Next, we computed voxel time series data from the peak voxels in each region, per participant and response, in order to probe for group differences. Theta, alpha, and gamma all showed symmetric bilateral clusters (one cluster per hemisphere), and since we did not have any laterality hypotheses, we averaged the time series across the hemispheres per oscillatory response in each participant prior to statistical analyses. This resulted in one time series per response, per participant. We then averaged the frequency power across the active window, and ran ANCOVAs to test for the effects of alcohol and cannabis use disorder. We utilized a 2 × 2 design to allow for the investigation of each main effect while controlling for the variance associated with all other variables. For example, the main effect of AUD therefore tests participants with AUD against those without AUD, controlling for CUD, sex, and age. For alpha oscillatory activity, ANCOVA revealed a significant main effect of AUD (F(1,40) = 8.12; p = 0.007; η2 = 0.163) but not CUD (F(1,40) = 3.11; p = 0.082), which indicated that
