Due to the small spatial volume of the STN, we expect some variability in the precise location of the contacts within the STN with some contacts possibly even being outside the STN borders. Nevertheless we decided not to bias results by pre-selecting the channels with the strongest spectral power or PAC as we observed that these could occur at different channels. We therefore averaged all measures across all three bipolar channels for each electrode, resulting for each subject in mean values for both left and right STN, and ON and OFF medication conditions, when data were available. For visualisation of the grand-average power spectral densities, we employed robust averaging to eliminate the influence of outliers (Wager et al., 2005).
