Task-related functional connectivity study of attentional control in monkeys

We aimed to determine differences in top-down and bottom-up control of attention using monkey fMRI and psychophysiological interaction (PPI) analyses. Three macaque monkeys were trained to covertly detect a random (0.6 probability) dimming occurring with equal probability at one out of four locations at 6 deg. eccentricity in each quadrant (all objects 0.5 x 0.5 deg.). The target-dimming was cued either in a bottom-up or top-down manner, using a change in color of the target itself or the fixation point (symbolic cue for each quadrant), respectively. Target but not distractor dimmings (appearing at the 3 other quadrants) had to be indicated with a manual response. The monkeys were scanned using an event-related paradigm on a 3 T Siemens Trio scanner and an 8-channel receive coil (contrast-agentenhanced (Vanduffel et al., 2001), 1.25 mm isotropic voxels). We measured cue related fMRI activations using standard GLM and taskspecific network interactions using PPI. In addition, based on the betastrength of the PPI interactions, we assessed network metrics (node/edge betweenness centrality; node strength) (Rubinov and Sporns, 2010) to characterize the topology of functional connectivity networks. The behavioral performance of all monkeys was better for bottom-up compared to the randomly-interleaved top-down cued trials (higher percent correct; shorter reaction times). While the GLM-defined activation patterns revealed only very marginal functional differences between key nodes within the parieto-frontal attentional network, task-based functional connectivity analyses showed a stronger contribution of LIP during bottom-up and FEF during top-down trials. In conclusion, subtle functional differences across nodes participating in attentional control might be easier to detect using task-based functional connectivity analyses compared to traditional GLM analyses. Put otherwise, it is not necessarily the degree of fMRI activation in the individual nodes but rather the interaction across nodes within a functional network which determines their functionality in attentional control.status: publishe