Abstract
Over a decade ago, Maurizio Corbetta and Gordon Shulman introduced the concept of the dorsal attention network, a set of regions that mediate top-down attention. Since then, the dorsal attention network has been reliably observed across studies of goal-directed attention. Nevertheless, there exists great variability in the location and degree of activation across individuals. The neural architecture that may underlie an individual's specific and idiosyncratic activation pattern remains unexplored. We hypothesized that an individual's connectivity pattern may be strongly associated with their particular pattern of activity, since connectivity is the principle neural component that defines the computational domain of a brain region. Here we used the pattern of intrinsic functional connectivity of an individual to model and predict the location and activation strength of their own dorsal attention network. We used the modeling approach first described by Saygin et al. 2012 and later extended in Osher et al. 2015 and Tavor et al. 2016. This model, which links activations patterns of single voxels as a function of connectivity to the rest of the brain, can be applied to other attentional tasks that are capable of recruiting the dorsal attention network, such as multiple object tracking, change detection, or visual 2-back tasks. This demonstrates that connectivity can produce comparable predictions of an individual's dorsal attention network as various localizers. Lastly, an analysis of the final model coefficients describes the connectivity patterns that best define an individual's dorsal attention network.
Meeting abstract presented at VSS 2017