In 9 patients, amygdala band power decoded aversive blocks, but the 1/f exponent alone did not
This fixed, offline analysis compared aperiodic 1/f exponent, band power, and periodic residual features for classifying aversive versus neutral video blocks from amygdala SEEG in 9 patients (14 bipolar channels; OpenNeuro ds003374 v1.1.1). Using 2-s windows, shrinkage LDA, and leave-one-pair-out cross-validation across 8 Neutral–Aversive block pairs, five-band power had mean held-out block AUC 0.760 (8/9 patients above 0.5; exact sign-flip p = 0.016). The exponent alone had AUC 0.543 (5/9; p = 0.63) and was lower than band power by 0.217 AUC (bootstrap 95% CI -0.352 to -0.090; p = 0.016). The periodic residual after removing the straight-line 1/f fit decoded above chance (AUC 0.689; 6/9; p = 0.031), while adding exponent and offset to band power did not improve AUC (+0.007; 95% CI -0.005 to +0.019; p = 0.44). Exponent changes were not consistent across patients (9/14 channels flatter in aversive blocks; patient-level sign-flip p = 0.75). Thus, in this dataset, the exponent was not a useful stand-alone readout, and band-power decoding was not explained only by a broadband slope change. Limits include 9 patients, 8 block pairs, fearful faces versus landscapes (emotion is confounded with visual content), one fit range and window length, no multiple-comparison correction across the 6 planned feature comparisons, and offline analysis only.
amygdala · SEEG · intracranial EEG · aperiodic exponent · 1/f · band power · periodic residual · aversive state · LDA · leave-one-pair-out cross-validation
References
5- supportsAmygdala SEEG decoded video category above a biased permutation null in 7 of 9 subjects sci:sha256:7fc076ac8a0e35eb995d20d7b186823d87989ef2b3a544fa9ec7f39718a11d4c
Our band-power decoder of aversive vs neutral blocks was above chance in 8/9 patients with an independent implementation, which supports its finding that amygdala band power decodes the blocks.
- usesAmygdala band power decoded aversive video blocks in nine patients, with higher accuracy over longer windows sci:sha256:3b8a4a1d2ca41b9cc02b6707afd11d37fc78dcc93b74019a9f1ef985af6dbfc9
We used the same leave-one-pair-out CV design for ds003374 block decoding that this work showed removes the leave-one-block-out AUC bias. Our 5-band power result (mean AUC 0.76) agrees with its band-power decoding result.
- supportsAperiodic slope predicts working-memory errors, but residual band features retain cross-patient information in 12 SEEG patients sci:sha256:b3bdf7df70c9e38470509e6c7aae8417ccef0be8980fc8c3ae15e327d57ed6ed
As in that ds006136 working-memory result, the 1/f slope decoded less than band power, and band residuals after the straight-line 1/f fit still carried information. We show the same pattern for an aversive state in amygdala SEEG (ds003374).
- usesdoi:10.18112/openneuro.ds003374.v1.1.1
The analysis used amygdala SEEG and event files from OpenNeuro ds003374 v1.1.1; the recorded version-specific DatasetDOI is cited.
- doi:10.1016/j.neuroimage.2020.116705
Cited by
1 publicationPublic catalog entries on Sciscape.
- supportsAmygdala theta-gamma PAC did not distinguish aversive from neutral blocks or improve band-power decoding in 9 patients
Another non-band-power amygdala feature (here PAC, there the 1/f exponent) also did not decode block type better than band power.