide, improve the coupling between brain activity and brain blood supply, improve cognition. Every step of that is biologically real and the last one is where the evidence thins out.

The characteristic dose is grams. The 2016 citrulline cycling trial used 2.4 g a day, the oxygen-uptake work six, the citrulline malate trial eight in a single dose, and the arginine AKG trial twelve grams a day — and found nothing.

The cognitive literature for this family in humans is essentially empty. The best test of the underlying idea is a thirteen-week trial of escalating dietary nitrate, a more reliable route to the same pathway, in overweight and obese older adults: neither cognition nor cerebral blood flow moved.

This is the family MemoHoney belongs to. Five of its six named ingredients are in it.

Bacopa monnieri, Ginkgo biloba, Panax ginseng, lion’s mane, rhodiola. This is the family with the longest randomised record in the category, and it is worth being clear that none of it is on this bottle.

Bacopa is the one most often cited for memory specifically, usually at 300 mg a day of a standardised extract over twelve weeks, and the trials that exist mostly measure delayed word recall. Ginkgo has been studied at scale in older adults with mixed and largely disappointing results at the prevention end.

A 2023 review of over-the-counter memory supplements covers this family in more depth than any retail page can, and anyone seriously shopping this shelf should read it before spending anything.

The failure mode of this family is extract standardisation. Two bottles both saying ‘bacopa 300 mg’ can contain quite different amounts of the compound the trials measured, and a label that does not name the standardisation is not comparable with one that does.

B6, B12, folate and niacin. Two quite different arguments live in this family and they get confused constantly.

The first is deficiency. Severe niacin deficiency causes pellagra, whose third symptom is dementia, and B12 deficiency is a genuine and reversible cause of cognitive impairment that a blood test finds. Correcting a real deficiency is medicine and it works.

The second is supplementation in people who are not deficient, and that is a much weaker case. The Chicago Health and Aging Project is the strongest observational signal for niacin: 3,718 older adults, higher dietary intake associated with less incident Alzheimer’s disease and slower decline over a median 5.5 years. Dietary. Observational.

The failure mode here is a ceiling. The adult tolerable upper intake level for niacin is 35 mg a day, set on flushing, and B6 has its own limit for a more serious reason. A reader stacking a multivitamin, a B-complex and a memory formula can cross those without ever seeing a warning.

EPA and DHA, from fish oil or algae. DHA is a structural component of neuronal membranes and the biological case is the most straightforward in the category.

Its characteristic dose is also the reason it is almost never a component of a blended memory capsule: a gram or more a day, as an oil. That does not share a capsule with five other things, which is why omega-3 is sold on its own in soft gels and why a memory blend claiming to include it is usually including a token amount.

Its failure mode is expectation. The large prevention trials in cognitively healthy older adults have been largely null, and the honest position is that it is a reasonable thing to have in a diet rather than a treatment for a memory complaint.

Alpha-GPC, citicoline, huperzine A, and the various choline salts. The reasoning is neurotransmitter-level rather than vascular: supply more choline, support acetylcholine synthesis, support memory formation.

This family is the one that most often appears alongside a prescription-drug-adjacent name, and huperzine A in particular is a cholinesterase inhibitor with a mechanism shared by actual dementia drugs. That is an argument for caution rather than enthusiasm: an ingredient with a drug-like mechanism deserves a drug-like level of scrutiny, and it rarely gets one on a supplement label.

None of this family is on this bottle either.

Three reasons, and only one of them is sinister.

Proprietary blends are legal. A manufacturer may declare a blend total rather than the amount of each component, and many do. It protects a formula from being copied, which is a real commercial concern.

Cost. The ingredients with the best evidence are frequently the most expensive per gram, and a label without numbers cannot be compared on cost per active gram with a label that has them. That is an advantage to whoever is spending less.

Physics. This is the one nobody says out loud. A capsule holds a finite amount of powder — four hundred to eight hundred milligrams for the common sizes — and several ingredient families have characteristic doses in the grams. Printing the numbers would make the arithmetic obvious, and the arithmetic is unflattering for any once-daily capsule containing six things.

None of that makes an unlabelled product bad. It makes it unevaluable, which is a different problem and in some ways a worse one.

Capsule counts assume 400-800 mg of fill per capsule shared between all ingredients. Trial doses are from the records cited on this page and, for bacopa, from the review cited above.

This table is the least commercial thing on this website and it is the most useful.

MemoHoney is a family-one product with one family-three ingredient and one unusual addition.

Five of its six names are nitric oxide amino acids, whose cognitive evidence in humans is the thinnest of the five families. Niacin is the B vitamin, and it is the only ingredient here that can sit at a useful amount inside a capsule. Beta-alanine belongs to none of the five families and is the genuinely interesting choice: two small randomised trials at 2.4 g a day for ten weeks, in older adults scoring at or below normal, with a real result in that subgroup.

That combination is unusual and it is not obviously wrong. What it is not is checkable, because no amount is printed for any of it.