Test-retest Reliability at Network Level
To examine whether and how the reliability estimates of ANT varied among intrinsically organized networks, we summarized the region-level ICC estimates by the seven networks defined by the Schefer-Yeo atlas (Figure 4 , supplementary Table S6 ).
We used a linear mixed-effect model to examine whether the seven networks have differential reliability in alerting, orienting, and executive networks. There was a significant interaction between the intrinsic and attention network contrast (F (12, 386) = 2.327,p = 0.007). Simple effect analysis revealed that the frontoparietal, default mode, and limbic network showed differential reliability for the three attention network contrasts (supplementary Table S7 ). The frontoparietal network exhibited higher ICC on the orienting contrast than the executive contrast (t = 2.449, p = 0.039) and alerting contrast (t386 = 2.557, p = 0.029). The default network showed inferior reliability on the executive contrast compared with the alerting (t386 = -2.839, p = 0.013) and orienting contrast (t386 = -2.687, p = 0.021).
In addition, we also examined the interaction between the ANT condition (congruent, incongruent, center cue, spatial cue, no cue) and the seven intrinsic networks (F (24, 772) = 7.129, p < 0.001). Congruent and incongruent conditions showed higher ICC than the other conditions in all intrinsic networks except for the limbic network (supplementary Table S8) . At least half of the regions in dorsal attention, ventral attention, visual attention, and somatomotor network can have acceptable reliability for the congruent and incongruent conditions. Finally, comprehensive pairwise comparisons among the eight contrasts supported that ICC of congruent and incongruent conditions were higher than all other contrasts for all intrinsic networks except for the limbic network (supplementaryTable S9). The reliability of the estimates also corresponded well to the effect size (BOLD percent signal change) estimates at the intrinsic network level (Figure S5) .