01
Name the claim precisely.
“A person reported seeing a room from above” and “the person obtained inaccessible information while unconscious” are different claims. The second needs evidence about the information, its availability, and the timing of awareness. Write down the narrowest claim the record actually supports.
02
Find the original record.
A compelling retelling can hide missing context. Open the original account or paper. Check dates, definitions, participant numbers, and the authors’ own limitations. Follow AI-extracted findings back to their source; an automated summary can omit qualifications or misread a result.
03
Ask how the information was collected.
In an experience report, look for contemporaneous notes and independent corroboration. In an experiment, examine controls, blinding, preregistration, and the outcome measures. In a review, check the inclusion criteria and how inconsistent findings were handled.
04
Look beyond the successful examples.
A collection selected for remarkable accounts cannot tell you how often those events occur in the general population. Ask about the denominator, excluded cases, missing follow-up, and unpublished results. A large archive and a representative sample are different things.
05
Compare explanations fairly.
List the ordinary and extraordinary explanations relevant to the specific evidence. Ask what each predicts and what observation would distinguish them. A failure to confirm one explanation is not automatically a confirmation of another.
06
Let uncertainty stay visible.
Treat your conclusion as proportional to the evidence. A case may be interesting, carefully documented, and still unresolved. Independent replication and stronger documentation can change its weight. Being open to revision is a strength of the inquiry.