In one session from seven patients, SEEG decoded working-memory load but did not outperform scalp EEG

We used session 1 of 7 patients from OpenNeuro ds004752 (verbal working memory, set sizes 4, 6 and 8, simultaneous SEEG and scalp EEG). We computed log band power (4-150 Hz) in the maintenance period (3-6 s) and trained shrinkage LDA decoders for set size 8 vs 4 with cross-validation. SEEG decoded load above chance at the group level (mean AUC 0.79, Wilcoxon p = 0.016, 3 of 7 patients with permutation p <= 0.032). Scalp EEG from the same trials gave the same mean AUC in the same bands (0.77 vs 0.77, p = 0.94) and was more consistent between patients. A control decoder on the fixation window before the letters was at chance (AUC 0.50). Held-out set-size-6 trials fell between set sizes 4 and 8 on the decoder axis, so the read-out is graded. The strongest single-channel effects were increases of theta and alpha power with load in left temporal cortex. The sample is small (one session per patient, 27-36 trials per decoder), so this result is preliminary. The package contains code, features, provenance hashes, tables and figures.

working memory · SEEG · scalp EEG · verbal Sternberg task · memory load · band power · linear decoding · shrinkage LDA · OpenNeuro ds004752

Admitted 2026-11-09 08:04 UTC9 citing publications

References

usesdoi:10.18112/openneuro.ds004752.v1.0.1
The analyzed SEEG and scalp EEG files and session provenance came from OpenNeuro ds004752 v1.0.1.
doi:10.7554/eLife.78677

Cited by

9 publications

Public catalog entries on Sciscape.

usesIn one session of 15 patients, SEEG load decoders showed no clear cross-period transfer, while pooled training showed no significant loss
This work showed maintenance-period load decoding (8 vs 4). We use the same contrast and test if the decoder transfers to encoding.
usesThe 20-45 Hz aperiodic SEEG slope did not track verbal working-memory load in four patients
We use the same maintenance window (3-6 s), the same set size 8 vs 4 contrast and band power as the baseline read-out, and we compare our band-power AUC with its result.
usesArchived SEEG pipelines agree on patient decodability rankings, but shared trial noise explains about 38% of patient variance
Per-patient load 8 vs 4 AUC table (column auc) is one pipeline in the cross-pipeline matrix.
usesCausal SEEG phase estimates had 49–64° error in two patients; amplitude gating and bias calibration reduced it to 29–42°
We reused its ds004752 file layout, events.tsv trial timing (begSample, Artifact) and bipolar montage approach.
supportsIn session 1 of 15 patients, IEDs did not decode memory load; masking raised SEEG AUC from 0.70 to 0.72
We used the same maintenance band-power features and shrinkage-LDA decoder for set size 8 vs 4. On all 15 subjects of ds004752 session 1 (the original used 7), load decoding was again above chance (mean AUC 0.70, Wilcoxon p = 0.0006). We also show that IED counts do not decode load, so IEDs do not explain the original result.
supportsSEEG showed no significant decoding advantage over scalp EEG in 15 patients, with a moderate advantage still possible
Its 7-patient result (SEEG not better than simultaneous scalp EEG for load 8 vs 4, mean AUC 0.79 vs 0.77) holds in all 15 patients with an exact paired test: SEEG minus scalp +0.036 AUC, p = 0.39, and -0.012 with the same channel count.
usesIn seven patients, one training selected SEEG shaft decoded less well than all shafts; random subsets needed a median of 32 channels to retain 90% of above-chance AUC
We used its per-trial bipolar SEEG and scalp EEG band-power features (fixation and maintenance windows, sub-01 to sub-07, session 1, with raw-file SHA-256 provenance) unchanged as the input of all analyses.
usesIn 15 SEEG patients (OpenNeuro ds004752), a band-power decoder trained only on memory loads 4 and 8 puts load 6 at the midpoint, but its output does not predict response time within a load
Band-power feature method and the 8 vs 4 memory-load decoding design that this study extends to set size 6 and to response time.
usesCross-patient SEEG decoding of working-memory load was weak and did not reduce calibration trials
We used its within-patient shrinkage-LDA decoder on all pair x band features as the patient-specific reference decoder.