With the intention of appreciating the different activations revealed by the EEG–fMRI analysis in terms of spatial maps, we also analyzed BOLD data with a classical approach (Linden et al., 1999), based on a rare/frequent contrast, and also with a more recent approach based on direct P300-BOLD correlation (Eichele et al., 2005; Horovitz et al., 2002). A general linear model (GLM) analysis (Friston et al., 1995) was performed for both of them, using different predictors. In the first case, the predictor was created convolving a time-course containing 0 for frequent events and 1 for rare events with the same HRF defined previously. In the second case, the P300 reference time-course was used as predictor. Fixed-effect analysis was used for obtaining group-level maps for both the rare/frequent contrast and P300-BOLD correlation methods. This allowed comparing the resulting maps, by means of the spatial correlation coefficient, with the ICA group maps, which were intrinsically based on a fixed-effect analysis.
