The present study examined differences between the ASD and control groups in default network intrinsic functional connectivity. Both groups showed robust connectivity throughout the default network. However, relative to controls, the ASD subjects showed alterations in functional connectivity. Specifically, the ASD group showed weaker connectivity than controls between the posterior cingulate cortex and the right superior frontal gyrus. In addition, the ASD group showed stronger connectivity relative to the control group between the posterior cingulate cortex and two areas: the right temporal lobe and the right parahippocampal gyrus. Moreover, in an exploratory analysis, the extent of altered connectivity within the ASD group was associated with core ASD symptoms. Within the ASD group, poorer social functioning was associated with weaker connectivity between the posterior cingulate cortex and the right superior frontal gyrus. Finally, more severe restricted and repetitive behaviors were associated with stronger connectivity between the posterior cingulate cortex and the right parahippocampal gyrus.
Unlike the present work that examined functional connectivity at rest, there are a growing number of studies that documented functional connectivity as ASD subjects performed specific tasks. Many of these task-driven FMRI studies found that subjects with ASD had weaker connectivity relative to controls (Villalobos et al., 2005
; Welchew et al., 2005
; Kana et al., 2006
; Just et al., 2007
; Kleinhans et al., 2008
; Koshino et al., 2008
), but others showed that ASD is associated with stronger connectivity between regions (Mizuno et al., 2006
; Turner et al., 2006
). The present study extends these findings by showing that ASD subjects demonstrate both weaker and stronger connectivity during rest.
Two other published studies examined functional connectivity during rest in an ASD sample. The first study evaluated a large sample of individuals (57 with ASD and 57 controls) during multiple 24 second rest periods between blocks of tasks (Cherkassky et al., 2006
). Relative to the control group, the ASD group showed weaker connectivity between the ventral anterior cingulate cortex and the posterior cingulate and the ventral anterior cingulate cortex and the precuneus. Moreover, when connectivity between the parahippocampal gyrus and the other structures of the default network was averaged together, the ASD group showed overall weaker connectivity. The discrepancy in findings between the present study and Cherkassky et al (2006)
may be due to differences in the procedures. Specifically, whereas the present study acquired resting connectivity data over a sustained period of time (10 minutes), Cherkassky et al examined connectivity during multiple 24 second periods of rest between blocks of a task. Further work is necessary to understand the functioning of the default network over long versus short periods of time.
In the second study that examined ASD functional connectivity during rest, Kennedy and Courchesne (2008)
acquired FMRI data over a sustained period of time (7 min, 10 s). In the analysis, activation from three seed ROIs (posterior cingulate cortex, medial prefrontal cortex, and angular gyrus) were averaged together and group differences in connectivity between this averaged activation and the default network were examined. Relative to controls, the ASD subjects showed weaker connectivity selectively in the medial prefrontal cortex and the left angular gyrus. Although the analytic procedures differ from the present work, the findings are somewhat complementary. Kennedy and Courchesne (2008)
found selective areas of weaker connectivity when multiple seed regions were combined. In the present study, we found a selective area of weaker connectivity and two selective areas of stronger connectivity when only the posterior cingulate cortex was used as the seed. Task driven FMRI studies have found that subjects with ASD have areas of weaker connectivity (Just et al., 2007
; Koshino et al., 2008
) and stronger connectivity (Mizuno et al., 2006
; Turner et al., 2006
). Kennedy and Courchesne (2008)
shows that ASD is related to selective areas of weaker connectivity. The present study builds on this finding and indicates that ASD is associated with specific areas of weaker connectivity and also specific areas of stronger connectivity.
Task-driven FMRI studies show that social functioning relies on multiple neural structures, including the amygdala, fusiform gyrus, anterior cingulate and ventral prefrontal cortex (Winston et al., 2002
; Eisenberger, Lieberman and Williams, 2003
; Iacoboni et al., 2004
). Moreover, in ASD, social impairment is associated with altered functional connectivity, activation and morphology within these same areas (Ohnishi et al., 2000
; Dapretto et al., 2006
; Munson et al., 2006
; Nacewicz et al., 2006
; Kleinhans et al., 2008
). The present findings suggest that components of the default network, notably the posterior cingulate cortex and superior frontal gyrus, also may play a role in social function in ASD. Since these results were found in the absence of a task, it is possible that connectivity between the posterior cingulate cortex and superior frontal gyrus may relate to some aspect of offline processing of social function, such as memory consolidation involving social interactions, or maintenance that supports social processing.
Relative to social function, little work has examined brain correlates of restricted and repetitive behaviors in ASD. One functional imaging study reported a negative correlation between left superior temporal sulcus activation in response to task of biological motion and the repetitive behavior score (Freitag et al., 2008
). In addition, two studies linked disturbances in anterior cingulate function and restricted and repetitive behaviors (Shafritz et al., 2008
; Thakkar et al., 2008
). Turning to the present study, further investigation is required to understand the finding that connectivity between the posterior cingulate cortex and parahippocampal gyrus correlated positively with restricted and repetitive behaviors. The hyper-connectivity may give rise to these behavioral abnormalities. On the other hand, the connectivity patterns may be the result of the restricted and repetitive behaviors; for example, increased connectivity may be a neural manifestation of a constant effort to control these behaviors.
There are at least four limitations to the present work. First, the procedures used in the present study for examining resting connectivity make it impossible to know precisely what processes underlie the group differences in the default network. It has been persuasively argued that the default network during rest is primarily involved in intrinsic processes that are independent of awareness (Fox and Raichle, 2007
; Raichle and Snyder, 2007
; Buckner et al., 2008
). Nevertheless, it is possible that there may be some systematic differences between groups in stimulus-independent thoughts that underlie our present findings. Second, as in other fMRI studies, the degree of connectivity merely indexes correlations in activation between regions. Thus, these results do not demonstrate that there are group differences in how brain structures communicate with one another. Moreover, motion, physiology, and activation of the seed may contribute to the group differences in connectivity. However, there were no group differences in motion; physiological variations were regressed out of the time series data; and the time series data from the seed were normalized. Thus, these potential influences are unlikely to be factors in the present study. Third, the sample size in the study was small. Therefore, these findings should be considered preliminary until replicated. And, fourth, most subjects with ASD received psychotropic medication. Use of medication is exceptionally high in ASD (Oswald and Sonenklar, 2007
). Thus, excluding these subjects would lead to an unrepresentative sample, potentially with a very different symptom profile. Since subjects used different classes of medications, it was possible to evaluate whether each class influenced the results. Specifically, as documented in the results section, ASD subjects using particular classes of medications were removed from the analyses and the same results remained, indicating that the medication did not underlie the results.
Future directions for this line of research include the following. First, using diffusion tensor imaging (DTI), it would be possible to examine the integrity of white matter tracts within the default network in ASD. To date, multiple DTI studies have reported white matter alterations in participants with ASD (Barnea-Goraly et al., 2004
; Alexander et al., 2007
; Ben Bashat et al., 2007
; Keller, Kana and Just, 2007
; Lee et al., 2007
; Sundaram et al., 2008
; Thakkar et al., 2008
). Moreover, recently, two groups used DTI and showed that white matter tracts mapped onto specific connections of the default network (Mandl et al., 2008
; Greicius et al., 2009
). Combining DTI and resting connectivity with an ASD sample would help to determine whether the altered functional connectivity in ASD is instantiated anatomically. Second, the present study used a predefined seed in the posterior cingulate cortex that was based on investigations of healthy adults. It is possible that individuals with ASD may have default network connectivity that is most pronounced in a different area. Future research may wish to use self-organizing map algorithms to evaluate differences in connectivity without a predefined model (Ngan and Hu, 1999
; Peltier, Polk and Noll, 2003
). Third, since these procedures do not involve a cognitive task, they can be replicated on a sample of low functioning individuals. From this, it would be possible to associate functional connectivity of the default network to level of adaptive functioning. Fourth, age and developmental level could also be correlated with connectivity. In particular, these procedures are suitable for very young children and so early risk and manifestations of ASD could be related to functional connectivity. And fifth, following the approach which recently showed that polymorphisms of the serotonin transporter gene are associated with gray matter volume in ASD (Wassink et al., 2007
), it may be fruitful to link ASD-relevant polymorphisms to alterations in intrinsic functional connectivity.
Despite evidence that the default network at rest commands greater metabolic activity than cognitive or affective tasks (Raichle and Mintun, 2006
), it is unclear what function(s) the default network subserves. Recently, studies reported altered functioning of the default network in multiple clinical populations, including depression (Greicius et al., 2009
; Grimm et al., 2009
; Sheline et al., 2009
), attention deficit/hyperactivity disorder (Castellanos et al., 2008
), schizophrenia (Garrity et al., 2007
; Zhou et al., 2007
), and mild cognitive impairment (Sorg et al., 2007
). Moreover, developmental studies have documented that children and older adults, too, show differences in default network activation relative to young adults (Andrews-Hanna et al., 2007
; Fair et al., 2008
; Thomason et al., 2008
). Combined with the present findings and the studies reported above on ASD, the default network appears to be altered in many different populations. Thus, it appears likely that the default network does not subserve a narrow, circumscribed function. Instead, the default network may support many different functions or it may be involved in a fundamental capacity of the central nervous system.