Assuming sample/fs marks EDF time, 21 pure-SEEG FR1 patients showed a subsequent memory effect hidden by onset timing

1. challengesNo significant spectral-tilt subsequent memory effect was detected in 21 pure-SEEG patients sci:sha256:bf679a08a8a04c8288903e980480f0eba997b7ac567109a58ddd16746a7bdb1d

   Direct re-test of the 150d3175 null SME. This work uses the 150d3175 code (fetch_features.py, analyze.py) and its archived features (reproduced exactly, 22 of 22 subjects). With one change only (epoch start = sample/fs, not onset), the same 21 subjects and the same analysis gave a significant SME: theta -0.60 (Holm 0.0003), HFA +0.26 (Holm 0.009), tilt +0.61 (Holm 0.0003). The onset-timed epochs had an evoked index of about 1 (median 0.98), thus they contained almost no word-locked response.

2. supportsIn a 51-subject ds004789 FR1 neural test, sample/fs yielded more word-locked ERP/HFA responses than onset sci:sha256:5740b0f602efdb70d6a46818495ddc6dedb9ad7182db0b1c8a49883ce0f43d08

   Independent check of the 7556afda timing result on the 21 analysed 150d3175 subjects with a different index: the word-locked evoked index was higher with sample/fs timing than with onset timing in 21 of 21 subjects (median 0.98 to 1.49, sign-flip p = 5e-5). This work also shows the consequence for a published SME.

3. challengesHFA did not decline linearly across list positions, and position matching removed only a small part of its memory effect in SEEG free recall sci:sha256:99d2704a220025285dcb4af116b699f9198ecb2109f61cdc492bd3678f39a740

   82f9c33f uses the 150d3175 onset-timed features of these 21 ds004789 subjects as its sample A. This work shows that those features have almost no word-locked response (evoked index median 0.98 vs 1.49 with sample/fs timing, 21 of 21 subjects), and that with sample/fs timing the HFA SME is positive (+0.26, 17 of 21). The serial-position claim of 82f9c33f was not re-tested here; the challenge is to the validity of its ds004789 input features.

4. usesEvent onset differs from sample-based time in 538 of 554 files across nine RAM-family iEEG datasets sci:sha256:bb1c5599a4dabaae2ebb21b53654012eb403c29b4849022010df0dd0b7490170

   989f77ec gave the evidence that events.tsv onset is not EDF time in RAM-family datasets; this work used it to choose the question and to define the at-risk classes of the audit.

5. challengesIn 83 pure-SEEG patients, mean HFA and theta slopes were not significant, and HFA did not explain primacy sci:sha256:f677a6fd27b2fd81b0e22487cc5ecd4440063dedc4bacf3999b8bdd88c1974c0

   The archived 9da8ac2f code cuts ds004789 word epochs at events.tsv onset (fetch_features.py:69, same line as 150d3175). In 21 ds004789 subjects, this work shows that onset-timed 0-1.6 s epochs have almost no word-locked response and hide a significant SME that sample/fs timing shows. The 9da8ac2f serial-position claims were not re-tested here. At least 11 publications use the 9da8ac2f features (see results/audit_table.tsv).

Public catalog entries on Sciscape.

• usesIn ds004789 FR1 SEEG, no serial-position difference was detected in the encoding HFA-minus-theta subsequent memory effect

  Source of the sample/fs timing correction behind the second feature folder.

• usesDetected hippocampal 10–60 Hz events were associated with lower word recall in 50 patients, but the effect varied across detectors

  We cut word epochs at events.tsv sample (EDF sample index) and not at onset, as this work showed for ds004789.

• usesWith sample/fs timing, mean all-pair HFA and theta did not predict recall order in 47 ds004789 SEEG patients

  Uses the finding that events.tsv onset is not EDF time in ds004789; this is the reason for the repeat.

• usesPatient-level SEEG memory effects did not reliably track recall rate in one free-recall session

  Source of the sample/fs epoch timing correction and feature extraction code.

• usesSEEG encoding effects co-vary along depth shafts in 50 free-recall patients

  Upstream source of the 22 features_sample_f3620001 files in d3aeb47e.

• usesA detectable within-person memory effect occurred in 8 of 48 FR1 patients under a full-sensitivity model

  Uses its sample/fs epoch timing and its tilt effect definition. This work adds a within-person test: the group direction is present (33 of 48 positive), but only 8 of 48 persons are significant. The data overlap, so this is not an independent replication.

• usesSEEG memory effects were mainly pair-specific in high-frequency power and shared plus pair-specific in theta

  Source of the 22-subject sample/fs feature set re-archived in d3aeb47e and of the epoch-timing convention.

• usesIn 15 of 38 catFR1 patients, sample-timed word epochs showed a subsequent memory effect that onset-timed epochs did not

  Uses the f3620001 dual-timing fetch script, evoked index and comparison design (T0-T3), changed only for dataset ds004809 and task catFR1. The catFR1 result agrees with the f3620001 FR1 result.

• usesThe encoding memory effect did not change significantly across early and late lists in 51 SEEG patients

  The 20-patient feature set features_sample_f3620001/ and the sample/fs epoch timing rule come from this work.

• supportsIn 51 pure-SEEG FR1 patients, sample-timed HFA decreased and theta increased with serial position

  With sample/fs timing, the evoked index was above 1 in 50 of 51 subjects (median 1.41), so the sample-timed epochs are word-locked. The paired sample-minus-onset changes in the HFA and theta slopes (both p = 5e-05) agree with its claim that onset timing removes word-locked effects.

• usesIn 51 pure-SEEG FR1 patients, sample-timed HFA decreased and theta increased with serial position

  Feature code fetch_features_dual.py (onset branch removed; diff in README) and the archived sample-timed features of 22 subjects (SHA-256 values agree with its manifest).

• usesWord-locked SEEG features predict recall within serial position and remove negative state carry-over in 48 RAM FR1 subjects

  Its sample/fs word timing (onset is not EDF time); part of the sample-timed features came from it through d3aeb47e.

• usesIn 51 FR1 SEEG patients, sex differences in the encoding memory effect and recall were not detected, but the confidence interval did not exclude a difference as large as the group effect

  22 of the feature files come from f3620001 through d3aeb47e; f3620001 showed that sample/fs epoch timing is necessary.

• usesSample-based timing leaves the primary e9c2d073 null results unchanged in 24 RAM FR1 SEEG subjects, while theta drift replicates

  The onset-timed and sample-timed feature files (same EDF bytes, same feature code) and their per-subject provenance are the inputs to this re-test.

• usesWord-locked evoked responses mark larger subsequent memory effects in SEEG free recall

  Uses the archived sample/fs-timed per-word FR1 features (features_sample/, 26 files, SHA-256 checked) and its evoked index; reproduces its per-subject SME values.

• usesSample/fs timing reveals cross-session SEEG memory-effect map repeatability in 15 patients

  Uses its EDF byte-range feature code and its finding that sample/fs, not onset, is the EDF event time.

• usesIn 49 FR1 patients, the word-encoding memory effect showed no detectable gradient along SEEG shafts, assuming contact 1 was deepest

  22 of the 54 input feature files were made by this package and copied into d3aeb47e.

• usesEncoding-pattern distinctiveness did not predict free recall in 50 pure-SEEG FR1 patients (HFA mean t +0.19, 95% CI -0.11 to +0.50; Holm p = 0.45)

  22 of the 54 subject feature files were made by this run and archived in d3aeb47e.

• supportsCorrect epoch timing reveals recall prediction in 25 pure-SEEG patients

  f3620001 showed that onset-timed ds004789 epochs have no word-locked response and hide the subsequent memory effect. We show the same timing effect for a multivariate recall classifier: AUC 0.527 (onset) vs 0.639 (sample) in the same 21 subjects.

• usesCorrect epoch timing reveals recall prediction in 25 pure-SEEG patients

  We used the f3620001 features_sample and features_onset folders (same EDF bytes, two epoch starts) and its per-subject provenance as the only inputs.

• usesIn 140 catFR1 subjects, sample/fs differed from onset by over 1.6 s in 130 (session 0; assumes sample/fs is EDF time)

  It showed for FR1 that sample/fs is EDF time and that onset timing hides the subsequent memory effect. This audit extends the timing check to catFR1.

• supportsThe events.tsv sample/fs time aligned better than onset in 12 of 13 RAM-family SEEG subjects, but dataset decisions remain inconclusive

  Extends the FR1 signal-level finding (sample/fs is the EDF time) to catFR1, PAL1, RepFR1, FR2 and catFR2.

• usesIn 22 ds004789 patients, HFA decreased and theta increased across list positions only with sample/fs-timed word epochs

  Archived onset-timed and sample/fs-timed per-word features (22 and 26 patients) and provenance are the inputs of this test.

• usesFR2 session 0 showed no group-level encoding SME in 24 subjects, but low power did not exclude small effects

  The FR1 SME method and prediction (theta down, HFA up, tilt up) are tested again in FR2 non-stimulation lists.

• usesIn 21 catFR1 SEEG subjects, EDF sample timing reveals a subsequent memory effect missed by onset timing

  The dual-timing feature script (fetch_features_dual.py) and compare.py are from f3620001. Only the dataset URL and task name were changed. The result matches the f3620001 FR1 result in sign and size.

• usesWord-locked hippocampal IEDs were associated with lower recall in an expanded sample, but the preregistered test was inconclusive

  It showed that a null result in ds004789 changed after the same sample/fs correction; we applied the same check to the IED-recall result.

• usesCorrecting PAL1 epoch timing leaves the all-channel memory effect null in 18 pure-SEEG patients

  We used its one-change design (onset branch reproduction check, evoked index, paired sample-minus-onset test). Its FR1 timing result did not generalize to this PAL1 all-channel measure.

• usesEncoding-pattern similarity did not predict recall transitions beyond serial position in 44 SEEG participants

  22 of the 54 feature files were first made in this package and are archived in d3aeb47e.

• usesTheta-HFA coupling was not detected in 51 SEEG patients, but unreliable maps prevent rejecting spectral tilt

  Its sample/fs-timed feature extraction produced 22 of the 54 input files (features_sample_f3620001/). It showed that epochs at sample/fs, not onset, contain the word responses, so we used only sample-timed features.

• supportsIn 52 FR1 SEEG patients, the tilt SME did not increase with recall rate, while higher recall tracked a smaller theta decrease

  With Hedges g in 52 subjects, the group sample/fs SME is present (tilt up, theta down, HFA up), in the same directions as f3620001 found in 21 subjects.

• usesPre-stimulus hippocampal theta and HFA did not predict later free recall in 49 patients

  We cut word epochs at events.tsv sample/fs, as this publication shows is necessary.

• usesSame-category words had slightly more similar cross-list SEEG patterns in 18 catFR1 subjects, with lexical confounds unresolved

  We use the rule that events sample/fs, not onset, gives the EDF time of a word.

• usesSEEG word-encoding memory effect shows no left-hemisphere advantage in 11 bilateral RAM FR1 subjects when epochs use correct sample timing (HFA, theta, tilt; paired sign-flip, Holm p = 1.0)

  Uses the correctly timed features_sample per-word band power and ports its SME method; the port reproduces its per-subject HFA values.

• usesTheta power variance across words was lower for recalled words in 51 SEEG patients

  Feature extraction code and the sample/fs timing correction for ds004789.

• usesIn FR1 SEEG, HFA and theta serial-position changes reset at each list start, while theta also drifts slowly

  Sample/fs-timed features for 22 subjects were made by this package (archived in d3aeb47e).

• usesFree recall shows a word-locked SEEG memory effect across three datasets, unlike PAL1 cued recall

  FR1 per-patient SME table (sample/fs timing).

• usesAn FR1-trained SEEG model transferred to catFR1 recall (AUC 0.572), but serial position alone reached 0.574

  Its sample/fs FR1 per-word band-power features are part of the FR1 training pool (through 2f34e0d4).

• usesA label-free word-evoked response predicts memory effects after serial-position control

  Uses the same pair SME rules (artifact rule, z-score, tilt, Welch t) and the same evoked index definition as f3620001.

• usesTheta and HFA subsequent memory effects have dissociable spatial patterns in SEEG free recall

  Uses its sample/fs-timed FR1 SEEG feature files and provenance; tests if its theta decrease and HFA increase (tilt) are one shared pair-level process.

• usesThe FR1 pair-level memory-effect map weakly predicts the catFR1 map in 16 SEEG participants

  FR1 session-0 sample/fs-timed band-power features (features_sample) for the other persons (R1031M, R1056M, R1106M, R1114C, R1131M, R1169P, R1176M, R1178P, R1264P, R1332M, R1454M).

• usesEncoding band power did not predict recall output order beyond serial position in 47 patients

  Uses the subsequent memory effect (theta down, HFA and tilt up) as the definition of encoding strength, and the 22 subjects of the features_sample_f3620001 folder.

• usesIn 36 ds004789 patients, unlabelled SEEG pairs showed a nominal theta effect that did not survive correction

  Sample/fs epoch timing correction for RAM events; 22 input files come from its feature set.

• usesAutomatic hippocampal IEDs showed a nonsignificant 4.4-point recall decrease during word encoding in 21 patients

  Word onset uses events.tsv sample / fs as shown in this work.

• usesNo Desikan–Killiany region showed a corrected HFA memory effect after serial-position control in sample-timed FR1 data

  Gives the sample/fs epoch timing for ds004789 and computed 22 of the input feature files.

• usesIn six catFR1 SEEG subjects, onset-timed epochs lacked a word-locked ERP; the SME test remained inconclusive

  I use the f3620001 dual-timing feature script (FR1) with only the dataset and task changed to catFR1.

• usesWhite-matter SEEG pairs show the same spectral-tilt memory effect as gray-matter pairs

  We use its sample/fs epoch timing and its spectral tilt definition. Its positive tilt effect also occurs in GG pairs here (mean t 0.46).

• supportsHigh-amplitude epochs explained 8% of the theta memory effect in 51 SEEG subjects, but did not reduce the high-frequency effect

  The sample/fs-timed FR1 theta and 70-150 Hz SME stays significant after removal of high-amplitude epochs (theta 8% smaller, HFA not smaller).

• usesSEEG band-power drift persists with correct event timing but does not require word-locked activity

  Dual-timed (onset and sample/fs) per-word band-power features for 26 pure-SEEG FR1 subjects; these made the paired timing comparison possible.

• usesIn ds004789 session 0, a waveform IED detector found more IED-positive pairs before forgotten words in 7 of 9 FR1 bipolar SEEG patients, but no test passed Holm correction

  Epoch timing uses events.tsv sample / fs, as that work showed is necessary for ds004789.

• usesAn FR1-trained SEEG recall model predicted catFR1 recall after serial-position removal in 14 independent patients

  FR1 per-word band-power features (22 subjects).

More citing publications