In summary, analyses of temporal dynamics of BLP correlation reveal two important temporal properties about resting-state networks. First, cross-network interactions occurs between one network in a state of strong internal correlation, and a subset of nodes of another network that is not strongly coherent at that moment. BMS-777607 supplier It appears that some nodes can break away from their usual RSN and transiently correlate with one of the networks that tend to cross-interact, especially DMN. Second, networks spend a variable fraction of time in a state of high internal correlation, and this property seems to

inversely relate to their tendency to couple with other networks. Interestingly, the DMN, the most interacting network, spends on average less time in a state of high internal correlation (20% in α; 36% in β) than the VAN (53% in α; 56% in β), the least interacting network. This result is remarkable given these two networks are topographic neighbors yet display Selleckchem Bortezomib very different patterns of temporal interaction. We used MEG to examine the nonstationary properties of band limited power (BLP) time series correlation within and across functional networks defined by prior fMRI studies. Six segregated RSNs (DMN, DAN, VAN, visual, somatomotor, and language), showing topography similar to that fMRI RSNs, were recovered by computing voxel-wise BLP temporal correlation maps. Correlation maps for

each network were computed in epochs Rolziracetam of strong within-network correlation (MCWs). The dynamics of network interactions were studied during each network’s MCWs. Of all networks, the DMN showed the highest degree of cross-network interactions and this property was especially pronounced in the β (14–25 Hz) band. Among all DMN nodes, the PCC was the region manifesting the highest degree of cross-network interaction. This interaction involved subsets of nodes from other networks during periods in which their internal correlation was low. More generally, different networks exhibited different degree of temporal nonstationarity that appeared to be inversely related

to the degree of cross-network coupling. The following discussion considers three main issues: (1) the dynamics of functional segregation and integration of RSNs; (2) the DMN, and the PCC in particular, as a functional core of the brain; and (3) the significance of β band rhythms in functional integration. First, we consider some methodological factors that may potentially influence our findings. Studying the covariance structure of spontaneous cortical activity with MEG is challenging for several well-known reasons. MEG data are often contaminated by several artifacts including physiologic noise (respiration, heart), head and eye movements, and environmental noise. The impact of artifacts is important in resting state studies because averaging in phase with events cannot be used to improve the signal-to-noise ratio.