Established July 2026 · one working paper (design complete) · experiment pending
This research area treats contemplative traditions as engineering material: cognitive architectures refined over long periods, described extensively in the corpora on which large language models are trained, and therefore present in model representations — activatable, testable, and usable as harness mechanics. The approach is operational throughout. No claim in this program depends on any doctrinal or metaphysical position being accepted; the traditions are examined for the same reason one examines any long-lived engineering practice — the designs survived their users.
The methodological precedent is Francisco Varela's program in cognitive science: contemplative traditions engaged as rigorous method rather than belief, by researchers with practice training. The work here is practitioner-authored, from formal lineage training; that fact is disclosed for scope, and nothing further rests on it.
| contemplative structure | the agent-systems problem it addresses |
|---|---|
| Vow mechanics and graded empowerment — capability access granted in witnessed, staged degrees, transmitted with provenance, revocable | trust calibration and graded autonomy: authority as something demonstrated and conferred in stages, never assumed |
| The witness requirement — a practitioner's attainment is certified by the teacher, never self-certified | oversight: self-assessment is structurally insufficient for acceptance; verification must be external to the system being verified |
| Generation-stage practice — a sustained literature on role-simulation that becomes operative, together with the safety structures (vows) built around it | persona and boundary engineering for systems that are, mechanically, simulators |
| Treasure-text authorship (terma) — a genre mechanism by which new material enters a canon under sanctioned attribution | the provenance and legitimation of generated content; corpus traditions as generative systems with their own admission rules |
The program's first paper: rather than asking a model to reason about a contemplative concept, the harness has it hold one as a boundary while working, with the environment enforcing what the framing describes. A five-condition experiment separates roleplay, enforcement, and their coupling; results will be published here with verbatim transcripts.
This page is a research-program statement. Teaching and practice materials are not part of SSRI and are not hosted here. Behavioral claims made by this program are tested against logged runs; mechanism-level claims about model internals are stated as hypotheses until instrumented.