The Detection of Phase Amplitude Coupling during Sensory Processing.

The structure of the engaging sensory paradigm. Each trial started with a 1,500, 2,500, or 3,500 ms baseline period in which a square black box (the “porthole”) was centrally presented. This was followed by presentation of the visual grating stimulus (two cycles/degree) around the central porthole for 1,500 ms. A picture of an alien (target) or astronaut (non-target) was then shown within the porthole for 500 ms. Participants were instructed to respond after the appearance of an alien picture (maximum response time: 1,500 ms). Correct or incorrect responses were conveyed to the participant through audio-visual feedback in which the porthole turned green (correct) or red (incorrect) and a correct/incorrect tone was played. The times corresponding to the analyzed baseline and visual grating periods are labeled in orange/blue, respectively.

Illustration of the phase amplitude coupling (PAC) analysis procedure. The V1 time-series were filtered to obtain estimates of phase and amplitude, using a narrow (±1 Hz) bandwidth for the phase and a variable bandwidth (±0.4 times the center frequency) for the amplitude. Phase and amplitude information were obtained via the Hilbert transform. The coupling between phase and amplitude was then quantified using Mean Vector Length, Kullback-Leiber, or Phase Locking Value algorithms to produce a Modulation Index value.

Whole-brain oscillatory power changes following the presentation of the visual grating are marked by (A) increases in the gamma-band (40–60 Hz) and (B) decreases in the alpha-band (8–13 Hz), localized primarily in the ventral occipital cortex. Power maps were thresholded at a value which allowed prominent patterns of power changes to be determined, indicated by the white dotted line. Time-courses were extracted from bilateral visual area V1, defined using the atlas region shown in (C) from the HCP-MMP 1.0 parcelation (Glasser et al., 2016). (D) These V1 responses showed reductions in alpha/beta power and increases in gamma-band (40–70 Hz) power.

Phase-amplitude comodulograms produced by statistically comparing modulation index (MI)-values from 300 to 1,500 ms post-grating onset to a 1,200 ms baseline period, using four separate approaches. Comodulograms for (A) raw MI values and (B) MI values normalized by surrogate data are shown separately. The black dotted line represents significantly different phase-amplitude coupling frequencies (p < 0.05; for details of non-parametric cluster-based statistics see Section Methods).

Results of the simulated PAC analysis. (A) Phase-amplitude comodulograms produced using the MVL-MI-Canolty, MVL-MI-Özkurt, PLV-MI-Cohen, and KL-MI-Tort algorithms were able to successfully detect the 1.2 s of simulated coupling between 10 Hz phase and 50–70 Hz amplitude. (B) The coupling between 10 Hz phase and 60 Hz amplitude was calculated as a function of simulated data trial length. For trial data under 1 s, all four algorithms produced artificially inflated PAC.