| dc.description.abstract |
In our environment, there are so many things to see, e.g., computer screen, buildings, trees and cars in the street. In this busy scenery, we do not process all the information equally, but rather filter out some information and focus more on certain characteristics in the whole scene. In this process, attention plays an important role, and underlying neural correlate is the matter of interest.
We focus on investigating how attention changes the connectivity of the fMRI signal in the human brain. Prior studies examined this question, yet most studies used short trial interval (<20s) in examining the connectivity during attention. The short trial interval excludes the slow fMRI fluctuations (<0.1Hz) that showed segmented connectivity structure in the resting-state studies supported by the neurophysiological observations.
In the thesis, we introduce an ultra-long trial (2-3mins) to examine connectivity during task conditions, in attention demanding task. In the first study, we asked whether trial length affects the functional connectivity (FC) strength in general during attention task compared to visually matched condition as control. We observed that the long trial interval (2mins) condition showed nearly twice the FC strength compared to short traditional trials (20s). Moreover, attention reorganized the FC as enhanced positive FC between dorsal attention network (DAN) and visual network (VIS) and decreased negative FC between default mode network (DMN) and DAN/VIS, but reduced positive FC within VIS. Notably, the reorganization is frequency dependent: FC changed relied more on slow frequency (0.004-0.05Hz) for the connection between DAN and VIS and high frequency (0.05-0.2Hz) for decorrelation within VIS.
In the second study, we addressed the question whether FC strength relies on visual hierarchical distance in visual processing and attention task. We observed a gradient of connectivity, such that DAN connected strongly with high visual region (e.g., V5/MT) that degrades towards lower visual region (e.g., V1). A reversed effect was observed between DMN and VIS, revealing that DMN connected strongly negatively with high visual region that degrades its negative connectivity strength towards lower visual region. More interestingly, we implemented general linear model to the FC strength that showed attention modulates multiplicatively and addictively the connectivity strength along this visual hierarchy.
In the third study, we observed how attention changes the connectivity in different features, e.g., color and motion attention. Here, we used seed-to-whole brain connectivity with regressing out the mean signal from the whole brain. First, we observed that V4 and V5/MT selectively connected to the task positive network, including DAN and visual regions, and negatively connected to the DMN. Then, feature-specific analysis showed that color compared to motion attention, selectively connects the V4 to DAN more than V5/MT to DAN, with selective negative connections between V4 and DMN than V5/MT and DMN. This suggest that feature-based attention led the brain communicate specifically cooperative (positive) way, but also competitive (negative) way.
Taken together, attention not only reorganizes the connectivity in frequency dependent way, modulates differentially along the visual hierarchy as well as feature-specific manner. More interestingly, our results showed advantages of using long trial block experiment to detect important network connectivity change during attention. Not only applying frequency dependent analysis, but implementation of the GLM in comparing conditions, as well as, regressing out the mean signal from the whole brain for seed-to-whole brain connectivity analysis. All these methods that is used in the thesis can be extended to examine brain connectivity structure noninvasively, that may show important findings in other cognitive tasks, such as decision making or memory tasks. |
en |
