Literature Review: Interoception and Emotional Granularity: Reading the Body’s Signals

This systematic review synthesizes research on interoception—the perception of internal bodily signals—and its critical relationship to emotional granularity, which is the precision in differentiating emotional states. It finds that interoception functions as a predictive process within a neural network involving the insula and anterior cingulate cortex, forming the basis for constructed emotions according to theoretical frameworks. Greater interoceptive accuracy predicts finer emotional granularity, which in turn supports better emotion regulation, whereas impairments are linked to conditions like alexithymia and various psychopathologies. Developmentally, interoceptive competence is not innate but is scaffolded through early caregiving relationships and co-regulation. Furthermore, cultural context, language, and socialization meaningfully shape both the neural processing of bodily signals and the construction of emotional experiences. Interoceptive ability is trainable through practices like mindfulness, and its disruption is a transdiagnostic mechanism across numerous clinical conditions. The review ultimately advocates for standardized measurement, culturally sensitive clinical integration, and early, caregiver-involved interventions to advance health outcomes.

1. Introduction

The relationship between the body and the mind has occupied philosophers, clinicians, and scientists for centuries, yet it is only within the past two decades that empirical research has begun to illuminate the precise mechanisms by which internal bodily signals shape subjective experience, emotional life, and psychological health. Interoception — broadly defined as the sensing, interpretation, and integration of signals arising from within the body [1, 2] — has emerged as one of the most generative and rapidly evolving constructs in contemporary affective neuroscience and clinical psychology [3, 4]. What was once a peripheral concern in sensory science has become central to understanding how human beings construct emotional meaning, regulate affective states, and maintain a coherent sense of self [5, 6]. The convergence of neuroimaging advances, computational modeling, psychophysiological measurement, and clinical observation has transformed interoception from a largely theoretical curiosity into an empirically tractable domain with direct implications for mental and physical health [7, 8].

The past decade in particular has witnessed several developments that render a systematic review both timely and necessary. First, the emergence of predictive coding and active inference frameworks has fundamentally reoriented how researchers conceptualize the brain’s relationship to the body, shifting the dominant model from one of passive signal detection toward one of dynamic, hierarchical prediction and error correction [9, 10, 11]. Second, the construct of emotional granularity — the degree to which individuals differentiate their emotional experiences into discrete, fine-grained categories rather than broad, undifferentiated states — has been theoretically and empirically linked to interoceptive processes [12, 13], opening new questions about how bodily awareness scaffolds the richness of emotional life [14, 15]. Third, growing clinical evidence has implicated disrupted interoceptive processing across a remarkably broad range of psychiatric and somatic conditions, from anxiety and depression to eating disorders, chronic pain, and autism spectrum conditions [8, 16], pressing researchers to move beyond descriptive accounts toward mechanistic explanations [17]. Fourth, a significant methodological debate has emerged around how interoception is operationalized and measured, with accumulating evidence suggesting that different assessment approaches — heartbeat detection tasks, questionnaire-based measures, and physiological indices — tap partially dissociable constructs rather than a unitary ability [18, 19, 13]. This fragmentation of the measurement landscape has made cross-study synthesis both more challenging and more essential. Fifth, and most recently, developmental research has begun to illuminate how interoceptive capacities emerge and are shaped from infancy through adolescence, revealing that the caregiving environment functions as a primary scaffold for children’s developing bodily self-awareness — a finding that fundamentally repositions interoceptive competence as a relationally acquired capacity rather than a fixed biological endowment [20, 21]. Sixth, cross-cultural and cultural neuroscience research has begun to demonstrate that interoceptive processing and the construction of emotional experience from bodily signals are not universal biological constants but are meaningfully shaped by cultural context, language, and socialization — with emerging neuroimaging evidence showing that even the anterior insular cortex, the brain’s primary interoceptive hub, is recruited differently during emotional feeling depending on cultural background [22]. This cultural dimension fundamentally complicates any account of interoception and emotional granularity built exclusively on data from Western, educated populations, and presses the field toward greater attention to the cultural embeddedness of bodily self-awareness.

Against this backdrop, the present systematic review examines literature published between 2002 and 2025, synthesizing findings across sixty studies to address four interconnected research questions. First, what neural and physiological mechanisms underlie interoceptive awareness [23, 24]? Second, how does interoceptive accuracy relate to emotional differentiation and regulation [25, 26]? Third, what individual differences predict interoceptive ability, and can this ability be deliberately trained [27]? Fourth, how do disruptions in interoception manifest across psychiatric and somatic conditions, and what are the implications for clinical intervention [8, 17]?

The review is organized around six thematic domains, each representing a distinct but mutually informing strand of inquiry. The first addresses the neural architecture of interoception, tracing the ascending and descending pathways through which visceral information reaches cortical awareness and establishing the foundational neurobiological models that have shaped the field [7, 23]. The second examines predictive coding, active inference, and other computational frameworks that have reconceptualized interoception as a generative, inference-driven process rather than a simple readout of peripheral signals [9, 11, 28]. The third turns to the methodological challenges of measuring interoception, evaluating the psychometric properties of dominant paradigms and the theoretical consequences of how the construct is operationalized [18, 19]. The fourth explores the relationship between interoception, emotion regulation, and emotional granularity, examining how the precision and interpretation of internal signals shape the differentiation of emotional experience and the capacity for affective self-regulation [12, 15, 29]. The fifth covers contemplative practices and body-based interventions, reviewing evidence that interoceptive awareness is not a fixed trait but a trainable capacity responsive to deliberate practice, including mindfulness, somatic therapies, and movement-based approaches [27]. The sixth and final theme examines clinical disruptions of interoception, mapping the ways in which both hyper- and hypo-interoceptive processing contribute to the onset, maintenance, and expression of psychopathological and somatic conditions [8, 16, 17].

Taken together, the evidence reviewed here suggests that interoception occupies a foundational position in the architecture of emotional life [6, 4] — not merely as a physiological substrate, but as an active interpretive process through which the body becomes legible to the self. Understanding this process, and what happens when it breaks down, is increasingly recognized as essential to explaining how human beings feel, think, and suffer.

2. Methodology

The literature informing this review was identified through a systematic search of the OpenAlex database, using five targeted keyword queries designed to capture the principal dimensions of interoception and emotional granularity as a research domain. These queries addressed, respectively, the neural mechanisms and neuroimaging correlates of interoceptive awareness, the relationship between interoceptive accuracy and emotional differentiation and regulation, training-based interventions and individual differences in interoception, interoceptive dysfunction in psychiatric and somatic conditions, and task-based paradigms such as heartbeat detection in relation to emotional awareness. Each query was constructed to draw from recent and established literature across both basic and clinical perspectives, reflecting the breadth of the topic.

From an initial pool of 158 candidate records returned across these searches, relevance scoring was applied with a threshold of 0.6, yielding 30 papers for inclusion. Papers were evaluated against a set of quality filters designed to balance scholarly rigour with contemporary relevance: older papers were required to have accumulated at least five citations, while a recency quota ensured that at least 35% of the final corpus derived from work published within the preceding two years. The date range of included papers spans 2002 to 2025, allowing the review to situate current empirical findings within the longer theoretical development of the field.

Citation network expansion was also pursued as a supplementary discovery strategy. Although forward and backward citation searches were conducted and a substantial number of candidate papers were examined — with 53 ultimately identified as relevant and 152 rejected at this stage — the process reached its collection target before additional rounds of expansion were necessary. The coverage delta recorded at this stage was 0.26, indicating meaningful but bounded bibliographic growth beyond the initial keyword results.

Three papers that were initially selected could not be retrieved and were replaced with alternative papers drawn from the existing candidate pool, ensuring that the intended corpus size was maintained without compromising the selection logic. All 30 papers in the final corpus underwent full-text analysis; no papers were limited to abstract-only assessment, and there were no unresolved retrieval failures.

Supplementary Developmental Literature

An additional targeted search was conducted to address the developmental origins of interoception and emotional granularity — a dimension that emerged as underrepresented in the initial corpus despite its theoretical centrality. This supplementary search focused on peer-reviewed empirical and theoretical work published between 2008 and 2025 addressing the emergence of interoceptive awareness across infancy, childhood, and adolescence [30, 20, 31]; caregiver and relational antecedents of interoceptive development [32, 33, 34]; developmental trajectories of emotional granularity and differentiation [29, 26, 35]; and the intersection of atypical interoceptive development with psychopathology risk [36, 37, 38, 17]. Notably, psychometric work validating multidimensional interoceptive awareness measures in youth aged 7–17 has revealed that body trust declines significantly across adolescence and that reliability profiles differ substantially from adult norms [20], underscoring why developmentally sensitive measurement frameworks are essential to this area of inquiry. From this search, 14 additional papers were identified that contributed distinct evidential or theoretical content not captured by the primary corpus. These papers were subjected to the same full-text analysis and thematic classification procedures applied to the initial set, and their findings were integrated throughout the relevant sections of the review rather than confined to a single developmental chapter.

Supplementary Cross-Cultural and Cultural Neuroscience Literature

A targeted supplementary search addressed the cross-cultural dimensions of interoception and emotional granularity, a domain that emerged as theoretically central for evaluating the generalizability of findings derived predominantly from Western samples. This search focused on peer-reviewed work published between 2002 and 2025, examining cross-cultural differences in interoceptive awareness and somatic attention [39, 40, 41], cultural neuroscience of insular cortex function and interoceptive processing [22], the role of language and emotion concepts in constructing affective experience across cultures [42, 43], and cultural variation in somatization and alexithymia [44, 39].

From this supplementary search, 8 additional papers were identified that contributed distinct evidential or theoretical content not captured by the primary or developmental corpora. These papers underwent the same full-text analysis and thematic classification procedures applied to the initial set. Their findings were integrated throughout relevant sections of the review—particularly concerning emotional granularity, measurement, and clinical disruptions—rather than being confined to a single cross-cultural chapter.

Thematic Organisation

The combined corpus of papers was organised into six thematic clusters, reflecting the multi-layered structure of scholarship at the intersection of interoception and emotional granularity. These clusters broadly correspond to the conceptual architecture of the review: neural and predictive processing accounts of interoception [10, 9]; measurement approaches and task paradigms [18, 13]; the relationship between interoceptive signals and affective differentiation [12, 25]; individual differences and trait-level predictors; intervention and training studies [26]; and clinical applications spanning psychiatric and somatic symptom conditions [8, 4, 45]. This organisation was derived from the content structure of the corpus rather than imposed a priori, allowing the thematic groupings to reflect genuine patterns in how the literature has developed rather than a predetermined framework.

Together, these methodological decisions — selective but broad querying, relevance-based filtering, citation expansion, quality thresholds that weight both impact and recency, and supplementary developmental and cross-cultural searches — were designed to produce a corpus that is analytically tractable while remaining representative of a field that continues to evolve rapidly at the intersection of affective neuroscience [7], clinical psychology [17], and embodied cognition [10].

3. Neural Architecture of Interoception: Pathways, Brain Regions, and Foundational Models

The neural architecture of interoception—how the body communicates its physiological state to the brain and how the brain integrates, interprets, and acts upon those signals—has emerged as one of the most consequential topics in affective neuroscience and cognitive science. This section traces the conceptual and empirical development of interoceptive neuroscience from its definitional foundations through the delineation of ascending pathways, the identification of key cortical hubs, and the latest structural and connectivity evidence situating interoception within broader brain network organisation.

Defining Interoception: From Sherrington to Craig and Beyond

The conceptual history of interoception is inseparable from its scientific trajectory. Early work by Sherrington in 1906 defined interoception narrowly as visceral sensation—a modest claim that persisted largely unchallenged for nearly a century [2]. A pivotal reconceptualisation arrived with Craig’s landmark 2002 paper, which reframed interoception as a distinct sensory system detecting the physiological condition of the entire body, not merely visceral organs [46]. Craig argued that this system underpins subjective feeling states and is fundamentally tied to emotional experience [5, 3], effectively elevating interoception from a peripheral concept to a central concern of affective neuroscience. A subsequent bibliometric analysis confirmed the magnitude of this conceptual shift, documenting a dramatic increase in the use of the term “interoception” after 2002—a shift traceable almost entirely to Craig’s influence [2].

Yet Craig’s framework itself was subject to refinement. Ceunen and colleagues argued that interoception should be understood as a subjective central nervous system product that is not exclusively dependent on afferent sensory input, proposing that integration of homeostatic and non-homeostatic signals at the mid-insula is constitutive of interoceptive experience rather than incidental to it [2]. This view resonated with broader efforts to extend interoception beyond strictly homeostatic functions toward its roles in affect, cognition, and mental health [4]. By 2020, this trajectory toward a broader definition culminated in a comprehensive proposal that interoception encompasses bidirectional communication between brain and internal organs—including descending efferent regulatory pathways alongside the canonical ascending afferent routes [1, 8]. This bidirectional framing represents a significant departure from the predominantly feedforward models that characterised the early literature [11], emphasising that the brain is not merely a passive recipient of bodily signals but an active regulator shaping the very signals it receives [9].

Ascending Interoceptive Pathways: Vagal, Spinal, and Subcortical Architecture

Understanding how peripheral signals reach cortical representation requires attention to the specific anatomical routes involved. Craig’s foundational account described lamina I spinothalamic projections carrying homeostatic afferent signals—including temperature, pain, and visceral sensations—through the posterior thalamus toward the posterior insula [46]. Complementary neuroimaging and anatomical work has since confirmed that these thalamocortical projections terminate in a posterior-to-anterior gradient across insular cortex, with the posterior insula receiving primary interoceptive input and the anterior insula supporting higher-order integration [47, 8]. These pathways are characterised by biological properties quite unlike those of exteroceptive systems. Feldman and colleagues emphasised that interoceptive pathways feature sparse receptor distribution, thin unmyelinated C-fibres, and early signal compression at brainstem and midbrain relay stations before information reaches cortical levels [7]. This profile is further elaborated by Chen and colleagues, who note that interoceptive afferents—including both vagal and spinal routes—converge on the nucleus tractus solitarius (NTS) in the brainstem as a primary subcortical integration hub, before ascending projections reach the thalamus and cortex [1]. This early compression has important implications: it means that the cortex receives a substantially pre-processed, compressed representation of bodily state rather than high-fidelity sensory data, a property that distinguishes interoception from vision or audition in fundamental ways.

The vagal route constitutes a parallel and complementary ascending pathway. Porges’s polyvagal theory, which has undergone substantial empirical and theoretical development over several decades, identifies the brainstem—specifically nucleus ambiguus and the dorsal motor nucleus of the vagus—as critical relays through which autonomic information ascends to influence higher brain function [48]. A distinctive contribution of polyvagal theory is its evolutionary framing: mammals are argued to have uniquely developed a ventral vagal complex with myelinated cardioinhibitory fibres from nucleus ambiguus, enabling social engagement behaviours absent in phylogenetically older species [48, 49]. Heart rate variability and respiratory sinus arrhythmia are foregrounded as meaningful neural indices of brainstem regulation rather than statistical noise—a position supported by systematic reviews linking higher vagal tone to improved interoceptive sensitivity and emotional regulation [48, 50], and further corroborated by neuroimaging evidence connecting HRV to insula–default mode network functional connectivity [51].

The Insular Cortex as the Primary Interoceptive Hub

Convergent evidence positions the insular cortex as the principal cortical site of interoceptive integration. Craig’s 2002 framework proposed a posterior-to-anterior processing gradient within the insula, whereby posterior regions receive and map primary interoceptive signals while anterior regions re-represent these signals in progressively more integrated, affectively laden forms [46]. This gradient model has proven remarkably durable. The 2020 review by Chen and colleagues confirmed it, distinguishing between a posterior insula that integrates interoceptive with sensorimotor and proprioceptive data, and an anterior insula that connects to emotional and cognitive circuits [1]. The anterior insular cortex (AIC) in particular emerged as a site of interest for its role in subjective awareness of bodily states and its extensive bidirectional connectivity with prefrontal, limbic, and autonomic systems [52]. Neuroanatomically, the AIC also features a disproportionate concentration of von Economo neurons—specialized cells enabling rapid long-distance communication—that may underpin its unique integrative capacity [52].

Wiens’s 2005 review added an important empirical dimension by showing that brain-imaging studies employing heartbeat detection tasks consistently found that AIC and anterior cingulate cortex (ACC) activation correlated with individual differences in interoceptive accuracy and with trait measures of negative affect [5]. Complementing this, neuroimaging work by Pollatos and colleagues identified AIC and ACC as the principal nodes activated during cardioafferent signal processing, with signal strength in these regions tracking subjective bodily awareness [24, 23]. Notably, Wiens’s review also challenged the strict James-Lange view that peripheral autonomic feedback is necessary for emotional experience: patients with spinal cord injury or pure autonomic failure showed only minimal emotional impairment despite markedly reduced peripheral feedback [5]. More recent work by Gu and colleagues further clarified that AIC lesions, but not ACC lesions, correlate with deficits in emotional awareness, establishing a functional dissociation between these two co-activated regions [52]. This tension—between the demonstrated importance of AIC in interoceptive awareness and the apparent dispensability of peripheral feedback for emotion—remains one of the productive ambiguities driving the field.

Anterior Insula, Emotional Awareness, and Temporal Processing

The anterior insular cortex has been implicated not only in interoceptive accuracy but in the temporal architecture of conscious experience. Craig proposed in 2009 that the AIC serially represents emotional feelings across successive moments, generating what he termed the “specious present” and serving as a neural basis for time perception in the seconds-to-subsecond range—with AIC and bilateral putamen functioning as a mental real-time clock [53]. This proposal extended his foundational model by linking interoception to the subjective flow of time, suggesting that the AIC’s role is not confined to snapshot bodily representation but encompasses the dynamic temporal integration of feeling states. Complementary structural work on primate insular organisation confirms that the insula progresses from granular tissue posteriorly to agranular tissue anteriorly, with increasingly compressed and integrated representations arising along this gradient—an architecture that may underpin the AIC’s capacity to bind discrete interoceptive moments into coherent affective episodes [47].

Subsequent meta-analytic work by Gu and colleagues, analysing 47 fMRI studies using activation likelihood estimation, identified a functional dissociation within the anterior insula: right AIC activation appeared selectively associated with pain and negative emotion conditions, suggesting preferential processing of energy-demanding aversive states [52]. Importantly, this work demonstrated that AIC and anterior cingulate cortex are not functionally interchangeable, with AIC showing a specific profile that distinguishes it from the broader salience network [52, 54]. A more recent neurobiological model introduced the concept of repetition suppression to explain why rhythmic and predictable interoceptive signals—like heartbeat—are typically suppressed from conscious awareness: because visceral signals are inherently rhythmic, they become highly predictable, and only deviations from homeostatic expectation—prediction errors—ascend the neural hierarchy to update the brain’s internal model and reach awareness [7]. This same architecture, the authors argue, explains the characteristically diffuse, temporally persistent, and poorly localised quality of affective states [7, 10].

Insula–Default Mode Network Connectivity and Structural Correlates

The most recent empirical work situates insular interoceptive function within large-scale network architecture. A 2025 neuroimaging study of healthy adults found a significant positive correlation between insula and default mode network functional connectivity (r = 0.493, p = 0.003), and, crucially, that left insula volume correlated significantly with heart rate variability (r = 0.365, p = 0.016) [51]. These findings bridge several previously parallel literatures: structural properties of the insula are linked to autonomic regulation as indexed by HRV, while functional connectivity anchors the insula within networks associated with self-referential processing and mind-wandering [55, 56]. This convergence suggests that the insula’s role as an interoceptive hub [1, 57] is not anatomically isolated but is expressed through, and perhaps constituted by, its embedding in broader cortical networks.

Taken together, the literature reviewed here traces a coherent if still-evolving architecture: interoceptive signals originate peripherally, are compressed and relayed through vagal and spinal routes with early subcortical processing, reach the posterior insula for initial cortical mapping [47, 8], and are progressively re-represented in the anterior insula in forms that engage emotional, temporal, and self-referential processing [53, 52, 23]. The discovery of structural links between insular volume and HRV [51], combined with the polyvagal emphasis on brainstem regulation as a meaningful neural signal [48], increasingly positions interoception as a multilevel, network-embedded process rather than a simple sensory relay—a reconceptualisation that sets the terms for all downstream research in the field.

4. Predictive Coding, Active Inference, and Computational Frameworks of Interoception

Predictive coding and active inference frameworks have fundamentally reoriented how researchers conceptualize interoception, transforming it from a passive sensory channel into a dynamic, generative process through which the brain actively constructs and regulates its model of the body’s internal milieu. Rather than treating visceral signals as straightforward readouts of peripheral physiology, these theoretical accounts reconceptualize bodily perception as the product of hierarchical inference, where descending predictions suppress ascending prediction errors, and where mismatches between anticipated and actual states drive both perception and action. This computational turn has generated some of the field’s most formally precise explanatory frameworks, with cascading implications for accounts of emotion, selfhood, consciousness, and psychiatric dysfunction.

Predictive Coding and the Suppression of Interoceptive Signals

Early theoretical work in this tradition established the foundational claim that conscious bodily experience is not built from raw sensory signals but from the brain’s attempts to explain those signals away. [58] offered a landmark formulation: conscious presence — the felt sense of existing as an embodied agent in a real world — arises from the successful suppression of interoceptive prediction errors via top-down predictions generated in the anterior insular cortex (AIC). On this account, the phenomenological quality of “being present” reflects not heightened sensory access but the ongoing minimization of discrepancy between the brain’s generative model of the body and incoming visceral signals. This framing was subsequently extended by [10] and [9], who formalized interoceptive inference as a form of active inference in which the brain not only models but actively regulates visceral states, with precision-weighting determining how much influence ascending interoceptive signals exert on conscious experience. Crucially, [58] applied this logic to depersonalization/derealization disorder, reinterpreting AIC hypoactivation not as a failure to detect interoceptive signals but as a consequence of pathologically imprecise interoceptive predictions — a subtler computational dysfunction than earlier clinical descriptions had suggested. This account aligns with broader neurobiological evidence linking insular dysfunction to disrupted bodily self-consciousness [8].

This suppression logic received neurobiological grounding in subsequent work. [7] demonstrated that interoceptive signals are intrinsically rhythmic and temporally predictable, rendering them especially susceptible to repetition suppression effects that attenuate conscious access precisely because they are anticipated. This observation reinforces the predictive coding model: the brain’s familiarity with the body’s baseline dynamics becomes the mechanism through which those dynamics recede from awareness, surfacing only when predictions are violated — a principle also articulated in the context of interoceptive predictions more broadly by [11]. Importantly, [7] also identified structural features that distinguish interoceptive from exteroceptive pathways — sparse receptor distributions, thin unmyelinated fibers, and early-stage signal compression in the brainstem and midbrain — properties that constrain the precision of ascending signals well before they reach cortex and that must be incorporated into any biologically adequate generative model. These anatomical constraints are consistent with evidence that interoceptive afferents project through lamina I spinothalamic pathways to subcortical and cortical targets via routes that impose substantial compression and transformation on raw visceral signals [1].

Active Inference and Hierarchical Generative Models

The theoretical ambitions of this framework extended considerably beyond interoceptive neuroscience with the development of active inference as a unified account of sentient behavior. [9] argued that bodily states are not merely sensed but actively regulated: descending predictions from deep hierarchical generative models constitute efference that coordinates autonomic and somatic responses, making interoceptive inference an inherently action-oriented process rather than a purely perceptual one. This relocates emotion from a readout of peripheral physiology to an emergent property of hierarchical inference on the causes of interoceptive signals — a move that aligns with and formally specifies constructionist accounts of affect [6]. The framework positions the autonomic nervous system less as a reflex arc and more as an effector system through which the brain enacts its predictions about the body’s required internal state.

This synthesis reached its most comprehensive articulation in [59], which presented active inference as a unified theory of perception, action, planning, and learning under a single principle of variational free energy minimization. By grounding the framework in the statistical physics of biological self-organization and formalizing the role of Markov blankets in separating internal from external states [21], [59] showed how organisms come to embody models of their environments — including their own bodies — through ongoing cycles of prediction and correction. The account is notable for bridging computational formalism with evolutionary and developmental considerations, situating active inference not as a neuroscientific hypothesis alone but as a principle governing self-organizing systems at multiple scales.

A critical regulatory mechanism running through all these accounts is precision-weighting: the differential scaling of prediction errors according to their estimated reliability. High-precision signals command greater updating of generative models; low-precision signals are effectively silenced. Computational modeling work has operationalized this mechanism empirically, demonstrating that interoceptive training increases the precision weighting afforded to afferent cardiac signals (ΔIP₂) and that these increases correspond to reductions in anxiety symptoms [60] — providing quantitative traction on what had previously remained a largely theoretical construct. [27] operationalized this distinction in a clinically and contemplatively relevant register, proposing that mindfulness-based practices induce a shift from active inference — in which the brain alters bodily states to conform to prior expectations — toward perceptual inference, in which priors are revised to accommodate incoming sensations. This reframing suggests that contemplative interventions modify not the sensory periphery but the relative precision assigned to ascending interoceptive signals, thereby increasing the influence of bottom-up evidence on conscious experience and bodily regulation.

Aphantasia, Selfhood, and Disrupted Interoceptive Prediction

One of the more striking recent extensions of this framework concerns aphantasia — the absence of voluntary visual mental imagery — which [61] reconceptualized as a failure of interoceptive prediction rather than a purely visual or mnemonic deficit. In this 2025 account, reduced interoceptive precision in the insula [11, 10] disrupts the capacity of prefrontal top-down signals to sufficiently activate parietal and visual cortices during imagery generation. Aphantasia thus becomes a disorder not of image storage or retrieval but of the embodied scaffolding that normally confers the subjective vividness and sense of agency over imagined content. Empirical support for this view comes from Monzel et al. [62], who found — across 468 participants — that aphantasics and hypophantasics scored significantly lower on interoceptive measures, with emotional awareness and body listening showing the strongest associations with imagery vividness, and that mental imagery statistically mediated the relationship between interoception and autobiographical memory richness. Complementing this, Silvanto and Nagai [63] demonstrated that a general sense of agency mediates the relationship between subjective interoception and mental imagery vividness across the full aphantasia–hyperfantasia spectrum, linking bodily self-awareness to individual differences in voluntary image generation. This argument generalizes the Seth-Critchley presence model [58] to voluntary cognition, proposing that agency over mental imagery is itself an interoceptive inference phenomenon — a claim that substantially broadens the explanatory reach of the framework.

Computational Psychiatry: Hyperprecise Priors and Context Rigidity

The clinical implications of hierarchical interoceptive inference were systematically developed by [17], who identified two computational failure modes underlying interoceptive psychopathology: hyperprecise priors, in which expectations are so strongly weighted that ascending prediction errors cannot update the generative model, and context rigidity, in which the system fails to adjust prior expectations when environmental conditions shift. Complementary work on hierarchical Bayesian modelling has demonstrated that increased precision weighting for afferent bodily signals is empirically measurable and correlates with maladaptive interoceptive processing [60], while active inference accounts of the emotional brain situate both failure modes within a broader architecture linking autonomic regulation to top-down belief control [9]. These mechanisms, reinforced through cycles of mental rehearsal and hierarchical feedback, account for the persistence and cross-context stability of symptoms in conditions such as panic disorder, somatic symptom disorder, and PTSD [17, 28]. The framework proposed by [17] translates abstract Bayesian principles into intervention targets, suggesting that effective therapies must not merely correct erroneous beliefs but restructure the precision-weighting architecture that determines how those beliefs respond to disconfirming evidence — a goal further elaborated in computational accounts of functional neurological and somatic disorders [64, 65].

Dynamical Systems and the Architecture of Psychiatric State Transitions

Complementing the Bayesian tradition, a dynamical systems perspective has been brought to bear on the question of how regulatory systems — including interoceptive inference — transition between stable and disordered states. [66] imported concepts from ecology and climate science, including tipping points, alternative stable states, and critical slowing down, to model psychiatric conditions as attractors in a high-dimensional state space. On this account, a person may inhabit a healthy attractor or a disordered one, and the transition between them is not a smooth, proportionate response to stress but a sudden shift that becomes increasingly likely as resilience erodes. Crucially, this erosion is detectable in advance: as a system approaches a tipping point, it recovers more slowly from small perturbations and exhibits increasing autocorrelation and variance — a signature known as critical slowing down [66]. Dynamic indicators of resilience derived from this signature have shown empirical promise for predicting transitions into depressive episodes and state shifts in bipolar disorder [66], with dense time-series data from ecological momentary assessment emerging as a key methodology for capturing these individual-level fluctuations [67]. A complementary cultural-ecosocial systems perspective has further argued that psychiatric trajectories must be understood within multilevel environmental and social contexts, not solely as properties of the individual nervous system [68].

This dynamical systems framing converges with the active inference account in at least one important respect: both treat psychiatric conditions as consequences of systemic properties — attractor topology on one side, precision weighting on the other [17, 9] — rather than as categorical deficits, and both imply that understanding individual trajectories requires dense, temporally resolved data rather than cross-sectional snapshots [67].

Taken together, these accounts constitute the field’s most mathematically disciplined explanatory layer. They converge on interoception as an active, hierarchical, precision-regulated process [11, 10] whose perturbations ramify into emotion dysregulation, disordered selfhood, and psychiatric illness — while leaving open important empirical questions about how generative model parameters are acquired, modified, and operationalized for experimental measurement.

5. Measuring Interoception: Tasks, Psychometrics, and Individual Differences

Measuring interoception presents a fundamental methodological challenge: the construct is inherently multidimensional, and the tools used to operationalize it have not always kept pace with theoretical sophistication. Early work in the field tended to treat interoception as a unitary capacity indexed primarily by performance on heartbeat perception tasks, but subsequent psychometric research, neuroimaging studies, and individual-difference approaches have progressively dismantled this assumption, revealing that accuracy, sensibility, and metacognitive awareness are dissociable dimensions with distinct neural substrates and moderating variables. The most recent work, extending into 2025–2026, has further refined this picture by demonstrating that even within a single paradigm, distinct cognitive processes leave separable electrophysiological signatures—complicating any remaining attempts to treat interoceptive ability as a single underlying trait.

Heartbeat Perception Tasks: Tracking, Discrimination, and Their Dissociation

The heartbeat counting (tracking) task and the heartbeat discrimination task have been the dominant objective measures of interoceptive accuracy for several decades, yet they do not appear to tap the same underlying process. A pivotal contribution from [18] examined both paradigms within the same sample of 80 healthy participants, finding that while performance on each task exceeded chance at the group level, inter-individual variability was substantial and the two tasks showed only modest correlation with one another. Crucially, metacognitive interoceptive awareness—the correspondence between confidence and accuracy—was significant at the group level for the tracking task but not for the discrimination task, suggesting that participants’ subjective certainty about their performance reflects qualitatively different processes depending on the paradigm employed. This dissociation has important implications: studies that rely exclusively on one task may be measuring something partially idiosyncratic to that paradigm rather than capturing a domain-general interoceptive faculty [13]. Consistent with this concern, systematic reviews of interoceptive accuracy measures have highlighted the limited convergent validity across paradigms, underscoring that no single behavioural index can be treated as a comprehensive proxy for cardiac interoceptive ability [69].

The construct validity of the heartbeat counting task in particular has attracted sustained criticism. Because the task is susceptible to heart rate knowledge, breath-holding strategies, and general psychophysiological arousal, performance scores may confound genuine cardiac signal detection with peripheral physiological artifacts or cognitive strategies that bypass interoception altogether [17]. Confidence ratings, moreover, have been shown to be independently associated with distinct neural and psychological variables from accuracy scores, further complicating the interpretation of any single task metric [70]. Such considerations motivate the use of multiple convergent measures and underscore the importance of not treating any single paradigm as a gold-standard proxy for interoceptive ability.

Metacognitive Awareness and Confidence Calibration

A recurring theme across this literature is the need to distinguish first-order interoceptive accuracy from second-order metacognitive awareness—the degree to which individuals “know what they know” about their own bodily signals. [18] operationalized this distinction formally, showing that interoceptive accuracy and metacognitive interoceptive awareness are empirically separable and may carry different functional and clinical implications. High accuracy with poor metacognitive calibration, or conversely high confidence with low accuracy, represent distinct profiles that aggregate measures would obscure. Empirical support for this dissociation comes from clinical populations: in autistic children, for instance, greater social-affective symptom severity was independently predicted by elevated interoceptive confidence rather than by objective accuracy—a pattern explaining 35.8% of symptom variance and suggesting that miscalibrated confidence is the more clinically salient dimension [37]. Similarly, interoceptive sensibility and accuracy have been shown to dissociate across anxiety and depression profiles [71], further underscoring that these constructs should not be collapsed.

This metacognitive layer has broader relevance for psychopathology. Computational accounts of interoceptive dysfunction frame aberrant interoception as a failure of predictive inference, in which miscalibrated precision weighting—not merely inaccurate signal detection—drives symptoms [17, 9, 72]. Within such a framework, confidence calibration becomes theoretically primary: the critical deficit lies not in whether the body generates clear signals, but in whether the brain appropriately weights and updates its predictions about them [10, 28]. Measuring interoception without assessing metacognitive calibration therefore risks missing the most clinically relevant dimension.

EEG Correlates: Dissociating Accuracy from Sensibility

Neurophysiological approaches have provided a complementary window into interoceptive processing, and early EEG work established that heartbeat-evoked potentials (HEPs) differentiate individuals with varying interoceptive accuracy. [24] demonstrated that good heartbeat perceivers showed significantly larger HEP amplitudes in the 250–450 ms post-R-wave window compared to poor perceivers, with dipole source localization pointing to the anterior cingulate cortex, medial frontal gyrus, inferior parietal cortex, and right insula as key generators. Converging neuroimaging evidence has since confirmed that these same regions — particularly the insula and anterior cingulate — are central nodes in the broader neural circuitry supporting interoceptive awareness and cardioafferent signal processing [23]. This body of work established a neural signature for interoceptive accuracy at a time when the field lacked a strong mechanistic account of why some individuals perceive their heartbeat more reliably than others.

A critical conceptual advance came with the formal tripartite distinction between interoceptive accuracy (objective performance), sensibility (self-reported bodily attention, typically assessed via questionnaire measures such as the Body Awareness Questionnaire or the MAIA [73, 19]), and awareness (metacognitive correspondence between confidence and accuracy), which demonstrated empirically that these dimensions can be dissociated and show only selective interrelationships — particularly in high-accuracy individuals [18]. Building on this framework, recent work has substantially extended the electrophysiological picture by dissecting the neural correlates of accuracy and sensibility independently. [70] employed task-related alpha power (TRP) during the heartbeat tracking task and found that these two dimensions engage opposite patterns of cortical oscillatory activity. Higher interoceptive sensibility — the subjective tendency to attend to and report bodily sensations — was associated with stronger global alpha power decreases, consistent with a broadly active, cognitively effortful processing mode in which alpha suppression indexes widespread cortical recruitment [74]. Higher interoceptive accuracy, by contrast, was linked to relatively greater alpha power concentrated at left posterior regions (T7, P7, P3), suggesting more automatic or efficient processing that does not require widespread cortical recruitment — a pattern consistent with the gating-by-inhibition framework, wherein alpha increases at task-irrelevant sites reflect selective inhibition rather than global engagement [70]. Importantly, these effects survived covariate control for systolic blood pressure, BMI, gender, and age, and were specific to the alpha band, with no comparable associations emerging in theta or beta frequencies. This double dissociation provides compelling evidence that accuracy and sensibility are not simply high and low ends of the same dimension but reflect functionally distinct modes of engaging with cardiac information.

Multidimensional Self-Report: Development and Validation of the MAIA

Recognising that objective tasks capture only a narrow slice of the interoceptive construct, researchers have developed multidimensional self-report instruments. The Multidimensional Assessment of Interoceptive Awareness (MAIA), developed by [75] through a mixed-methods process involving focus groups with mind-body practitioners and patient populations followed by expert panel review and psychometric evaluation, operationalizes interoception across eight dimensions including noticing, not-distracting, not-worrying, attention regulation, emotional awareness, self-regulation, body listening, and trusting. A critical innovation of the MAIA was its explicit differentiation of adaptive and maladaptive forms of somatic attention — prior instruments had conflated anxiety-driven hypervigilance with mindful, accepting bodily awareness, thereby obscuring theoretically meaningful distinctions [75, 19]. The instrument demonstrated sensitivity to experience level, differentiating mind-body practitioners from general populations, and has since become the dominant self-report framework in the field, having been translated into over twenty languages and deployed across a wide range of clinical and non-clinical populations [73, 41].

A revised version, the MAIA-2 [73], addressed persistent psychometric weaknesses in the original instrument — particularly the low internal consistency of the Not-Distracting and Not-Worrying subscales, which had been attributed to their reliance on reverse-scored items and small item counts. Developed with a sample of over 1,000 English-speaking adults, the MAIA-2 added six new items (three per problematic subscale) drawn from the original focus group item pool, producing a 37-item version with Cronbach’s alphas ranging from 0.64 to 0.83 across subscales and acceptable model fit indices (RMSEA = .055, SRMR = .064) [73]. Despite these revisions, the Noticing and Not-Worrying subscales continued to fall below the conventional reliability threshold of 0.70, foreshadowing concerns that would recur in subsequent cross-cultural validation work.

[13] validated an Italian translation of the MAIA in a sample of 321 psychology students, confirming acceptable reliability and a factorial structure consistent with versions validated in other languages. However, the relationship between MAIA scores and objective heartbeat perception was weak: only the Attention Regulation subscale showed a modest correlation with accuracy, and no association emerged with emotional susceptibility. This pattern of near-independence between self-report and objective performance has been replicated across multiple studies and constitutes one of the field’s most robust and troubling findings — individuals’ beliefs about their own interoceptive sensitivity are poor proxies for their actual cardiac signal detection ability [18, 13, 19].

A more recent Czech adaptation of the MAIA, conducted with 431 participants and employing longitudinal network modelling alongside confirmatory factor analysis, replicated the eight-factor structure with only minor modifications but introduced a methodologically significant refinement [76]. Where earlier validation studies had relied primarily on factor-analytic approaches, the network modelling approach allowed examination of directional relationships among MAIA subscales. The results were theoretically provocative: attitudinal mindsets — specifically Not-Distracting and Not-Worrying — functioned as source nodes within the network, predicting the development of other interoceptive skills rather than being predicted by them [76]. This challenges the implicit assumption in the original MAIA framework that neutral, non-anxious bodily perception flows naturally from cultivated awareness skills; instead, it suggests that the willingness to attend without distraction or worry may be a prerequisite for the development of more sophisticated interoceptive competencies. The Czech study also found that wellbeing appeared as a precondition for interoceptive skill acquisition rather than its downstream product, and that the Not-Distracting, Not-Worrying, and Noticing subscales showed poor internal consistency — a finding that partially replicates concerns raised in prior validation work, including the MAIA-2 development study [73], and raises questions about whether these subscales function as coherent latent constructs or instead reflect heterogeneous collections of items whose apparent unity is context-dependent [76]. These cross-cultural psychometric findings carry implications beyond measurement: if attitudinal openness is upstream of interoceptive skill, then clinical interventions may need to target evaluative and avoidant attitudes toward bodily experience before expecting gains in perceptual accuracy — a sequencing question with direct relevance for the intervention literature reviewed in Section 7.

Complementing the MAIA’s dimensional approach, the Levels of Emotional Awareness Scale (LEAS) provides a performance-based assessment of emotional awareness — a construct closely intertwined with interoceptive processing — that avoids the limitations inherent in self-report measurement [35]. The LEAS conceptualises emotional awareness as a hierarchically organised socio-emotional skill, assessable through written responses to emotionally evocative scenarios rather than through subjective ratings. The scale demonstrates robust psychometric properties, with internal consistency ranging from 0.75 to 0.88, inter-rater reliability exceeding 0.81, and satisfactory test-retest reliability at four weeks [35]. Critically, higher emotional awareness as indexed by the LEAS correlates with greater cardiac parasympathetic tone, more differentiated somatic symptom reporting, superior impulse control, and heightened engagement of the anterior cingulate cortex during emotional processing — converging with the neuroimaging evidence reviewed in Section 3 on AIC and ACC as key interoceptive-emotional integration sites [35]. The LEAS thus represents a methodological complement to objective heartbeat tasks and subjective questionnaires, capturing a dimension of interoceptive-emotional competence — the complexity and differentiation of emotional representations — that neither accuracy scores nor sensibility ratings alone can access.

Neural and Demographic Moderators: Age, Insula, Anxiety, Sleep, and Gender

Individual differences in interoceptive ability are not randomly distributed but are systematically patterned by age, neuroanatomy, affective state, and gender. Functional neuroimaging work has demonstrated that interoceptive attention to heartbeat and skin temperature recruits overlapping but distinct cortical networks, including the supplementary motor area, dorsal anterior cingulate, and mid-to-anterior insula for cardiac monitoring, with more posterior insular and somatosensory engagement for temperature monitoring [77]. The insula, in particular, serves as a primary cortical hub for integrating ascending visceral signals into conscious bodily awareness [56, 1], and its large-scale connectivity with the anterior cingulate and prefrontal regions supports both interoceptive inference and affective regulation [57]. The insula’s structural properties, not merely its functional engagement, also predict interoceptive outcomes: [78] found in a lifespan sample of 70 adults that greater insula volume predicted higher interoceptive sensibility independent of age, while accuracy was not predicted by either structural or demographic variable. Older age also independently predicted greater sensibility, suggesting that accumulated bodily experience or age-related shifts in interoceptive inference may enhance subjective awareness even where objective detection remains unchanged — a dissociation consistent with broader evidence that aging differentially affects interoceptive accuracy and metacognitive confidence [79].

Anxiety emerged as a critical moderator in the same study: greater interoceptive sensibility was associated with shorter sleep duration and lower sleep efficiency specifically among individuals with moderate-to-high anxiety, whereas this relationship was absent or reversed in low-anxiety individuals [78]. This interaction is theoretically important because it suggests that heightened bodily awareness is not inherently adaptive—its consequences depend on the affective context in which it is deployed. Relatedly, evidence from interoceptive-emotion research indicates that the direction and valence of attention to bodily signals is shaped by individual differences in emotional processing styles, with higher interoceptive awareness facilitating cognitive reappraisal under some conditions but amplifying distress under others [15].

Gender differences constitute another unresolved puzzle. Women reliably perform more poorly than men on objective heartbeat perception tasks, yet consistently outperform men on emotional recognition and processing measures [80] — a paradox that superficially contradicts somatic marker accounts linking interoception to emotional competence. [81] propose that this apparent contradiction is resolved by attending to differential language socialization: women may have developed richer emotional vocabulary and labeling skills that operate relatively independently of raw physiological signal detection, such that emotional ability is partly decoupled from the interoceptive accuracy substrate assumed by many theoretical models. This account has significant methodological implications, as it suggests that studies conflating emotional processing with interoceptive accuracy may be measuring two partially distinct constructs.

A further individual-difference finding with implications for the broader theoretical landscape concerns the relationship between interoceptive awareness and self-objectification. [82] found a significant negative correlation between objectively measured heartbeat perception accuracy and self-objectification in a sample of 50 female university students (r = −.31), supporting the theoretical proposition derived from objectification theory that an external, appearance-focused orientation toward the body may erode internal bodily attunement. Importantly, interoceptive awareness and private body consciousness were uncorrelated with each other yet independently predicted self-objectification, suggesting they capture meaningfully distinct dimensions of body-related self-awareness that should not be conflated in research design [82]. This finding extends the gender-difference literature by identifying a culturally shaped psychological orientation — self-objectification — as a moderator of interoceptive accuracy, rather than treating gender effects as purely biological. It also provides an external validity anchor for heartbeat perception tasks, demonstrating that objectively measured interoceptive accuracy relates predictably to theoretically grounded psychological constructs beyond the interoception domain itself.

Developmental Trajectories: Interoceptive Measurement from Infancy Through Adolescence

The measurement challenges described above become considerably more acute when extended to developmental populations, where the tools available and the constructs they can access are constrained by participants’ age, verbal capacity, and cognitive maturity. Empirical work on cardiac interoception in infants has yielded notably cautious results. [30] examined whether 5- to 7-month-old infants could discriminate heartbeat-synchronous from asynchronous stimuli using both looking-time paradigms and heartbeat-evoked potentials (HEPs) across emotional and self-related contexts. The findings were striking in their restraint: no behavioural preference for synchronous stimuli and no HEP modulation by emotional context were detected, although infants did show a clear preference for trimodal audiovisuocardiac stimuli over bimodal visuocardiac presentations, suggesting attentional sensitivity to the richness of multisensory input rather than to cardiac contingency per se [30]. This dissociation — heightened sensitivity to multimodal complexity without evidence of specific cardiac discrimination — serves as a sober empirical counterpoint to theoretical claims about early-emerging interoception and underscores the methodological difficulty of isolating cardiac signal processing in pre-verbal populations. Within the active inference framework, these null findings are interpretable as reflecting an immature hierarchical predictive architecture in which interoceptive priors have not yet been sufficiently shaped by experience to generate detectable cardiac discrimination [17, 31, 10]. The embodied cognition perspective similarly holds that the minimal self — including its interoceptive foundations — is scaffolded gradually through sensorimotor experience and caregiver interaction rather than being present in full form from birth [31].

By middle childhood and adolescence, measurable individual differences in interoceptive sensitivity are already functionally coupled to emotionally relevant behaviour. [36] examined 46 healthy children and adolescents aged 9 to 16 using heartbeat detection accuracy alongside psychophysiological measures of heart rate variability. Their cross-sectional findings revealed that better interoceptive sensitivity was significantly associated with lower use of maladaptive emotion regulation strategies — particularly rumination and self-devaluation — while higher resting HRV (indexed by RMSSD and HF-HRV) was positively associated with external regulation strategies, especially social support seeking [36]. Critically, interoceptive sensitivity and HRV emerged as dissociable constructs with distinct behavioural correlates, cautioning against their conflation in developmental models and replicating in a younger sample the accuracy-sensibility dissociations observed in adult populations [18, 70]. The study also identified maternal psychopathology as a contextual factor shaping children’s interoceptive profiles — a finding that positions interoceptive development as socially embedded from its earliest measurable stages [32, 33] and that connects to the caregiver-mediated processes discussed in Section 6. Consistent with this picture, Oldroyd and colleagues found that attachment-related processes — including parental sensitivity and the quality of early dyadic regulation — were significantly associated with children’s interoceptive awareness, providing independent evidence that the caregiving environment calibrates bodily self-knowledge from infancy onward [32].

These developmental findings carry implications for the broader measurement debate. If interoceptive accuracy is not reliably detectable in infants, emerges in measurable form by middle childhood, and is already functionally linked to emotion regulation strategy selection by adolescence, then the critical window for the establishment of interoceptive competence likely spans the first decade of life — a period during which no validated objective paradigm currently exists for continuous measurement [69, 83]. This gap extends to self-report instruments as well: Jones and colleagues’ adaptation of the Multidimensional Assessment of Interoceptive Awareness for youth aged 7–17 (MAIA-y) found that three of eight subscales — Noticing, Not-Distracting, and Not-Worrying — demonstrated poor internal consistency (Cronbach’s α = .36–.47), particularly in younger children, and that the eight-factor structure achieved only partial scalar invariance across age groups [20]. A notable developmental trend also emerged: older youth reported significantly lower trust in their bodily sensations, suggesting puberty-related shifts in interoceptive awareness that existing instruments are poorly equipped to track [20]. Together, these psychometric limitations confirm that the measurement gap is not merely one of objective paradigms but encompasses the subjective and multidimensional assessment of interoception across development. This gap is not merely methodological; it is structural. Harrison and colleagues have argued persuasively that the NIMH’s Research Domain Criteria (RDoC) framework contains a significant lacuna: despite strong evidence that sensory processing abnormalities — including interoceptive dysregulation — are transdiagnostic features of mental health conditions [17, 8], no dedicated sensory processing domain exists within RDoC [16]. The authors demonstrated that sensory processing meets all three RDoC inclusion criteria — identifiable psychological constructs, underlying neural circuits, and demonstrated clinical relevance — and advocated for its formal inclusion in an updated framework [16]. For developmental research specifically, this institutional gap has consequences: without a recognised domain, interoceptive measures risk being classified as peripheral rather than central variables in large-scale studies of childhood psychopathology, limiting the accumulation of normative developmental data needed to track interoceptive trajectories across the early years.

Gaps and Limitations

Several important gaps remain. The predominance of cross-sectional designs limits causal inference about whether, for example, insula volume drives interoceptive ability or whether repeated interoceptive practice shapes neural architecture over time [78]. The weak correspondence between self-report and objective measures represents an unresolved psychometric problem: either the constructs are genuinely distinct and both are valid targets of measurement, or one class of measure is substantially confounded [13, 18]. Existing neuroimaging studies are typically underpowered — a concern compounded by the fact that neural correlates of interoception are distributed across insular, cingulate, and somatosensory regions that require adequate sample sizes to detect reliably [77, 57] — and the alpha-band EEG findings, while theoretically compelling, await replication in larger and more diverse samples [70]. The near-absence of longitudinal designs tracking interoceptive capacity from pre-verbal infancy through adolescence represents the field’s most significant developmental gap, as does the lack of validated objective paradigms for continuous measurement across the full paediatric age range [30, 36]. The cross-cultural psychometric evidence — while confirming the broad structural replicability of the MAIA across Italian [13] and Czech [76] populations — remains confined largely to European samples. Studies validating the MAIA in Japanese populations have documented meaningful cross-cultural variation in interoceptive sensibility profiles [41], and broader reviews of cross-cultural differences in somatic awareness suggest that non-Western populations show systematically distinct patterns of interoceptive attention and accuracy [39, 40]. The subscale-level instabilities documented in European validation studies thus raise sharper questions about whether the instrument’s conceptual architecture fully generalises to non-Western cultural contexts where the relationship between bodily attention, emotional awareness, and attitudinal dispositions may be patterned differently. Finally, the field lacks consensus on which combination of task and self-report measures constitutes an adequate operationalization of interoception for any given research question — a methodological pluralism that, while arguably appropriate given the construct’s multidimensionality [19], complicates cross-study synthesis.

Yet it would be a mistake to frame these measurement challenges as primarily technical failures awaiting better instruments. The fragmentation documented here — across paradigms, dimensions, demographic groups, developmental stages, and cultural contexts — reflects something real about interoception itself: a construct whose multidimensionality is not an artefact of inadequate operationalization but a genuine feature of the phenomenon being studied. Accuracy, sensibility, metacognitive calibration, attitudinal openness, and emotional awareness are not interchangeable proxies for a single latent trait; they are partially independent capacities with distinct neural substrates [77, 70], developmental trajectories, and clinical consequences [84, 50]. Accepting this multidimensionality as substantive rather than merely inconvenient reframes the stakes of resolving these gaps: progress in measurement is not simply a prerequisite for cleaner data, but a condition for answering the questions that matter most — how interoceptive capacity shapes the differentiation and regulation of emotional experience, how disruptions in that capacity manifest across culturally diverse populations, and how clinical interventions can be sequenced and targeted to the specific interoceptive dimensions most relevant to a given individual’s profile [85]. These are precisely the questions taken up in the sections that follow.

6. Interoception, Emotion Regulation, and Emotional Granularity

Interoception — the perception of signals arising from within the body — has emerged as a foundational mechanism linking somatic experience to the richness, differentiation, and regulation of emotional life. Early theoretical work positioned bodily feedback as central to emotion itself, with the James-Lange tradition holding that peripheral autonomic changes constitute, rather than merely accompany, feeling states. Yet empirical challenges to this strong somatic view prompted a more nuanced picture: evidence from spinal cord injury patients and individuals with pure autonomic failure demonstrated that substantial reductions in peripheral autonomic activity produced only minimal impairment in subjective emotional experience, complicating simple feedback accounts [5]. Rather than abandoning the body’s role, however, subsequent decades of research progressively reframed the question — not whether interoception matters to emotion, but precisely how, through which pathways, and subject to what individual and contextual moderators.

The Theory of Constructed Emotion: Interoceptive Prediction as the Foundation of Emotional Meaning

The theoretical framework that most directly bridges interoceptive processing and emotional granularity is Barrett’s Theory of Constructed Emotion (TCE), which reconceptualizes emotions not as discrete biological entities triggered by dedicated neural circuits but as meaning-laden categories actively constructed by the brain through predictive processing [6]. On this account, the brain’s fundamental task is not to react to the world but to anticipate it: the brain continuously generates predictions about the causes of interoceptive signals, implementing allostasis by modelling the body’s metabolic and physiological needs before they arise and coordinating autonomic, immune, and endocrine responses accordingly. These ongoing interoceptive predictions constitute what TCE terms core affect — a continuous, fluctuating stream of bodily sensation characterized along dimensions of valence (pleasure–displeasure) and arousal (activation–deactivation) that forms the basic substrate of all affective experience [6].

Crucially, core affect does not itself constitute emotion. Discrete emotional experiences — instances of “anger,” “guilt,” “awe,” or “disappointment” — emerge only when the brain categorizes interoceptive predictions using available emotion concepts: learned, culturally transmitted, and context-sensitive conceptual structures that impose categorical order on otherwise undifferentiated affective signals [6, 42]. Language plays a particularly important role in this process, with emotion words functioning as abstract category anchors that hold emotion concepts together even in the absence of strong perceptual regularities in the body or brain [42]. Barrett’s framework thus positions emotions as constructed in the moment rather than detected from pre-existing biological essences, with instances of any given emotion category showing remarkable variability in their physiological, neural, and behavioral profiles rather than conforming to fixed signatures. Critically, large-scale neuroimaging meta-analyses support this constructionist account: no discrete emotion category has been found to activate a specific brain region consistently and exclusively, with every region implicated in one emotion also activating across others — evidence that the brain implements emotion through distributed networks supporting core affect, conceptualization, and attention rather than through dedicated emotion circuits [86]. Physical changes in the body and brain acquire emotional significance only when they are conceptualized — that is, when socially shared conceptual knowledge confers psychological functions that the physical changes cannot perform by their nature alone [87].

This constructionist account carries several implications that are fundamental to the present review. First, it positions interoception not merely as a correlate of emotion but as its generative substrate: without interoceptive predictions, there is no core affect, and without core affect, there is nothing for emotion concepts to categorize. Second, it formally explains why emotional granularity — the capacity to make fine-grained distinctions among emotional states — varies across individuals and contexts. Granularity, in TCE’s terms, reflects the precision and richness of the conceptual repertoire an individual can bring to bear on interoceptive experience [6, 29]. An individual with a sparse or rigid set of emotion concepts will tend to experience affect in broad, undifferentiated terms (“I feel bad”), whereas an individual with a rich, context-sensitive conceptual repertoire will construct more precisely differentiated experiences (“I feel disappointed” versus “I feel resentful”) — a distinction that empirical work on emotional granularity confirms is both measurable and consequential [29, 26]. Third, because TCE is itself a predictive processing account — emotion construction operates through the same hierarchical Bayesian inference described in active inference frameworks [9] — it provides theoretical coherence between the computational accounts of interoception reviewed in Section 4 and the empirical findings on emotional differentiation examined here. The active inference principle that descending predictions regulate bodily states through precision-weighted error correction is, within TCE, the very mechanism through which allostasis is implemented and through which core affect is generated as a by-product of the brain’s ongoing regulatory activity [6, 9].

Language and Emotion Concepts as Constitutive of Emotional Experience

The constructionist claim that emotion concepts mediate the transformation of interoceptive signals into discrete emotional experiences has received direct empirical support from research on the role of language in emotion. The Conceptual Act Theory (CAT) elaborated by [42] proposes that emotion words and the conceptual knowledge they carry are not post-hoc labels applied to emotions that have already formed through biological programmes but constitutive ingredients of emotional experience itself. On this account, the brain continuously generates predictions about the causes of interoceptive and sensory signals, and emotion concepts — including culturally learned categories such as “anger,” “schadenfreude,” or “saudade” — supply the categorical templates through which undifferentiated core affect is resolved into discrete, recognisable emotional experiences [42, 43]. When access to emotion word meanings is experimentally impaired — through verbal shadowing tasks that occupy the phonological loop, or through studies of patients with semantic impairments — participants show corresponding deficits in perceiving discrete emotions in facial expressions, suggesting that the recognition of another person’s emotional state depends on deploying one’s own emotion concepts rather than simply decoding a universal signal [42]. Conversely, labelling an affective state with a specific emotion word produces measurable physiological and experiential consequences distinct from those produced by alternative labels [25], implicating language not merely in representation but in the very regulation and differentiation of felt experience. Evidence from alexithymia research further corroborates this view: individuals who lack fluent access to emotion vocabulary show disrupted semantic-interoceptive integration and corresponding impairments in emotional differentiation [88].

The implications for emotional granularity are direct: if emotion concepts are constitutive of emotional experience, then the linguistic and conceptual resources available to an individual — resources that are profoundly shaped by cultural socialisation and language structure — determine the granularity with which that individual can differentiate affective states [26]. Languages that lexicalise distinct emotional states unavailable in other languages may afford different emotional experiences to their speakers [43], and socialisation into richer emotion vocabularies should produce measurably finer-grained emotional differentiation [42, 29]. This prediction aligns with and extends TCE’s framework by specifying language as the primary vehicle through which emotion concepts are acquired, transmitted, and deployed in the categorisation of interoceptive experience [6, 42].

Embodied Affectivity and the Pre-Reflective Roots of Emotional Differentiation

Complementing the emphasis on conceptual and linguistic scaffolding, a phenomenological tradition insists that the body’s dynamic relationship with its environment constitutes a distinct and foundational layer of emotional experience. [89] develop the concept of embodied affectivity, proposing that emotions arise through a circular feedback cycle between affective affordances in the environment and bodily resonance — the felt changes in posture, movement, muscular tonus, and visceral sensation through which organisms register the significance of situations. On this account, bodily states are not merely the downstream expression of emotions that have been constituted at a cognitive or conceptual level; rather, bodily dynamics are co-constitutive elements that actively shape the qualitative character of feeling [89]. Empirical evidence marshalled in support includes demonstrations that experimentally induced postural changes influence memory recall for emotionally valenced material, that suppression of facial muscular activity modifies humour ratings — findings consistent with the broader facial feedback literature [42] — and that bodily mimicry forms the substrate of interpersonal empathy through what the authors term “interaffectivity” — an interbodily resonance that precedes and grounds reflective understanding. This account resonates with interoceptive research demonstrating that individual differences in the accuracy of bodily signal detection are associated with the richness and specificity of emotional experience [5, 3], suggesting that the pre-reflective somatic layer described by Fuchs and Koch has measurable functional consequences for emotional differentiation.

The relationship between the embodied affectivity and conceptual act frameworks is one of complementarity rather than strict opposition. Both reject the view that discrete emotions are preformed biological programmes, and both treat emotion as constructed rather than discovered. Yet they differ in where they locate the primary medium of construction: [42] foreground conceptual and linguistic knowledge, while [89] foreground pre-reflective bodily dynamics that may operate prior to and independently of explicit conceptualisation. This raises a genuinely open theoretical question about the direction and primacy of these constitutive relations — a question that bears directly on how emotional granularity should be understood. If the embodied layer is indeed more foundational, as [89] argue, then the capacity for fine-grained emotional differentiation may depend not only on the richness of one’s emotion vocabulary but also on the quality of pre-reflective bodily engagement with the environment [12, 25] — a possibility that connects to the contemplative and body-based intervention evidence reviewed in Section 7. The phenomenological evidence from Irish men’s lived experience of emotional differentiation, which revealed a pre-articulation phase involving inward bodily attention preceding linguistic categorisation [90], provides empirical support for this layered model in which embodied and conceptual processes operate in sequence rather than in isolation.

Emotion Regulation as Constitutive Rather Than Corrective

A significant reconceptualisation relevant to the interoception-granularity link concerns the nature of emotion regulation itself. Influential theoretical work by [91] challenged prevailing models that treat regulation as a higher-order executive process imposed upon emotional activation from above, arguing instead that emotion regulation is a component of emotional activation itself — embedded within it rather than reactive to it. On this view, separating “the emotion” from “the regulation of the emotion” is analytically misleading; regulation is a property of the system’s ongoing dynamics distributed across physiological, attentional, and social systems simultaneously rather than an executive faculty that intervenes upon them [91]. This reconceptualisation has direct consequences for understanding granularity: if emotional states are always already shaped by regulatory processes operating at multiple levels, then the capacity to make fine-grained distinctions between emotions is not simply a post-hoc labelling skill but reflects the organisation of emotional experience from its inception. Emotional differentiation, on this view, is inseparable from the regulatory and physiological systems that constitute emotion in the first place.

The developmental implications of this systems-level framing are substantial. If regulation is constitutive rather than corrective, then caregiver co-regulation in early life does not merely supplement a developing self-regulatory system — it is part of the medium through which the child’s emotional and interoceptive systems are organised. Taipale’s [33] tripartite analysis of affect regulation makes this point with particular clarity: in early infancy, caregiver regulation is not assistance to a nascent self-regulatory system but its functional replacement, and self-regulatory capacity only emerges through the gradual internalisation of patterns first provided entirely by the caregiver. Crucially, this internalisation extends to the categorical organisation of emotional experience itself: sensitive caregivers who respond differentially to distinct forms of infant distress scaffold the infant’s ability to discriminate between discrete emotional states — the very capacity that constitutes emotional granularity [33]. Disruptions to co-regulation therefore do not simply leave a developing regulatory system unsupported; they shape the architecture of that system as it is being built [91]. This framing connects directly to evidence that attachment processes shape the quality of interoceptive engagement from early childhood onward — including through effects on the anterior insula, anterior cingulate cortex, and orbitofrontal cortex, regions central to both interoceptive processing and emotional awareness [32] (see below) — and it helps explain why the relationship between interoceptive accuracy and emotional granularity is neither simple nor unidirectional.

Interoceptive Attention, Emotional Awareness, and Strategy Selection

A critical conceptual contribution of more recent empirical work has been the disaggregation of interoception into distinct components — accuracy, attention, and awareness — each with potentially different consequences for emotional functioning. Early operationalizations tended to conflate these, typically using heartbeat detection paradigms as a proxy for the whole of interoceptive ability [18]. The tripartite distinction between interoceptive accuracy (objective signal detection), sensibility (self-reported bodily attention), and awareness (metacognitive correspondence between the two) has since been formalized as a principled empirical framework, with evidence that the three dimensions can be meaningfully dissociated yet also show selective interrelationships, particularly among individuals with high objective accuracy [18]. A multi-method validation study using the Multidimensional Assessment of Interoceptive Awareness (MAIA) across 321 university students found that interoceptive accuracy, measured via heartbeat perception, showed only a weak correlation with the Attention Regulation subscale and no relationship with emotional susceptibility as measured by self-report instruments [13]. This dissociation between accuracy and awareness dimensions implies that knowing one’s heartbeat does not straightforwardly translate into emotionally relevant self-knowledge — a finding that is interpretable within TCE as reflecting the distinction between the precision of ascending interoceptive signals and the conceptual apparatus available to categorize them.

More recent work has pursued the downstream consequences of interoceptive attention specifically. Using experience-sampling methods combined with one-time laboratory measurement, a 2022 study found that interoceptive attention was positively associated with the habitual use of multiple emotion regulation strategies, and that this relationship was mediated by emotional awareness [92]. This mediated pathway is theoretically significant: it frames emotional awareness as the interpretive bridge between detecting internal signals and flexibly deploying regulatory resources. Individuals who attend to bodily states are, on this account, better positioned to identify what they are feeling and thereby select contextually appropriate strategies — a mechanism consistent with the broader emphasis on interoceptive signals as inputs to predictive affective inference [7], and corroborated by evidence that interoceptive awareness specifically facilitates cognitive reappraisal as a regulatory strategy [15]. Within the constructionist framework, this pathway can be understood as one in which interoceptive attention provides the raw affective signal, emotional awareness reflects the successful application of emotion concepts to that signal, and strategy selection follows from the differentiated appraisal that such categorization enables [6]. Systematic reviews of the broader literature further confirm that interoceptive capacity and vagal tone together predict regulatory flexibility across multiple populations [50].

Critically, this attention-awareness-strategy pathway operates across development. In children and adolescents aged 9 to 16, interoceptive sensitivity as measured by heartbeat detection accuracy is already significantly associated with lower use of maladaptive regulation strategies — particularly rumination and self-devaluation — while heart rate variability independently predicts the use of externally oriented strategies such as social support seeking [36]. The dissociability of these two physiological contributors to regulation in youth mirrors the accuracy-sensibility dissociations observed in adults [18], and it suggests that the mediating role of emotional awareness documented in adult samples may have precursors in the developing links between bodily signal detection and regulatory repertoire that are established during childhood.

Attachment, Caregiving, and the Social Scaffolding of Interoceptive-Emotional Links

The constructionist claim that emotion concepts are necessary for transforming undifferentiated interoceptive signals into discrete emotional experiences [6] raises a developmental question: where do emotion concepts come from, and how is the link between bodily signals and emotional meaning established in the first place? Converging evidence points to the caregiving relationship as a primary site where this connection is forged. [32] provided direct evidence that attachment orientation systematically predicts the quality of interoceptive engagement: anxious attachment was associated with heightened noticing of bodily sensations and elevated emotional awareness, but this heightened attention came coupled with increased worry about bodily states, suggesting an amplified but distrustful mode of interoceptive engagement. Avoidant attachment, by contrast, predicted lower attention to bodily cues and reduced trust in internal signals, reflecting a suppressive stance toward somatic information [32]. Most strikingly, maternal rejection of children’s negative emotions was associated with measurable incongruence between subjective emotional reports and objective physiological arousal — evidence that caregiver responses can literally decouple children from their own bodies, producing a functional disconnection between felt experience and physiological reality [32]. This early dyadic regulation of affect, wherein the caregiver’s responsiveness scaffolds the infant’s capacity to integrate and interpret bodily states, is foundational to later autonomous emotional processing [33].

These findings are directly interpretable within TCE’s framework [6]. If emotion construction requires both interoceptive signals and emotion concepts, then caregiving practices that invalidate or override children’s bodily experiences may impair both components simultaneously: reducing the child’s trust in ascending interoceptive signals — the precision-weighting dimension [9, 72] — while also impoverishing the conceptual repertoire available for categorising those signals [42] (the conceptual dimension). The result is a double deficit in which undifferentiated interoceptive states are neither clearly perceived nor precisely named — the profile that characterises low emotional granularity. This developmental pathway suggests that the relationship between interoceptive accuracy and emotional granularity documented in adults [25, 12] may be partially constituted by early relational experiences that either support or undermine the integration of bodily signals with emotional meaning.

HRV as a Psychophysiological Marker of Regulatory Capacity

Parallel to the psychological literature, psychophysiological research has examined heart rate variability (HRV) as an index of the organism’s capacity for flexible, context-sensitive self-regulation. A systematic review conducted under PRISMA guidelines and drawing on five electronic databases synthesized evidence indicating that higher HRV consistently predicts better emotion regulation, particularly favouring cognitive reappraisal over suppression, and greater capacity for emotional downregulation more broadly [50]. This reappraisal advantage is mechanistically grounded in interoceptive processing: individuals with greater interoceptive awareness show both heightened initial emotional reactivity to negative stimuli and, critically, significantly larger reductions in arousal when deploying reappraisal — a pattern accompanied by corresponding modulation of neural indices such as the P3 and slow-wave components [15]. Critically, higher interoceptive accuracy also predicted regulatory capacity, especially in supporting reappraisal. Notably, HRV and interoceptive sensitivity appear to constitute complementary rather than redundant pathways: in adolescent samples, HRV preferentially predicts interpersonal and externally oriented regulation strategies, while interoceptive sensitivity more specifically buffers against maladaptive strategies such as rumination — and crucially, the two measures are themselves uncorrelated [36]. This convergence of HRV and interoceptive accuracy findings suggests that the autonomic nervous system and the conscious monitoring of bodily states constitute partially overlapping but distinct contributors to regulatory flexibility — a distinction with implications for interventions targeting either pathway.

Neural Substrates: AIC and ACC in Integrated Interoceptive-Emotional Awareness

The neural architecture supporting interoceptive-emotional integration has been progressively delineated. Early brain imaging studies using heartbeat detection tasks found that activation in the anterior insula cortex (AIC) and anterior cingulate cortex (ACC) correlated with individual differences in interoceptive ability and with trait measures of negative affect [5]. Subsequent work linking interoceptive awareness more directly to subjective emotional experience confirmed that AIC activation during bodily signal processing predicts the intensity of felt emotion [23]. A meta-analysis of 47 fMRI studies employing activation likelihood estimation methodology subsequently refined this picture, demonstrating that right AIC activation appears selectively in pain and negative emotion conditions, and revealing a functional dissociation whereby AIC — but not ACC — showed increased activation for painful relative to neutral stimuli in carefully controlled paradigms [52]. This selectivity implies that the AIC is not a generic salience detector but is preferentially engaged by energy-demanding, negatively valenced states.

A neuroanatomical modeling framework extending this work proposed that the AIC serially represents emotional feelings across successive moments, generating what has been termed a “specious present” and serving as a neural substrate for time perception in the seconds-to-subseconds range [53]. This temporal integration function links interoceptive awareness to the continuous threading of affective experience through time — a dimension largely absent from earlier accounts focused on discrete emotion episodes. Most recently, a neurobiological model tracing interoceptive information flow from peripheral receptors to cortex has emphasized the distinctive architecture of interoceptive pathways: sparse receptor distribution, thin unmyelinated fibers, and early compression in the brainstem and midbrain before reaching cortex, yielding signals that are inherently rhythmic and predictable, and therefore subject to repetition suppression in conscious awareness [7]. This suppression mechanism offers a principled explanation for why interoceptive signals so often remain below the threshold of explicit attention unless specifically directed there. Consistent with this, research on interoceptive attention shows that directed focus on bodily signals is associated with greater habitual use of emotion regulation strategies, mediated by enhanced emotional awareness, and — crucially — with greater situational flexibility in strategy selection during high-intensity negative events [92]. This contextualizes why interoceptive training or attention interventions might be necessary to unlock regulatory benefits, a conclusion further supported by neuroimaging evidence that MBSR-based interoceptive training produces practice-specific changes in insula–sensorimotor and ACC connectivity that correspond with reduced difficulties in emotion regulation [93].

Cross-Cultural Variation in Interoceptive-Emotional Construction and Neural Architecture

The theoretical claim that emotional granularity depends on culturally acquired emotion concepts and culturally shaped patterns of bodily attention generates a strong prediction: the neural and psychological pathways through which interoceptive signals become emotional experiences should vary across cultural groups. Emerging evidence from cultural neuroscience provides direct support for this prediction. In a landmark fMRI study, [22] simultaneously recorded brain activity and electrocardiographic data while Chinese, American, and Asian American participants viewed emotionally evocative narrative stimuli. The results revealed a compelling double dissociation: correlations between emotional feeling strength and cardiac arousal were consistent across all three cultural groups, indicating that the basic interoceptive channel linking physiological state to subjective experience is pan-cultural in character. Yet the specific subregion of the anterior insula recruited during conscious emotional feeling diverged sharply along cultural lines: ventral anterior insula (vAI) activity correlated with feeling strength across all groups, but dorsal anterior insula (dAI) activity — associated with more cognitively elaborated interoceptive processing that integrates bodily signals with higher-order social and conceptual content — correlated with feeling strength only in the American and Asian American participants, not in the Chinese group [22].

This finding carries substantial implications for understanding how emotional granularity is realised in the brain. Because cardiac arousal was statistically equivalent across groups, the cultural difference cannot be attributed to peripheral physiological variation; it must reflect divergence at the level of central interoceptive construction. Chinese participants’ differential reliance on the ventral rather than dorsal anterior insula suggests a different attentional relationship to bodily signals during emotional experience — one potentially oriented toward holistic, diffuse somatic awareness rather than the focal, analytic interoceptive attention that characterises Western cognitive styles and that heartbeat detection tasks reward. This neural dissociation helps resolve a long-standing paradox in cross-cultural interoception research: non-Western, particularly East Asian, populations frequently report heightened somatic awareness and a greater tendency toward somatic expression of distress, yet often perform less accurately on laboratory measures of interoceptive accuracy such as heartbeat detection tasks [39]. Linguistic evidence illuminates the cultural depth of this somatic orientation — Chinese emotional expressions systematically metaphorise visceral organs as the locus of emotional experience, while Japanese maps emotional progression directly onto bodily locations (anger originating in the belly, moving to the chest, and finally reaching the head), suggesting that non-Western somatic salience is scaffolded by culturally transmitted conceptual and linguistic schemas rather than by superior bottom-up signal detection [39]. Ma-Kellams’ [39] review identifies the likely mechanism: East Asian cultures cultivate context-dependent, holistic thinking patterns that habitually privilege environmental and relational cues over isolated internal states, and this attentional style — while conferring other cognitive advantages — creates a systematic liability for the kind of focal signal detection that heartbeat counting tasks demand. Crucially, however, this account is complicated by evidence from Ubukata et al. [40], who found that Japanese participants actually outperformed European participants on heartbeat detection under methodologically controlled conditions using adapted task instructions designed to prevent reliance on time estimation and prior heart-rate knowledge — suggesting that the relationship between cultural cognitive style and interoceptive accuracy is more nuanced than a simple East-low, West-high gradient, and may depend on which specific aspects of interoceptive processing a given task recruits. If different cultural groups recruit distinct subregions of the anterior insula during emotional feeling, then what appears as lower interoceptive “accuracy” in some paradigms may instead reflect a different mode of interoceptive-emotional construction — one that does not map cleanly onto the signal-detection paradigms developed within Western psychophysiology [22, 39].

These neural findings are reinforced by large-scale behavioural evidence demonstrating that cultural context shapes the expressivity and intensity with which emotional experience is communicated. An analysis of 129,222 consumer reviews across six datasets found that Western customers demonstrate consistently higher positive sentiment, greater emotional intensity, and more pronounced dominance in emotional communication compared to Eastern customers, with gender-based emotional disparities more pronounced within Western than Eastern cultural contexts [94]. From an interoceptive-emotional construction standpoint, this moderation is significant because emotional expressivity and interoceptive awareness are functionally linked: cultures and gender norms that encourage emotional disclosure may simultaneously facilitate greater attentional engagement with internal bodily states, while those that value emotional restraint may shape a different pattern of interoceptive-to-emotional translation [94]. The convergence between the neural differentiation observed across cultural groups in imaging paradigms [22] and the behavioural differentiation in expressive emotional patterns [94] suggests that cultural modulation operates coherently across levels of analysis, from neural architecture to observable communicative behaviour.

The stability versus malleability of the conceptual structures through which interoceptive signals are categorised has been further illuminated by research examining affective knowledge under conditions of sustained collective stress. A longitudinal study of Italian participants during the COVID-19 pandemic found that negative affectivity remained significantly elevated relative to pre-pandemic baselines, with particularly pronounced increases in nervousness, irritability, and fear [95]. Yet the foundational dimensional architecture of affective knowledge — organised around valence and arousal dimensions — remained structurally intact, indicating that while the intensity and patterning of emotional experience shifted substantially under pandemic conditions, the deep conceptual scaffolding of the emotion system was not fundamentally disrupted [95]. Females showed stronger interoceptive awareness during the pandemic period, particularly in emotional awareness and attention regulation, alongside increased intercorrelations among interoceptive subscales — suggesting that chronic stress may reorganise the internal structure of interoceptive self-monitoring in a gender-differentiated manner [95]. This dissociation between the stability of dimensional affective architecture and the malleability of interoceptive sensitivity parallels the neural dissociation observed across cultural groups: in both cases, certain structural features of emotional experience appear robust, while the processing emphasis placed on interoceptive signals proves context-sensitive and demographically variable.

Together, these cross-cultural and contextual findings reinforce TCE’s central claim that the path from interoceptive signal to emotional experience is mediated by conceptual knowledge — but they add the critical qualification that this conceptual mediation is itself culturally constituted. The emotion concepts through which interoceptive predictions are categorised, the neural pathways through which this categorisation occurs, and the social norms governing which bodily signals are attended to and how they are expressed are all shaped by the cultural ecology in which development unfolds. Any account of emotional granularity that treats it solely as an individual cognitive capacity — without attending to the cultural, linguistic, and social contexts that furnish the conceptual repertoire upon which granularity depends — will therefore remain incomplete.

Gender, Language Socialization, and Differential Emotional Labeling

A particularly generative tension in this literature concerns the apparent paradox of gender differences in interoceptive accuracy and emotional ability. Women typically demonstrate poorer performance on heartbeat detection tasks than men, yet consistently outperform men on emotional processing and recognition tasks — a contradiction that resists resolution within simple somatic feedback accounts [81]. A theoretical review synthesizing research on gender differences in interoception and emotional ability proposes that differential language socialization shapes how individuals label internal physiological versus emotional states, such that women’s richer emotional vocabularies may compensate for, or operate independently of, lower interoceptive accuracy [81]. Empirical support for this dissociation between interoceptive accuracy and emotional verbalization comes from research demonstrating that individuals with higher interoceptive accuracy use significantly more emotion-related words and produce richer, more differentiated co-occurrence networks when describing both their own and others’ emotional states, while those with lower interoceptive accuracy tend to rely on top-down, context-driven labeling processes rather than bottom-up bodily signals [25]. This suggests that emotional verbalization is doubly determined: by the precision of interoceptive input and by the linguistic and conceptual resources brought to bear on it. This argument is directly congruent with both TCE’s central claim that the path from interoceptive signal to emotional experience is mediated by conceptual knowledge [6] and CAT’s demonstration that emotion words are constitutive of emotional experience rather than merely descriptive of it [42]: women’s richer emotional vocabularies may reflect a larger and more differentiated repertoire of emotion concepts, which in turn supports finer-grained categorization of interoceptive states independently of the precision of those states at the signal-detection level. The gender paradox thus dissolves once one accepts that emotional granularity depends not only on the quality of interoceptive input but also — and perhaps equally — on the conceptual and linguistic resources available to categorize it [81, 6, 42]. The cross-cultural behavioural evidence further contextualises this gender effect: gender-based emotional disparities are more pronounced within Western than Eastern cultural contexts [94], a pattern that is echoed in cross-cultural research documenting broader variation in somatic awareness and interoceptive processing styles across individualist and collectivist societies [39], suggesting that the magnitude of the gender paradox is itself culturally moderated — women’s advantage in emotional differentiation may be amplified in cultural environments that socialise females toward greater emotional disclosure and richer emotional vocabularies while simultaneously measuring interoceptive accuracy through paradigms calibrated to focal, analytic signal detection.

Phenomenological research has deepened this picture by revealing the lived texture of what emotional differentiation actually involves. A hermeneutical phenomenological study of Irish men found that emotional differentiation unfolds in two analytically distinct phases: a pre-articulation phase involving inward attention to bodily sensations and contextualisation against prior experiences, followed by an articulation phase characterised by recursive, dialectical translation into language [90]. Participants consistently reported that language is inadequate for capturing the full texture of emotional experience, yet simultaneously necessary for imposing order on internal chaos, with metaphor emerging as a particularly important vehicle during moments of heightened emotional intensity [90]. Crucially, the cultural context of Irish masculinity constrained not merely the expression of emotion but participants’ very willingness to engage in the inward bodily attention that pre-articulation requires — a finding that challenges models treating emotional granularity as primarily an interoceptive or cognitive skill, insisting instead that it is fundamentally socially embedded, shaped by norms governing which internal states are worth attending to and which are best left undifferentiated [90]. This phenomenological evidence converges with the gender-socialization account of [81], the cross-cultural neural evidence of [22], and the attachment-mediated findings of [32] to underscore that the path from body to emotion is culturally and relationally constrained at every step.

Emotional Granularity: From Undifferentiated Arousal to Fine-Grained Experience

The question of how bodily signals become emotionally differentiated — emotional granularity — thus emerges as a central concern integrating multiple strands of this literature. The Theory of Constructed Emotion provides the formal theoretical architecture for this integration: granularity is not a peripheral by-product of emotional life but a reflection of how precisely the brain categorizes interoceptive predictions using its available repertoire of emotion concepts [6]. High granularity — the ability to experience emotions with precision and context-specificity — is consistently associated with more effective coping, better mental health outcomes, and greater regulatory flexibility [26]. Empirical support for this link is also provided by evidence that interoceptive sensibility and emotional conceptualization jointly predict the quality and differentiation of emotional experience, such that individuals who both attend to and have richer conceptual frameworks for bodily states report more finely grained affective episodes [12]. Low granularity, conversely, characterizes conditions in which individuals rely on coarse, non-specific affective categories — a pattern with wide clinical consequences. Alexithymia, for instance, has been reconceptualized as a general deficit of interoception in which the inability to differentiate bodily signals cascades into impaired emotional differentiation [96], while reduced levels of emotional awareness have been documented across mood, anxiety, and personality disorders [35]. In functional neurological disorder, individuals appear to categorize high-arousal interoceptive states through bodily or health-related concepts rather than emotional ones, producing somatic symptoms where more granular emotion construction would generate psychologically tractable affective experiences [64]. This clinical observation illustrates TCE’s core claim that the same interoceptive signals can give rise to radically different experiential outcomes depending on the conceptual categories through which they are processed.

Importantly, recent evidence demonstrates that emotional granularity is not a fixed trait but a dynamic, practice-responsive capacity. A 14-day intensive ambulatory assessment study found that both positive and negative emotional granularity increased significantly over the study period, with effect sizes of d = 0.50 for positive granularity and d = 0.32 for negative granularity, suggesting that the repeated act of attending to and labelling emotional experience in real time scaffolds finer emotional differentiation [29]. However, a more nuanced dose-response pattern emerged: greater engagement with experience sampling prompts showed a 98% posterior probability of positive association with increased positive granularity, yet a greater than 99% probability of negative association with increased negative granularity [29]. This asymmetry implies that sustained attention to negative emotional experience may function differently — perhaps introducing fatigue, avoidance, or recursive rumination that disrupts rather than refines differentiation. Resting respiratory sinus arrhythmia further moderated individual change trajectories, anchoring emotional granularity within a psychophysiological framework and reinforcing the systemic nature of affective differentiation [29]. These findings carry a direct methodological implication: intensive ambulatory assessment is not a passive observational window but an active perturbation of the emotional system, reshaping the very representational structures it aims to measure — a consideration with both scientific and clinical significance.

The empirical findings reviewed across this section converge with the constructionist account. The AIC’s role in temporal integration [53], the suppression of predictable interoceptive signals [7], the culturally specific patterns of anterior insular recruitment during emotional feeling [22], and the mediating function of emotional awareness between interoceptive attention and strategy use [92] together suggest that fine-grained emotional experience requires not merely intact peripheral signaling but active cortical elaboration and conceptual categorization — and that the neural pathways through which this elaboration occurs are themselves culturally shaped. Language socialization effects on emotional labeling [81, 42], cultural constraints on inward bodily attention [90, 94], embodied affective dynamics that precede conceptual elaboration [89], and the attachment-mediated shaping of interoceptive trust [32] further underscore that the path from bodily state to named emotion is neither direct nor uniform across individuals, social contexts, or cultural ecologies. Importantly, emotional granularity appears to be a trainable capacity rather than a fixed trait: both intensive self-monitoring [29] and contemplative practices rooted in Western mindfulness traditions and Buddhist philosophical frameworks have been proposed as vehicles for cultivating granularity, on the grounds that such practices refine the emotion concepts through which interoceptive experience is categorized, promoting acceptance, decentering, and the dereification of habitual affective patterns [26]. Consistent with this, three months of contemplative training has been shown to produce differential changes in self-reported interoceptive awareness, including increased noticing and non-reactive attention to bodily states [85] — precisely the attentional prerequisites for finer emotional differentiation. This trainability claim connects directly to the intervention evidence reviewed in Section 7 and reinforces the constructionist insight that emotional expertise is, at its core, a matter of conceptual precision applied to the body’s signals.

Taken together, the evidence reviewed here charts a cumulative and increasingly integrated account of how reading the body’s signals enables, constrains, and is culturally shaped in its contribution to emotional life. The Theory of Constructed Emotion provides the unifying theoretical mechanism: interoceptive predictions generate core affect, emotion concepts categorize that affect into discrete experiences, and the precision of this categorization process — emotional granularity — determines the richness and adaptiveness of emotional functioning. The developmental evidence reviewed here adds a critical temporal dimension: this categorization process is not established de novo in adulthood but is scaffolded through early caregiving relationships that shape both the precision of interoceptive signals and the richness of emotion concepts from the earliest years of life. The cross-cultural evidence adds an equally critical contextual dimension: the conceptual repertoire upon which granularity depends is furnished by language, culture, and socialisation, such that emotional granularity is not a purely individual cognitive achievement but a culturally constituted capacity whose expression varies systematically across populations and contexts.

7. Contemplative Practices, Body-Based Interventions, and Training Interoceptive Awareness

The preceding section concluded that emotional granularity is not a fixed trait but a trainable capacity — one that increases with sustained self-monitoring and that contemplative practices have been proposed to cultivate by refining the emotion concepts through which interoceptive experience is categorized [29, 26]. That conclusion raises an immediate question: what are the mechanisms through which such training operates, and what does the clinical evidence show about its efficacy? Section 7 takes up both questions in turn. It traces the evidence from early neuroimaging demonstrations of how mindfulness reshapes the neural architecture of self-referential processing, through theoretical frameworks that reconceive contemplative practice as a fundamental shift in the brain’s inferential stance toward internal bodily signals, to structured clinical translations that operationalize these insights for populations with alexithymia, trauma, and regulatory deficits. Together, these threads build the mechanistic and empirical foundation beneath the trainability claim advanced at the close of Section 6.

Contemplative practices occupy a distinctive position in the broader interoception literature: they represent not merely passive observations of bodily experience but deliberately engineered interventions that reshape the neural and psychological architecture of self-referential processing. Early foundational work established that structured training programs could produce measurable changes in how the brain processes internal signals, while subsequent research translated these mechanistic insights into clinical frameworks targeting alexithymia, trauma, and emotional dysregulation. The most recent evidence, including a 2025 clinical trial, has begun to quantify these gains with greater empirical precision, confirming and extending the theoretical propositions advanced nearly two decades earlier.

Mindfulness Training and the Neural Modes of Self-Reference

A landmark contribution to understanding how contemplative practice reshapes interoceptive processing came from neuroimaging research demonstrating that mindfulness training produces qualitatively distinct neural states of self-reference [55]. Comparing novices with participants who had completed an eight-week Mindfulness-Based Stress Reduction (MBSR) programme — combining meditation, yoga, and stress management education [97] — this work identified two separable neural modes: a narrative mode anchored in the medial prefrontal cortex (mPFC), associated with self-referential elaboration and story-telling about one’s experience, and an experiential mode marked by reduced mPFC activity and heightened engagement of right-lateralized networks including the lateral prefrontal cortex, insula, and somatosensory cortices [55]. Critically, mindfulness-trained participants showed markedly stronger reductions in mPFC activity during experiential focus than novices, suggesting that sustained practice actively decouples conceptually mediated self-reference from present-moment bodily awareness. A subsequent neuroimaging study corroborated this account, finding that MBSR training specifically enhanced dorsal anterior insula responsiveness during interoceptive attention while strengthening negative connectivity between the dorsomedial prefrontal cortex and posterior insula [97]. The finding that the insula — widely regarded as the primary cortical substrate for interoceptive representation, integrating ascending visceral signals with top-down contextual predictions to generate conscious bodily feeling [52] — was preferentially engaged during the experiential mode provided an early neural bridge between contemplative training and interoceptive sensitivity.

From Neural Mechanisms to Theoretical Synthesis: Active Versus Perceptual Inference

Building on this neural evidence, subsequent theoretical work proposed a more integrative account of how contemplative practices alter the fundamental computational stance toward internal bodily signals [27]. Rather than framing mindfulness simply as enhanced attention, this synthesis distinguished among multiple interoceptive constructs — including coherence, attentional tendency, sensitivity, accuracy, sensibility, and regulatory capacity — and argued that these dimensions are selectively modifiable through different forms of practice [27]. The central theoretical contribution was the distinction between active inference, in which the organism generates actions to bring bodily states into line with prior predictions, and perceptual inference, in which predictions are updated to accommodate incoming sensory evidence [9, 59]. Contemplative practices, the argument runs, cultivate a disposition toward perceptual inference: the practitioner learns to attend to sensations as they are rather than immediately acting to resolve them, thereby progressively recalibrating the generative model that shapes ongoing interoceptive experience [27]. This account integrates neatly with the predictive coding framework developed in adjacent theorizing, which holds that conscious presence itself arises from successful suppression of interoceptive prediction errors through top-down predictions originating in the anterior insular cortex [58]. Critically, this suppression operates through the selective modulation of precision weighting — the relative gain assigned to ascending sensory signals versus descending predictions — such that mindfulness practice may function by recalibrating the precision applied to interoceptive streams rather than merely redirecting attentional resources [72, 11]. Seen through this lens, mindfulness practice is not merely attentional training but a recalibration of the precision-weighting applied to ascending interoceptive signals, gradually loosening the grip of habitual predictive models that may have become maladaptively rigid [10].

From the perspective of the Theory of Constructed Emotion, this recalibration has a further consequence: by increasing the precision and differentiation of interoceptive signals available to consciousness, contemplative practice also enriches the raw material upon which emotion concepts operate, thereby supporting finer-grained emotional categorization [6, 26]. Wilson-Mendenhall and Dunne have argued explicitly that mindfulness practices cultivate emotional granularity through mechanisms that parallel the constructionist account — including acceptance, decentering, and the dereification of habitual affective categories — refining the emotion concepts through which interoceptive experience is interpreted and enabling more precise, context-sensitive emotional construction [26]. Computational modeling work further supports this view, demonstrating that iterative updating of emotion concept representations through active inference can progressively sharpen affective categorization in a manner consistent with granularity gains [98]. This convergence between contemplative science and constructionist emotion theory suggests that the benefits of mindfulness-based interventions extend beyond attentional or regulatory gains to encompass the very structure of emotional experience itself. Notably, empirical evidence from intensive ambulatory assessment confirms that sustained self-monitoring of emotional states — a practice closely related to contemplative self-observation — does indeed increase emotional granularity over time, with positive granularity showing particular responsiveness to repeated engagement [29]. The asymmetric effects observed for positive versus negative granularity in that work, however, underscore that the relationship between attention and differentiation is not uniformly beneficial: attending closely to negative emotional experience may introduce recursive rumination rather than refined differentiation [99, 29], a finding with direct implications for how contemplative protocols are designed and implemented.

Clinical Translation: Mindful Awareness in Body-Oriented Therapy

The theoretical advances described above created a principled rationale for bottom-up clinical interventions that engage the body as a primary therapeutic site. Mindful Awareness in Body-Oriented Therapy (MABT) represents perhaps the most systematically articulated such approach, organizing intervention into three progressive stages — awareness, access, and appraisal — that scaffold the client’s capacity to detect, sustain attention to, and ultimately derive meaning from internal bodily sensations [84]. The foundational claim of MABT is that interoceptive awareness functions as a critical bridge between raw somatic sensation and cognitive-emotional processing, and that many clinical populations — particularly those with histories of trauma or substance use — present with disrupted access to this bridge [84]. Empirical evidence supports this disruption: research across patients with PTSD, major depression, and somatic symptom disorder consistently identifies elevated body dissociation — a habitual disregarding of internal signals — as a significant mediator between traumatic childhood experiences and emotion dysregulation, with body dissociation accounting for the relationship even when other interoceptive measures show no significant group differences [100]. By building interoceptive skill incrementally, MABT aims to restore the somatosensory-emotional link that underpins effective emotion regulation. Importantly, this framework positions interoceptive training not as an adjunct to conventional psychotherapy but as a substrate-level intervention that addresses the bodily foundations of psychological functioning [84].

A recurring and clinically significant concern running through this literature is the recognition that heightened interoceptive awareness is not unconditionally beneficial. For individuals with trauma histories, turning attention toward the body may amplify distress rather than resolve it, particularly in the absence of adequate regulatory scaffolding [84, 27]. This caveat is theoretically coherent within the predictive coding framework: if interoceptive prediction errors are chronically elevated — as may be the case in post-traumatic states — then increasing the precision-weighting of ascending signals without simultaneously strengthening top-down regulatory capacity could exacerbate rather than attenuate dysregulation [58]. Active inference accounts of interoceptive psychopathology formalise this risk further, identifying hyperprecise priors — excessively strong bodily expectations that resist updating despite contradictory sensory evidence — and context rigidity as the two core dysfunctions that sustain chronic somatic error signals across anxiety, depression, and trauma-related conditions [17]. On this view, body-based interventions must target not only attentional deployment but the miscalibrated generative models that determine how incoming signals are weighted and interpreted. The staged architecture of MABT directly addresses this risk by ensuring that regulatory capacity is built alongside, rather than after, awareness expansion [84]. Developmental evidence reinforces this concern from a different angle: attachment research shows that caregivers who reject children’s emotional expressions can produce a functional disconnection between physiological arousal and subjective emotional experience [32], suggesting that for some adults, the interoceptive suppression that therapeutic interventions seek to reverse may have originated as an adaptive response to an invalidating relational environment. Restoring interoceptive sensitivity in the absence of a sufficiently safe relational context — whether in therapy or in the individual’s broader social world — may therefore recapitulate rather than resolve the original dilemma.

The network modelling evidence from cross-cultural MAIA validation work introduces a further consideration relevant to intervention sequencing. If attitudinal openness — specifically the willingness to attend to bodily signals without distraction or worry — functions as an upstream prerequisite for the development of more sophisticated interoceptive skills rather than as their downstream product [76], then body-based interventions may achieve their effects not primarily by sharpening bodily attention but by first modifying the evaluative and avoidant attitudes that ordinarily obstruct interoceptive processing [19]. This directional logic has substantive implications for how contemplative protocols are structured: interventions that begin with attitudinal cultivation — non-judging, non-reactivity, acceptance — before introducing focused interoceptive attention may be more effective than those that reverse this sequence, a hypothesis that remains empirically untested but theoretically well-motivated [27, 101].

Empirical Gains and Recent Clinical Evidence

The most recent empirical contribution to this theme provides direct clinical corroboration of the theoretical propositions outlined above. A 2025 pilot clinical trial examined the effects of a structured body-based mindfulness programme on alexithymia, dispositional mindfulness, and distress symptoms in 73 non-clinical adults [102]. The intervention produced significant reductions in overall alexithymia scores, with the most pronounced gains on the difficulty identifying and describing feelings subscales — the components most directly linked to disrupted interoceptive access [102]. Improvements extended to four of the five facets of dispositional mindfulness assessed: observing, describing, acting with awareness, and non-reacting; the non-judging facet alone showed no significant change, a finding the authors attribute to the programme’s emphasis on sensorimotor engagement over evaluative reappraisal [102]. These results not only confirm the clinical efficacy of body-oriented mindfulness training but also provide granular evidence that different facets of mindfulness are differentially amenable to body-based approaches — a finding consistent with Bornemann et al.’s [85] controlled longitudinal study, which likewise demonstrated that three months of body-scan and breath-meditation training produced robust gains in regulatory and integrative dimensions of interoceptive awareness (self-regulation, attention regulation, body listening; d = 0.40–0.72) while leaving basic noticing sensitivity largely unchanged. Such differential profiles refine the undifferentiated claims that characterised earlier enthusiasm for MBSR as a general-purpose intervention.

Taken together, this body of research traces a coherent trajectory from early neuroimaging demonstrations that mindfulness decouples narrative self-reference from present-moment bodily processing [55], through theoretical frameworks reconceiving contemplative practice as a shift in inferential strategy toward interoceptive signals [27, 58], to structured clinical translations that operationalize these insights for populations with alexithymia and regulatory deficits [84, 102]. Neuroimaging evidence from a randomised controlled pilot of MABT — which documented reduced insula deactivation during interoceptive attention tasks and enhanced connectivity between the right insula and the dorsal attention network, with connectivity gains correlating positively with self-reported interoceptive awareness — offers a preliminary mechanistic bridge between behavioural and neural levels of analysis [103]. Remaining limitations include the small sample sizes and predominantly non-clinical composition of the 2025 trial [102], the largely theoretical rather than experimental status of the active-versus-perceptual inference distinction [27], and the absence of longitudinal neuroimaging data linking MABT-style interventions to the specific default-mode decoupling documented in foundational mindfulness research [55, 103]. Establishing this mechanistic chain across levels of analysis — from predictive coding architecture to clinical outcome — remains the field’s most pressing integrative challenge.

8. Clinical Disruptions of Interoception: Psychopathology, Somatic Conditions, and Psychiatric Applications

Early theoretical work drew a clear line between interoceptive dysfunction and psychopathology, establishing a foundation that subsequent computational and clinical research would both deepen and complicate. The seminal synthesis by [104] proposed that anxiety and depression are not merely disorders of mood or cognition but of amplified, noisy interoceptive self-referential beliefs, wherein the anterior insula integrates distorted bodily signals with self-relevant processing through dense connections to the medial prefrontal cortex. This neurobiological framing recast common psychiatric conditions as disorders of body-brain inference rather than purely cortical phenomena—a conceptual shift that catalyzed a decade of increasingly sophisticated theoretical development.

Predictive Coding Frameworks and the Architecture of Interoceptive Failure

Building on the interoceptive foundations laid by [104], formal predictive coding accounts extended the explanatory architecture considerably. [58] proposed that conscious presence itself arises from the successful suppression of interoceptive prediction errors within the anterior insular cortex (AIC), and that depersonalization/derealization disorder (DPD) reflects not simple prediction mismatch but pathologically imprecise interoceptive predictions. Where earlier models emphasized amplified error signals, this framework highlighted that hypoactivation of the AIC in DPD results from a failure to generate sufficiently precise top-down predictions, leaving prediction errors unsuppressed and the sense of embodied selfhood attenuated [10]. The implication is that DPD and anxiety may represent opposing poles of the same inferential dysregulation—hyperprecision in the latter, hypoprecision in the former—a polarity with direct empirical support from hierarchical Bayesian modelling of cardiac signal weighting in anxious populations [60].

This polarity became theoretically productive. [9] elaborated the active inference account of emotion and selfhood, arguing that bodily states are actively regulated through descending predictions from deep generative models rather than passively monitored through ascending afferent signals. By the late 2010s, [17] synthesised these threads into a clinically actionable taxonomy, identifying two core interoceptive dysfunctions: hyperprecise priors, in which excessively strong expectations resist updating in light of new bodily evidence, and context rigidity, an inability to flexibly revise expectations as environmental demands shift. Crucially, this framework is explicitly transdiagnostic, accounting for the overlapping interoceptive signatures observed across panic disorder, PTSD, somatic symptom disorders, and substance use.

The transdiagnostic character of this framework has been independently corroborated by a comprehensive review systematically challenging the parcellation of interoceptive dysfunction into condition-specific profiles [105]. Marshalling evidence across autism spectrum disorder, eating disorders, PTSD, borderline personality disorder, and depression, [105] demonstrates that atypical interoception—whether manifesting as heightened sensitivity, reduced accuracy, or impaired integration of bodily signals—is a common antecedent and maintaining factor across a wide range of psychiatric presentations. This convergence between the computational taxonomy of [17] and the empirical review of [105] substantially strengthens the case for treating interoceptive dysfunction as a core transdiagnostic process.

The Theory of Constructed Emotion extends this transdiagnostic logic by specifying a further mechanism through which interoceptive dysfunction produces psychopathology: the failure of emotion construction itself. When predictive models for allostasis break down, interoceptive prediction errors may be categorised through impoverished or maladaptive conceptual repertoires, yielding psychiatric symptoms rather than psychologically tractable emotional experiences [106, 6]. Depression has been reconceptualised within this framework not as heightened emotional reactivity but as a breakdown in the brain’s capacity to accurately predict and regulate energy needs—a state of chronic dyshomeostasis producing persistent metabolic inefficiency experienced as fatigue, anhedonia, and withdrawal [106, 107]. Stephan et al. [107] formalise this as allostatic self-efficacy failure, wherein impaired confidence in the brain’s ability to restore homeostasis compounds the subjective burden of energy dysregulation. In functional neurological disorder, interoceptive prediction errors during high-arousal states appear to be categorised through somatic and health-related concepts rather than emotional ones, producing neurological symptoms where more granular emotion construction would generate manageable affective experiences [64].

Somatization as Culturally Coherent Interoceptive-Emotional Construction

The constructionist account of symptom formation has typically been framed in clinical literature as reflecting a deficit: a failure of emotional granularity that produces physical symptoms where more differentiated emotion construction would generate psychologically tractable experiences. Cross-cultural evidence, however, demands a more nuanced appraisal. The neural finding that Chinese participants recruit the ventral anterior insula preferentially during emotional feeling, while American participants additionally engage the dorsal anterior insula—associated with more cognitively elaborated interoceptive processing [22]—suggests that cultures differ not only in which emotion concepts are available for categorising interoceptive states but in the neural pathways through which such categorisation occurs. This neuroimaging evidence converges with linguistic and anthropological data: Chinese emotional vocabulary consistently metaphorises visceral organs as the locus of experience, and non-Western populations more broadly demonstrate heightened somatic awareness alongside distinct interoceptive attentional styles [39]. What clinicians in Western settings classify as somatization may therefore represent the normal operation of a culturally calibrated interoceptive-emotional system rather than its failure: the same interoceptive prediction errors will be categorised through whatever conceptual repertoire is culturally available, and cultures providing rich somatic categories but fewer psychologised emotion concepts will naturally produce more somatic expressions of distress [6, 42, 44].

This cultural reframing does not imply that somatic expressions are clinically inconsequential. Rather, it challenges the implicit hierarchy positioning psychological articulation as the normative endpoint and somatic expression as developmental failure. Notably, Kleinman’s influential study of Chinese neurasthenic patients found that 87% met DSM criteria for depression—suggesting not experiential absence but diagnostic mismatch, where genuine distress is expressed through culturally legitimate somatic idioms rather than the affective vocabulary Western instruments presuppose [44]. The Conceptual Act Theory’s demonstration that emotion concepts are constitutive of emotional experience [42] implies that individuals socialised into contexts where distress is principally articulated through bodily idioms may genuinely experience their suffering differently—not merely report it differently. Diagnostic frameworks that treat somatization as a deficit to be corrected through emotional literacy training, without attending to the cultural legitimacy of somatic expression, risk imposing culturally specific norms of emotional articulation as if they were universal standards of psychological health [44, 39]. The cross-cultural neuroscience reviewed here [22, 42] thus introduces a necessary complication into any transdiagnostic account: what counts as miscalibrated interoceptive inference may itself be culturally indexed.

Autonomic Regulation, Vagal Tone, and Emotional Dysregulation

Complementing the Bayesian tradition, a separate but convergent line of inquiry has focused on the autonomic correlates of interoceptive dysfunction. The systematic review by [50] synthesised evidence showing that higher heart rate variability (HRV) consistently predicts more adaptive emotional regulation strategies—particularly cognitive reappraisal over suppression—and that interoceptive accuracy supports emotional downregulation capacity, linking cardiac vagal tone to the quality of emotional inference rather than simply to physiological arousal. Experimental work corroborates this directionality: individuals with greater interoceptive awareness demonstrate enhanced ability to deploy reappraisal under stress [15], and in younger populations, higher resting HRV has been associated with more flexible deployment of regulatory strategies across contexts [36].

The clinical framework provided by polyvagal theory offers an organisational lens for these findings. [48] traces the development of polyvagal theory from early physiological observation to a neuroanatomically grounded account centred on the ventral vagal complex (VVC)—a phylogenetically recent mammalian system with myelinated cardioinhibitory fibres enabling social engagement behaviours. HRV and respiratory sinus arrhythmia, in this account, are meaningful neural signals reflecting brainstem regulatory capacity. States of chronic threat or shutdown, characteristic of trauma-related and anxiety disorders, reflect hierarchical downregulation of VVC function, leading to autonomic inflexibility that mirrors the context rigidity described in computational accounts [17]. This autonomic rigidity is further understood through active interoceptive inference [9], wherein dysregulated autonomic states bias the precision-weighting of interoceptive predictions, amplifying threat signals and constraining adaptive response. The convergence between polyvagal and active inference frameworks [48, 50] suggests that autonomic dysregulation and Bayesian misfiring are mechanistically linked rather than independent pathways to psychopathology.

Developmental Origins of Clinical Interoceptive Disruption

The transdiagnostic frameworks reviewed above have been primarily developed with reference to adult populations. A growing body of developmental evidence now reveals that the roots of these disruptions extend considerably earlier, positioning the caregiving environment and early biological exposures as formative influences on the interoceptive systems that later become clinically relevant.

A striking developmental pattern emerges from meta-analytic evidence on emotional self-awareness in autism. Across 47 studies, autistic adults showed significantly poorer emotional self-awareness relative to neurotypical peers (d = 1.16), yet no significant differences were observed in autistic children aged twelve years and under (d = 0.03) [38]. This divergence across the lifespan implicates rising social demands during adolescence and accumulated comorbid mental health difficulties—rather than a fixed neurological deficit—as the primary drivers of observable group differences [38]. Corroborating this developmental sensitivity, psychometric work validating a youth-adapted measure of interoceptive awareness (MAIA-y) across 429 children and adolescents aged 7–17 found a significant negative linear age trend in body trust, with older youth reporting progressively lower trust in their bodily sensations [20]. Converging evidence from cardiac interoception research further indicates that interoceptive sensitivity plays a meaningful role in emotion regulation even in early development, with heart rate awareness linked to emotional reactivity outcomes in youth [36]. This erosion of bodily trust across adolescence—observed in neurotypical development as well—suggests that puberty-related shifts in interoceptive awareness may constitute a transdiagnostically relevant vulnerability window, not one confined to autism alone [20]. Within the systems-level reconceptualisation of emotion regulation proposed by [91], where regulation is constitutive of emotional activation rather than superimposed upon it, the adolescent emergence of emotional awareness deficits in autism may reflect a mismatch between increasingly complex social-emotional demands and available regulatory resources across multiple interconnected systems.

The caregiving ecology in which interoceptive systems develop has also been theorised as a direct pathway to specific clinical presentations. [34] situates interoceptive vulnerability within an evolutionary framework, proposing that “somaticity”—a tendency to rely on external social referencing rather than internal interoceptive signals—evolved as an adaptive mechanism for modulating affiliative tendencies in response to ancestral pathogenic threats. Environments characterised by parental psychological control—intrusive, manipulative caregiving that invalidates children’s emotional expression—reproduce conditions that imprint somaticity by signalling that internal signals are unreliable or socially dangerous guides to behaviour [34]. Empirical support for the developmental sensitivity of interoceptive trust to the caregiving context is provided by Oldroyd et al. [32], who document that attachment-related processes in childhood are directly associated with interoceptive awareness, and by evidence that traumatic childhood experiences are linked to impaired embodied self-awareness via reduced body connection [108]. Extending this line of evidence, Schmitz et al. [100] demonstrate more specifically that traumatic childhood experiences are associated with a tendency to disregard one’s own bodily signals, framing interoceptive avoidance as a direct sequela of adverse early environments rather than a constitutional trait. The suppression of interoceptive trust is thus reframed not as a deficit but as a developmentally programmed response to a specific caregiving ecology. This evolutionary perspective resonates with the attachment-related findings reviewed in Section 6 [32], offering a deeper explanatory layer for why maternal rejection of emotion and parental psychopathology produce the interoceptive profiles they do.

At the most distal developmental window, emerging evidence implicates the gut-brain axis and prenatal biological exposures as upstream determinants of later interoceptive and emotional difficulties. A large prospective longitudinal analysis of the ABIS cohort found that gut microbiome composition and metabolic signatures within the first year of life were meaningful predictors of later neurodevelopmental conditions [109]. Early infections requiring penicillin were associated with substantially elevated odds of speech disorder (OR = 3.89), ADHD (OR = 3.27), and intellectual disability (OR = 2.44), while maternal smoking during pregnancy carried odds ratios reaching 3.72 for ASD and 3.31 for ADHD, with effects increasing dose-dependently [109]. These findings implicate gut-brain axis dysregulation, immune-mediated signalling, and prenatal metabolic exposures as potential early-life determinants of the neurodevelopmental foundations upon which later interoceptive and emotional capacities are built. The vagus nerve—a primary anatomical substrate of interoceptive signalling—begins developing prenatally, and its functional maturation is sensitive to precisely the perinatal conditions identified by the ABIS findings [49]. Polyvagal perspectives further emphasise that vagal tone, established through early autonomic development, shapes the capacity for social engagement and self-regulation across the lifespan [48]. The pathway from gut dysbiosis through neurodevelopmental vulnerability to the interoceptive and emotional difficulties documented by [105] and [38] represents a plausible and testable developmental cascade.

Stress, Interoceptive Reorganisation, and Clinical Vulnerability

The relationship between sustained environmental stress and interoceptive-emotional functioning has been illuminated by naturalistic research conducted during the COVID-19 pandemic. A longitudinal study tracking affective knowledge and interoceptive awareness in Italian participants found that negative affectivity remained significantly elevated relative to pre-pandemic baselines approximately one year after the initial outbreak, with particularly pronounced increases in nervousness, irritability, and fear [95]. Critically, females demonstrated stronger interoceptive awareness during the pandemic period—particularly across emotional awareness and attention regulation dimensions—alongside increased intercorrelation among interoceptive subscales, suggesting that chronic collective stress may consolidate or intensify the coactivation of interoceptive processes in a gender-differentiated manner [95]. These findings underscore that interoceptive awareness is not a fixed trait but a dynamic system responsive to environmental context [4].

Within the precision-weighting framework [17], the stress-induced intensification of interoceptive coactivation may represent an adaptive sharpening of bodily monitoring under threat that, if chronically maintained, risks tipping into the hyperprecise prior configuration associated with anxiety and somatic symptom disorders. Specifically, hyperprecise priors—excessively strong expectations about bodily states that resist updating despite contradictory sensory evidence—are proposed as a core mechanism sustaining both anxiety and somatic symptom presentations [17]. A related dysfunction, context rigidity, describes an impaired ability to flexibly recalibrate interoceptive expectations when circumstances change, generating persistent prediction errors and chronic aversive feeling states that perpetuate psychopathological cycles [17, 9].

Sleep Disturbance, Ageing, and the Modulating Role of Anxiety

An underexplored frontier concerns the relationship between interoceptive sensibility and sleep. A recent cross-sectional investigation [78] examined 70 healthy adults aged 18–79 and found that older age and greater insula volume predicted higher interoceptive sensibility—the subjective sense of bodily awareness—but not interoceptive accuracy, the objective detection of internal signals [18]. This dissociation implies that ageing inflates confidence in bodily monitoring without a corresponding improvement in signal detection [79], a pattern consistent with the development of hyperprecise priors [17, 10, 28]—whereby the brain assigns excessive weight to top-down predictions relative to incoming interoceptive signals. More strikingly, anxiety significantly moderated the relationship between interoceptive sensibility and sleep efficiency: heightened bodily awareness was associated with poorer sleep specifically among individuals with moderate-to-high anxiety [78], suggesting that sleep disruption may be a downstream consequence of anxiety-amplified interoceptive inference [104] rather than an independent symptom domain.

Mental Imagery, Aphantasia, and Interoceptive Precision in Neurodevelopmental Contexts

The transdiagnostic precision-weighting logic has recently been extended into the domain of mental imagery. [110] proposed that individual differences in imagery vividness—including the extreme cases of aphantasia and hyperfantasia—may reflect variation in interoceptive awareness as a fundamental generative mechanism, with this relationship further mediated by the general sense of agency, implicating bodily self-awareness and perceived control over mental processes in voluntary image generation [63]. A predictive coding account elaborates this mechanism: the insula regulates the precision of interoceptive signals and modulates top-down predictions from prefrontal cortex, such that reduced interoceptive precision prevents imagery from reaching conscious awareness by failing to activate parietal and visual cortices [61]. Consistent with this, empirical work across 468 participants found that aphantasics and hypophantasics score lower on interoceptive measures, with emotional awareness and body listening showing the strongest associations with imagery vividness, and mediation analyses confirming that mental imagery mediates the relationship between interoceptive processing and autobiographical memory richness [62]. These findings suggest that the same precision-weighting mechanisms that produce anxiety or depersonalisation when severely miscalibrated [17] may, at less extreme degrees of deviation, account for wide population variation in imagery ability—extending the transdiagnostic framework [110, 63] into territory not traditionally classified as psychopathological.

Synthesis and Outstanding Challenges

Taken together, this body of work describes a coherent transdiagnostic architecture in which miscalibrated interoceptive inference—whether hyperprecise, hypoprecise, context-rigid, or autonomically dysregulated—produces a wide spectrum of clinical presentations [17, 9, 58]. The convergence of active inference accounts [9, 17], autonomic frameworks [48, 50], constructionist emotion theory [6, 106], developmental evidence [38, 34, 109, 105], and cross-cultural neuroscience [22, 42] around shared constructs of precision, agency, and hierarchical prediction represents significant theoretical maturation from the early synthesis of [104]. The cultural evidence adds a further complication: what counts as miscalibrated inference may itself be culturally indexed. Substantial gaps nonetheless remain. The causal direction between interoceptive dysfunction and psychiatric onset is rarely established; the accuracy-sensibility dissociation requires longitudinal tracking [78]; the developmental mechanisms driving adolescent emergence of emotional self-awareness deficits in autism remain uncharacterised [38]; and the translation of computational models into clinical protocols remains nascent [28, 1]. Whether interoceptive retraining [84, 103], vagal stimulation [50, 60], or relational and agency-focused therapies can recalibrate precision weighting in clinically meaningful ways—and at what developmental stage such recalibration remains possible—constitutes the most pressing empirical frontier.

9. Discussion

The field of interoception research has undergone a quiet but consequential transformation. What began as a neuroanatomical question about how visceral signals reach cortical awareness has expanded into one of the most integrative frameworks in contemporary cognitive neuroscience and clinical psychology. Synthesizing across the six thematic areas covered in this review, several cross-cutting conclusions emerge that collectively redefine where understanding now stands — and where it must go next.

Perhaps the most significant conceptual shift of recent years is the consolidation of predictive coding and active inference as the dominant theoretical scaffolding for interoception. Earlier models positioned the brain primarily as a receiver of ascending visceral signals, with regions like the insular cortex serving as relay and integration hubs. The predictive processing turn has inverted this picture: the brain is now understood to actively generate predictions about internal states, with ascending signals functioning primarily as prediction error updates. This reframing does not simply repackage old findings — it generates genuinely new hypotheses. Crucially, it bridges what had previously been treated as separate literatures: the neural architecture of interoception, its computational logic, its role in emotion, and its disruption in psychopathology all become expressions of the same underlying inferential process. The insight that psychiatric conditions such as depression, anxiety, eating disorders, and somatic symptom disorders may share a common substrate in miscalibrated interoceptive inference — rather than in disorder-specific neural lesions — represents a meaningful convergence that was not clearly articulable even three years ago.

This theoretical unification, however, has outpaced the methodological infrastructure available to test it. The measurement theme in this review reveals a persistent and troubling fragmentation: accuracy tasks, self-report sensibility measures, and metacognitive awareness indices frequently dissociate from one another, and the psychometric properties of widely used paradigms remain contested. The heartbeat detection and heartbeat discrimination tasks, long the workhorses of interoceptive accuracy research, continue to generate divergent findings that may reflect task-specific cognitive demands as much as genuine interoceptive ability. Recent work has begun to address this through latent variable approaches and Bayesian signal-detection frameworks, but no consensus gold standard has emerged. The network modelling approach applied to MAIA validation — revealing that attitudinal openness may be upstream of interoceptive skill rather than its consequence [76] — represents a methodological innovation that could reshape understanding of how different interoceptive sub-constructs relate to one another, but it requires replication across diverse cultural and clinical populations. The implication is sobering: theoretical progress has been faster than empirical traction, and claims about individual differences in interoception — whether as predictors of emotional regulation capacity or as therapeutic targets — rest on a measurement foundation that remains unsettled.

The relationship between interoception and emotional granularity represents one of the more productive integrations to emerge from recent work. The Theory of Constructed Emotion provides the theoretical architecture for this integration, formally specifying how core affect — generated through interoceptive prediction — is transformed into discrete emotional experiences through conceptual categorization, with granularity reflecting the precision and richness of the conceptual repertoire brought to bear on interoceptive states [6]. The capacity to differentiate discrete emotional states from undifferentiated arousal appears meaningfully linked to the fidelity with which individuals can perceive and interpret internal signals, but also — critically — to the conceptual and linguistic resources available for categorizing those signals. This is not merely a correlational observation; it carries direct implications for intervention. Body-based and contemplative practices — mindfulness, yoga, somatic therapies — appear to improve not just interoceptive accuracy in narrow task senses, but the broader capacity to use bodily information adaptively in emotional contexts, potentially by refining both the precision of interoceptive signals and the emotion concepts through which they are interpreted [26]. Crucially, recent training studies suggest that these gains are domain-specific rather than globally transferable, implying that intervention design matters enormously. Training interoceptive attention without simultaneously cultivating the interpretive and regulatory skills to use that attention may produce heightened awareness without corresponding benefit — and in clinical populations with hypersensitivity profiles, potentially harm.

Developmental Origins and the Relational Constitution of Interoceptive Competence

A critical dimension that substantially extends and deepens the foregoing conclusions concerns the developmental origins of interoceptive competence and emotional granularity. The evidence reviewed in this paper reveals that interoception is not a fixed biological endowment that matures along a predetermined trajectory but a relationally constituted capacity that is actively shaped by the caregiving environment from the earliest stages of life. Attachment processes, parental mentalization, and the quality of co-regulatory exchanges emerge as primary determinants of how children learn to attend to, trust, and interpret their own bodily signals [32, 36, 91]. The reconceptualisation of emotion regulation as constitutive of emotional activation itself — rather than as a corrective process imposed from above — implies that caregiver co-regulation is not merely a supplement to developing self-regulation but part of the medium through which the child’s emotional and interoceptive systems are organised [91, 33]. When this co-regulatory scaffold is disrupted — through parental rejection of emotional expression, parental psychopathology, or patterns of insecure attachment — the result may be a double deficit in which interoceptive signals are neither clearly perceived nor precisely categorised, establishing the conditions for low emotional granularity that persist into adulthood. Critically, research on traumatic childhood experiences documents that body dissociation — a habitual learned disregard of bodily signals, likely acquired as a protective response to chronic stress — functions as the primary mediator between early adversity and adult emotion dysregulation, and is robustly elevated across clinical populations with depression, somatic symptom disorder, and PTSD relative to healthy controls [100]. This pattern underscores that the deficit is not simply one of interoceptive accuracy but of metacognitive trust in one’s own bodily signals, precisely the dimension that early relational experience is best positioned to shape [100, 111].

This developmental perspective reframes the trainability findings from the intervention literature in important ways. Adult interoceptive training — whether through mindfulness, body-based therapy, or intensive self-monitoring — may in part represent the reconstruction of a capacity that was inadequately scaffolded in childhood rather than the acquisition of an entirely novel skill. The active inference framework supports this interpretation: if children’s interoceptive priors are shaped by early caregiving experiences, and if those priors become progressively more entrenched across development [17], then adult interventions face the challenge of recalibrating inferential architectures that have been consolidated over decades rather than merely training novel perceptual abilities. This may explain why contemplative and body-based interventions show moderate but variable effect sizes — for instance, three months of body-focused contemplative training produces significant improvements in the regulatory dimensions of interoceptive awareness (Self-Regulation: d = 0.72; Attention Regulation: d = 0.54; Body Listening: d = 0.40) but no significant change in basic bodily signal noticing [85] — and why the benefits of interoceptive training are not uniformly distributed across individuals. Notably, training gains are predicted more strongly by participants’ appreciation and integration of practice than by sheer practice volume [85], suggesting that motivational and relational factors continue to modulate interoceptive learning even in adulthood. Those whose early relational environments produced more rigid or suppressive interoceptive priors — manifest particularly as body dissociation and distrust of bodily signals [100] — may require qualitatively different or more extended intervention than those whose developmental history was more supportive. The developmental evidence thus carries a direct clinical implication: intervention programmes aimed at enhancing interoceptive awareness and emotional granularity in adults should be designed with sensitivity to the relational origins of the capacities they seek to restore, while prevention-oriented programmes in early childhood should prioritise caregiver-focused components that strengthen parental mentalization of children’s bodily states alongside any direct interoceptive training.

Cultural Generalizability and the Limits of WEIRD-Derived Models

A second dimension that qualifies and extends the review’s primary conclusions concerns the cultural generalizability of the neural, computational, and clinical models that currently dominate interoception research. The overwhelming majority of the empirical work reviewed in this paper — from the foundational neuroanatomical accounts of Craig [46] through the predictive coding frameworks of Seth and colleagues [58] to the clinical taxonomies of Paulus and colleagues [17] — has been developed using Western, educated, industrialised, rich, and democratic (WEIRD) samples. The question of whether this theoretical architecture generalises across culturally diverse populations is not a peripheral methodological concern but a substantive scientific one, because the constructionist frameworks that underpin much of the review’s integrative argument explicitly predict that culture should modulate both the content and the process of interoceptive-emotional construction.

The available cross-cultural evidence suggests that this prediction is borne out at multiple levels of analysis. At the neural level, the demonstration that Chinese and American participants recruit distinct subregions of the anterior insula during emotional feeling — despite equivalent peripheral cardiac arousal — provides direct evidence that the neural architecture of interoceptive-emotional processing is culturally modulated rather than universal [22]. This is a finding of considerable theoretical consequence. The anterior insula has been treated throughout this review as the primary cortical hub for interoceptive integration, and the computational models built upon it — including the interoceptive predictive coding model of conscious presence [58] and the active inference account of emotion [9] — have been specified without reference to cultural variation in how this hub operates. If dorsal and ventral anterior insular subregions are differentially recruited depending on cultural background, then the computational dynamics of interoceptive inference — including the precision-weighting of prediction errors and the conceptual categories through which they are resolved — may vary systematically across populations in ways that current models do not accommodate.

At the behavioural level, cross-cultural evidence reveals that interoceptive accuracy itself varies meaningfully between populations: a direct comparison of Japanese and European participants found significant group differences in heartbeat detection performance, suggesting that cultural context shapes not only the interpretation but the measurable precision of interoceptive signalling [40]. A broader review of somatic awareness and interoceptive accuracy across cultures further documents that East Asian populations exhibit reliably different patterns of bodily attention and somatic representation compared to Western samples [39]. Beyond interoception narrowly construed, emotional expressivity, the gender patterning of emotional communication, and the attentional orientation toward bodily signals all vary meaningfully across cultural contexts [94, 81]. These are not superficial reporting differences but reflect the cultural socialisation of the very emotion concepts and bodily attention habits that constructionist theory identifies as constitutive of emotional experience [42, 6]. The Conceptual Act Theory’s prediction — that languages and cultures with richer or differently structured emotion lexicons will produce different patterns of emotional granularity — remains largely untested in neuroscientific settings, yet it represents one of the most consequential hypotheses in the field. Illustrating the point, analyses of the Turkish emotion term heyecan demonstrate that culturally specific emotional concepts carry distinct conceptual structures that do not map cleanly onto the affective categories presupposed by English-language research instruments [43]. If emotion concepts are constitutive of emotional experience, and if those concepts are culturally furnished, then the construct of emotional granularity itself cannot be treated as a culture-free individual difference variable: it is, by the field’s own theoretical logic, culturally constituted.

The clinical implications of this cultural constitution are substantial and have been insufficiently reckoned with in the existing literature. Cultures differ not merely in emotional lexicons but in the fundamental idioms through which psychological distress is expressed and communicated — with some populations exhibiting markedly greater tendencies toward somatic rather than psychologised presentation of emotional suffering [44], a distinction with direct consequences for how interoceptive-emotional assessments are interpreted. If the emotion concepts through which interoceptive signals are categorised are culturally furnished, then clinical assessments of emotional granularity — and by extension, diagnostic frameworks that treat low granularity as a transdiagnostic risk marker — will perform differently, and potentially misleadingly, across cultural groups. The differential recruitment of anterior insular subregions documented by [22], combined with cross-cultural variation in interoceptive accuracy [40, 39], suggests that the neural and behavioural signatures currently used to anchor computational models of interoceptive inference may not function as culturally invariant biomarkers. Intervention programmes designed to enhance interoceptive awareness and emotional differentiation — however well-validated in WEIRD populations — cannot be assumed to generalise without modification to individuals whose cultural background has shaped fundamentally different interoceptive-emotional architectures. This is not a counsel of relativism but a call for precision: culturally sensitive assessment and intervention design will require specification of which aspects of interoceptive processing are universal in their broad outlines and which are genuinely culturally particular in their instantiation, a distinction that the field does not yet have the empirical tools to draw cleanly.

Integrative Summary: Three Pillars of the Review’s Contribution

Taken together, the evidence synthesised in this review advances understanding of interoception and emotional granularity along three mutually reinforcing pillars. The first is the neural-computational architecture of interoceptive inference: the consolidation of predictive coding and active inference frameworks [9, 10, 11] has provided a principled, generative account of how the brain constructs internal states and how miscalibration of precision-weighting at this level [72, 17] propagates into the wide spectrum of psychiatric presentations documented across the clinical literature. The second pillar is the developmental-relational constitution of interoceptive competence: the caregiving environment, attachment processes, and early co-regulatory exchanges [32, 33] are not merely contextual factors in the story of interoception but are constitutive of the interoceptive and emotional architecture that individuals bring to adulthood [34, 112], with direct consequences for the efficacy and design of later intervention. The third pillar is the cultural-linguistic mediation of interoceptive-emotional experience: the emotion concepts, attentional habits, and bodily norms furnished by cultural membership [42, 12] shape both the content and the neural instantiation of interoceptive inference, such that the construct of emotional granularity [29] — and the clinical models that depend upon it — must be understood as culturally situated rather than culturally neutral.

These three pillars are not independent. The precision-weighting mechanisms formalised in computational models [9, 11] are the very mechanisms through which relational early experience consolidates interoceptive priors, and the cultural-linguistic environment [42] is the medium through which both developmental and adult-life influences on emotional granularity are transmitted and sustained. The integration of these perspectives represents, in the view advanced here, the most important theoretical development in the field since the early neuroimaging synthesis that established the insula’s centrality to interoceptive processing [23, 14, 22].

Priorities and Forward-Looking Conclusions

Against this background, several evidence-based priorities emerge clearly. The most immediate is the standardisation of interoceptive measurement. The persistent dissociation between accuracy, sensibility, and metacognitive awareness indices [18, 13] — and the contested psychometric properties of the paradigms used to assess each [19, 73] — means that the field’s theoretical advances cannot be matched by commensurate empirical traction until a more principled, psychometrically validated, and cross-culturally tested measurement framework is established. The network modelling innovations already applied to self-report instruments [76] and the Bayesian signal-detection approaches developed for accuracy tasks [60, 28] represent promising directions, but their consolidation into a usable consensus toolkit requires coordinated effort across research groups.

The second priority is the integration of interoceptive assessment into clinical practice in a culturally sensitive manner. The evidence reviewed here supports the value of interoceptive profiling — distinguishing hyperprecise from hypoprecise from context-rigid inferential styles [17, 9] — as a means of personalising intervention. However, the cross-cultural neural evidence [22] and the constructionist theoretical framework [42, 6] jointly require that such profiling be normed and interpreted with reference to the cultural context of the individual being assessed, rather than against WEIRD-derived universal benchmarks [39, 40]. Clinical implementation that ignores cultural variation in interoceptive-emotional architecture risks both misdiagnosis and the imposition of interventions whose implicit model of healthy interoceptive functioning is not appropriate to the population served [41].

The third priority, and perhaps the one with the greatest potential for long-term impact, is the development and rigorous evaluation of caregiver-focused early intervention programmes. The developmental evidence makes clear that the interoceptive and emotional capacities that clinical and contemplative interventions struggle to recalibrate in adults are, in their origins, products of the early relational environment [32, 36, 91]. Strengthening parental mentalization of children’s bodily and emotional states [111, 33] — equipping caregivers to serve as reliable, responsive co-regulators of interoceptive experience — represents the most proximal lever available for preventing the consolidation of the maladaptive interoceptive priors that underlie so much of the psychopathological burden documented in this review [8]. Whether at the level of neural architecture, developmental trajectory, or cultural constitution, the evidence converges on a single, generative conclusion: how we come to know and interpret the signals of the body is not a private biological fact but a socially constituted, developmentally sensitive, and culturally embedded achievement — and it is one that can, with appropriately designed support, be meaningfully shaped for the better.

---

References

[1] Chen, W., Schloesser, D., Arensdorf, A., et al. (2021). The Emerging Science of Interoception: Sensing, Integrating, Interpreting, and Regulating Signals within the Self. Trends in Neurosciences. [2] Ceunen, E., Vlaeyen, J., Van Diest, I. (2016). On the Origin of Interoception. Frontiers in Psychology. [3] Critchley, H., Garfinkel, S. (2017). Interoception and emotion. Current Opinion in Psychology. [4] Tsakiris, M., Critchley, H. (2016). Interoception beyond homeostasis: affect, cognition and mental health. Philosophical Transactions of the Royal Society B Biological Sciences. [5] Wiens, S. (2005). Interoception in emotional experience. Current Opinion in Neurology. [6] Barrett, L. (2016). The theory of constructed emotion: an active inference account of interoception and categorization. Social Cognitive and Affective Neuroscience. [7] Feldman, M., Bliss‐Moreau, E., Lindquist, K. (2024). THE NEUROBIOLOGY OF INTEROCEPTION AND AFFECT. Trends in Cognitive Sciences. [8] Quadt, L., Critchley, H., Garfinkel, S. (2018). The neurobiology of interoception in health and disease. Annals of the New York Academy of Sciences. [9] Seth, A., Friston, K. (2016). Active interoceptive inference and the emotional brain. Philosophical Transactions of the Royal Society B Biological Sciences. [10] Seth, A. (2013). Interoceptive inference, emotion, and the embodied self. Trends in Cognitive Sciences. [11] Barrett, L., Simmons, W. (2015). Interoceptive predictions in the brain. Nature reviews. Neuroscience. [12] Ventura‐Bort, C., Wendt, J., Weymar, M. (2021). The Role of Interoceptive Sensibility and Emotional Conceptualization for the Experience of Emotions. Frontiers in Psychology. [13] Calì, G., Ambrosini, E., Picconi, L., et al. (2015). Investigating the relationship between interoceptive accuracy, interoceptive awareness, and emotional susceptibility. Frontiers in Psychology. [14] Terasawa, Y., Fukushima, H., Umeda, S. (2011). How does interoceptive awareness interact with the subjective experience of emotion? An fMRI Study. Human Brain Mapping. [15] Füstös, J., Gramann, K., Herbert, B., et al. (2012). On the embodiment of emotion regulation: interoceptive awareness facilitates reappraisal. Social Cognitive and Affective Neuroscience. [16] Harrison, L., Kats, A., Williams, M., et al. (2019). The Importance of Sensory Processing in Mental Health: A Proposed Addition to the Research Domain Criteria (RDoC) and Suggestions for RDoC 2.0. Frontiers in Psychology. [17] Paulus, M., Feinstein, J., Khalsa, S. (2019). An Active Inference Approach to Interoceptive Psychopathology. Annual Review of Clinical Psychology. [18] Garfinkel, S., Seth, A., Barrett, A., et al. (2014). Knowing your own heart: Distinguishing interoceptive accuracy from interoceptive awareness. Biological Psychology. [19] Mehling, W. (2016). Differentiating attention styles and regulatory aspects of self-reported interoceptive sensibility. Philosophical Transactions of the Royal Society B Biological Sciences. [20] Jones, A., Silas, J., Todd, J., et al. (2021). Exploring the Multidimensional Assessment of Interoceptive Awareness in youth aged 7-17 years. Journal of Clinical Psychology. [21] Cheadle, J., Davidson-Turner, K., Goosby, B. (2024). Active Inference and Social Actors: Towards a Neuro-Bio-Social Theory of Brains and Bodies in Their Worlds. KZfSS Kölner Zeitschrift für Soziologie und Sozialpsychologie. [22] Immordino‐Yang, M., Yang, X., Damásio, H. (2014). Correlations between social-emotional feelings and anterior insula activity are independent from visceral states but influenced by culture. Frontiers in Human Neuroscience. [23] Pollatos, O., Gramann, K., Schandry, R. (2006). Neural systems connecting interoceptive awareness and feelings. Human Brain Mapping. [24] Pollatos, O., Kirsch, W., Schandry, R. (2005). Brain structures involved in interoceptive awareness and cardioafferent signal processing: A dipole source localization study. Human Brain Mapping. [25] Suzuki, N., Yamamoto, T. (2023). The influence of interoceptive accuracy on the verbalization of emotions. Scientific Reports. [26] Wilson‐Mendenhall, C., Dunne, J. (2021). Cultivating Emotional Granularity. Frontiers in Psychology. [27] Farb, N., Daubenmier, J., Price, C., et al. (2015). Interoception, contemplative practice, and health. Frontiers in Psychology. [28] Allen, M., Levy, A., Parr, T., et al. (2022). In the Body’s Eye: The computational anatomy of interoceptive inference. PLoS Computational Biology. [29] Hoemann, K., Barrett, L., Quigley, K. (2021). Emotional Granularity Increases With Intensive Ambulatory Assessment: Methodological and Individual Factors Influence How Much. Frontiers in Psychology. [30] Weijs, M., Daum, M., Lenggenhager, B. (2023). Cardiac interoception in infants: Behavioral and neurophysiological measures in various emotional and self-related contexts. Psychophysiology. [31] Musculus, L., Tünte, M., Raab, M., et al. (2021). An Embodied Cognition Perspective on the Role of Interoception in the Development of the Minimal Self. Frontiers in Psychology. [32] Oldroyd, K., Pasupathi, M., Wainryb, C. (2019). Social Antecedents to the Development of Interoception: Attachment Related Processes Are Associated With Interoception. Frontiers in Psychology. [33] Taipale, J. (2016). Self-regulation and Beyond: Affect Regulation and the Infant–Caregiver Dyad. Frontiers in Psychology. [34] Gagliardi, M. (2024). The role of developmental caregiving programming in modulating our affiliation tendency and the vulnerability to social anxiety and eating disorders. Frontiers in Psychology. [35] Lane, R., Smith, R. (2021). Levels of emotional awareness: Theory and measurement of a sociolemotional skill. Journal of Intelligence. [36] De Witte, N., Sütterlin, S., Braet, C., et al. (2016). Getting to the Heart of Emotion Regulation in Youth: The Role of Interoceptive Sensitivity, Heart Rate Variability, and Parental Psychopathology. PLoS ONE. [37] Palser, E., Fotopoulou, A., Pellicano, E., et al. (2019). Dissociation in How Core Autism Features Relate to Interoceptive Dimensions: Evidence from Cardiac Awareness in Children. Journal of Autism and Developmental Disorders. [38] Huggins, C., Donnan, G., Cameron, I., et al. (2021). Emotional self-awareness in autism: A meta-analysis of group differences and developmental effects. Autism. [39] Ma‐Kellams, C. (2014). Cross-cultural differences in somatic awareness and interoceptive accuracy: a review of the literature and directions for future research. Frontiers in Psychology. [40] Ubukata, S., Watanabe, K., Isomura, T. (2023). Cultural Differences in Interoceptive Accuracy: Comparison Between Japan and Europe. Japanese Psychological Research. [41] Shoji, M., Mehling, W., Hautzinger, M., et al. (2018). Investigating Multidimensional Interoceptive Awareness in a Japanese Population: Validation of the Japanese MAIA-J. Frontiers in Psychology. [42] Lindquist, K., MacCormack, J., Shablack, H. (2015). The role of language in emotion: predictions from psychological constructionism. Frontiers in Psychology. [43] Baş, M. (2024). The conceptual nature of the Turkish emotion term ‘Heyecan’. Language and Cognition. [44] Ryder, A., Yang, J., Heine, S. (2002). Somatization vs. Psychologization of Emotional Distress: A Paradigmatic Example for Cultural Psychopathology. Online Readings in Psychology and Culture. [45] Moneta, M., Kaechele, H. (2023). A theoretical and clinical perspective of an embodied view in psychotherapy of somatic symptoms disorders. Research in Psychotherapy Psychopathology Process and Outcome. [46] Agnella Craig (2002). How do you feel? Interoception: the sense of the physiological condition of the body. Nature reviews. Neuroscience. [47] Evrard, H. (2019). The Organization of the Primate Insular Cortex. Frontiers in Neuroanatomy. [48] Porges, S. (2025). Polyvagal theory: a journey from physiological observation to neural innervation and clinical insight. Frontiers in Behavioral Neuroscience. [49] Cerritelli, F., Frasch, M., Antonelli, M., et al. (2021). A Review on the Vagus Nerve and Autonomic Nervous System During Fetal Development: Searching for Critical Windows. Frontiers in Neuroscience. [50] Pinna, T., Edwards, D. (2020). A Systematic Review of Associations Between Interoception, Vagal Tone, and Emotional Regulation: Potential Applications for Mental Health, Wellbeing, Psychological Flexibility, and Chronic Conditions. Frontiers in Psychology. [51] Nogueira, M., Magalhães, J., Sampaio, A., et al. (2025). Examining Insula–Default Mode Network Functional Connectivity and Its Relationship with Heart Rate Variability. Brain Sciences. [52] Gu, X., Hof, P., Friston, K., et al. (2013). Anterior Insular Cortex and Emotional Awareness. The Journal of Comparative Neurology. [53] Craig, A. (2009). Emotional moments across time: a possible neural basis for time perception in the anterior insula. Philosophical Transactions of the Royal Society B Biological Sciences. [54] Seeley, W., Menon, V., Schatzberg, A., et al. (2007). Dissociable Intrinsic Connectivity Networks for Salience Processing and Executive Control. Journal of Neuroscience. [55] Farb, N., Segal, Z., Mayberg, H., et al. (2007). Attending to the present: mindfulness meditation reveals distinct neural modes of self-reference. Social Cognitive and Affective Neuroscience. [56] Tisserand, A., Philippi, N., Botzung, A., et al. (2023). Me, Myself and My Insula: An Oasis in the Forefront of Self-Consciousness. Biology. [57] Kleckner, I., Zhang, J., Touroutoglou, A., et al. (2017). Evidence for a Large-Scale Brain System Supporting Allostasis and Interoception in Humans. [58] Seth, A., Suzuki, K., Critchley, H. (2012). An interoceptive predictive coding model of conscious presence. Frontiers in Psychology. [59] Pezzulo, G., Parr, T., Friston, K. (2022). Active Inference: The Free Energy Principle in Mind, Brain, and Behavior. Biological Psychology. [60] Suksasilp, C., McLanachan, A., Quadt, L., et al. (2023). A hierarchical Bayesian model reveals increased precision weighting for afferent cardiac signals, and reduced anxiety, as a function of interoceptive training. [61] Silvanto, J. (2025). Interoception, insula, and agency: a predictive coding account of aphantasia. Frontiers in Psychology. [62] Monzel, M., Nagai, Y., Silvanto, J. (2025). The role of subjective interoception in autobiographical deficits in aphantasia. [63] Silvanto, J., Nagai, Y. (2025). General sense of agency mediates the relationship between subjective interoception and mental imagery in aphantasia and hyperfantasia. [64] Jungilligens, J., Paredes-Echeverri, S., Popkirov, S., et al. (2022). A new science of emotion: implications for functional neurological disorder. Brain. [65] Bohlen, L., Shaw, R., Cerritelli, F., et al. (2021). Osteopathy and Mental Health: An Embodied, Predictive, and Interoceptive Framework. Frontiers in Psychology. [66] Scheffer, M., Bockting, C., Borsboom, D., et al. (2024). A Dynamical Systems View of Psychiatric Disorders—Practical Implications: A Review. JAMA Psychiatry. [67] Hoemann, K., Wormwood, J., Barrett, L., et al. (2023). Multimodal, Idiographic Ambulatory Sensing Will Transform our Understanding of Emotion. Affective Science. [68] Gómez-Carrillo, A., Kirmayer, L. (2023). A cultural-ecosocial systems view for psychiatry. Frontiers in Psychiatry. [69] Unknown (2023). The New Measures of Interoceptive Accuracy: A Systematic Review and Assessment. [70] Rominger, C., Fink, A., Perchtold‐Stefan, C., et al. (2025). The Interoceptive Brain: Confidence Ratings and Accuracy Scores are Independently and Differently Associated With Task-Related Alpha Power During the Heartbeat Tracking Task. Psychophysiology. [71] Dunn, B., Stefanovitch, I., Evans, D., et al. (2010). Can you feel the beat? Interoceptive awareness is an interactive function of anxiety- and depression-specific symptom dimensions. Behaviour Research and Therapy. [72] Ainley, V., Apps, M., Fotopoulou, A., et al. (2016). ‘Bodily precision’: a predictive coding account of individual differences in interoceptive accuracy. Philosophical Transactions of the Royal Society B Biological Sciences. [73] Mehling, W., Acree, M., Stewart, A., et al. (2018). The Multidimensional Assessment of Interoceptive Awareness, Version 2 (MAIA-2). PLoS ONE. [74] Kerr, C., Sacchet, M., Lazar, S., et al. (2013). Mindfulness starts with the body: somatosensory attention and top-down modulation of cortical alpha rhythms in mindfulness meditation. Frontiers in Human Neuroscience. [75] Mehling, W., Price, C., Daubenmier, J., et al. (2012). The Multidimensional Assessment of Interoceptive Awareness (MAIA). PLoS ONE. [76] Klocek, A., Řiháček, T., Cígler, H. (2025). Czech Version of the Multidimensional Assessment of Interoceptive Awareness (MAIA): Psychometric Evaluation and Network Model. Mindfulness. [77] Stern, E., Grimaldi, S., Muratore, A., et al. (2017). Neural correlates of interoception: Effects of interoceptive focus and relationship to dimensional measures of body awareness. Human Brain Mapping. [78] Seo, A., Nishiyama, R., Lee, K., et al. (2025). Factors Linking Interoception and Sleep Across the Adult Lifespan. Psychophysiology. [79] Ulus, G., Aisenberg, D. (2022). Interoception in Old Age. Brain Sciences. [80] Herbert, B., Herbert, C., Pollatos, O. (2011). On the Relationship Between Interoceptive Awareness and Alexithymia: Is Interoceptive Awareness Related to Emotional Awareness?. Journal of Personality. [81] Freya Prentice, Hannah Hobson, R. L. Spooner, et al. (2022). Gender differences in interoceptive accuracy and emotional ability: An explanation for incompatible findings. Neuroscience & Biobehavioral Reviews. [82] Ainley, V., Tsakiris, M. (2013). Body Conscious? Interoceptive Awareness, Measured by Heartbeat Perception, Is Negatively Correlated with SelfObjectification. PLoS ONE. [83] Georgiou, E., Matthias, E., Kobel, S., et al. (2015). Interaction of physical activity and interoception in children. Frontiers in Psychology. [84] Price, C., Hooven, C. (2018). Interoceptive Awareness Skills for Emotion Regulation: Theory and Approach of Mindful Awareness in Body-Oriented Therapy (MABT). Frontiers in Psychology. [85] Bornemann, B., Herbert, B., Mehling, W., et al. (2015). Differential changes in self-reported aspects of interoceptive awareness through 3 months of contemplative training. Frontiers in Psychology. [86] Lindquist, K., Wager, T., Kober, H., et al. (2012). The brain basis of emotion: A meta-analytic review. Behavioral and Brain Sciences. [87] Lisa Feldman Barrett (2012). Emotions are real.. Emotion. [88] Montolio, M., Visser, M., Zareba, M. (2024). Socio-emotional difficulties observed in alexithymia reflect altered interactions of the semantic and monoaminergic neuromodulatory brain networks. [89] Fuchs, T., Koch, S. (2014). Embodied affectivity: on moving and being moved. Frontiers in Psychology. [90] O’Leary, D., McAuliffe, A. (2025). AM I OK? Investigating the Lived Experience of Emotional Differentiation in a Sample of Irish Men. Counselling and Psychotherapy Research. [91] Ross A. Thompson, Marc D. Lewis, Susan D. Calkins (2008). Reassessing Emotion Regulation. Child Development Perspectives. [92] Tan, Y., Wang, X., Blain, S., et al. (2023). Interoceptive attention facilitates emotion regulation strategy use. International Journal of Clinical and Health Psychology. [93] Guu, S., Chao, Y., Huang, F., et al. (2023). Interoceptive awareness: MBSR training alters information processing of salience network. Frontiers in Behavioral Neuroscience. [94] Truong, V. (2025). Examining Gender and Cultural Influences on Customer Emotions. Journal of Consumer Behaviour. [95] Barca, L., Iodice, P., Chaigneau, A., et al. (2025). Emotional distress and affective knowledge representation one year after the COVID-19 outbreak. PLoS ONE. [96] Brewer, R., Cook, R., Bird, G. (2016). Alexithymia: a general deficit of interoception. Royal Society Open Science. [97] Farb, N., Segal, Z., Anderson, A. (2012). Mindfulness meditation training alters cortical representations of interoceptive attention. Social Cognitive and Affective Neuroscience. [98] Smith, R., Parr, T., Friston, K. (2019). Simulating emotions: An active inference model of emotional state inference and emotion concept learning. [99] Vago, D., Silbersweig, D. (2012). Self-awareness, self-regulation, and self-transcendence (S-ART): a framework for understanding the neurobiological mechanisms of mindfulness. Frontiers in Human Neuroscience. [100] Schmitz, M., Back, S., Seitz, K., et al. (2023). The impact of traumatic childhood experiences on interoception: disregarding one’s own body. Borderline Personality Disorder and Emotion Dysregulation. [101] Gibson, J. (2019). Mindfulness, Interoception, and the Body: A Contemporary Perspective. Frontiers in Psychology. [102] Quinto, R., Russo, F., Scafuto, F., et al. (2025). Effects of a Body-Based Mindfulness Program on Alexithymia, Dispositional Mindfulness, and Distress Symptoms: A Pilot Clinical Trial. Behavioral Sciences. [103] Price, C., Sevinc, G., Farb, N. (2023). Within-Person Modulation of Neural Networks following Interoceptive Awareness Training through Mindful Awareness in Body-Oriented Therapy (MABT): A Pilot Study. Brain Sciences. [104] Paulus, M., Stein, M. (2010). Interoception in anxiety and depression. Brain Structure and Function. [105] Brewer, R., Murphy, J., Bird, G. (2021). HiRes. Neuroscience & Biobehavioral Reviews. [106] Shaffer, C., Westlin, C., Quigley, K., et al. (2022). Annual Review of Clinical Psychology: Allostasis, Action, and Affect in Depression: Insights from the Theory of Constructed Emotion. Annual Review of Clinical Psychology. [107] Stephan, K., Manjaly, Z., Mathys, C., et al. (2016). Allostatic Self-efficacy: A Metacognitive Theory of Dyshomeostasis-Induced Fatigue and Depression. Frontiers in Human Neuroscience. [108] Schmitz, M., Bertsch, K., Löffler, A., et al. (2021). Body connection mediates the relationship between traumatic childhood experiences and impaired emotion regulation in borderline personality disorder. Borderline Personality Disorder and Emotion Dysregulation. [109] Ahrens, A., Hyötyläinen, T., Petrone, J., et al. (2024). Infant microbes and metabolites point to childhood neurodevelopmental disorders. Cell. [110] Juha Silvanto, Yoko Nagai (2025). How Interoception and the Insula Shape Mental Imagery and Aphantasia. Brain Topography. [111] Fonagy, P., Campbell, C., Luyten, P. (2023). Attachment, Mentalizing and Trauma: Then (1992) and Now (2022). Brain Sciences. [112] Schore, A. (2014). Early interpersonal neurobiological assessment of attachment and autistic spectrum disorders. Frontiers in Psychology.