DANDI:002034 · Encore Science

Claims 16

From the paper Limitations of pupil diameter as a proxy for modes of locus coeruleus activity · claims file · checked 2026-10-21: ready

data-units-sessions
The analysis used 42 LC single units from 32 sessions in monkey Oz and 63 units from 51 sessions in monkey Ci, recorded with pupil diameter during passive fixation with occasional startling beeps.
N = 42 single units from 32 sessions in monkey Oz and 63 units from 51 sessions in monkey Ci
Results, paragraph 1
How it will be judged
Counts of units and sessions in the code's include lists that hold beep trials with LC spikes and pupil. Tolerance zero, as these are counts of recorded units and sessions, not estimates.
fig2-row1-baseline-lc-pupil
Trial-by-trial baseline LC firing rate is positively correlated with baseline pupil diameter within sessions, after correcting both for linear drift, in both monkeys.
trial-by-trial measurements of baseline LC activity were positively correlated with simultaneously recorded measurements of baseline pupil diameter over the course of a session
Results, paragraph 2; Figure 2, row 1
How it will be judged
Per unit, the partial Spearman correlation of row 1 (opening comment). Reproduced when, in each monkey, the median coefficient is above zero and the population test gives p < 0.05 under T1; partially reproduced when this holds in one monkey only. The p-values are reported beside the paper's.
fig2-row2-evoked-lc-pupil
Baseline-subtracted evoked LC responses are positively correlated with baseline-subtracted evoked pupil responses within units, in both monkeys.
Likewise, baseline-subtracted evoked LC responses were positively correlated with baseline-subtracted evoked pupil responses (p = 0.0135 for monkey Oz, p = 0.0018 for monkey Ci)
Results, paragraph 2; Figure 2, row 2
How it will be judged
Per unit, the Spearman correlation over beep trials of evoked firing rate with evoked pupil, under E1 and under E2. Reproduced when, in each monkey, the median coefficient is above zero and the population test (T1) gives p < 0.05, under E1 and under E2; partially reproduced when this holds in one monkey only. When E1 and E2 give different outcomes the claim is blocked as methods_underspecified.
fig2-row3-pupil-evoked-vs-baseline
Baseline-subtracted evoked pupil responses are negatively correlated with baseline pupil diameter (drift residuals) within sessions, in both monkeys.
baseline-subtracted evoked pupil responses were negatively correlated with baseline pupil diameter (using residuals; p = 1.16×10-6 for monkey Oz, p = 5.02×10-9 for monkey Ci; Figure 2, row 3)
Results, paragraph 3; Figure 2, row 3
How it will be judged
Per session, the Spearman correlation over beep trials (each trial once) of evoked pupil with the baseline pupil residual, under E1 and under E2. Reproduced when, in each monkey, the median coefficient is below zero and the population test (T1) gives p < 0.05, under E1 and under E2; partially reproduced when this holds in one monkey only. When E1 and E2 give different outcomes the claim is blocked as methods_underspecified. Sessions are the paper's 51 for Ci; the code's 50 (legacy removal, opening comment) is shown as a variant.
fig2-row4-lc-evoked-vs-baseline
Baseline-subtracted evoked LC responses are negatively correlated with baseline LC activity (drift residuals) within units, in both monkeys.
Similarly, baseline-subtracted evoked LC responses were negatively correlated with baseline LC activity (using residuals; p = 5.88×10-7 for monkey Oz, p = 1.04×10-11 for monkey Ci; Figure 2, row 4)
Results, paragraph 3; Figure 2, row 4
How it will be judged
Per unit, the Spearman correlation over beep trials of evoked firing rate with the baseline firing-rate residual. Reproduced when, in each monkey, the median coefficient is below zero and the population test gives p < 0.05 under T1; partially reproduced when this holds in one monkey only.
fig2-row5-baseline-lc-vs-evoked-pupil
Evoked pupil responses are not reliably related to baseline LC activity; the median per-unit correlation does not differ from zero in either monkey.
evoked pupil responses were rarely positively or negatively related to baseline LC activity in a reliable manner (Wilcoxon rank-sum test for H0... p = 0.894 for monkey Oz, p = 0.904 for monkey Ci)
Results, paragraph 4; Figure 2, row 5
How it will be judged
A claim of absence. Per unit, the Spearman correlation over beep trials of evoked pupil with the baseline firing-rate residual. The paper's test is the population test at 0.05 per monkey, T1 (opening comment). The evoked pupil has two readings (E1, E2), both run. Reproduced when the test gives p >= 0.05 in both monkeys; not reproduced when it gives p < 0.05 in either. When E1 and E2 give different outcomes the claim is blocked as methods_underspecified. The median coefficient per monkey is reported either way, and T2 and the code's evoked pupil are shown as variants.
fig2-row6-baseline-pupil-vs-evoked-lc
Baseline pupil diameter is not reliably related to evoked LC responses across monkeys; the median per-unit correlation does not differ from zero in Oz (p = 0.915) and does in Ci (p = 0.008).
Likewise, baseline pupil diameter was rarely positively or negatively related to evoked LC responses reliably across monkeys (Figure 2, bottom row) (p = 0.915 for monkey Oz, p = 0.008 for monkey Ci)
Results, paragraph 4; Figure 2, row 6
How it will be judged
Per unit, the Spearman correlation over beep trials of evoked firing rate with the baseline pupil residual. Population test per monkey at 0.05, T1 as main, T2 as variant (as in fig2-row5-baseline-lc-vs-evoked-pupil). The claim is that the relationship is not reliable across monkeys. Reproduced when Oz gives p >= 0.05 and Ci gives p < 0.05, the paper's pattern; partially reproduced when neither monkey reaches p < 0.05 (still no relationship across monkeys, but Ci's is not found); not reproduced when both monkeys give p < 0.05 with medians of the same sign; partially reproduced in the other cases (Oz p < 0.05 with Ci p >= 0.05, or both p < 0.05 with medians of opposite sign), where no relationship holds across monkeys but the paper's pattern is not found. Medians and p-values are reported.
cross-epoch-no-reliable-relation
Despite consistent within-epoch relationships between LC activity and pupil size, there are no reliable relationships between the two across epochs (baseline of one against evoked of the other).
Thus, despite consistent relationships between LC activity and pupil size within each epoch, there were no reliable relationships between the two when comparing across epochs
Results, paragraph 4, last sentence; Discussion, paragraph 3
How it will be judged
Reproduced when rows 1 and 2 are reproduced (within-epoch relations in both monkeys) and neither cross-epoch row (rows 5 and 6) shows p < 0.05 with medians of the same sign in both monkeys. Not reproduced when either cross-epoch row does; partially reproduced when the cross-epoch rows hold but a within-epoch row holds in one monkey only.
range-baseline-pupil-evoked-lc
The correlation between baseline pupil and evoked LC responses does not depend on the range of evoked LC responses in either monkey.
We found that the relationship between baseline pupil diameter and evoked LC responses was not systematically impacted by the range of evoked LC responses for either monkey (Spearman’s rho = 0.02, p = 0.89 for monkey Oz; rho = 0.07, p = 0.589 for monkey Ci)
Results, paragraph 5 (first additional analysis)
How it will be judged
A claim of absence. Across units, per monkey, the Spearman correlation between each unit's row 6 coefficient and the range (maximum minus minimum) of its evoked firing rates over beep trials, as in compute_manuscript_stats.m (build_evoked_range_by_unit). The paper says the range is per session; the code takes it per unit, which is followed. Reproduced when p >= 0.05 in both monkeys; not reproduced when p < 0.05 in either. Rho is reported.
range-baseline-lc-evoked-pupil
The correlation between baseline LC activity and evoked pupil responses does not depend on the range of evoked pupil responses in Ci (rho = 0.18, p = 0.16), but in Oz it becomes more negative as that range grows (rho = -0.33, p = 0.03).
the relationship between baseline LC activity and evoked pupil responses was not systematically impacted by the range of evoked pupil responses for monkey Ci (rho = 0.18, p = 0.16). However, we did observe a modest impact for monkey Oz (rho = -0.33, p = 0.03)
Results, paragraph 5 (first additional analysis)
How it will be judged
Across units, per monkey, the Spearman correlation between each unit's row 5 coefficient and the range (maximum minus minimum) of its evoked pupil responses over beep trials, as in compute_manuscript_stats.m, under E1 and E2. Reproduced when, under both, Ci gives p >= 0.05 and Oz gives p < 0.05 with rho within tolerance of -0.33; partially reproduced when Ci gives p >= 0.05 and Oz gives p < 0.05 with a negative rho outside the tolerance; not reproduced otherwise. When E1 and E2 give different outcomes the claim is blocked as methods_underspecified.
figs1-raw-baselines-similar
Using raw baseline measures in place of drift residuals gives results similar to those of Figure 2.
we obtained similar results when using raw baseline measurements as opposed to residuals (see Supplemental Figure 1)
Results, paragraph 6 (second additional analysis); Supplemental Figure 1
How it will be judged
The six Figure 2 coefficients recomputed with raw baseline pupil and raw baseline firing rate in place of their residuals, as the code's "baseline" mode does (PaperFigures/plot_S1_summary.m; row 1 stays the partial Spearman controlling for trial time), and the population test per monkey. "Similar" is judged on the paper's own Figure 2 pattern, the sign of the median and whether p < 0.05, for 12 row and monkey cells. Reproduced when all 12 match the pattern the paper reports with residuals (rows 1 to 4 significant in the reported direction in both monkeys, row 5 not significant in either, row 6 not significant in Oz and significant in Ci); partially reproduced when 10 or 11 match; not reproduced below 10. The population test is T1 (T2 shown as a variant). Run under E1 and under E2; when they give different outcomes the claim is blocked as methods_underspecified.
figs2-llr-pupil-baseline-evoked
The relationship between baseline and evoked pupil responses is better described by a quadratic than a linear fit in only a minority of sessions (7 of 32 for Oz, 6 of 51 for Ci).
The relationship between baseline and evoked pupil responses was better described by a non-linear fit for a minority of sessions (LLR > critical value of 3.841... 7/32 sessions for monkey Oz, 6/51 sessions for monkey Ci)
Results, paragraph 7 (third additional analysis); Supplemental Figure 2, first panel
How it will be judged
Per session, beep trials each once, evoked pupil regressed on the baseline pupil residual with a linear and a quadratic model fit by maximum likelihood (the code's fitlme with formulas y ~ x and y ~ x*x, compute_manuscript_stats.m); LLR = 2 times the difference in log-likelihood, counted above 3.841. Sessions with fewer than 4 valid trials or 3 distinct predictor values are left out, as in the code. Run under E1 and E2, and with all sessions (the paper's 51 for Ci) or the code's 50 Ci sessions (legacy removal, opening comment; compute_manuscript_stats.m prints 6/50). Tolerance of one session, as 10% of these counts is under one session and would demand an exact count. Reproduced when both counts are within tolerance and below half the sessions under one reading; partially reproduced when the counts are outside tolerance but below half the sessions.
figs2-llr-lc-baseline-evoked
LC evoked responses are rarely better described by a quadratic than a linear function of baseline LC activity (2 of 42 units for Oz, 3 of 63 for Ci).
LC neuronal responses were rarely better characterized using a non-linear fit (2/42 neurons for monkey Oz, 3/63 for monkey Ci)
Results, paragraph 7; Supplemental Figure 2, second panel
How it will be judged
Per unit, beep trials, evoked firing rate regressed on the baseline firing-rate residual, linear against quadratic as in figs2-llr-pupil-baseline-evoked (prepare_figure_analysis_tables.m, compute_llr). The code drops rows 43 and 65 of its unit table before counting (a "legacy" removal, leaving 61 Ci units); the paper's denominator is 63. Readings, all units (the paper's denominator) and the code's removal. Tolerance of one unit, for the reason in figs2-llr-pupil-baseline-evoked. Reproduced when both counts are within tolerance under one reading; partially reproduced when they are outside tolerance but below a tenth of the units.
figs2-llr-cross-epoch
Across-epoch relationships are rarely better described by a quadratic than a linear fit (baseline LC against evoked pupil, 4 of 42 units for Oz and 4 of 63 for Ci; baseline pupil against evoked LC, 2 of 42 and 5 of 63), so the null cross-epoch findings are not due to a non-linear relationship.
across-epoch comparisons showed similar trends that favored a linear relationship (baseline LC versus evoked pupil... 4/42 for monkey Oz, 4/63 for monkey Ci; baseline pupil versus evoked LC... 2/42 for monkey Oz, 5/63 for monkey Ci)
Results, paragraph 7; Supplemental Figure 2, third and fourth panels
How it will be judged
Per unit, beep trials, evoked pupil regressed on the baseline firing-rate residual, and evoked firing rate on the baseline pupil residual, linear against quadratic as in figs2-llr-pupil-baseline-evoked, under E1 and E2, and with all units or the code's removal of two (figs2-llr-lc-baseline-evoked). Tolerance of one unit, for the reason given there. Reproduced when all four counts are within tolerance under one reading; partially reproduced when they are outside tolerance but each below a tenth of the units.
fig3-pooled-within-epoch
With data z-scored per session or unit and pooled across trials, sessions and units for each monkey, the within-epoch relationships of Figure 2 hold (baseline LC with baseline pupil and evoked LC with evoked pupil positive; evoked against baseline negative for pupil and for LC).
we applied our single-unit analyses (Figure 2) to data pooled across all trials, sessions, and units (Figure 3). The results are qualitatively similar for the pooled and single-unit analyses
Results, paragraph 8 (fourth additional analysis); Figure 3, columns 1 to 4
How it will be judged
As in compute_manuscript_stats.m (Figure 3 block). Firing rates z-scored per unit over its baseline, raw evoked and fixation-only rates (zscore_tables.m), evoked rate as z-scored raw evoked minus z-scored baseline; pupil as stored, which the code uses. The Methods and the Figure 3 legend z-score pupil per session; when the stored trace is not so z-scored, the pupil samples are z-scored per session as the Methods say (main reading) and the stored values are shown as the code's variant. Drift residuals recomputed. Per monkey, a Spearman correlation over pooled trials for each column, beep and fixation-only trials pooled for column 1, beep trials each once per session for column 3. The Figure 3 legend draws linear regression lines; the code's test, Spearman, is followed. Reproduced when all four columns give p < 0.05 in the sign of the corresponding Figure 2 row in both monkeys, under E1 and E2; partially reproduced when six or seven of the eight column and monkey cells do. The code's raw-baseline version (PaperFigures/plot_S1_combined.m) is shown as a variant.
fig3-pooled-cross-epoch-null
In the pooled data, neither baseline pupil against evoked LC nor baseline LC against evoked pupil shows a reliable relationship across monkeys, also when units are split by the sign of their significant baseline LC-pupil relationship.
in neither case showing a reliable relationship between baseline pupil and evoked LC responses or baseline LC activity and evoked pupil responses across monkeys. This result also held when LC units were separated based on the direction of their baseline relationship with pupil size
Results, paragraph 8; Figure 3, columns 5 and 6
How it will be judged
A claim of absence. Pooled data as in fig3-pooled-within-epoch; Spearman correlation per monkey of evoked firing rate with the baseline pupil residual (column 6) and of the baseline firing-rate residual with evoked pupil (column 5), at 0.05, in all units and in the two groups of units whose row 1 partial Spearman correlation has p < 0.05 with positive or with negative sign (Figure 3 legend). The code's per-group pooled Spearman is in PaperFigures/plot_S1_combined.m (rows 3 and 4, raw baselines; groups from the z-scored stats table after the legacy removal of two units); the Figure 3 script itself is not in the repository (commented out in make_all_figures.m), so the residual version follows that code with residuals in place of raw baselines, as the Figure 3 block of compute_manuscript_stats.m uses them. "Across monkeys" is read, as in fig2-row6-baseline-pupil-vs-evoked-lc, as significant with the same sign in both monkeys. Reproduced when no column in no group is so; not reproduced when any is. Rho and p for every cell are reported, under E1 and E2.

Pipeline

Paper: done (1 of 1)
Claims: done
Claims check: done (ready)
Attempt: todo
Questions: none
Verified: none