DANDI:001347

Late eye closure, normalized by early eye closure (0.1 to 0.2 s after airpuff onset), decreases during ketamine (n = 4 human participants

Figure 1E; Table S3, Fig. 1E; Results, Precisely timed affective behavior in mice and humans, paragraph 3

What the paper reports

• Paired t-test, t = 3.703, p = 0.0342 (n = 4 participants, 38 trials each)

Trials 3 to 40, ratio of trial means (Open A a, Open B a) gives saline 0.706 (s.e.m. 0.084) and ketamine 0.426 (0.121), lower in 4 of 4 participants, paired t = 3.703, p = 0.0342, the values of Table S3. The three other readings of Open A and Open B give 0.703 to 0.705 and 0.424 to 0.426, p = 0.030 to 0.035.

How the test differed from the paper’s method

• The eye-closure traces are taken as deposited (already normalised to the open eye); the eye sizes behind them are not in the files, so the normalisation cannot be repeated.
• Saline is the pre_trials_data series by the code's label (p50-condition-labels-link).

Early eye closure (during 0.1 to 0.2 s after puff onset) does not change between saline and ketamine sessions

fig. S1B; Table S3, fig. S1 B

What the paper reports

• Paired t-test, t = 1.622, p = 0.2032 (n = 4 participants, 38 trials each)

Early eye closure 0.839 (s.e.m. 0.014) on saline and 0.777 (0.042) on ketamine, difference 0.062, Cohen dz 0.81; paired t = 1.622, p = 0.2032, not significant, as in Table S3. Trials 2 to 40 give t = 1.617, p = 0.2042.

How the test differed from the paper’s method

• As fig1e.

Late eye ... closure (during 0.3 to 0.8 s after puff onset) decreases with ketamine relative to saline

fig. S1C; Table S3, fig. S1 C

What the paper reports

• Paired t-test, t = 3.653, p = 0.0354 (n = 4 participants, 38 trials each)

Late eye closure 0.591 (s.e.m. 0.066) on saline and 0.343 (0.103) on ketamine, paired t = 3.653, p = 0.0354, lower in 4 of 4; trials 2 to 40 give t = 3.813, p = 0.0317.

How the test differed from the paper’s method

• As fig1e.

Dissociation increases during ketamine, as measured using the Clinician-Administered Dissociative States Scale (CADSS)

fig. S1D; Table S3, fig. S1 D

What the paper reports

• Paired t-test, t = 7.034, p = 0.006 (n = 4 participants)

CADSS 0.50 (s.e.m. 0.29) on saline and 9.75 (1.25) on ketamine, paired t = 7.034 (ketamine higher), p = 0.0059. The scores are the code's constants (analyze_airpuff_p50.py), read from the code record; nothing in the dataset bears on them.

How the test differed from the paper’s method

• CADSS scores from the code record, as the criterion allows; the dataset holds none.

behavior in the independent iEEG-implanted inpatient cohort displayed a multiphasic eyepuff response with a ketamine-elicited selective decrease in the affective phase of the response

Figure 2C; Results, Brain-wide activity screen during emergence of affect in human beings, paragraph 3

Means over the 7 participants of each participant's median trace. Late eye closure 0.541 preinfusion, 0.306 ketamine, 0.518 postinfusion, lower on ketamine in 7 of 7 against both (paired t = 2.908, p = 0.027 and 3.447, p = 0.014). Early eye closure 0.877, 0.850, 0.875; ketamine against preinfusion paired t = 1.318, p = 0.236. With the windows shifted 0.05 s later (the code's plotting origin) late 0.523, 0.284, 0.503 and early p = 0.210.

How the test differed from the paper’s method

• The paper reports no test; the criterion's paired t-test of early eye closure is applied to the per-participant window means of the median traces.

inclusion criteria were met for participants exhibiting: 1) a significant change of late eye closure, normalized to early eye closure, between preinfusion and ketamine conditions, and 2) a significant change of late eye closure, normalized to early eye closure, between ketamine and post infusion conditions, in the direction corresponding to a recovery toward preinfusion

fig. S8A; Table S3, fig. S8 A; Methods, Drug infusion (inclusion criteria)

What the paper reports

• SD168, preinfusion against ketamine: Paired t-test with fdr-bh correction, t = 5.8087, p = 1.3 × 10⁻⁵, Benjamini-Hochberg-corrected (n = 24 trials)
• SD168, postinfusion against ketamine: Paired t-test with fdr-bh correction, t = 3.2763, p = 0.0033, Benjamini-Hochberg-corrected (n = 24 trials)
• SD172, preinfusion against ketamine: Paired t-test with fdr-bh correction, t = 5.6281, p = 1.8 × 10⁻⁶, Benjamini-Hochberg-corrected (n = 39 trials)
• SD172, postinfusion against ketamine: Paired t-test with fdr-bh correction, t = 11.7475, p = 6.5 × 10⁻¹⁴, Benjamini-Hochberg-corrected (n = 39 trials)
• SD181, preinfusion against ketamine: Paired t-test with fdr-bh correction, t = 10.5245, p = 4.1 × 10⁻¹¹, Benjamini-Hochberg-corrected (n = 30 trials)
• SD181, postinfusion against ketamine: Paired t-test with fdr-bh correction, t = 5.8804, p = 2.2 × 10⁻⁶, Benjamini-Hochberg-corrected (n = 30 trials)
• SD185, preinfusion against ketamine: Paired t-test with fdr-bh correction, t = 4.7462, p = 4.8 × 10⁻⁵, Benjamini-Hochberg-corrected (n = 31 trials)
• SD185, postinfusion against ketamine: Paired t-test with fdr-bh correction, t = 6.8624, p = 2.6 × 10⁻⁷, Benjamini-Hochberg-corrected (n = 31 trials)
• SD186, preinfusion against ketamine: Paired t-test with fdr-bh correction, t = 2.7912, p = 0.0236, Benjamini-Hochberg-corrected (n = 12 trials)
• SD186, postinfusion against ketamine: Paired t-test with fdr-bh correction, t = 2.6265, p = 0.0236, Benjamini-Hochberg-corrected (n = 12 trials)
• SD211, preinfusion against ketamine: Paired t-test with fdr-bh correction, t = 7.1858, p = 1.4 × 10⁻⁸, Benjamini-Hochberg-corrected (n = 39 trials)
• SD211, postinfusion against ketamine: Paired t-test with fdr-bh correction, t = 10.4531, p = 2 × 10⁻¹², Benjamini-Hochberg-corrected (n = 39 trials)
• SD245, preinfusion against ketamine: Paired t-test with fdr-bh correction, t = 5.2553, p = 1.2 × 10⁻⁵, Benjamini-Hochberg-corrected (n = 39 trials)
• SD245, postinfusion against ketamine: Paired t-test with fdr-bh correction, t = 2.9258, p = 0.0058, Benjamini-Hochberg-corrected (n = 39 trials)

All 7 participants show a significant decrease from preinfusion to ketamine and a significant recovery (14 of 14 corrected tests p < 0.05 in the stated direction), with the trial counts of Table S3 (24, 39, 30, 31, 12, 39, 39) and 21 of 21 means within tolerance. Thirteen of the 14 t values equal Table S3's; SD245's preinfusion against ketamine test gives t = 5.288 where the table has 5.2553.

How the test differed from the paper’s method

• The per-trial values are recomputed from the deposited trial windows; they equal the stored per-trial values to rounding. The file session called baseline is the preinfusion session for SD168, SD172 and SD245, as the code renames it (airpuff_utils.load_airpuff_behavior).

Lack of meeting the behavioral inclusion criteria appears to have arisen from a variety of potentially independent reasons

fig. S8B; Table S3, fig. S8 B; Methods, Drug infusion, last paragraph

What the paper reports

• SD163, preinfusion against ketamine: Paired t-test with fdr-bh correction, t = 0.3794, p = 0.7065, Benjamini-Hochberg-corrected
• SD170, preinfusion against ketamine: Paired t-test with fdr-bh correction, t = 0.1429, p = 0.8871, Benjamini-Hochberg-corrected
• SD178, preinfusion against ketamine: Paired t-test with fdr-bh correction, t = 1.7376, p = 0.0941, Benjamini-Hochberg-corrected
• SD184, preinfusion against ketamine: Paired t-test with fdr-bh correction, t = 0.9389, p = 0.3555, Benjamini-Hochberg-corrected
• SD207, postinfusion against ketamine: Paired t-test with fdr-bh correction, t = 2.1669, p = 0.0555, Benjamini-Hochberg-corrected
• SD215, preinfusion against ketamine: Paired t-test with fdr-bh correction, t = 1.2251, p = 0.2281, Benjamini-Hochberg-corrected
• SD242, postinfusion against ketamine: Paired t-test with fdr-bh correction, t = 0.2170, p = 0.8294, Benjamini-Hochberg-corrected

None of SD163, SD170, SD178, SD184, SD207, SD215 and SD242 meets both criteria; over all 14 participants the ones that do are the 7 of figs8a. The statistics Table S3 lists for the excluded participants are matched (for example SD207 postinfusion against ketamine |t| = 2.167, corrected p = 0.0555; SD242 |t| = 0.217, p = 0.8294).

How the test differed from the paper’s method

• As figs8a.

the iEEG-recorded participants also reported fully reversible induction of dissociation (measured with the CADSS instrument) during recordings

Figure 2D; Table S3, Fig. 2D; Results, Brain-wide activity screen during emergence of affect in human beings, paragraph 3

What the paper reports

• preinfusion against ketamine: Paired t-test with fdr-bh correction, t = 4.807, p = 0.004, Benjamini-Hochberg-corrected (n = 7 participants)
• postinfusion against ketamine: Paired t-test with fdr-bh correction, t = 4.512, p = 0.004, Benjamini-Hochberg-corrected (n = 7 participants)

CADSS 0.86 before, 22.43 during and 0.86 after ketamine (7 participants), paired t = 4.807 and 4.513, both corrected p = 0.0040. Without SD185's retrospective rating the infusion mean is 21.00. The scores are the code record's; nothing in the dataset bears on them.

How the test differed from the paper’s method

• CADSS scores from the code record (data/cadss_across_pts.csv), as the criterion allows.

We observed that the overall CADSS dissociation score was correlated at the individual participant level with inhibition of the affective eyepuff response (fig. S14G)

fig. S14G; Table S3, fig. S14 G; Results, Brain-wide activity screen during emergence of affect in human beings, paragraph 9

What the paper reports

• Spearman rank-order correlation coefficient, rho = 0.744, p = 0.014 (n = 10 participants)

Ten participants remain after excluding SD170, SD184, SD211 and SD215 (preinfusion median below 0.5). Reading (a), the code's eye-closure change, gives Spearman rho = 0.744, p = 0.0136; reading (b), ketamine minus preinfusion, rho = 0.561, p = 0.0916, not significant. With all 14 participants rho = 0.355, p = 0.213.

How the test differed from the paper’s method

• CADSS scores from the code record. The claim holds under reading (a) of the criterion only.

The code's four non-iEEG participants (h009, h011, h012, h013) are the dataset's four, each with one saline and one ketamine session, and every dataset series has 40 trials. Which series is saline rests on the code alone (it reads pre_trials_data as saline); the files do not label it, and the first visit was saline for three participants and ketamine for one, so the series names do not follow visit order.

How the test differed from the paper’s method

• Link claim written by this attempt.

All 14 iEEG participants with behaviour in the dataset are in the code record's CADSS table (data/cadss_across_pts.csv), each with a total before ketamine, during infusion and at 110 min; the table also holds 7 participants without behaviour in the dataset. Matched by participant identifier; nothing in the dataset can be compared with the scores.

How the test differed from the paper’s method

• Link claim written by this attempt.

Late eye closure, normalized by early eye closure (0.1 to 0.2 s afterpuff onset), decreases during ketamine (n = 5 mice

Figure 1K; Table S3, Fig. 1K

What the paper reports

• Paired t-test, t = 4.737, p = 0.009 (n = 5 mice, 15 trials each)

Open I reading (a), code trials: saline 0.262 (s.e.m. 0.065), ketamine -0.068 (s.e.m. 0.056), paired t = 4.737, p = 0.0091, identical to Table S3. Reading (b), trials 23-40: 0.272 and -0.064, t = 4.751, p = 0.0090. Per-mouse values printed; pairing by list position follows the code and cannot be checked in the data.

How the test differed from the paper’s method

• Per-mouse late/early is mean(late)/mean(early) over trials (eye_trace_utils.get_data_by_trial with mean_ref=True and plot_drug_comparison_bar_from_df), as M-EYE states.
• Open I reading (b) is run as trials 23 to 40 (18 trials, fig. S2B/S2C legends), because the sub-mouse files carry no trial times with which to apply the Methods' 3-min rule.

dissociative drugs such as ketamine and phencyclidine [(PCP), uniquely enabled for mice] similarly reduced late but not early eye closure

Figure 1L; Table S3, Fig. 1L; Results, Precisely timed affective behavior in mice and humans, paragraph 4

What the paper reports

• saline, pre against infusion: Paired t-test with fdr-bh correction, t = 0.73, p = 0.504, Benjamini-Hochberg-corrected (n = 5 mice)
• ketamine, pre against infusion: Paired t-test with fdr-bh correction, t = 9.53, p = 0.002, Benjamini-Hochberg-corrected (n = 5 mice)
• PCP, pre against infusion: Paired t-test with fdr-bh correction, t = 3.41, p = 0.041, Benjamini-Hochberg-corrected (n = 5 mice)

Reading (a): saline t = 0.733 BH p = 0.504; ketamine t = 9.525 BH p = 0.0020; PCP t = 3.409 BH p = 0.0406; all six means within tolerance; ketamine and PCP lower in 5/5 mice. Reading (b): BH p 0.626, 0.0032, 0.0363.

How the test differed from the paper’s method

• As fig1k.

late-phase eye closure in mice accumulated across a sequence of eyepuffs and decayed slowly after the final eyepuff in the sequence (Fig. 1M), demonstrating persistence (fig. S2E) and scalability (fig. S2, F and G)

Figure 1M; fig. S2E and S2F; Table S3, fig. S2 E; Results, Precisely timed affective behavior in mice and humans, paragraph 4

20 preinfusion trials per saline mouse (100 trials, 86 reopen): pooled KM median 15.45 s from 0.3 s after the eighth puff (reading b, within 15.16 +/- 1.52) and 13.05 s from the code's start 3.3 s after it (reading a, outside). Late eye closure on the eighth puff (mean 0.517) above the first (0.244) in 5/5 mice. The code's own run uses all 40 trials of each saline session (200 trials) and gives exactly 15.16 s from its start frame, while Table S3 says 20 trials per mouse, 100 trials; finding recorded. Threshold 1.1/1.2/1.3 x baseline gives 14.92/15.45/16.84 s (reading b). The printed lines of fig1_persistence.py set ntrials = 40 and start at frame 6060 (3.3 s after the eighth puff); the 200 trials give 15.16 s from that start and 18.16 s from 0.3 s after the eighth puff, so the pass under reading (b) is not the paper's computation. Whether the criterion's 20 preinfusion trials are the paper's is asked as a question for a reader.

How the test differed from the paper’s method

• Trials 1-20 of each saline session as the criterion and Table S3 say; the code (fig1_persistence.py) uses all 40.

Late (normed by early) eye closure scales with puff number but not puff intensity, comparing air puffs of different pressure

fig. S2G and S2M; Table S3, fig. S2 G

What the paper reports

• 8 PSI puff 1 against puff 8: Paired t-test with fdr-bh correction, t = 4.834, p = 0.0038, Benjamini-Hochberg-corrected (n = 8 mice)
• 18 PSI puff 1 against puff 8: Paired t-test with fdr-bh correction, t = 9.243, p = 10⁻⁴, Benjamini-Hochberg-corrected (n = 8 mice)
• 28 PSI puff 1 against puff 8: Paired t-test with fdr-bh correction, t = 11.633, p = 10⁻⁴, Benjamini-Hochberg-corrected (n = 8 mice)
• puff 8, 8 against 18 PSI: Paired t-test with fdr-bh correction, t = 0.024, p = 0.982, Benjamini-Hochberg-corrected (n = 8 mice)
• puff 8, 8 against 28 PSI: Paired t-test with fdr-bh correction, t = 1.106, p = 0.458, Benjamini-Hochberg-corrected (n = 8 mice)
• puff 8, 18 against 28 PSI: Paired t-test with fdr-bh correction, t = 0.718, p = 0.596, Benjamini-Hochberg-corrected (n = 8 mice)

Puff 1 vs 8: t = -4.834, -9.243, -11.633 (8, 18, 28 PSI), BH p 0.0038, 0.0001, <0.0001. Puff 8 between pressures: BH p 0.98 (8 vs 18), 0.46 (8 vs 28), 0.60 (18 vs 28). All six means within tolerance; statistics identical to Table S3. 28 PSI sessions are the preinfusion halves of the ketamine 35 mg/kg sessions.

How the test differed from the paper’s method

• Per-puff late/early follows plot_puff_comparison_bar_trials (mean_ref=True), trials 3 to 17 of the first 20 (fig1_psi.py which_pre).

these robust dissociative effects begin around 35 mg/kg ketamine and 10 mg/kg PCP (fig. S2, H to J)

fig. S2H to S2J; Table S3, fig. S2 H; Results, Precisely timed affective behavior in mice and humans, paragraph 4

What the paper reports

• saline: Paired t-test with fdr-bh correction, t = 0.980, p = 0.425, Benjamini-Hochberg-corrected (n = 5 mice)
• ketamine 50 mg/kg: Paired t-test with fdr-bh correction, t = 8.488, p = 0.011, Benjamini-Hochberg-corrected (n = 5 mice)
• ketamine 35 mg/kg: Paired t-test with fdr-bh correction, t = 3.843, p = 0.0159, Benjamini-Hochberg-corrected (n = 8 mice)
• ketamine 25 mg/kg: Paired t-test with fdr-bh correction, t = 1.232, p = 0.357, Benjamini-Hochberg-corrected (n = 5 mice)
• ketamine 13 mg/kg: Paired t-test with fdr-bh correction, t = 1.270, p = 0.357, Benjamini-Hochberg-corrected (n = 5 mice)
• ketamine 6 mg/kg: Paired t-test with fdr-bh correction, t = 0.329, p = 0.759, Benjamini-Hochberg-corrected (n = 5 mice)
• PCP 20 mg/kg: Paired t-test with fdr-bh correction, t = 6.080, p = 0.016, Benjamini-Hochberg-corrected (n = 5 mice)
• PCP 10 mg/kg: Paired t-test with fdr-bh correction, t = 5.404, p = 0.016, Benjamini-Hochberg-corrected (n = 5 mice)
• PCP 5 mg/kg: Paired t-test with fdr-bh correction, t = 2.682, p = 0.11, Benjamini-Hochberg-corrected (n = 5 mice)
• PCP 1 mg/kg: Paired t-test with fdr-bh correction, t = 2.064, p = 0.18, Benjamini-Hochberg-corrected (n = 5 mice)

Reading (a): ketamine 50 (BH p 0.0106), 35 (0.0159), PCP 20 (0.0159), 10 (0.0159) drop; saline 0.43, ketamine 25 0.36, 13 0.36, 6 0.76, PCP 5 0.11, PCP 1 0.18 do not; 10/10 as claimed, t values identical to Table S3. Reading (b): 10/10.

How the test differed from the paper’s method

• As fig1k.

buprenorphine and diazepam did not show the same disconnection of reflexive and affective eye closure (fig. S5, A to H)

fig. S5A to S5C; Table S3, fig. S5 C; Results, Precisely timed affective behavior in mice and humans, paragraph 4

What the paper reports

• saline: Paired t-test with fdr-bh correction, t = 0.9797, p = 0.4784, Benjamini-Hochberg-corrected (n = 5 mice)
• ketamine: Paired t-test with fdr-bh correction, t = 8.4882, p = 0.0053, Benjamini-Hochberg-corrected (n = 5 mice)
• PCP: Paired t-test with fdr-bh correction, t = 6.0804, p = 0.0092, Benjamini-Hochberg-corrected (n = 5 mice)
• buprenorphine: Paired t-test with fdr-bh correction, t = 0.0878, p = 0.9342, Benjamini-Hochberg-corrected (n = 5 mice)
• diazepam: Paired t-test with fdr-bh correction, t = 1.6120, p = 0.3038, Benjamini-Hochberg-corrected (n = 5 mice)

Reading (a): ketamine t = 8.4882 BH p = 0.0053, PCP t = 6.0804 BH p = 0.0092, buprenorphine t = 0.0878 BH p = 0.9342, diazepam t = 1.6120 BH p = 0.3038, saline BH p = 0.4784; identical to Table S3. Reading (b): 0.0046, 0.0098, 0.97, 0.17.

How the test differed from the paper’s method

• As fig1k.

general anesthesia, achieved through addition of xylazine to ketamine (a common combination treatment when general anesthesia is desired), abolished both reflexive and affective responses (Fig. 1P)

Figure 1P; fig. S5D and S5E; Table S3, fig. S5 D and E; Results, Precisely timed affective behavior in mice and humans, paragraph 4

What the paper reports

• early eye closure, pre against K/X: Paired t-test with fdr-bh correction, t = 8.6017, p = 0.006, Benjamini-Hochberg-corrected (n = 5 mice)
• late eye closure, pre against K/X: Paired t-test with fdr-bh correction, t = 12.0602, p = 0.0016, Benjamini-Hochberg-corrected (n = 5 mice)

Reading (a), code scale: K/X early 0.773 to 0.025 (t = 8.6017, BH p = 0.0060), late 0.470 to 0.024 (t = 12.0602, BH p = 0.0016), BH over six drugs; identical to Table S3. Methods scale and reading (b) give K/X means 0.023 to 0.027 and BH p <= 0.003.

How the test differed from the paper’s method

• Scale factor as figs2bc; the outcome does not depend on it.

under either of the two dissociative agents, ketamine and PCP, reflexive eyeblinks remained present but affective eye closure responses were greatly reduced in magnitude, did not accumulate, and did not persist following the eyepuff series (Fig. 1, N and O)

Figure 1N and 1O; fig. S5F and S5G; Table S3, fig. S5 F; Results, Precisely timed affective behavior in mice and humans, paragraph 4

What the paper reports

• decay rate, ketamine: Paired t-test with fdr-bh correction, t = 4.5198, p = 0.0266, Benjamini-Hochberg-corrected (n = 5 mice)
• decay rate, PCP: Paired t-test with fdr-bh correction, t = 5.3534, p = 0.0266, Benjamini-Hochberg-corrected (n = 5 mice)
• decay rate, saline: Paired t-test with fdr-bh correction, t = 0.8204, p = 0.7634, Benjamini-Hochberg-corrected (n = 5 mice)
• decay rate, buprenorphine: Paired t-test with fdr-bh correction, t = 0.4065, p = 0.8243, Benjamini-Hochberg-corrected (n = 5 mice)
• decay rate, diazepam: Paired t-test with fdr-bh correction, t = 0.2370, p = 0.8243, Benjamini-Hochberg-corrected (n = 5 mice)

Decay rate (21.2-23.2 s fit, reading a): ketamine 3.53 to 11.89 (t = -4.5198), PCP 3.04 to 13.20 (t = -5.3534), BH p = 0.0266 each; saline 0.76, buprenorphine and diazepam 0.82; all fits converged; legend window (to 23.0 s) and reading (b) agree. Early eye closure on the first puff remains (0.451 ketamine, 0.427 PCP). The criterion's last part fails as written: infusion late eye closure on the eighth puff is slightly above that on the first under ketamine (-0.007 against -0.025, difference +0.017, 95% CI -0.012 to +0.046, 4/5 mice, paired p = 0.175) and PCP (0.000 against -0.010, +0.010, 95% CI -0.019 to +0.039, 4/5, p = 0.404), against a saline rise of +0.350 (95% CI +0.167 to +0.534, 5/5, p = 0.006); the ketamine and PCP rises are below saline's by 0.333 and 0.341 (Welch p = 0.0067 and 0.0062). On a test reading of "not above" (no significant rise, interval including zero) the part holds; read as point values it fails.

How the test differed from the paper’s method

• Assumption, "not above" in the criterion's last part is read as a test (no significant rise from the first to the eighth puff), since both values sit at zero and the statement is that closure does not accumulate; read as point values the outcome would be partially reproduced. The drug against saline contrast treats sessions as independent, as the files carry no mouse label.
• Decay fit as eye_trace_utils.fit_exponential_decay with curve_fit default initial values (the code passes none).

Eyepuff results were repeatable across days (fig. S2, K and L)

fig. S2K and S2L; Table S3, fig. S2 L; Results, Precisely timed affective behavior in mice and humans, paragraph 4

What the paper reports

• normalised affective day 1 against day 2: Paired t-test with fdr-bh correction, t = 0.74, p = 0.6569, Benjamini-Hochberg-corrected
• normalised affective day 1 against day 3: Paired t-test with fdr-bh correction, t = 1.38, p = 0.5638, Benjamini-Hochberg-corrected
• normalised affective day 1 against day 4: Paired t-test with fdr-bh correction, t = 1.24, p = 0.5638, Benjamini-Hochberg-corrected
• normalised affective day 1 against day 5: Paired t-test with fdr-bh correction, t = 0.48, p = 0.6569, Benjamini-Hochberg-corrected
• reflexive day 1 against day 2: Paired t-test with fdr-bh correction, t = 0.91, p = 0.7425, Benjamini-Hochberg-corrected
• reflexive day 1 against day 3: Paired t-test with fdr-bh correction, t = 0.35, p = 0.7425, Benjamini-Hochberg-corrected
• reflexive day 1 against day 4: Paired t-test with fdr-bh correction, t = 3.31, p = 0.119, Benjamini-Hochberg-corrected
• reflexive day 1 against day 5: Paired t-test with fdr-bh correction, t = 0.63, p = 0.7425, Benjamini-Hochberg-corrected
• affective day 1 against day 2: Paired t-test with fdr-bh correction, t = 0.02, p = 0.9879, Benjamini-Hochberg-corrected
• affective day 1 against day 3: Paired t-test with fdr-bh correction, t = 1.13, p = 0.6418, Benjamini-Hochberg-corrected
• affective day 1 against day 4: Paired t-test with fdr-bh correction, t = 3.19, p = 0.1324, Benjamini-Hochberg-corrected
• affective day 1 against day 5: Paired t-test with fdr-bh correction, t = 0.63, p = 0.7475, Benjamini-Hochberg-corrected

0 of 12 tests with BH p < 0.05 under reading (a) (smallest 0.119, reflexive day 1 vs 4, |t| = 3.31) and under reading (b) (smallest 0.118); t values identical to Table S3. Days and mouse pairing come from the code's lists.

How the test differed from the paper’s method

• Days compared by get_drug_comparison_pvals with do_paired=True (fig1_days.py).

in both male and female mice (fig. S2N)

fig. S2N; Table S3, fig. S2 N; Results, Precisely timed affective behavior in mice and humans, paragraph 4

What the paper reports

• male pre against saline: T-test with fdr-bh correction, t = 1.080, p = 0.3827, Benjamini-Hochberg-corrected (n = 5 mice)
• female pre against saline: T-test with fdr-bh correction, t = 0.980, p = 0.3827, Benjamini-Hochberg-corrected (n = 5 mice)
• preinfusion male against female: T-test with fdr-bh correction, t = 2.079, p = 0.2851, Benjamini-Hochberg-corrected (n = 5 and 5 mice)
• saline male against female: T-test with fdr-bh correction, t = 0.983, p = 0.3827, Benjamini-Hochberg-corrected (n = 5 and 5 mice)

BH p 0.2851 (pre male vs female, t = -2.079), 0.3827 (saline male vs female, t = -0.983; male pre vs saline, t = -1.080; female pre vs saline, t = -0.980); identical to Table S3. Reading (b): smallest 0.35. Sex is not in the files (all "O"); it comes from fig1_sex.py, whose female group is the saline sessions 282-286.

How the test differed from the paper’s method

• Tests as get_sex_comparison_pvals (independent between sexes, paired within).

this assay was not modulated by anticipation of the precise timing of upcoming eyepuffs (fig. S2O)

fig. S2O; Table S3, fig. S2 O; Results, Precisely timed affective behavior in mice and humans, paragraph 4

Skip-puff sessions 411, 412, 413, 420, 421 are in the dataset; SKIP_PUFFS read from fig1_skippuff.py (10 preinfusion trials per mouse per skipped position) and confirmed by the data (link claim). Skipped sixth puff: mean rise 0.0008 (s.e.m. 0.0079), level 0.204 above baseline; skipped seventh: rise 0.0004 (s.e.m. 0.0035), level 0.182; the delivered puff before rises 0.369 and 0.373.

How the test differed from the paper’s method

• The criterion's reading as fixed in the claims file; the rise compared with the across-mouse s.e.m. of the rise.

We intercepted suprabulbar eyepuff-induced neural activity by inhibiting the mouse ventral posteromedial (VPM) thalamus, which transmits ocular-tactile input to the cortex (29), and found reduced affective eye closure relative to the reflexive response (fig. S6, A to J)

fig. S6C and S6F; Table S3, fig. S6 C and F; Results, Brain-wide activity screen during emergence of affect in human beings, paragraph 1

What the paper reports

• late/early, no injection against saline: Paired t-test with fdr-bh correction, t = 0.45, p = 0.6727, Benjamini-Hochberg-corrected (n = 5 mice)
• late/early, saline against muscimol: Paired t-test with fdr-bh correction, t = 3.85, p = 0.0289, Benjamini-Hochberg-corrected (n = 5 mice)
• late/early, no injection against muscimol: Paired t-test with fdr-bh correction, t = 3.79, p = 0.0289, Benjamini-Hochberg-corrected (n = 5 mice)
• fast early (0 to 40 ms), saline against muscimol: Paired t-test with fdr-bh correction, t = 0.37, p = 0.8898, Benjamini-Hochberg-corrected (n = 5 mice)

Late/early, eighth puff: no injection 0.633, saline 0.602, muscimol 0.105; saline vs no injection t = -0.455 BH p = 0.6727, saline vs muscimol t = 3.845 BH p = 0.0289, no injection vs muscimol t = 3.789 BH p = 0.0289. Fast early (0-40 ms): saline vs muscimol t = -0.367 BH p = 0.8898 with the script's independent t-tests (the values of Table S3, which calls them paired); paired t-tests give t = -0.379, BH p = 0.9013.

How the test differed from the paper’s method

• The fast-early test follows the script (get_drug_comparison_pvals without do_paired, i.e. independent t-tests); Table S3 labels it paired; both shown, same conclusion.

All 142 experiment ids named by the experiment lists of the ten mouse scripts are sub-mouse files, and every sub-mouse file is listed; no session is shorter than the ntrials its script reads (103 files of 40 trials, 39 of 20, the 20-trial ones listed only by fig1_psi.py and fig1_muscimol.py). Ids 1020 to 1027 serve as ketamine 35 mg/kg (40 trials) and as 28 PSI (first 20 trials); 282 to 286 serve as saline and as the female group. Label checks from the data, eighth puff, trials 6-20 against 23-37: all 5 ketamine 50 and all 5 PCP 20 sessions drop in late/early (median change -0.695 and -0.668) while the saline median change is 0.027; all 5 K/X sessions have infusion early (max 0.062) and late (max 0.051) below 0.1; the skipped puff inferred from the data (smallest early response) matches SKIP_PUFFS in 100% of infusion trials and 95% of preinfusion trials of the skip sessions. Not checkable from the data: the mouse behind each session (the scripts pair sessions by list position), sex, day, dose and pressure.

How the test differed from the paper’s method

• Link claim written by this attempt (criterion in the link_claims section).

Early eye closure (during 0.1 to 0.2 s after puff onset) decreases with ketamine but remains present ... closure (during 0.3 to 0.8 s after puff onset) is abolished by ketamine

fig. S2B and S2C; Table S3, fig. S2 B and C

What the paper reports

• early, saline against ketamine: Paired t-test, t = 6.854, p = 0.002 (n = 5 mice, 15 trials each)
• late, saline against ketamine: Paired t-test, t = 4.951, p = 0.008 (n = 5 mice, 15 trials each)

Reading (a), code scale: early saline 0.778, ketamine 0.451 (t = 6.854, p = 0.0024); late saline 0.203, ketamine -0.025 (t = 4.951, p = 0.0078); ketamine early above zero in every mouse. Both test statistics equal Table S3 to three decimals, so the per-mouse values are the paper's, yet the saline late mean is 0.203 (s.e.m. 0.047, median 0.19) against Table S3's 0.34 +/- 0.07, outside tolerance under every reading and scale (0.186 to 0.208). Three of four means within tolerance. The paper's own numbers imply about 0.20 (Fig. 1K saline late/early 0.26 times early 0.78), so the expected 0.34 may be a misprint; asked as a question for a reader (questions/001347.yaml).

How the test differed from the paper’s method

• Absolute values use the code's scale factor (one, since get_mean_traces computes the baseline after prox_norm); the Methods' 1/(1 - raw baseline of trials 1-2) is shown as a variant (early saline 0.722, ketamine 0.463, t = 2.873). The code scale reproduces Table S3's t values exactly; the late-saline mismatch holds under both.

In response to eyepuffs, mice showed affective eye closure that was reduced by ketamine (Fig. 4B and fig. S28A)

Figure 4B; Figure 5B; fig. S28A; Table S3, Fig. 4B and Fig. 5B; Results, Brain-wide activity screen during emergence of affect in mice, paragraph 2

10 of the 13 sessions of Table S3 are in the dataset (9 mice); the 001326 sessions of 2023-02-25 (sub-3) and 2023-07-06 and 2023-08-22 (sub-5) have no behaviour file. Criterion's method (baseline subtracted over the 1 s before the puff, no normalisation, stored trial ranges 1-20 and 23-30, mean over trials): first-puff late eye closure 0.174 +/- 0.061 preinfusion and 0.147 +/- 0.034 on ketamine (1.96 s.e.m.), intervals overlap; difference 0.027, paired 95% CI -0.039 to 0.094, lower in 6 of 10 sessions. Accumulation (late after puff 8 minus puff 1) is lower on ketamine, 0.046 against 0.058 (lower in 5 of 10). The intervals also overlap with ddof 0, ketamine trials 23 to 32, and the median over trials. The printed code shows that fig1_npx_behavior.py, which the criterion names, and its plot_npx_eyeclosure take a median over the stored trial ranges without baseline or normalisation, one session per figure, while make_puffseries_behavior_plot in analysis.py divides each trace by its maximum (lines 78 and 80). With that normalisation the intervals separate (0.630 +/- 0.080 against 0.318 +/- 0.061, lower in 10 of 10; accumulation 0.200 against 0.087). The stored first-puff puff_aff is 0.737 +/- 0.042 against 0.358 +/- 0.038 and puff_aff/puff_ref 0.948 +/- 0.042 against 0.634 +/- 0.069, lower in 10 of 10, but these are raw window means (puff_ref equals the raw 0 to 0.3 s mean), and the raw eye closure in the second before the first puff also falls, from 0.563 to 0.209. Fig. 5B's 11 sessions are not identifiable; the accumulation uses all 10.

How the test differed from the paper’s method

• The across-session summary follows the criterion (baseline subtraction with the 1 s before the puff, trials averaged within session, Table S3); the first puff 5 s after each events-on (confirmed from t_trial_eyeclosure and the data); trials 1-20 and 23-30 as stored in the files, 23-32 as a variant; baseline mean with nanmean (2 of 400 trials have a missing sample).
• The max normalisation of analysis.py (Zenodo 10.5281/zenodo.14876694, make_puffseries_behavior_plot) is run as a variant; it is not in the criterion, Table S3 or fig1_npx_behavior.py.
• 10 of 13 sessions (data missing for 3).

eyepuff sequences elicited a persistent reduction in reward seeking (fig. S4)

fig. S4; Results, Precisely timed affective behavior in mice and humans, paragraph 4

No lick data. Looked in: every asset path of 001347 at 0.250528.0702 (none with lick or reward); the only acquisition of the 142 sub-mouse files (eye_closure); every acquisition of the 10 Neuropixels-rig files; the 13 ecephys files of 001326 (units, electrodes, stimulus intervals and the ketamine time, no behaviour); all 142 sub-mouse sessions are accounted for by the scripts' lists, so none is an unlabelled lick session; the code record has no file that mentions licking.

the brain-wide eyepuff-triggered response pattern was absent in response to an unexpected auditory control

fig. S13D; fig. S19B; Results, Brain-wide activity screen during emergence of affect in human beings, paragraph 3

Control-session trial means of the 275 channels of the 5 participants with a control (SD181, SD185, SD186, SD211, SD245; 151 control trials) projected on the fig2g factors, channels excluded per factor as in fig2i. Mean control loading is below the preinfusion mean of the same channels for 6 of 6 factors (ratios 0.33 to 0.70) and for 30 of 30 participant and factor pairs. With preinfusion loadings from all preinfusion trials (the same trial use as the control) the control is still lower for 6 of 6 factors (ratios 0.342, 0.490, 0.536, 0.446, 0.766, 0.977; factor 5 0.299 against 0.306) and for 24 of 30 pairs (the exceptions on factors 2 and 5).

How the test differed from the paper’s method

• Preinfusion loadings of the test half of the trials (fig2i); all preinfusion trials as a variant.
• Control means read from sub-H_ses-sgramPerChannelPerTrial.nwb (all control trials, code X_ctrl); the dataset has no deposited control loadings to compare with.

Factoring the responses into six components (Fig. 2G) was found to be optimal, with an explained variance ... of 0.47 on held-out trials and diminishing returns with more components

Figure 2F and 2G; figs. S17 and S18; Table S3, Fig. 2G; Methods, Factorization

458 channels from 7 participants (SD168 86, SD172 97, SD181 58, SD185 43, SD186 47, SD211 44, SD245 83) in the 21 regions. Refitting NMF (3 positive, 3 negative components) on the mean of the preinfusion and postinfusion training halves reproduces the deposited factors to within 3e-13 and the deposited loadings exactly, so the dataset's per-channel file is the input the paper factored. Held-out explained variance with six components is 0.466 under the code's reading (training reconstruction against test data, decode_utils.nmf_k_determination with do_refit_W_for_test = False) and 0.535 with loadings refitted on the test data. The test curve gains 0.081 (2 to 4 components) and 0.056 (4 to 6), then 0.014 or less per step. Elbow at 6 by the largest fall in gain (both readings), at 8 by the largest distance from the chord over 2 to 28 components, and at 6 by the chord rule when the range ends at 16, 20 or 24 components. The criterion holds under the largest-fall rule, which is assumed. Per participant the test explained variance is 0.292 (SD186) to 0.654 (SD211). Both Open G readings give the code's numbering [0, 2, 4, 5, 1, 3]. 12 of the 28 fits stop at scikit-learn's default 200 iterations, as in the paper's code.

How the test differed from the paper’s method

• The fit starts from the dataset's per-channel train and test means (sub-H_ses-sgramPerChannel.nwb), not from the LFP; these were shown to equal the means of odd and even trials of the per-trial file (code split 1 and 0) for one channel per participant and session.
• Held-out explained variance follows decode_utils.nmf_k_determination (lines 136 to 291), which compares the training reconstruction with the test data; the refit reading is shown as a variant.
• Assumed elbow rule: the largest fall in the gain of test explained variance between successive steps. The paper and code give none (the code hard-codes k_elbow = 3); the chord-distance rule gives 8 over the full range of fig. S17A.

Two of these factors (1 and 2) showed decreased representation during ketamine followed by post infusion recovery (Fig. 2I)

Figure 2I; Table S3, Fig. 2I; Results, Brain-wide activity screen during emergence of affect in human beings, paragraph 6

What the paper reports

• factor 0, infusion against mean of pre and post: Mixed-effect linear model, grouped by subject, with fdr-bh correction across factors, coefficient = -0.049, p = 0.121, Benjamini-Hochberg-corrected (n = 312 channels, 7 participants)
• factor 1, infusion against mean of pre and post: Mixed-effect linear model, grouped by subject, with fdr-bh correction across factors, coefficient = -0.125, p = 2.6 × 10⁻⁵, Benjamini-Hochberg-corrected (n = 260 channels, 7 participants)
• factor 2, infusion against mean of pre and post: Mixed-effect linear model, grouped by subject, with fdr-bh correction across factors, coefficient = -0.115, p = 5.5 × 10⁻⁸, Benjamini-Hochberg-corrected (n = 185 channels, 7 participants)
• factor 3, infusion against mean of pre and post (model did not converge): Mixed-effect linear model, grouped by subject, with fdr-bh correction across factors, coefficient = -0.041, p = 0.016, Benjamini-Hochberg-corrected (n = 158 channels, 5 participants)
• factor 4, infusion against mean of pre and post: Mixed-effect linear model, grouped by subject, with fdr-bh correction across factors, coefficient = 0.013, p = 0.406, Benjamini-Hochberg-corrected (n = 370 channels, 6 participants)
• factor 5, infusion against mean of pre and post: Mixed-effect linear model, grouped by subject, with fdr-bh correction across factors, coefficient = 0.013, p = 0.341, Benjamini-Hochberg-corrected (n = 378 channels, 7 participants)
• factor 1, preinfusion against infusion (post hoc): Mixed-effect linear model, grouped by subject, with fdr-bh correction across factors, coefficient = 0.190, p = 1.8 × 10⁻⁹, Benjamini-Hochberg-corrected
• factor 1, postinfusion against infusion (post hoc): Mixed-effect linear model, grouped by subject, with fdr-bh correction across factors, coefficient = 0.061, p = 0.046, Benjamini-Hochberg-corrected
• factor 2, preinfusion against infusion (post hoc): Mixed-effect linear model, grouped by subject, with fdr-bh correction across factors, coefficient = 0.140, p = 1.2 × 10⁻⁸, Benjamini-Hochberg-corrected
• factor 2, postinfusion against infusion (post hoc): Mixed-effect linear model, grouped by subject, with fdr-bh correction across factors, coefficient = 0.089, p = 5.2 × 10⁻⁵, Benjamini-Hochberg-corrected

Infusion against the mean of pre and post (BH over six): factor 0 -0.049 (p 0.121, 312 channels), factor 1 -0.125 (2.56e-05, 260), factor 2 -0.115 (5.51e-08, 185), factor 3 -0.041 (0.0159, 158; the convergence flag of this fit, and of the post-hoc factor 3 fit against postinfusion, varies between runs), factor 4 0.013 (0.406, 370), factor 5 0.013 (0.341, 378). Post hoc (BH over 12), infusion minus postinfusion -0.061 (0.0464) for factor 1 and -0.089 (5.16e-05) for factor 2, so both recover. All six means within tolerance. Every coefficient, p and channel count equals Table S3; Table S3 reports post-hoc coefficients with the opposite sign (comparison minus infusion) and gives 5 participants for factor 3 and 6 for factor 4, where all 7 participants have kept channels for every factor here (fewest 13 channels, SD186, for factor 3 and 26, SD211, for factor 4). Factor 1 falls from pre to infusion in 7 of 7 participants, factor 2 in 6 of 7 (SD181 rises 0.199 to 0.294). Post-hoc tests are significant for factor 0 (pre, 0.036), factor 4 (pre 0.000299, post 0.0464) and factor 5 (pre 0.00985), as in Table S3.

How the test differed from the paper’s method

• The loading threshold is applied to the loading of the averaged pre/postinfusion test spectrogram (code W_prepost, script lines 507 to 528), which the claims file words as the mean of the pre and post loadings.

occurrence of factor 1 ... was suppressed substantially in limbic, dorsal, and ventral attention networks, whereas occurrence of factor 2 ... was suppressed in somatomotor, dorsal, and ventral attention networks

Figure 2J; fig. S19D and S19E; Results, Brain-wide activity screen during emergence of affect in human beings, paragraph 6

Reading (a) holds for both factors. Factor 1 change limbic -0.184, dorsal attention -0.225, ventral attention -0.266; factor 2 somatomotor -0.108, dorsal attention -0.190, ventral attention -0.157. Reading (b) holds for factor 2 and not for factor 1 (three most negative default -0.269, ventral attention, dorsal attention; limbic fourth); factor 1 falls in all six networks (-0.083 to -0.269). Dorsal attention rests on 7 channels from 3 participants (factor 1) and 5 from 2 (factor 2). The matrix equals the deposited Fig. 2J matrix.

How the test differed from the paper’s method

• Channel Yeo7 labels from the code record (link-ieeg-yeo-labels).

Permutation cluster ... testing resolved ketamine-altered spectrotemporal clusters which reached statistical significance in the limbic, ventral attention, and somatomotor Yeo7 resting-state networks

Figure 3B; fig. S21A; Table S3, Fig. 3B; Results, Brain-wide activity screen during emergence of affect in human beings, paragraph 7

Trial counts as in the criterion (178, 214, 153, 214, 214, 214). Limbic, ventral attention and somatomotor have significant clusters (lowest p 0.00195, the 1/512 floor; BH 0.0039); frontoparietal 0.027, default 0.045, dorsal attention 0.10, none below 0.01. t maps and contours equal the deposited ketChangePermutationClusterTest maps exactly in all six networks. Cluster-forming threshold 0.05 adds dorsal attention; 0.001 drops somatomotor. Inside the clusters the limbic change is negative and the somatomotor change positive in each of the 7 participants.

How the test differed from the paper’s method

• Uses the dataset's per-network per-trial spectrograms as deposited; their channel selection (both contacts in the same region) cannot be checked from the dataset.

These analyses consistently identified distinctive ketamine-elicited spectrotemporal pattern changes in the insular cortex (INS), cingulate cortex (ACC, MCC), orbitofrontal cortex (ORB), posteromedial cortex (PMC), supramarginal gyrus (SMG), motor cortex (MOT), and subcortical structures including the hippocampus (HIPP), basal ganglia (BG, primarily caudate), and thalamus (THAL)

Figure 3C and 3D; figs. S22 to S25; Table S1; Results, Brain-wide activity screen during emergence of affect in human beings, paragraph 8

Trial counts equal the criterion's for all 21 regions. Reading (a), BH across 21 regions: INS, ACC, MCC, ORB, PMC, SMG, MOT, HIPP, BG, THAL ANT and THAL POS significant (lowest p 0.00195, corrected 0.0037); IFG, MFG, SS, PHG, STG, STS, MTG, ITG, TP, AMY not (10 of 10; 10 of 11 counting THAL POS among the others). The significant set is the same with BH across all 35 regions plus PRECG (one participant), which equals the set of regions with a deposited contour; t maps equal the deposited ones in all 33 deposited regions. ACC, MCC and MOT rest on 2 participants. Threshold 0.001 keeps all listed but THAL POS; 0.05 adds MTG and STG. Reading (b) not run, as (a) holds.

How the test differed from the paper’s method

• Uses the dataset's per-region per-trial spectrograms as deposited. The paper's correction family appears to be every region in its file (33 deposited, 35 in the per-region file); the criterion's 21 is used, with 35 shown.

all participants with strong dissociation (CADSS score >10) exhibited low-frequency (<10 Hz) oscillations induced by ketamine

fig. S27E to S27I; Results, Brain-wide activity screen during emergence of affect in human beings, paragraph 9

Approximate test on the 20-min LFP files, CADSS from the code record (SD168 9, SD172 30, SD181 10, SD185 0, SD186 33, SD211 26, SD245 39). Main reading (baseline the preinfusion and postinfusion files, ketamine the second half of the infusion file, prominence 0.1): every high-CADSS participant has at least one channel with a ketamine-unique oscillation below 10 Hz, SD172 28 of 97, SD186 4 of 47, SD211 6 of 44, SD245 8 of 83 (CADSS > 10, 4 of 4) and SD181 33 of 58 (>= 10, 5 of 5). This holds in 50 of 54 variants under each threshold; the 4 exceptions use the combined baseline with the wider-peak overlap rule, and SD245 has none in each. All hits are at new frequencies; with 3 ketamine segments the rank-sum route cannot reach p < 0.001 (lowest 0.0022). The detection is not specific to ketamine: low-CADSS participants also have such channels (SD168 5 of 86, SD185 2 of 43), and the second half of the postinfusion file against the preinfusion file gives them in all 7 participants (60 channels, against 97 for the infusion file with the same baseline); only 2 of the 4 participants with CADSS > 10 have more on infusion than in that reference. Second part (not required), exact label permutation, channel fraction 0.240 high against 0.054 low under >= 10 (p 0.048 over 21 assignments), 0.170 against 0.214 under > 10 (p 0.629 over 35). Cohort (b) cannot be separated from (a).

How the test differed from the paper’s method

• Assumption, windows as figs15 (ketamine the second half of the infusion file, about 30 to 40 min after infusion start, the criterion's fallback window); baseline the preinfusion and postinfusion files combined (Methods, "pre- and post-infusion windows were combined"), the preinfusion file alone and its first half as variants. Sleep and ictal periods not excluded (no annotations).
• Assumptions where the Methods give no value, each with variants where it could matter, FOOOF fixed aperiodic mode fitted on 1 to 40 Hz (fooof 1.1.1), find_peaks prominence 0.1 (0.05 and 0.2 run), width at half prominence, 80% overlap on the narrower peak (wider run), single-linkage clustering, cluster frequency the median centre, one height per segment.
• The label permutation is enumerated exactly over all assignments in place of 10,000 shuffles.

458 of 458 channels found; atlas region equals the channel-name region for 458 of 458. Of 126 participant, network and region counts in the dataset's per-network file, 120 equal the label file's and in the other 14 rows the label file has more electrodes (including labels marked EXCLUDE), never fewer. With these labels the deposited channel loadings give the deposited Fig. 2H and 2J matrices exactly (maximum difference 0.0). 221 of the 458 channels carry a single Yeo7 network label; the others (Depth, mixed labels) are excluded from network means as in the code.

factors 0 to 2 loaded heavily onto attentional and somatomotor networks whereas factors 3 to 5 instead loaded heavily onto limbic, default-mode, and frontoparietal networks

Figure 2H; Results, Brain-wide activity screen during emergence of affect in human beings, paragraph 5

Row-normalised preinfusion loadings. Factors 0 to 2 carry 0.671 of dorsal attention (8 channels), 0.651 of ventral attention (57) and 0.662 of somatomotor (37); factors 3 to 5 carry 0.536 of default (72), 0.495 of limbic (28) and 0.463 of frontoparietal (19). Four of six rows meet the criterion (partial range). The refit loadings are identical to the deposited ones and the matrix to the deposited Fig. 2H matrix, so the two failing rows are in the paper's own numbers. Both Open G readings give the same numbering. By participant the share of factors 0 to 2 in limbic runs 0.27 to 0.73.

How the test differed from the paper’s method

• Channel Yeo7 labels from the code record (link-ieeg-yeo-labels).

Frontal cortex exhibited the largest magnitudes of change, led by ... orbitofrontal cortex

fig. S15G to S15I; Results, Brain-wide activity screen during emergence of affect in human beings, paragraph 4

Approximate test on the 20-min LFP files of the 7 participants. Main reading (baseline the preinfusion file before its first puff, the Baseline file for SD168, SD172 and SD245; ketamine the second half of the infusion file, about 30 to 40 min; log power): the median magnitude is largest in MCC (2.644, 2 participants), then IFG (2.267, 6), ORB (2.141, 5) and MFG (2.048, 2), against a median region of 1.386 over 21 regions. The frontal regions rank 2 to 4, above the median region, but ORB is third, so the first part of the criterion fails. Over 24 variants (baseline part, ketamine window, log or linear power, SD245 file) ORB, IFG and MFG are above the median region in 24 and ORB is first in 2 (both with the whole preinfusion file as baseline and the infusion file after the last puff); MCC is first in 20 and MFG in 2. Within participants ORB is the largest region in SD172, SD181 and SD186 and third in SD185 and SD245. MCC rests on SD185 (1 channel) and SD245 (14 channels). Cohort (b) cannot be separated from (a), as the 7 participants with LFP files are the included ones. The criterion takes the median over participants; the fig. S15H legend reports the magnitude as mean +/- s.e.m. Under the mean MCC is still first (2.644), then IFG (2.367) and ORB (2.262), so the outcome is the same (a question for a reader is in questions/001347.yaml).

How the test differed from the paper’s method

• Assumption, the ketamine window is the second half of the infusion file (600 to 1200 s), placed at about 30 to 40 min after infusion start from the Methods (iEEG ketamine infusion protocol, "The eyepuff assay was administered preceding infusion, at approximately 30 minutes after the infusion start, and 110 minutes after infusion start") and the 10 min of recording before the first puff in every file; the files carry no infusion start time. Other windows run as variants.
• The 10 min of the preinfusion file before its first puff stand in for the 25-min pre-ketamine baseline (fig. S15B legend), which does not fit in the 20-min files; preinfusion windows exclude the eyepuff assay as in the Methods' phase-locking paragraph. The whole file is a variant.
• Multitaper spectrogram with the fig. S27 parameters (NW 30, 59 tapers, 30-s windows, 15-s stride), since the paper gives none for fig. S15; log10 power assumed, linear a variant; 1 to 100 Hz without 58 to 62 Hz. Bipolar within-shank channels with a region label, out-of-brain, excluded and the code's bad_ch contacts removed. Sleep and ictal periods not excluded (no annotations).

Despite similar magnitude of measured eyelid closure in spontaneous versus puff-triggered eyeblinks, the ERP pattern was not seen in spontaneous eyeblinks

fig. S12A to S12D; Table S3, fig. S12 C and D; Results, Brain-wide activity screen during emergence of affect in human beings, paragraph 3

The behaviour files hold continuous eye-closure traces, puff on/off events, per-trial closure and trial lists, but no spontaneous blink onsets (no field mentions blinks in any behaviour file). The onsets exist only as a hand-made list in the code record (for_reference/analyze_airpuff_ieeg_natural_blinks.py lines 42 to 79; 52 in the baseline sessions, matching Table S3). The claims file says to block in this case.

Across species we observed a decrease, caused by ketamine, in eyepuff-triggered persistence that recovered post infusion (Fig. 6, C and D)

Figure 6C; Table S3, Fig. 6C; fig. S37C; Results, Conserved emotion-related population dynamics across mammalian species, paragraph 3

What the paper reports

• preinfusion against ketamine: Paired t-test with fdr-bh correction, t = 5.887, p = 0.002, Benjamini-Hochberg-corrected (n = 7 participants)
• postinfusion against ketamine: Paired t-test with fdr-bh correction, t = 3.171, p = 0.019, Benjamini-Hochberg-corrected (n = 7 participants)

Main reading (Open C b, the code's z-scoring; Open D a, white matter excluded; the code's bad-channel list; per-trial 2-s baseline; SD245 Baseline file as in the code) gives 1.363, 0.411, 0.724, t = 5.876 (BH p = 0.002) preinfusion against ketamine and t = 3.172 (BH p = 0.019) postinfusion against ketamine, the paper's 1.36, 0.41, 0.72, t = 5.887 and 3.171. Ketamine below preinfusion in 7 of 7 participants, postinfusion above ketamine in 6 of 7. Of 32 combinations (Open C x Open D x bad list x trial baseline x SD245 file) the tests hold in 27 and tests plus means in 15, all 15 under Open C b (8 of 8 with the code's list, 7 of 8 with the dataset-only channel set; the exception has ketamine 0.482 against an upper bound of 0.48). Under Open C a (Methods) the preinfusion mean is 1.616 to 1.661, outside tolerance, in all 16 readings, and the postinfusion test holds in 11 of 16.

How the test differed from the paper’s method

• Bad channels are not marked in the dataset; the code's list (ketman/analyze/utils.py, bad_ch, read from the code record at run time) is used in the main reading, and every result is also shown with only the dataset's out-of-brain and EXCLUDE labels removed.
• Preinfusion session per the code (ketman/DATASETS.py airpuff_sessions with the rename of 'baseline' to 'preinfusion' in load_and_organize_raw_traces and load_patients_erp), which for SD168, SD172 and SD245 is the Baseline file of the day before; SD245's Preinfusion file is run as a variant.
• Each trial's epoch is baseline-subtracted by its mean over the 2 s before the puff before the weights and averages are computed (load_patients_erp, do_baseline_sub=True), a code detail not in the Methods; shown with and without.
• Bipolar pairs are formed between neighbouring contacts of each electrode group (shank).
• 60-Hz notch with Q = 20 (Methods) where the code uses Q = 30.

whereas similar analysis of a dimension encoding fast eyepuff-triggered activity exhibited no decrease during infusion (Fig. 6, E and F

Figure 6E; Table S3, Fig. 6E; Results, Conserved emotion-related population dynamics across mammalian species, paragraph 3

What the paper reports

• preinfusion against ketamine: Paired t-test with fdr-bh correction, t = 0.432, p = 0.681, Benjamini-Hochberg-corrected (n = 7 participants, 12 to 40 trials each per condition)
• postinfusion against ketamine: Paired t-test with fdr-bh correction, t = 0.576, p = 0.585, Benjamini-Hochberg-corrected (n = 7 participants)

Main reading as fig6c with the fast window gives 4.120, 4.006, 3.866, t = 0.433 (BH p = 0.680) preinfusion against ketamine and t = -1.072 (BH p = 0.650) postinfusion against ketamine; the paper gives 4.12, 4.00, 3.86, t = 0.432, p = 0.681. Pre minus ketamine mean 0.114, Cohen dz 0.16, 95% CI of the difference -0.529 to 0.756. Under Open C b both tests are non-significant in all 16 readings; with the dataset-only channel set they are non-significant in all 16 readings of both Open C readings. Only with the code's bad-channel list and the Methods z-scoring (Open C a) does postinfusion against ketamine reach BH p = 0.029 to 0.039 (postinfusion higher), in 8 of 8 such readings. The preinfusion against ketamine test is not significant in any of the 32 readings (smallest BH p 0.131). The postinfusion statistic matches the fig. S37B beta row of Table S3 (t = 1.063, p = 0.657), not the Fig. 6E row (t = 0.576, p = 0.585). The absence thus rests on the code's z-scoring (Open C b): under the Methods z-scoring with the code's bad-channel list it fails in all 8 readings.

How the test differed from the paper’s method

• As fig6c.
• Assumed Open C reading b (the code's z-scoring over -1.0 to 1.5 s and the 0.05-s shifted summary axis), because the readings differ with the code's bad-channel list and this one reproduces the paper's values to the third decimal.

All power bands (except for broadband LFP) exhibit significant decrease of persistent population projection from preinfusion to ketamine. Beta and delta further exhibit significant recovery during postinfusion.

fig. S37C; Table S3, fig. S37 C; Results, Conserved emotion-related population dynamics across mammalian species, paragraph 3

What the paper reports

• delta pre against ketamine: Paired t-test with fdr-bh correction, t = 3.930, p = 0.015, Benjamini-Hochberg-corrected
• delta post against ketamine: Paired t-test with fdr-bh correction, t = 2.502, p = 0.046, Benjamini-Hochberg-corrected
• theta pre against ketamine: Paired t-test with fdr-bh correction, t = 3.246, p = 0.035, Benjamini-Hochberg-corrected
• theta post against ketamine: Paired t-test with fdr-bh correction, t = 1.935, p = 0.101, Benjamini-Hochberg-corrected
• alpha pre against ketamine: Paired t-test with fdr-bh correction, t = 4.620, p = 0.007, Benjamini-Hochberg-corrected
• alpha post against ketamine: Paired t-test with fdr-bh correction, t = 1.777, p = 0.126, Benjamini-Hochberg-corrected
• low gamma pre against ketamine: Paired t-test with fdr-bh correction, t = 6.252, p = 0.002, Benjamini-Hochberg-corrected
• low gamma post against ketamine: Paired t-test with fdr-bh correction, t = 0.383, p = 0.715, Benjamini-Hochberg-corrected
• high gamma pre against ketamine: Paired t-test with fdr-bh correction, t = 6.870, p = 0.001, Benjamini-Hochberg-corrected
• high gamma post against ketamine: Paired t-test with fdr-bh correction, t = 1.189, p = 0.279, Benjamini-Hochberg-corrected
• broadband pre against ketamine: Paired t-test with fdr-bh correction, t = 2.625, p = 0.079, Benjamini-Hochberg-corrected
• broadband post against ketamine: Paired t-test with fdr-bh correction, t = 0.790, p = 0.46, Benjamini-Hochberg-corrected

Main reading (Open C b, white matter excluded, code bad list) BH p preinfusion against ketamine delta 0.015 (t = 3.929), theta 0.035 (3.241), alpha 0.007 (4.621), low gamma 0.002 (6.233), high gamma 0.001 (6.863), broadband 0.078 (2.629); postinfusion against ketamine delta 0.047, beta 0.019; all twelve means within tolerance and close to Table S3. Dataset-only channel set, same readings, 5 of 5 filtered bands, broadband 0.086, delta recovery 0.048. With white matter kept (Open D b, which also reproduces fig6c) 4 of 5 filtered bands hold (theta BH p = 0.084 to 0.102), the delta recovery fails (0.25) and broadband is significant (0.008), a partial reproduction. Under Open C a broadband is significant (0.002 to 0.017).

How the test differed from the paper’s method

• As fig6c; each band's weights are computed in that band.
• Assumed Open D reading a (white matter excluded, the code's use_all_channels = False), since the two readings that reproduce fig6c give different outcomes here.

Right: No ... significant differences across conditions of mean projection during early window

fig. S37B; Table S3, fig. S37 B

What the paper reports

• delta pre against ketamine: Paired t-test with fdr-bh correction, t = 0.817, p = 0.859, Benjamini-Hochberg-corrected
• theta pre against ketamine: Paired t-test with fdr-bh correction, t = 0.274, p = 0.793, Benjamini-Hochberg-corrected
• alpha pre against ketamine: Paired t-test with fdr-bh correction, t = 0.848, p = 0.508, Benjamini-Hochberg-corrected
• beta pre against ketamine: Paired t-test with fdr-bh correction, t = 0.432, p = 0.681, Benjamini-Hochberg-corrected
• gamma pre against ketamine: Paired t-test with fdr-bh correction, t = 1.245, p = 0.273, Benjamini-Hochberg-corrected
• high gamma pre against ketamine: Paired t-test with fdr-bh correction, t = 0.053, p = 0.96, Benjamini-Hochberg-corrected
• broadband pre against ketamine: Paired t-test with fdr-bh correction, t = 0.816, p = 0.761, Benjamini-Hochberg-corrected
• delta post against ketamine: Paired t-test with fdr-bh correction, t = 0.183, p = 0.859, Benjamini-Hochberg-corrected
• theta post against ketamine: Paired t-test with fdr-bh correction, t = 0.428, p = 0.793, Benjamini-Hochberg-corrected
• alpha post against ketamine: Paired t-test with fdr-bh correction, t = 0.703, p = 0.508, Benjamini-Hochberg-corrected
• beta post against ketamine: Paired t-test with fdr-bh correction, t = 1.063, p = 0.657, Benjamini-Hochberg-corrected
• gamma post against ketamine: Paired t-test with fdr-bh correction, t = 1.207, p = 0.273, Benjamini-Hochberg-corrected
• high gamma post against ketamine: Paired t-test with fdr-bh correction, t = 0.371, p = 0.96, Benjamini-Hochberg-corrected
• broadband post against ketamine: Paired t-test with fdr-bh correction, t = 0.319, p = 0.761, Benjamini-Hochberg-corrected

Under Open C b no test in any of the 7 bands gives BH p < 0.05 in any of the 4 Open D and bad-list readings; main reading delta t = 0.817, BH p = 0.859 (Table S3 0.817, 0.859). Under Open C a the low-gamma preinfusion against ketamine test is significant in all 4 readings (BH p = 0.020 to 0.049, ketamine lower) and, with the code's list, the beta postinfusion test (0.034, 0.039), so under the Methods z-scoring the absence fails in all four readings; the outcome rests on the code's z-scoring.

How the test differed from the paper’s method

• As fig6c.
• Assumed Open C reading b, as for fig6e.

Beta power yields a CD projection that is most similar to that of the mouse

fig. S37A; Table S3, fig. S37 A; Methods, Coding Dimension

Human mean preinfusion persistent projections (Open C b, white matter excluded, code bad list) averaged into 10-ms bins from -0.5 to 1.5 s, against the mean of 13 mouse sessions of DANDI:001326 (mouse_npx.py; first puff of each series, per-neuron variances). Pearson r beta 0.901, broadband 0.894, low gamma 0.769, high gamma 0.657, alpha 0.617, delta 0.580, theta 0.464. With the pooled-variance mouse weights beta 0.889, broadband 0.871. Over 16 readings (Open C, Open D, mouse variance, first puff or all puffs) beta is highest in 10, broadband in 4 (all-puff mouse average, or white matter kept with per-neuron variances), low gamma in 2 (Methods z-scoring with pooled variance). The margin of beta over broadband is small.

How the test differed from the paper’s method

• Assumed the human readings that reproduce fig6c and the mouse first-puff projection (the mouse code's limit=1); the readings give different outcomes.
• All 13 sessions of DANDI:001326 used; the paper does not say which sessions enter fig. S37A. Mouse processing (unit selection, weights, projection) is the entry's mouse_npx.py; its docstring gives the sources.
• The human trace is averaged into the mouse's 10-ms bins; the paper gives no time range or resampling.

In both humans and mice, we observed that ketamine robustly accelerated the baseline intrinsic timescale of the population (Fig. 6, G and H)

Figure 6G; fig. S34B; Table S3, Fig. 6G; Results, Conserved emotion-related population dynamics across mammalian species, paragraph 4

What the paper reports

• Paired t-test, t = 4.74, p = 0.0032 (n = 7 participants)

Open E reading b (first 10 min of the epoch, fit from lag one sample; code) with the fig6c weights and SD245's Preinfusion file (the file the code's timescale path loads) gives 0.115 s and 0.092 s, t = 4.78, p = 0.003 (paper 0.11, 0.09, t = 4.74, p = 0.0032), shorter in 7 of 7. Reading a (8 min before the first puff, from lag 0) gives 0.117 and 0.089, p = 0.016, also within tolerance. The test holds in 28 of 32 readings (fails only with white matter kept, the code's list and SD245's Baseline file, p = 0.058 to 0.080). Every reading with SD245's Baseline file has the preinfusion mean outside tolerance (0.122 to 0.143 s). Dataset-only channel set, reading b, 0.120 and 0.092, p = 0.0005.

How the test differed from the paper’s method

• As fig6c for the weights.
• Timescale input is the beta power of the z-scored LFP (not z-scored again) projected on the weights, without mean removal before the autocorrelation (get_cd_timescales, extract_timescale).

cross-patient regression coefficient is significantly positive for preinfusion air puff sessions, but not significant for ketamine sessions

fig. S38G to S38L; Table S3, fig. S38 H and K

What the paper reports

• persistent projection on pre-puff PLV, preinfusion: Mixed-effects linear model, coefficient = 12.12, p = 6 × 10⁻⁴, 95% CI 5.19 to 19.04 (n = 214 trials, 7 participants)
• persistent projection on pre-puff PLV, infusion: Mixed-effects linear model, coefficient = -2.15, p = 0.361, 95% CI -6.78 to 2.47 (n = 214 trials, 7 participants)
• affective eye closure on pre-puff PLV, preinfusion: Mixed-effects linear model, coefficient = 3.35, p = 0.0017, 95% CI 1.26 to 5.45 (n = 214 trials, 7 participants)
• affective eye closure on pre-puff PLV, infusion: Mixed-effects linear model, coefficient = -0.065, p = 0.96, 95% CI -2.83 to 2.7 (n = 214 trials, 7 participants)

Main reading (code bad list, PLV over all kept channels including white matter, Open C b, per-trial 2-s baseline, per-trial 0.5-0 s pre-puff subtraction), 214 trials of 7 participants, persistence on pre-puff PLV 12.38 (CI 3.09 to 21.68, p = 0.009) preinfusion and -6.21 (p = 0.35) during ketamine; affective eye closure 2.57 (p = 0.003) and 0.03 (p = 0.98). Paper 12.12 (p = 0.0006), -2.15 (0.36), 3.35 (0.0017), -0.065 (0.96). Eye closure meets its part in every reading. The persistence part is met with the code's list in 8 of 8 readings with the pre-puff subtraction and 2 of 8 without; dataset-only, 4 of 8 and 2 of 8 (with the PLV excluding white matter, either the preinfusion coefficient has p >= 0.05 or the infusion one p < 0.05, for example p = 0.03). Without the subtraction the preinfusion coefficient is mostly 2.5 to 6 with p >= 0.05.

How the test differed from the paper’s method

• Assumed the per-trial pre-puff subtraction of the projection (Methods, Coding Dimension, 'baseline subtracted, using a window from 0.5 to 0 s'), which the code comments out for single trials (compute_all_cd); adopted because it gives the paper's coefficient.
• Pre-puff PLV computed on the session-filtered beta phase over the 100 samples before each puff, averaged over all edges between kept channels; the code's per-trial function (compute_plv_per_puff_tr) filters each epoch and uses rolling windows whose parameters are not given.
• Affective eye closure is the per-trial 'affective' series of the behaviour file.

Participation is global, and uniformly distributed ... across networks

Figure 6B; fig. S35A to S35D; Table S3, fig. S35 C and D; Results, Conserved emotion-related population dynamics across mammalian species, paragraph 3

What the paper reports

• persistent, default: Bootstrap shuffle (network labels) with fdr-bh correction across networks, p = 0.974, Benjamini-Hochberg-corrected
• persistent, dorsal attention: Bootstrap shuffle (network labels) with fdr-bh correction across networks, p = 0.974, Benjamini-Hochberg-corrected
• persistent, frontoparietal: Bootstrap shuffle (network labels) with fdr-bh correction across networks, p = 0.792, Benjamini-Hochberg-corrected
• persistent, limbic: Bootstrap shuffle (network labels) with fdr-bh correction across networks, p = 0.974, Benjamini-Hochberg-corrected
• persistent, somatomotor: Bootstrap shuffle (network labels) with fdr-bh correction across networks, p = 0.974, Benjamini-Hochberg-corrected
• persistent, ventral attention: Bootstrap shuffle (network labels) with fdr-bh correction across networks, p = 0.974, Benjamini-Hochberg-corrected
• fast, somatomotor: Bootstrap shuffle (network labels) with fdr-bh correction across networks, p = 0, Benjamini-Hochberg-corrected
• fast, dorsal attention: Bootstrap shuffle (network labels) with fdr-bh correction across networks, p = 0.044, Benjamini-Hochberg-corrected
• fast, frontoparietal: Bootstrap shuffle (network labels) with fdr-bh correction across networks, p = 0.018, Benjamini-Hochberg-corrected

Main reading (beta weights, white matter excluded, code bad list) 570 channels, 275 with a single Yeo7 label; counts Default 87, Dorsal Attention 18, Frontoparietal 23, Limbic 34, Somatomotor 50, Ventral Attention 63 (claims file 88, 18, 23, 34, 50, 63). Persistent, no network BH p < 0.05 (smallest 0.684, frontoparietal, median 0.350); fast, somatomotor median 0.446 above every shuffle (BH p = 0), frontoparietal under-represented (BH p = 0.030). Dataset-only channel set, persistent smallest BH p 0.516, fast somatomotor 0.404 (BH p = 0). The measured values are medians (the legend's measure); the means are 0.437 and 0.339.

How the test differed from the paper’s method

• Yeo7 labels from the code record (with_labels); channels with a mixed or 'Depth' label take part in the shuffle but belong to no network, as in cd_utils.plot_cd_weights.
• Shuffle seed 1 with numpy's default_rng (the code uses np.random.seed(1) with permutation).

All 986 contacts and all 911 within-shank bipolar pairs of the 7 participants are found by name in anat/pts_ch_labels.pkl (pt_elec_names_mono, pt_elec_names_bipo), and the record's region label equals the dataset's electrode location for 100% of contacts in every participant.

How the test differed from the paper’s method

• Link claim written by this attempt (criterion in link_claims).

in both mice and humans, we discovered that ketamine decreased coupling among these sites (Fig. 6, I and J

Figure 6I; Table S3, Fig. 6I; Results, Conserved emotion-related population dynamics across mammalian species, paragraph 5

What the paper reports

• Paired t-test, t = 3.344, p = 0.0155 (n = 7 participants)

Open F reading b (10 min before each eyepuff session, Table S3), persistent sites the top 10% of |beta persistent weight|, white matter excluded, code bad list, 184 edges; ketamine/ preinfusion ratio 0.960 (s.e.m. 0.014), paired t = 2.834, p = 0.030, lower in 6 of 7; the z-scored change (Methods) -0.082, p = 0.034. Across 8 readings (white matter, bad list, SD245 file) the test holds in all (p = 0.0013 to 0.040) and the ratio is 0.926 to 0.963, outside 0.979 +/- 0.0061 in every reading. Reading a (25-min windows before and during the infusion) cannot be run, as the dataset holds only the 20-min eyepuff epochs. Sliding 5-s windows (Methods) with a 2.5-s stride give 0.9597 (p = 0.0158) and with a 1-s stride 0.9593 (p = 0.0146); over all 24 readings the test holds (p = 0.0011 to 0.0401) and the ratio is 0.9255 to 0.9633, never within tolerance.

How the test differed from the paper’s method

• PLV in 5-s windows with a 5-s stride, read from the fig. S38 code comment "using 5 sec w+s plv data" (for_reference/analyze_airpuff_ieeg_top_cd_network.py line 230); the Methods give 5-s sliding windows without a stride, and strides of 2.5 and 1 s are run. No exclusion of epileptic or sleep periods; the paper's code for Fig. 6I is not released.
• Persistent sites per cd_utils.cd_pctile (int(10% of channels) largest |weight|).

ketamine-induced decreases in coupling are greater in magnitude between high persistence sites, compared to between low ... persistence sites

fig. S38B and S38C; Table S3, fig. S38 B

What the paper reports

• delta: Mixed-effects linear model with fdr-bh correction, coefficient = -0.0038, p = 0.7, Benjamini-Hochberg-corrected, 95% CI -0.024 to 0.016 (n = 716 edges, 7 participants)
• theta: Mixed-effects linear model with fdr-bh correction, coefficient = -0.016, p = 0.16, Benjamini-Hochberg-corrected, 95% CI -0.038 to 0.005 (n = 716 edges, 7 participants)
• alpha: Mixed-effects linear model with fdr-bh correction, coefficient = -0.028, p = 0.021, Benjamini-Hochberg-corrected, 95% CI -0.05 to -0.007 (n = 716 edges, 7 participants)
• beta: Mixed-effects linear model with fdr-bh correction, coefficient = -0.038, p = 2.6 × 10⁻⁴, Benjamini-Hochberg-corrected, 95% CI -0.057 to -0.02 (n = 716 edges, 7 participants)
• gamma: Mixed-effects linear model with fdr-bh correction, coefficient = -0.023, p = 0.019, Benjamini-Hochberg-corrected, 95% CI -0.04 to -0.006 (n = 716 edges, 7 participants)
• high gamma: Mixed-effects linear model with fdr-bh correction, coefficient = -0.012, p = 0.16, Benjamini-Hochberg-corrected, 95% CI -0.028 to 0.003 (n = 716 edges, 7 participants)

Outcome the ketamine minus preinfusion PLV change z-scored by each edge's preinfusion window SD (the figure's 'phase locking (z)' units). Main reading (Table S3 bands, weights of each band, white matter excluded, code bad list) beta -0.0452 (95% CI -0.0810 to -0.0094, BH p = 0.0266), within -0.038 +/- 0.019; alpha BH p = 0.0572; low gamma positive (p = 0.5118). Over 64 readings (as figs38a) beta is significant and within tolerance in 26. Alpha, beta and gamma all hold with beta in tolerance in 3, all with the fig. S38 code's bands, the weights of each band, white matter kept and no bad-channel list (stride 5 s, alpha p = 0.0009, beta -0.0299, CI -0.0573 to -0.0024, p = 0.0398, gamma p = 0.0398; also stride 1 s and ceil sites). In the code's bands with white matter excluded beta is not significant in any reading with the weights of each band. With the beta weights for every band beta holds in 11 of 20 readings and alpha in none. The raw change (PLV units) meets the criterion in no reading. The edge counts differ from the paper's 716, as for figs38a. Bonferroni (the code's pval_bonf) does not give the same pattern: in the main reading beta has Bonferroni p = 0.0797, and in the code-band reading that meets the full criterion beta has 0.1991 and gamma 0.1989.

How the test differed from the paper’s method

• As figs38a.

Preinfusion, out-of-task phase locking is not significantly different among edges between low puff-triggered persistence sites (low CD, denoted as L) vs. among edges between high persistence sites

fig. S38A; Table S3, fig. S38 A

What the paper reports

• delta: Mixed-effects linear model with fdr-bh correction, coefficient = 0.0017, p = 0.76, Benjamini-Hochberg-corrected, 95% CI -0.005 to 0.008 (n = 716 edges, 7 participants)
• theta: Mixed-effects linear model with fdr-bh correction, coefficient = 0.0026, p = 0.76, Benjamini-Hochberg-corrected, 95% CI -0.004 to 0.009 (n = 716 edges, 7 participants)
• alpha: Mixed-effects linear model with fdr-bh correction, coefficient = 0.0017, p = 0.76, Benjamini-Hochberg-corrected, 95% CI -0.006 to 0.009 (n = 716 edges, 7 participants)
• beta: Mixed-effects linear model with fdr-bh correction, coefficient = 0.0016, p = 0.76, Benjamini-Hochberg-corrected, 95% CI -0.005 to 0.008 (n = 716 edges, 7 participants)
• gamma: Mixed-effects linear model with fdr-bh correction, coefficient = 0.0018, p = 0.76, Benjamini-Hochberg-corrected, 95% CI -0.004 to 0.007 (n = 716 edges, 7 participants)
• high gamma: Mixed-effects linear model with fdr-bh correction, coefficient = -0.00060, p = 0.83, Benjamini-Hochberg-corrected, 95% CI -0.006 to 0.005 (n = 716 edges, 7 participants)

Preinfusion PLV (10 min before the session) on high against low edge identity, mixedlm with a random intercept per participant, BH over bands. With the weights computed in each band (the criterion and the fig. S38B legend) the absence holds in 4 of 44 readings (band set as Table S3 or the fig. S38 code's (1, 4) to (71, 95), int or ceil site counts or white matter removed after ranking, four channel readings, strides of 5, 2.5 and 1 s, with or without dropping edges between neighbouring bipolar channels), all 4 with white matter excluded and the adjacent-edge filter; in the other 40 at least one band differs, theta and alpha both in 38 (higher among high-persistence edges, largest coefficient per reading 0.0084 to 0.0293). Main reading (Table S3 bands, white matter excluded, code bad list, 368 edges) theta 0.0227 (BH p = 0.012), alpha 0.0140 (0.033), low gamma 0.0094 (0.042), high gamma 0.0161 (0.012); in the code's bands theta, alpha and high gamma differ (BH p 0.003, 0.043, 0.043). With the beta weights ranking the sites in every band, as the code comment says (lines 231 to 232, "version 24-06-07 uses beta CD weights 0.5-1.5 to sort top v bot CD electrodes"), no band differs in any of 20 readings (largest absolute coefficient 0.0048 to 0.0091; the paper's 0.0006 to 0.0026). Per-participant site and edge counts are printed; no reading gives the paper's 716 edges (339 to 1020), and no channel set from the dataset or the code record's list gives 716 with int, ceil or round (nearest 724).

How the test differed from the paper’s method

• Edge sets formed here (top and bottom int(10%) of |weight| per participant and band, cd_utils.cd_pctile, white matter excluded per the Methods, and kept as a variant); the code reads them from a precomputed file (top_v_bot_cd_plv_edges.h5, lines 229 to 234) that is not in the record.
• Bands as the claims file (gamma 30 to 50 Hz) and the fig. S38 code's bands as a variant; PLV windows as fig6i. The code's beta-weight sorting of sites, which conflicts with the legend, is a variant.

Clusters with more rapid preinfusion dynamics were minimally altered or unaltered by ketamine, whereas clusters with slower and more persistent preinfusion dynamics were substantially altered (Fig. 4G)

Figure 4G; fig. S31A and S31C; Table S3, Fig 4G, fig. S31 A; Results, Brain-wide activity screen during emergence of affect in mice, paragraph 5

6553 puff-modulated cells from 13 sessions and 10 mice, 11 Leiden clusters (paper 12), each with cells from all 10 mice. Ketamine trials 23 to 32: non-overlap fraction rises with preinfusion decay time (Spearman rho = 0.47, p = 0.14); three slowest clusters 0.507, three fastest 0.368. Trials 22 to 32: rho = 0.27 (0.475 against 0.355). Resolutions giving 13 and 17 clusters: rho = 0.54 and 0.33, slowest above fastest. The correlation is positive but not significant in any version, and the fastest clusters still separate over 0.255 to 0.515 of the 0 to 2 s window, so "minimally altered" holds only relatively.

How the test differed from the paper’s method

• Scanpy 1.12.5 with leidenalg (resolution 1), random_state 0, numpy seed 42 (code); the code's Scanpy version is not pinned.
• z-scoring within cluster and condition: the transform set by the cluster mean trace is applied to each cell, intervals over cells (claims file); the code's interval is over the trace's bins.
• Decay time of the cluster average by the code's find_decay (25% of the normalised peak; 2 s when the peak z is below 1 or no fall within 2 s).

Measuring neural activity along this dimension revealed emotion-like properties of scalability through a saturating rise across the series of eyepuffs and of persistence through a decay lasting tens of seconds following each eyepuff series before returning to baseline (Fig. 5D)

Figure 5D; Table S3, Fig. 5D; Results, Brain-wide activity screen during emergence of affect in mice, paragraph 6

Main reading (dimension of fig5e): before ketamine the 1 s before the eighth puff exceeds the 1 s before the second by 0.846 (95% interval ± 0.170; every session positive, 0.354 to 1.291); 10 to 20 s after the last puff 0.838 ± 0.169, above zero; on ketamine the accumulation is -0.048 ± 0.114. All 16 readings (window 5 to 6 s or 3 to 4 s, variance, fast-dimension source, ketamine trials) pass all three conditions. The saturation is not part of the criterion.

How the test differed from the paper’s method

• Dimension, sign, epochs and exclusion as in fig5e; the signal is unfiltered (the Methods' 200-ms Butterworth smoothing is for display only); the 23 to 25 s stretch is left out as in the code (plot_affective_projection_late).

Not quoted: The variance explained values (0.954 for the neural fit; 0.870 and 0.927 for the behavioural fits) appear only as figure labels in the extracted text; the statement reads the Results paragraph 7 and the figure legends.

Neural fit (fig5e main dimension, 11 sessions): persistence model 0.881, within 0.954 ± 0.095 (0.947 with the code's 3 to 4 s window); broadcast and saturation alternatives 0.871 and 0.871 overall and 0.896 and 0.897 on the ketamine half against 0.918 (0.941, 0.941 against 0.947 with the code's window). Dose fits on DANDI:001347 eye closure: ketamine (6, 13, 25, 35 mg/kg) 0.930, PCP (1, 5, 10 mg/kg) 0.870, within 10% of 0.927 and 0.870 when 0.927 is ketamine (Fig. 5H); the 10% tolerance also admits the other assignment, so it cannot decide which panel carries which value. Adding the highest dose gives 0.941 and 0.919; adding saline as well 0.885 and 0.884.

How the test differed from the paper’s method

• Dose conditions: the code's just_ket and just_pcp orderings in mouse_eyepuff/scripts/ fig1_dose.py (assumed to be the conditions of Fig. 5H and 5I); session ids per dose from the same file's experiments dict, read at run time from Zenodo 14920570.
• Behavioural traces: prox_norm, 1/(1 - baseline) scaling and the median over trials 23 to 37 (code get_mean_traces as called in fig1_dose.py with agg_fn = np.median, which_post 22..36); mean over mice (Methods); 200 frames/s averaged to 10 ms bins; -1 to 43 s, 750 ms after each puff removed (code theoretical_behavior, filter_reflex). The driver of Fig. 5H and 5I is not in either code record.
• Neural data: the across-session mean of the fig5e signal (unsmoothed), preinfusion and ketamine (trials 23 to 32) concatenated; null model with the condition multipliers bounded to [0.99999, 1] (code compare_models, analysis.py lines 2813 to 2852 and 2867 to 2882, printed in the notebook).
• The dose session lists are checked by the link claim mouse-session-labels-link, which covers fig1_dose.py with the other mouse scripts.

Across species we observed a decrease, caused by ketamine, in eyepuff-triggered persistence that recovered post infusion (Fig. 6, C and D)

Figure 6D; Table S3, Fig. 6D

What the paper reports

• pre against ketamine: Paired t-test with fdr-bh correction, t = 5.662, p = 4 × 10⁻⁴, Benjamini-Hochberg-corrected (n = 11 sessions, 9 mice)
• post against ketamine: Paired t-test with fdr-bh correction, t = 2.657, p = 0.024, Benjamini-Hochberg-corrected (n = 11 sessions, 9 mice)

11 sessions from 9 mice (mouse 5 excluded, see deviations). Main reading (per-neuron variance, first puff of each series, trials 23 to 32 and 35 to 40): 0.867 ± 0.059 pre, -0.002 ± 0.133 ketamine, 0.515 ± 0.219 post, each within tolerance; pre against ketamine t = 5.48, corrected p = 0.0005; post against ketamine t = -2.54, corrected p = 0.029. The code's pooled variance gives 0.855, 0.019, 0.513, t = 5.64 and -2.61, corrected p = 0.0004 and 0.026, close to Table S3 (t = 5.662, 2.657; p = 0.0004, 0.024). With the Methods' trials (22 to 32, 34 to 40) the fall holds (corrected p = 0.0008) but post against ketamine is not significant with the first puff (corrected p = 0.17; 0.017 when all 8 puffs are averaged). Averaged within mouse (9 mice) corrected p = 0.0033 and 0.030; all 13 sessions 0.0001 and 0.032.

How the test differed from the paper’s method

• Epochs: the Methods (Comparison of rates) give trials 22 to 32 and 34 to 40, Table S3 gives 10 and 6 trials, and the code's slices (analysis.py, plot_coding_dimension_early_late, ket_range (2,12) and late_range (14,20) of the post-ketamine series) give 23 to 32 and 35 to 40. Assumed the Table S3 and code epochs, as two sources agree; both are shown and differ on the recovery test.
• First puff of each series only for the projection: code (limit = 1 for the puff); the Methods do not say. The all-puff average is shown as a variant.
• Coding-dimension variance per neuron (Methods, Coding dimension, mouse peri-puff); the code's pooled scalar (compute_cd_cov) is a variant.
• Projection z-scored over the 3-s window with the 1-s baseline subtracted (code); the Methods do not describe the normalisation.
• Excluded animal: the Methods exclude one stressed animal without naming it; mouse 5 is assumed, the only choice giving 11 sessions from 9 mice and 7 tone sessions from 7 mice.
• Rates z-scored per unit over the whole recording (assumed; the code's all_zfr driver is not in the record); a z-scored projection does not depend on it.

In both humans and mice, we observed that ketamine robustly accelerated the baseline intrinsic timescale of the population (Fig. 6, G and H)

Figure 6H; Table S3, Fig. 6H

What the paper reports

• Paired t-test, t = 3.357, p = 0.0073 (n = 11 sessions, 9 mice)

Main reading (per-neuron variance, series 11 to 20 and 23 to 32, fit from lag 0, per-series fits averaged): 0.437 ± 0.077 s pre and 0.162 ± 0.079 s ketamine, within tolerance; t = 3.51, p = 0.0057 (Table S3 t = 3.357, p = 0.0073); 9 of 11 sessions shorter; no fit failed. All 16 readings give p between 0.004 and 0.038 with ketamine shorter; preinfusion means 0.323 to 0.557 s, some outside tolerance. Within mouse t = 3.14, p = 0.014. With all 13 sessions the test fails (t = -0.98, p = 0.35) because one ketamine fit for mouse 5 (sub-5 ses-20230822) returns 154.8 s, far beyond the 2.5-s fitting window; without that session 0.410 against 0.151 s, p = 0.0046.

How the test differed from the paper’s method

• Series 11 to 20 and 23 to 32: code (stimulus_timescales, pre 10:20 and post 2:12), matching Table S3's 10 and 10 trials; the Methods do not say which. Methods epochs (1 to 20, 22 to 32) run as a variant.
• Fit from lag 0 (Methods, 'from the peak of the autocorrelation'); the code's start at lag one bin (extract_timescale) is a variant. Per-series fits averaged (Methods, Table S3); the code's concatenation of the series is a variant. p0 b = 0.2 (Methods; code t0 = 0.2).
• Window: the 30 s before the first puff of each series (Methods); the code's shifted psth driver is not in the record.

the rise times of average neural activity were not altered by ketamine (Fig. 4I and fig. S31, D to F), whereas regional times to decay from peak response were shortened (Fig. 4H and fig. S31G)

Figure 4H and 4I; fig. S31D to S31G; Table S3, Fig. 4H, I; Results, Brain-wide activity screen during emergence of affect in mice, paragraphs 4 and 5

13 sessions, 36 regions with peak d' of at least 1.3 (30 with 2). Decay: median change -0.189 s, bootstrap interval of the median -0.231 to -0.150 s, 34 of 36 regions shorter (d' 2: -0.186 s, -0.228 to -0.149). Rise: median change -0.009 s, interval of the median -0.018 to -0.002 s, excluding zero; 29 of 36 regions rise earlier. Under the other reading of the interval (2.5 to 97.5 percentiles of the code's pooled distribution over regions and bootstrap samples) rise includes zero (-0.083 to 0.068) but decay does too (-0.486 to 0.088). No reading satisfies both conditions; the decay shortening is clear and the rise change is small (a twentieth of the decay change) but not zero. Methods ketamine trials (22 to 32) give the same (-0.183 and -0.009 s). 0 of the 8 combinations of ketamine trials, d' threshold and interval reading satisfy both parts: under the interval of the median decay holds in 4 and rise in 0, under the pooled reading rise holds in 4 and decay in 0. The subcortical-to-frontal ordering is not in the criterion and was not judged.

How the test differed from the paper’s method

• Clusters from the fig4g refit (11 clusters, the paper 12); no cluster dropped (the code skips its cluster 11 for low activity).
• Regions: layer suffixes merged (code reduce_region), at least 30 modulated cells (Methods; the code uses more than 30), 'NA', 'STR', 'MB' and 'HY' left out (code, plotting).
• Ketamine series 23 to 32 (code region_time_course, puff_psth_ket[:, 2:12]); 22 to 32 run as a variant. Bootstrap 1000 resamples of puff replicates, seed 0 (standard choice).
• Interval reading: the claims file's 'its bootstrap 95% interval' is run as the interval of the median over regions across bootstrap samples and as the pooled distribution's percentiles.

in both mice and humans, we discovered that ketamine decreased coupling among these sites (Fig. 6, I and J

Figure 6J; Table S3, Fig. 6J; fig. S30

What the paper reports

• Paired t-test, t = 7.463, p = 2.2 × 10⁻⁵ (n = 11 sessions, 9 mice)

Methods reading (per-neuron-variance dimension, cells with |w| above the 90th percentile, mean Pearson r within each 30-s window averaged over series, series 1 to 10 and 23 to 32): ratio 0.852 ± 0.128, t = 1.10, p = 0.30, outside 0.49 ± 0.07; 31 to 76 cells per session. The variance of the coding dimension decides the outcome: with the code's pooled variance, signed selection and median |r| over concatenated windows (the code's full reading) the ratio is 0.559 ± 0.067, t = 6.32, p = 8.7e-05, within tolerance (Table S3 t = 7.463, p = 2.2e-5); pooled variance with the Methods' selection and averaging gives 0.576 ± 0.101, p = 0.0025. With per-neuron variance only the code's selection rule gives a reduction (0.832, p = 0.005 per series; 0.724, p = 0.012 concatenated), outside tolerance. With the Methods' series (1 to 20, 22 to 32) concatenated readings give no reduction (ratios 0.92 to 1.45). The Methods rule averages signed correlations near zero: over its 8 readings and 11 sessions the preinfusion mean r is 0.0069 to 0.0833 (median 0.0231, 31 of 88 values below 0.02), the per-session ratio runs 0.15 to 3.50 and the session-mean ratio 0.576 to 1.450 across readings; the code's median |r| gives preinfusion values of 0.0169 to 0.0841. With all 13 sessions, Methods 0.779 +/- 0.124 (p = 0.11), code 0.560 +/- 0.061 (12 sessions with a value, p = 3e-05). Partially reproduced: the claim holds under the code's reading the criterion names as a variant and not under the Methods reading, whose ratio of near-zero means is unstable.

How the test differed from the paper’s method

• Series 1 to 10 and 23 to 32: code (plot_coding_dimension_early_late, pre_bins pid 0..9 and ket_bins 22+pid), matching Table S3's 10 and 10 trials; Methods epochs run as a variant.
• Correlations within each 30-s window averaged over series (Methods, Table S3 'Values averaged within session across trials'); the code's concatenation is a variant.
• Coding-dimension variance per neuron (Methods); the code's pooled scalar is a variant. The test is the code's paired t-test of 1 against the normalised ketamine value.
• Session exclusion as in fig6d.

Whereas fast tone-related neural activity was unchanged on ketamine (fig. S34D), persistent tone-related activity decreased on ketamine and then recovered (fig. S34E)

fig. S34C to S34H; Table S3, fig. S34 C to H; Results, Conserved emotion-related population dynamics across mammalian species, paragraph 6

What the paper reports

• fast, pre against ketamine: Paired t-test with fdr-bh correction, t = 1.088, p = 0.318, Benjamini-Hochberg-corrected (n = 7 sessions, 7 mice)
• fast, post against ketamine: Paired t-test with fdr-bh correction, t = 1.438, p = 0.318, Benjamini-Hochberg-corrected (n = 7 sessions, 7 mice)
• persistent, pre against ketamine: Paired t-test with fdr-bh correction, t = 6.942, p = 9 × 10⁻⁴, Benjamini-Hochberg-corrected (n = 7 sessions, 7 mice)
• persistent, post against ketamine: Paired t-test with fdr-bh correction, t = -3.753, p = 0.009, Benjamini-Hochberg-corrected (n = 7 sessions, 7 mice)
• timescale, pre against ketamine: Paired t-test, t = 5.479, p = 0.0015 (n = 7 sessions, 7 mice)
• coupling, pre against ketamine: Paired t-test, t = 6.217, p = 8 × 10⁻⁴ (n = 7 sessions, 7 mice)

7 tone sessions from 7 mice. Fast tone activity not significantly changed (main corrected p = 0.16 pre against ketamine and 0.43 post against ketamine; every reading above 0.08). Persistent tone activity falls from 1.639 ± 0.053 to 0.481 ± 0.162 (t = 7.30, corrected p = 0.0007) and recovers (t = -2.91, corrected p = 0.027); preinfusion mean outside 1.531 ± 0.081 (the code's pooled variance gives 1.512 to 1.522, within; t = 7.10, p = 0.0008 with the code's split, as Table S3's 6.942, 0.0009). Timescale 0.304 ± 0.043 to 0.115 ± 0.027 s (t = 5.43, p = 0.0016); preinfusion within tolerance, ketamine outside 0.089 ± 0.012; all 16 timescale readings p < 0.04. Coupling not reduced under the Methods' reading (ratio 1.238 ± 0.451, p = 0.64); reductions only with pooled variance and median |r| (per series 0.741, p = 0.041; over the code's windows before puff series 0.535 ± 0.104, p = 0.006, within 0.46 ± 0.08). Over the 30 s before each tone series (Methods, Table S3) the code's rule gives 0.664 ± 0.163 (p = 0.105) concatenated with the injection split, 1.019 ± 0.241 (p = 0.94) with the code's split, and 0.741 ± 0.092 (p = 0.041) and 0.759 ± 0.090 (p = 0.047) per series, all outside tolerance; the coupling part fails over the tone windows. Preinfusion signed r of the Methods rule is 0.0136 to 0.0516 per session. With the 9 tone sessions (mouse 5 included) persistent activity falls (corrected p = 0.0008), the timescale falls (p = 0.033) and the code's rule over tone windows gives 0.613 to 0.889. The eye-closure part of the statement is not in the criterion and not tested; the Neuropixels-rig behaviour files hold puff trials only.

How the test differed from the paper’s method

• Tone epochs: up to 10 tone series before the split, tone series 2 to 6 and 8 to 10 after it (code, ket_range (1,6) and late_range (7,10) for 'tone'). Split at the file's injection time, giving Table S3's 10 preinfusion trials (assumed); the code's split (30 s after puff series 20) gives 9 and is run as a variant.
• Tone dimensions on the puff-modulated cells (code get_windowed_tone_psth selects with the puff tests).
• Timescale and coupling over the 30 s before each tone series (Methods, Table S3); the code's coupling uses the windows before puff series 1 to 10 and 23 to 32 even for the tone dimension, run as a variant.
• Variance, reference window, lag and averaging readings as in fig6d, fig6f, fig6h, fig6j.

These transient increases on ketamine were not different from those prior to ketamine, whereas late changes along the emotion-like dimension (seconds after each eyepuff) were reduced (Fig. 5E)

Figure 5D and 5E; Table S3, Fig. 5E; Results, Brain-wide activity screen during emergence of affect in mice, paragraph 6

What the paper reports

• 0 to 0.25 s, pre against ketamine: Paired t-test with fdr-bh correction, t = 1.209, p = 0.254, Benjamini-Hochberg-corrected (n = 11 sessions, 9 mice)
• 2 to 3 s, pre against ketamine: Paired t-test with fdr-bh correction, t = 2.874, p = 0.033, Benjamini-Hochberg-corrected (n = 11 sessions, 9 mice)

Criterion's dimension (target 5 to 6 s after the last puff onset, per-neuron variance, fast dimension from puffs 5 to 8, ketamine trials 23 to 32): early 0.899 ± 0.300 against 0.362 ± 0.160 (t = 1.95), late 0.343 ± 0.143 against 0.009 ± 0.076 (t = 1.90), both corrected p = 0.086. Late is lower on ketamine but not significant, and both preinfusion means are outside tolerance. Over the 8 readings with the 5 to 6 s window the corrected late p runs 0.066 to 0.158. The code's window (bins 400:500 of a post-series window whose plotting offsets place its start 1 s before the last puff, so 3 to 4 s after it) with pooled variance gives early 0.608 and 0.308 (t = 1.30, corrected p = 0.22), late 0.534 and 0.004 (t = 2.77, corrected p = 0.039), all four means within tolerance, close to Table S3 (t = 1.209 and 2.874, p = 0.254 and 0.033). With the 3 to 4 s window and per-neuron variance the early test is also significant (corrected p = 0.036). The 3 to 4 s offset is an inference from the printed code lines (analysis.py 432, 1436 to 1438, 1482 to 1494, 2833 to 2838); the code that builds the window is not in the record. With all 13 sessions the main reading gives uncorrected p = 0.039 (late) and 0.022 (early). Whether the criterion should name the two windows as readings is asked as a question for a reader.

How the test differed from the paper’s method

• Target window 1 s starting 5 s after the last puff onset, as the criterion and the Methods (Emotion-like neural dimension) state; the code's window (analysis.py, get_affective_projections, post_psth bins 400:500, read with plot_affective_projection_late, start = 500 plotted at 25 s) is inferred to be 3 to 4 s after the last puff and run as a variant.
• Sign of the dimension set so that puff 8 exceeds puff 1 and QR sign correction (code my_qr, get_affective_projections); the Methods do not say.
• Ketamine trials 23 to 32 (code ket_proj_normed [2:12], Table S3 10 trials); 22 to 32 run as a variant.
• Session exclusion and rate z-scoring as in fig6d.

whereas similar analysis of a dimension encoding fast eyepuff-triggered activity exhibited no decrease during infusion (Fig. 6, E and F

Figure 6F; Table S3, Fig. 6F

What the paper reports

• pre against ketamine: Paired t-test with fdr-bh correction, t = -0.548, p = 0.595, Benjamini-Hochberg-corrected (n = 11 sessions, 9 mice)
• post against ketamine: Paired t-test with fdr-bh correction, t = 2.468, p = 0.066, Benjamini-Hochberg-corrected (n = 11 sessions, 9 mice)

Pre against ketamine is not significant in any of 8 readings (main: t = -0.43, corrected p = 0.67; difference -0.054, 95% CI -0.296 to 0.189, dz = -0.13). Ketamine against post is significant in every reading (main: t = 4.07, corrected p = 0.0045; post lower by 0.322, dz = -1.23; range of corrected p 0.0045 to 0.016), where Table S3 gives t = 2.468, p = 0.066. This is a claim of absence judged on the paper's two tests, and the ketamine against post test finds a difference, so the claim is not reproduced; the pre against ketamine part, the ketamine effect itself, holds. Main means 4.226 ± 0.097, 4.279 ± 0.137, 3.957 ± 0.130 are below the paper's 4.48, 4.59, 4.32 by more than the reported s.e.m.; the code's reading (pooled variance, 70-ms reference) gives 4.433, 4.496, 4.161, within them. Within mouse: corrected p = 0.36 and 0.020; 13 sessions 0.57 and 0.010.

How the test differed from the paper’s method

• Reference window 1 s before each puff (Methods, Coding dimension, mouse peri-puff: 'average activity in the 1s before each puff'); the code uses the 70 ms before (bins 93:100), run as a variant.
• Epochs, variance, z-scoring and session exclusion as in fig6d.