NBS is a procedure for non-parametric multiple comparisons to investigate the differences of network connectivity (Zalesky et al., 2010, 2012). This approach is very promising because conventional mass-univariate tests often fail to find connectivity differences because they require adjustment for multiple comparisons (e.g. FDR correction). In our initial study (unreported), no significant differences were found from the pairwise comparisons of SL connectivity with 406 connections from 29 nodes and FDR p < 0.05, which potentially due to a weak contrast-to-noise ratio in the network (Zalesky et al., 2010). So NBS, a less conservative approach than FDR, was applied which detected a subnetwork consisting of fronto-centro-parietal connections (Fig. 4). It should be noted that no individual connections within this subnetwork were significant with connection-wise comparisons, but only the subnetwork as a whole was significant because NBS are effective only for connected and structured components.
