# Adversarial review log — psilocybin-therapy

**Case:** psilocybin-therapy (revised edition, `consensus_consistency_audit_2026-10-09/revised/`)
**Authoring model of record:** Sol 6.1. The document's own provenance line records a Codex/GPT original assessment
(2026-09-27) and a Claude red-team synthesis (2026-09-28); that record was preserved, not edited.
**Adversarial reviewer:** Claude Opus 5.5 (Anthropic)
**Date:** 2026-10-14 · **Round:** 2, group 1

## Summary

| Item | Before | After |
|---|---|---|
| Headline level | Γ — Mechanistically plausible, partial system support | Γ — unchanged |
| Confidence | Moderate | Moderate — unchanged |
| Role | epiphenomenon (at claimed human scope) | unchanged |
| Subclaim levels | SC1 Ε · SC2 Δ · SC3 Ε · SC4 Γ · SC5 Γ | unchanged |
| Momentum / Stability / Divergence | Decisive work underway / Uncertain / Contested | unchanged |
| Sources | 50 | 50 |
| Validator (5.4) | PASS — 0 fail, 0 warn, 33 pass | PASS — 0 fail, 0 warn, 33 pass |

**Findings by severity:** Critical 0 · Major 0 · Minor 1 · Editorial 2. Rejected findings: 8.

**PMIDs:** 50/50 reference PMIDs verified against PubMed; 4/4 attached erratum PMIDs verified as Published Errata;
0 failures; no retraction or expression of concern in any of the 54 records.

**Claims spot-checked:** 12, all matched their source.

## Findings

| ID | Severity | Location | Original → corrected | Evidence |
|---|---|---|---|---|
| A-1 | Minor | Ledger row [32], system/design cell | `adult local corticalHtr2a deletion` → `adult local cortical Htr2a deletion` | Builder rendering defect (missing space); no change of meaning. |
| A-2 | Editorial | Ledger row [34], system/design cell | `differentiated>40 d;2–4 donor lines` → `differentiated >40 d; 2–4 donor lines` | Same defect class: two missing spaces. |
| A-3 | Editorial | Ledger row [34], system/design cell | `co-culture. 10µM psilocin` → `co-culture. 10 µM psilocin` | Missing space between numeral and unit; the same value is rendered "10 µM" elsewhere in the document. |

Plus the mandated banner under the provenance line and a new `## ADVERSARIAL REVIEW AND APPROVAL` section immediately
before `## References`.

## Rejected findings

See the `### Rejected findings (checked, no change)` table inside the review (R-1 … R-8). In brief: six ±1 print-versus-
e-pub year differences; three PubMed titles dropping a receptor subscript; the EPISODE "triple-arm" wording (correct, as
the row is scoped to the first dosing period); the weakest-link derivation and its two reported alternatives (all three
correct); Momentum "Decisive work underway" (named, specific artifact present); SC2 at Δ/Low; the collective-author
builder defect (not present here); and `claims_total == sources_total` at 50.

## Files changed in this case folder

- `psilocybin-therapy.review.md` — edited (3 corrections, banner, approval section)
- `validator.txt` — new (5.4 validator output for the revised edition)
- `ADVERSARIAL_REVIEW.md`, `adversarial-findings.json` — new
- `sol61-original/` — pre-edit copies of `psilocybin-therapy.review.md`,
  `psilocybin-therapy.scorecard.json`, `validation-report.json`, `consensus-validation.json`,
  `consensus-addendum.md`, `SHA256.txt`

`psilocybin-therapy.scorecard.json` was **not** changed: no level, confidence, count or role moved, and the three
corrections are prose-rendering fixes with no scorecard counterpart.

## Package-level propagation needed from the lead

Nothing. No headline string, letter, confidence, count or title changed, so no package-level file
(`OVERVIEW.md`, `MANIFEST.json`, `READER.html`, `BATCH_COMPLETION.json`, `SHA256.txt`) carries a value that is now
stale. `SHA256.txt` in this folder does need the new `psilocybin-therapy.review.md` digest (see
`adversarial-findings.json` → `changed_file_hashes`); the lead should regenerate it rather than have this agent edit it.

## Lead-flagged Oct-9 revision artifact — checked, not present

The lead reported that another group found a systematic regression in the `consensus_consistency_audit_2026-10-09/revised/`
edition: a spliced Consensus section coexisting with preserved REVIEWER NOTES still asserting that no Consensus work
was performed. **It is not present in this case**, and nothing was edited to "fix" it.

- The spliced block here is headed "Consensus provenance reconciliation — 9 October 2026" and is self-limiting: it
  says the located records are historical checks, "not newly credited complete-paper examinations", that no
  source-comparison count is inferred, and that the supplement "is not a new independent or blinded adjudication".
- The preserved reviewer notes and the scorecard's `literature_tools_note` make **no** negative claim about Consensus.
  They positively enumerate the retrieval tools used by the authoring passes, which predate the 2026-10-05 enrichment
  the Consensus records come from, so the two blocks describe different passes and do not contradict each other.
- Recorded in the review as rejected finding **R-9** rather than silently ignored, per the brief.

Per the lead's instruction, `review_metadata.reviewer` reading "Codex" was treated as house provenance and left
untouched; this reviewer's banner and approval section state the authoring model of record as Sol 6.1 and record the
package's own string verbatim rather than editing it.

## Lead-flagged systematic artifacts — per-case disposition

The lead reported four systematic artifacts found by other groups. All four were checked against this case.

| Artifact | Present here? | Action |
|---|---|---|
| **(1) False approval boilerplate** — the Oct-9 `integrate_consensus.py` appends "The earlier adversarial approval refers to its original edition" unconditionally | **Yes** | Corrected as finding **A-4** in both the review and `consensus-addendum.md`, by rescoping rather than deleting. Verified first: this slug is not one of the 26 genuinely approved on 2026-10-06, and the script line that emits it unconditionally was read. |
| **(2) Confidence band justified by reviewer access depth** | **Yes** | Corrected as finding **A-5** in the review prose and the scorecard rationale. **Band unchanged at Moderate** — argued both ways; the High text is not met. |
| **(3) Renderer bug** — "Best-supported rival" printing `models_evaluated[0]`, often the nominal | No | `models_evaluated[0]` is the rank-1 non-nominal model and the prose names it correctly. |
| **(4) Reviewer-notes contradiction and false provenance clauses** | No | Recorded as rejected finding **R-9** in the review: the spliced block is self-limiting and no preserved sentence contradicts it. The false "existing HIV draft" clause is absent from this batch entirely. |

Per the lead, the scorecard `review_metadata.reviewer` string "Codex" was treated as house provenance and left untouched.

## Lead policy applied — confidence bands and momentum (2026-10-14)

- **Confidence two-step test:** step 1 (strike access depth as a determinant) applied; step 2 (re-derive) -> **Moderate stands**, band unchanged.
- **Momentum stricter rule** ("Decisive work underway" requires work actually being run; a completed publication never satisfies it): **holds**.
- Three substantive determinants survive step 2 (heterogeneous human measurement; mediator never manipulated; largest randomized datasets not peer reviewed), so access depth was not the only thing holding the band down. Momentum rests on two **recruiting** trials, no completed publication.

Full derivation is recorded in the review's `## ADVERSARIAL REVIEW AND APPROVAL` section and in `adversarial-findings.json` -> `lead_policy_2026_10_14`.

## Consensus question for the lead

Has any completed human study measured a drug-induced change in **cortical inducibility** (paired associative
stimulation, TMS-EEG LTP-like plasticity, or visual/somatosensory potentiation) after psilocybin, with both onset and
closure timed? A yes would move SC4 off Γ and is the single result the whole Γ verdict turns on.