Neural dynamics underlying varying attentional control facing invariant cognitive task upon invariant stimuli.

Even when performing invariant behavioral task repeatedly on invariant physical stimuli, our behavioral performance always changes as manifested in varying response times (RTs), which is associated with fluctuations in attentional control and thus the underlying self-organization states of the human brain. In a visuospatial task of the present fMRI study, physical stimuli differed across six levels of spatial scope, but were kept invariant within each level. The slower RTs with larger spatial area attended suggested higher demands on visuospatial attention. The slower RTs within each level, however, implicated worse attentional control since both the task and the physical stimuli were kept invariant within each level. The imaging results showed that slower RTs within each of the six levels were associated with higher but later activations in the frontoparietal network, and higher but later deactivations in the default-mode network (DMN). These findings thus for the first time suggested that the within-level variance of attentional control corresponded to dynamic changes in the frontoparietal network and the DMN, in terms of not only the height but also the latency of neural activity. Moreover, although the two networks are anti-correlated in terms of the height of neural activity, they are tightly coupled in terms of the temporal dynamics. Based on the current results, we proposed a tentative hypothesis on the optimal working mode of the frontoparietal attentional control system in the human brain: even a lower height of neural activity in frontoparietal network can significantly improve behavioral performance as long as it starts relatively early.