Our standard
How we read the evidence
Brain health is a field where confident claims travel faster than the research behind them. This page explains how we decide what to publish, how we label the strength of what we publish, and what we will not say regardless of how well it would perform.
Everyday first, terminology second
Every idea we publish is explained twice: first as something you already recognise from ordinary life, then as the mechanism underneath it, with the scientific term attached. This is an editorial rule, not a stylistic preference. A concept you can only repeat is not a concept you can use, and terminology introduced before understanding tends to produce the first without the second.
Three levels, assigned by a person
Where an article or program makes a claim strong enough to act on, we label how well supported that claim currently is. The levels below are assigned editorially, against the criteria listed, by someone who has read the underlying work. They are never derived automatically from the text — an automatic label would be a statement about research that nobody actually checked, which is the exact failure this page exists to prevent.
A level describes the state of the evidence, not how useful the idea might be for you. Early-evidence findings are not worthless; they are unfinished. The label is there so you can calibrate how much to bet on something.
Strong evidence
Replicated across multiple good-quality human studies.- Supported by several independent human studies, including randomised trials or large prospective cohorts where the question allows them.
- Findings have been replicated by teams with no shared funding or authorship.
- The effect direction is consistent across studies, even where the effect size varies.
- Reviews or meta-analyses point the same way.
Promising evidence
Consistent human findings, but the picture is still incomplete.- Multiple human studies point the same way, but the trials are small, short, or run in narrow populations.
- Mechanism is plausible and supported, but long-term outcomes are not yet established.
- Replication exists but is limited, or effect sizes vary widely between studies.
Early evidence
Interesting, not settled — treat as a hypothesis, not a plan.- Largely animal work, mechanistic studies, or single small human trials.
- Not yet replicated independently, or replication attempts have been mixed.
- Frequently over-reported in popular coverage relative to what the data supports.
What we will not publish
We do not diagnose or treat conditions, and we do not write personalised protocols. Where a subject is genuinely sensitive — extended fasting, ketogenic diets, cannabis, mental health, cognitive decline — the page carries an additional, specific notice, and the guidance stops at the point where an individual clinical judgement would be required.
We do not sell supplements, and we have no commercial relationship with anything we describe. Where we link to a book, it is one we wrote. Where the honest answer is that the research does not yet support a recommendation, we say that instead of filling the gap with a plausible-sounding one.
Corrections and review
Where an article carries a review date, it means a person re-read it against current research on that date. Evidence moves; a piece written two years ago may be describing a picture that has since shifted, and an unreviewed date is a signal to treat it accordingly. If you find something you believe is wrong or out of date, we would rather hear it than not.
This standard sits alongside our medical disclaimer, which sets out the limits of everything published here.