qishizhe/digital-drosophila
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

Audit calyx-specific PN contact scope without changing frozen baseline

Browse Source
This commit is contained in:
Digital Drosophila Archive
2026-09-30 00:27:22 +08:00
parent d6843e675d
commit 3213aada79
77 changed files with 4152 additions and 14 deletions
Download Patch File Download Diff File Expand all files Collapse all files
docs/progress/phase4ab-technical-results.md
+52
View File
@@ -0,0 +1,52 @@
# Phase 4AB — calyx contact-scope diagnostic
Status: COMPLETE-SCOPE-COUNTERFACTUAL / NO-DYNAMICS-CANDIDATE.
Diagnostic interventions only; no calibrated biological inhibition, no training, no parameter winner.
Eligible PNs: 229/259. CA mask: 20263 edges / 337099 contacts.
Affected KCs: 3760. Mixed-ROI affected edges: 3323; retained other contacts: 28975.
Broader PN→KC control: 21882 edges / 384555 contacts. All-fast-PN-output control: 116211 edges / 1268111 contacts.
| Odor | Condition | Active KC | KC late spikes | KC82312 late | VA6 late |
|---|---|---:|---:|---:|---:|
| ethyl (R)-(-)-3-hydroxybutanoate | frozen_reference | 160 | 114 | 60 | 213 |
| ethyl (R)-(-)-3-hydroxybutanoate | calyx_PN_KC_contact_block | 1 | 39 | 0 | 214 |
| ethyl (R)-(-)-3-hydroxybutanoate | all_PN_KC_block | 1 | 39 | 0 | 214 |
| ethyl (R)-(-)-3-hydroxybutanoate | whole_PN_output_block | 0 | 0 | 0 | 0 |
| 3-penten-2-one | frozen_reference | 172 | 118 | 60 | 213 |
| 3-penten-2-one | calyx_PN_KC_contact_block | 2 | 39 | 0 | 213 |
| 3-penten-2-one | all_PN_KC_block | 1 | 39 | 0 | 213 |
| 3-penten-2-one | whole_PN_output_block | 0 | 0 | 0 | 0 |
| (2R)-hexan-2-ol | frozen_reference | 144 | 117 | 61 | 213 |
| (2R)-hexan-2-ol | calyx_PN_KC_contact_block | 1 | 39 | 0 | 213 |
| (2R)-hexan-2-ol | all_PN_KC_block | 1 | 39 | 0 | 213 |
| (2R)-hexan-2-ol | whole_PN_output_block | 0 | 0 | 0 | 0 |
| gamma-propyl-gamma-butyrolactone | frozen_reference | 151 | 117 | 60 | 212 |
| gamma-propyl-gamma-butyrolactone | calyx_PN_KC_contact_block | 2 | 39 | 0 | 212 |
| gamma-propyl-gamma-butyrolactone | all_PN_KC_block | 1 | 40 | 0 | 212 |
| gamma-propyl-gamma-butyrolactone | whole_PN_output_block | 0 | 0 | 0 | 0 |
| 2-methyl butyl acetate | frozen_reference | 139 | 118 | 61 | 213 |
| 2-methyl butyl acetate | calyx_PN_KC_contact_block | 1 | 39 | 0 | 212 |
| 2-methyl butyl acetate | all_PN_KC_block | 1 | 39 | 0 | 212 |
| 2-methyl butyl acetate | whole_PN_output_block | 0 | 0 | 0 | 0 |
| ethyl (S)-(+)-3-hydroxybutanoate | frozen_reference | 163 | 119 | 61 | 214 |
| ethyl (S)-(+)-3-hydroxybutanoate | calyx_PN_KC_contact_block | 1 | 39 | 0 | 214 |
| ethyl (S)-(+)-3-hydroxybutanoate | all_PN_KC_block | 1 | 39 | 0 | 214 |
| ethyl (S)-(+)-3-hydroxybutanoate | whole_PN_output_block | 0 | 0 | 0 | 0 |
Late means 500–1,000 ms; 120 ms old calibration odor input; one seed, six old odors. No held-out generalization claim.
Identical scheduled requests across conditions. Delivered request-associated output differences versus reference: 459 (summed over all comparisons). See per-trial table; this is state/refractory feedback, not a changed RNG.
## Interpretation limits
CA selection uses exact primary_post labels plus matched APL→PN/ROI metadata, not bouton identities or inferred receptors. Unknown ROIs are not reclassified.
Whole-PN control removes fast output, not somatic spikes or the frozen graded APL input. It is not a complete PN knockout.
Reduction/silencing is not restored information or associative memory. No condition becomes a formal model. Fixed-event replay freezes other recorded events/APL; dynamic trials allow natural recurrent feedback.
Local-inhibition motivation: [Prisco et al.](https://doi.org/10.7554/eLife.74172), [Amin et al.](https://doi.org/10.7554/eLife.56954). These sources do not establish the complete-block operator as physiological.
## Verification
Six original trials reproduced exactly (whole spike events, watched state and final counts/V/g). Four first-odor repeats exact. Five deterministic-v2 frozen reference trials exact. Finite state throughout.
Original 230,239 files and both Phase 4AA manifests rechecked before/after. Base CSR/count/sign, frozen LIF/APL, eligibility-v2 and 14,232 plastic mask preserved.
Timing includes diagnostic CPU/GPU observation overhead and is not a runtime speed benchmark. ComfyUI was not stopped or unloaded.
Top-level acceptance gates remain 5/10 (50%); anatomical diagnosis is not evidence of learning success.
docs/progress/phase4ab-terminal-scope.md
+36
View File
@@ -0,0 +1,36 @@
# Phase 4AB 机制解读
正式状态：**COMPLETE-SCOPE-COUNTERFACTUAL / NO-DYNAMICS-CANDIDATE**。
这一轮区分了“PN→KC 传入位置”与“维持 PN 放电的上游反馈”。没有修复或批准新的 APL 生理模型，也没有训练。
| 诊断范围 | 活跃 KC（6 个旧气味） | 目标 KC 晚期 spikes | VA6 PN 晚期 spikes |
|---|---:|---:|---:|
| 原模型 | 139–172 | 60–61 | 212–214 |
| 仅明确 calyx 接触 | 1–2 | 0 | 212–214 |
| 同批 PN 的全部 KC 输出 | 1 | 0 | 212–214 |
| 同批 PN 的全部快速输出 | 0 | 0 | 0 |
晚期窗口是 500–1,000 ms。每个气味沿用 seed 131101；总计 24 组诊断、4 次精确重复，不是 5-seed 学习验证。
## 能说明什么
仅 calyx 接触阻断后，目标 KC82312 的晚期 spikes 范围为 0，VA6 仍为 212–214。
同时 KC 总活跃数降为 1–2，原模型为 139–172。不能把这种接近静默的结果包装成编码或记忆改善。
即使两个局部阻断条件的事件序列相同，其电压/g 仍可能不同；逐气味比较见 narrow-vs-allKC.csv。
宽范围 PN 快速输出阻断影响反馈路径，不等价于局部 calyx 调制；它也没有关闭 PN 胞体或 frozen graded APL 的 PN 输入。
## 仍不能说明什么
完全阻断只是工程极端对照，不是 APL 的实测释放抑制比例。没有估计 receptor、bouton、局部扩散范围或时间常数。
未使用 held-out 气味，未检验学习、保持、反转或动作输出。不得选取这次“让活动更少”的操作作为正式脑模型。
剩余 late KC 的 ID/type 已单独保存；其来源不能仅凭残余计数猜测。
## 下一步边界
先继续 upstream feedback / compartment-specific physiology 的证据约束，区分 calyx 信号传递与触角叶持续放电，再确定可检验的局部动力学假设。
不能靠增加整体抑制、调 eta/threshold、扩大 plastic mask 或让 KC 全静默宣称修复。当前不进入学习、游戏或视觉。
相关原始研究：[Prisco et al., 2021](https://doi.org/10.7554/eLife.74172)、[Amin et al., 2020](https://doi.org/10.7554/eLife.56954)。它们支持研究局部 APL 作用，不直接验证本实验完全阻断幅度。
顶层验收仍为 5/10（50%）；本轮推进的是机制定位与可重复诊断能力。最终完整性状态见 regression/final-integrity.json。