Abstract
Recent work by Gmeindl et al.1 showed that the medial superior parietal lobule (mSPL) is transiently engaged during both cue-driven and uncued (purely voluntary) shifts of spatial attention. We applied multivoxel pattern classification (MVPC) to their data, training the classifier on the cued shift events and testing the classifier on the uncued shifts, and found that spatiotemporal patterns of BOLD activity in mSPL were consistent for both cued and purely voluntary shifts of attention. To further examine the neural basis of attentional control, we used MVPC within mSPL to compute a decision time series that reflects, on a moment by moment basis, the degree of shift-like activity. We then entered this decision time series as a regressor in a univariate (voxelwise) GLM analysis. This analysis revealed significant covariation in several regions, including bilateral caudate and prefrontal cortex. We compared this procedure to a whole brain correlation analysis using the mean mSPL BOLD timecourse as a seed region. Only a subset of the regions identified in the correlation analysis was identified with the multivariate decision time series analysis. Furthermore, we identified task-related regions that were not revealed in a standard univariate GLM using regressors denoting only the time of task events. This novel multivoxel decision time series analysis thus provides both greater specificity than a conventional correlation, and greater sensitivity than a standard GLM analysis. The results suggest that the control of spatial attention extends beyond the well-known frontoparietal network to include the basal ganglia.
Gmeindl, Gao, Yantis, & Courtney (2008, November). Neural mechanisms of purely voluntary covert shifts of attention. Paper presented at the 38th annual meeting of the Society for Neuroscience, Washington, D.C.