Similar observations have been reported using fMRI in human visual cortex by Nir and coworkers (2006). Interregional temporal correlations as seen through the lens of fMRI are largely stationary (but see Smith et al., 2012), due to the severe low-pass filter of the neurovascular coupling. In contrast, MEG BLP correlations are patently non-stationary (de Pasquale et al., 2010 and de Pasquale et al., 2012). Several “endogenous” mechanisms such as noise and synaptic delays (Deco and Corbetta, 2011 and Deco et al., 2011), neuromodulation (Marder,

2011), and cognitive activity have Sunitinib in vivo been proposed to explain the nonstationary dynamics of functional connectivity (Sporns, 2011). Here, we show that sensory inputs and their interaction with cognitive activity (event boundary detection) strongly affect the variability of temporal correlation in visual cortex. The power spectral density of α BLP correlation time course especially at the lower frequencies (0.01 < Hz < 0.2) was strongly enhanced by watching the movie as compared to visual fixation (Figure 8B). Furthermore, we observed a significant lagged correlation between the detection of event boundaries in the movie, as judged by an independent

group of observers (Figure 8F) and the α BLP correlation time course. This lagged correlation was not simply due to the chance temporal interaction of BLP and psychophysical time series. It was also not due to low level sensory changes in the movie Dasatinib as indicated by the analysis on the luminance time course. Our interpretation is that it may reflect longer time-scale (tens of seconds) Cytidine deaminase adjustments in functional connectivity induced by specific high-level sensory or cognitive events. This is in keeping with

the event segmentation theory (EST) (Zacks et al., 2007), in which event segmentation may be the product of an adaptive mechanism that makes prediction about upcoming information by means of the integration between sensory clues and previous knowledge about event parts or actor’s goals and plans. This could involve top-down mechanisms to visual cortex that could modulate the strength of ongoing BLP correlation. We do not believe though nonstationary fluctuations in visual cortex BLP correlation are directly involved in processing event boundaries; rather, they reflect the time-varying influence on cortical noise reduction of task-specific processes. It is as if visual cortex takes a short break after detection of an event boundary. This study provides a glimpse in the long-lasting adjustments of functional connectivity induced by natural vision on resting-state activity. RSN are similar to a complex space landscape formed by peak and valleys that slowly change over time.