2.12. Estimating participant-specific ERSP at each domain

The ERSP matrix of each IC was 200 cells along the time domain (1.26 s before and 7.26 s after the heat onset, truncated by the sinusoidal wavelet transformation), and 100 cells along the frequency domain (3 Hz – 90 Hz) on a logarithmic scale. For a given domain, for each participant per condition, a mean ERSP matrix was developed by first choosing all participant-specific dipoles that contributed to the domain and then averaging across the dipoles. Permutation t-test with 1000 samples were applied to the subject-mean projected ERSP measures to determine significant differences between groups for each condition as done in our previous work (Chung et al., 2017; Misra et al., 2017b). Between group contrast plots (time x frequency) were therefore generated for the low pain and moderate pain conditions for each domain. All p-values from all contrast plots for all domains were then concatenated and simultaneously corrected using the FDR correction to stringently control for multiple comparisons (Benjamini and Hochberg, 1995). That is, following the permutation tests, p-values from each matrix (100 × 200 elements = 20,000 elements per contrast) for each contrast plot (2 per domain) for all domains (5 domains) were concatenated and simultaneously corrected using the FDR correction. P-values were therefore corrected across 200,000 elements.