mind upload rq60 rq by rq deep dossiers - yasufumi-nakata/mind-upload GitHub Wiki

Mind-upload RQ60 RQ-by-RQ Deep Dossiers (EEG-DATA × auto-research-funds)

汎用要約ではなく、RQ単位で実装可能な深掘り

This learning page is generated for GitHub Wiki. The public portal is managed on mind-upload.com.

  • Updated: 2026-04-04 / Role: RQ-by-RQ Deep Dossier

Mind-upload RQ60 RQ-by-RQ Deep Dossiers (EEG-DATA × auto-research-funds)

本ページは、60RQ を1件ずつ深掘りするための固定ドシエ集です。
汎用横断サマリは最小化し、各RQで 検証命題 境界 応募テーマ 主データ KPI 停止条件 を先に固定します。

  • 更新日: 2026-04-04
  • 出典: mind-upload/wiki/mind-upload-rq60-deep-evaluation-cards.md(60行を再構成)

2026-04-04 08:07 JST latest actual-pull route-hold note

This run still does not change the current public six-question batch. An actual-pull reread after mind-upload@418c8f5d0ef9e871531d9cfb8548887e57f54d68, EEG-DATA@dc6d833b83e9a5e6a181edfee5349f561f65b043, and auto-research-funds@58d8738be5d55cff13d7bd626479783d6818de1f still leaves the current D10 / D03 / D02 / D01 family and the outside-current-six order U7-1 -> U1-2 -> U4-2 -> U8-1 unchanged.

  • The newest pulled deltas still deepen route discipline and backup space rather than replacing the current route family: mind-upload tightens the front-door human-proxy comparison, EEG-DATA still keeps 30,929 merged-unique rows, 28,892 high-confidence rows, and 24,909 directly usable rows while promoting HeartBEAM and figshare EEG at rest and with TMS only as backup families, and auto-research-funds adds the JGC-S foreign-student scholarship only as an adjacent runway without changing the practical shortlist.
  • Recounting the ### Ux-y headings on this page still returns 60, so this run keeps the one-row-at-a-time dossier structure intact.
  • The public six still stay U13-2 / U13-5 / U0-2 / U0-3 / U14-4 / U14-2.
  • The solve-first order still stays U13-2 -> U13-5 -> U0-2 -> U0-3 -> U14-4 -> U14-2.
  • The route-safe access reading stays narrow in one important way: D10 still stays the only clear public-open primary anchor while D03 / D02 / D01 still stay gated-support planning anchors, and none of the newest repo-wide carry-forward closes the recall-validity gap under U13-5 or the rerun-support gap under U14-4 / U14-2.
  • Browser-level live checks with a browser user agent still returned HTTP/2 200 for issue.html, mind-upload-current-public-six-rq-brief.html, and mind-upload-rq-solvability-bridge.html; the current Last-Modified header was Fri, 03 Apr 2026 22:10:18 GMT on those public surfaces, while the checked bodies still surfaced the earlier 01:04 JST hold note before this source sync.
  • The funding bridge still stays narrow: Kura Fund and Nakatani Foundation graduate scholarship remain the active near-term lanes, Kashinome Scholarship remains the student-runway side lane, Nakatani Foundation research grant remains next-cycle watch, Secom General Research Grant remains closed for the current cycle, and a fresh Todoist api/v1/sync(items, projects) audit (items=206, projects=2, Kura=2, Nakatani family=11, Kashinome=2, Secom=3, Doumei=0, Kenkyu no Susume=0) still did not justify a new task.

2026-04-03 05:12 JST latest actual-pull route-hold note

This run still does not change the current public six-question batch. An actual-pull reread after mind-upload@a9fc5d1cb691f3669c40803604e78cf87fab4eed, EEG-DATA@8c3ace28bad0b3bac259caa440994e5d05bfc7d1, and auto-research-funds@35756f8381c47daa24bb70bb7a1ff88d1dcf76bc still leaves the current D10 / D03 / D02 / D01 family and the outside-current-six order U7-1 -> U1-2 -> U4-2 -> U8-1 unchanged.

  • The newest pulled deltas still deepen route discipline and support pages rather than replacing the current route family: mind-upload is now at a docs: sync github wiki export head that keeps the exported route mirrors aligned without changing the active queue, EEG-DATA carries forward a VR navigation dataset output refresh on top of a recent EESM17 metadata deepening while still keeping 30,900 merged-unique rows, 28,863 high-confidence rows, and 24,885 directly usable rows, and auto-research-funds adds JSPS overseas fellow current-card snapshots on top of a recent JSSF research publication refresh while now reporting 8,594 aid IDs, 4,307 rows, and 3,957 official-source rows without changing the practical shortlist.
  • Recounting the ### Ux-y headings on this page still returns 60, so this run keeps the one-row-at-a-time dossier structure intact.
  • The public six still stay U13-2 / U13-5 / U0-2 / U0-3 / U14-4 / U14-2.
  • The solve-first order still stays U13-2 -> U13-5 -> U0-2 -> U0-3 -> U14-4 -> U14-2.
  • The route-safe access reading stays narrow in one important way: D10 still stays the only clear public-open primary anchor while D03 / D02 / D01 still stay gated-support planning anchors, and none of the newest repo-wide carry-forward closes the recall-validity gap under U13-5 or the rerun-support gap under U14-4 / U14-2.
  • Browser-level live checks with a browser user agent still returned HTTP/2 200 for issue.html, mind-upload-current-public-six-rq-brief.html, and mind-upload-rq-solvability-bridge.html; the current Last-Modified header was Thu, 02 Apr 2026 19:17:56 GMT on those public surfaces, while the checked bodies still surfaced the earlier 04:05/04:06 JST hold note before this source sync.
  • The funding bridge still stays narrow: Kura Fund and Nakatani Foundation graduate scholarship remain the active near-term lanes, Kashinome Scholarship remains the student-runway side lane, Nakatani Foundation research grant remains next-cycle watch, Secom General Research Grant remains closed for the current cycle, and a fresh Todoist api/v1/sync(items, projects) audit (items=203, projects=2, Kura=2, Nakatani graduate scholarship=7, Kashinome=2, Secom=3) still did not justify a new task.

2026-04-03 04:06 JST latest actual-pull route-hold note

This run still does not change the current public six-question batch. An actual-pull reread after mind-upload@f5e13ebdc4748ce683d905c37cabfdb3c53036bb, EEG-DATA@53f5c8166faf0d5fcebfb5c39741d4992eb82a39, and auto-research-funds@1f190fec93a9180a27443b1fe2b31f9488dfbf74 still leaves the current D10 / D03 / D02 / D01 family and the outside-current-six order U7-1 -> U1-2 -> U4-2 -> U8-1 unchanged.

  • The newest pulled deltas still deepen route discipline and support pages rather than replacing the current route family: mind-upload carries forward a BCSFB / human-proxy composition literature sync outside the current six, EEG-DATA keeps 30,900 merged-unique rows, 28,863 high-confidence rows, and 24,885 directly usable rows after a catalog refresh that does not replace any current anchor family, and auto-research-funds refreshes current-card and registry layers around Itoh domestic scholarship without changing the practical shortlist.
  • Recounting the ### Ux-y headings on this page still returns 60, so this run keeps the one-row-at-a-time dossier structure intact.
  • The public six still stay U13-2 / U13-5 / U0-2 / U0-3 / U14-4 / U14-2.
  • The solve-first order still stays U13-2 -> U13-5 -> U0-2 -> U0-3 -> U14-4 -> U14-2.
  • The route-safe access reading stays narrow in one important way: D10 still stays the only clear public-open primary anchor while D03 / D02 / D01 still stay gated-support planning anchors, and none of the newest repo-wide carry-forward closes the recall-validity gap under U13-5 or the rerun-support gap under U14-4 / U14-2.
  • Browser-level live checks with a browser user agent still returned HTTP/2 200 for issue.html, mind-upload-current-public-six-rq-brief.html, and mind-upload-rq-solvability-bridge.html; the current Last-Modified header was Thu, 02 Apr 2026 18:14:39 GMT on those public surfaces.
  • The funding bridge still stays narrow: Kura Fund and Nakatani Foundation graduate scholarship remain the active near-term lanes, Kashinome Scholarship remains the student-runway side lane, Nakatani Foundation research grant remains next-cycle watch, Secom General Research Grant remains closed for the current cycle, and a fresh Todoist api/v1/sync(items, projects) audit (items=203, projects=2, Kura=2, Nakatani graduate scholarship=7, Kashinome=2, Secom=3) still did not justify a new task.

2026-04-02 18:06 JST latest actual-pull route-hold note

This run still does not change the current public six-question batch. An actual-pull reread after mind-upload@85b6603b569b, EEG-DATA@a7f9390492fd, and auto-research-funds@b6c0d0f4eeb7 still leaves the current D10 / D03 / D02 / D01 family and the outside-current-six order U7-1 -> U1-2 -> U4-2 -> U8-1 unchanged.

  • The newest pulled deltas still deepen metadata and support pages rather than replacing the current route family: mind-upload only syncs U7-2 public bridge wording, EEG-DATA deep-dives OSF gazx2 confidence EEG metadata and OpenNeuro ds007081 while now reporting 30,892 merged-unique rows, 28,855 high-confidence rows, and 24,880 directly usable rows in the current README snapshot, and auto-research-funds refreshes its README summary to 8,590 aid IDs, 4,301 rows, and 3,951 official-source rows without changing the practical shortlist.
  • Recounting the ### Ux-y headings on this page still returns 60, so this run keeps the one-row-at-a-time dossier structure intact.
  • The public six still stay U13-2 / U13-5 / U0-2 / U0-3 / U14-4 / U14-2.
  • The solve-first order still stays U13-2 -> U13-5 -> U0-2 -> U0-3 -> U14-4 -> U14-2.
  • The route-safe access reading stays narrow in one important way: D10 still stays the only clear public-open primary anchor while D03 / D02 / D01 still stay gated-support planning anchors, and none of the latest repo-wide carry-forward closes the recall-validity gap under U13-5 or the rerun-support gap under U14-4 / U14-2.
  • Browser-level live checks with a browser user agent still returned HTTP/2 200 for issue.html, mind-upload-current-public-six-rq-brief.html, and mind-upload-rq-solvability-bridge.html; the current Last-Modified header was Thu, 02 Apr 2026 08:22:50 GMT on those public surfaces.
  • The funding bridge still stays narrow: Kura Fund, Nakatani Foundation graduate scholarship, and Kashinome Scholarship remain the active near-term lanes, Nakatani Foundation research grant remains next-cycle watch, Secom General Research Grant remains closed for the current cycle, and a fresh Todoist api/v1/sync(items, projects) audit (items=203, projects=2, Kura=2, Nakatani graduate scholarship=7, Kashinome=2, Secom=3) still did not justify a new task.

2026-04-02 11:09 JST latest actual-pull route-hold note

This run still does not change the current public six-question batch. An actual-pull reread after mind-upload@208e59e2180c, EEG-DATA@00d85380f3f, and auto-research-funds@f4ba35518400 still leaves the current D10 / D03 / D02 / D01 family and the outside-current-six order U7-1 -> U1-2 -> U4-2 -> U8-1 unchanged.

  • The current mind-upload head is the earlier 10:07 route-hold packet refresh itself, while the newest upstream movement is on EEG-DATA: chore: deep-dive ChineseEEG-2 and PhysioNet auditory EEG. The refreshed summary still reports 30,887 merged-unique rows, 28,850 high-confidence rows, and 24,875 directly usable rows, and auto-research-funds keeps the same practical shortlist while its repository stats still read 8,585 aid IDs, 4,293 rows, and 3,943 official-source rows.
  • Recounting the ### Ux-y headings on this page still returns 60, so this run keeps the one-row-at-a-time dossier structure intact.
  • The public six still stay U13-2 / U13-5 / U0-2 / U0-3 / U14-4 / U14-2.
  • The solve-first order still stays U13-2 -> U13-5 -> U0-2 -> U0-3 -> U14-4 -> U14-2.
  • The route-safe access reading stays narrow in one important way: D10 remains the only clear public-open primary anchor while D03 / D02 / D01 still stay gated support anchors in public wording.
  • The newest public-open adjuncts only tighten row-specific backup coverage rather than replace the current route family: ds007591 and ds007602 strengthen speech-side backup options for U13-2 but still expose only three released participants each, ChineseEEG-2 strengthens public reading/listening alignment coverage for U13-5 but still does not expose the intended recall-labeled branch, and PhysioNet auditory-eeg is clearly public-open but remains single-day 4-channel coverage rather than a replacement for D02 in U0-2 / U0-3.
  • Browser-level live checks with a browser user agent still returned 200 for issue.html, mind-upload-current-public-six-rq-brief.html, and mind-upload-rq-solvability-bridge.html; the current Last-Modified header was Thu, 02 Apr 2026 01:22:44 GMT on those public surfaces.
  • The funding bridge still stays narrow: Kura Fund, Nakatani Foundation graduate scholarship, and Kashinome Scholarship remain the active near-term lanes, Secom General Research Grant remains closed for the current cycle, and a fresh Todoist api/v1/sync(items, projects) audit (items=203, Kura=2, Nakatani graduate scholarship=7, Kashinome=2, Secom=3) still did not justify a new task.

2026-04-02 10:07 JST latest actual-pull route-hold note

This run still does not change the current public six-question batch. An actual-pull reread after mind-upload@be507c810b11, EEG-DATA@3f20f48cca6e, and auto-research-funds@fb32c05163f4 still leaves the current D10 / D03 / D02 / D01 family and the outside-current-six order U7-1 -> U1-2 -> U4-2 -> U8-1 unchanged.

  • The newest pulled deltas still deepen metadata and support pages rather than replacing the current route family: mind-upload refines EEG foundation setup-equivalence boundaries outside the current six, EEG-DATA deep-dives the HBN EEG release 11 route while now reporting 30,887 merged-unique rows, 28,850 high-confidence rows, and 24,871 directly usable rows, and auto-research-funds keeps the same practical shortlist while its repository stats still read 8,585 aid IDs, 4,293 rows, and 3,943 official-source rows.
  • Recounting the ### Ux-y headings on this page still returns 60, so this run keeps the one-row-at-a-time dossier structure intact.
  • The public six still stay U13-2 / U13-5 / U0-2 / U0-3 / U14-4 / U14-2.
  • The solve-first order still stays U13-2 -> U13-5 -> U0-2 -> U0-3 -> U14-4 -> U14-2.
  • The route-safe access reading stays narrow in one important way: the refreshed local EEG-DATA catalog keeps widening, but the live-access read still leaves D10 as the only clear public-open primary anchor while D03 / D02 / D01 still stay gated support anchors in public wording.
  • In this run, the funding-facing copy is also re-synchronized to this file's Dxx + DOI + dataset name + access class wording, so the public-side and funding-side RQ-by-RQ dossiers now point to the same canonical row packets again.
  • Browser-level live checks with a browser user agent still returned 200 for issue.html, mind-upload-current-public-six-rq-brief.html, and mind-upload-rq-solvability-bridge.html; the current Last-Modified header was Thu, 02 Apr 2026 00:15:06 GMT on those public surfaces.
  • The funding bridge still stays narrow: Kura Fund, Nakatani Foundation graduate scholarship, and Kashinome Scholarship remain the active near-term lanes, Secom General Research Grant remains closed for the current cycle, and a fresh Todoist api/v1/sync(items, projects) audit (items=203, Kura=2, Nakatani graduate scholarship=7, Kashinome=2, Secom=3) still did not justify a new task.

2026-04-01 20:09 JST latest post-pull route-hold note

This run still does not change the current public six-question batch. An actual-pull reread after mind-upload@da9644095ba6, EEG-DATA@1ecc5cdc0131, and auto-research-funds@b5cd37febbbb still leaves the current D10 / D03 / D02 / D01 family and the outside-current-six order U7-1 -> U1-2 -> U4-2 -> U8-1 unchanged.

  • The newest pulled deltas still deepen metadata and support pages rather than replacing the current route family: mind-upload tightens human-proxy composition evidence roles outside the current six, EEG-DATA adds an Iowa Gambling EEG family deep dive while still reporting merged_unique=30873, high_confidence=28836, and access_counts_high.immediate=24486, and auto-research-funds adds the Nagahisa research grant snapshot plus a README-stat refresh without changing the practical shortlist.
  • Recounting the ### Ux-y headings on this page still returns 60, so this run keeps the one-row-at-a-time dossier structure intact.
  • The public six still stay U13-2 / U13-5 / U0-2 / U0-3 / U14-4 / U14-2.
  • The solve-first order still stays U13-2 -> U13-5 -> U0-2 -> U0-3 -> U14-4 -> U14-2.
  • The route-safe access reading stays narrow in one important way: the refreshed local EEG-DATA catalog keeps widening, but the live-access read still leaves D10 as the only clear public-open primary anchor while D03 / D02 / D01 still stay gated support anchors in public wording.
  • Browser-level live checks with a browser user agent still returned 200 for issue.html, mind-upload-current-public-six-rq-brief.html, and mind-upload-rq-solvability-bridge.html; the current Last-Modified header was Wed, 01 Apr 2026 10:21:50 GMT on those public surfaces.
  • The funding bridge still stays narrow: Kura Fund, Nakatani Foundation graduate scholarship, and Kashinome Scholarship remain the active near-term lanes, Secom General Research Grant remains closed for the current cycle, and a fresh Todoist api/v1/sync(items) audit (items=203, Kura=2, Nakatani graduate scholarship=7, Kashinome=2, Secom=3) still did not justify a new task.

2026-04-01 17:06 JST latest post-pull route-hold note

This run still does not change the current public six-question batch. An actual-pull reread after mind-upload@c5a0550, EEG-DATA@82893f6ca6e, and auto-research-funds@ec6f7d068 still leaves the current D10 / D03 / D02 / D01 family and the outside-current-six order U7-1 -> U1-2 -> U4-2 -> U8-1 unchanged.

  • The newest pulled deltas still deepen metadata rather than replacing the current route family: mind-upload deepens astrocyte PET route splits outside the current six, EEG-DATA now reports merged_unique=30871, high_confidence=28834, and access_counts_high.immediate=24484, its experiment-metadata report still covers 18,312 / 28,832 rows with at least one experiment-facing field (63.51% coverage), and auto-research-funds adds the Megachips 2026 scholarship snapshot plus a README-stat refresh without changing the practical shortlist.
  • Recounting the ### Ux-y headings on this page still returns 60, so this run keeps the one-row-at-a-time dossier structure intact.
  • The public six still stay U13-2 / U13-5 / U0-2 / U0-3 / U14-4 / U14-2.
  • The solve-first order still stays U13-2 -> U13-5 -> U0-2 -> U0-3 -> U14-4 -> U14-2.
  • The route-safe access reading stays narrow in one important way: the refreshed local EEG-DATA catalog still renders D03 / D02 as catalog-side immediate rows and D01 as a mixed immediate/check-needed pair, but the live-access read remains stricter, so D10 stays the only clear public-open primary anchor while D03 / D02 / D01 still stay gated support anchors in public wording.
  • Browser-level live checks with a browser user agent still returned 200 for issue.html, mind-upload-current-public-six-rq-brief.html, mind-upload-rq-solvability-bridge.html, and u8-1-closed-loop-delay-tolerance-route.html; the current Last-Modified header is Wed, 01 Apr 2026 07:16:09 GMT on those public surfaces.
  • The funding bridge still stays narrow: Kura Fund, Nakatani Foundation graduate scholarship, and Kashinome Scholarship remain the active near-term lanes, Secom General Research Grant remains closed for the current cycle, and a fresh Todoist api/v1/sync(items) audit (items=203, Kura=2, Nakatani family=11, Kashinome=2, Secom=3) still did not justify a new task.

2026-04-01 14:08 JST latest post-pull route-hold note

This run still does not change the current public six-question batch. An actual-pull reread after mind-upload@ccef40f, EEG-DATA@deefd5f95c6, and auto-research-funds@f5f76c485 still leaves the current D10 / D03 / D02 / D01 family and the outside-current-six order U7-1 -> U1-2 -> U4-2 -> U8-1 unchanged.

  • The newest EEG-DATA head still deepens metadata rather than replacing the current route family: the refreshed summary now reports merged_unique=30869, high_confidence=28832, and access_counts_high.immediate=24482, and the new experiment-metadata report covers 18,312 / 28,832 rows with at least one experiment-facing field (63.51% coverage).
  • Recounting the ### Ux-y headings on this page still returns 60, so this run keeps the one-row-at-a-time dossier structure intact.
  • The public six still stay U13-2 / U13-5 / U0-2 / U0-3 / U14-4 / U14-2.
  • The solve-first order still stays U13-2 -> U13-5 -> U0-2 -> U0-3 -> U14-4 -> U14-2.
  • The route-safe access reading stays narrow in one important way: the refreshed local EEG-DATA catalog still renders D03 / D02 as catalog-side immediate rows and D01 as a mixed immediate/check-needed pair, but the live-access read remains stricter, so D10 stays the only clear public-open primary anchor while D03 / D02 / D01 still stay gated support anchors in public wording.
  • Browser-level live checks with a browser user agent still returned 200 for issue.html, mind-upload-current-public-six-rq-brief.html, mind-upload-rq-solvability-bridge.html, and u8-1-closed-loop-delay-tolerance-route.html; the current Last-Modified header is Wed, 01 Apr 2026 04:20:02 GMT on those public surfaces.
  • The funding bridge still stays narrow: Kura Fund, Nakatani Foundation graduate scholarship, and Kashinome Scholarship remain the active near-term lanes, Secom General Research Grant remains closed for the current cycle, and a fresh Todoist api/v1/sync(items) audit (items=203, Kura=2, Nakatani family=11, Kashinome=2, Secom=3) still did not justify a new task.

2026-04-01 12:06 JST latest post-pull route-hold note

This run still does not change the current public six-question batch. The latest pulled deltas were reread before writing this note: mind-upload@9f4bc8e760b1, EEG-DATA@1324f9e1ae2b, and auto-research-funds@6548b1650535 still leave the current D10 / D03 / D02 / D01 family and the outside-current-six order U7-1 -> U1-2 -> U4-2 -> U8-1 unchanged.

  • The newest EEG-DATA additions still deepen metadata rather than replace the current route family: the refreshed summary now reports merged_unique=30867, high_confidence=28830, and access_counts_high.immediate=24480.
  • Recounting the ### Ux-y headings on this page still returns 60, so this run keeps the one-row-at-a-time dossier structure intact.
  • The public six still stay U13-2 / U13-5 / U0-2 / U0-3 / U14-4 / U14-2.
  • The solve-first order still stays U13-2 -> U13-5 -> U0-2 -> U0-3 -> U14-4 -> U14-2.
  • Browser-level live checks with a browser user agent still returned 200 for issue.html, mind-upload-current-public-six-rq-brief.html, mind-upload-rq-solvability-bridge.html, and u8-1-closed-loop-delay-tolerance-route.html; the current Last-Modified header is Wed, 01 Apr 2026 02:26:28 GMT on those public surfaces.
  • The funding bridge still stays narrow: Kura Fund, Nakatani Foundation graduate scholarship, and Kashinome Scholarship remain the active near-term lanes, Secom General Research Grant remains closed for the current cycle, and the latest canonical funding-shortlist audit still did not justify a new Todoist route.

2026-04-01 11:06 JST latest post-pull route-hold note

This run still does not change the current public six-question batch. The latest pulled deltas were reread before writing this note: mind-upload@aa21ab9855c1, EEG-DATA@cccd137a3468, and auto-research-funds@3a691196cb9f still leave the current D10 / D03 / D02 / D01 family and the outside-current-six order U7-1 -> U1-2 -> U4-2 -> U8-1 unchanged.

  • The newest EEG-DATA additions still deepen metadata rather than replace the current route family: the refreshed summary now reports merged_unique=30866, high_confidence=28829, and access_counts_high.immediate=24479.
  • Recounting the ### Ux-y headings on this page still returns 60, so this run keeps the one-row-at-a-time dossier structure intact.
  • The public six still stay U13-2 / U13-5 / U0-2 / U0-3 / U14-4 / U14-2.
  • The solve-first order still stays U13-2 -> U13-5 -> U0-2 -> U0-3 -> U14-4 -> U14-2.
  • Browser-level live checks still returned 200 for issue.html, mind-upload-current-public-six-rq-brief.html, mind-upload-rq-solvability-bridge.html, and u8-1-closed-loop-delay-tolerance-route.html; Last-Modified advanced to Wed, 01 Apr 2026 01:21:12 GMT for issue.html / u8-1 and Wed, 01 Apr 2026 01:21:13 GMT for the two bridge pages, so the current public-facing route stays aligned.
  • The funding bridge still stays narrow: Kura Fund, Nakatani Foundation graduate scholarship, and Kashinome Scholarship remain the active near-term lanes, Nakatani Foundation research grant, Secom General Research Grant, and Brain Science Foundation research grant remain outside the current-cycle lane, and a fresh Todoist api/v1/sync(items) audit (items=203) still did not require a new task.

2026-04-01 03:09 JST latest post-pull route-hold note

This run still does not change the current public six-question batch. The latest pulled deltas were reread before writing this note: mind-upload@31835fcd8daa, EEG-DATA@12eef8ca4ef0, and auto-research-funds@040e5b8798fd still leave the current D10 / D03 / D02 / D01 family and the outside-current-six order U7-1 -> U1-2 -> U4-2 -> U8-1 unchanged.

  • mind-upload split temporal routes for invasive language BCIs outside the current public six, EEG-DATA added a semantic-text-relevance deep dive that widens language-related backup coverage without replacing the current route family, and auto-research-funds refreshed Obayashi / Yokogaku cards without changing the practical shortlist.
  • Recounting the ### Ux-y headings on this page still returns 60, so this run keeps the one-row-at-a-time dossier structure intact.
  • The public six still stay U13-2 / U13-5 / U0-2 / U0-3 / U14-4 / U14-2.
  • The solve-first order still stays U13-2 -> U13-5 -> U0-2 -> U0-3 -> U14-4 -> U14-2.
  • Browser-level live checks at 2026-04-01 03:09 JST still returned 200 for issue.html, mind-upload-current-public-six-rq-brief.html, mind-upload-rq-solvability-bridge.html, and u8-1-closed-loop-delay-tolerance-route.html, so the current public-facing route stays aligned.
  • The funding bridge still stays narrow: Kura Fund, Nakatani Foundation graduate scholarship, and Kashinome Scholarship remain the active near-term lanes, Nakatani Foundation research grant, Secom General Research Grant, and Brain Science Foundation research grant remain outside the current-cycle lane, and a fresh Todoist api/v1/sync(items) audit (items=203) still did not require a new task.

2026-03-31 21:09 JST latest post-pull route-hold note

This run still does not change the current public six-question batch. The latest pulled deltas were reread before writing this note: mind-upload@ed6f8203e46d, EEG-DATA@20fbbad9959f, and auto-research-funds@77f189662745 still leave the current D10 / D03 / D02 / D01 family and the outside-current-six order U7-1 -> U1-2 -> U4-2 -> U8-1 unchanged.

  • mind-upload tightened dataset-benchmark object rules and clarified the FAQ RNA route split, EEG-DATA added request-gated HBUED emotion-recognition coverage plus a continuous-pursuit online BCI metadata deep dive, and auto-research-funds added Teraura Sayoko outbound scholarship, Gakusho scholarship, and README-stat refreshes, but none of those changes replace the current route family.
  • Recounting the ### Ux-y headings on this page still returns 60, so this run keeps the one-row-at-a-time dossier structure intact.
  • The public six still stay U13-2 / U13-5 / U0-2 / U0-3 / U14-4 / U14-2.
  • The solve-first order still stays U13-2 -> U13-5 -> U0-2 -> U0-3 -> U14-4 -> U14-2.
  • Browser-level live checks at 2026-03-31 21:09 JST still returned 200 for issue.html, mind-upload-current-public-six-rq-brief.html, and u8-1-closed-loop-delay-tolerance-route.html, so the current public-facing route stays aligned.
  • The funding bridge still stays narrow: Kura Fund, Nakatani Foundation graduate scholarship, and Kashinome Scholarship remain the active near-term lanes, Nakatani Foundation research grant, Secom General Research Grant, and Brain Science Foundation research grant remain outside the current-cycle lane, and no new Todoist item was created in this session.

2026-03-31 19:07 JST latest post-pull route-hold note

This run still does not change the current public six-question batch. The latest pulled deltas were reread before writing this note: mind-upload@b7b6c0498ae1, EEG-DATA@24e85cc20f09, and auto-research-funds@0fddd53bd03c still leave the current D10 / D03 / D02 / D01 family and the outside-current-six order U7-1 -> U1-2 -> U4-2 -> U8-1 unchanged.

  • mind-upload strengthened the front-door human-proxy split, EEG-DATA refreshed the catalog to merged_unique=30860, high_confidence=28823, and access_counts_high.immediate=24471, and auto-research-funds refreshed Kioxia/Takayanagi funding records, but none of those changes replace the current route family.
  • Recounting the ### Ux-y headings on this page still returns 60, so this run keeps the one-row-at-a-time dossier structure intact.
  • The public six still stay U13-2 / U13-5 / U0-2 / U0-3 / U14-4 / U14-2.
  • The solve-first order still stays U13-2 -> U13-5 -> U0-2 -> U0-3 -> U14-4 -> U14-2.
  • Browser-level live checks at 2026-03-31 19:09 JST still returned 200 for issue.html and mind-upload-current-public-six-rq-brief.html, and the live issue.html paragraph still names the current six in the same order.
  • The funding bridge still stays narrow: Kura Fund, Nakatani Foundation graduate scholarship, and Kashinome Scholarship remain the active near-term lanes, Nakatani Foundation research grant and Secom General Research Grant are already closed for the current cycle, and no new Todoist item was created in this session.

2026-03-31 17:06 JST latest post-pull route-hold note

This run still does not change the current public six-question batch. The latest pulled deltas were reread before writing this note: mind-upload@4731680627ff, EEG-DATA@3e7a53d9ca8, and auto-research-funds@ef2924014a65 still leave the current D10 / D03 / D02 / D01 family and the outside-current-six order U7-1 -> U1-2 -> U4-2 -> U8-1 unchanged.

  • The newest EEG-DATA additions widen the support map rather than the route: the NTU Dataverse adult-to-infant unidirectional neural-coupling raw/preprocessed releases are immediately downloadable and useful for future social-learning or paired-observer rows, while OpenNeuro ds007146 still stays uncertain_public because the visible object tree does not yet verify the claimed raw EEG / Ear-EEG coverage.
  • Recounting the ### Ux-y headings on this page still returns 60, so this run keeps the one-row-at-a-time dossier structure intact.
  • The public six still stay U13-2 / U13-5 / U0-2 / U0-3 / U14-4 / U14-2.
  • The solve-first order still stays U13-2 -> U13-5 -> U0-2 -> U0-3 -> U14-4 -> U14-2.
  • Browser-level live checks with a browser user agent still returned 200 for issue.html, mind-upload-current-public-six-rq-brief.html, and u8-1-closed-loop-delay-tolerance-route.html.
  • The funding bridge still stays narrow: Kura Fund, Nakatani Foundation graduate scholarship, and Kashinome Scholarship remain the active near-term lanes, Secom General Research Grant is already closed for the current cycle (2026-03-31 12:00 JST) and stays only as a carried-forward watch record, and no new Todoist item was created in this session.

2026-03-31 16:05 JST latest post-pull route-hold note

This run still does not change the current public six-question batch. The latest pulled deltas were reread before writing this note: mind-upload@a11240d38415 tightened the connectome / timing-state route cards, EEG-DATA@f318ac8ce42d added an OSF eucqf deep-dive refresh, and auto-research-funds@c95cc61fdfc9 added new Takeda / Nohken funding-card material. None of those changes replace the current D10 / D03 / D02 / D01 anchor family or the outside-current-six order U7-1 -> U1-2 -> U4-2 -> U8-1.

  • The public six still stay U13-2 / U13-5 / U0-2 / U0-3 / U14-4 / U14-2.
  • The solve-first order still stays U13-2 -> U13-5 -> U0-2 -> U0-3 -> U14-4 -> U14-2.
  • Browser-level live checks in this run still returned 200 for issue.html, mind-upload-current-public-six-rq-brief.html, and u8-1-closed-loop-delay-tolerance-route.html, so this run adds no new public-site prose.
  • The Todoist full-item audit still showed enough coverage (items=205, Kura=2, Nakatani family=11, Kashinome=2, Secom family=3), so no new task was added.

2026-03-31 00:04 JST overnight route-hold note

This run still does not change the current public six-question batch. The added value is not a new batch switch but a stronger reason for the hold: the latest EEG-DATA upstream additions remain outside the current speech / identity / reproducibility anchor family, D10 remains the only clearly public-open primary anchor, and D03/D02/D01 now have to be read only as subscription-gated support anchors rather than immediate public anchors.

  • The public six still stay U13-2 / U13-5 / U14-4 / U0-2 / U0-3 / U14-2.
  • The solve-first order still stays U13-2 -> U13-5 -> U0-2 -> U0-3 -> U14-4 -> U14-2.
  • Official-window recheck at 2026-03-31 00:04 JST still supports the same narrow funding bridge: Kura Fund = 2026-04-17 17:00, Nakatani Foundation graduate scholarship = 2026-04-01 - 2026-05-25 15:00, Kashinome = 2026-04-09, and Secom General Research Grant = 2026-03-31 12:00.
  • Todoist action still stays conservative in this run. The full-item audit continues to show the current route already covered (Kura=2, Nakatani family=11, Kashinome=2, Secom family=3), so no new task was added.
  • Earlier March 30 notes on this page that called D03/D02/D01 immediate public anchors should now be read as superseded by the March 31 access audit.

2026-03-30 live evidence note for the current public six

This run keeps the current public batch unchanged. The new contribution is row-level evidence wording that can be quoted safely in public pages without inflating the claim ceiling. Each active row now has one explicit EEG-DATA reason, one bounded first artifact, and one funding sentence that stays consistent with the current shortlist.

RQ EEG-DATA reason that still justifies the row First artifact that can be closed now Funding sentence that still fits No-go boundary that stays explicit
U13-2 D10 3M-CPSEED remains the strongest current speech anchor because the local mirror still exposes an immediate OpenNeuro route with overt, silent-intended, and imagined speech across 20 participants x 4 sessions, plus raw BIDS and phase-split derivatives. Brain-minus-prior control table and mimic-separation note. Keep Kura Fund first and Nakatani Foundation graduate scholarship second, because the row still reads naturally as a bounded young-researcher / doctoral-track speech-decoding benchmark rather than a solved mind-upload claim. Do not translate this row into evidence of preserved internal mechanism, preserved identity, or successful mind preservation.
U13-5 D03 Multimodal Fusion EEG + Eye-Tracking remains only a gated planning anchor because the intended paired EEG and gaze route is still useful, but the visible DataPort metadata centers a 55-student cognitive-load dataset and does not expose a recall-specific label on the public page. Perception-to-recall degradation note and transfer-gap table. Keep Kura Fund first and Nakatani Foundation graduate scholarship second, because the row is still a decoder-transfer benchmark with a clear failure mode rather than a broader mechanism claim. Do not write this as evidence that shared decoder structure implies shared internal mechanism across perception and recall.
U0-2 D02 CSTE remains the cleanest intended identity-drift floor because the local mirror still places it in Biometric / Authentication and Cross-session / Transfer / Domain adaptation, with 10 participants, even though the current DataPort page is subscription-gated. Offset/jitter distribution and state-feature collapse audit. Keep Nakatani Foundation graduate scholarship first and Kashinome Scholarship second, because the row still fits a student-facing continuity / drift audit better than a broad systems proposal. Do not elevate timing stability or re-identification robustness into legal, philosophical, or normative identity continuity.
U0-3 The same D02 anchor remains correct because threshold stability only becomes interpretable after the identity row stays on a cross-session / cross-task biometric floor instead of a single-session score. Threshold-stability note and overfit-exclusion rule. Keep Nakatani Foundation graduate scholarship first and Kashinome Scholarship second, because the row still reads as a doctoral-track identity-drift benchmark with bounded operational claims. Do not present a stable threshold as a sufficient condition for personhood continuity or branch-identity resolution.
U14-4 D01 Acquisition delay of wireless EEG remains acceptable only as a bounded entry point, because the visible package is a latency/synchronization dataset with README_en plus device/run .mat files rather than a rerun benchmark by itself. Filled Model Card / Dataset Card examples for the current batch. Keep Kura Fund first and Nakatani Foundation research grant as next-cycle watch, because the row is still strongest as a reproducibility/disclosure package rather than a current open BME result claim. Do not turn card completeness into field-wide reproducibility, deployment readiness, or standards adoption.
U14-2 The same D01 anchor remains enough to freeze the split boundary only as a latency-first floor; broader split or rerun wording still needs D13 as the first public-open support family. Fixed-split rerun contract and negative-case trail. Keep Kura Fund first and Nakatani Foundation research grant as next-cycle watch, because the row still functions as a bounded reproducibility-contract artifact. Do not present exploration outputs as confirmatory evidence or imply broader reproducibility than the current anchor can support.
  • Todoist action stayed intentionally conservative in this run. A fresh read-only API recheck still found the existing Kura, Nakatani graduate scholarship, Kashinome, Nakatani research grant, and Secom tasks already present, so the correct action was to avoid duplication rather than add another reminder layer.

2026-03-30 13:08 JST row packets for the current public six

The current public batch still does not change. What changes in this addendum is the working granularity for the six active rows: each one now has one submission-ready title, one explicit anchor reason, one bounded first artifact, and one sentence about what still blocks stronger wording.

RQ Submission-ready title Current first artifact Why the present anchor is enough now Funding bridge that still fits What still blocks stronger wording
U13-2 Neural-contribution audit for imagined speech with a brain-minus-prior control Brain-minus-prior control table and mimic-separation note D10 3M-CPSEED still keeps overt, silent-intended, and imagined speech inside one public release, so the first artifact can compare a neural contribution against a prior-only shortcut instead of collapsing into a generic decoder score. Kura Fund, then Nakatani Foundation graduate scholarship The row still does not preserve internal mechanism, preserved identity, or successful mind preservation.
U13-5 Perception-to-recall transfer-gap benchmark for mimic separation Perception-to-recall degradation note and transfer-gap table D03 Multimodal Fusion EEG + Eye-Tracking still provides the intended paired EEG + gaze route, but the visible DataPort metadata is still a cognitive-load package and does not expose a recall-specific label on the public page, so it only freezes a bounded planning branch point rather than an immediate public anchor. Kura Fund, then Nakatani Foundation graduate scholarship The row still cannot support a stronger preserved-mechanism reading without an external recall-validity layer and a richer public-open adjunct benchmark.
U0-2 Temporal-validity-aware identity drift audit with offset, jitter, and state-feature collapse Offset/jitter distribution and state-feature collapse audit D02 CSTE remains cross-session, cross-task, and biometric, but the current DataPort page is subscription-gated, so identity-like wording can first be reduced only to a bounded timing-and-feature audit. Nakatani Foundation graduate scholarship, then Kashinome Scholarship The row still cannot cross the line into legal, philosophical, or normative identity continuity.
U0-3 Task-specific threshold stability and overfit-exclusion benchmark for identity drift Threshold-stability note and overfit-exclusion rule The same D02 anchor keeps threshold movement on the same cross-session / cross-task floor as U0-2, which is the minimum condition for interpreting threshold stability at all. Nakatani Foundation graduate scholarship, then Kashinome Scholarship The row still cannot treat a stable threshold as sufficient for personhood continuity or branch-identity resolution.
U14-4 Card-integrated reproducibility contract for the current public batch Filled Model Card / Dataset Card examples D01 Acquisition delay of wireless EEG is narrow, but it is still enough to force latency and measurement fields into concrete card examples that a rerun contract can actually reference; stronger public wording would still need D13 as the first public-open rerun family. Kura Fund; Nakatani Foundation research grant as next-cycle watch The row still cannot justify field-wide reproducibility, deployment readiness, or standards uptake wording.
U14-2 Fixed-split rerun contract for exploration versus confirmation Fixed-split rerun contract and negative-case trail The same D01 route is still sufficient to make an exploration / confirmation split operational as a narrow latency floor, but stronger rerun wording would still need D13 as the first public-open rerun family. Kura Fund; Nakatani Foundation research grant as next-cycle watch The row still cannot support broad rerun-robustness wording or confirmatory language for exploratory outputs.

今回の再検証ログ(2026-03-30 02:05 JST, one-RQ closure gates)

本runでは、current public 6RQ の batch 自体は変えずに、「今この6件で何を閉じたら public-safe に進捗と呼べるか」を RQ ごとに明文化いたしました。ここで固定するのは solved claim ではなく、1RQ=1最初の成果物=1非主張境界 の closure gate です。

  • current public 6RQ と solve-first 順は維持:
    • U13-2 -> U13-5 -> U0-2 -> U0-3 -> U14-4 -> U14-2
  • official / Todoist 判断も据え置き:
    • くら基金=2026-04-17 17:00
    • 中谷財団 大学院生奨学金=2026-04-01 - 2026-05-25 15:00
    • 樫の芽会=2026-04-09
    • セコム一般研究助成=2026-03-31 12:00
    • 中谷財団 研究助成=next-cycle watch
    • current route / watch route を支える Todoist task は既存で足りているため、新規追加なし。
RQ 今回の closure gate この gate を先に閉じる理由 主アンカー claim ceiling を上げる前に次に要るもの
U13-2 brain-minus-prior control table と mimic-separation note language decoder を一括りにせず、neural contribution の最小監査を先に閉じる必要があるため。 D10 external language-evaluation layer と control-condition audit
U13-5 perception-to-recall degradation note と transfer-gap table branch point を先に固定しないと mimicry / mechanism の過読が残るため。 D03 recall-label validity と adjunct speech/control benchmark
U0-2 offset/jitter distribution と state-feature collapse audit identity-like wording へ measurement error が混入する境界を先に切る必要があるため。 D02 stronger longitudinal adjunct benchmark
U0-3 threshold-stability note と overfit-exclusion rule 閾値の外的妥当性を閉じないと本人性ドリフト読解が静的一致へ戻りやすいため。 D02 session-external rerun pack と cohort extension
U14-4 filled Model Card / Dataset Card examples rerun contract を実物で埋めない限り disclosure claim が抽象論のまま残るため。 D01 adjunct rerun benchmark と leak-audit exemplars
U14-2 fixed-split rerun contract と negative-case trail exploration / confirmation split を運用文書へ落とさないと reproducibility claim が曖昧なままだからです。 D01 multi-benchmark rerun trail と split-violation examples

2026-03-30 05:07 JST addendum (current six-RQ anchor-strength recheck)

This run does not change the current public six-question batch. It only tightens how the present anchors should be read after a post-pull review of EEG-DATA, a live check of mind-upload.com/issue.html, an official-window recheck, and a read-only Todoist recheck. The operational rule stays narrow: D10 remains the only clearly public-open primary anchor for first artifacts, while D03, D02, and D01 remain subscription-gated support anchors that are still useful for the current public route but should not carry broader mechanism or field-wide reproducibility wording by themselves.

RQ Why this anchor still stays in the current batch What the current anchor is good enough to close now What must still stay outside the claim Best next adjunct before stronger wording
U13-2 D10 3M-CPSEED stays the strongest current speech-side anchor because it is public, BIDS-structured, and exposes overt, silent-intended, and imagined speech inside the same release. A brain-minus-prior control route and a narrow mimic-separation note. Preserved internal mechanism or any mind-preservation wording. An external language-evaluation layer plus a richer control family.
U13-5 D03 stays acceptable only as a bounded multimodal support anchor because it combines EEG with eye-tracking and related physiology in a subscription-gated release whose visible metadata still does not expose a recall-specific label. A perception-to-transfer branch note that stays narrow and public-safe. Any stronger mechanism-preservation reading. ChineseEEG-2 plus ZuCo 2.0 and the public robot-speech family, with SparrKULee kept as the larger auditory-only stress-test and CuBiAAD only as a later access-gated adjunct.
U0-2 D02 CSTE remains the cleanest current identity-drift floor because it is explicitly cross-session and cross-task with a biometric framing, even though the current DataPort page is subscription-gated. A state-feature collapse audit plus bounded subject-stability failure analysis. Philosophical or legal continuity claims, and stronger timing language than the dataset actually supports. A longer-horizon adjunct benchmark with more explicit timing and synchronization logs.
U0-3 The same D02 anchor keeps the threshold rule on the same cross-session / cross-task identity-drift floor as U0-2. Threshold stability and overfit-exclusion as a bounded benchmark artifact. Treating a stable threshold as a sufficient condition for personhood continuity. A session-external rerun pack and a larger cohort extension.
U14-4 D01 remains a narrow but useful disclosure anchor because a wireless-delay dataset is enough to force concrete latency and measurement fields into a Model Card / Dataset Card example. A disclosure-first rerun checklist with concrete latency fields. Any field-wide reproducibility or organizational-adoption wording. A New Benchmark Dataset Towards Ubiquitous P300 ERP-based BCI Applications as the first required public-open rerun family, then D08 for grounded validation examples.
U14-2 The same D01 route is still enough to keep the exploration/confirmation split operational and measurable at the latency-contract level. A fixed-split rerun contract plus a negative-case trail that stays narrow. Broader rerun robustness claims. The same public P300 benchmark-style family as the first required support, plus D08 for stronger split-violation grounding.
  • Funding-route judgment did not widen in this run: Kura Fund and the Nakatani Foundation graduate scholarship still carry the immediate bridge, Kashinome Scholarship stays runway-only support, and the Nakatani Foundation research grant plus Secom General Research Grant stay watch routes.
  • The read-only Todoist recheck still found existing task coverage for the current practical routes, so this run added no new task.
  • 高齢社会課題解決研究(AI) still remains outside the current EEG-first route and still stays deferred rather than added.

今回の再検証ログ(2026-03-29 19:07 JST)

本runでは、pull 後の current public 6RQ を再度据え置いたまま、official funding window と Todoist 実 API の「重複なく足りているか」を確認し、さらに broader funding watch が既存 task とどこまで接続済みかを整理いたしました。方針は今回も総論追加ではなく、1RQ=1検証命題=1主アンカー=1第一資金ルート=1非主張境界 を崩さないことです。

  • current public 6RQ と solve-first 順は維持:
    • U13-2 -> U13-5 -> U0-2 -> U0-3 -> U14-4 -> U14-2
  • official funding window の再確認:
    • くら基金: 2026-04-17 17:00 まで open
    • 中谷財団 大学院生奨学金: detail page で 2026-04-01 - 2026-05-25 15:00
    • 樫の芽会: 2026-04-09 まで open
    • 中谷財団 研究助成: 2026年度募集は終了済み(2025-06-01 - 2025-07-07)
  • Todoist 判断:
    • current route に必要な task は今回も既存で、新規 task は追加しておりません。
    • 実 API では current route 周辺で 11 task がヒットしましたが、これは 中谷財団 大学院生奨学金 / 大学院生奨学金給付中谷財団 研究助成 周辺に重複系の watch があるためです。現行 public 6RQ に必要な最小 task は既に足りているため、今回の正しい処理は 追加しない ことでした。
  • broader funding watch の扱い:
    • ブレインサイエンス振興財団 島津科学技術振興財団 精密測定技術振興財団 服部報公会 リカケンホールディングス は、Todoist 側に既存 watch があることを確認いたしました。
    • 高齢社会課題解決研究(AI) は今回の API 再確認では task 未登録でしたが、current public 6RQ の immediate route ではなく、しかもこの run では 2026 サイクルの official recurrence を独立再確認していないため、新規追加は見送りました。
  • public 反映:
    • live mind-upload.com/issue.html は引き続き current public 6RQ と solve-first order を自然な英語で保持していたため、本runでも公開ページ本文の差し替えは行っておりません。

今回の再検証ログ(2026-03-29 15:15 JST)

本runでは、current public 6RQ を据え置いたまま、anchor strengthroadmap-only gap をドシエ入口へ明示いたしました。総論を増やすのではなく、「どの RQ が今の EEG-DATA で最も深く前に進みやすいか」をさらに明確にするための補足です。

  • solve-first 順は維持: U13-2 U13-5 U0-2 U0-3 U14-4 U14-2
  • anchor の強弱:
    • clear public-open primary: D10 3M-CPSEED
    • gated planning primary for identity rows: D02 CSTE
    • bounded support: D03 Multimodal Fusion EEG + Eye-Tracking, D01 Acquisition delay of wireless EEG
  • 実務含意:
    • U13-2/U0-2/U0-3 は現行 EEG-DATA だけでも first artifact が比較的閉じやすい行です。
    • U13-5/U14-4/U14-2 は public route には残しますが、D03/D01 のまま claim ceiling を上げすぎず、DOI + dataset name + access class を正本にして narrow artifact から進めます。U13-5 は recall-validity layer、U14-4/U14-2D13 が次の必須 support です。
  • roadmap gap:
    • tech_roadmap.md 上の U2/U5/U6/U9 は、まだこの RQ60 practical dossier 層に翻訳されておりません。current public batch の差し替え対象ではなく、次に増やす internal translation backlog として扱います。

今回の再検証ログ(2026-03-29 13:03 JST)

本runでは、mind-upload@12d3d42 EEG-DATA@8d3d7db889a auto-research-funds@eaa5633d0 まで pull 後に、live mind-upload.com/issue.html、主要制度の公式募集ページ、Todoist 実 API を再確認し、current public batch を据え置く判断を明文化いたしました。

  • current public 6RQ(本run正本): U13-2 U13-5 U14-4 U0-2 U0-3 U14-2
  • 主アンカー固定は維持:
    • U13-2: D10
    • U13-5: D03
    • U14-4: D01
    • U0-2: D02
    • U0-3: D02
    • U14-2: D01
  • funding route の再確認:
    • U13-2/U13-5: くら基金 / 中谷財団 大学院生奨学金
    • U14-4/U14-2: くら基金 / 中谷財団 研究助成(next-cycle watch)
    • U0-2/U0-3: 中谷財団 大学院生奨学金 / 樫の芽会(runway 補助のみ)
  • Todoist判断:
    • くら基金 中谷財団 大学院生奨学金 樫の芽会 中谷財団 研究助成募集開始確認/応募締切 8 task は 13:03-13:05 JST の実 API 再確認でも既存でした。本runでは新規 task を追加しておりません。
  • 補足:
    • live public route と source page の文面は今回の bounded 6RQ と一致していたため、本runでは mind-upload.com 側の public source page を更新しておりません。

今回の再検証ログ(2026-03-29 12:06 JST)

本runでは、pull 後に live mind-upload.com/issue.html を再確認し、current public batch を U13-2/U13-5/U14-4/U0-2/U0-3/U14-2 として内部ドシエの入口を同期し直しました。汎用横断の総論は増やさず、ここでも 1RQ=1検証命題=1応募テーマ=1主データ を維持いたします。

  • current public 6RQ(本run正本): U13-2 U13-5 U14-4 U0-2 U0-3 U14-2
  • 主アンカー固定:
    • U13-2: D10
    • U13-5: D03
    • U14-4: D01
    • U0-2: D02
    • U0-3: D02
    • U14-2: D01
  • funding route の再固定:
    • U13-2/U13-5: くら基金 / 中谷財団 大学院生奨学金
    • U14-4/U14-2: くら基金 / 中谷財団 研究助成(next-cycle watch)
    • U0-2/U0-3: 中谷財団 大学院生奨学金 / 樫の芽会(runway 補助のみ)
  • Todoist判断:
    • くら基金 中谷財団 大学院生奨学金 樫の芽会 中谷財団 研究助成募集開始確認/応募締切 task は実 API でも既存確認できたため、本runでは新規 task 追加を行っておりません。
  • 補足:
    • 中谷財団 研究助成 は official page 上で 2026年度 募集終了が明記されていたため、この run では U14-4/U14-2 の immediate route から外し、next-cycle watch としてのみ残します。
    • 07:18 に残っている U0-1/U1-2/U4-4/U7-1/U8-4/U11-1 は earlier internal candidate であり、current public route ではありません。

今回の再検証ログ(2026-03-29 07:18 JST)

本runでは、mind-upload.com 側でそのまま公開しても不自然になりにくい6RQを先に固定し、その6件だけを EEG-DATA 実在面・現行助成面・Todoist登録状況まで含めて再検証いたしました。汎用横断の総論は増やさず、今回も 1RQ=1検証命題=1応募テーマ=1主データ を維持いたします。

  • 深掘り再固定6RQ(本run): U0-1 U1-2 U4-4 U7-1 U8-4 U11-1
  • この 07:18 候補は、12:06 の live 再確認により current public batch U13-2/U13-5/U14-4/U0-2/U0-3/U14-2 へ superseded されました。
  • 主アンカー固定:
    • U0-1: D02
    • U1-2: D08
    • U4-4: D05
    • U7-1: D11
    • U8-4: D20
    • U11-1: D14
  • Todoist判断:
    • 関連する主要制度の task は既に登録済みであることを確認したため、本runでは新規 task 追加を行っておりません。

今回の再検証ログ(2026-03-29 05:55 JST)

本runでは、既存の 6RQ 深掘りキューを維持したまま、各RQを EEG-DATA の実在 bucket と auto-research-funds の現行 card 実体へ再固定いたしました。汎用横断の総論は増やさず、今回も 1RQ=1検証命題=1応募テーマ=1主データ を維持いたします。

  • 深掘り再固定6RQ(本run): U0-2 U1-4 U7-6 U8-6 U11-3 U13-2
  • 主アンカー固定:
    • U0-2: D02
    • U1-4: D08
    • U7-6: D11
    • U8-6: D12
    • U11-3: D17
    • U13-2: D10
  • 現行助成面の固定:
    • G1/G2 は routing key として維持いたします。
    • 実体参照は G3-G6 に戻し、正本一覧は auto-research-funds/wiki/Mind-Upload-Current-Funding-Shortlist.md といたしました。
  • 再固定した実データ条件:
    • U0-2: repeated sessions / same-subject labels / task variation / timestamp-QC / device metadata
    • U1-4: high-density EEG / intervention ground-truth / head-model or iEEG disclosure / rerunnable pipeline
    • U7-6: same-trial synchronization / BIDS-sidecar provenance / missing-modality policy / subject-session count
    • U8-6: online labels / latency-jitter log / fixed-vs-recalibrated decoder disclosure / failure registry
    • U11-3: same-cohort calibration / continuous recordings / hypnogram or sedation-stage labels / pharmacology or perturbation log
    • U13-2: stimulus-audio-transcript alignment / overt-vs-covert control / leak-free split / BIDS or equivalent provenance

今回の再検証ログ(2026-03-29 02:04 JST)

本runでは、現行 mind-upload の site deepening audit を強く反映する 6RQ だけを追加で固定いたしました。汎用横断の要約は増やさず、1RQ=1検証命題=1応募テーマ=1主データ を維持いたします。

  • 深掘り固定6RQ(本run): U0-2 U1-4 U7-6 U8-6 U11-3 U13-2
  • 意図:
    • U0-2Temporal Validity Cardstate annotation の分離を、同一性RQへ直結させるためです。
    • U1-4 は inverse route を field visibility / conductivity / solver uncertainty / validation class の4ゲートで読み直すためです。
    • U7-6 は multimodal gain と bundle robustness を混同しないよう、missing-modality policy を主張単位へ昇格させるためです。
    • U8-6same-session fastfixed-decoder durability、さらに rescue-mode recalibration を分けて読む必要があるためです。
    • U11-3 は理論比較を 4-gate で通さない限り、意識proxyの過読が残るためです。
    • U13-2 は language-facing route を一括りにせず、brain-minus-prior baseline を含む模倣分離へ戻すためです。
  • 主データ固定:
    • U0-2: D02
    • U1-4: D08
    • U7-6: D11
    • U8-6: D12
    • U11-3: D17
    • U13-2: D10
  • 根拠にした現行site論点:
    • site_deepening_audit_2026-03-21_temporal_validity_state_trait_drift
    • site_deepening_audit_2026-03-27_measurement_modeling_inverse_gate_sync
    • site_deepening_audit_2026-03-26_multimodal_bundle_robustness
    • site_deepening_audit_2026-03-26_closed_loop_temporal_deployability_split
    • site_deepening_audit_2026-03-27_consciousness_theory_map_gate_sync
    • site_deepening_audit_2026-03-26_literature_brain_decode_route_split
  • 再計数: RQ見出し=60、判定内訳 A/B/C=17/25/18 を維持いたしました。
  • 参照データ運用: 提出時は数値IDではなく Dxx + DOI + dataset名 + access区分 を正本として扱います。
  • 助成キー運用: G1-G6 の固定レーンは維持し、今回は G1/G4G2/G3/G5 を中心に、実装しやすい A/B を優先いたします。

今回の再検証ログ(2026-03-29 01:37 JST)

本runでは、A/B 判定の中でも「今の EEG-DATA で最初の実験と助成テーマを切りやすいもの」を優先し、可解性優先パックとして 6RQ を固定いたしました。汎用横断の要約は増やさず、1RQ=1検証命題=1応募テーマ=1主データ を継続しています。

  • 深掘り固定6RQ(本run): U1-1 U4-2 U7-1 U11-1 U13-5 U14-4
  • 意図:
    • U1-1U4-2 は、逆問題/因果の中核主張を EEG 側だけで閉じやすいためです。
    • U7-1U14-4 は、後続RQの再現性・監査運用を底上げする基盤RQであるためです。
    • U11-1U13-5 は、B 判定でも失敗条件を前もって書けるため、助成文面へ落とし込みやすいためです。
  • 主データ固定:
    • U1-1: ID 56
    • U4-2: ID 19
    • U7-1: ID 676
    • U11-1: ID 735
    • U13-5: ID 13
    • U14-4: ID 6
  • 再計数: RQ見出し=60、判定内訳 A/B/C=17/25/18 を維持しました。
  • 参照データ運用: 提出時は ID 再採番ドリフト回避のため DOI + dataset名 + access区分 を必須といたします。
  • 助成キー運用: G1-G6 の固定レーンを継続し、今回は G1/G4G1/G3 を中心に、実装しやすい A/B を先に詰めます。

判定内訳(再計数)

  • A: 17
  • B: 25
  • C: 18

使い方(1RQずつ深掘りする運用)

  1. 対象RQの 検証命題判定境界 を先に固定する。
  2. 主データ で最小再現実験を実施し、初期KPI を埋める。
  3. 停止条件 に抵触したら主張を下方修正し、応募テーマを分割する。
  4. 最低成果物 を完成させてから次RQへ進む。

今回の再検証ログ(2026-03-28 09:02 JST)

本runでは、汎用横断の更新を増やさず、1RQ=1検証命題=1応募テーマ=1主データ の方針を維持したまま、前runと重複しにくい6RQを新規の深掘り固定キューとして設定しました。

  • 深掘り固定6RQ(本run): U0-3 U1-4 U4-1 U8-6 U13-2 U15-3
  • 目的: 同一性閾値運用、逆問題の公開基準、観測相関→因果判定、閉ループ再学習頻度、模倣/機構分離、法域差運用をそれぞれ1件ずつ詰める。
  • 再計数: RQ見出し=60、判定内訳 A/B/C=17/25/18 を維持。
  • 参照データ運用: 提出時はID再採番ドリフト回避のため DOI + dataset名 + access区分 を必須とする。
  • 応募キー運用: G1-G6 の固定レーンを継続し、A は再現実装、B は外部依存境界つき実装、C は制度/法務・他モダリティ接続で提出する。

RQ-by-RQ Deep Dossiers(60件)

U0-1 同一性判定を『観測一致』と『介入応答一致』に分離したとき、どちらを必要条件・十分条件に置くか。

  • 検証命題: 観測一致(EEG特徴量)と介入一致(刺激/課題変更時の応答)を分離して二軸ROCを作る。
  • EEG-DATA判定境界: A(直接検証可能)EEG-DATA単独で主要仮説まで検証可能(外部依存なし)。
  • auto-research-funds応募テーマ: 若手研究助成(同一性KPI)
  • 第一応募先 / 予備応募先: G2 (GR-2026-014) / G3 (9Lx4dPK6a4k2gOb7)
  • 主データ(ID): 11
  • 補助データ(推奨ID): ID 11, 29, 696
  • 初期KPI(U0標準): セッション外AUC 観測一致率 介入応答一致率
  • 停止条件: interval coverage80-100% 帯を継続逸脱するか、導電率・電極欠損・ノイズ摂動後に 手法順位一致率 < 0.80 へ低下する場合は、主張を降格する。
  • 最低成果物: Intervention protocol and rebuttal condition definition
  • Run update (2026-03-29 07:18 JST): This section is now locked to D02 CSTE as the primary anchor, with D11/D23 reserved only for cross-session and physiological side-channel robustness checks.
  • First-pass KPI fixed in this run: a public pass requires both observation-match AUC and intervention-response match rate to clear threshold; a single-axis pass is rejected by design.

U0-2 時間同期(ms単位)と状態表現(行動・神経活動・生理)の対応をどう固定するか。

  • 検証命題: D02 CSTE (10.21227/j162-nh61 / subscription-gated support) を主に、D11D23 を補助として、時刻同期ずれが状態特徴量と同一性AUCへ伝播する許容帯を固定する。
  • EEG-DATA判定境界: A(直接検証可能)EEG-DATA単独で主要仮説まで検証可能(外部依存なし)。
  • auto-research-funds応募テーマ: 博士進学を前提にした mind-upload 向け EEG 本人性ドリフト評価
  • 第一応募先 / 予備応募先: 中谷財団 大学院生奨学金 / 樫の芽会 奨学金(runway 補助のみ)
  • 主データ: D02 CSTE: A Dataset for Cross-Sessions and Cross-Tasks EEG Biometrics
  • 補助データ(推奨): D11 A multi-session simultaneous EEG-fMRI dataset with online experience sampling, D23 EEG, PPG, GSR signals for rehabilitation fatigue detection
  • 実データ条件: repeated sessions same-subject labels task variation timestamp/QC device metadata を最低条件に固定いたします。
  • 初期KPI(U0標準): clock offset p95 state-feature collapse rate resync recovery rate
  • 停止条件: 同期補正後も offset/jitter の95パーセンタイルが 20 ms を超え、状態特徴相関維持率 < 0.90 に留まる場合は、同一性主張を「同期依存の部分成立」に降格する。
  • 最低成果物: Time synchronization audit report (offset/jitter distribution)

U0-3 同一性判定の閾値をタスク別にどう設定し、過学習モデルをどう除外するか。

  • 検証命題: D02 CSTE (10.21227/j162-nh61 / subscription-gated support) を主に、D11D23 を補助として、タスク別に同一性閾値を推定し、過学習モデルはセッション外性能で除外する。
  • EEG-DATA判定境界: A(直接検証可能)EEG-DATA単独で主要仮説まで検証可能(外部依存なし)。
  • auto-research-funds応募テーマ: 博士進学を前提にした mind-upload 向け EEG 本人性ドリフト評価
  • 第一応募先 / 予備応募先: 中谷財団 大学院生奨学金 / 樫の芽会 奨学金(runway 補助のみ)
  • 主データ: D02 CSTE: A Dataset for Cross-Sessions and Cross-Tasks EEG Biometrics
  • 補助データ(推奨): D11 A multi-session simultaneous EEG-fMRI dataset with online experience sampling, D23 EEG, PPG, GSR signals for rehabilitation fatigue detection
  • 初期KPI(U0標準): session-out AUC threshold stability overfit exclusion rate
  • 停止条件: 主データ単独で再現しない場合は、前処理・分割・同期ログを固定して再試行し、それでも再現不能なら主張を下方修正する。
  • 最低成果物: Threshold search notebook and selection rationale table

U0-4 分岐/複製ケースでの『同一個体』定義をどの評価軸に帰着させるか。

  • 検証命題: ID 11/509で「同一個体判定」指標を生体同定精度に限定して運用限界を明記する。
  • EEG-DATA判定境界: B(部分検証(外部データ併用で成立))EEG主解析は可能だが、外部データまたは制度情報の併用が必須。
  • auto-research-funds応募テーマ: 共同研究助成(同一性×責任)
  • 第一応募先 / 予備応募先: G2 (GR-2026-014) / G3 (9Lx4dPK6a4k2gOb7)
  • 主データ(ID): 11
  • 補助データ(推奨ID): ID 11, 29, 696
  • 初期KPI(U0標準): セッション外AUC 観測一致率 介入応答一致率
  • 停止条件: EEG側で主要KPIが成立しても、外部依存の根拠が接続できない場合は「部分成立」に止めて応募文面を分割する。
  • 最低成果物: Identity determination two-axis report (observation match/intervention match)
  • Run update (2026-03-29): This section is now locked to D02 CSTE as the primary anchor, with D10/D12 kept only as support datasets for cross-session and longitudinal stress tests.
  • External dependency kept separate in this run: legal personhood and responsibility allocation remain outside the EEG claim and must stay in the proposal as an external dependency.

U1-1 EEG/MEG逆問題での不良設定性を、どの事前分布で制御するか。

  • 検証命題: ID 56の既知刺激位置を使って、事前分布ごとの逆解誤差を比較する。
  • EEG-DATA判定境界: A(直接検証可能)EEG-DATA単独で主要仮説まで検証可能(外部依存なし)。
  • auto-research-funds応募テーマ: 研究助成(逆問題ベンチ)
  • 第一応募先 / 予備応募先: G1 (GR-2026-013) / G4 (Drbm6vBRDJkn0NGJ)
  • 主データ(ID): 56
  • 補助データ(推奨ID): ID 56, 676, 1839
  • 初期KPI(U1標準): 逆解誤差 不確実性被覆率 手法順位一致率
  • 停止条件: 主データ単独で再現しない場合は、前処理・分割・同期ログを固定して再試行し、それでも再現不能なら主張を下方修正する。
  • 最低成果物: Inverse solution comparison table and uncertainty distribution chart

U1-2 頭蓋導電率・電極配置・ノイズ構造の不確実性を、推定不確実性へどう伝播させるか。

  • 検証命題: D08 で頭蓋導電率・電極欠損・ノイズ摂動を与え、interval coverage error increment rank stability の校正曲線を固定する。
  • EEG-DATA判定境界: A(直接検証可能)EEG-DATA単独で主要仮説まで検証可能(外部依存なし)。
  • auto-research-funds応募テーマ: 研究助成(逆問題の不確実性伝播校正)
  • 第一応募先 / 予備応募先: G1 (GR-2026-013) / G4 (Drbm6vBRDJkn0NGJ)
  • 主データ(ID): D08 Simultaneous human intracerebral stimulation and HD-EEG
  • 補助データ(推奨ID): D11 A multi-session simultaneous EEG-fMRI dataset with online experience sampling, D19 CerebellarTMSEEGData
  • 初期KPI(U1標準): interval coverage error increment rank stability
  • 停止条件: 導電率・電極欠損・ノイズ摂動を繰り返したときに interval coverage が崩れるか、rank stability が反転する場合は、不確実性主張を bounded benchmark まで降格する。
  • 最低成果物: Inverse problem reproduction report (error/uncertainty)
  • Run update (2026-03-29): This section is now locked to D08 Simultaneous human intracerebral stimulation and HD-EEG as the primary anchor, with D11/D19 reserved for perturbation checks that do not replace the main ground-truth signal.
  • First-pass KPI fixed in this run: interval coverage is the public pass/fail metric, and the claim must be downgraded when repeated conductivity, missing-electrode, or noise perturbations break coverage stability.

U1-3 同じデータで異なる逆解法(MNE, beamformer, Champagne 等)が乖離した場合の判定規則をどう置くか。

  • 検証命題: ID 56/676でMNE/beamformer/Champagneの乖離率をタスク別に比較する。
  • EEG-DATA判定境界: B(部分検証(外部データ併用で成立))EEG主解析は可能だが、外部データまたは制度情報の併用が必須。
  • auto-research-funds応募テーマ: 研究奨励金(手法比較)
  • 第一応募先 / 予備応募先: G1 (GR-2026-013) / G4 (Drbm6vBRDJkn0NGJ)
  • 主データ(ID): 56
  • 補助データ(推奨ID): ID 56, 676, 1839
  • 初期KPI(U1標準): 逆解誤差 不確実性被覆率 手法順位一致率
  • 停止条件: EEG側で主要KPIが成立しても、外部依存の根拠が接続できない場合は「部分成立」に止めて応募文面を分割する。
  • 最低成果物: Inverse problem reproduction report (error/uncertainty)

U1-4 推定値だけでなく、信頼区間/後方分布を公開基準に含めるか。

  • 検証命題: ID 56を主に、ID 676/1839を補助として 95%信用区間後方分布幅 の公開が再現性を損なわないかを評価する。
  • EEG-DATA判定境界: B(部分検証(外部データ併用で成立))EEG主解析は可能だが、外部データまたは制度情報の併用が必須。
  • auto-research-funds応募テーマ: 研究助成(逆問題の不確実性公開基準)
  • 第一応募先 / 予備応募先: G1 (GR-2026-013) / G4 (Drbm6vBRDJkn0NGJ)
  • 主データ(ID): 56
  • 補助データ(推奨ID): ID 56, 676, 1839
  • 実データ条件: 今回の主アンカーは D08 simultaneous intracerebral stimulation + HD-EEG とし、high-density EEG intervention ground-truth head-model or iEEG disclosure rerunnable pipeline を必須条件にいたします。
  • 初期KPI(U1標準): 95%被覆率 後方分布幅 手法順位一致率
  • 停止条件: 95%被覆率80-100% 帯を継続逸脱するか、区間公開を入れた後に 手法順位一致率 < 0.80 へ低下する場合は、公開基準を「点推定 + 限定区間」に降格する。
  • 最低成果物: Inverse solution comparison table and uncertainty distribution chart

U3-1 ニューロン中心モデルに対して、グリア/体液性調節を追加した際の予測改善量をどう測るか。

  • 検証命題: ID 13/39でEEG単独 vs マルチモーダルの性能差を測り、境界拡張の必要量を推定する。
  • EEG-DATA判定境界: B(部分検証(外部データ併用で成立))EEG主解析は可能だが、外部データまたは制度情報の併用が必須。
  • auto-research-funds応募テーマ: 探索的若手研究助成(境界 proxy)
  • 第一応募先 / 予備応募先: G2 (GR-2026-014) / G6 (46z9VPE4wnkrvEJR)
  • 主データ(ID): 13
  • 補助データ(推奨ID): ID 13, 16, 39
  • 初期KPI(U3標準): 統合条件での性能差 境界拡張時の再現率 外部依存の残差
  • 停止条件: EEG側で主要KPIが成立しても、外部依存の根拠が接続できない場合は「部分成立」に止めて応募文面を分割する。
  • 最低成果物: Boundary expansion evaluation report (single/integrated comparison)

U3-2 connectome完全性と機能予測性能の関係を、種横断でどう比較するか。

  • 検証命題: EEG-DATA単独ではconnectome完全性を直接検証できないため、補助指標としての再現性評価に限定する。
  • EEG-DATA判定境界: B(部分検証(外部データ併用で成立))EEG主解析は可能だが、外部データまたは制度情報の併用が必須。
  • auto-research-funds応募テーマ: 探索的若手研究助成(境界 proxy)
  • 第一応募先 / 予備応募先: G2 (GR-2026-014) / G6 (46z9VPE4wnkrvEJR)
  • 主データ(ID): 13
  • 補助データ(推奨ID): ID 13, 16, 39
  • 初期KPI(U3標準): 統合条件での性能差 境界拡張時の再現率 外部依存の残差
  • 停止条件: EEG側で主要KPIが成立しても、外部依存の根拠が接続できない場合は「部分成立」に止めて応募文面を分割する。
  • 最低成果物: Boundary expansion evaluation report (single/integrated comparison)

U3-3 身体・環境結合を除去したモデルで失われる機能をどう定量化するか。

  • 検証命題: ID 16/39で身体・環境ループを含む条件差を比較し、除去時の性能低下を測る。
  • EEG-DATA判定境界: B(部分検証(外部データ併用で成立))EEG主解析は可能だが、外部データまたは制度情報の併用が必須。
  • auto-research-funds応募テーマ: 探索的若手研究助成(境界 proxy)
  • 第一応募先 / 予備応募先: G2 (GR-2026-014) / G6 (46z9VPE4wnkrvEJR)
  • 主データ(ID): 13
  • 補助データ(推奨ID): ID 13, 16, 39
  • 初期KPI(U3標準): 統合条件での性能差 境界拡張時の再現率 外部依存の残差
  • 停止条件: EEG側で主要KPIが成立しても、外部依存の根拠が接続できない場合は「部分成立」に止めて応募文面を分割する。
  • 最低成果物: Boundary expansion evaluation report (single/integrated comparison)

U3-4 『必要最小構成』の判定を理論的主張ではなくデータでどう固定するか。

  • 検証命題: 必要最小構成の最終判定はEEG単独では不可。まずID 13/39で代理KPIを固定する。
  • EEG-DATA判定境界: C(EEG-DATA単独では不可(制度・他モダリティ必須))制度・法務・他モダリティが主対象で、EEGは補助証拠に限定。
  • auto-research-funds応募テーマ: 探索的若手研究助成(境界 proxy)
  • 第一応募先 / 予備応募先: G2 (GR-2026-014) / G6 (46z9VPE4wnkrvEJR)
  • 主データ(ID): 13
  • 補助データ(推奨ID): ID 13, 16, 39
  • 初期KPI(U3標準): 統合条件での性能差 境界拡張時の再現率 外部依存の残差
  • 停止条件: EEG補助証拠のみで制度/理論主張を確定しない。外部モダリティまたは制度根拠が揃うまで確定主張を保留する。
  • 最低成果物: Boundary expansion evaluation report (single/integrated comparison)

U3-5 glymphatic/meningeal lymphatic 系を含むとき、予測精度と説明可能性はどの程度改善するか。

  • 検証命題: glymphatic/meningeal系はEEG-DATAに直接計測がなく、同時代謝データ併用前提の探索扱い。
  • EEG-DATA判定境界: C(EEG-DATA単独では不可(制度・他モダリティ必須))制度・法務・他モダリティが主対象で、EEGは補助証拠に限定。
  • auto-research-funds応募テーマ: 探索的若手研究助成(境界 proxy)
  • 第一応募先 / 予備応募先: G2 (GR-2026-014) / G6 (46z9VPE4wnkrvEJR)
  • 主データ(ID): 13
  • 補助データ(推奨ID): ID 13, 16, 39
  • 初期KPI(U3標準): 統合条件での性能差 境界拡張時の再現率 外部依存の残差
  • 停止条件: EEG補助証拠のみで制度/理論主張を確定しない。外部モダリティまたは制度根拠が揃うまで確定主張を保留する。
  • 最低成果物: Boundary expansion evaluation report (single/integrated comparison)

U3-6 免疫監視(髄膜免疫・炎症性シグナル)を除外したモデルは、どの時点で長期予測が破綻するか。

  • 検証命題: 免疫監視除外モデルの長期破綻はEEG-DATAだけでは不可。縦断臨床データの外部連携が必要。
  • EEG-DATA判定境界: C(EEG-DATA単独では不可(制度・他モダリティ必須))制度・法務・他モダリティが主対象で、EEGは補助証拠に限定。
  • auto-research-funds応募テーマ: 探索的若手研究助成(境界 proxy)
  • 第一応募先 / 予備応募先: G2 (GR-2026-014) / G6 (46z9VPE4wnkrvEJR)
  • 主データ(ID): 13
  • 補助データ(推奨ID): ID 13, 16, 39
  • 初期KPI(U3標準): 統合条件での性能差 境界拡張時の再現率 外部依存の残差
  • 停止条件: EEG補助証拠のみで制度/理論主張を確定しない。外部モダリティまたは制度根拠が揃うまで確定主張を保留する。
  • 最低成果物: Boundary expansion evaluation report (single/integrated comparison)

U4-1 観測データ由来の相関を因果グラフへ持ち上げる識別条件は何か。

  • 検証命題: ID 19/1839で入力操作に対する応答差を計測し、観測相関との差を分離する。
  • EEG-DATA判定境界: B(部分検証(外部データ併用で成立))EEG主解析は可能だが、外部データまたは制度情報の併用が必須。
  • auto-research-funds応募テーマ: 研究助成(介入同値)
  • 第一応募先 / 予備応募先: G1 (GR-2026-013) / G4 (Drbm6vBRDJkn0NGJ)
  • 主データ(ID): 19
  • 補助データ(推奨ID): ID 19, 56, 1839
  • 初期KPI(U4標準): 介入あり/なし差 反事実誤差 反証閾値超過率
  • 停止条件: EEG側で主要KPIが成立しても、外部依存の根拠が接続できない場合は「部分成立」に止めて応募文面を分割する。
  • 最低成果物: Intervention protocol and rebuttal condition definition

U4-2 介入実験(刺激・抑制・入力撹乱)で検証可能な最小因果主張は何か。

  • 検証命題: D05 を主データに、intervention-versus-control differencesign agreement を最小因果主張として事前登録し、D08/D19 で cross-check する。
  • EEG-DATA判定境界: A(直接検証可能)刺激条件と非刺激条件の比較をEEG-DATA内で完結できるため、最小主張はEEG単独で検証可能。
  • auto-research-funds応募テーマ: 研究助成(最小介入 causal verification)
  • 第一応募先 / 予備応募先: G1 (GR-2026-013) / G4 (Drbm6vBRDJkn0NGJ)
  • 主データ(ID): D05 EEG cortical responses after HiPi
  • 補助データ(推奨ID): D08 Simultaneous human intracerebral stimulation and HD-EEG, D19 CerebellarTMSEEGData
  • 初期KPI(U4標準): 介入あり/なし差 符号一致率 反事実誤差
  • 停止条件: sign agreement が崩れるか、intervention-versus-control difference の95%CIが0をまたぐ場合は、因果主張を exploration only へ降格する。
  • 最低成果物: Intervention protocol and rebuttal condition definition
  • Run update (2026-03-31): This section is now locked to D05 EEG cortical responses after HiPi as the primary anchor, with D08/D19 kept only as support for cross-checking the same intervention family.
  • First-pass KPI fixed in this run: intervention-versus-control difference, sign agreement, and counterfactual error are the public audit numbers, and persistent sign reversal immediately downgrades the claim.

U4-3 active inferenceやDCMの理論予測を、反事実評価にどう接続するか。

  • 検証命題: ID 676を主に、介入あり/なしで反事実誤差(PEHE)と予測尤度差を比較し、理論順位を固定する。
  • EEG-DATA判定境界: B(部分検証(外部データ併用で成立))EEG主解析は可能だが、外部データまたは制度情報の併用が必須。
  • auto-research-funds応募テーマ: 研究奨励金(反事実評価)
  • 第一応募先 / 予備応募先: G1 (GR-2026-013) / G4 (Drbm6vBRDJkn0NGJ)
  • 主データ(ID): 19
  • 補助データ(推奨ID): ID 19, 56, 676
  • 初期KPI(U4標準): 介入あり/なし差 反事実誤差 反証閾値超過率
  • 停止条件: EEG側で主要KPIが成立しても、外部依存の根拠が接続できない場合は「部分成立」に止めて応募文面を分割する。
  • 最低成果物: Intervention protocol and rebuttal condition definition
  • Run update (2026-03-29): This section is now locked to D05 EEG cortical responses after HiPi as the primary anchor, with D08/D19 kept as support only for cross-checking the intervention family.
  • External dependency kept separate in this run: common I/O specifications and the intervention protocol remain outside the EEG claim and must be carried as explicit external deliverables.

U4-4 同値判定の失敗条件(falsification)をどの水準で宣言するか。

  • 検証命題: ID 19/56で ΔAUC > 0.03 または 介入応答の符号反転率 > 10% を失敗宣言閾値として事前登録する。
  • EEG-DATA判定境界: B(部分検証(外部データ併用で成立))EEG主解析は可能だが、外部データまたは制度情報の併用が必須。
  • auto-research-funds応募テーマ: 共同研究助成(反証設計)
  • 第一応募先 / 予備応募先: G1 (GR-2026-013) / G4 (Drbm6vBRDJkn0NGJ)
  • 主データ(ID): 19
  • 補助データ(推奨ID): ID 19, 56, 1839
  • 初期KPI(U4標準): 介入あり/なし差 反事実誤差 反証閾値超過率
  • 停止条件: ΔAUC > 0.03 または 介入応答の符号反転率 > 10% が再現するか、条件別に反証閾値が揺れて固定できない場合は、因果同値主張を降格する。
  • 最低成果物: Causal verification report (minimum intervention claim)
  • First-pass KPI fixed in this run: delta-AUC and sign-reversal rate are the public falsification metrics, and the proposal must declare them before reading the result.
  • Run update (2026-03-29 07:18 JST): This section is now locked to D05 EEG cortical responses after HiPi as the primary anchor, with D08/D19 kept only for intervention-family cross-checks.
  • First-pass KPI fixed in this run: the falsification-first rule is public and explicit, so exceeding either preregistered ΔAUC or sign reversal rate immediately downgrades the claim.

U7-1 BIDS拡張で同期・QC・刺激ログをどこまで必須化するか。

  • 検証命題: ID 676/783で clock offset sampling drift stimulus onset QC fail code の4項目を必須化し、欠落時は不採択にする。
  • EEG-DATA判定境界: A(直接検証可能)EEG-DATA単独で主要仮説まで検証可能(外部依存なし)。
  • auto-research-funds応募テーマ: 研究助成(マルチモーダル標準化)
  • 第一応募先 / 予備応募先: G1 (GR-2026-013) / G3 (9Lx4dPK6a4k2gOb7)
  • 主データ(ID): 676
  • 補助データ(推奨ID): ID 676, 783, 4878
  • 初期KPI(U7標準): 必須項目充足率 欠損ログ率 再解析成功率
  • 停止条件: clock offset sampling drift stimulus onset QC fail code の4必須項目を再構成できないか、再解析成功率 < 0.90 の条件が残る場合は、BIDS拡張必須化を降格する。
  • 最低成果物: Time synchronization audit report (offset/jitter distribution)
  • First-pass KPI fixed in this run: the four-field BIDS extension is the minimum public contract, and missing any one field blocks the claim.
  • Run update (2026-03-29 07:18 JST): This section is now locked to D11 A multi-session simultaneous EEG-fMRI dataset with online experience sampling, with D15/D23 used only to test whether the same audit rule survives a second modality family.
  • First-pass KPI fixed in this run: required-field fill rate and reanalysis success rate are the public pass/fail metrics, and missing any of clock offset, sampling drift, stimulus onset, or QC fail code blocks the claim.

U7-2 LSL等の時刻同期誤差を検証可能な指標に落とせるか。

  • 検証命題: ID 676を主に、ID 6/4878を補助として offset/jitterERP peak shift再解析一致率 に写像し、release block 可能な同期誤差指標へ落とし込む。
  • EEG-DATA判定境界: A(直接検証可能)EEG-DATA単独で主要仮説まで検証可能(外部依存なし)。
  • auto-research-funds応募テーマ: 研究助成(同期誤差メトリクス標準)
  • 第一応募先 / 予備応募先: G1 (GR-2026-013) / G3 (9Lx4dPK6a4k2gOb7)
  • 主データ(ID): 676
  • 補助データ(推奨ID): ID 676, 6, 4878
  • 初期KPI(U7標準): offset中央値 jitter p95 peak-latency drift
  • 停止条件: 補正後の event-alignment 誤差95パーセンタイルが 15 ms を超えるか、再解析一致率 < 0.90 の条件が残る場合は、同期監査を「警告のみ」に留めて release block へ昇格しない。
  • 最低成果物: Time synchronization audit report (offset/jitter distribution)

U7-3 アーチファクト除去(ASR, ZapLine等)の設定差が結果へ与える影響をどう監査するか。

  • 検証命題: ID 676(試験ストレスEEG)とID 4878(学習法前後resting-state)で ASR cutoff ZapLine line-frequency ICA除去閾値 を振り、前処理差分が推論結果へ与える移動量を定量化する。
  • EEG-DATA判定境界: A(直接検証可能)前処理条件差分の評価はEEG-DATA内で完結でき、外部制度情報を要しない。
  • auto-research-funds応募テーマ: 研究奨励金(前処理差分)
  • 第一応募先 / 予備応募先: G1 (GR-2026-013) / G3 (9Lx4dPK6a4k2gOb7)
  • 主データ(ID): 676
  • 補助データ(推奨ID): ID 676, 783, 4878
  • 初期KPI(U7標準): 再現率 前処理差分感度 同期監査異常率
  • 停止条件: 基準パイプライン比で 再現率 -5pt を超える設定が3条件以上発生した場合、当該前処理はデフォルト候補から除外する。
  • 最低成果物: Reproducibility audit report (synchronization/QC/preprocessing difference)
  • Run update (2026-03-29): This section is now locked to D11 A multi-session simultaneous EEG-fMRI dataset with online experience sampling, with D12/D24 used only to test whether the preprocessing rule survives a second protocol family.
  • First-pass KPI fixed in this run: preprocessing-drift sensitivity is the main audit number, and any configuration that breaks ranking stability across repeated runs is removed from the default pipeline.

U7-4 モダリティ間アライメント失敗時の再計測/除外基準をどう固定するか。

  • 検証命題: ID 676/783でアライメント失敗時の再計測/除外条件を判定木化する。
  • EEG-DATA判定境界: A(直接検証可能)EEG-DATA単独で主要仮説まで検証可能(外部依存なし)。
  • auto-research-funds応募テーマ: 共同研究助成(再計測基準)
  • 第一応募先 / 予備応募先: G1 (GR-2026-013) / G3 (9Lx4dPK6a4k2gOb7)
  • 主データ(ID): 676
  • 補助データ(推奨ID): ID 676, 783, 4878
  • 初期KPI(U7標準): 再現率 前処理差分感度 同期監査異常率
  • 停止条件: 主データ単独で再現しない場合は、前処理・分割・同期ログを固定して再試行し、それでも再現不能なら主張を下方修正する。
  • 最低成果物: Reproducibility audit report (synchronization/QC/preprocessing difference)

U7-5 前処理差分をCIで自動比較する場合、どの再現率低下をリリースブロック閾値にするか。

  • 検証命題: ID 676で前処理CIを回し、基準パイプライン比で 再現率 -5pt 以上低下を release block 閾値に固定する。
  • EEG-DATA判定境界: A(直接検証可能)EEG-DATA単独で主要仮説まで検証可能(外部依存なし)。
  • auto-research-funds応募テーマ: 研究助成(CI再現性)
  • 第一応募先 / 予備応募先: G1 (GR-2026-013) / G3 (9Lx4dPK6a4k2gOb7)
  • 主データ(ID): 676
  • 補助データ(推奨ID): ID 676, 783, 4878
  • 初期KPI(U7標準): 再現率 前処理差分感度 同期監査異常率
  • 停止条件: 主データ単独で再現しない場合は、前処理・分割・同期ログを固定して再試行し、それでも再現不能なら主張を下方修正する。
  • 最低成果物: Threshold search notebook and selection rationale table

U7-6 モダリティ欠損(EEG欠損・fMRI欠損)条件でも同等結論を保てる最小観測セットは何か。

  • 検証命題: ID 676/783でEEG欠損・fMRI/fNIRS欠損をシミュレートし最小観測セットを求める。
  • EEG-DATA判定境界: A(直接検証可能)EEG-DATA単独で主要仮説まで検証可能(外部依存なし)。
  • auto-research-funds応募テーマ: 研究助成(欠損ロバスト性)
  • 第一応募先 / 予備応募先: G1 (GR-2026-013) / G3 (9Lx4dPK6a4k2gOb7)
  • 主データ(ID): 676
  • 補助データ(推奨ID): ID 676, 783, 4878
  • 実データ条件: 今回の主アンカーは D11 multi-session simultaneous EEG-fMRI with online experience sampling とし、same-trial synchronization BIDS-sidecar provenance missing-modality policy subject-session count を最低限開示いたします。
  • 初期KPI(U7標準): 再現率 前処理差分感度 同期監査異常率
  • 停止条件: 主データ単独で再現しない場合は、前処理・分割・同期ログを固定して再試行し、それでも再現不能なら主張を下方修正する。加えて missing-modality policycomplete-case bias を分離できない場合は、「complete-case 限定の部分成立」までに降格いたします。
  • 最低成果物: Reproducibility audit report (synchronization/QC/preprocessing difference)

U8-1 閉ループBCIの遅延許容域を制御理論的にどう同定するか。

  • 検証命題: D20 で遅延・位相ずれに対する stability rate stop-intervention rate recovery time を測り、bounded delay-tolerance band を定義する。
  • EEG-DATA判定境界: B(部分検証(外部データ併用で成立))EEG主解析は可能だが、外部データまたは制度情報の併用が必須。
  • auto-research-funds応募テーマ: 研究助成(閉ループ遅延許容域と safety-stop)
  • 第一応募先 / 予備応募先: G2 (GR-2026-014) / G5 (871pw3rLjNPKgqA0)
  • 主データ(ID): D20 Closed-loop auditory stimulation targeting REM oscillations
  • 補助データ(推奨ID): D21 NeuroSimo: closed-loop EEG/EMG-guided TMS, D12 Longitudinal MI-BCI training with transcutaneous spinal stimulation
  • 初期KPI(U8標準): stability rate stop-intervention rate recovery time
  • 停止条件: stability rate が事前宣言した floor を下回るか、recovery time が operating band を継続逸脱する場合は、主張を local-controller result まで降格する。
  • 最低成果物: Closed loop safety KPI dashboard
  • Run update (2026-03-31): This section is now locked to D20 Closed-loop auditory stimulation targeting REM oscillations as the primary anchor, with D21/D12 kept only as support for trigger-stop-recovery cross-checks.
  • First-pass KPI fixed in this run: stability rate, stop-intervention rate, and recovery time are the public safety numbers, and drifting beyond the declared operating band immediately downgrades the claim.

U8-2 オンライン較正と概念ドリフト対策をどう組み込むか。

  • 検証命題: ID 2412(real-time BCI)を主に、ID 696で被験者内変動を補完し、再較正間隔(セッションごと/日次/週次)ごとの安定性と運用コストの境界を推定する。
  • EEG-DATA判定境界: B(部分検証(外部データ併用で成立))EEG側の性能ドリフトは評価できるが、運用SLA/臨床安全域の最終決定は外部要件が必要。
  • auto-research-funds応募テーマ: 研究奨励金(再較正)
  • 第一応募先 / 予備応募先: G2 (GR-2026-014) / G5 (871pw3rLjNPKgqA0)
  • 主データ(ID): 2412
  • 補助データ(推奨ID): ID 6, 696, 1972, 2412
  • 初期KPI(U8標準): 安定率 停止介入率 回復時間
  • 停止条件: 週次較正で 停止介入率 が日次較正比で20%以上悪化し、かつ 回復時間 が閾値を超える場合は週次運用案を棄却する。
  • 最低成果物: Closed loop safety KPI dashboard

U8-3 個体差と日内変動をまたぐ安定性をどの指標で評価するか。

  • 検証命題: ID 696で個体内・個体間分散を分けて長期安定性KPIを設計する。
  • EEG-DATA判定境界: B(部分検証(外部データ併用で成立))EEG主解析は可能だが、外部データまたは制度情報の併用が必須。
  • auto-research-funds応募テーマ: 博士課程研究助成(長期ドリフト)
  • 第一応募先 / 予備応募先: G2 (GR-2026-014) / G5 (871pw3rLjNPKgqA0)
  • 主データ(ID): 6
  • 補助データ(推奨ID): ID 6, 1972, 2412, 696
  • 初期KPI(U8標準): 安定率 停止介入率 回復時間
  • 停止条件: EEG側で主要KPIが成立しても、外部依存の根拠が接続できない場合は「部分成立」に止めて応募文面を分割する。
  • 最低成果物: Operational stability report (safety shutdown/recovery time)

U8-4 異常検知とフェイルセーフを評価契約へどう組み込むか。

  • 検証命題: ID 2412を主に、ID 1972/696を補助として異常検知遅延と fail-safe 発火/回復を同一契約で固定する。
  • EEG-DATA判定境界: B(部分検証(外部データ併用で成立))EEG主解析は可能だが、外部データまたは制度情報の併用が必須。
  • auto-research-funds応募テーマ: 共同研究助成(閉ループ異常検知とフェイルセーフ契約)
  • 第一応募先 / 予備応募先: G2 (GR-2026-014) / G5 (871pw3rLjNPKgqA0)
  • 主データ(ID): 2412
  • 補助データ(推奨ID): ID 2412, 1972, 696
  • 初期KPI(U8標準): 異常検知遅延 フェイルセーフ発火率 回復時間
  • 停止条件: 異常検知遅延 の95パーセンタイルが 250 ms を超えるか、回復時間 の中央値が 5 秒 を超える場合は、評価契約を「研究内安全手順」に留めて運用契約へ昇格しない。
  • 最低成果物: Operational stability report (safety shutdown/recovery time)
  • Run update (2026-03-29): This section is now locked to D20 Closed-loop auditory stimulation targeting REM oscillations as the primary anchor, with D01/D21 used only for device-delay and closed-loop controller cross-checks.
  • External dependency kept separate in this run: site shutdown ownership and operational safety procedure remain outside the EEG claim and must stay as external approval work.
  • Run update (2026-03-29 07:18 JST): This section remains in the current public six-RQ batch, and abnormal-detection latency p95 stays paired with fail-safe trigger rate and recovery time as the public pass/fail bundle.

U8-5 ヒューマンオーバーライドを導入したとき、誤作動率と回復時間をどうKPI化するか。

  • 検証命題: ID 2412で手動オーバーライド時の誤作動率と回復時間をKPI化する。
  • EEG-DATA判定境界: B(部分検証(外部データ併用で成立))EEG主解析は可能だが、外部データまたは制度情報の併用が必須。
  • auto-research-funds応募テーマ: 研究奨励金(オーバーライド)
  • 第一応募先 / 予備応募先: G2 (GR-2026-014) / G5 (871pw3rLjNPKgqA0)
  • 主データ(ID): 6
  • 補助データ(推奨ID): ID 6, 1972, 2412, 696
  • 初期KPI(U8標準): 安定率 停止介入率 回復時間
  • 停止条件: EEG側で主要KPIが成立しても、外部依存の根拠が接続できない場合は「部分成立」に止めて応募文面を分割する。
  • 最低成果物: Closed loop safety KPI dashboard

U8-6 週〜月スケール運用での再学習頻度を、性能劣化と安全余裕のトレードオフでどう最適化するか。

  • 検証命題: ID 696/2412で再学習間隔を変え、性能劣化と安全余裕のトレードオフを推定する。
  • EEG-DATA判定境界: B(部分検証(外部データ併用で成立))EEG主解析は可能だが、外部データまたは制度情報の併用が必須。
  • auto-research-funds応募テーマ: 研究助成(再学習最適化)
  • 第一応募先 / 予備応募先: G2 (GR-2026-014) / G5 (871pw3rLjNPKgqA0)
  • 主データ(ID): 6
  • 補助データ(推奨ID): ID 6, 1972, 2412, 696
  • 実データ条件: 今回の主アンカーは D12 longitudinal MI-BCI + spinal stimulation とし、online labels latency-jitter log fixed-vs-recalibrated decoder disclosure failure registry を必須条件にいたします。
  • 初期KPI(U8標準): 安定率 停止介入率 回復時間
  • 停止条件: EEG側で主要KPIが成立しても、外部依存の根拠が接続できない場合は「部分成立」に止めて応募文面を分割する。さらに credit assignment probe がなく、user learning decoder update interface shaping を分離できない場合は、「co-adaptive partial success」に降格いたします。
  • 最低成果物: Closed loop safety KPI dashboard

U10-1 Landauer下限を神経計算でどう適用/解釈するか。

  • 検証命題: Landauer下限の直接検証はEEG-DATA単独では不可。ID 735を理論整合の事前検証に使う。
  • EEG-DATA判定境界: C(EEG-DATA単独では不可(制度・他モダリティ必須))制度・法務・他モダリティが主対象で、EEGは補助証拠に限定。
  • auto-research-funds応募テーマ: 理論研究奨励(情報熱力学)
  • 第一応募先 / 予備応募先: G2 (GR-2026-014) / G6 (46z9VPE4wnkrvEJR)
  • 主データ(ID): 735
  • 補助データ(推奨ID): ID 735, 783, 1011
  • 初期KPI(U10標準): 理論整合チェック完了率 proxy妥当性 外部計測依存率
  • 停止条件: EEG補助証拠のみで制度/理論主張を確定しない。外部モダリティまたは制度根拠が揃うまで確定主張を保留する。
  • 最低成果物: Theory consistency memo and proxy correlation analysis

U10-2 非平衡熱力学指標と神経情報処理効率の対応をどう定義するか。

  • 検証命題: 非平衡熱力学指標の主評価は不可。ID 783/1011で代謝proxyとの弱い相関を探索する。
  • EEG-DATA判定境界: C(EEG-DATA単独では不可(制度・他モダリティ必須))制度・法務・他モダリティが主対象で、EEGは補助証拠に限定。
  • auto-research-funds応募テーマ: 理論研究奨励(情報熱力学)
  • 第一応募先 / 予備応募先: G2 (GR-2026-014) / G6 (46z9VPE4wnkrvEJR)
  • 主データ(ID): 735
  • 補助データ(推奨ID): ID 735, 783, 1011
  • 初期KPI(U10標準): 理論整合チェック完了率 proxy妥当性 外部計測依存率
  • 停止条件: EEG補助証拠のみで制度/理論主張を確定しない。外部モダリティまたは制度根拠が揃うまで確定主張を保留する。
  • 最低成果物: Theory consistency memo and proxy correlation analysis

U10-3 理論式を実データ(神経活動・代謝)へ落とし込む観測設計をどう作るか。

  • 検証命題: 神経活動+代謝同時計測はID 783で部分検証可能だが、熱散逸実測は外部計測が必要。
  • EEG-DATA判定境界: C(EEG-DATA単独では不可(制度・他モダリティ必須))制度・法務・他モダリティが主対象で、EEGは補助証拠に限定。
  • auto-research-funds応募テーマ: 理論研究奨励(情報熱力学)
  • 第一応募先 / 予備応募先: G2 (GR-2026-014) / G6 (46z9VPE4wnkrvEJR)
  • 主データ(ID): 735
  • 補助データ(推奨ID): ID 735, 783, 1011
  • 初期KPI(U10標準): 理論整合チェック完了率 proxy妥当性 外部計測依存率
  • 停止条件: EEG補助証拠のみで制度/理論主張を確定しない。外部モダリティまたは制度根拠が揃うまで確定主張を保留する。
  • 最低成果物: Theory consistency report (thermodynamic proxy)

U10-4 WBE計算コスト評価に熱力学制約をどう統合するか。

  • 検証命題: WBE計算コストへ熱制約を統合する実証はEEG-DATA外。ID 735でモデル妥当性確認まで。
  • EEG-DATA判定境界: C(EEG-DATA単独では不可(制度・他モダリティ必須))制度・法務・他モダリティが主対象で、EEGは補助証拠に限定。
  • auto-research-funds応募テーマ: 理論研究奨励(情報熱力学)
  • 第一応募先 / 予備応募先: G2 (GR-2026-014) / G6 (46z9VPE4wnkrvEJR)
  • 主データ(ID): 735
  • 補助データ(推奨ID): ID 735, 783, 1011
  • 初期KPI(U10標準): 理論整合チェック完了率 proxy妥当性 外部計測依存率
  • 停止条件: EEG補助証拠のみで制度/理論主張を確定しない。外部モダリティまたは制度根拠が揃うまで確定主張を保留する。
  • 最低成果物: Theory consistency memo and proxy correlation analysis

U11-1 理論間で比較可能な入出力仕様をどう定義するか。

  • 検証命題: ID 842/859で PCI近似・LZ複雑性・摂動応答の3指標を同一I/O仕様に揃え、順位保存率80%以上を採択条件にする。
  • EEG-DATA判定境界: B(部分検証(外部データ併用で成立))EEG主解析は可能だが、外部データまたは制度情報の併用が必須。
  • auto-research-funds応募テーマ: 研究助成(意識指標比較)
  • 第一応募先 / 予備応募先: G2 (GR-2026-014) / G4 (Drbm6vBRDJkn0NGJ)
  • 主データ(ID): 735
  • 補助データ(推奨ID): ID 735, 842, 859
  • 初期KPI(U11標準): 理論間順位保存率 境界症例符号一致率 計算予算内完走率
  • 停止条件: 理論間順位保存率 < 0.80 となるか、境界状態で 符号一致率 が不安定な場合は、shared I/O 仕様を「暫定比較仕様」に降格する。
  • 最低成果物: Awareness index comparison report (with failure condition)
  • First-pass KPI fixed in this run: rank-preservation rate under shared I/O is the public comparison metric, and theory choice stays outside the EEG claim.
  • Run update (2026-03-29 07:18 JST): This section is now locked to D14 PK-NMM EEG simulation during propofol anesthesia, with D16/D17 reserved only for sleep and DoC cross-state checks.
  • First-pass KPI fixed in this run: the public pass/fail metric is rank-preservation rate across shared I/O, and the claim must be downgraded when calibration cannot preserve ordering across anesthesia, sleep, and DoC.

U11-2 PCIやIIT近似計算の計算量制約をどう扱うか。

  • 検証命題: IIT厳密計算は除外し、ID 735/842で PCI近似 LZ wSMI を計算予算(1症例あたりGPU 2時間以内)付きで比較する。
  • EEG-DATA判定境界: C(EEG-DATA単独では不可(制度・他モダリティ必須))制度・法務・他モダリティが主対象で、EEGは補助証拠に限定。
  • auto-research-funds応募テーマ: 理論研究奨励(IIT近似)
  • 第一応募先 / 予備応募先: G2 (GR-2026-014) / G4 (Drbm6vBRDJkn0NGJ)
  • 主データ(ID): 735
  • 補助データ(推奨ID): ID 735, 842, 859
  • 初期KPI(U11標準): 理論間順位保存率 境界症例符号一致率 計算予算内完走率
  • 停止条件: EEG補助証拠のみで制度/理論主張を確定しない。外部モダリティまたは制度根拠が揃うまで確定主張を保留する。
  • 最低成果物: Theoretical comparison I/O specifications and calculation budget

U11-3 理論予測の対立点を単一実験計画へどう落とすか。

  • 検証命題: ID 735/859で理論対立点を同一入力条件へ寄せる試験計画を作る。
  • EEG-DATA判定境界: B(部分検証(外部データ併用で成立))EEG主解析は可能だが、外部データまたは制度情報の併用が必須。
  • auto-research-funds応募テーマ: 共同研究助成(理論対立検証)
  • 第一応募先 / 予備応募先: G2 (GR-2026-014) / G4 (Drbm6vBRDJkn0NGJ)
  • 主データ(ID): 735
  • 補助データ(推奨ID): ID 735, 842, 859
  • 実データ条件: 今回の主アンカーは D17 DoC EEG biomarker pilot とし、same-cohort calibration continuous recordings hypnogram or sedation-stage labels pharmacology or perturbation log を必須条件にいたします。
  • 初期KPI(U11標準): 理論間順位保存率 境界症例符号一致率 計算予算内完走率
  • 停止条件: EEG側で主要KPIが成立しても、外部依存の根拠が接続できない場合は「部分成立」に止めて応募文面を分割する。加えて same-cohort calibration が取れない、または construct / perturbation / calibration / incremental の4ゲート分解ができない場合は、「proxy comparison memo」に降格いたします。
  • 最低成果物: Awareness index comparison report (with failure condition)

U11-4 意識指標を臨床/研究で運用する際の失敗条件をどう明示するか。

  • 検証命題: ID 859で偽陽性/偽陰性を含む臨床失敗条件の閾値を先に定義する。
  • EEG-DATA判定境界: B(部分検証(外部データ併用で成立))EEG主解析は可能だが、外部データまたは制度情報の併用が必須。
  • auto-research-funds応募テーマ: 臨床系若手助成(失敗条件定義)
  • 第一応募先 / 予備応募先: G2 (GR-2026-014) / G4 (Drbm6vBRDJkn0NGJ)
  • 主データ(ID): 735
  • 補助データ(推奨ID): ID 735, 842, 859
  • 初期KPI(U11標準): 理論間順位保存率 境界症例符号一致率 計算予算内完走率
  • 停止条件: EEG側で主要KPIが成立しても、外部依存の根拠が接続できない場合は「部分成立」に止めて応募文面を分割する。
  • 最低成果物: Theoretical comparison I/O specifications and calculation budget

U12-1 分岐後主体の識別子を何に基づいて付与するか。

  • 検証命題: 分岐後識別子の法的定義はEEG-DATA単独では不可。ID 11を技術IDの安定性評価に限定利用。
  • EEG-DATA判定境界: C(EEG-DATA単独では不可(制度・他モダリティ必須))制度・法務・他モダリティが主対象で、EEGは補助証拠に限定。
  • auto-research-funds応募テーマ: 制度・倫理系研究助成(技術監査接続)
  • 第一応募先 / 予備応募先: G2 (GR-2026-014) / G6 (46z9VPE4wnkrvEJR)
  • 主データ(ID): 11
  • 補助データ(推奨ID): ID 11, 509, 696
  • 初期KPI(U12標準): 分岐時整合率 監査証跡欠損率 同意伝播遅延
  • 停止条件: EEG補助証拠のみで制度/理論主張を確定しない。外部モダリティまたは制度根拠が揃うまで確定主張を保留する。
  • 最低成果物: Branch-based audit log design document

U12-2 責任・権利・同意の継承ルールをどの時点で分岐させるか。

  • 検証命題: 責任・権利・同意の継承規則は制度設計課題であり、EEG-DATAは検証補助のみ。
  • EEG-DATA判定境界: C(EEG-DATA単独では不可(制度・他モダリティ必須))制度・法務・他モダリティが主対象で、EEGは補助証拠に限定。
  • auto-research-funds応募テーマ: 制度・倫理系研究助成(技術監査接続)
  • 第一応募先 / 予備応募先: G2 (GR-2026-014) / G6 (46z9VPE4wnkrvEJR)
  • 主データ(ID): 11
  • 補助データ(推奨ID): ID 11, 509, 696
  • 初期KPI(U12標準): 分岐時整合率 監査証跡欠損率 同意伝播遅延
  • 停止条件: EEG補助証拠のみで制度/理論主張を確定しない。外部モダリティまたは制度根拠が揃うまで確定主張を保留する。
  • 最低成果物: System audit requirements table (technical log compatible)

U12-3 心理的連続性基準と法的個体基準の不一致をどう扱うか。

  • 検証命題: 心理的連続性と法的個体の不一致はデータだけでは裁定できず、規範設計が主対象。
  • EEG-DATA判定境界: C(EEG-DATA単独では不可(制度・他モダリティ必須))制度・法務・他モダリティが主対象で、EEGは補助証拠に限定。
  • auto-research-funds応募テーマ: 制度・倫理系研究助成(技術監査接続)
  • 第一応募先 / 予備応募先: G2 (GR-2026-014) / G6 (46z9VPE4wnkrvEJR)
  • 主データ(ID): 11
  • 補助データ(推奨ID): ID 11, 509, 696
  • 初期KPI(U12標準): 分岐時整合率 監査証跡欠損率 同意伝播遅延
  • 停止条件: EEG補助証拠のみで制度/理論主張を確定しない。外部モダリティまたは制度根拠が揃うまで確定主張を保留する。
  • 最低成果物: System audit requirements table (technical log compatible)

U12-4 技術評価(性能)と人格評価(帰属)をどう接続するか。

  • 検証命題: 性能KPIと人格帰属KPIの接続は法務要件が主で、EEGは監査ログ項目の補助。
  • EEG-DATA判定境界: C(EEG-DATA単独では不可(制度・他モダリティ必須))制度・法務・他モダリティが主対象で、EEGは補助証拠に限定。
  • auto-research-funds応募テーマ: 制度・倫理系研究助成(技術監査接続)
  • 第一応募先 / 予備応募先: G2 (GR-2026-014) / G6 (46z9VPE4wnkrvEJR)
  • 主データ(ID): 11
  • 補助データ(推奨ID): ID 11, 509, 696
  • 初期KPI(U12標準): 分岐時整合率 監査証跡欠損率 同意伝播遅延
  • 停止条件: EEG補助証拠のみで制度/理論主張を確定しない。外部モダリティまたは制度根拠が揃うまで確定主張を保留する。
  • 最低成果物: Branch-based audit log design document

U12-5 分岐主体間で記憶編集・再同期が起きた場合、法的主体IDを再編する基準は何か。

  • 検証命題: 記憶編集・再同期時のID再編ルールはEEGデータでは直接検証不可。
  • EEG-DATA判定境界: C(EEG-DATA単独では不可(制度・他モダリティ必須))制度・法務・他モダリティが主対象で、EEGは補助証拠に限定。
  • auto-research-funds応募テーマ: 制度・倫理系研究助成(技術監査接続)
  • 第一応募先 / 予備応募先: G2 (GR-2026-014) / G6 (46z9VPE4wnkrvEJR)
  • 主データ(ID): 11
  • 補助データ(推奨ID): ID 11, 509, 696
  • 初期KPI(U12標準): 分岐時整合率 監査証跡欠損率 同意伝播遅延
  • 停止条件: EEG補助証拠のみで制度/理論主張を確定しない。外部モダリティまたは制度根拠が揃うまで確定主張を保留する。
  • 最低成果物: Branch-based audit log design document

U12-6 同意撤回が発生したとき、複数分岐主体への権限剥奪を技術的にどう実装・監査するか。

  • 検証命題: 同意撤回の権限剥奪監査はシステム設計課題で、EEG-DATAは模擬ログ検証まで。
  • EEG-DATA判定境界: C(EEG-DATA単独では不可(制度・他モダリティ必須))制度・法務・他モダリティが主対象で、EEGは補助証拠に限定。
  • auto-research-funds応募テーマ: 制度・倫理系研究助成(技術監査接続)
  • 第一応募先 / 予備応募先: G2 (GR-2026-014) / G6 (46z9VPE4wnkrvEJR)
  • 主データ(ID): 11
  • 補助データ(推奨ID): ID 11, 509, 696
  • 初期KPI(U12標準): 分岐時整合率 監査証跡欠損率 同意伝播遅延
  • 停止条件: EEG補助証拠のみで制度/理論主張を確定しない。外部モダリティまたは制度根拠が揃うまで確定主張を保留する。
  • 最低成果物: System audit requirements table (technical log compatible)

U13-1 brain-to-text成功を『意味復元』と『因果再現』へ分解できるか。

  • 検証命題: ID 509でovert/imaginedを比較し、意味一致と因果一致の乖離を測る。
  • EEG-DATA判定境界: B(部分検証(外部データ併用で成立))EEG主解析は可能だが、外部データまたは制度情報の併用が必須。
  • auto-research-funds応募テーマ: 研究助成(模倣分離)
  • 第一応募先 / 予備応募先: G1 (GR-2026-013) / G4 (Drbm6vBRDJkn0NGJ)
  • 主データ(ID): 509
  • 補助データ(推奨ID): ID 509, 13, 65
  • 初期KPI(U13標準): 意味一致率 因果一致率 リーク検出率
  • 停止条件: EEG側で主要KPIが成立しても、外部依存の根拠が接続できない場合は「部分成立」に止めて応募文面を分割する。
  • 最低成果物: Intervention protocol and rebuttal condition definition
  • Run update (2026-03-29): This section is now locked to D10 3M-CPSEED as the primary anchor, with D03/D09 reserved for multimodal and control-condition support only.
  • First-pass KPI fixed in this run: the public report must show semantic-match rate, causal-match rate, and their gap on a leak-free split, and an imitation-only pass is rejected by design.

U13-2 LLMの幻覚・整合性検査を神経デコード評価へどう接続するか。

  • 検証命題: D10 3M-CPSEED (10.18112/openneuro.ds006465.v2.0.0 / immediate) を主に、EEG data recorded during spoken and imagined speech interaction with a simulated robot (10.5281/zenodo.14645652 / immediate)ChineseEEG-2 (10.57760/sciencedb.CHNNeuro.00001 / immediate) を公開補助、D03 を gated multimodal comparator として hallucination labelneural inconsistency を同時監査し、意味一致だけの合格を禁止する。
  • EEG-DATA判定境界: B(部分検証(外部データ併用で成立))EEG主解析は可能だが、外部データまたは制度情報の併用が必須。
  • auto-research-funds応募テーマ: mind-upload 仮説の国際共同検証に向けた EEG brain-to-text 模倣分離ベンチ立ち上げ
  • 第一応募先 / 予備応募先: くら基金 若手研究者奨励助成 / 中谷財団 大学院生奨学金
  • 主データ: D10 3M-CPSEED: An EEG-based Dataset for Chinese Pinyin Production in Overt, Silent-intended, and Imagined Speech
  • 補助データ(推奨): EEG data recorded during spoken and imagined speech interaction with a simulated robot, ChineseEEG-2, D03 Multimodal Fusion EEG + Eye-Tracking
  • 実データ条件: 今回の主アンカーは D10 3M-CPSEED とし、stimulus/audio/transcript alignment overt-vs-covert control leak-free split BIDS or equivalent provenance を必須条件にいたします。
  • 初期KPI(U13標準): semantic alignment brain-minus-prior gap misalignment rate
  • 停止条件: 内部状態差検出率 < 0.60 または 幻覚率neural inconsistency の相関区間が 0 を跨ぐ場合は、「意味一致のみ」の主張へ下方修正する。
  • 最低成果物: Mimic separation evaluation script and control condition table

U13-3 同一出力でも内部機構が異なるケースをどう検出するか。

  • 検証命題: ID 65/509で同一出力・異内部状態ケースを抽出する判定器を設計する。
  • EEG-DATA判定境界: B(部分検証(外部データ併用で成立))EEG主解析は可能だが、外部データまたは制度情報の併用が必須。
  • auto-research-funds応募テーマ: 研究奨励金(shortcut検出)
  • 第一応募先 / 予備応募先: G1 (GR-2026-013) / G4 (Drbm6vBRDJkn0NGJ)
  • 主データ(ID): 509
  • 補助データ(推奨ID): ID 509, 13, 65
  • 初期KPI(U13標準): 意味一致率 因果一致率 リーク検出率
  • 停止条件: EEG側で主要KPIが成立しても、外部依存の根拠が接続できない場合は「部分成立」に止めて応募文面を分割する。
  • 最低成果物: Imitation separation evaluation report (meaning/mechanism)

U13-4 模倣性能の上限を因果評価でどこまで抑制できるか。

  • 検証命題: ID 509で模倣スコア単独合格を禁止し、因果整合指標との二重閾値を運用する。
  • EEG-DATA判定境界: B(部分検証(外部データ併用で成立))EEG主解析は可能だが、外部データまたは制度情報の併用が必須。
  • auto-research-funds応募テーマ: 研究助成(因果整合KPI)
  • 第一応募先 / 予備応募先: G1 (GR-2026-013) / G4 (Drbm6vBRDJkn0NGJ)
  • 主データ(ID): 509
  • 補助データ(推奨ID): ID 509, 13, 65
  • 初期KPI(U13標準): 意味一致率 因果一致率 リーク検出率
  • 停止条件: EEG側で主要KPIが成立しても、外部依存の根拠が接続できない場合は「部分成立」に止めて応募文面を分割する。
  • 最低成果物: Intervention protocol and rebuttal condition definition

U13-5 視覚知覚と視覚想起で同一デコーダを使ったとき、意味復元精度の劣化パターンはどこで分岐するか。

  • 検証命題: D03 Multimodal Fusion EEG + Eye-Tracking (10.21227/6106-6120 / subscription-gated support) を gated planning anchor とし、D10ChineseEEG-2ZuCo 2.0SparrKULeeEEG data recorded during spoken and imagined speech interaction with a simulated robot を public-open support として、知覚条件で学習した同一デコーダを想起条件へ転送し、分岐点を時間窓別に特定する。ただし公開 metadata 上は recall-specific label が見えていないため、想起妥当性は外部 layer を必須とする。
  • EEG-DATA判定境界: B(部分検証(外部データ併用で成立))デコーダ劣化の定量化は可能だが、想起内容の主観ラベル妥当性には外部行動/報告系が必要。
  • auto-research-funds応募テーマ: mind-upload 仮説の国際共同検証に向けた EEG brain-to-text 模倣分離ベンチ立ち上げ
  • 第一応募先 / 予備応募先: くら基金 若手研究者奨励助成 / 中谷財団 大学院生奨学金
  • 主データ: D03 Data of Multimodal Fusion System for Cognitive Load Assessment in Psychological Experiments: An EEG and Eye-Tracking Study
  • 補助データ(推奨): D10 3M-CPSEED, ChineseEEG-2, ZuCo 2.0, SparrKULee, EEG data recorded during spoken and imagined speech interaction with a simulated robot
  • 初期KPI(U13標準): perception-to-recall degradation point transfer gap control-condition stability
  • 停止条件: 想起条件で 意味一致率 が知覚条件比で一貫してチャンス水準近傍となる場合、または recall-validity layer が接続できない場合、同一デコーダ主張を撤回し分離学習または planning note へ切り替える。
  • 最低成果物: Mimic separation evaluation script and control condition table

U13-6 プロンプト誘導・データリーク・shortcut学習を分離検出する対照実験をどう設計するか。

  • 検証命題: ID 65/13でリーク対照実験を設計し、shortcut検出率を評価する。
  • EEG-DATA判定境界: B(部分検証(外部データ併用で成立))EEG主解析は可能だが、外部データまたは制度情報の併用が必須。
  • auto-research-funds応募テーマ: 研究助成(リーク分離実験)
  • 第一応募先 / 予備応募先: G1 (GR-2026-013) / G4 (Drbm6vBRDJkn0NGJ)
  • 主データ(ID): 509
  • 補助データ(推奨ID): ID 509, 13, 65
  • 初期KPI(U13標準): 意味一致率 因果一致率 リーク検出率
  • 停止条件: EEG側で主要KPIが成立しても、外部依存の根拠が接続できない場合は「部分成立」に止めて応募文面を分割する。
  • 最低成果物: Leak audit results and repartition trail

U14-1 データ/コード/評価環境の固定粒度をどこまで要求するか。

  • 検証命題: ID 6で測定遅延再現試験を自動化し、最小追試パックの基礎にする。
  • EEG-DATA判定境界: A(直接検証可能)EEG-DATA単独で主要仮説まで検証可能(外部依存なし)。
  • auto-research-funds応募テーマ: 研究助成(再現性運用)
  • 第一応募先 / 予備応募先: G1 (GR-2026-013) / G3 (9Lx4dPK6a4k2gOb7)
  • 主データ(ID): 6
  • 補助データ(推奨ID): ID 6, 56, 719
  • 初期KPI(U14標準): 追試成功率 リーク検出率 再実行コスト
  • 停止条件: 主データ単独で再現しない場合は、前処理・分割・同期ログを固定して再試行し、それでも再現不能なら主張を下方修正する。
  • 最低成果物: Supplementary test operation report (including negative cases)

U14-2 探索研究と検証研究を運用上どう分離するか。

  • 検証命題: D01 Acquisition delay of wireless EEG (10.21227/dv1p-vq18 / subscription-gated support) を latency/disclosure floor とし、D13 を必須 public-open support、D08 を grounded support として、探索/検証の分離運用を行い、同一条件再実行の再現率を示す。
  • EEG-DATA判定境界: A(直接検証可能)EEG-DATA単独で主要仮説まで検証可能(外部依存なし)。
  • auto-research-funds応募テーマ: EEG 再現契約監査パックとしての exploration / confirmation split 固定
  • 第一応募先 / 予備応募先: くら基金 若手研究者奨励助成 / 中谷財団 研究助成(next-cycle watch)
  • 主データ: D01 Acquisition delay of wireless EEG
  • 補助データ(推奨): D13 recalibration-free-p300-eeg, D08 Simultaneous human intracerebral stimulation and HD-EEG
  • 初期KPI(U14標準): fixed-split rate rerun success rate explore-to-confirm carryover gap
  • 停止条件: D01 単独の latency floor から D13 支援の rerun trail へ接続できない場合、または split 違反例が D13 上でも安定しない場合、主張を latency-floor note へ下方修正する。
  • 最低成果物: Supplementary test operation report (including negative cases)

U14-3 leaderboardでのリーク・過適合・報告バイアスをどう監査するか。

  • 検証命題: ID 719を主に、ID 6/56を補助として leak challenge を再現し、split 固定前後の順位変動を監査する。
  • EEG-DATA判定境界: A(直接検証可能)EEG-DATA単独で主要仮説まで検証可能(外部依存なし)。
  • auto-research-funds応募テーマ: 研究助成(leaderboard leak audit benchmark)
  • 第一応募先 / 予備応募先: G1 (GR-2026-013) / G3 (9Lx4dPK6a4k2gOb7)
  • 主データ(ID): 719
  • 補助データ(推奨ID): ID 719, 6, 56
  • 初期KPI(U14標準): リーク検出率 split再現率 順位変動幅
  • 停止条件: リーク注入条件 の検出率が 0.80 未満、または split 固定後も 順位変動幅 > 3 が残る場合は leaderboard 契約を見送り、内部監査トラックへ留める。
  • 最低成果物: Leak audit results and repartition trail

U14-4 Model Card / Dataset Card を評価契約へどう統合するか。

  • 検証命題: D01 Acquisition delay of wireless EEG (10.21227/dv1p-vq18 / subscription-gated support) を latency/disclosure floor とし、D13 を必須 public-open support、D08 を grounded support として、Model Card・Dataset Card 必須化を rerun contract へ接続し監査項目化する。
  • EEG-DATA判定境界: A(直接検証可能)EEG-DATA単独で主要仮説まで検証可能(外部依存なし)。
  • auto-research-funds応募テーマ: EEG 再現契約監査パックとしての card-integrated reproducibility contract
  • 第一応募先 / 予備応募先: くら基金 若手研究者奨励助成 / 中谷財団 研究助成(next-cycle watch)
  • 主データ: D01 Acquisition delay of wireless EEG
  • 補助データ(推奨): D13 recalibration-free-p300-eeg, D08 Simultaneous human intracerebral stimulation and HD-EEG
  • 初期KPI(U14標準): required-card fill rate rerun success rate leak-audit coverage
  • 停止条件: D01 単独の disclosure floor から D13 支援の rerun examples へ接続できない場合、または card 欄の追加が traceability を改善しない場合、主張を disclosure-floor note へ下方修正する。
  • 最低成果物: Model/Dataset Card template and filled-in examples

U14-5 再現失敗ケースを否定例レジストリとして公開し、再試行サイクルをどう運用するか。

  • 検証命題: ID 6/56/719で再現失敗ケースを failure_id dataset_doi split_hash failure_type rerun_window owner で管理し、否定例の再試行SLAを固定する。
  • EEG-DATA判定境界: B(部分検証(外部データ併用で成立))失敗検出と分類はEEGで実行可能だが、公開範囲・同意・是正責任は制度運用の併設が必要。
  • auto-research-funds応募テーマ: 研究奨励金(否定例レジストリ)
  • 第一応募先 / 予備応募先: G1 (GR-2026-013) / G3 (9Lx4dPK6a4k2gOb7)
  • 主データ(ID): 719
  • 補助データ(推奨ID): ID 6, 56, 719
  • 初期KPI(U14標準): 追試成功率 リーク検出率 再実行コスト
  • 停止条件: 連続2サイクルで rerun_window 超過率が30%を超えた場合、公開運用を一時停止しワークフロー再設計を優先する。
  • 最低成果物: Negative Example Registry First Edition (CSV+Operating Rules)

U14-6 コンテナ固定(OS・依存ライブラリ・乱数種)を必須化した場合、再現コスト増分をどこまで許容するか。

  • 検証命題: ID 719でコンテナ固定時の計算コスト増分を測定し、許容上限を設定する。
  • EEG-DATA判定境界: A(直接検証可能)EEG-DATA単独で主要仮説まで検証可能(外部依存なし)。
  • auto-research-funds応募テーマ: 研究助成(再現コスト上限)
  • 第一応募先 / 予備応募先: G1 (GR-2026-013) / G3 (9Lx4dPK6a4k2gOb7)
  • 主データ(ID): 6
  • 補助データ(推奨ID): ID 6, 56, 719
  • 初期KPI(U14標準): 追試成功率 リーク検出率 再実行コスト
  • 停止条件: 主データ単独で再現しない場合は、前処理・分割・同期ログを固定して再試行し、それでも再現不能なら主張を下方修正する。
  • 最低成果物: Fixed container reproduction procedure manual and cost measurement

U15-1 神経データの機微性をどの法概念で扱うか(個人情報・生体情報・人格情報)。

  • 検証命題: 神経データ機微性の法概念マッピングは法務課題で、EEG-DATA単独検証は不可。
  • EEG-DATA判定境界: C(EEG-DATA単独では不可(制度・他モダリティ必須))制度・法務・他モダリティが主対象で、EEGは補助証拠に限定。
  • auto-research-funds応募テーマ: 制度設計助成(神経データ法制)
  • 第一応募先 / 予備応募先: G2 (GR-2026-014) / G6 (46z9VPE4wnkrvEJR)
  • 主データ(ID): 49
  • 補助データ(推奨ID): ID 49, 2412, 3419
  • 初期KPI(U15標準): 制度要件被覆率 監査項目実装率 停止/更新条件明確度
  • 停止条件: EEG補助証拠のみで制度/理論主張を確定しない。外部モダリティまたは制度根拠が揃うまで確定主張を保留する。
  • 最低成果物: System integrated audit report (suspension/renewal conditions)

U15-2 neurorightsを技術監査項目へどう写像するか。

  • 検証命題: neurorights監査項目化は制度設計が主で、ID 2412は技術ログ項目定義にのみ有効。
  • EEG-DATA判定境界: C(EEG-DATA単独では不可(制度・他モダリティ必須))制度・法務・他モダリティが主対象で、EEGは補助証拠に限定。
  • auto-research-funds応募テーマ: 制度設計助成(神経データ法制)
  • 第一応募先 / 予備応募先: G2 (GR-2026-014) / G6 (46z9VPE4wnkrvEJR)
  • 主データ(ID): 49
  • 補助データ(推奨ID): ID 49, 2412, 3419
  • 初期KPI(U15標準): 制度要件被覆率 監査項目実装率 停止/更新条件明確度
  • 停止条件: EEG補助証拠のみで制度/理論主張を確定しない。外部モダリティまたは制度根拠が揃うまで確定主張を保留する。
  • 最低成果物: System audit requirements table (technical log compatible)

U15-3 法域差(EU/US/JP等)を跨ぐ最小共通運用をどう定義するか。

  • 検証命題: 法域差を跨ぐ最小運用はデータでは確定できず、規制比較テーブル整備が必要。
  • EEG-DATA判定境界: C(EEG-DATA単独では不可(制度・他モダリティ必須))制度・法務・他モダリティが主対象で、EEGは補助証拠に限定。
  • auto-research-funds応募テーマ: 制度設計助成(神経データ法制)
  • 第一応募先 / 予備応募先: G2 (GR-2026-014) / G6 (46z9VPE4wnkrvEJR)
  • 主データ(ID): 49
  • 補助データ(推奨ID): ID 49, 2412, 3419
  • 初期KPI(U15標準): 制度要件被覆率 監査項目実装率 停止/更新条件明確度
  • 停止条件: EEG補助証拠のみで制度/理論主張を確定しない。外部モダリティまたは制度根拠が揃うまで確定主張を保留する。
  • 最低成果物: System audit requirements table (technical log compatible)

U15-4 技術進展に応じた停止条件・更新条件をどうガバナンス化するか。

  • 検証命題: ID 49/3419の運用ログを補助根拠に、停止条件 更新条件 例外承認 監査責任者 を1ページ規約として固定し、技術更新時の再審査手順を定義する。
  • EEG-DATA判定境界: C(EEG-DATA単独では不可(制度・他モダリティ必須))EEGは閾値設定の補助証拠に留まり、最終判断は法務/倫理/運用ガバナンスで決定する。
  • auto-research-funds応募テーマ: 制度設計助成(停止・更新ガバナンス)
  • 第一応募先 / 予備応募先: G2 (GR-2026-014) / G6 (46z9VPE4wnkrvEJR)
  • 主データ(ID): 49
  • 補助データ(推奨ID): ID 49, 2412, 3419
  • 初期KPI(U15標準): 制度要件被覆率 監査項目実装率 停止/更新条件明確度
  • 停止条件: 監査責任の不在または例外承認ログ欠落が確認された時点で、制度トラック応募を停止し規約更新を先行する。
  • 最低成果物: System integrated audit report (suspension/renewal conditions)

運用メモ

  • C 判定RQは、EEGを補助証拠として扱い、制度・法務・他モダリティの成立前に確定主張しない。
  • データセットID再採番に備え、提出時は DOI + データセット名 + access区分 を併記する。
  • 次runでは、本ページから 6RQ を選び、各RQの成果物を実ファイルとして残す。

2026-03-26 12:03 JST 再検証ログ(本run / 1RQずつ深掘りキュー再固定)

  • 60RQ 全件を対象とする主テーブル(### Ux-y または | Ux-y<br>)の件数は 60 を維持し、欠損・重複はありません。
  • 判定内訳 A/B/C=17/25/18 は前runから不変です。
  • EEG-DATA との照合は DOI + データセット名 + access区分 を正本にし、数値IDは探索補助に限定する運用を継続します。
  • 本runの深掘り優先キューは U0-4 -> U1-1 -> U4-2 -> U8-1 -> U13-3 -> U14-3 に再固定しました。
  • 各RQでは 検証命題 判定境界 応募テーマ 主データ 最低成果物 を先に固定してから実装します。

2026-03-27 09:30 JST 再検証ログ(本run / 非重複6RQの深掘り固定)

  • 作業開始前に auto-startupmain と、mind-upload / auto-research-funds / EEG-DATA の各 maingit pull --ff-only origin main を実行し、最新化済みであることを確認しました。
  • RQ_TOTAL=60A/B/C=17/25/18 を再確認し、1RQ=1検証命題=1応募テーマ=1主データ の運用を維持しました。
  • 本runの深掘り優先キューを、前run(U0-4/U1-1/U4-2/U8-1/U13-3/U14-3)と重複しない U1-2 -> U3-1 -> U7-4 -> U8-5 -> U11-4 -> U14-6 に固定しました。
  • U1-2 は不確実性伝播の校正、U3-1 は統合時改善量の主張上限、U7-4 は再計測/除外判定木、U8-5 はオーバーライド安全KPI、U11-4 は失敗条件の先行宣言、U14-6 はコンテナ固定コスト境界をそれぞれ主軸に設定しました。
  • 参照データ運用は、数値IDを探索補助に限定し、提出正本を DOI + dataset名 + access区分 に固定しました。

2026-03-28 07:02 JST 再検証ログ(本run / 非重複6RQの深掘り固定)

  • 作業開始前に auto-startupmain と、mind-upload / auto-research-funds / EEG-DATA の各 maingit pull --ff-only origin main を実行し、最新化済みであることを確認しました。
  • RQ_TOTAL=60A/B/C=17/25/18 を再確認し、1RQ=1検証命題=1応募テーマ=1主データ の運用を維持しました。
  • 本runの深掘り優先キューを、直近バッチ(U0-4/U1-1/U3-3/U8-6/U12-4/U15-4)と重複しない U0-2 -> U1-4 -> U4-3 -> U7-1 -> U13-2 -> U14-5 に固定しました。
  • U0-2 はms同期誤差の許容帯固定、U1-4 は後方分布公開の提出必須化、U4-3 は反事実誤差での理論順位監査、U7-1 はBIDS必須項目の欠落停止、U13-2 は模倣判定の二軸化、U14-5 は否定例レジストリの再試行SLA固定を主軸に設定しました。
  • 参照データ運用は継続して、数値IDを探索補助に限定し、提出正本を Dxx + DOI + dataset名 + access区分 に固定しました。

2026-03-28 08:02 JST 再検証ログ(本run / 非重複6RQの深掘り固定)

  • 作業開始前に auto-startup / mind-upload / EEG-DATA / auto-research-funds の各 main を pull し、最新状態で再検証しました(ローカル変更があるリポジトリは --autostash で保護)。
  • ### Ux-y 見出しを再計数し、RQ_TOTAL=60(欠損 0 / 重複 0)を再確認しました。
  • 判定内訳 A/B/C=17/25/18 は不変で、方針 1RQ=1検証命題=1応募テーマ=1主データ を維持します。
  • 本runの深掘り優先キューは、直近バッチと重複しにくい U0-1 -> U1-2 -> U4-4 -> U8-2 -> U12-3 -> U15-4 に固定しました。
  • U0-1 は同一性二軸ROCの合格条件固定、U1-2 は不確実性伝播の被覆率監査、U4-4 は反証閾値の先行宣言、U8-2 は再較正運用帯の数値化、U12-3 は心理連続性ログの制度分離、U15-4 は停止/更新ガバナンス規約化を主軸に設定しました。
  • データ参照運用は継続して、数値IDを探索補助に限定し、提出正本を Dxx + DOI + dataset名 + access区分 に固定します。

今回の再検証ログ(2026-03-28 16:02 JST)

本runでは、汎用横断サマリを増やさず、1RQ=1検証命題=1応募テーマ=1主データ の実行粒度を維持したまま、深掘り対象を次の6RQに固定しました。

  • 深掘り固定6RQ(本run): U0-4 U1-1 U3-3 U8-4 U14-1 U15-4
  • 目的: 分岐同一性、逆問題の事前分布制御、身体/環境ループ境界、閉ループの異常停止、安全な再現運用、停止/更新ガバナンスをそれぞれ1件ずつ詰める。
  • 再計数: RQ見出し=60、判定内訳 A/B/C=17/25/18 を維持。
  • 参照データ運用: 提出時のデータ参照は DOI + dataset名 + access区分 を必須とする。
  • 応募キー運用: G1-G6 の固定レーンを継続し、A は再現実装、B は外部依存境界つき実装、C は制度/法務・他モダリティ接続で提出する。