Humans generate thousands of thoughts each day, many of which concern other people. Yet it remains unclear whether these ongoing thoughts vary systematically across social contexts that demand meaningful social behavior, and whether they reflect individual differences in social preferences such as generosity and fairness. To address this question, we combined multidimensional experience sampling (MDES) with an altruism task administered online or in person (total N = 320). Participants made monetary sharing decisions across social contexts that varied in partner cues (distressed, neutral, unidentifiable), while intermittently reporting their ongoing thoughts using MDES. Dimensionality reduction of MDES responses using principal component analysis revealed a small set of common, interpretable, and robust thought patterns characterizing moment-to-moment cognition during social choice. Critically, the expression of these thought patterns tracked variation in social context, generosity, and model-based parameters indexing fairness preferences. Moreover, these patterns generalized to a separate non-social task that did not require social behavior, indicating that they reflect common patterns of cognition rather than task-specific artifacts. Together, these findings provide initial evidence that ongoing thought states offer reliable insights into social preferences, introducing a novel framework for studying the link between ongoing cognition and social behavior in the laboratory and everyday life.

Abstract Individual differences modulate our thoughts and emotional experiences, yet how thought and emotion interact in daily life remains largely unclear. We leverage alexithymia, a trait reflecting atypical emotional awareness, to reveal these interactions in naturalistic settings. Using multi-dimensional experience sampling via smartphones, we captured moment-to-moment thought patterns and concurrent affective states (valence, arousal, stress) in people’s daily life (N = 190 undergraduate students, age range = 18 to 36, 159 females). Using Principal Component Analysis and Linear Mixed Models, we identified four thought dimensions that relate differently to these affective states: future-self orientation, intrusive distraction, sensory engagement, and task-focus. Alexithymia modulated these relationships. High overall alexithymia predicted fewer future-self-oriented thoughts and greater variability in sensory engagement across affective and social contexts, while difficulty identifying feelings selectively reduced future-self orientation during intense sadness, and externally oriented thinking rendered thought patterns less sensitive to affective context. By mapping affective experiences onto thought dimensions, these findings uncover cognitive pathways that connect to emotional well-being, providing a scalable framework for understanding variability in human affective experience.

To navigate the world, we store knowledge about relationships between concepts and retrieve this information flexibly to suit our goals. The semantic control network, comprising left inferior frontal gyrus (IFG) and posterior middle temporal gyrus (pMTG), is thought to orchestrate this flexible retrieval by modulating sensory inputs. However, interactions between semantic control and input regions are not sufficiently understood. Moreover, pMTG's well-formed structural connections to IFG and visual cortex suggest it as a candidate region to integrate control and input processes. We used magnetoencephalography to investigate oscillatory dynamics during semantic decisions to pairs of words, when participants (both sexes) did or did not know the type of semantic relation between them. IFG showed increases and decreases in oscillatory activity to prior task knowledge, while pMTG only showed positive task knowledge effects. Furthermore, IFG provided sustained feedback to pMTG when task goals were known, while in the absence of goals this feedback was delayed until receiving bottom-up input from the second word. This goal-dependent feedback coincided with an earlier onset of feedforward signaling from visual cortex to pMTG, indicating rapid retrieval of task-relevant features. This pattern supports a model of semantic cognition in which pMTG integrates top-down control from IFG with bottom-up input from visual cortex to activate task-relevant semantic representations. Our findings elucidate the separate roles of anterior and posterior components of the semantic control network and reveal the spectro-temporal cascade of interactions between semantic and visual regions that underlie our ability to flexibly adapt cognition to the current goals.

Abstract Post-stroke semantic aphasia is characterised by multimodal semantic deficits and reflects disruption of a distributed brain network spanning frontal and temporal regions. Connectivity gradients, which capture key dimensions of whole-brain variation in functional connectivity, offer a promising framework for understanding the global impact of stroke on brain function. This study investigated whether changes in connectivity gradients following stroke can explain semantic aphasia deficits. First, we evaluated whether lesion-location and lesion-load information from structural MRI could predict the gradient changes observed in resting-state fMRI, as a proof-of-principle analysis. Second, we tested whether simulated gradient changes predict the severity of semantic impairment. Results show that post-stroke gradient changes simulated from structural MRI are correlated with actual changes in resting-state fMRI, particularly for the principal gradient that separates unimodal and heteromodal regions. Semantic deficits were related to simulated connectivity changes along this gradient: left prefrontal areas involved in controlled semantic retrieval exhibited stronger connectivity to unimodal cortex in patients with more severe deficits. Semantic deficits also correlated with changes in the second gradient, which distinguishes visual and motor cortex. Particularly, the right parahippocampal gyrus, typically visually biased—showed reduced visual connectivity in more impaired patients. These results help explain controlled semantic retrieval deficits in semantic aphasia. More broadly, the findings suggest that functional connectivity gradients capture post-stroke reorganisation of global brain networks linked to cognitive impairment, and that these changes can be estimated from structural MRI alone, enhancing clinical utility of gradient-based approaches. Highlights Functional connectivity gradients explain the multimodal impairments in semantic aphasia from a dimensional perspective, using the unimodal–transmodal and motor–visual axes. Post-strokë functional changës arë ëxplorëd through altërations in connëctivity gradiënt patterns. Cortical lesion information from structural MRI can be used to simulate changes in connectivity gradients, offering potential clinical relevance.

Human cognition relies on two modes: a perceptually coupled mode where mental states are driven by sensory input and a perceptually decoupled mode featuring self-generated mental content. Imagined states that evoke mental imagery are thought to be supported primarily by reinstated activity in sensory cortex, but transmodal systems are also implicated in imagery-related processes like mind-wandering, recollection, and imagining the future. During a precision fMRI experiment, participants imagined different scenarios in the scanner, then rated their mental states using multi-dimensional experience sampling. Thinking involving scenes evoked activity within parts of the canonical default network, while imagining speech evoked activity within the language network. In each domain, imagining-related activity overlapped with activity evoked by viewing scenes or listening to speech, respectively; however, this overlap was predominantly within transmodal association networks, rather than adjacent unimodal sensory networks. We conclude that the engagement of transmodal networks supports self-generated mental states involving different forms of mental imagery.

A debate is raging in human neuroscience over the foundational principles of cortical organisation, chiefly whether the cortex is best characterised by discrete areas or continuous gradients. Though fuelled by recent insights from in vivo neuroimaging, the debate is in fact centuries old. Classic works in neuroanatomy and neurophysiology hold insights that can still shape our theories today. Tracing the origins of these schools of thought also sheds light on where prevailing models simplify cortical organisation for purposes of utility. In this scoping historical review, traversing across methods as varied as vivisection, skull morphology, neuropathological lesions, histology and modern neuroimaging, we trace a persistent tension between modular and integrative accounts of cortical organisation. Yet several principles consistently emerge, even among competing schools, (i) hard boundaries are the exception rather than the rule, (ii) parcellations and gradients can be thought of as complementary descriptions operating at different spatial scales, and (iii) a more realistic cortical map would be multi-dimensional. We close by reflecting on how a multi-scale, multi-dimensional conceptualisation of the cortex offers a flexible framework for investigating how dynamic cortical function maps onto relatively stable cortical structure.

While listening toan audiobook, listeners' attention may shift between the text and self-generated thoughts occurring during mind-wandering. Despite the growing use of audiobooks, little is known about how listeners process text when listening to it. The present study investigated the types of thought patterns that listeners have while listening to an audiobook, and whether and how these are reflected in eye movements. Participants (NStudy1 = 63, NStudy2 = 58) listened to an audio recording of a novel and responded to a 13-item mind-wandering questionnaire and a simple memory question 30 times during the listening task. In study 2, participants' eye movements were also recorded while they looked at a fixation cross on a screen. A principal components analysis (PCA) of the mind-wandering questionnaire responses produced four components in Study 1, and three components in Study 2. Three components were replicated across the two studies: Immersion, Mind-wandering, and Sub-vocalization. We then examined how these thought patterns were reflected in eye movementsin Study 2 including: fixation duration, fixation count, fixation dispersion, saccade amplitude, blink duration, and blink count. The results showed that higher levels of immersion was characterized by fewer and less dispersed fixations on the screen, shorter saccades, and longer blinks. Mind-wandering was related to more dispersed fixations. Sub-vocalization resulted in more fixations, higher dispersion across the screen, and more blinks. The results suggest that eye movements reflect shifts in attentional focus while listening to a literary text. The results provide important information about the processes underlying literary experience.

The posterior superior temporal sulcus (pSTS) processes information from the eyes and the mouth that support social perception. To investigate the laterality of how these mechanisms function, we performed three experiments on lip and eye-gaze discrimination. In Experiment 1, participants (n = 18) performed lip-position and eye-gaze discrimination tasks in static facial expressions while transcranial magnetic stimulation (TMS) was delivered over the left and right pSTS. Results showed a double dissociation in which disruption of the left pSTS impaired the lip-position task, while disruption of the right pSTS impaired the eye-gaze matching task. In Experiment 2, participants (n = 16) performed a lip-reading task using dynamic video clips of a speaker while TMS was delivered over the left and right pSTS. Task performance was impaired when TMS was delivered over the left pSTS only. In Experiment 3, participants (n = 256) underwent resting-state functional magnetic resonance imaging. Results demonstrated that the left pSTS exhibited greater connectivity to language processing brain areas in the left hemisphere. In contrast, the right pSTS exhibited greater connectivity to visual areas specialized for face processing and spatial attention processing. Our study suggests that lip and eye-gaze discrimination are preferentially lateralized across the bilateral pSTS.

The brain undergoes profound structural and functional development from childhood to adolescence. Convergent evidence suggests that neurodevelopment proceeds in a hierarchical manner, characterized by heterogeneous maturation patterns across brain regions and networks. However, the maturation of the intrinsic spatiotemporal propagations of brain activity remains largely unexplored. This study aims to bridge this gap by delineating spatiotemporal propagations from childhood to early adulthood. By leveraging a recently developed approach that captures time-lag dynamic propagations, we characterized intrinsic dynamic propagations along three axes: sensory-association (S-A), ‘task-positive’ to default networks (TP-D), and somatomotor-visual (SM-V) networks, which progress towards adult-like brain dynamics from childhood to early adulthood. Importantly, we demonstrated that as participants mature, there is a prolonged occurrence of the S-A and TP-D propagation states, indicating that they spend more time in these states. Conversely, the prevalence of SM-V propagation states declines during development. Notably, top-down propagations along the S-A axis exhibited an age-dependent increase in occurrence, serving as a superior predictor of cognitive scores compared to bottom-up S-A propagation. These findings were replicated across two independent cohorts (Human Connectome Project: Development and Nathan Kline Institute-Rockland Sample), emphasizing the robustness and generalizability of these findings. Our results provide new insights into the developmental progression of functional dynamics during youth and their role in supporting cognition. Here, the authors show that brain dynamics mature from childhood to adulthood with increased hierarchical propagation and reduced visual-somatomotor dynamics. Top-down hierarchical flow strengthens throughout this developmental period.

Humans have thousands of thoughts every day—often shaped by personal memories, goals, and environmental cues—that are deeply intertwined with emotional functioning. Yet, it remains unclear how these naturally occurring thoughts relate to people’s ability to dynamically regulate their emotions and their emotional well-being in daily life. To address this question, the present study explored whether and when common patterns of ongoing thought are associated with differences in emotion regulation (ER) variability—the degree of moment-to-moment change in the use of different regulatory strategies—across individuals and settings, as well as negative affect in everyday life. Over seven days, 120 participants completed 3,286 experience sampling surveys assessing ongoing thoughts, ER variability, and negative affect in real-world contexts (data collected in 2023–2024). Principal components analysis identified common “thought patterns,” which were then used to predict participants’ ER variability and negative affect. One thought pattern—characterized by negative, intrusive, and distractive content—was associated with lower ER variability and heightened negative affect, as captured by higher levels of anxiety, stress, and loneliness. Importantly, the link between this thought pattern and negative affect was moderated by momentary ER variability. Social activities, such as texting or engaging in conversations, similarly buffered the adverse relationship between this thought pattern and negative affect in daily life. These findings demonstrate how common, recurring patterns of ongoing thought relate to both ER variability and emotional well-being, underscoring the potential protective role of social interactions and variable use of diverse regulatory strategies from our ER repertoire.

Everyday cognition depends on the brain's capacity to shift between sensing the external world and constructing it from memory. To achieve this, large-scale cortical systems must flexibly integrate incoming sensory signals with internally generated representations. Here, we show that this flexibility is reflected in the macroscale architecture of the default mode network (DMN). Using convergent analyses across three independent fMRI datasets spanning directional connectivity, intrinsic organization, and task-evoked responses, we identify spatially distinct DMN subregions that are preferentially engaged during perceptual decisions about faces or memory-guided decisions based on previously seen images. These subregions correspond to a microarchitectural distinction, captured via directional and intrinsic connectivity profiles: regions preferentially engaged during face perception align with receiver-like, afferent-biased zones that show strong intrinsic connectivity across the heteromodal cortex, a profile that might support information integration during perceptually guided decision-making. In contrast, memory-guided, perceptually decoupled decisions differentially engage sender-like, efferent-biased zones that show broader connectivity with perceptual-motor and attentional systems beyond the DMN. This double dissociation demonstrates a systematic association between DMN connectivity and engagement during perceptually coupled versus memory-guided cognitive processes, providing an organizational account of how DMN architecture relates to flexible human thought.

Understanding the mechanisms of brain function and dysfunction is at the core of the neuroscience mission. However, the field's grasp of causal relationships between brain properties has been hindered by a focus on single modalities that neglects the complex interplay between the features found at different neural scales. Progress in neuroinformatics and the increasing availability of open datasets have helped overcome this limitation by facilitating the contextualization of brain maps against cellular, metabolic and network features. Despite the rapid uptake of data contextualization methods proposing that quantification of spatial similarity between brain maps may shed light on pathways of structure-function coupling, development and disease, their potential pitfalls have received little attention. In the context of neuroimaging research, these limitations include reliance on often small-sample and non-representative reference datasets, repeated use of the same brain maps across studies, and problems with intermodal and interindividual alignment. Applying data contextualization without considering these limitations can lead to circular reasoning, overfitting and correlational overreach, and limits the interpretation of findings to the properties of the source data. Here we provide a Roadmap of practical guidelines operating at the level of study design, analysis pipelines and interpretation of findings to encourage the development of best practices in data contextualization. A more informed use of brain map correlation approaches will improve mechanistic investigations and our understanding of causal relationships between brain properties.

When at rest, the mind becomes preoccupied with self-generated thoughts, commonly known as mind-wandering. While the social, autobiographical, and temporal features of these thoughts have been extensively studied, little is known about how frequently the wandering mind turns toward the interoceptive and somatic body. To map this underexplored component of "body-wandering," we conducted a large-scale neuroimaging study in 536 healthy participants, expanding a retrospective multidimensional experience sampling approach to include probes targeting visceral and somatomotor thoughts. Our findings reveal a robust interindividual dimension of body-wandering characterized by negative affect, high autonomic arousal, and a reduction in socially oriented thoughts. Despite this negative tone, individual differences in the propensity for body-wandering thoughts were associated with lower self-reported symptoms of ADHD and depression. Multivariate functional connectivity analyses further revealed that affective, body-oriented thoughts are related to a pattern of thalamocortical connectivity interlinking somatomotor and interoceptive-allostatic cortical networks. Collectively, these results demonstrate that self-generated thoughts exhibit core embodied features which are linked to the ongoing physical and emotional milieu of the visceral body.

Flexible cognition requires the adaptive retrieval of conceptual knowledge, spanning a continuum from proximal to distal semantic associations. However, the neural dynamics that facilitate this flexibility remain poorly understood. Here, combining computational linguistics with functional magnetic resonance imaging (fMRI) and machine learning methods, we derive a whole-brain signature that captures graded variations in semantic distance. This domain-specific neural model revealed three distinct large-scale cognitive systems whose interactions coordinate semantic retrieval: a left-lateralised frontotemporal language and a bilateral frontoparietal control network, both recruited for distant associations, and a medial default mode memory network, facilitating access to proximal relations. Importantly, we show that adaptive retrieval across the continuum of semantic distance is facilitated by a dynamic coordination mechanism. As semantic distance increases, representational patterns converge across the three cognitive systems. These findings provide a unifying model of the neural architecture underlying semantic processing, revealing how dynamic interactions between competing cognitive systems enable flexible knowledge retrieval.

ABSTRACT Human behavior is highly flexible, allowing efficient performance across a wide range of task contexts. A distributed set of frontal and parietal regions, commonly termed the multiple-demand network (MDN), is consistently engaged during diverse cognitively demanding tasks and is thought to support this flexibility. However, it remains unclear how patterns of MDN engagement relate to the qualitative features of ongoing cognition experienced during task performance. To address this issue, we examined the reliability of self-reported experiential features sampled during performance of a broad range of tasks. Across tasks, we found little evidence that particular patterns of thought were intrinsically more reliable than others, nor that individual tasks were associated with stable, characteristic thought profiles. Instead, the reliability of specific experiential features varied systematically across task contexts, with the same patterns showing high stability in some tasks and low stability in others. We next asked whether stable patterns of thought were associated with distinct neural signatures. We found that patterns of brain activity resembling the MDN tended to be present for tasks in which deliberate task focus was high, and when distraction was lower, adding to an emerging body of research suggesting that coordinated activity within frontal and parietal regions helps to establish a stable goal-focused mode of thoughts and actions.

Abstract Prevailing models propose that sustained focus is a product of top-down control that amplifies task-relevant representations while suppressing distraction. We challenge this view, demonstrating that the neural signatures of mental focus differ across cognitive modes. In two neuroimaging studies, we used a paradigm that creates conflict between external task goals and self-generated thought, to identify the neural correlates of task focus in contexts varying in memory engagement. In Study 1, arithmetic and comprehension tasks were associated with distinct neural architectures of task focus: arithmetic-related focus recruited visual–motor regions, while comprehension-related focus was characterized by the recruitment of memory systems and reduced activation of control regions. Study 2 extended these findings, with both externally directed comprehension and internal memory retrieval associated with a shared neural profile of reduced control engagement, despite reliance on distinct memory systems. Furthermore, focus covaried with activity in heteromodal semantic regions across both reading and listening, suggesting semantic engagement is a primary marker of being on task during comprehension. Our findings indicate that sustained focus during memory-guided cognition is characterized by mutual disengagement between control and memory networks. This discovery motivates a departure from traditional executive-control accounts, since task focus in meaningful contexts is associated with the recruitment of representational structures in long-term memory rather than active top-down supervision. Significance Statement Does concentration always require effort? While traditional models suggest that sustained focus is a product of top-down executive supervision, we show that the brain employs fundamentally different architectures for focus depending on the cognitive mode. In two neuroimaging studies, focus was associated with distinct neural profiles – reduced engagement of memory systems in arithmetic, and control deactivation alongside the activation of distinct memory systems in comprehension and autobiographical recall. Heteromodal semantic regions were associated with focus during reading and listening, consistent with an association between semantic network engagement and task focus during comprehension. These findings indicate that sustained attention occurs across multiple neural configurations linked to different cognitive modes, challenging the view that top-down supervision is a universal requirement.

Humans generate thousands of thoughts each day, many of which concern other people. Yet it remains unclear whether these ongoing thoughts vary systematically across social contexts that demand meaningful social behavior, and whether they reflect individual differences in social preferences such as generosity and fairness. To address this question, we combined multidimensional experience sampling (MDES) with an altruism task administered online or in person (total N = 320). Participants made monetary sharing decisions across social contexts that varied in partner cues (distressed, neutral, unidentifiable), while intermittently reporting their ongoing thoughts using MDES. Dimensionality reduction of MDES responses using principal component analysis revealed a small set of common, interpretable, and robust thought patterns characterizing moment-to-moment cognition during social choice. Critically, the expression of these thought patterns tracked variation in social context, generosity, and model-based parameters indexing fairness preferences. Moreover, these patterns generalized to a separate non-social task that did not require social behavior, indicating that they reflect common patterns of cognition rather than task-specific artifacts. Together, these findings provide initial evidence that ongoing thoughts offer a reliable window into social preferences, introducing a novel framework for studying the link between ongoing cognition and social behavior in the laboratory and everyday life.

Contemporary accounts of semantic cognition propose that conceptual knowledge is supported by a heteromodal conceptual store and controlled retrieval processes. However, it remains unclear how the neural basis of semantic control varies across modalities. Recent models of cortical organisation suggest that control networks are distributed along a unimodal-to-heteromodal cortical gradient, with the semantic control network (SCN) located in more heteromodal cortex than the domain-general multiple demand network (MDN). We used fMRI to examine how these networks respond to semantic control demands in visual and auditory tasks. Participants judged the semantic relatedness of spoken and written word pairs. On half of the trials, a task cue specified the semantic feature to guide retrieval; on the remaining trials, no such cue was given. The SCN showed greater activation when task knowledge was available, consistent with a role in the top-down control of semantic retrieval across modalities. In contrast, the MDN showed greater activation for spoken words, likely reflecting increased demands in speech perception. These findings demonstrate a dissociation between control networks, with SCN involvement modulated by task structure and MDN activity influenced by perceptual difficulty.

Motor learning induces alterations in neural activity that can persist long after the effects of such learning have faded. These persistent neural alterations are thought to manifest behaviorally as "savings," or faster relearning, via access to a latent motor memory. How the human brain forms and retrieves these latent memories, and the specific neural systems involved, remains unresolved. Here, using human functional MRI and a two-day sensorimotor adaptation paradigm, we show that savings are associated with the reinstatement of a large-scale cortical manifold structure formed during initial learning. Notably, this neural reinstatement effect was not observed across sensorimotor systems but was localized to regions of the default mode network (DMN). Moreover, the specific dynamics of DMN activity were linked to inter-subject differences in patterns of learning and relearning across days. These results suggest that motor savings arises from the re-expression of DMN activity patterns associated with initial learning, establishing a key role for this network in motor memory formation and retrieval. This finding, paralleling reinstatement principles from other memory domains (episodic memory, fear conditioning) and anticipated by recent computational models of motor learning, suggests a common mechanism for the flexible recall and reuse of stored memories across diverse behavioral contexts.

How does the brain support the complex processes that allow us to read? Using predictive modeling we establish that visual and association cortex are closer together in individuals with stronger oral reading ability. These findings indicate that large-scale cortical geometry provides a scaffold that supports the coordinated processing required to read.