, 2008, Schoffelen and Gross, 2009 and Siegel et al., 2008). Furthermore, we investigated functional http://www.selleckchem.com/products/CP-690550.html modulations rather than absolute levels of synchronization. This subtracted out the spatial pattern of synchronization induced by the limited spatial resolution that is common to any two conditions compared. Another crucial but often ignored problem is that interaction measures of neural population signals depend on the relative weighting of different signal components (Nunez and Srinivasan, 2006). Specifically, they depend on the weighting of the neural signal of interest relative to

noise and neural signals that are not of interest. Thus, even if the true interaction between the signal components remains constant, changes in the components’ amplitudes may alter their relative weighting and cause a change in the measured interaction between the population signals. We addressed this problem by comparing changes in synchrony to concurrent changes in signal amplitude (see Supplemental Experimental Procedures available online). Second, we devised a new analysis approach that allows for identifying networks of synchronized cortical regions (Figure S2 available online). In brief,

we employed permutation statistics to identify cortical networks as continuous clusters in a high-dimensional interaction space (see Experimental Procedures). This allowed for directly identifying networks across a full pairwise cortico-cortical space. We applied this approach to source-level coherence estimated from EEG (Gross et al., 2001), which quantifies the frequency-specific phase consistency between regions. This allowed us to effectively Ku-0059436 supplier image synchronized cortical networks across space, time, and frequency. Importantly, no a priori assumptions had to be made about the time and frequency of synchronization or about the number, size, location, and spatial structure of the synchronized networks. We first applied this network-identification approach either to contrast cortico-cortical coherence between the stimulation and baseline

intervals. This revealed a widespread but highly structured cortical network (Figures 3A and 3B, permutation-test, p = 0.0245) that showed enhanced beta-band coherence (15–23 Hz) during stimulation. The network consisted of a largely symmetric pattern of cortical regions spanning extrastriate visual areas implicated in the processing of visual motion as well as higher order association areas. Bilaterally, it included frontal regions consistent with the FEF, posterior parietal cortices along the intraparietal sulcus (IPS), lateral occipitotemporal cortices consistent with the middle temporal area (MT+), and medially extrastriate visual cortex near the transversal occipital sulcus (see Table S1 available online). Beta-band coherence in this network was enhanced for about 1 s around the time of bar overlap (Figure 3B).