This study was supported by NIH grants R01MH085828 (C.E.L.S., W.A.S.) and R01MH058847 (W.A.S.). “
“Two complementary principles, segregation of function and dynamic integration, coexist within the brain (Friston, 2002). Segregation of function has been recognized for well over 100 years. Thus, there exist different neural

systems specialized for sensory processing (in multiple modalities), motor control, and various aspects of cognition (e.g., attention, episodic memory, affective evaluation). Neuroimaging, first with positron emission tomography (PET), and later with functional magnetic resonance imaging (fMRI), has greatly advanced our understanding of functional segregation within the human brain. Remarkably, the topography of segregated functional systems can be discerned by analysis of spontaneous covariation of the blood oxygenation level (BOLD) signal acquired in the “resting state,” that is, in the absence Target Selective Inhibitor Library solubility dmso of imposed task structure (Biswal et al., 1997, Deco and Corbetta, 2011, Fox et al., 2005 and Smith et al., 2009). These topographies now are widely referred to as resting state networks (Beckmann et al., 2005). However, effective behavior depends on the dynamic integration of sensory, motor, and cognitive functions at multiple spatial and temporal scales (Tononi, 2004). The low find protocol temporal resolution of fMRI precludes studying integration

on a timescale well matched to behavior. Hence, our understanding of principles governing dynamic integration across anatomically segregated functional domains remains limited. Invasive electrophysiology (extracellular spike or local field

potential recording) offers a means of studying network interactions at high those temporal resolution (Engel et al., 2001) but is restricted in spatial coverage, which prevents simultaneous monitoring of multiple, distributed brain networks. Noninvasive electrophysiology (electroencephalography [EEG] and magneto-encephalography [MEG]) offers high temporal resolution and wide coverage, thereby enabling the study of network interactions on a behaviorally relevant timescale (Varela et al., 2001). Recent studies using high temporal resolution methods suggest that the apparent temporal stationarity of resting state networks (RSNs), as observed through the slow temporal filter of the BOLD signal, obscures a much richer structure both in the time- and frequency-domain. (Brookes et al., 2011a, Brookes et al., 2011b, de Pasquale et al., 2010, He et al., 2008, Liu et al., 2010, Mantini et al., 2007 and Nir et al., 2008). These complex spatio-temporal patterns theoretically provide the substrate for the integration of information within and across networks during active behavior. However, little information currently exists regarding how this communication manifests in the resting state. The present study builds on the observation that RSNs, when studied at high temporal resolution with MEG (de Pasquale et al.