Primary results from the current study revealed reduced negative connectivity between the fronto-parietal network and default mode network in OCD. Patients also showed more positive connectivity between fronto-parietal seeds and several areas outside DMN, including somato-motor areas (pre-postcentral gyrus and posterior insula) and other FPN regions (anterior insula/frontal operculum and lateral frontal cortex). However, not all group differences reflected increased connectivity in OCD, as patients showed less positive connectivity than control subjects within
DMN itself, although these effects only emerged at lower levels of significance. Negative connectivity among controls was evident despite the fact that global normalization was not used in the analysis 
, and group differences were not due to head motion 
. Importantly, the factorial design gave us greater ability to control for confounding factors frequently co-occuring in OCD, such as the presence of medication and depression, and post-hoc multiple regression analyses indicated that these variables did explain group differences. Overall, these data suggest that intrinsic functional relationships between large-scale brain systems are altered in OCD, specifically pointing to a disruption of the competitive interactions typically found at rest between fronto-parietal and default mode networks.
Research on DMN has garnered much interest since the initial observation of reduced blood flow within several anatomically-widespread brain regions during cognitive tasks versus passive viewing 
. Subsequent studies have associated this network with a core set of regions including PCC/precuneus, ventral and dorsal medial frontal cortex, posterior inferior parietal cortex, and medial and lateral temporal cortex 
. Activation in these regions not only decreases during externally-directed attention, but also increases when attention is directed inward, such as when subjects think about their own personal qualities and take the perspective of others (theory of mind/mentalizing), remember personal episodes from their past, imagine future events, and construct scenes 
. Experiments investigating the relationship between fronto-parietal network regions involved in executive/control processes and DMN indicate that efficient processing of external stimuli requires the suspension of introspectively-oriented cognition 
, and that this inverse pattern of activity appears even when the brain is at rest 
Recent evidence has identified dissociable sub-networks within FPN, with dorsal regions of anterior insula and medial frontal cortex forming a “core" network involved in implementing and maintaining attention to external task demands and detecting salient events 
, and lateral frontal and parietal regions performing more “executive" functions such as working memory, planning, and cognitive control 
. Right insula in particular may function as a “switch" between modes of processing, triggering the activation of executive regions and suppression of default mode regions when salient external events are detected 
. While anterior insula may respond somewhat generally to external task demands, it does appear to be particularly sensitive to stimuli signaling potential risk 
; as such, this area may be important for switching attention away from an internal focus toward the external environment after detecting potentially harmful situations. Though speculative, the current findings of attenuated negative connectivity between anterior insula and DMN suggest a potential neural basis for the difficulty OCD patients may have in efforts to disengage from internal mental processes in order to respond more appropriately to salient external information related to potential risk (such as that informing them that dreaded events have not or will not occur). These findings extend work identifying greater connectivity between anterior insula and VMPFC during task 
to include alterations in connectivity with several regions of DMN at rest, and raise the possibility that abnormal intrinsic connectivity contributes to the DMN hyperactivity found in OCD during cognitive tasks 
Connectivity differences with FPN occurred across a number of DMN regions, including PCC, pIPL, DMPFC/aMFC, and parahippocampus. Despite the consistent activation of these brain regions across several different tasks involving internally-directed cognition 
, there appears to be some measure of dissociation between different nodes within DMN, particularly between anterior medial frontal and posterior parietal/temporal regions 
. Although the current study was not aimed at distinguishing among different DMN subsystems, the findings of aberrant connectivity between FPN and several DMN regions suggests that altered resting-state connectivity in OCD is not specific to a particular DMN subsystem.
OCD patients also showed greater positive interactions between FPN seeds and several areas outside of DMN, including pre-postcentral gyrus and posterior insula, which are part of a somatosensory/interoceptive and motor network 
. Increased connectivity between fronto-parietal regions and somato-motor network could contribute to obsessive-compulsive symptoms involving sensitivity to physical sensations, particularly those related to disgust or internal urges, although such interpretations need to be tested directly. The finding of hyper-connectivity between anterior insula and thalamus is also of interest given the importance of the fronto-striatal-thalamic circuitry in OCD 
. Whereas positive connectivity between these regions was greater in the patient group overall, it was inversely related to OC symptom severity, suggesting a compensatory mechanism because patients with lesser severity of symptoms showed the greatest difference from controls. However, this finding should be interpreted with caution as we did not correct for the number of tests that were performed to examine correlations with symptom severity.
Although the general absence of correlations with Y-BOCS scores might seem surprising, this may indicate that group differences reflect stable biomarkers of OCD not sensitive to symptom severity differences, similar to the mechanism suggested for the error-related negativity 
. The current analysis focused on identifying connectivity patterns where OCD patients differed from both healthy controls and remitted depressed subjects (see and Table S2
). However, the current data cannot address whether similar disturbances would be found in patients with active major depression, or whether remitted depressed patients would show differences in connectivity compared to healthy controls in brain regions other than those explored in the current analysis.
Unlike group differences found with FPN seeds, OCD patients showed less positive connectivity than controls subjects within DMN, a finding that was not due to differential motion. Although this effect was found only at a lowered threshold, it is consistent with two prior reports 
, and is intriguing given the finding that negative affect reduces connectivity within DMN 
. These results suggest that OCD may be characterized by a complex pattern of hyper- and hypo-connectivity among large-scale networks.
To our knowledge, this is the first report of altered intrinsic connectivity between distributed regions of fronto-parietal and default mode networks in OCD. However, there are several limitations of the current study, many of which could be addressed by future research. Given that results were not corrected for multiple seed comparisons, replications using a larger sample size are necessary. In addition, OCD and controls groups were not matched on generalized depression/anxiety and education levels. Although post-hoc inclusion of these factors in multiple regressions indicated that these effects were not driving the reported group differences in connectivity, future studies would benefit from investigating the effects of these variables on brain connectivity. Of particular interest for the study of psychopathology, hyperactivity in DMN has been identified in other psychiatric disorders (for a review, see 
), and the current investigation cannot address whether the findings are specific to OCD. It is possible that changes in DMN across a range of dysfunction could be due to abnormal relationships between FPN and DMN, which would be consistent with the emerging recognition that overlapping physiological mechanisms are found among comorbid disorders 
. Many anxiety disorders 
, including OCD 
, are characterized by increased activation of anterior insula; as such, the hyper-connectivity between insula and DMN revealed by the current study may be more related to an anxiety phenotype than other psychopathologies, although direct comparisons between disorders will be needed to test this hypothesis. In addition, although we did not have enough subjects to examine different symptom dimensions in OCD, it is likely that some DMN-based cognitive processes (e.g., scene construction and imagination) are more directly related to certain OCD subtypes (e.g., those involving intrusive imagery) than others (e.g., those with symmetry/ordering concerns). Future work should be aimed at obtaining a large and diverse cohort of patients to determine whether these alterations in rs-fcMRI differ based on symptom dimension. Finally, although interpretations regarding the functional significance of resting-state interactions rely on a large body of literature in cognitive neuroscience, it will be necessary to directly test the relationship between rs-fcMRI and behavior as well as task-evoked activity in order to elucidate the cognitive processes associated with these abnormal interactions. In particular, assessment of subjects' thoughts or feelings during rest will improve the ability to link resting-state connectivity to behavior. Despite these issues, the current results highlight the importance of inter-regional interactions between large-scale networks in OCD, revealing alterations of connectivity that may provide promising leads for the development of novel treatments using behavioral training or neuromodulation to target specific patterns of dysfunctional connectivity. Future work will be aimed at replicating and extending the current findings in order to determine the generality of the effect and its behavioral significance.