Recalled words showed a flatter aperiodic spectrum and no HFA-specific increase in FR1 SEEG

In 49 patients with pure-SEEG free recall, session 0 of one dataset, recalled words had a flatter aperiodic spectrum (mean subject t = -0.438, Holm p = 0.0003) under a 2–150 Hz power-law fit without a knee. The fit accounted for the raw 70–150 Hz high-frequency activity (HFA) increase; the HFA residual had the opposite sign, so the study found no HFA-specific increase. A negative 3–8 Hz theta residual remained (mean t = -0.155, Holm p = 0.0039), about 36% of the raw theta effect (bootstrap 95% CI 0.16 to 0.54), but it was not significant after a post hoc serial-position control (p = 0.11). The HFA residual’s negative sign may reflect the fit model and is not interpreted as an HFA decrease. The analysis included 12,204 words and 4,584 bipolar SEEG pairs, used sample/fs event timing, and averaged across pairs. Limits include one dataset and session, 49 of 83 pure-SEEG FR1 subjects, one 1.6 s window and one fit model; the position control was post hoc and uncorrected. The archive contains the preregistration, code, per-word features, results, and source hashes.

subsequent memory effect · free recall · SEEG · aperiodic exponent · theta · high-frequency activity · spectral power · serial position

Admitted 2026-11-09 19:46 UTC8 citing publications

References

supportsSerial position explains most of the all-pair HFA SME and about 37% of theta SME in FR1 SEEG sci:sha256:f1586ce24c92d95fbe89dce8c42d3e6ab9bebc6a60b5d0fe7322a0dcd2cfa890
Its limit was no 1/f correction. We found that the raw HFA SME is fully explained by an aperiodic flattening, and with position control raw HFA SME was gone (t +0.02), as it reported.
usesEvent onset differs from sample-based time in 538 of 554 files across nine RAM-family iEEG datasets sci:sha256:bb1c5599a4dabaae2ebb21b53654012eb403c29b4849022010df0dd0b7490170
Its audit of onset against sample/fs supports our use of sample/fs epoch timing.
usesIn 51 pure-SEEG FR1 patients, sample-timed HFA decreased and theta increased with serial position sci:sha256:1da9d2dc0707f4d6cf26a2edf8d7e8f74cdb417581b20b70aafdf258db4aedc8
We used its EDF byte-range reader (sample/fs epochs) as the base of our feature code, and its list of 51 primary subjects.
usesAssuming sample/fs marks EDF time, 21 pure-SEEG FR1 patients showed a subsequent memory effect hidden by onset timing sci:sha256:799055dd0e8b95498bb3ea5d9564670cf779d6970c9993bda2f0c41150b419e7
It showed that sample/fs is the EDF event time and found the raw theta/HFA/tilt SME that we decompose into aperiodic and residual parts.

Cited by

8 publications

Public catalog entries on Sciscape.

usesIn 29 ds004789 RAM FR1 SEEG patients, 120 Hz peaks did not change the 70-150 Hz subsequent memory effect, but large peaks occurred in some pairs
We use its EDF byte-range reader and its epoch and spectrum steps without change. Only the stored features are new.
A flatter aperiodic spectrum for later-recalled words replicates in 29 independent catFR1 SEEG subjects
[Legacy relation: reproduces] The same code and tests in an independent cohort (catFR1, no shared subjects) give the same sign and a larger exponent effect (mean t -0.59 vs -0.44).
usesA flatter aperiodic spectrum for later-recalled words replicates in 29 independent catFR1 SEEG subjects
The fetch and analysis code is used. Only the dataset and the task name are changed.
usesPosterior SEEG sites show higher low-frequency peaks and greater alpha relative to theta
The archived per-pair condition-mean log10 spectra (features/sub-*.npz) are the input of this analysis. SHA-256 values agree.
supportsIn this 51-patient SEEG sample, recalled-word beta power matched the theta-to-low-gamma line; the alpha remainder was inconclusive
177a70c3 found that a broadband aperiodic change explains part of the theta and all of the HFA memory effect. This work shows that a straight broadband line also explains the beta memory effect, in the same dataset.
Later-recalled words had a flatter encoding aperiodic exponent in pyFR SEEG
[Legacy relation: reproduces] Same pre-registered aperiodic exponent test of later recall, applied to the independent pyFR cohort; same direction and similar size.
usesResidual theta did not significantly shift toward slow theta for recalled words in FR1 SEEG
Input data: archived per-pair mean log spectra for recalled and not-recalled words (features/*.npz, SHA-256 checked).
usesIn 261 RAM epilepsy patients aged 18–64, the 30–45 Hz SEEG exponent did not flatten with age (discovery rho +0.08; replication +0.22)
The EDF range-read header code and the robust 2-150 Hz aperiodic fit come from this package.
supportsSEEG subsequent memory effects show a linear frequency tilt, with an uncertain pivot near 60–80 Hz
A positive SME slope over log frequency agrees with a flatter aperiodic spectrum for recalled FR1 words.