Repository navigation
The turn was starved of time: omega_tau 1.5 → 2.5 restores the animal's own depth - #215
Merged
Merged
Conversation
…l's own depth Three measured-correct mechanisms pressed against one ceiling -- taxis outcomes never moved because the turn the circuit asks for is shallower than the animal's. The hunt (issue #196, tools/turn_depth.py) found where the depth goes, and it is not where anyone was pushing. Amplitude is healthy: reversal durations sit at their reference, the transient earns 0.84 of full amplitude on average, and corr(amplitude, reorientation) is 0.04. Doubling the current makes turns SHALLOWER -- saturating one side of the head pool disrupts the oscillation, exactly as the omega dossier's held-drive table warned. Time is the lever: at tau 1.5 s the ventral bias decays inside one undulation period and the wave overwrites it before it has carried down the body. The sweep reads 13.6 / 20.3 / 34.8 / 56.1 percent of reversals over 120 degrees at tau 1.5 / 2.0 / 2.5 / 3.0, and 2.5 restores the ~35% of Gray, Hill & Bargmann 2005 -- which is the mechanism's own original fit target (32%), silently slid off as later mechanisms changed what a turn completes to. This is maintenance of that fit, not a new mechanism. The paired battery at 2.5 (three assay families, identical seeds) prices it at nothing measurable: every trajectory guard clean, |heading drift| numerically improving 2.87 -> 0.87 deg/s, nociception all no-effect, chemotaxis all no-effect with the point estimates leaning better. The grazing determinism pin moved with the legitimate model change and is re-pinned (0.057295 -> 0.021999, history extended in the docstring); the runtime inherits the constant through OMEGA_DECAY and conformance reproduces to 5e-13 mm. The hunt also paid for an instrument lesson twice: a reorientation window flush against the reversal reads the turn's interior -- a still-rotating heading -- and at tau 2.5 it inverted the sign of a paired comparison. tools/ethogram.py's estimator now skips five seconds past the event, and tools/turn_depth.py measures at the commanded reversal edge where events do not fragment. NEXT.md's taxis-magnitude item flips from "blocked on turn depth" to "re-run at power against the deeper turn". Closes #196. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01CBroMXcYdfkyb5buJPcKvL
YesterdaysLemon
marked this pull request as ready for review
August 27, 2026 11:08
YesterdaysLemon
pushed a commit
that referenced
this pull request
Aug 28, 2026
…r, and refuted The standing diagnosis said taxis outcomes were small because the turn was shallow, and clearing the ceiling (#215) was supposed to re-open them. A new runner (tools/chemo_power.py -- every trial completes, cached to a resumable JSONL instead of dying to the pooled batch timeout that left the compare batteries at 4 paired animals of 16) measured the promise on all sixteen paired seeds: no effect on any chemotaxis metric, point estimates leaning worse, CI -0.045 [-0.154, +0.034]. The n=4 "trending better" was the noise its own MDE line warned about. The deep turn stays -- it was adopted on fit-maintenance grounds that stand -- but the taxis-magnitude hunt moves: conditioning was weak even before (pirouette down/up ratio 1.41 against the strong modulation of real worms) and reads 0.79 on the deep-turn animal, so the next questions are whether deep omegas contaminate the mechanical reversal detector (the estimator lesson, a third time) and how strongly dC/dt actually gates the pirouette machinery. Also in this stretch: the scorecard freshness workflow made its first autonomous commit to main after the PYTHONPATH fix (#217) -- the emitted table now measures the post-PVD, deep-turn animal, the buffer ghost row is gone, and peak curvature shows the sharper omega (14.5 /mm against the animal's 9.8, worth an eye). Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01CBroMXcYdfkyb5buJPcKvL
YesterdaysLemon
added a commit
that referenced
this pull request
Aug 28, 2026
…r, and refuted (#218) The standing diagnosis said taxis outcomes were small because the turn was shallow, and clearing the ceiling (#215) was supposed to re-open them. A new runner (tools/chemo_power.py -- every trial completes, cached to a resumable JSONL instead of dying to the pooled batch timeout that left the compare batteries at 4 paired animals of 16) measured the promise on all sixteen paired seeds: no effect on any chemotaxis metric, point estimates leaning worse, CI -0.045 [-0.154, +0.034]. The n=4 "trending better" was the noise its own MDE line warned about. The deep turn stays -- it was adopted on fit-maintenance grounds that stand -- but the taxis-magnitude hunt moves: conditioning was weak even before (pirouette down/up ratio 1.41 against the strong modulation of real worms) and reads 0.79 on the deep-turn animal, so the next questions are whether deep omegas contaminate the mechanical reversal detector (the estimator lesson, a third time) and how strongly dC/dt actually gates the pirouette machinery. Also in this stretch: the scorecard freshness workflow made its first autonomous commit to main after the PYTHONPATH fix (#217) -- the emitted table now measures the post-PVD, deep-turn animal, the buffer ghost row is gone, and peak curvature shows the sharper omega (14.5 /mm against the animal's 9.8, worth an eye). Claude-Session: https://claude.ai/code/session_01CBroMXcYdfkyb5buJPcKvL Co-authored-by: Claude <noreply@anthropic.com>
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Closes #196 — the highest-leverage item on the board: the shared ceiling under every taxis magnitude.
The hunt
New instrument
tools/turn_depth.pymeasures reorientation at the animal's own commanded reversals. Diagnosis first:The lever is time
At tau 1.5 s the ventral bias decays inside one undulation period; the wave overwrites it before it has carried down the body.
2.5 s restores the animal's ~35% (Gray, Hill & Bargmann 2005) — which is the mechanism's own fit target (32%), silently slid off as later mechanisms changed what a turn completes to. This is maintenance of the original fit, not a new mechanism.
The paired price: nothing measurable
Three assay families on identical seeds: every trajectory guard clean (|drift| numerically improves 2.87 → 0.87 °/s; net/path, speed unchanged), nociception all no-effect, chemotaxis all no-effect with point estimates leaning better (CI +0.049, approach +5.6 mm at n=4). NEXT.md's taxis item flips to "re-run at power against the deeper turn."
Bookkeeping
tools/ethogram.pynow skips 5 s past the event.OMEGA_DECAY; conformance reproduces to 5e-13 mm. All 49 behaviour + world tests pass locally.🤖 Generated with Claude Code
https://claude.ai/code/session_01CBroMXcYdfkyb5buJPcKvL
Generated by Claude Code