What the instrument returned

Every figure the repositories render, and the journeys.

Artifacts

Q1 — excess over each null against window length

Dose–response, with the detection floor drawn

How wide each arm of the state vector is

Distance from each animal's own baseline, by day

Tracking-flag load per rendered label

The four-method composite, reference arm marked

Journeys

+0.1067[+0.0783, +0.1373]provisional

context contrast, frame occupancy

+0.0245

the same contrast on bout rate

NOT_A_RESULT

the day-0 scalar

refused, and says why

What moves with context is how long behaviours are held, not how often they start.

Frame mass is the share of frames a label holds; event mass is the share of runs. They cross. Sizing anything by frame mass ranks stillness first, which is the bias the program exists to measure — so the tiles on Clips are sized by event mass and show both numbers.

Learning curve

Every MDE gate passes before the slope is read: the design had power to see a quarter of a plausible effect, on both occupancy arms.

gate: PASS, for every row below.

measure	MDE / day	plausible effect
discrimination (frame)	0.0379	0.1949
discrimination (bout)	0.1396	0.5910
stabilisation (frame)	0.0359	0.1949
stabilisation (bout)	0.1383	0.5910

And every group-level slope is null — both measures, both occupancy arms agree, which is the opposite of the Journeys section above: there, frame and bout occupancy split.

verdict: FAIL, for every row below.

measure	mean slope / day	95% CI	p
discrimination (frame)	+0.0096	[-0.0189, +0.0397]	0.429
discrimination (bout)	+0.0341	[-0.0838, +0.1484]	0.480
stabilisation (frame)	-0.0234	[-0.0485, +0.0026]	0.0475
stabilisation (bout)	-0.0252	[-0.1265, +0.0755]	0.6035

Top: per-animal slopes with the group mean and 95% CI. Bottom: individual per-animal trajectories on frame mass, coloured by that animal's own slope. A null group mean is not a null animal — individual slopes span roughly a unit peak-to-peak on frame occupancy and up to 3.4 on bout.

1.048

discrimination (frame) peak-to-peak

[-0.370, 0.677]

3.400

discrimination (bout) peak-to-peak

[-1.857, 1.543]

0.725

stabilisation (frame) peak-to-peak

[-0.446, 0.280]

2.653

stabilisation (bout) peak-to-peak

[-1.672, 0.981]

89 report-split animals, CFD days 3–7, MoSeq's unrestricted arm (2.82% artifact mass, same caveat as everything else MoSeq-labelled here). Provisional on the four-gate re-run. Per-animal values: results/per_animal_slopes.csv.