The default mode network (DMN) and semantic network (SN) are two of the most extensively studied systems, and both are increasingly used as clinical biomarkers in neurological studies. There are strong theoretical reasons to assume a relationship between the networks, as well as anatomical evidence that they might rely on overlapping cortical regions, such as the anterior temporal lobe (ATL) or angular gyrus (AG). Despite these strong motivations, the relationship between the two systems has received minimal attention. We directly compared the SN and DMN using a large (n = 69) distortion-corrected functional MRI (fMRI) dataset, spanning a range of semantic and nonsemantic tasks that varied input modality. The results showed that both networks fractionate depending on the semantic nature of the task, stimulus type, modality, and task difficulty. Furthermore, despite recent claims that both AG and ATL are semantic hubs, the two areas responded very differently, with results supporting the role of ATL, but not AG, in semantic representation. Specifically, the left ATL was positively activated for all semantic tasks, but deactivated during nonsemantic task performance. In contrast, the left AG was deactivated for all tasks, with the level of deactivation related to task difficulty. Thus, ATL and AG do not share a common interest in semantic tasks, but, rather, a common "disinterest" in nonsemantic tasks. The implications for the variability in the DMN, its cognitive coherence, and interpretation of resting-state fMRI data are discussed.semantic network | default mode network | distortion-corrected fMRI | anterior temporal lobe | angular gyrus T wo substantial bodies of research literature, spanning cognitive and clinical neuroscience fields, have been dedicated to exploring the function and components of the semantic network (SN) and the default mode network (DMN). The DMN is an anatomically defined network that shows task-related deactivation during many goal-directed tasks (i.e., rest > task) (1) and can be reliably delineated using techniques such as independent components analysis (ICA) of resting-state functional MRI (fMRI) (2). The SN is a fronto-temporo-parietal network that is sensitive to semantic content in comparisons of semantic tasks > rest/nonsemantic control tasks (3). Although investigations of the DMN and SN have been primarily independent of each other, there are good reasons to compare the two networks directly. First, the networks might share common cognitive functions. One prominent theory suggests that during "rest," the brain is engaged in the activation of rich conceptual representations, and thus default-mode processing places strong demands on the semantic system (4). Secondly, the DMN and SN engage some common anatomical areas. The DMN consistently includes medial prefrontal cortex, parietal areas [angular gyrus (AG), precuneus, posterior cingulate cortex (PCC)], and, somewhat more variably, the lateral anterior temporal lobe (ATL) and hippocampus (1, 5). Some of these areas are considered...