A collection built for researchers studying spiritual awakening, non-ordinary states of consciousness, and transformative experience. Every account was read by two independent readers working from one written rulebook, and how far the two of them agreed is published beside every figure.
Stories of Awakening is a systematically built collection of long-form video interviews with people describing their own spiritual awakening or mystical experience, drawn from independent archives.
The two figures differ because people were interviewed more than once. The unit of analysis is the person, so every percentage on this site is a share of people, never a share of interviews.
Each account was selected to meet the following criteria: a speaker sharing their own direct experience, in sufficient depth to be coded (minimum transcript length enforced per source), in a first-person interview format.
Accounts in which somebody nearly died are not here. They moved to a separate near-death study in August 2026, so this collection describes only people who did not nearly die. That removes a confound rather than a subject: injuries from the event were being read as physical features of the awakening, which inflated the exposure in exactly the accounts most likely to report difficulty afterwards.
The collection covers five dimensions, features in all. Each dimension was read in its own pass, and no pass was shown the tags from any other:
Every feature is reported twice. Clearly described is the share of people both readers recorded it for. Possibly described is the share at least one of them did. The gap between the two is not a confidence interval and not a margin of error. It is the amount of disagreement between two readers about what a transcript says.
The collection was built through a documented pipeline across five phases. The full methodology is at methodology.html, including the coding prompt text, the agreement figures, and a description of all known limitations.
Two readers, one rulebook. That is the whole design, and it is a change from what this site reported before. The earlier version also showed two numbers per feature, but those came from one reader working to two different standards, so the gap between them measured the standards rather than the accounts. The two numbers here come from two independent readers working from the same rulebook, so the gap is real disagreement about what a transcript says.
of the features reached the conventional bar for agreement between two readers. They are psychedelics, leaving the body, hearing guidance, light, emptiness, the body moving on its own, energy in the body, and visions. Every one is something a person saw, heard or felt. The features the two readers did not agree on are the interpretive ones: presence and stillness, service to others, letting time pass, pulling away from people. Every one of those needs a judgment about what an experience meant.
That split is the result, and it is reported as one. Concrete experience can be counted from interviews at this scale. Interpretive categories cannot, and a figure for one of them should be read as a description of how the readers behaved as much as a description of the people.
Accuracy was measured by hand rather than asserted. 400 applied tags were drawn at random from across all five passes, frozen at the draw with the seed recorded, and each one was read against the passage it was taken from. 2.2% were not supported by that passage (95% interval 1.0 to 3.8, resampled over people). That is a false-positive rate and nothing more. It measures how often an applied tag is unsupported by its own evidence, and it cannot see a feature a reader missed, because a missing tag leaves no passage to inspect. False negatives are the harder problem here, and nothing on this site measures them.
The collection is suited to descriptive and correlational questions about what people describe and what follows it, and to questions about the instrument itself. Examples:
This is a self-report collection of public video interviews, not a clinical or population sample. The source population skews toward people willing and able to sit for a long-form video interview, toward Western and English-speaking contexts, and toward people who have found their experience to be ultimately meaningful (people who found it purely harmful are less likely to appear in an interview archive at all).
Two further limits belong beside every figure. People interviewed more than once have more chances to trip any feature, simply from having talked for longer, so every figure is published alongside a version that excludes them. And both readers read the same transcript, so their errors are not independent: anything that raises the chance of a tag, a long interview, a vivid speaker, a clean transcript, a host who probed hard, raises it for both of them. That biases an association away from zero, not toward it, so the numbers here should not be read as a conservative floor. These limits are set out in full in the methodology.
This site provides aggregated figures, how common each feature is, and categorical breakdowns. Individual transcripts are not hosted here; they are available through the original source archives listed on the Sources page.
The data and pipeline code are not yet publicly released, but collaboration inquiries are welcome. If you are working on research related to spiritually transformative experiences and would like to discuss access or collaboration, please reach out via the About page.
For citation purposes, please use: