From “Using functional connectivity changes associated with cognitive fatigue to delineate a fatigue network”:
It is not entirely clear why increasing fatigue should be associated with decreasing connectivity between specific frontal areas while also being assocaited with increasing connectivity between frontal and more posterior regions. However, the hub regions identified here are part of several networks that are critical for the control of behavior: the salience network (SN, which includes the anterior insula)*, the reward network (RN, which includes the striatum and vmPFC) 47,48 and the cognitive control network (CCN, which includes the DLPFC and anterior insula)*. One possibilty is that as subjects repeatedly perform the tasks across successive runs, cognitive resources decrease. This leads to a change in the balance between the amount of reward subjects are receiving (which is simply the intrinsic reward of doing well) and the effort required, and this change is detected by the RN.
I guess the reward juice is in limited supply? In this study, subjects did recall tasks that were very similar to each other. I wonder if different kinds of tasks require different reward resources.