How Two5 reads evidence
A sentence about brains can be careful or it can be advertising. This page is the filter. It describes a standard. It does not describe a laboratory Two5 already runs.
What we look for first
For a question about groups of people and an outcome, we start with guidelines and systematic reviews, then the trials those reviews depend on. A single encouraging paper does not outrank a review that includes it and still calls the evidence weak. Official public-health guidance outranks a company blog, including ours.
That order is not universal. A question about how a puzzle works is answered with a worked example, not a meta-analysis. A question about dementia risk is not answered with a worked example.
What we record about a study
- Who was in it: age, cognitive status, and whether they match the people we are talking to.
- What they actually did, for how long, compared with what.
- What was measured: the trained task, another test, daily function, or a diagnosis.
- How large and how consistent the result was, and whether anyone has repeated it.
- What the authors say they cannot conclude.
Transfer, cause, and scores
Correlation is not causation. People who choose mentally active hobbies can differ in education, health, and social life. Near transfer means better performance on a task like the one you practised. Far transfer means a gain on something you did not practise. A clinical outcome means a health result a clinician would recognise. A Two5 score, when rounds exist, will be the first of these at most.
Self-reported “I feel sharper” is not a validated endpoint. Product metrics — signups, sessions, points — are not evidence of cognitive benefit. We will not publish a dementia-risk score from gameplay.
Vulnerable groups and future studies
Research with older adults, or with people who have impaired decision-making, needs qualified partners, ethics review, informed consent, accessible procedures, and data rules that match the sensitivity of the data. None of that apparatus is in place at Two5 today. Interest in talking about it is not a protocol.
The public write-up of that boundary is the About page. The internal version, including how claims are logged, is in the editorial policy in the repository.
Sources used on the evidence guides
- World Health Organization, 2019 guideline on risk reduction of cognitive decline and dementia.
- U.S. National Institute on Aging pages on prevention and cognitive health.
- Simons and colleagues, 2016, on brain-training programs.
- Owen and colleagues, 2010, a direct test of a commercial program.
Read the guides: what the research supports and whether brain games prevent dementia.