Word-locked evoked responses mark larger subsequent memory effects in SEEG free recall
We tested whether the subsequent memory effect (SME) in SEEG free recall is larger in bipolar pairs with a word-locked evoked response. We used archived per-word features from OpenNeuro ds004789 (RAM FR1), session 0: 25 pure-SEEG subjects and 1,851 bipolar pairs. Epochs were 0–1.6 s from events.tsv sample/fs. Bands were theta (3–8 Hz) and high-frequency activity (HFA, 70–150 Hz); tilt was z(HFA) − z(theta). For each pair, we calculated a Welch t statistic (recalled minus not recalled). Within each subject, we measured Spearman rho across pairs between log evoked index and pair t. A sign-flip test across subjects and Holm correction covered three outcomes. The mean rho was +0.215 for tilt (95% CI +0.155 to +0.274; 24/25 positive; Holm p=0.00015), −0.208 for theta (CI −0.300 to −0.113; 19/25 negative; p=0.0008), and +0.092 for HFA (CI +0.023 to +0.159; 15/25 positive; p=0.017). In the pre-registered, uncorrected tercile comparison, top minus bottom third mean pair t was +0.58 for tilt (CI 0.40 to 0.75; 22/25; p=5e−5), −0.57 for theta (CI −0.81 to −0.33; p=0.0003), and +0.27 for HFA (CI 0.08 to 0.47; p=0.011). Adjustment for raw signal size gave similar correlations. A post-hoc within-atlas-region analysis also retained the association (tilt z +0.17, CI 0.09 to 0.25, 19/24, p=0.0003; theta −0.15, p=0.0009; HFA +0.10, p=0.026). The evoked index does not use recall labels and may rank pairs for memory decoding, but no decoder was tested. The index and SME use the same words. A recall-dependent ERP may affect both, especially theta; theta power also contains ERP power. The study used only FR1 session 0 and one time window, and lacked raw signals to calculate the index from separate words. Correct EDF timing is assumed to be events.tsv sample/fs. These results show an association in this dataset and analysis, not validated decoder performance.
SEEG · subsequent memory effect · evoked response · free recall · theta · high-frequency activity · memory decoding
References
2- usesAssuming sample/fs marks EDF time, 21 pure-SEEG FR1 patients showed a subsequent memory effect hidden by onset timing sci:sha256:799055dd0e8b95498bb3ea5d9564670cf779d6970c9993bda2f0c41150b419e7
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.
- usesdoi:10.18112/openneuro.ds004789.v3.1.0
The archived per-word features used in this analysis are identified as coming from OpenNeuro ds004789 RAM FR1, session 0. The saved exact-version metadata records this dataset DOI and a CC0 license.
Cited by
2 publicationsPublic catalog entries on Sciscape.
- usesIn 49 FR1 patients, the word-encoding memory effect showed no detectable gradient along SEEG shafts, assuming contact 1 was deepest
That study related the SME to the word-evoked response size of a pair. This study tests a different pair property: position along the shaft.
- A label-free word-evoked response predicts memory effects after serial-position control
[Legacy relation: reproduces] Repeats the evoked-index versus pair-SME test of 586bff5f with the same rules in 31 subjects that were not in its 25-subject sample; the raw tilt, theta and HFA links replicate (Holm p <= 0.0003).
- supportsA label-free word-evoked response predicts memory effects after serial-position control
Adds a serial-position control: the link stays significant (tilt rho +0.134, Holm p 0.0003), so it is not only a primacy effect, but it is about 25% smaller than the raw link.