Proceedings of the National Academy of Sciences of the United States of America
May 26, 2009
Kalina Christoff, Alan M Gordon, Jonathan Smallwood et al.
1,832 citations
Mind wandering, which occupies a large portion of waking life, involves parallel recruitment of both default and executive brain networks, two systems previously thought to work in opposition. Using fMRI with experience sampling during a task, activation in default network regions, particularly medial prefrontal cortex, was linked to subjective reports of mind wandering and to performance errors. Executive network recruitment also occurred, especially when participants lacked meta-awareness of their mind wandering. The findings suggest that mind wandering represents a unique mental state where these networks cooperate rather than oppose each other.
Nature reviews. Neuroscience
July 5, 2021
Jonathan Smallwood, Boris C. Bernhardt, Robert Leech et al.
970 citations
The default mode network (DMN) is a set of brain regions in the parietal, temporal, and frontal cortex that typically reduce activity during attention-demanding tasks but increase activity during complex cognition linked to memory or abstract thought. These regions are located farthest from sensory and motor systems. The paper considers how knowledge of the DMN's topographic characteristics can be used to better understand its contributions to cognition and behavior.
Brain Research
April 15, 2011
Jonathan Smallwood, Kevin Brown, Ben Baird et al.
394 citations
The ability to generate and sustain an internal train of thought independent of external reality frees an agent from acting only on immediate environmental events. This paper proposes that such thought arises from cooperation between autobiographical information from the default mode network and a frontal-parietal control network that sustains and buffers internal trains of thought against disruption. This hypothesis explains two features: first, access to the top-down control system is a prerequisite of conscious experience, explaining why its activation and default mode activity are often observed together during internally guided thought.
Proceedings of the National Academy of Sciences
August 27, 2018
Mladen Sormaz, Charlotte Murphy, Hao-ting Wang et al.
291 citations
The default mode network (DMN), a set of brain regions traditionally linked to off-task or mind-wandering states, actually contributes to detailed, task-relevant cognition during active tasks. Using fMRI, participants performed working-memory tasks while reporting their thoughts. Patterns of neural activity showed that distinctions between on- and off-task thought involved regions near sensory and motor cortex, not the DMN. However, the level of detail in ongoing thought corresponded to activity patterns within the DMN during memory maintenance. These findings indicate the DMN supports detailed cognition under active task conditions, challenging the view that it is solely task-negative.
NeuroImage
January 12, 2018
Charlotte Murphy, Elizabeth Jefferies, Shirley-Ann Rueschemeyer et al.
288 citations
The default mode network supports cognition independent of immediate environment and higher-order conceptual representations. A novel paradigm manipulated perceptual information availability and representational complexity during decision-making. Brain regions including left and right angular gyri and left middle temporal gyrus responded when cognition combined stimulus independence with multi-modal information. These default mode network sites showed stronger response to demanding memory judgments than easier perceptual tasks, contradicting the view they support automatic cognition. These regions were at the extreme end of a macroscale gradient from sensorimotor to transmodal cortex, suggesting functional distance from sensory input enables conceptually rich cognitive states in absence of input.
PLoS ONE
June 30, 2015
Mahiko Konishi, Donald George McLaren, Haakon Engen et al.
210 citations
The default mode network (DMN) is more active when people make decisions based on stored information from a previous trial than when they respond to current perceptual input. Using fMRI, the study found increased BOLD activity in the medial prefrontal cortex and posterior cingulate cortex when individuals recalled the location of a shape from the prior trial, and decreased activity when they judged the current location. This challenges the view of the DMN as merely task-negative or emotion-sensitive. Instead, the DMN's core hubs enable cognition to be guided by internal representations rather than immediate sensory input, explaining its involvement in higher-order thought such as imagining the future or considering another's perspective.
Social Cognitive and Affective Neuroscience
March 21, 2017
Giulia L. Poerio, Mladen Sormaz, Hao-ting Wang et al.
185 citations
Mind-wandering involves cognition detached from the immediate environment. The default mode network (DMN), a set of brain regions far from sensory and motor areas, is implicated but its exact role is unclear. This study examined how individual differences in resting-state DMN connectivity relate to performance on memory, social, and planning tasks and to variability in spontaneous thought. Poor external engagement was linked to stronger coupling between medial and dorsal DMN subsystems, while decoupling of the core from the cerebellum predicted detailed memory retrieval. Both patterns predicted off-task future thoughts. The DMN supports stimulus-independent cognition through strong subsystem coupling and allows memory representations to form conscious experience.
Journal of Cognitive Neuroscience
April 17, 2014
Benjamin Baird, Jonathan Smallwood, Antoine Lutz et al.
169 citations
When the mind wanders, the brain's response to sensory input becomes less consistent from one moment to the next, particularly in theta-band oscillations over parietal regions. Using an experience sampling method with continuous EEG, the authors found that task-unrelated thought reduced the P1 event-related potential, replicating prior work. Time-frequency analysis showed decreased theta-band phase-locking, indicating lower temporal fidelity in the brain's response to stimuli. Additionally, mind-wandering required more cortical processing to re-engage with a task. These results suggest that attentional states shape neural processing of sensory input by affecting the consistency of oscillatory responses.
iScience
February 2, 2021
Jonathan Smallwood, Adam Turnbull, Hao-ting Wang et al.
130 citations
The landscape of ongoing thought is heterogeneous and shaped by both personal traits and environmental context. Recent work shows that attention and control systems organize experience in response to changing demands, while the default mode network contributes not only to task-negative or episodic content but also to the vividness of experience in both task contexts and spontaneous self-generated states. Multiple neural systems reflect the landscape of ongoing thought, and it is important to distinguish processes that shape how experience unfolds from those that regulate it.
Neuroimage
April 25, 2022
Manesh Girn, Leor Roseman, Boris Bernhardt et al.
112 citations
LSD and psilocybin flatten the brain's normal hierarchical organization, reducing the functional separation between sensory and higher-order networks and increasing cross-talk between them. Analyzing two resting-state fMRI datasets, the principal gradient of cortical connectivity—which normally runs from basic sensory areas to complex association regions—was significantly compressed under both drugs compared to placebo. This flattening was driven by decreased specialization at both ends of the hierarchy: default mode and frontoparietal networks at the top, and somatomotor networks at the bottom. The findings support a proposed mechanistic model of the psychedelic state and demonstrate that macroscale connectivity gradients can be acutely altered by a pharmacological intervention.
Frontiers in human neuroscience
January 1, 2013
Micah Allen, Jonathan Smallwood, Joanna Christensen et al.
100 citations
Mind-wandering, or task-unrelated thoughts (TUTs), is common and often impairs performance on demanding tasks, but new findings show it can also enhance metacognitive abilities. Using the Error Awareness Task (EAT), researchers found that individual differences in average TUTs strongly predicted stop accuracy, while variability in TUTs specifically predicted error awareness. Brain imaging revealed that both response inhibition and TUT ratings activated the medial prefrontal cortex (mPFC) of the default mode network (DMN), but in distinct dorsal areas, suggesting functional segregation. Co-activation of salience and default mode regions during error awareness linked monitoring to TUTs. The results suggest that fluctuations between internal and external thought, rather than constant focus, characterize individuals with greater metacognitive monitoring, and balancing these modes may optimize task performance.
Scientific Reports
December 3, 2020
Theodoros Karapanagiotidis, Diego Vidaurre, Andrew J. Quinn et al.
82 citations
When people are not engaged in an explicit task, they experience a variety of self-generated thoughts, such as planning or reminiscing. Using machine learning to analyze brain activity from resting-state fMRI scans, researchers identified distinct neural states that recur over time. Two of these states predicted different patterns of thinking. One neural state, resembling activity seen during demanding tasks, was linked to problem-solving about the future. Another state, associated with less demanding conditions, was tied to intrusive thoughts about the past. These two states fell at opposite ends of a brain hierarchy related to cognitive demand. The findings show that tracking moment-to-moment changes in brain function can help classify self-generated mental states and that these states align with the brain's response to cognitive tasks.
bioRxiv (Cold Spring Harbor Laboratory)
May 3, 2020
Manesh Girn, Leor Roseman, Boris C. Bernhardt et al.
27 citations
preprint
LSD and psilocybin flatten the brain's hierarchical organization, reducing the functional separation between sensory and higher-order cognitive networks. Using a non-linear dimensionality reduction technique on resting-state fMRI data, the authors found that both drugs compressed the principal gradient of cortical connectivity, which normally spans from unimodal (sensory) to transmodal (association) cortex. This flattening was driven by decreased differentiation at both ends of the hierarchy—default and frontoparietal networks at the upper end and somatomotor networks at the lower end—and was accompanied by increased crosstalk between unimodal and transmodal regions. Changes in the principal gradient under LSD tracked self-reported ego-dissolution. The findings support a mechanistic model of the psychedelic state and demonstrate that macroscale connectivity gradients are sensitive to serotonergic modulation.
Journal of Neuroscience
April 8, 2024
Ximing Shao, Katya Krieger‐Redwood, Meichao Zhang et al.
20 citations
The default mode network (DMN) typically deactivates during external tasks but supports semantic cognition. Analysis of four human fMRI datasets shows that its subsystems respond differently: the frontotemporal subsystem activates across domains and modalities, especially during abstract verbal tasks, and shows more tuned states with higher semantic retrieval demands. The medial temporal subsystem activates for both perceptually coupled scenes and decoupled autobiographical memory, with stronger responses to picture associations, supporting scene construction. The core DMN consistently deactivates, particularly for externally oriented tasks. These distinct responses relate to each subsystem's location on intrinsic connectivity gradients, revealing complementary roles in semantic cognition.
Consciousness and cognition
February 1, 2025
Anusha Garg, Shivang Shelat, Madeleine E Gross et al.
13 citations
Thinking aloud while letting the mind wander does not substantially alter the stream of consciousness compared to thinking silently. In two studies with 111 and 102 participants, people who verbalized their ongoing thoughts showed no significant differences in meta-awareness or how often their topics shifted. Of 21 thought qualities and 18 content topics examined, only three qualities (private thoughts, mind blanking, and session difficulty) and one topic (partner, intimacy, love, and sexual matters) differed between conditions. Cognitive load also did not differ. The findings indicate that the Think Aloud method is a reliable and minimally reactive tool for studying the natural flow of thoughts in task-absent settings.
NeuroImage
May 15, 2019
Léa M Martinon, Jonathan Smallwood, Deborah McGann et al.
People spend a lot of time engaged in internally generated thoughts like mind-wandering, but studying these private experiences is methodologically difficult because they are not directly observable. This review argues that combining self-reports with neuroimaging techniques—such as EEG, ERP, and MRI—provides a more rigorous way to investigate patterns of ongoing thought. Imaging methods reveal the presence of covert mental states and help distinguish different aspects of experience, offering critical information that purely psychological approaches cannot. Triangulating brain data with subjective reports is essential for building an empirically grounded account of the stream of consciousness.
Annual review of psychology
January 3, 2015
Jonathan Smallwood, Jonathan W Schooler
Conscious experience is fluid, rarely staying on one topic for long. Mind wandering illustrates this, as attention shifts from a current task to unrelated thoughts and feelings. Its content and relation to meta-cognition determine functional outcomes. Mind wandering involves perceptual decoupling to escape the moment, with content arising from episodic and affective processes, regulated by executive control. It carries costs, linked to errors, and benefits, associated with creativity and future planning. Strategies can minimize downsides while maintaining productive aspects.
Brain and cognition
February 1, 2008
Jonathan Smallwood, Merrill McSpadden, Bryan Luus et al.
Principal component analysis of response times during a continuous performance task can distinguish mental states. Three components accounted for 58.8% of the variance in response time data. The second-largest component was a marker for mind wandering: it was stronger under slow stimulus presentation (paralleling the distribution of mind-wandering reports) and more powerful before failures in response inhibition (a behavioral marker of mind wandering) and less powerful before reports of task-focused thinking. The analysis appears to be a viable tool for differentiating mental states and may help identify neural substrates underlying mind wandering.