The capacity to identify aspects of meaning that overlap across multiple concepts may relate to individual differences in the strength of intrinsic connectivity within and between distinct brain networks supporting semantic cognition. This study examined a semantic summation task, which tested the capacity to detect weak overlapping aspects of meaning,in 76 participants who were also scanned with resting-state fMRI. We examined associations between summation and the intrinsic connectivity of semantically-relevant default mode and control network regions. These networks are implicated in information integration and controlled retrieval respectively. We found higher intrinsic connectivity between default and control networks was associated with better performance in the summation task. The same pattern of coupling between semantic default mode and control networks was not associated with more efficient retrieval of individual weak as opposed to strong associations in an additional cohort of around 200 participants, suggesting this pattern is specific to the summation of multiple concepts, rather than semantic task difficulty. Finally, higher connectivity within the default mode network was associated with better performance when selecting a word that was strongly-related to a single probe item, supporting the role of this network in more automatic aspects of semantic retrieval.