Item

Abstract

Attention allows our brain to focus its limited resources on a selected task. It does so by the selective modulation of neuronal activity of task-relevant cortical areas, and by the simultaneous change of communication between sets of regions. Previous fMRI evidence in the human brain showed selective increases in functional connectivity within visual cortex and between visual with fronto-parietal networks. However, these studies examined relatively brief attention periods that are affected by task-induced signal transients. We designed an experiment involving very long (2 min) trials of attention and rest that would allow us to discard initial transient periods (30 s), and to analyze slow (0.004-0.05Hz) and fast frequency (0.05-0.2Hz) bands of fMRI signals driving connectivity changes. We analyzed functional connectivity between visual regions, the dorsal attention network, and the resting state network. We found that attention increased long-distance connectivity between the dorsal attention network and visual regions and within the resting-state network. It decreased connectivity between resting-state and attention networks, but also between distinct visual regions and within distinct parts of the same visual region (left/right, dorsal/ventral parts). The change in connectivity strength was correlated with hierarchical distance, such that the increase between the dorsal attention network and visual cortex was more pronounced for higher than lower visual regions. Correspondingly, the decrease within visual cortex was the more pronounced the closer the hierarchical proximity was between neighboring regions, and was highest within regions. A frequency-segregated analysis showed that long-distance effects between dorsal attention network and visual regions were mediated by slow and fast fluctuations, whereas only fast fluctuations mediated de-correlation among visual regions. These results may pinpoint two fundamentally distinct effects of attention on connectivity. A long-range facilitation of information flow between distinct networks, and a short-range de-correlation within sensory cortex that may indicate and increase of information and decrease of redundancy.