# U0-3 threshold-stability route packet This page does not change the current public six-question batch. It makes only the fourth active row more explicit, so readers can see why `U0-3` stays next after `U0-2`, why `D02 CSTE` now stays only as a gated support anchor, and what the next public-safe artifact can honestly close. ## What still stays fixed after the 14:10 JST actual-pull reread on April 4, 2026 - An actual-pull recheck at `2026-04-04 14:10 JST` after `mind-upload@cafeab4bc2a5`, `EEG-DATA@e0695bcac83d`, and `auto-research-funds@a4852d89376c` still keeps `U0-3` fourth and still keeps `D02` only as the planned gated-support anchor for the present threshold-governance question. - The newest pulled deltas remain carry-forward rather than route replacement. `mind-upload` tightens glossary excitability and ionic terms on top of the earlier EEG foundation-model source-status split without widening the route, `EEG-DATA` still reports `30,930` merged-unique rows, `28,893` high-confidence rows, `24,909` immediate rows, and `24,539` high-confidence immediate rows after the newest metadata deep dives, and `auto-research-funds` adds `JELA Diakonia` scholarship cards without changing the practical shortlist. - None of those additions replaces the intended threshold floor. The newest carry-forwards still do not supply the `cross-session / cross-task` human-biometrics framing that keeps `D02` primary in planning, while `ds004148` and `ds007216` still stay the main public-open threshold stress-test ladders. - Browser-level live checks in this run still returned `HTTP/2 200` for the current public entry pages and this row packet with `Last-Modified: Sat, 04 Apr 2026 04:18:04 GMT`; the checked front-door pages already surfaced `2026-04-04 12:02 JST` wording while this row packet still surfaced `2026-04-04 09:05 JST` wording before this source sync. A fresh read-only Todoist audit still showed the existing funding-task coverage as sufficient (`items=206`, `projects=2`, `Kura=2`, `Nakatani family=11`, `Kashinome=2`, `Secom=3`, `Brain Science Foundation=8`, `Okawa=2`, `Japan Industrial Science Research Institute=1`, `Doumei=0`, `Kenkyu no Susume=0`), so no new Todoist item was justified for this row in this sync. ## What still stays fixed after the 09:05 JST actual-pull reread on April 4, 2026 - An actual-pull recheck at `2026-04-04 09:05 JST` after `mind-upload@e571567f017d`, `EEG-DATA@fbbb0d8c01e0`, and `auto-research-funds@d7f6fb9bc158` still keeps `U0-3` fourth and still keeps `D02` only as the planned gated-support anchor for the present threshold-governance question. - The newest pulled deltas remain carry-forward rather than route replacement. `mind-upload` now just carries the earlier `08:07 JST` route-hold sync on top of a recent pretraining-overlap audit, `EEG-DATA` still reports `30,929` merged-unique rows, `28,892` high-confidence rows, `24,909` immediate rows, and `24,539` high-confidence immediate rows after the `OpenNeuro ds007020` deep dive, and `auto-research-funds` adds deeply reviewed April 4 scholarship updates without changing the practical shortlist. - None of those additions replaces the intended threshold floor. The newest carry-forwards still do not supply the `cross-session / cross-task` human-biometrics framing that keeps `D02` primary in planning, while `ds004148` and `ds007216` still stay the main public-open threshold stress-test ladders. - Browser-level live checks in this run still returned `HTTP/2 200` for the current public entry pages and this row packet with `Last-Modified: Sat, 04 Apr 2026 00:11:41 GMT`; the checked front-door pages already surfaced `2026-04-04 08:07 JST` wording while this row packet still surfaced `2026-04-04 01:04 JST` wording before this source sync. A fresh read-only Todoist audit still showed the existing funding-task coverage as sufficient (`items=206`, `projects=2`), so no new Todoist item was justified for this row in this sync. ## What still stays fixed after the 01:04 JST actual-pull reread on April 4, 2026 - An actual-pull recheck at `2026-04-04 01:04 JST` after `mind-upload@9b6a8b971ba0`, `EEG-DATA@56d765da60b`, and `auto-research-funds@3c8569e28` still keeps `U0-3` fourth and still keeps `D02` only as the planned gated-support anchor for the present threshold-governance question. - The newest pulled deltas remain carry-forward rather than route replacement. `mind-upload` now carries the earlier `00:06 JST` route reread plus field-level artifact-stack disclosure tightening, `EEG-DATA` still reports `30,921` merged-unique rows, `28,884` high-confidence rows, `24,900` immediate rows, and `24,530` high-confidence immediate rows after deepening `Numbers` and `PRIOS`, and `auto-research-funds` adds current coverage for `同盟育成会` and `研究のすゝめ奨学金` without changing the practical shortlist. - None of those additions replaces the intended threshold floor. The newest carry-forwards still do not supply the `cross-session / cross-task` human-biometrics framing that keeps `D02` primary in planning, while `ds004148` and `ds007216` still stay the main public-open threshold stress-test ladders. - Browser-level live checks in this run still returned `HTTP/2 200` for the current public entry pages and this row packet with `Last-Modified: Fri, 03 Apr 2026 21:11:06 GMT`; the checked row packet still surfaced `Updated: 2026-04-02` before this source sync even though the front-door pages already surfaced the `2026-04-04 01:04 JST` basis. A fresh read-only Todoist audit still showed the existing funding-task coverage as sufficient (`items=206`, `projects=2`, `Kura=2`, `Nakatani family=11`, `Nakatani graduate scholarship=7`, `Kashinome=2`, `Secom=3`, `Brain Science Foundation=8`, `Okawa=2`, `Japan Industrial Science Research Institute=1`, `Keio graduate scholarship=2`, `G-7=1`, `ENEOS=1`, `同盟会=0`, `研究のすゝめ=0`), so no new Todoist item was justified for this row in this sync. ## What still stays fixed after the 23:08 JST actual-pull reread on April 2, 2026 - An actual-pull recheck at `2026-04-02 23:08 JST` after `mind-upload@f5d1d1dd1683`, `EEG-DATA@ee1463d28288`, and `auto-research-funds@ec24a355abbc` still keeps `U0-3` fourth and still keeps `D02` only as the planned gated-support anchor for the present threshold-governance question. - The newest pulled deltas remain carry-forward rather than route replacement. `mind-upload` clarifies the causal-versus-human-proxy bridge outside the current six, `EEG-DATA` now reports `30,896` merged-unique rows, `28,859` high-confidence rows, and `24,882` directly usable rows after candidate-curation refinement plus the `OpenNeuro ds007615` deep dive, and `auto-research-funds` adds `Geo-Kagaku` current cards without changing the practical shortlist. - None of those additions replaces the intended threshold floor. The latest public-open carry-forwards still do not supply the `cross-session / cross-task` human biometrics framing that keeps `D02` primary in planning, and `ds004148` plus `ds007216` still stay the main public-open stress-test ladders. - Browser-level live checks still returned `HTTP/2 200` for the current public entry pages with `Last-Modified: Thu, 02 Apr 2026 13:12:21 GMT`, and a fresh read-only Todoist audit still showed the existing funding-task coverage as sufficient, so this page only refreshes the row-specific route-hold wording. ## What still stays fixed after the 12:04 JST actual-pull reread on April 2, 2026 - An actual-pull recheck at `2026-04-02 12:04 JST` after `mind-upload@dceb8bc`, `EEG-DATA@04b7fda167e`, and `auto-research-funds@4442e348a` still keeps `U0-3` fourth and still keeps `D02` only as the planned gated-support anchor for the present threshold-governance question. - The newest pulled deltas remain carry-forward rather than route replacement. `mind-upload` syncs a spectroscopy front door outside the current six, `EEG-DATA` now reports `30,888` merged-unique rows, `28,851` high-confidence rows, and `24,506` immediate rows after `THINGS-EEG`, an OSF orientation route, and a rat ECoG deep dive, and `auto-research-funds` only refreshes repository stats. - None of those additions replaces the intended threshold floor. PhysioNet `auditory-eeg` is public-open but still too short-horizon and too low-channel to replace the `cross-session / cross-task` identity-drift floor, while `THINGS-EEG` and rat ECoG remain the wrong task family for the present threshold-stability packet. `ds004148` and `ds007216` still stay the main public-open stress-test ladders. - Browser-level live checks still returned `200` for the current public entry pages with `Last-Modified: Thu, 02 Apr 2026 02:37:04 GMT`, and a fresh read-only Todoist audit still showed the existing funding-task coverage as sufficient, so this page only refreshes the row-specific route-hold wording. ## What was rechecked for this row - `D02 CSTE` still fixes the intended repeated-session, cross-session / cross-task identity-drift floor that matches `U0-2`, but the current IEEE DataPort page is subscription-gated, so it can no longer be described here as an immediate public route. - The current funding bridge remains narrow and unchanged: `Nakatani Foundation graduate scholarship` stays first and `Kashinome Scholarship` stays second as a runway-support route. - The public six-question batch also stays unchanged: `U13-2 / U13-5 / U0-2 / U0-3 / U14-4 / U14-2`. ## Why `U0-3` still stays fourth - It is the threshold-governance row that only becomes interpretable after `U0-2` has already fixed the timing floor. - If this row fails, stronger threshold-stability wording should stay blocked because the route would still be unable to separate bounded identity drift from split-specific overfit. - The row can keep threshold selection narrow and benchmark-like before any broader continuity language is attempted. ## Why `D02` stays in scope only as a gated support anchor | Candidate | What it helps with | Why it is not the first public anchor | |---|---|---| | `D02 CSTE` (`10.21227/j162-nh61`, subscription-gated support) | It still fixes the intended repeated-session and cross-task floor shared with `U0-2`, which keeps the `threshold stability + overfit exclusion` question in scope if access is available. | The current IEEE DataPort page is subscription-gated, so it no longer counts as an immediate public anchor in this run. | | `A test-retest resting and cognitive state EEG dataset` (`10.18112/openneuro.ds004148.v1.0.1`, immediate) | It gives a richer session-external retest ladder that may later stress-test how stable the threshold rule remains outside the present biometric framing. | It is useful as a retest adjunct, but it is not the cleanest direct identity-drift floor for the first public threshold packet. | | `A multi-session simultaneous EEG-fMRI dataset with online experience sampling` (`10.18112/openneuro.ds007216.v1.0.0`, immediate) | It gives denser multi-session state labels and richer logging that may later support threshold-stability stress tests under stronger state drift. | It is multimodal and experience-sampling oriented rather than the narrowest current EEG-only threshold floor, so it stays a support anchor. | Among public-open backups, `ds004148` still stays the cleanest retest ladder and `ds007216` still stays the cleanest multi-session provenance ladder. They are strong threshold stress-test families, but they still do not replace the intended identity-drift floor that keeps `D02` primary in planning. ## What the first artifact can honestly close now | Artifact piece | What it closes now | What still stays outside the claim | |---|---|---| | `Threshold-stability note` | A bounded report of how much the decision threshold moves across tasks and sessions on the same identity-drift floor as `U0-2`. | Any statement that threshold stability by itself settles personhood continuity or branch identity. | | `Overfit-exclusion rule` | A narrow rule for rejecting models that only look stable under split-specific or single-session conditions. | Any claim that one accepted threshold already proves a broader continuity benchmark. | | `Selection-rationale table` | A public-safe record of why one threshold survives cross-session reading better than nearby alternatives. | Any theory-level conclusion that goes beyond the present EEG artifact. | ## First-pass KPI bundle and stop rule | Item | Current row-level fixation | |---|---| | `First-pass KPI bundle` | `session-out AUC`, `threshold stability`, `overfit-exclusion rate` | | `Stop rule` | Do not widen this row if the threshold order flips across sessions or if accepted models remain split-specific after the overfit screen is applied. | ## Funding wording that still fits | Route | Submission wording that stays honest | Why it still fits this row | |---|---|---| | `Nakatani Foundation graduate scholarship` | `Doctoral-track threshold-stability and overfit-exclusion benchmark for identity drift in the mind-upload route` | The row still fits a doctoral-track identity-drift benchmark when kept narrow and audit-first. | | `Kashinome Scholarship` | `Master's-stage threshold-governance rerun pack for identity drift in the mind-upload route` | It still fits only as a runway-support layer rather than the main long-horizon route. | ## External dependency tasks that stay outside this session - A session-external rerun pack is still needed before this row can support stronger threshold wording. - A larger cohort extension is still needed before the route can claim broader threshold robustness. - Any personhood-continuity or branch-identity interpretation remains outside the present EEG-only route. ## What changed in this run - The public batch did not change. - The main change in this sync is that `U0-3` now carries the same `2026-04-04 09:05 JST` route basis as the latest front-door reading and records that the live row packet had still been stuck at the earlier `2026-04-04 01:04 JST` wording before this source update. - The row still records the anchor-choice rationale, one compact first-pass KPI bundle, and one explicit stop rule instead of leaving them implicit. - This is still a route-setting update, not a solved-claim update.