Figure 2

Principle components confirmed that bilateral frontal and temporal lobe dipoles were necessary to explain >98% of the topographic variance. The 4-dipole-model [3] was fitted to the ERP grand mean for each group. Each dipole was initially constrained to the hypothesis and then allowed to move within the model without constraint. Stereotaxic coordinates were calculated with the method of Garneron [57] and group solutions placed on the modified Montreal Neurological Institute atlas. The residual variance (RV) and best fit for the BESA-calculated group dipole solution (RV<2%) is plotted in the middle. On the left of the figure, for the young patients (EOS [N = 28]) and age-matched controls (C-EOS [N = 22]), is the time course for activity of the four MMN dipole moments whose location is marked in the left/right frontal (LF/RF) and left/right temporal lobes (LT/RT) on axial, coronal and sagittal sections of the brain atlas in the middle and on the right. To improve the accuracy of source-analysis headshape was controlled by digitizing electrode sites relative to the nasion, left and right periauricular skull-landmarks with ultrasound (Zebris, Munich).