Coverage and design limits
This page exists to make the confound visible, not to work around it.
Area and subject are partially confounded. Of 12 recorded areas, 0 appear in all 3 subjects and 2 appear in only one. A between-area difference is therefore not separable from a between-subject difference by modelling alone, and no area-latency hierarchy is claimed anywhere in Analysis 6A. Pooled area medians, wherever they appear, are descriptive.
| Subject | sessions | eligible units | areas |
|---|---|---|---|
| M1 | 7 | 1,530 | 10 |
| M2 | 10 | 1,598 | 8 |
| M3 | 5 | 1,002 | 4 |
Unit counts are not independent biological replicates. Units within a session share a subject, a session, a probe insertion, and a preprocessing pass. Every inferential claim in Analysis 6A reduces each session to a single number before testing; the pooled unit-level percentages that appear on other pages are labelled descriptive for exactly this reason.
Session-level census
| subject_public | session_public | n_units_eligible | n_omission_detected | n_latency_resolved | n_abs_dT_gt_50 | resolved_up | resolved_down |
|---|---|---|---|---|---|---|---|
| M1 | S01 | 179 | 54 | 14 | 6 | 6 | 8 |
| M1 | S02 | 232 | 62 | 25 | 16 | 11 | 14 |
| M1 | S03 | 213 | 62 | 8 | 2 | 5 | 3 |
| M1 | S04 | 246 | 47 | 26 | 13 | 18 | 8 |
| M1 | S05 | 171 | 21 | 6 | 1 | 2 | 4 |
| M1 | S06 | 247 | 69 | 24 | 14 | 18 | 6 |
| M1 | S07 | 242 | 39 | 14 | 3 | 11 | 3 |
| M2 | S08 | 142 | 9 | 3 | 1 | 2 | 1 |
| M2 | S09 | 220 | 29 | 8 | 2 | 7 | 1 |
| M2 | S10 | 111 | 13 | 4 | 1 | 0 | 4 |
| M2 | S11 | 200 | 44 | 16 | 8 | 3 | 13 |
| M2 | S12 | 78 | 2 | 0 | 0 | 0 | 0 |
| M2 | S13 | 205 | 13 | 0 | 0 | 0 | 0 |
| M2 | S14 | 151 | 6 | 0 | 0 | 0 | 0 |
| M2 | S15 | 230 | 11 | 2 | 2 | 0 | 2 |
| M2 | S16 | 99 | 16 | 6 | 2 | 1 | 5 |
| M2 | S17 | 162 | 16 | 1 | 0 | 0 | 1 |
| M3 | S18 | 250 | 0 | 0 | 0 | 0 | 0 |
| M3 | S19 | 248 | 84 | 26 | 23 | 2 | 24 |
| M3 | S20 | 179 | 10 | 1 | 0 | 0 | 1 |
| M3 | S21 | 167 | 10 | 1 | 0 | 0 | 1 |
| M3 | S22 | 158 | 20 | 2 | 2 | 0 | 2 |