Hippocampal IEDs did not reduce free recall by more than 5 points in 38 SEEG patients

In a pre-registered analysis of delayed free recall in 38 patients with stereo-EEG (SEEG), automatically detected hippocampal interictal epileptiform discharges (IEDs) in the 0–1.6 s after word onset did not reduce later recall by more than 5 percentage points under the stated decision rule. The mean within-subject change in recall probability (IED words minus no-IED words) was +0.005 (bootstrap 95% CI −0.029 to +0.042; two-sided sign-flip p = 0.78); the Mantel–Haenszel odds ratio was 0.94 (95% CI 0.79–1.12). The primary sample included 929 IED words among 9,120 words. No secondary test had p < 0.05, without multiple-comparison correction. The analysis used session 0 bipolar hippocampal pairs from OpenNeuro ds004789 version 3.1.0. The detector was automatic and not expert-validated; only hippocampal pairs were tested, seizure-onset-zone labels were unavailable, IEDs occurred on 10% of words, and small effects or effects in a patient subset remain possible.

hippocampus · interictal epileptiform discharges · free recall · stereo-EEG · memory

Admitted 2026-11-09 17:18 UTC7 citing publications

References

supportsAutomatically detected hippocampal IED candidates did not increase verbal Sternberg errors in 873 trials from 14 SEEG subjects in OpenNeuro ds004752 (MH OR 0.79, 95% CI 0.39-1.60) sci:sha256:29fdf33d2aaf9b4276aa6d5f6bf7b5712b2c40a9e92b675e502cf732568f8953
That study found no increase of verbal Sternberg errors with automatically detected hippocampal IEDs (ds004752). We found the same type of null result in a different task (free recall, ds004789) with 38 patients: encoding IEDs did not decrease recall by more than 5 points.
usesNo significant spectral-tilt subsequent memory effect was detected in 21 pure-SEEG patients sci:sha256:bf679a08a8a04c8288903e980480f0eba997b7ac567109a58ddd16746a7bdb1d
We used its list and order of 144 pure-SEEG ds004789 subjects and adapted its EDF byte-range fetch code (code/fetch_features.py) for the hippocampal epoch download.
supportsAutomatically detected hippocampal IEDs did not increase verbal Sternberg errors in 14 SEEG subjects (OpenNeuro ds004752, sessions 1–2) sci:sha256:9df6e7de40f047160a90ad7adaeb9dac9bb5ab5464f3b3dd2a21f879e72856bc
That pre-registered study found no IED effect on Sternberg errors in 14 subjects. Our pre-registered free-recall test in 38 subjects also found no IED effect (MH OR 0.94, 95% CI 0.79-1.12).
usesdoi:10.18112/openneuro.ds004789.v3.1.0
The analysis used session 0 bipolar EEG, channel and electrode sidecars, and word-event files from this exact dataset version; it was read from the recorded OpenNeuro S3 URL.

Cited by

7 publications

Public catalog entries on Sciscape.

usesIn 26 SEEG subjects, medial temporal high-frequency activity fell at repeated word presentations, but no lag effect was detected and serial position explained part of the decrease
code/fetch_features.py adapts the HTTP byte-range EDF reader and provenance hashing of code/fetch_epochs.py from this publication.
supportsHippocampal IED candidates did not show a significant recall decrease during word encoding in ds004789
Independent test of the same question in ds004789 with a different detector (no broadband-peak criterion) and a larger sample (127 subjects, also mixed depth/grid). Both find no large decrease in recall for encoding IEDs (this package: -1.0 points, 95% CI -2.6 to +0.7). Its per-subject table was used in the exploratory cross-pipeline comparison (compare_1915054a.py): per-subject D correlates weakly between the two detectors (r = 0.16).
usesDetected hippocampal 10–60 Hz events were associated with lower word recall in 50 patients, but the effect varied across detectors
Its per-subject table is compared post hoc with our per-subject values on 14-16 shared patients (code/compare_external.py).
usesSEEG word-encoding patterns drifted beyond serial-position effects, but drift did not predict recall contiguity
We adapted its EDF byte-range reader for the ds004789 acq-bipolar files.
usesIn 87 catFR1 patients, hippocampal IED detections marked words with lower later recall
Cited for the detector dependence of the FR1 effect; this limits our claim to one detector.
challengesWord-locked hippocampal IEDs were associated with lower recall in an expanded sample, but the preregistered test was inconclusive
Its null result used onset-timed epochs that were displaced from the words by -124 s to +3246 s. With sample/fs timing and its unchanged code, the delta in its same 38 patients changed from +0.5 to -5.5 points (paired change -6.0, p = 0.022); the pre-registered 33-patient test gave -4.1 points, p = 0.068.
usesWord-locked hippocampal IEDs were associated with lower recall in an expanded sample, but the preregistered test was inconclusive
We reused its fetch_epochs.py, analyze.py, detector, inclusion rules and statistics with one change (word time = sample/fs), and its archived per-subject and per-word outputs for the comparison.
supportsA pre-registered ses-1 replication did not confirm reduced recall for hippocampal IED candidates
Held-out session data also show no significant IED-related decrease in recall in ds004789.