The rail has names: the head pool saturates, and the clamp holds no dynamics - #206
Merged
Merged
Conversation
…ynamics The scorecard printed "some neuron is on a clamp 75%/86% of the window -- should be neither" for as long as the line has existed. The new census mode of tools/clamp_occupancy.py names the culprits: the head motor pool and nobody else. SMB/SMD/RMD sit on each rail 28-41% of crawl instants per cell, RIV 6-8%, every other neuron under 3%, and the A/B cords at the 0% the original instrument already measured. The control answers what it means. With v_clamp widened to (-100, 65) mV the gait does not move -- frequency, wavelength, TWI, antagonism and curvature all within seed noise -- so the rail is a ceiling the head oscillation saturates against, not a hidden nonlinearity doing dynamical work: the drive saturates downstream through the activation nonlinearity before the extra headroom matters. The clamp stays where the ion physics puts it, and now the provenance says why that is safe. Bookkeeping: the scorecard's annotation now names the pool instead of raising an eyebrow, and worm/params.py::v_clamp carries the mechanism note. No model change; every number in this commit is a comment or an instrument. Closes #202. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01CBroMXcYdfkyb5buJPcKvL
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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 #202.
The question
The scorecard has printed this forever, with its own raised eyebrow:
A graded, sodium-channel-free network has no business on a voltage rail — and
tools/clamp_occupancy.pyhad already established the A/B cords spend ~0% there. So who is it?The answer (measured, seeds 0/1/3, 30 s crawls)
The head motor pool, and nobody else.
The per-seed "some neuron railed" fractions (70–78% hi / 83–89% lo) tie out exactly with the scorecard's 75%/86%.
What it means
The load-bearing control: widen the clamp to (−100, 65) mV and the gait does not move — freq 0.68→0.67, wavelength 0.83→0.84, TWI +0.88→+0.88, antagonism −0.73→−0.70, curvature RMS 4.52→4.51 (seed means). The head oscillation slams into the ceiling a third of every cycle, but the wave it produces doesn't depend on where the ceiling is: the drive saturates downstream through the activation nonlinearity before the extra headroom matters. The clamp is a physiological guardrail, not hidden dynamics — nothing behavioural secretly rests on it.
What changed
tools/clamp_occupancy.pygrows acensusmode (per-neuron occupancy for all 302 + the widened-clamp gait control) and records the reading in its docstring.worm/params.py::v_clampcarries the mechanism note.No model change — every number in this diff is a comment or an instrument.
🤖 Generated with Claude Code
https://claude.ai/code/session_01CBroMXcYdfkyb5buJPcKvL
Generated by Claude Code