{ "dataset": { "archive": "dandi", "id": "000576", "label": "DANDI:000576" }, "page": "datasets/dandi/000576/", "generated": "2026-11-05T07:41:23+00:00", "caution": "A failure to reproduce is not a finding that the paper is wrong. It may be a failure of the attempt.", "outcome_counts": { "reproduced": 0, "partially_reproduced": 3, "not_reproduced": 12, "blocked": 3 }, "outcome_counts_with_outside_data": { "reproduced": 0, "partially_reproduced": 1, "not_reproduced": 0, "blocked": 6 }, "outcome_counts_elsewhere": { "reproduced": 0, "partially_reproduced": 0, "not_reproduced": 0, "blocked": 0 }, "outcome_counts_links": { "reproduced": 0, "partially_reproduced": 0, "not_reproduced": 0, "blocked": 0 }, "attempts": [ { "id": "000576-fedele-2020-01", "dataset": { "archive": "dandi", "id": "000576", "version": "0.231010.1811", "doi": "10.48324/dandi.000576/0.231010.1811" }, "key": "dandi:000576", "label": "DANDI:000576", "dataset_version": "0.231010.1811", "paper": "10.1016/j.neuroimage.2020.116705", "paper_title": "The relation between neuronal firing, local field potentials and hemodynamic activity in the human amygdala in response to aversive dynamic visual stimuli", "date": "2026-10-30", "kind": "agent", "accountable": "CatalystNeuro", "agent": { "harness": "Claude Code", "model": "claude-opus-5-5", "protocol": "pilot-12" }, "runner": { "platform": "GitHub Actions", "url": "https://github.com/catalystneuro/encore/actions/runs/41128517946" }, "tier": "agent_verified", "deprecated": "", "deprecated_on": null, "entry": "000576-fedele-2020", "entry_title": "iEEG, neurons and BOLD in the human amygdala during aversive videos (Fedele, Tzovara et al. 2020)", "description": "Tests the findings of Fedele, Tzovara et al. (2020) on dandiset 000576, which holds the bipolar amygdala iEEG of 9 patients (14 amygdalae) and the spike times of 42 units from 5 of them during a block design of aversive and neutral videos. It tests the delta and high gamma responses, the modulation indices and their relation to the seizure onset zone, and the firing of the units. The BOLD time courses and the continuous microwire signals are not in the dandiset.", "notebooks": [ { "path": "reproduce_fedele_2020.ipynb", "title": "Reproducing Fedele, Tzovara et al. (2020) from dandiset 000576", "executed": "notebooks/000576-fedele-2020/reproduce_fedele_2020.html" } ], "outcomes": [ { "claim": "fig1b-delta-decrease", "statement": "In amygdalae outside the SOZ, delta power (1-4 Hz) is lower during AVERSIVE than during NEUTRAL sequences, and the reduction lasts through the AVERSIVE sequence.", "source": "Results 2.1, paragraph 1; Figure 1B", "quote": "In nSOZ amygdalae, DELTA band power was consistently reduced throughout the duration of the AVERSIVE sequence (Fig. 1B)", "statistics": [ { "test": "linear mixed model per time point, cluster-based permutation correction (Fig. 1 legend calls it cluster analysis)", "p": 0.05, "p_relation": "<" } ], "kind": "finding", "inferred": "", "outcome": "blocked", "blocked_by": "methods_underspecified", "notebook": "reproduce_fedele_2020.ipynb", "notes": "The criterion blocks the claim when the readings disagree, and they do; it holds in 6 of the 12 mixed-model readings of the claims file. With the z-score over the whole 26 s trial no reading finds a delta cluster (smallest cluster p 0.88 in all 14 amygdalae, 1.0 in the 7 nSOZ). With z-scoring against the 2 s baseline, AVERSIVE is below NEUTRAL with significant clusters covering 14.3 s of the 24 s block in the nSOZ amygdalae and 15.6 s in all 14. With z-scoring over all trials of the amygdala, 17.1 s in all 14 and none in the nSOZ amygdalae (smallest p 0.072). The block mean of AVERSIVE minus NEUTRAL is negative under every reading (-0.03 to -1.39). The variants agree with the baseline reading (baselinepool 13.3 and 16.8 s, trialsd 7.0 and 16.1 s). The unpaired t-test variant gives the same pattern. The question for the authors is in the findings (methods_unavailable, z-scoring).", "deviations": [ "The paper's mixed model Power ~ condition + (1|Patients) is computed in its within-patient form (patient fixed effects), which gives the same condition estimate for this balanced design (8 AVERSIVE and 9 NEUTRAL trials per amygdala); the notebook compares it with statsmodels' MixedLM at four time points (t -2.125 against -2.123 and so on).", "Labels are permuted within each patient, the same permutation for both amygdalae of a patient (Methods 4.11, \"permuted the labels of neutral and aversive trials across all participants\", 500 permutations); the unpaired t-test variant permutes over all trials.", "The test is run over the 24 s block (0 to 24 s from onset, Fig. 1 \"locked to stimulus onset\"); the baseline clip is left out. A choice, stated in the notebook.", "Outcomes are judged on the three z-scorings of the claims file (trial, baseline, amygdala). Two variants are shown and enter no count of readings, trialsd (the baseline-corrected envelope divided by its SD in the trial) and baselinepool (the epoch's baseline mean and the SD of the baseline clips pooled over the amygdala's epochs, a stable form of baseline z-scoring). The two other baselines of Methods 4.8 (mean of all epochs, previous epoch) are also run as variants.", "Envelopes are filtered per 26 s epoch with 2 s mirror padding, as the epochs are not contiguous in the files, and kept at 10 Hz after the 1 s smoothing (standard choices)." ], "assumed": false, "disagreed": false, "data": "dataset", "where": null, "proposed": null }, { "claim": "fig1c-hgp-increase-first-2s", "statement": "High gamma power (above 60 Hz) is higher during AVERSIVE than during NEUTRAL sequences in the first 2 s of the sequence.", "source": "Results 2.1, paragraph 1; Figure 1C", "quote": "The HGP was enhanced in response to aversive stimuli during the first 2 s of the AVERSIVE sequence presentation (Fig. 1C, unpaired t-tests, corrected for multiple comparisons using random permutations, p < 0.05)", "statistics": [ { "test": "unpaired t-tests, corrected for multiple comparisons using random permutations (Results); linear mixed model with cluster-based permutation (Methods 4.11)", "p": 0.05, "p_relation": "<" } ], "kind": "finding", "inferred": "", "outcome": "not_reproduced", "blocked_by": null, "notebook": "reproduce_fedele_2020.ipynb", "notes": "None of the 12 mixed-model readings and none of the 12 unpaired t-test readings of the claims file finds a significant HGP cluster, nor do the two z-scoring variants. The largest mixed-model t in the first 2 s is 3.01 at 0.95 s after onset (baseline z-scoring, nSOZ amygdalae), above the threshold at 13 time points, and its cluster has permutation p 0.062; the smallest cluster p of all mixed-model readings is 0.056. With the HGP band limited to 60-150 or 60-250 Hz (variants of the open upper edge) the only significant HGP clusters are AVERSIVE above NEUTRAL from about 6 to 8 s (p 0.020 and 0.042) and, with baselinepool, 3.1 to 4.7 s, not in the first 2 s. Among the baselines of Methods 4.8 (variants), the previous epoch as baseline with z-scoring over the amygdala gives an HGP cluster from 0.6 to 5.5 s in the nSOZ amygdalae (p 0.002); as blocks alternate, that baseline is always a block of the other condition. With the trialsd variant two more HGP clusters begin in the first 2 s, both in the nSOZ amygdalae, 0.6 to 1.9 s with the mean of all epochs as baseline (p 0.028) and 0.5 to 8.4 s with the previous epoch (p 0.002). These three are the only HGP clusters that begin in the first 2 s, and all use a baseline other than the 1 s clip. Per-amygdala time courses are in the notebook.", "deviations": [ "As fig1b-delta-decrease. HGP is a 60 Hz high-pass (the claims file's reading); two band-pass variants are shown and do not decide the outcome." ], "assumed": false, "disagreed": false, "data": "dataset", "where": null, "proposed": null }, { "claim": "other-bands-no-effect", "statement": "No AVERSIVE against NEUTRAL difference is found in the theta, alpha, beta and gamma (30-60 Hz) bands.", "source": "Results 2.1, paragraph 1; bands in Methods 4.8", "quote": "Since no effect was observed in the other spectral bands (see Methods 7.8), in the following we focus on DELTA power and HGP", "statistics": [], "kind": "finding", "inferred": "", "outcome": "not_reproduced", "blocked_by": null, "notebook": "reproduce_fedele_2020.ipynb", "notes": "The readings of the claims file disagree. Under the z-score over the whole trial no band has a cluster, nor has delta. Under baseline z-scoring the mixed model finds clusters with AVERSIVE below NEUTRAL in alpha in the 7 nSOZ amygdalae (13.3 to 15.9 s, p 0.016) and in theta and alpha in all 14 (p 0.002 to 0.028); under z-scoring over the amygdala all four bands have clusters in all 14 and none in the nSOZ amygdalae. Every reading that shows the delta effect of Fig. 1B (6 of 12) shows a cluster in another band. Of the variants, baselinepool in the nSOZ amygdalae shows the delta effect and no other-band cluster; in all 14 it has theta and alpha clusters.", "deviations": [ "Assumed that the paper's pipeline is one under which its own delta result (Fig. 1B) appears, as the paper reports the delta effect and the absence in other bands from the same analysis (Results 2.1, paragraph 1). The outcome is judged on the readings of the claims file that show the delta effect; the trial reading shows no cluster in any band, delta included. The first revision adopted baseline z-scoring alone; the verifier found that the author's own choice, and it is no longer used.", "Outcomes are judged on the three z-scorings of the claims file (trial, baseline, amygdala). Two variants are shown and enter no count of readings, trialsd (the baseline-corrected envelope divided by its SD in the trial) and baselinepool (the epoch's baseline mean and the SD of the baseline clips pooled over the amygdala's epochs, a stable form of baseline z-scoring). The two other baselines of Methods 4.8 (mean of all epochs, previous epoch) are also run as variants.", "As fig1b-delta-decrease for the test." ], "assumed": true, "disagreed": false, "data": "dataset", "where": null, "proposed": null }, { "claim": "table2-mi-values", "statement": "The modulation indices of DELTA and HGP for each of the 14 amygdalae are as printed in Table 2.", "source": "Table 2, columns Modulation index (MI) of iEEG, DELTA and HGP", "quote": "", "statistics": [], "kind": "finding", "inferred": "Values read from Table 2, which no sentence states one by one. Minus signs are garbled in the extracted text; the signs used here agree with the counts the Results give.", "outcome": "not_reproduced", "blocked_by": null, "notebook": "reproduce_fedele_2020.ipynb", "notes": "No reading gives all 28 values within tolerance (at best 4 of 28 among the claims file's z-scorings, z-scoring over the amygdala with fixed windows; 6 with the trialsd variant) or all 28 signs (at best 23 of 28). The measured values are those of the first reading the criterion lists that defines all 28, z-scoring over the trial with fixed windows (1 of 28 within tolerance, 16 signs). The group cluster window defines no HGP MI, as there is no HGP group cluster; the per-amygdala clusters define 1 to 3 MIs per band. Re-mapping subjects 2 and 7, or 7 to the paper's 5 and 8, does not bring the values closer under the claims file's z-scorings (0 to 4 within tolerance); the trialsd variant goes from 6 to 5 and 7.", "deviations": [ "Window readings (a) group clusters and (b) per-amygdala clusters are the claims file's; (c) fixed windows, 0 to 2 s for HGP and 0 to 24 s for delta, is taken from the Results' text (\"occurred at stimulus onset in HGP and was sustained throughout the trial in DELTA\", Results 2.1 paragraph 2; \"HGP enhancement in the first 2 s or sustained delta decrease\", Methods 4.11) and counted as a reading.", "Where a window reading finds no significant window for an amygdala, its MI is undefined (the first revision set it to 0); a claim is judged on the readings that define every MI it needs.", "Outcomes are judged on the three z-scorings of the claims file (trial, baseline, amygdala). Two variants are shown and enter no count of readings, trialsd (the baseline-corrected envelope divided by its SD in the trial) and baselinepool (the epoch's baseline mean and the SD of the baseline clips pooled over the amygdala's epochs, a stable form of baseline z-scoring). The two other baselines of Methods 4.8 (mean of all epochs, previous epoch) are also run as variants." ], "assumed": false, "disagreed": false, "data": "dataset", "where": null, "proposed": null }, { "claim": "fig1d-delta-hgp-mi-correlation", "statement": "Across the 14 amygdalae, the DELTA modulation index and the HGP modulation index are negatively correlated.", "source": "Results 2.1, paragraph 2; Figure 1D", "quote": "MI and HGP-MI were negatively correlated across all amygdalae", "statistics": [ { "test": "Spearman correlation", "statistic": "r = -0.67", "p": 0.01, "n": "14 amygdalae" } ], "kind": "finding", "inferred": "", "outcome": "not_reproduced", "blocked_by": null, "notebook": "reproduce_fedele_2020.ipynb", "notes": "Only window (c) defines both MIs. Under it r = -0.30 (p = 0.30), -0.12 (p = 0.69) and -0.12 (p = 0.68) for z-scoring over the trial, against the baseline and over the amygdala; the variants give -0.33 and -0.04. None is within tolerance or has p < 0.05. The printed Table 2 values give r = -0.74 (p = 0.003).", "deviations": [ "MI readings as in table2-mi-values; z-scorings as in fig1b-delta-decrease.", "Assumption. Window (c), fixed windows of 0 to 2 s for HGP and 0 to 24 s for delta, is the only window reading that defines the MIs this claim needs, so the outcome rests on it. It is taken from the Results' text (Results 2.1, Methods 4.11, \"HGP enhancement in the first 2 s\"), while the paper defines the MI over \"the interval where the two curves are significantly separated\" (Table 2 legend), which these data give for no HGP MI. The question is for the authors (methods_unavailable, MI window)." ], "assumed": true, "disagreed": false, "data": "dataset", "where": null, "proposed": null }, { "claim": "nsoz-delta-hgp-mi-inverse", "statement": "Among the 7 amygdalae outside the SOZ, the stronger the HGP enhancement (HGP MI), the stronger the reduction of DELTA power (more negative DELTA MI).", "source": "Results 2.1, paragraph 2; Table 2", "quote": "Specifically for amygdalae not included in the SOZ, we observed that the stronger the HGP enhancement, the stronger the reduction in DELTA power (Table 2)", "statistics": [], "kind": "finding", "inferred": "", "outcome": "not_reproduced", "blocked_by": null, "notebook": "reproduce_fedele_2020.ipynb", "notes": "Over the 7 nSOZ amygdalae, window (c) gives r = -0.57 (trial), -0.14 (baseline) and +0.07 (amygdala), negative in 2 of the 3 readings that define the MIs, so not under every reading. Windows (a) and (b) define no HGP MI for these amygdalae. The variants give -0.07 and +0.21. The printed Table 2 values give r = -0.68 (p = 0.094).", "deviations": [ "MI readings as in table2-mi-values; z-scorings as in fig1b-delta-decrease.", "Assumption. Window (c), fixed windows of 0 to 2 s for HGP and 0 to 24 s for delta, is the only window reading that defines the MIs this claim needs, so the outcome rests on it. It is taken from the Results' text (Results 2.1, Methods 4.11, \"HGP enhancement in the first 2 s\"), while the paper defines the MI over \"the interval where the two curves are significantly separated\" (Table 2 legend), which these data give for no HGP MI. The question is for the authors (methods_unavailable, MI window)." ], "assumed": true, "disagreed": true, "data": "dataset", "where": null, "proposed": null }, { "claim": "table2-delta-reduced-12-of-14", "statement": "DELTA power is reduced (negative MI) in 12 of the 14 amygdalae, whether inside or outside the SOZ.", "source": "Results 2.1, paragraph 3; Table 2", "quote": "DELTA power was reduced in 12 of 14 amygdalae (negative MI in Table 2), irrespective of whether the amygdalae were within the SOZ or not", "statistics": [], "kind": "finding", "inferred": "", "outcome": "partially_reproduced", "blocked_by": null, "notebook": "reproduce_fedele_2020.ipynb", "notes": "Of the 6 readings that define all 14 delta MIs, 5 give 11 or 12 negative with negative MIs inside and outside the SOZ (baseline z-scoring with both group windows and the fixed window, 11, positive in amygdalae 2, 10 and 11; z-scoring over the amygdala with the all-14 group window and the fixed window, 12, positive in 2 and 10). Z-scoring over the trial with the fixed window, the first reading the criterion lists that defines them and given as measured, gives 10 (positive in 2, 9, 10 and 11), so the count is not in range under every reading. Table 2 has amygdalae 2 and 4 positive. The per-amygdala window defines 1 or 2 delta MIs. The variants give 10 to 13.", "deviations": [ "MI readings as in table2-mi-values; z-scorings as in fig1b-delta-decrease.", "Assumption. The only reading that fails uses window (c), fixed windows of 0 to 2 s for HGP and 0 to 24 s for delta, taken from the Results' text (Results 2.1, Methods 4.11) and not from the claims file. The claims file's window readings that define all 14 delta MIs (group clusters under baseline z-scoring in both sets and under z-scoring over the amygdala in all 14) give 11, 11 and 12, all in range. So the outcome rests on counting window (c) as a reading. The question is for the authors (methods_unavailable, MI window)." ], "assumed": true, "disagreed": true, "data": "dataset", "where": null, "proposed": null }, { "claim": "table2-hgp-mi-positive-nsoz", "statement": "All 7 amygdalae outside the SOZ have a positive HGP MI, while the response is absent in amygdalae within the SOZ (6 of 7 negative).", "source": "Results 2.1, paragraph 3; Table 2 and its legend", "quote": "All amygdalae outside SOZ showed a positive HGP-MI in response to AVERSIVE videos, while this response was absent in amygdalae within the SOZ (p 1\u20444 0.009)", "statistics": [ { "about": "HGP MI, nSOZ against SOZ amygdalae", "test": "not named", "p": 0.009, "n": "7 nSOZ and 7 SOZ amygdalae" } ], "kind": "finding", "inferred": "", "outcome": "not_reproduced", "blocked_by": null, "notebook": "reproduce_fedele_2020.ipynb", "notes": "Window (c) gives 4, 4 and 5 of 7 nSOZ amygdalae with a positive HGP MI and 3, 5 and 7 of 7 SOZ amygdalae with a negative one (trial, baseline, amygdala); no reading gives both counts. The measured values are those of the trial reading. Under z-scoring over the amygdala Fisher's test gives p = 0.021 but the nSOZ count is 5. Windows (a) and (b) define no HGP MI for all amygdalae.", "deviations": [ "MI readings as in table2-mi-values; z-scorings as in fig1b-delta-decrease.", "Assumption. Window (c), fixed windows of 0 to 2 s for HGP and 0 to 24 s for delta, is the only window reading that defines the MIs this claim needs, so the outcome rests on it. It is taken from the Results' text (Results 2.1, Methods 4.11, \"HGP enhancement in the first 2 s\"), while the paper defines the MI over \"the interval where the two curves are significantly separated\" (Table 2 legend), which these data give for no HGP MI. The question is for the authors (methods_unavailable, MI window)." ], "assumed": true, "disagreed": false, "data": "dataset", "where": null, "proposed": null }, { "claim": "fig1e-hgp-not-delta-distinguishes-soz", "statement": "The HGP MI, but not the DELTA MI, distinguishes amygdalae within the SOZ from those outside it.", "source": "Figure 1E legend", "quote": "HGP but not DELTA distinguishes between amygdalae within SOZ and outside SOZ (p 1\u20444 0.019, not corrected)", "statistics": [ { "about": "HGP MI, SOZ against nSOZ", "test": "not named", "p": 0.019, "corrected": "not corrected", "n": "7 nSOZ and 7 SOZ amygdalae" } ], "kind": "finding", "inferred": "", "outcome": "not_reproduced", "blocked_by": null, "notebook": "reproduce_fedele_2020.ipynb", "notes": "Under no reading that defines both MIs (window c) and neither test does the HGP MI differ between nSOZ and SOZ amygdalae at p < 0.05 (smallest p 0.084, baseline z-scoring, t-test); the DELTA MI does not differ either (p 0.08 to 0.95). With the baselinepool variant the rank-sum test gives HGP p = 0.035 and DELTA p = 0.48. Run on the printed Table 2 values the same tests give HGP p = 0.004 and 0.002 but also DELTA p = 0.035 and 0.013; asked in questions/000576.yaml.", "deviations": [ "MI readings as in table2-mi-values; z-scorings as in fig1b-delta-decrease.", "Assumption. Window (c), fixed windows of 0 to 2 s for HGP and 0 to 24 s for delta, is the only window reading that defines the MIs this claim needs, so the outcome rests on it. It is taken from the Results' text (Results 2.1, Methods 4.11, \"HGP enhancement in the first 2 s\"), while the paper defines the MI over \"the interval where the two curves are significantly separated\" (Table 2 legend), which these data give for no HGP MI. The question is for the authors (methods_unavailable, MI window)." ], "assumed": true, "disagreed": false, "data": "dataset", "where": null, "proposed": null }, { "claim": "hgp-soz-sensitivity-specificity", "statement": "HGP enhancement indicates amygdalar functional integrity (outside the SOZ) with sensitivity 7/7 (100%), specificity 6/7 (86%) and accuracy 13/14 (93%).", "source": "Results 2.1, paragraph 3; Methods 4.11, paragraph 4", "quote": "With sensitivity (7/7 1\u20444 100% CI [66\u2013100]%), high specificity (6/7 1\u20444 86% CI [42 99]%) and accuracy 93% (13/14, CI [66\u201399]%), HGP enhancement indicated amygdalar functional integrity", "statistics": [ { "about": "sensitivity, 7 of 7", "ci": { "level": 0.95, "low": 66, "high": 100, "unit": "%" } }, { "about": "specificity, 6 of 7", "ci": { "level": 0.95, "low": 42, "high": 99, "unit": "%" } }, { "about": "accuracy, 13 of 14", "ci": { "level": 0.95, "low": 66, "high": 99, "unit": "%" } } ], "kind": "finding", "inferred": "", "outcome": "not_reproduced", "blocked_by": null, "notebook": "reproduce_fedele_2020.ipynb", "notes": "Window (c) gives sensitivity 4/7, 4/7 and 5/7, specificity 3/7, 5/7 and 7/7 and accuracy 7/14, 9/14 and 12/14 (trial, baseline, amygdala; the trial reading given as measured); no reading gives 7/7, 6/7 and 13/14. The Clopper-Pearson 95% intervals of the paper's counts are 59-100, 42-100 and 66-100%, against the printed 66-100, 42-99 and 66-99%; no single interval method run (one-sided exact, Wilson, Jeffreys) gives all three printed intervals. Asked in questions/000576.yaml.", "deviations": [ "MI readings as in table2-mi-values; z-scorings as in fig1b-delta-decrease.", "Assumption. Window (c), fixed windows of 0 to 2 s for HGP and 0 to 24 s for delta, is the only window reading that defines the MIs this claim needs, so the outcome rests on it. It is taken from the Results' text (Results 2.1, Methods 4.11, \"HGP enhancement in the first 2 s\"), while the paper defines the MI over \"the interval where the two curves are significantly separated\" (Table 2 legend), which these data give for no HGP MI. The question is for the authors (methods_unavailable, MI window)." ], "assumed": true, "disagreed": false, "data": "dataset", "where": null, "proposed": null }, { "claim": "fig1f-hgp-more-specific-than-delta", "statement": "The HGP response is more specific than the DELTA response for indicating amygdalar dysfunction within the SOZ, by a chi-square test that is not significant (p about 0.2).", "source": "Figure 1F legend", "quote": "The HGP response exceeds the DELTA response in specificity when indicating the amygdala\u2019s dysfunction within SOZ (p~0.2 HGP 5/5 DELTA 1/5, chi2-test)", "statistics": [ { "test": "chi-square test", "p": 0.2, "n": "HGP 5/5, DELTA 1/5 (as printed)" } ], "kind": "finding", "inferred": "", "outcome": "blocked", "blocked_by": "methods_underspecified", "notebook": "reproduce_fedele_2020.ipynb", "notes": "The readings disagree on which response is more specific, which the criterion blocks. Window (c), the only one that defines the MIs of all 7 SOZ amygdalae, gives HGP more specific under baseline z-scoring (5/7 against 2/7, chi-square p = 0.11) and z-scoring over the amygdala (7/7 against 1/7, p = 0.001), and a tie under z-scoring over the trial (3/7 each). The variants give HGP (6/7 against 1/7 and 3/7). Reading (b), the printed counts 5/5 against 1/5, gives HGP more specific with p = 0.0098 (0.053 with Yates' correction), not about 0.2; no reading gives a p near 0.2.", "deviations": [ "MI readings as in table2-mi-values; z-scorings as in fig1b-delta-decrease.", "Assumption. Window (c), fixed windows of 0 to 2 s for HGP and 0 to 24 s for delta, is the only window reading that defines the MIs this claim needs, so the outcome rests on it. It is taken from the Results' text (Results 2.1, Methods 4.11, \"HGP enhancement in the first 2 s\"), while the paper defines the MI over \"the interval where the two curves are significantly separated\" (Table 2 legend), which these data give for no HGP MI. The question is for the authors (methods_unavailable, MI window)." ], "assumed": false, "disagreed": true, "data": "dataset", "where": null, "proposed": null }, { "claim": "fig2-participant4-bold-correlations", "statement": "In the left amygdala of Participant 4, the BOLD time course is negatively correlated with DELTA power and positively correlated with HGP over the video sequence.", "source": "Results 2.2, paragraph 2; Figure 2A and 2C; Table 2, amygdala 6", "quote": "BOLD response and DELTA power were negatively correlated during presentation of video sequence (correlation of BOLDDELTA time-courses r 1\u20444 \ufffd0.48 p < 0.01), while BOLD and HGP were positively correlated (correlation BOLD-HGP Spearman\u2019s r 1\u20444 0.41 p < 0.01)", "statistics": [ { "about": "BOLD against DELTA", "test": "Spearman correlation", "statistic": "r = -0.48", "p": 0.01, "p_relation": "<", "corrected": "FDR (Table 2 legend)" }, { "about": "BOLD against HGP", "test": "Spearman correlation", "statistic": "r = 0.41", "p": 0.01, "p_relation": "<", "corrected": "FDR (Table 2 legend)" } ], "kind": "finding", "inferred": "", "outcome": "blocked", "blocked_by": "data_missing", "notebook": null, "notes": "Needs the BOLD time course of the amygdala from the clinical fMRI (Methods 4.6, 4.10). The paper names no deposit of it; the dandiset holds no fMRI, nor does the iEEG-BIDS version of the same data on OpenNeuro (ds003374, snapshot 1.1.1, which holds only ieeg folders), and the data descriptor lists only iEEG, spikes and electrode positions.", "deviations": [], "assumed": false, "disagreed": false, "data": "with_inputs", "where": null, "proposed": null }, { "claim": "table2-bold-correlations-group", "statement": "Across amygdalae with BOLD, DELTA power and BOLD are negatively correlated in 4 of 5 nSOZ amygdalae, while HGP and BOLD are positively correlated in 3 of 5 nSOZ and 1 of 4 SOZ amygdalae.", "source": "Results 2.2, paragraph 3; Table 2; Figure 2C", "quote": "DELTA power and BOLD responses showed a negative correlation in 4/5 nSOZ amygdalae, while the correlation of HGP and BOLD was positive in the 3/5 the nSOZ amygdalae and 1/4 SOZ amygdalae (Table 2, Fig. 2C)", "statistics": [], "kind": "finding", "inferred": "", "outcome": "blocked", "blocked_by": "data_missing", "notebook": null, "notes": "Needs the BOLD time courses, not public (see fig2-participant4-bold-correlations).", "deviations": [], "assumed": false, "disagreed": false, "data": "with_inputs", "where": null, "proposed": null }, { "claim": "table2-bold-correlation-values", "statement": "The Spearman correlations of BOLD with DELTA power and with HGP for each of the 9 amygdalae with BOLD are as printed in Table 2.", "source": "Table 2, columns Correlation BOLD vs. DELTA and BOLD vs. HGP, r", "quote": "", "statistics": [], "kind": "finding", "inferred": "Values read from Table 2, which no sentence states one by one. Minus signs are garbled in the extracted text (the same glyph as plus or minus); a table cell is read as a minus, and the signs agree with the counts of Results 2.2 (4/5, 3/5, 1/4).", "outcome": "blocked", "blocked_by": "data_missing", "notebook": null, "notes": "Needs the BOLD time courses, not public (see fig2-participant4-bold-correlations).", "deviations": [], "assumed": false, "disagreed": false, "data": "with_inputs", "where": null, "proposed": null }, { "claim": "bold-hgp-uncorrelated-low-mi", "statement": "The two nSOZ amygdalae without a significant HGP-BOLD correlation have a very low HGP MI.", "source": "Results 2.2, paragraph 3; Table 2 (amygdalae 10 and 12)", "quote": "the two nSOZ amygdalae that did not show a significant correlation between HGB and BOLD, had a remarkably low MI for HGP (MI < 0.05, Table 2)", "statistics": [], "kind": "finding", "inferred": "", "outcome": "blocked", "blocked_by": "data_missing", "notebook": null, "notes": "Needs the BOLD-HGP correlations, which need the BOLD time courses, not public (see fig2-participant4-bold-correlations).", "deviations": [], "assumed": false, "disagreed": false, "data": "with_inputs", "where": null, "proposed": null }, { "claim": "neuron-count", "statement": "Spike sorting gave 50 putative neuron clusters from 9 amygdalae, of which 41 had a firing rate above 1 Hz and were analysed.", "source": "Results 2.3, paragraph 1; Methods 4.9", "quote": "We identified 50 putative clusters of neurons (Niediek et al., 2016) from 9 amygdalae and included those with firing rate >1 Hz (N 1\u20444 41 neurons) in further analysis", "statistics": [], "kind": "finding", "inferred": "", "outcome": "partially_reproduced", "blocked_by": null, "notebook": "reproduce_fedele_2020.ipynb", "notes": "The dandiset holds 42 units, 42 above 1 Hz over the epochs each was recorded in (reading a, given as measured) and 40 over the whole 442 s (reading b); both within 4.1 of 41. The 50 putative clusters cannot be tested, as only units above about 1 Hz were deposited. Units are in 8 amygdalae, not 9, so the third expected value fails. The measured value is the first expected value only. The units come from 5 patients (1, 2, 3, 4, 7), which matches the abstract's 5 participants, but not the amygdalae of Table 1, which lists MUA for patients 5 and 8 and none for 7. Their mean rate over the observed epochs is 4.65 +- 3.91 Hz (median 3.13), against the data descriptor's 1.66 +- 2.65 Hz; 31 units are above 2 Hz.", "deviations": [ "Epochs in which a unit has no spike at all are taken as not recorded (16 units of subject 7 after epoch 11, 4 units of subject 2 in epoch 1), as their rates elsewhere are 1 to 16 Hz." ], "assumed": false, "disagreed": false, "data": "dataset", "where": null, "proposed": null }, { "claim": "fig3a-firing-increase-aversive", "statement": "All 41 neurons fire more during AVERSIVE than during NEUTRAL sequences (FRV 0.26 plus or minus 0.15).", "source": "Results 2.3, paragraph 2; Figure 3A", "quote": "All 41 clusters of neurons increased their firing rate during the AVERSIVE condition (Firing Rate Variation, FRV 1\u20444 0.26 \ufffd0.15)", "statistics": [], "kind": "finding", "inferred": "", "outcome": "not_reproduced", "blocked_by": null, "notebook": "reproduce_fedele_2020.ipynb", "notes": "Over the 24 s blocks and observed epochs 27 of 42 units have FRV > 0, mean 0.033 +- 0.690 Hz (Wilcoxon against 0, p = 0.50); over the 26 s trials, or with all epochs, 23 to 26 positive and means 0.007 to 0.032 Hz. By subject the mean FRV runs from -0.21 to 0.62 Hz. Relative to each unit's mean rate it is 0.061 +- 0.231 (a variant).", "deviations": [ "Observed epochs as in neuron-count; the variant with all epochs is shown." ], "assumed": false, "disagreed": false, "data": "dataset", "where": null, "proposed": null }, { "claim": "fig3a-firing-gradual-sustained", "statement": "The neurons' firing rose gradually and stayed enhanced throughout the AVERSIVE sequences, not as an immediate response at onset.", "source": "Results 2.3, paragraph 2; Figure 3A", "quote": "Contrary to our expectation of an immediate spiking onset, the neurons increased their firing gradually and sustained enhanced firing throughout the AVERSIVE video sequences", "statistics": [], "kind": "finding", "inferred": "", "outcome": "not_reproduced", "blocked_by": null, "notebook": "reproduce_fedele_2020.ipynb", "notes": "The population AVERSIVE minus NEUTRAL rate (100 ms bins, 1 s smoothing) is 0.310 Hz over 0 to 2 s and 0.018 Hz over 2 to 24 s, with its first peak at 0.15 s and 4 of the twelve 2 s bins negative. Neither reading of the criterion holds. It is -0.078 Hz over the baseline clip.", "deviations": [], "assumed": false, "disagreed": false, "data": "dataset", "where": null, "proposed": null }, { "claim": "fig3b-firing-hgp-no-condition-difference", "statement": "Across neurons, the correlation of firing rate with macro-electrode HGP does not differ between the AVERSIVE and NEUTRAL conditions.", "source": "Results 2.3, paragraph 3; Figure 3B, lower boxplot", "quote": "Across all neurons, the covariation of neuronal firing with HGP was not significantly different between the two experimental conditions (Spearman\u2019s r 1\u20444 0.17 ... \ufffd0.28 in AVERSIVE, 0.18 \ufffd0.18 in NEUTRAL; Wilcoxon paired test, p > 0.05)", "statistics": [ { "test": "Wilcoxon paired test", "p": 0.05, "p_relation": ">", "n": "41 neurons" } ], "kind": "finding", "inferred": "", "outcome": "partially_reproduced", "blocked_by": null, "notebook": "reproduce_fedele_2020.ipynb", "notes": "The readings of the claims file disagree; the claimed absence holds in 4 of 6. Per block and averaged no time base finds a difference (p 0.72 to 0.83), nor does the 1 s time base over the blocks laid end to end (r = 0.085 +- 0.130 in AVERSIVE and 0.094 +- 0.124 in NEUTRAL, p = 0.54, given as measured). Over the blocks laid end to end on the 1.5 s and 15 s time bases, each block smoothed on its own, the test finds NEUTRAL higher (p = 0.012 and 0.023, median differences -0.087 and -0.131). Smoothing over the whole session first (the first revision's form, now a variant) gave p = 0.002 and 0.009. The means are within the stated spreads under every reading. Partially reproduced and not blocked, because the six readings are named by the claims file, so no method detail is left to choose; the test ran under each, and the claimed absence holds under some and fails under others, that is, it holds only in part. Unlike the criteria of fig1b-delta-decrease and fig1f-hgp-more-specific-than-delta, this criterion does not say to block when its readings disagree. The time base is still asked of the authors (methods_unavailable).", "deviations": [ "In the readings with the blocks laid end to end, each block is smoothed (or binned) on its own before concatenation, so no sample carries another block or a baseline clip.", "HGP is the Methods 4.8 envelope without per-trial normalisation; a per-block Spearman r is unchanged by any per-trial normalisation." ], "assumed": false, "disagreed": true, "data": "dataset", "where": null, "proposed": null }, { "claim": "fig3b-firing-muhgp-higher-aversive", "statement": "Across neurons, the correlation of firing rate with microwire high gamma power (muHGP) is higher in the AVERSIVE than in the NEUTRAL condition.", "source": "Results 2.3, paragraph 3; Figure 3B, upper boxplot", "quote": "the covariation of neuronal firing with \u03bcHGP was higher during AVERSIVE than during NEUTRAL conditions (Spearman\u2019s r 1\u20444 0.51 \ufffd0.18 during AVERSIVE, 0.43 \ufffd0.16 during NEUTRAL; Wilcoxon paired test, p < 0.01)", "statistics": [ { "test": "Wilcoxon paired test", "p": 0.01, "p_relation": "<", "n": "41 neurons" } ], "kind": "finding", "inferred": "", "outcome": "blocked", "blocked_by": "data_missing", "notebook": null, "notes": "Needs the continuous microwire signals. The dandiset holds only spike times and mean waveforms of the microwire data, as the data descriptor says was shared; the OpenNeuro version holds macro iEEG only.", "deviations": [], "assumed": false, "disagreed": false, "data": "dataset", "where": null, "proposed": null }, { "claim": "fig3c-ssi-frv-correlation", "statement": "Across neurons, the spatial synchronization index (SSI) is strongly positively correlated with the firing rate variability (FRV).", "source": "Results 2.3, paragraph 4; Figure 3C; Figure 4A", "quote": "We found a strong positive correlation across all participants between the SSI and FRV (Spearman\u2019s r 1\u20444 0.78, p < 1e-7)", "statistics": [ { "test": "Spearman correlation", "statistic": "r = 0.78", "p": 1e-07, "p_relation": "<", "n": "neurons with an SSI" } ], "kind": "finding", "inferred": "", "outcome": "not_reproduced", "blocked_by": null, "notebook": "reproduce_fedele_2020.ipynb", "notes": "Over the 41 units with another unit in their amygdala, Spearman r between SSI and FRV is 0.26 (p = 0.11) on the 1 s time base and 0.40 (p = 0.009) on the 15 s one, against 0.78 with p < 1e-7. Neither reading gives the value, and the 1 s reading misses p < 0.05, so the partial result of the criterion does not hold either. Dropping the 4 pairs likely on one wire gives 0.24 (p = 0.13) and 0.35 (p = 0.023).", "deviations": [ "SSI uses every other unit of the same amygdala. Methods 4.11 say \"on different microwires\"; the units table has no microwire column (columns offset, spike_times, obs_intervals, electrodes, electrode_group, waveform_mean, waveform_sd), each unit points only to the bipolar macro channel, and the file holds one electrode group. So the Results' definition, \"all neighboring neurons\", is assumed. A variant drops pairs likely on one wire by a spike train test (no spikes within 1 ms of each other where the 2 to 12 ms flanks predict at least 3); it flags 4 of 116 testable pairs, and the others fire within 1 ms as often as the flanks predict (median ratio 1.19)." ], "assumed": true, "disagreed": false, "data": "dataset", "where": null, "proposed": null }, { "claim": "fig3d-firing-correlations-modalities", "statement": "Across neurons, the ongoing firing rate correlates positively with muHGP (0.27), HGP (0.19) and BOLD (0.23), and the correlation is lowest with HGP.", "source": "Results 2.3, paragraph 5; Figure 3D and its inset", "quote": "Across all neurons and participants, the ongoing firing rate positively correlated with \u03bcHGP (Spearman\u2019s r 1\u20444 0.27 \ufffd0.15), HGP (Spearman\u2019s r 1\u20444 0.19 \ufffd 0.17) and BOLD (Spearman\u2019s r 1\u20444 0.23 \ufffd0.20) ... Across all recording modalities, the correlation of the firing rate and other measures was lowest with HGP (Wilcoxon\u2019s paired test, p < 1e-2)", "statistics": [ { "about": "correlation with HGP against those with muHGP and with BOLD", "test": "Wilcoxon paired test", "p": 0.01, "p_relation": "<", "n": "neurons, number not given (BOLD exists only for the amygdalae with fMRI)" } ], "kind": "finding", "inferred": "", "outcome": "partially_reproduced", "blocked_by": null, "notebook": "reproduce_fedele_2020.ipynb", "notes": "The firing-HGP Spearman r is 0.136 +- 0.241 with 15 s smoothing (reading a, given as measured) and 0.139 +- 0.195 on the 1.5 s time base, positive in 29 and 32 of 42 units, within 0.19 +- 0.17. The muHGP and BOLD correlations and the two tests that HGP is lowest cannot be computed, as the microwire signals and the BOLD time courses are not available. The measured value is the HGP mean, the second expected value; the first and third cannot be measured.", "deviations": [ "Only the HGP part is tested." ], "assumed": false, "disagreed": false, "data": "with_inputs", "where": null, "proposed": null }, { "claim": "fig4b-regression-firing-hgp", "statement": "The per-neuron correlation of firing rate with HGP is described by both FRV and SSI in a multiple linear regression.", "source": "Results 2.4, paragraph 1; Figure 4B", "quote": "We found that the co-modulation between neuronal firing and HGP is described by both FVR and SSI (Fig. 4B, r 1\u20444 0.59, p < 1e-4)", "statistics": [ { "test": "multiple linear regression", "statistic": "r = 0.59", "p": 0.0001, "p_relation": "<" } ], "kind": "finding", "inferred": "", "outcome": "not_reproduced", "blocked_by": null, "notebook": "reproduce_fedele_2020.ipynb", "notes": "With 15 s time bases for both the firing-HGP r and SSI, R = 0.635 (within 0.059 of 0.59) and model p = 5.6e-5; over the four readings R = 0.56 to 0.68 and model p from 6e-6 to 7e-4. SSI has p <= 2e-4 in every reading; FRV has p 0.07 to 0.65, with a negative coefficient in three, so the regression is described by SSI alone and the claim's \"both FRV and SSI\", which the criterion requires, fails in every reading. The SSI variant without likely same-wire pairs gives the same.", "deviations": [ "SSI as assumed in fig3c-ssi-frv-correlation." ], "assumed": true, "disagreed": true, "data": "dataset", "where": null, "proposed": null }, { "claim": "fig4c-regression-firing-bold", "statement": "The per-neuron correlation of firing rate with BOLD is described by FRV but not by SSI.", "source": "Results 2.4, paragraph 1; Figure 4C", "quote": "while the co-modulation between neuronal firing and BOLD is described by FRV (Fig. 4C, Spearman\u2019s r 1\u20444 0.69, p < 1e-4), but not SSI", "statistics": [ { "test": "multiple linear regression", "statistic": "Spearman's r = 0.69", "p": 0.0001, "p_relation": "<" } ], "kind": "finding", "inferred": "", "outcome": "blocked", "blocked_by": "data_missing", "notebook": null, "notes": "Needs the per-neuron correlation of firing with BOLD; the BOLD time courses are not public (see fig2-participant4-bold-correlations).", "deviations": [], "assumed": false, "disagreed": false, "data": "with_inputs", "where": null, "proposed": null }, { "claim": "fig4d-hgp-bold-covariation", "statement": "Across neurons, the correlation of firing with HGP is positively correlated with the correlation of firing with BOLD.", "source": "Results 2.4, paragraph 2; Figure 4D", "quote": "Fig. 4D displays the covariation of neuronal firing with BOLD (x-axis) and HGP (y-axis) (Spearman\u2019s r 1\u20444 0.72, p < 1e-3)", "statistics": [ { "test": "Spearman correlation", "statistic": "r = 0.72", "p": 0.001, "p_relation": "<", "n": "neurons, number not given (BOLD exists only for the amygdalae with fMRI)" } ], "kind": "finding", "inferred": "", "outcome": "blocked", "blocked_by": "data_missing", "notebook": null, "notes": "Needs the per-neuron correlation of firing with BOLD; the BOLD time courses are not public (see fig2-participant4-bold-correlations).", "deviations": [], "assumed": false, "disagreed": false, "data": "with_inputs", "where": null, "proposed": null } ], "findings": [ { "kind": "data_missing", "statement": "The BOLD time courses of the clinical fMRI, on which six of the paper's claims rest, are not public. The dandiset holds none, the OpenNeuro version (ds003374, 1.1.1) holds only iEEG, and the paper names no deposit.", "found": null, "status": null }, { "kind": "data_missing", "statement": "The continuous microwire signals behind the microwire high gamma power (muHGP) are not in the dandiset, which holds only spike times and mean waveforms of the microwire data.", "found": null, "status": null }, { "kind": "other", "statement": "Each session is deposited twice, as `_ecephys.nwb` and `_ecephys+image.nwb`; the second adds a link to the video and holds identical iEEG, spikes and trials. The 18 NWB files hold 9 recordings.", "found": null, "status": null }, { "kind": "metadata_error", "statement": "The iEEG is stated in volts, but its values (standard deviation 15 to 101, up to 1546) are microvolts.", "found": null, "status": null }, { "kind": "metadata_error", "statement": "The trial table gives trial 1 as 0 to 26 s, including its 2 s baseline clip, and trials 2 to 17 as their 24 s blocks only (starting 2 s into each 26 s epoch). The iEEG timestamps run without a gap from 0 to 442 s, but the 17 epochs are not contiguous; the signal jumps at each boundary (median step 12 to 36 against 1.4 to 1.9 within epochs).", "found": null, "status": null }, { "kind": "metadata_error", "statement": "Every unit's obs_intervals cover all 17 epochs, but the 16 units of subject 7 have no spike after 286 s (epochs 12 to 17) and four units of subject 2 none in epoch 1, at rates of 1 to 16 Hz elsewhere.", "found": null, "status": null }, { "kind": "metadata_error", "statement": "The ages of subjects 2 and 8 (39 and 24) differ from those in the paper and the data descriptor (48 and 27). Subject 2's file holds the same subject metadata as subject 7's (age 39, gliosis, AL and AR implanted, AR in the SOZ), the age being that of the paper's patient 7.", "found": null, "status": null }, { "kind": "metadata_error", "statement": "The electrode location of the right channel of subjects 1, 5 and 9 is \"Right Hippocampus, rostral hippocampus\", and both channels of subject 6 are \"unknown\", while their labels (AR, AL), the paper and the data descriptor place them in the amygdala.", "found": null, "status": null }, { "kind": "metadata_insufficient", "statement": "Units are assigned only to the bipolar macro channel of their amygdala; the microwire each was recorded on is not given, so the paper's spatial synchronisation index (neurons on different microwires) cannot be computed as defined. The units table has no column for it, and a spike train test of coincident firing flags only 4 of 116 testable pairs as likely sharing a wire.", "found": null, "status": null }, { "kind": "other", "statement": "The units are in 8 amygdalae of 5 patients (1 L and R, 2 L and R, 3 L, 4 L, 7 L and R). The paper's Results give 9 amygdalae, and its Table 1 lists multi-unit activity for patients 5 and 8 and none for 7. Three of the 8 (1 R, 2 R, 7 R, with 17 units) are SOZ amygdalae, while the paper's Discussion 3.4 says there were no microwire recordings from SOZ amygdalae. Their mean rate is 4.65 +- 3.91 Hz against the data descriptor's 1.66 +- 2.65 Hz, and two units of subject 7 are below 1 Hz over the whole session.", "found": null, "status": null }, { "kind": "methods_unavailable", "statement": "Question for the authors. How were the power envelopes \"z-scored within each trial\" (Methods 4.8), over the whole 26 s trial, against the baseline clip, or otherwise? A z-score over the whole trial makes the baseline correction void and removes the sustained delta decrease, and with it every delta effect here; a z-score against the baseline keeps it. The outcome of fig1b-delta-decrease is blocked on this. Methods 4.8 also says results were quantitatively similar with the mean of all epochs or of the previous epoch as baseline; here the HGP clusters that begin in the first 2 s (nSOZ amygdalae) come only with these baselines, the previous epoch with z-scoring over the amygdala (0.6 to 5.5 s) and, with the trialsd variant, the mean of all epochs (0.6 to 1.9 s) and the previous epoch (0.5 to 8.4 s); the 1 s baseline clip gives none.", "found": null, "status": null }, { "kind": "methods_unavailable", "statement": "Question for the authors. Over which window was the modulation index of each amygdala computed? With these data the group test finds no HGP cluster, and per-amygdala tests find clusters in 1 to 3 of 14 amygdalae, while Table 2 prints an MI for every amygdala; the claims on the HGP MI rest on fixed windows (0 to 2 s for HGP, the whole block for delta) taken from the Results' text, an assumption, and the only failing reading of table2-delta-reduced-12-of-14 uses them.", "found": null, "status": null }, { "kind": "methods_unavailable", "statement": "Question for the authors. On what time base, and over which span (the condition's blocks laid end to end, or per block), were the per-neuron correlations of firing with HGP in Fig. 3B computed? The test finds no condition difference per block or on the 1 s time base, and finds NEUTRAL higher on the 1.5 s and 15 s time bases over blocks laid end to end.", "found": null, "status": null }, { "kind": "paper_inconsistency", "statement": "Run on the paper's printed Table 2 values, the tests of Fig. 1E find the DELTA MI also different between nSOZ and SOZ amygdalae (rank-sum p = 0.035, t-test p = 0.013), against \"HGP but not DELTA distinguishes\"; the chi-square test of the printed counts 5/5 and 1/5 of Fig. 1F gives p = 0.0098 (0.053 with Yates' correction), not about 0.2; and the exact binomial interval for 7/7 is 59-100%, not the printed 66-100%, and no single binomial interval method gives all three printed intervals.", "found": null, "status": null } ], "reconciliation": null, "verifications": [ { "date": "2026-10-30", "kind": "agent", "by": "Independent agent run by CatalystNeuro (same harness, separate context, did not write the notebook)", "reviewed": "Round 1. The first executed notebook with the conclusion cells and the summary removed, the helper module and the claims file, with the paper. The verifier checked the archive's asset list and metadata for the blocked claims and tested the helper functions on synthetic inputs. It gave its own outcome for each of the 25 claims; 20 agree with the record of that round.", "agreed": 20, "of": 25, "disagreements": [ { "claim": "nsoz-delta-hgp-mi-inverse", "outcome": "not_reproduced", "reason": "r<0 in only 3 of 8 readings." }, { "claim": "table2-delta-reduced-12-of-14", "outcome": "partially_reproduced", "reason": "11-13 negative, with negative values both inside and outside the SOZ, under window (a) wherever it is defined and under (c) except trial z-scoring (10). It fails under every per-amygdala window (b) reading (0-2 negative; most MIs are set to 0). The criterion requires every reading." }, { "claim": "fig1f-hgp-more-specific-than-delta", "outcome": "blocked", "reason": "Under the claims file's MI windows, HGP is never more specific: (a) is undefined and (b) gives ties or reverses. Under the notebook's added window (c), HGP is more specific for 3 of 4 z-scorings. The chi-square p (0.001-0.11) never approaches the paper's ~0.2." }, { "claim": "fig3b-firing-hgp-no-condition-difference", "outcome": "partially_reproduced", "reason": "No difference at the Fig. 3A time base (both spans, p 0.37/0.83) or with any per-block span. The means are within the stated spreads. With 1.5 s and 15 s concatenated series, NEUTRAL > AVERSIVE at p=0.002 and 0.009." }, { "claim": "fig4b-regression-firing-hgp", "outcome": "not_reproduced", "reason": "R is within tolerance in 3 of 4 readings, but the FRV coefficient never reaches p<0.05 (0.07-0.65). The model p<1e-4 in only 2 of 4 readings." } ] }, { "date": "2026-10-30", "kind": "agent", "by": "Independent agent run by CatalystNeuro (same harness, separate context, did not write the notebook)", "reviewed": "Round 2. The full executed notebook of the first revision, the helper module, the claims file, the attempt record and the README, with the paper. The verifier checked the outcomes, the assumptions and every Result cell, the Summary, the README and the record's notes against the printed outputs, and the round-1 points. It agrees with all 25 outcomes. In reply to its points not tied to one claim, cells 6 and 86 now count five channel locations that do not name the amygdala (not four), cell 86 says the within-patient model gives the same condition estimate as the paper's mixed model (not that it equals it), and the README gives the run time printed by the notebook.", "agreed": 25, "of": 25, "disagreements": [] } ] } ], "entries": [ { "name": "000576-fedele-2020", "type": "reproduction", "title": "iEEG, neurons and BOLD in the human amygdala during aversive videos (Fedele, Tzovara et al. 2020)", "description": "Tests the findings of Fedele, Tzovara et al. (2020) on dandiset 000576, which holds the bipolar amygdala iEEG of 9 patients (14 amygdalae) and the spike times of 42 units from 5 of them during a block design of aversive and neutral videos. It tests the delta and high gamma responses, the modulation indices and their relation to the seizure onset zone, and the firing of the units. The BOLD time courses and the continuous microwire signals are not in the dandiset.", "datasets": [ "dandi:000576" ], "papers": [ "10.1016/j.neuroimage.2020.116705" ], "derived_from": [], "notebooks": [ { "path": "reproduce_fedele_2020.ipynb", "title": "Reproducing Fedele, Tzovara et al. (2020) from dandiset 000576", "executed": "notebooks/000576-fedele-2020/reproduce_fedele_2020.html" } ] } ] }