Skip to content

Sleep's surface, watched: the bout is the circuit's, the rhythm is the animal's - #212

Merged
YesterdaysLemon merged 1 commit into
mainfrom
claude/repo-overview-directions-t7gut2
Aug 27, 2026
Merged

YesterdaysLemon merged 1 commit into
mainfrom
claude/repo-overview-directions-t7gut2

Conversation

@YesterdaysLemon

Copy link
Copy Markdown
Owner

Closes #197.

What

A new instrument, tools/sleep_surface.py, does the thing nobody had done since the homeostat landed: watch the animal simply live with it — ten undisturbed minutes per seed, two plates (the default plate is the ecology; the pharynx fixture's dense lawn is the incubator where bout statistics accumulate).

The record (600 s × seeds 0/1/2 per plate)

dense   seed 0: onset 121 s, 6 bouts, 46.5 ± 0.5 s, interval 32.8 ± 0.4 s
        seed 1: onset  98 s, 3 bouts, 46.0 ± 0.0 s, interval 43.5 ± 10.5 s
        seed 2: onset  85 s, 4 bouts, 39.8 ± 11.4 s, interval 114 ± 114 s
default seed 0: onset 483 s, one 46 s bout    seed 2: onset 448 s, one 46 s bout
        seed 1: left the dish at 118 s (→ #211)

The reading

  • Bout duration belongs to the circuit: 46 s ≈ tau_sleep · ln(0.69/0.25) almost exactly — the discharge is the bout.
  • Bout timing belongs to the ecology: intervals run from metronomic (a seed parked on food) to 114 ± 114 s (a seed wandering off the lawn edge, stalling its own pressure build). On the default plate sleep is a rare late event (7–8 min to onset) because dopamine only builds while the roamer happens to feed. That variance is foraging, not an injected noise term — the lifelike way to come by it.
  • Both quiescence effectors show on every bout (pump 1.3–2.6 Hz awake vs ~0.4 asleep; speed to a third), and the depth curve (Sleep depth: arousal threshold should ride the standing FLP-11 #204) rides each one (peak FLP-11 0.86–0.89).
  • The gap against You 2008 is variance in bout durations — and the record says where it should come from if it ever matters: satiety reaching tau_sleep, not a noise dial.

Bycatch

A default-plate roamer pushed through the dish rim's 0.05 mm repulsion zone and diverged at 118 s — filed as #211 with the geometry. The instrument now reports a dish-leaver as a finding instead of dying on it.

No model change; one new instrument, its record, and the NEXT.md sleep item flipped from "unmeasured" to measured.

🤖 Generated with Claude Code

https://claude.ai/code/session_01CBroMXcYdfkyb5buJPcKvL


Generated by Claude Code

…e animal's

A new instrument (tools/sleep_surface.py) does the thing nobody had done
since the homeostat landed: watch the animal simply live with it, ten
undisturbed minutes per seed, on two plates.

The reading splits cleanly. Bout DURATION belongs to the circuit: 46 s,
which is tau_sleep * ln(0.69/0.25) almost exactly -- the discharge is the
bout. Bout TIMING belongs to the ecology: on the dense lawn the intervals
run from metronomic (a seed parked on food, 32.8 +- 0.4 s) to 114 +- 114 s
(a seed wandering off the lawn edge and stalling its own pressure build),
and on the default plate sleep is a rare late event -- 7-8 minutes to
first onset, because dopamine only builds while the roamer happens to
feed. That variance is foraging, not an injected noise term, which is the
lifelike way to come by it. Both effectors show on every bout (pump
1.3-2.6 Hz awake against ~0.4 asleep, speed to a third), and the depth
curve rides each one (peak FLP-11 0.86-0.89).

What the model still lacks against You 2008 is variance in bout durations
-- the discharge constant fixes them -- and the record says where that
should come from if it ever matters: satiety reaching tau_sleep, not a
noise dial.

The run also caught something sleep-unrelated: a default-plate roamer
pushed through the dish rim's 0.05 mm repulsion zone and diverged at
118 s. Filed as #211 with the geometry; the instrument now reports a
dish-leaver as a finding instead of dying on it.

Closes #197.

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01CBroMXcYdfkyb5buJPcKvL
@YesterdaysLemon
YesterdaysLemon marked this pull request as ready for review August 27, 2026 02:25
@YesterdaysLemon
YesterdaysLemon merged commit 38c5894 into main Aug 27, 2026
12 checks passed
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

Sleep's behavioural surface: bout statistics, arousal habituation, and the arena energy audit

2 participants