The Psyntient research program.

The central goal is to identify recurring physiological patterns associated with distinct experiential states, beginning with altered states as unusually informative windows into the range and structure of human experience. We then test whether those patterns generalize across people, contexts, cultures, and measurement methods.

If Archetypes survive independent validation and generalize across corpora, they could support four downstream research directions:

• → research corpora for frontier AI labs
• → evaluation data for brain–computer interface state inference
• → consumer neurofeedback and entrainment via Psyntient Ground
• → empirical anchor points for philosophy of mind

The research pipeline proceeds in five stages, each feeding the next:

1. 01

   Neural data collection

   EEG recordings captured on Psyntient Ground under standardized session protocols.

2. 02

   Phenomenological reporting

   Structured first-person reports submitted alongside each neural recording.

3. 03

   Machine learning clustering

   Unsupervised methods identify candidate neural patterns across participants.

4. 04

   Architect classification

   The Architect agent evaluates candidate clusters against phenomenological tags and assigns or proposes archetypes.

5. 05

   Taxonomy refinement

   The Architect updates the living taxonomy as new data enters the Archive.

The Archive separates operational checks from scientific validation. A packet first has to clear integrity, provenance, consent, and recording-quality checks. Candidate patterns are then tested for coding reliability, subject concentration, independent annotation, cross-corpus replication, and performance on data held out from discovery.

A candidate that fails is revised, split, merged, demoted, or retired. The record is not erased. This has already happened: after recalibrating a subject-confound threshold against the corpus itself, the system demoted two of four Archetypes that had passed the earlier rule.

• Q1. Can neural archetypes be reliably detected across individuals?
• Q2. How stable are these archetypes across contexts and over time?
• Q3. Can AI systems infer experiential states directly from neural signals with calibrated uncertainty?
• Q4. What level of resolution does experiential inference require, and where does additional resolution stop helping?

The dated results, unresolved gaps, and next tests are published on the validation status page. The public account explains the tests and outcomes without publishing proprietary implementation details.