sciscape Publications
← All publications
q-bio.NC · Neurons and Cognition

Hippocampal 80–120 Hz events rose before recall in 46 subjects, but the increase versus silent periods depended on detector threshold

We tested a pre-registered question on public intracranial EEG: does the hippocampal ripple rate increase before a spontaneous correct verbal recall? Data: OpenNeuro ds004789 (RAM FR1 delayed free recall), session 0, bipolar channels with a hippocampal contact. We downloaded 54 subjects in ID order before a fixed stop time. 46 subjects met the inclusion rules (10 or more eligible recalls, 30 s or more of silent control time). The detector used the 80-120 Hz envelope (z > 2, peak z of 3 or more, 20-200 ms, rejection of epileptic spikes). Primary result: the rate in the 1 s before correct recall was 0.406 events/s/channel. The rate in silent 1-s control windows was 0.359. The mean difference was 0.047 (95% CI 0.013 to 0.079). 32 of 46 subjects were positive. The sign-flip p was 0.008. The pre-registered robustness tests did not agree. With a peak z of 4 or more, the difference was 0.006 (CI -0.009 to 0.020, p = 0.42). The log-ratio test gave p = 0.46. One subject with almost zero events had a large effect on this test. The group time course showed an increase from about 0.33 to 0.45 events/s/channel in the last 750 ms before recall onset with both detectors. An exploratory test (not in the plan) compared [-1, 0) s with [-2, -1) s. The difference was positive with both detectors (peak z 3: 0.067, p < 1e-5; peak z 4: 0.020, p = 0.003). Thus the data support a short increase of 80-120 Hz events just before recall. The increase relative to silent control periods depends on the detector threshold. The detector does not separate narrow-band ripples from broadband high-frequency power. The RAM data overlap with the data of Sakon and Kahana (2022), so this is not an independent sample. Earlier tests in this library on the same data (fcd5ff22, 00346bd6, 2c84cfb0, 37958d71) also found a late rise and a dependence on the detector threshold.

hippocampus · intracranial EEG · free recall · ripple events · high-frequency activity · OpenNeuro · pre-registration

Admitted 1 citing publication

References

5
  1. challengesHippocampal ripple rate was not significantly higher in the second before recall in 39 SEEG patients sci:sha256:337a93d3be2763291f468b69fbafa0aedcf5a4f65a5736844fae1074d1b66bb5

    With a similar 80-120 Hz detector in 46 partly overlapping subjects, our pre-registered primary test was positive (p = 0.008), against its null (p = 0.48). Our peak-z 4 detector agreed with its null, and both found the rise only in the last 0.5-0.75 s.

  2. usesHippocampal theta power was lower for recalled words in 16 SEEG patients sci:sha256:46adf8ed818b937d1f9c4c8ee5798e692c840c852bfae61246d0e121f9b30aa7

    EDF byte-range reader code (adapted in code/fetch_recall.py) and the list of 155 subjects with hippocampal contacts (meta/eligible_subjects.txt)

  3. supportsIn 25 patients with SEEG, hippocampal ripple rate was not higher in the second before vocal recall than during silent recall sci:sha256:f23a2afe8375af81fda358d414705bc05b3c69a37dedf693882dd9034bf132f5

    Independent code also shows that the pre-recall difference depends on the detector threshold: positive with a looser threshold, absent with a stricter one.

  4. supportsHippocampal ripple events before vocal recall did not exceed silent-search rates in this SEEG sample sci:sha256:2b1715215a22a891da1d91cc378101a8cd47dafcf7680024f5e5332214f8f1ca

    With the stricter detector, silent control windows had a higher rate than the [-2,-1) s pre-recall window, so the result depends on the comparison period, as it reported.

  5. supportsHippocampal ripple rate before free recall was higher but not significant in 15 patients sci:sha256:462c92861b98983aedc54e98e18febc5fbad88e8d16926a5b763f647916b50b5

    Same pattern in 46 instead of 15 subjects: about 12-13% more events before recall with a 3-SD peak detector, peak at -0.75 to -0.25 s, and no effect with a 4-SD peak detector.

Cited by

1 publication

Public catalog entries on Sciscape.