Source Library · comparative phenomenology · companion to the Atlas
The Atlas of Described Experience sorted the library using thirty-six categories derived from twentieth-century psychometrics. That is circular: a map cannot find a kind of experience its probe set never named. So we did it again, letting a tradition supply the categories — the stage model of the Visuddhimagga, extracted from our own copy of the manual, in its own vocabulary. Both maps are drawn here in one space.
Buddhaghosa’s fifth-century manual is the most systematic phenomenology any premodern tradition produced. It defines two ladders: nine attainments of concentration, from access concentration through the four jhānas and the four immaterial bases to the cessation of perception and feeling; and sixteen knowledges of insight, each with stated criteria for recognising it. Stages six through eleven of the insight sequence — terror, danger, disenchantment, desire for deliverance, re-observation, equanimity — are the tradition’s own documented dark passage. No modern instrument has items for them.
We fed our copy of the text back through itself: located the pages where each stage is defined, and had a model extract the recognition markers in the manual’s own language. The result is a probe set of 26 stages and 93 sentences that no one on this project authored. It reads like this:
That restraint is a feature, not a failure. The questionnaire asks whether an experience had a quality — light, unity, timelessness — and almost any vivid passage has some. The manual asks whether an experience occupies a specific position in an ordered sequence, with the right factors present and the wrong ones absent. It is a diagnostic instrument, not a checklist, and it refuses most of what it is shown.
| Passages matching a stage | Concentration ladder | The dark passage (stages 6–11) | Other insight stages |
|---|
Read the two rows against each other. Inside the tradition, the model is used mostly to describe attainments of concentration — which is what you would expect of a manual written to guide a practice. Outside it, where nobody was following Buddhaghosa’s instructions, the stages that fit are overwhelmingly the ones describing things falling apart: terror, danger, disenchantment, the desire to escape one’s own experience. The technique is local. The crisis is not.
| Tradition | Stage matches | Most-matched stage |
|---|
Some of the passages the model claims, from outside its own tradition:
| Modern category | Times it appears among the nearest Atlas neighbours |
|---|
The model over-matches literature. A Pushkin lyric about breaking free of nets, and a line of Akhmatova’s about going silent into the earth, were both matched to insight stages. Weariness in a poem is not muñcitukamyatā-ñāṇa. Any claim built on this should be read at the level of the aggregate, never the individual match.
The inside/outside line is imperfect. The classifier was told that Theravāda, Mahāyāna and Vajrayāna sources count as inside the tradition, and it still labelled many Tibetan texts as outside it. Those are corrected here by tradition tag rather than by the classifier’s own judgment, which means the boundary is drawn by a crude list and not by a scholar.
These are soteriological maps. The Visuddhimagga is normative: it describes what should happen to someone progressing correctly, and it was used to authorise some experiences and pathologise others. Swapping the MEQ for the manual trades a psychometric bias for a monastic one. Neither is ground truth; the value is in reading them against each other.
And the translation problem is unchanged. Probes were built from an English translation of a Pali text and matched against machine translations of the corpus. Proximity here is still partly proximity of English. Probes written natively in Pali, Persian, Tibetan and Classical Chinese are the next thing that would actually move this forward.