by WMNLS in the source space, defined by a 3-dimensional regular grid with 4mm step. The brain areas included in all the task-related fMRI network maps are segmented, dilated by 5mm to account for the difference between electrophysiological and hemodynamic activity, and weighted at 140% with respect to all other locations [Liu et al., 1998]. The average EEG power in each area belonging to the fMRI networks is used for evaluating the spatial distribution of the sources associated with the ERP-IC time-course. For each ERP-IC, the noise level across the brain, calculated projecting to the source space the sensor covariance matrix of the noise, is used to scale the spatial maps to z-scores [Dale et al., 2000]. Finally, the statistical significance of the ERP activations is performed by means of permutation testing [Maris and Oostenveld, 2007].
