Figure 6.

Decoding of the stimulus envelope in sEEG and MEG power data. A, Performance of the decoding model (TRF), trained on the power signal (2-150 Hz) of sEEG data recorded during presentation of the second half of the stimulus (62 Hz tones) and generalized to the first half of the stimulus (83 Hz) tones and the post-stimulus silence. Performance is assessed by the coefficient of determination R² obtained between the envelope reconstructed from the neural data and the actual stimulus envelope. Gray distribution indicates performance of TRF model on surrogate data. B, Marginal distributions of TRF weights (in absolute value) of sEEG data. Values are averaged in the sEEG channels (left), time (middle), and frequencies (right). C, Performance of the decoding model trained on MEG data. D, Marginal distributions TRF weights (in absolute value) of MEG data.