20 years of the default mode network: A review and synthesis

This perspective reviews how the default mode network (DMN)—a set of interconnected brain regions typically suppressed during externally focused tasks and engaged during internally oriented cognition—was discovered and how understanding of its functions has evolved. The paper frames two core questions: what mental processes are suspended during externally focused cognition, and what are the direct cognitive roles of the DMN? It highlights the DMN’s contrast with sensory-motor and attention/control networks, its consistent deactivation during demanding tasks, and its engagement during internally directed processes (self-reflection, memory, mind wandering). The purpose is to synthesize two decades of evidence across domains (self, social cognition, episodic/autobiographical memory, language/semantics, mind wandering), to identify unifying themes, propose an integrative model in which the DMN constructs an internal narrative, and outline challenges and future directions.

The article synthesizes a large body of prior work. Key milestones include: (1) PET meta-analyses showing consistent task-induced decreases in posterior cingulate/precuneus, medial prefrontal cortex (mPFC), and inferior parietal regions during active tasks (Shulman et al., 1997; Mazoyer et al., 2001); (2) formulation of a “default mode” of brain function (Raichle et al., 2001); (3) identification of a cohesive intrinsic functional network with posterior cingulate cortex (PCC) coupling to mPFC and angular gyrus (AG) at rest (Greicius et al., 2003), establishing the DMN; (4) early functional interpretations linking DMN nodes to episodic memory (Andreasen et al., 1995), semantic processing (Binder et al., 1999), and self-referential monitoring (Gusnard & Raichle, 2001); (5) clinical evidence of DMN dysfunction in Alzheimer’s disease, depression, schizophrenia, with node-specific vulnerability (e.g., PCC/RSC/MTL in AD; ventral mPFC/subgenual cingulate in depression); (6) mechanistic accounts of DMN suppression via interactions with salience and frontoparietal networks (triple-network model), supported by dynamic causal modeling and electrophysiology; (7) convergent meta-analyses and lesion studies implicating DMN nodes in self-reference, theory of mind, episodic/autobiographical memory, language/semantic cognition, and mind wandering; (8) advances in naturalistic paradigms and inter-brain coupling demonstrating DMN’s role in narrative comprehension and social communication; (9) emerging views of DMN heterogeneity and subnetworks (e.g., dorsal/ventral PCC, RSC) and its hub-like, apex position in cortical gradients; (10) translational evidence from rodent optogenetics/fMRI showing causal suppression of rodent DMN by anterior insula stimulation.

This is a narrative review and synthesis integrating findings from: (a) task-based and resting-state fMRI (including meta-analyses of thousands of studies), (b) PET studies on task-induced deactivations, (c) intracranial EEG demonstrating neural suppression in DMN nodes, (d) computational and effective connectivity approaches (e.g., dynamic causal modeling, information-theoretic measures) characterizing network switching and directionality, (e) naturalistic paradigms assessing narrative processing and inter-subject neural coupling, and (f) translational rodent studies using optogenetic stimulation with concurrent fMRI and fiber photometry to test causal mechanisms of DMN suppression. The author organizes evidence around indirect (suppression, network switching) and direct (domain-specific activation) roles of DMN, and develops an integrative model positing that DMN integrates memory, language, and semantic representations to sustain an internal narrative.

• Discovery and characterization: DMN comprises interconnected cortical (PCC/precuneus, retrosplenial cortex, mPFC, angular gyrus, medial/lateral temporal regions) and subcortical nodes (thalamus, basal forebrain/nucleus accumbens). It shows reliable task-induced deactivation during externally focused, attentionally demanding tasks.
• Indirect role via suppression and switching: Inefficient DMN suppression predicts attentional lapses and slower responses; salience network (particularly right anterior insula) acts as a causal switching hub that disengages DMN and engages frontoparietal control during salient external events (shown via dynamic causal modeling, latency analyses, and causal outflow metrics).
• Neural origin of deactivation: Intracranial EEG reveals high-gamma suppression in DMN core (mPFC, PCC/RSC) during tasks, confirming neuronal basis of fMRI/PET deactivations.
• Cross-species causal validation: Optogenetic stimulation of anterior insula in rodents activates salience nodes and suppresses retrosplenial (DMN-homolog) regions and their connectivity, providing causal evidence for SN-mediated DMN suppression.
• Direct cognitive roles: Robust engagement of DMN in self-reference (mPFC, PCC, left AG), social cognition/theory of mind (mPFC, PCC, right AG/TPJ), episodic/autobiographical memory (hippocampus with PCC, RSC, mPFC, AG supporting distinct facets), language/semantic cognition (left AG/PGp, MTG, ATC; semantic judgment and sentence comprehension; predictive and narrative processing), and mind wandering (mPFC, PCC, MTL, AG; lesion data indicate causal contributions).
• Integrative internal narrative: DMN integrates and broadcasts memory, language, and semantic representations to sustain coherent internal narratives that contribute to self-construction and subjective continuity; DMN hub properties (short path length, apex of cortical gradients, higher net causal outflow to other networks) support global state changes and frames of thought.
• Heterogeneity/subnetworks: DMN is not unitary—posterior medial components (dorsal/ventral PCC, RSC) and other nodes form interdigitated subnetworks with distinct cognitive and network coupling profiles; multiplexed streams enable flexible coupling to language and control systems.
• Clinical relevance: Node-specific DMN disruptions relate to symptoms across disorders (e.g., AD memory loss linked to PCC/RSC/MTL; depression/rumination linked to ventral mPFC/subgenual cingulate); maladaptive switching contributes to attentional lapses and psychopathology.
• Developmental and ontogenetic perspective: Inner narrative function may emerge from internalization of self-directed speech (Vygotsky), suggesting developmental roots for DMN-supported internal cognition.

The synthesis reframes the DMN from a passive resting system to a dynamic, integrative network with both indirect and direct roles in cognition. Indirectly, the DMN is suppressed during externally focused, attentionally demanding tasks via salience-driven network switching, enabling cognitive control engagement. Directly, its nodes contribute to self-referential processing, social inference, episodic/autobiographical memory, and language/semantic operations. These roles converge on an overarching function: the DMN integrates and broadcasts representations to construct and sustain internal narratives that underlie a sense of self and guide social interaction. The model explains how failures of suppression lead to lapses and psychopathology, while appropriate switching permits flexible alternation between external processing and internal mentation. DMN hub properties and causal outflow support global state changes and frames of thought, reconciling its involvement in both internally directed and externally related processes (e.g., narrative comprehension, predictive language). Recognizing DMN heterogeneity and subnetworks clarifies node-specific contributions and resolves tensions with a unitary-function view.

The perspective consolidates two decades of research to propose that the DMN integrates episodic memory, language, and semantic systems to generate a coherent, ongoing internal narrative that supports self-construction and subjective continuity. It emphasizes a unified account linking indirect suppression/network switching with direct domain-specific functions. Future directions include: (1) elucidating fast spatiotemporal dynamics of DMN suppression/activation and global broadcasts with improved multimodal methods and modeling; (2) mapping DMN subnetworks, convergence zones, and individual variability; (3) leveraging intracranial recordings and causal manipulations (e.g., optogenetics in animal homologs) to uncover cellular/circuit mechanisms; (4) expanding developmental studies to track emergence of internal narrative, inner speech, and social cognition; and (5) refining clinical models where aberrant switching and node-specific dysfunction drive symptoms, to inform targeted interventions.

• Methodological constraints: fMRI’s limited temporal resolution hinders tracking rapid network switching and signal flow; EEG/MEG lack spatial precision for medial/deep DMN nodes; intracranial EEG coverage is sparse and rarely spans multiple DMN hubs in one individual.
• Causality and translation: Direct causal manipulations are largely infeasible in humans; rodent DMN homologs provide insights but cannot capture uniquely human aspects (rich mind wandering, language semantics).
• Heterogeneity: DMN comprises interdigitated subnetworks with distinct activation/deactivation patterns; incomplete mapping of convergence zones complicates interpretation of unitary vs. fractionated functions.
• State ambiguity: It remains unresolved whether global brain-state shifts are predominantly externally triggered or internally initiated, and how these dynamics alter ongoing mentation.
• Developmental gaps: Limited task-based pediatric studies constrain understanding of how DMN-supported internal narratives and self-related cognition emerge and evolve.