Fig 3. Within-site spectral-power associations on the sub-second timescale (within a 500 ms interval).

The sub-second-timescale spectral-power associations were computed at 512 Hz temporal resolution within each 500 ms interval and averaged across all intervals (per site per participant) and averaged across participants. Each 2D-association plot shows the strengths of the same-signed (positive values, warmer colors) or opposite-signed (negative values, cooler colors) power modulation at all test frequencies (x-axis) that temporally coincided with the top/bottom-15% power variation at each probe frequency (y-axis), computed as Ln-ratio relative to the DCT-phase-scrambled control (to remove artifacts due to partial wavelet overlap; see text for details). Any asymmetry across the diagonal indicates directional differences; X-Hz power modulation coincident with the top/bottom-15% Y-Hz power variation may be stronger/weaker than the Y-Hz power modulation coincident with the top/bottom-15% X-Hz power variation. A-C. Representative association patterns. The strong positive associations along the 45° lines (A and B) indicate that spectral-power associations occurred at constant frequency differences, f’s. The accompanying line plots show the association strengths for major frequency bands as a function of Δf(warmer-colored curves representing higher-frequency bands), averaged across the y-side and x-side of the diagonal as the associations were approximately symmetric about the diagonal. The plots clearly show that the association strengths peaked at specific Δfvalues. A. An example from a representative posterior site (PO8) characterized by the Δf~ 3 Hz associations for the θ-α frequencies (small dashed ellipse) and particularly by the Δf~ 10 Hz associations for the β-γ frequencies (large dashed ellipse), suggesting that the θ-α and β-γ frequencies are amplitude-modulated at ~3 Hz and ~10 Hz, respectively, at posterior sites (see text for details). B. An example from a representative lateral site (C6) primarily characterized by the Δf~ 16 Hz associations for the higher γ frequencies (large dashed ellipse), suggesting that the higher γ frequencies are amplitude-modulated at ~16 Hz at lateral sites. C. An example from a representative anterior site (AFz) characterized by the lack of notable associations, suggesting that different frequency bands operate relatively independently at anterior sites. D-H. Association patterns at all sites. D. 2D-association plots. E-H. Corresponding line plots showing the association strengths for major frequency bands as a function of f. The line plots are shown separately for the θ-α (E), β (β₁, β₂, β₃) (F), γ₁ (G), and γ₂ (H) bands to reduce clutter. The lower panels show representative 2D-association plots to highlight the specific bands included in the line plots. E. Line plots of association strengths for the θ and α frequencies, peaking at Δf~ 3 Hz at posterior sites (especially in the regions shaded in gray). F. Line plots of association strengths for the β (β₁, β₂, β₃) frequencies, peaking at Δf~ 10 Hz at posterior sites (especially in the region shaded in gray). G. Line plots of association strengths for the γ₁ frequencies, peaking at Δf~ 10 Hz at posterior sites (especially in the region shaded in gray). H. Line plots of association strengths for the γ₂ frequencies, peaking at Δf~ 10 Hz at posterior sites and Δf~ 16 Hz at lateral sites (especially in the regions shaded in gray). In all parts the shading on the line plots represents ±1 standard error of the mean with participants as the random effect.