The AS field is refused: a standing field buys turn depth at the price of every run - #209
Merged
Merged
Conversation
…e of every run The behavioural battery for the AS-class receptive field ran paired on identical seeds, three assay families, two gains (issue #193). At 1.0 the turns deepen exactly as the dossier hoped -- mean reorientation +31.9 degrees, fraction over 120 degrees up 23 points -- and everything around them breaks: heading drift triples on straight runs, spontaneous reversals in clean space double, and the pirouette food-conditioning ratio collapses from 1.77 to 1.02. At 0.5 the nociception and chemotaxis guards recover, but the navigation gains vanish with the harms and the drift is still detectable. The verdict is a mechanism lesson, not a tuning miss: a standing field rides the direction gate whenever it is open, so the same bias that deepens a reorientation also curves every straight run. Depth and drift arrive together at every gain, and no point on that axis is the animal. The field stays shipped at 0.0 (still provably inert, still pinned in the runtime registry). The follow-up hypothesis is recorded where the next reader will look: a turn-phase-gated field, built only while an omega is commanded, would keep the depth without touching the runs -- that thread lives with the turn-depth question (#196). The dossier at SensoryParams.as_field_gain carries both batteries in full. Closes #193. 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 26, 2026 23:59
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 #193.
The battery (paired on identical seeds,
tools/compare.py, three assay families, two gains)Gain 1.0 — the turns deepen exactly as the dossier hoped, and everything around them breaks:
Gain 0.5 — the guards recover and the gains vanish with them:
The verdict is a mechanism lesson, not a tuning miss
A standing field rides the direction gate whenever it is open, so the same bias that deepens a reorientation also curves every straight run. Depth and drift arrive together at every gain — no point on that axis is the animal.
What changed
SensoryParams.as_field_gaindossier carries both batteries in full; the field stays at 0.0 (provably inert, still pinned inRUNTIME_UNSUPPORTED).No model change — the mechanism was already off by default; this PR is the record of why it stays off.
🤖 Generated with Claude Code
https://claude.ai/code/session_01CBroMXcYdfkyb5buJPcKvL
Generated by Claude Code