rate of 500 or 512 Hz, based on the version of the Neuroscan collection system used (resampling was performed at 256 Hz—see below). The preprocessing was performed using custom scripts in Matlab (The MathWorks, Inc., Natick, MA). The following steps were performed on the entire EEG sweep: (i) data points were resampled to 256 Hz for harmonizing different sampling rates; (ii) bandpass filtering at 0.05–50 Hz to keep only the frequency range of interest; (iii) waveforms were “detrended” to remove upward/downward trending; and (iv) “de-meaning” was done by subtracting the gross mean from each data point in order to align the waveforms close to the zero-amplitude baseline. Then, the continuous data was segmented into 2 second epochs. Another batch of preprocessing steps were performed on each of the epochs: (i) detrending; (ii) baseline alignment by subtracting epoch mean from each data point; (iii) interpolation of missing data or “flat” channels by computing mean of surrounding nearest channels; (iv) removal of epochs with DC shift/drift involving voltage steps higher than 75 mV between any two adjacent sampling points; and (v) removal of possible EOG contaminated epochs if any data point was beyond the threshold of ± 100 μV or if the
