Abstract
Superior longitudinal fasciculus (SLF) is a major white matter tract connecting frontal and parietal cortices. Here we examined the role of SLF in the control of visual spatial attention. Diffusion MRI data were recorded from 19 subjects; from the same subjects, simultaneous EEG-fMRI data were also recorded while they performed a cued visual spatial attention task. In the task, each trial started with an auditory cue, directing subjects to attend to the left or right visual field. For validly cued trials, following a random cue-target interval (CTI), two rectangles were displayed, and the subjects were required to discriminate the orientation (vertical or horizontal) of the rectangle in the cued location. For invalidly cued trials, following a random CTI, one rectangle appeared in the uncued location, and the subjects were required to discriminate the orientation of the rectangle. The cue validity effect, which measures the behavioral benefits of deploying preparatory attention to the target location, was defined as invalid-cued RT minus valid-cued RT. The following results were found. First, SLF track-count showed a positive correlation with the cue validity effect. Second, SLF track-count was positively correlated with fMRI decoding accuracy (left versus right) within the dorsal attention network (DAN). Third, a statistical mediation analysis showed that SLF track-count was the mediator for the relationship between the neural representations in DAN and the cue validity effect. Fourth, subjects who showed larger alpha event-related desynchronization over the hemisphere contralateral to the attended location during the CTI were found to have larger SLF track-count. Taken together, these results suggest that SLF plays an important role in mediating the behavioral benefits of valid cuing in visual spatial attention, and it is doing so by enabling the neural representations of attended information in DAN and by facilitating the implementation of sensory biasing according to behavioral goals.
Acknowledgement: This work was supported by NIH Grant MH117991 and NSF Grant BCS-1439188. X. Hong was additionally supported by National Science Foundation of China (No. 61601294).