The Default Mode Network in Autism

Abstract

Autism spectrum disorder (ASD) is characterized by deficits in social communication and interaction. Since its discovery as a major functional brain system, the default mode network (DMN) has been implicated in a number of psychiatric disorders, including ASD. Here we review converging multimodal evidence for DMN dysfunction in the context of specific components of social cognitive dysfunction in ASD: 'self-referential processing' - the ability to process social information relative to oneself and 'theory of mind' or 'mentalizing' - the ability to infer the mental states such as beliefs, intentions, and emotions of others. We show that altered functional and structural organization of the DMN, and its atypical developmental trajectory, are prominent neurobiological features of ASD. We integrate findings on atypical cytoarchitectonic organization and imbalance in excitatory-inhibitory circuits, which alter local and global brain signaling, to scrutinize putative mechanisms underlying DMN dysfunction in ASD. Our synthesis of the extant literature suggests that aberrancies in key nodes of the DMN and their dynamic functional interactions contribute to atypical integration of information about the self in relation to 'other', as well as impairments in the ability to flexibly attend to socially relevant stimuli. We conclude by highlighting open questions for future research.

Keywords: Autism; default mode network; mentalizing; self-referential processing; social; theory of mind.

Figure 1. Functional and structural architecture of the DMN identified using multiple imaging modalities and methods

(A) Architecture of the default mode network (DMN), identified as regions of “task-induced deactivation” in the seminal meta-analysis by Shulman et al., 1999. Data derived from nine studies using [^15O] H2O positron emission tomography (PET) and reproduced as a surface rendering, as in Buckner et al., 2005. (Adapted from Bucker et al., 2008 (13)) (B) DMN topology is readily identifiable using resting-state functional magnetic resonance imaging (fMRI) data, here through the use of an independent component analysis (ICA) approach. (Adapted from Shirer et al., 2012 (131)) (C) Core midline DMN nodes, the medial prefrontal cortex (mPFC) and posterior cingulate cortex (PCC) are structurally connected via a major white matter pathway, the cingulum bundle. Fibers reconstructed using diffusion tensor imaging tractography. (Adapted from van den Heuvel et al., 2008(132)) (D) The strength of structural connections amongst DMN nodes can be quantified using diffusion imaging. Edge thickness and node size represents connection strength and node degree, respectively. PCC is most strongly connected node within the DMN (Adapted from Tao et al., 2015 (133)). (E) Spring graph illustrates the differing functional connectivity weights between DMN nodes, such that more strongly connected nodes are closer together in space and these midline “hubs” are embedded centrally within the network. (Adapted from Andrews-Hanna et al., 2010 (134))

Figure 2. The DMN overlaps spatially with regions recruited by component processes of social cognition

The result of a meta-analysis using over 1200 fMRI studies in the Neurosynth database(135), using the search term “default mode” (shown in red). Regions implicated with four additional search terms, (A) “social”, (B) “mentalizing”, (C) “self-referential”, and (D) “theory of mind”, are shown in blue. The overlap between the DMN and regions recruited by these unique aspects of social cognition are shown in yellow.

Figure 3. Hyperconnectivity with the PCC predicts social communication deficits in children with ASD

(A) Children with ASD demonstrate whole brain hyperconnectivity with PCC and retrosplenial cortex and hypoconnectivity with precuneus. *p<.01, **p<.005, ***p<.001. aLTC, anterolateral temporal cortex; DMPFC, dorsomedial prefrontal cortex; ERc, entorhinal cortex; LG, lingual gyrus; PHG, parahippocampal gyrus; pInsula, posterior insular cortex; PRc, perirhinal cortex; pSTS, posterior superior temporal sulcus; TempP, temporal pole. (Adapted from Lynch et al., 2013 (54)) (B) hyperconnectivity between the PCC and target regions including the right parahippocampal gyrus, left temporal pole, and right lingual gyrus predicted social impairments as measured by the Autism Diagnostic Observation Schedule (ADOS) social subscale in children with ASD. No such significant relationships were demonstrated by the retrosplenial cortex (RSC) and precuneus (PreC). *Significant. L, left; LG, lingual gyrus; n.s., not significant; PHG, parahippocampal gyrus; R, right; TempP, temporal pole. (Adapted from Lynch et al., 2013 (54))

Figure 4. The DMN transitions from an immature state in childhood to a cohesive network in adulthood

(A) Independent component analysis applied to resting-state fMRI data reveals stronger mPFC functional connectivity in a group of adults relative to a group of children aged 7–9 years. (Adapted from Supekar et al., 2010 (67)) (B) Comparison of mPFC functional connectivity strength in children (left) and adults (middle) confirms greater connectivity, especially with PCC, later in development (right). (Adapted from Fair et al., 2008 (136)) (C) Connectivity between the DMN and other systems of the brain change over development as well. Spring graph representation of brain network development using resting-state fMRI during three periods of development, left = 9 years, middle = 13 years, right = 25 years. Note that some DMN nodes (filled red) are isolated from one another in childhood but are more strongly integrated in adulthood. (Adapted from Power et al., 2010 (70)).

Figure 5. Structural, neurochemical, and cytoarchitectonic disorganization of key DMN nodes in ASD

(A) Right temporoparietal junction sulcus is shallower in children with ASD (red line) compared to neurotypical children (blue line). (Adapted from Dierker et al., 2015 (77)). (B) Relationship between TPJ gray matter volume and ability to assess interactions between two objects in a social motion experiment in adults with ASD (red) relative to neurotypical adult controls (blue). Note that the relationship between gray matter volume in ASD is significant, such that greater gray matter volume is associated with better performance. Adapted from David et al., 2014 (78)). (C) Increased cortical thickness of the bilateral mPFC in children with ASD relative to adults and neurotypical children is replicable across sites of the ABIDE dataset. (Adapted from Valk et al., 2015 (75)). (D) Differences in gyrification of the mPFC in ASD by sex. Males with ASD have reduced gyrification whereas females have increased gyrification relative neurotypical controls. (Adapted from Schaer et al., 2015 (76)) (E) Abnormal laminar pattering in post-mortem ASD PCC brain tissue (right) relative to healthy case (left). Note the general disorganization and poor differentiation between layer IV and V. (Adapted from Oblak et al., 2011 (104)) F) Reduced 3H-muscimol labeled GABA_A receptor binding density in the PCC in ASD (right) and neurotypical control (left) postmortem brain tissue, darker colors indicate greater receptor density. (Adapted from Oblak et al., 2011 (118))