The voxel-wise map closely resembles the topography of the DMN as reported with fMRI. Some regions (e.g., right angular gyrus and dorsal medial prefrontal cortex) were used as input to the algorithm for determination of epochs of MCWs, whereas others (e.g., ventral medial prefrontal cortex and left angular gyrus) were found independently. Also note the paucity of correlation elsewhere in the brain with a few exceptions of cross-network correlation in visual cortex, and left sensorimotor cortex. A different topography is obtained by seeding left posterior intraparietal sulcus (IPS) with right posterior IPS and right

frontal eye field (FEF) as additional inputs to the algorithm. Specific correlation selleck chemicals is observed in left FEF, a node that is not constrained (Figure 1C). The topography resembles that of the dorsal attention network (DAN) as described in fMRI. Control analyses

(see Supplemental Information and Figure S2A) show that the algorithm does not produce RSN as an artifact using arbitrary combinations of nodes not corresponding Ulixertinib ic50 to fMRI-derived RSNs (“fake RSNs”). Figure 1D shows the group-average topography of the wide band (1–150 Hz) BLP temporal correlation maps for six RSNs projected onto an inflated brain atlas surface (Van Essen et al., 2001). Each map represents within-network correlation evaluated during that network’s MCWs. The same results are also shown in axial view in Figure S2C. Importantly these maps contain regions that are significantly correlated in terms of wide-band BLP not very only across subjects, but also across nodes. The six networks may be listed as follows: the default mode network (DMN), the dorsal attention network (DAN), the ventral attention network (VAN), plus language

(LAN), somatomotor (SOM), and visual (VIS) networks. Note that each MEG BLP network exhibits topography consistent with the resting state fMRI literature (Cordes et al., 2000, Cordes et al., 2001, Damoiseaux et al., 2006, De Luca et al., 2006 and Greicius, 2008). Similar MEG networks have been reported in other recent resting-state studies (Brookes et al., 2011a, Brookes et al., 2011b, de Pasquale et al., 2010, He et al., 2010 and Liu et al., 2010). Next, we inquired whether the six presently studied RSNs are equivalent as regards cross-network BLP correlation dynamics. For each network, and for each frequency band, cross-network correlations were computed on each network’s MCWs. These quantities then were averaged over appropriate node pairs. The results, expressed as Z-score matrices, are shown in Figure 2 (see Experimental Procedures and Supplemental Information for details).