Attention networks engage the default mode network related to policy precision under uncertainty

In decision-making, choices impact the present and cascade into future decisions, highlighting the importance of confidence when making a decision. Here, we investigated the meta-network level neural correlates of this confidence by estimating the instantaneous changes in precision in selecting actions and comparing them with brain state trajectories. To confirm the relationship between behavioral and neural signals, we leveraged inter subject correlation to determine how similar the group shared components of time-series of precision-triggered meta-network occurrences are across participants and how this similarity changes during decision-making. We found that policy precision is a proper behavioral signal to explain the meta-network level neural dynamics. It positively correlated with the default mode network (DMN) dominant state, the occurrence of which is mutually exclusive with the dorsal attentional network-(DAN) and frontoparietal network (FPN)-dominant state, the activation of which is speculated to be associated with a highly uncertain state and arises from increased integration between the DAN, FPN, and DMN. Therefore, supporting the novel perspective that the DMN may reflect internal beliefs, these findings indicate that their integration promotes the DAN and FPN to exert attention to decrease uncertainty.