RESEARCH DOSSIER
Interoception
Abstract. Interoception comprises the sensing, integration, interpretation, and behavioral use of internal bodily signals across cardiovascular, respiratory, gastrointestinal, and other physiological systems. This critical narrative research monograph reconstructs its conceptual history, neurophysiology, distributed neural organization, homeostatic and predictive models, experimental paradigms, and psychological significance for emotion, attention, bodily self-experience, and mental health. It evaluates heartbeat counting and discrimination, psychometric self-report, respiratory and gastric perturbation, metacognitive confidence, and the limitations of clinical associations. Primary experiments, theoretical works, systematic reviews, and contemporary meta-analyses converge on the importance of bodily-state processing while revealing considerable task specificity and construct-validity problems. The monograph proposes a methodological research agenda across modalities, intervention designs, developmental and cultural contexts, and human–artificial feedback. It is a critical synthesis of published evidence rather than a report of new experiments.
Keywords: interoception; internal bodily signals; interoceptive accuracy; interoceptive sensibility; interoceptive awareness; visceral perception; interoceptive inference; insula; allostasis; heartbeat counting; heartbeat discrimination; respiratory interoception; gastrointestinal interoception; body awareness; emotion; anxiety; embodied self; psychophysiology
Interoception as a Psychological Research Object
Interoception concerns the sensing, integration, interpretation, and use of physiological signals arising within the organism. Its scientific object includes cardiovascular, respiratory, gastrointestinal, thermoregulatory, metabolic, and other bodily processes through which a living system monitors internal conditions and adjusts behavior. Contemporary usage encompasses both nonconscious processing and consciously accessible sensations, while different research traditions draw the boundary around internal signals differently. A scientifically useful account must specify the sensory source, the neural or behavioral process under investigation, the conscious or nonconscious level of analysis, and the behavioral consequence being explained. Such specificity is central to the conceptual criticism developed by Desmedt et al. (2025) and the interdisciplinary research agenda proposed by Khalsa et al. (2018).
The phenomenon becomes visible in ordinary psychological life through hunger, fullness, thirst, breathlessness, heartbeat sensations, visceral discomfort, and changes in temperature or energy. These experiences arise within ongoing regulation of blood gases, circulation, tissue needs, and the allocation of physiological resources. Sensory information is interpreted in light of learned expectations, attentional priorities, environmental threats, and prior bodily experiences. A feeling of rapid heartbeat may appear during exercise, anticipation, or fear, with meaning determined partly by context. The classification of every such event as a direct measure of interoceptive accuracy would collapse distinctions between bodily change, sensory access, subjective appraisal, and practical response. The historical synthesis by Ceunen et al. (2016) clarifies why these boundaries remain consequential.
This dossier investigates interoception as a multilevel psychological process, relating bodily signaling to awareness, affect, cognition, action, self-experience, and regulation. Its primary method is a critical narrative synthesis of original experimental studies, neuroimaging, psychometric research, conceptual analyses, and systematic reviews. The evidence is assessed by measurement validity, task specificity, causal leverage, replication, and population coverage. It neither reports a new experiment nor assumes that a single cardiac task represents perception across all bodily systems. Its organizing scientific question is how bodily states become usable information and under what conditions that information shapes experience and behavior, with the evidence roadmap of Khalsa et al. (2018) providing a reference for the multidisciplinary scope.
Bodily Signals, Sensory Channels, and Functional Boundaries
Signals relevant to interoception originate across organ systems and reach the central nervous system by heterogeneous pathways. Visceral mechanoreceptors and chemoreceptors, cardiopulmonary afferents, spinal pathways, vagal and glossopharyngeal afferents, and humoral mechanisms participate in the wider regulation of bodily condition. These signals can influence autonomic and endocrine responses without producing an explicit reportable feeling. A researcher studying respiratory load, gastric distension, cardiac phase, or inflammatory signaling therefore examines different physiological inputs, stimulus properties, and temporal scales. Craig's integrative neuroanatomical account Craig (2002) helped establish the importance of these ascending channels while also becoming a starting point for debates about whether the boundaries of interoception should be defined anatomically or phenomenologically.
Interoception overlaps with proprioception, nociception, somatosensation, and thermosensation in ways that cannot always be settled by assigning an afferent to a single category. Muscle state informs effort and fatigue, while cutaneous temperature and deep bodily temperature jointly participate in thermoregulation. Pain can involve somatic or visceral sources and becomes psychologically salient through attention, valuation, and action preparation. The physiologically narrow definition associated with viscera has given way to broader classifications, but broader inclusion also creates problems when instruments developed for one channel are presented as measures of a single general capacity. The analysis of historical terminology in Ceunen et al. (2016) and the framework debate in Desmedt et al. (2025) expose the stakes of these decisions.
Functional boundaries are especially important for distinguishing stimulus representation from conscious access. Arterial pressure changes may influence moment-to-moment threat processing, breathing may guide action before it is verbally described, and gastric signals may contribute to feeding decisions through learned associations. Each case involves a partly different relationship among receptor stimulation, central integration, conscious feeling, and behavioral adaptation. A rigorous study identifies which portion of that chain is measured. Experimental manipulation of physiological phase or load provides stronger leverage over some causal questions than self-report alone, although the manipulation may simultaneously change arousal or attention. The cardiophase experiment of Garfinkel et al. (2014) illustrates an important design that integrates physiological timing with behavioral and neural outcomes.
Historical Origins and the Transformation of the Concept
The language of interoception developed out of early twentieth-century physiological efforts to classify receptive fields and reflex organization. Charles Sherrington used the terms interoceptors and interoceptive receptor fields in The Integrative Action of the Nervous System in 1906, distinguishing receptors associated with the body's interior from exteroceptive and proprioceptive sources. The noun interoception became established later, with documented uses in scientific publications during the 1940s. The shift from receptor classifications toward an account of experienced bodily state is reconstructed in Ceunen et al. (2016). Historical precision matters here because retrospective claims that the modern multidimensional construct was already fully present in Sherrington flatten a century of conceptual change.
Nineteenth-century discussions of emotion, including William James's emphasis on sensed bodily change, prepared an intellectual context for studying how physiology participates in feeling. Twentieth-century psychophysiology then made signals such as heart rate and autonomic activity measurable during emotional and cognitive tasks. In 1981, Schandry described a heartbeat perception procedure and its relationship to emotional experience, helping consolidate a laboratory approach that became remarkably influential Schandry (1981). The experiment is historically important even though the task's inferential assumptions and construct validity have since been questioned. A method's formative role does not guarantee that its scores are specific indicators of the ability it was designed to measure.
A major conceptual expansion arrived with Craig's account of the physiological condition of the body and ascending interoceptive pathways Craig (2002). Later work connected the anterior insula to conscious feeling and bodily awareness Craig (2009). The subsequent research landscape merged human psychophysics, clinical psychology, neuroimaging, computational neuroscience, and contemplative science, bringing both theoretical progress and terminology disputes. The development from visceral receptors to conscious feeling and multidimensional processing is therefore best understood as an accumulation of partially overlapping research programs. Their claims must be compared at a level matching the tasks and data that produced them, rather than treating their different uses of interoception as interchangeable descriptions of one settled capacity.
The Physiology of Internal Regulation and Homeostasis
Physiological regulation operates through recurrent feedback between peripheral organs, spinal and brainstem systems, hypothalamic networks, and cortical mechanisms. Homeostasis describes the maintenance of viable physiological conditions through dynamic compensatory responses rather than an absence of change. Respiratory and cardiovascular adjustments can occur before awareness, while sustained challenges may enter consciousness as breathlessness, thirst, fatigue, or discomfort. Interoceptive signaling helps couple these regulatory systems to learning and goal-directed behavior. The anatomical account of Craig (2002) and the clinical integration offered by Khalsa et al. (2018) both emphasize that subjective report captures only a fraction of the continuous information relevant to maintaining internal conditions.
A central analytical distinction separates physiological perturbation, compensatory regulation, sensory encoding, and conscious interpretation. For example, a rise in respiratory resistance produces a measurable physical challenge, invokes respiratory motor and autonomic adjustments, and may generate distressing dyspnea. These components have different time courses and can be affected differently by expectations, conditioning, and attention. If a participant reports increased discomfort after an experimental manipulation, that finding supports an effect on experience but does not alone identify the causal neural pathway. Conversely, a physiological reaction without reportability remains relevant to internal-state processing. The inclusive account advanced by Desmedt et al. (2025) accommodates such nonconscious contributions while requiring operational clarity.
The science of bodily regulation benefits from work at several temporal resolutions. A heartbeat lasts seconds or less, fluctuations in respiratory rhythm unfold over breaths, meal-related satiety develops across minutes and hours, and endocrine changes can affect behavior over longer intervals. Studies that compress all these phenomena into a single sensitivity score obscure the distinction between fast afferent signaling and slower homeostatic inference. An evidence-based synthesis treats internal regulation as a network of coupled functions and then asks which measurements show cross-domain stability. The large-scale network analysis by Kleckner et al. (2017) offers one neural hypothesis for coordination across regulatory functions while remaining open to scrutiny about what functional connectivity establishes.
Insular Cortex and Distributed Neural Representation
The insular cortex occupies a prominent position in theories of interoception because it is engaged by numerous bodily and affective tasks. In a foundational functional magnetic resonance imaging study, Critchley and colleagues measured activity while participants judged the timing of their heartbeats. Their results implicated insular, somatomotor, and cingulate regions; activity in right anterior insular and opercular cortex was related to task accuracy Critchley et al. (2004). This is evidence for a neural contribution to a particular operational task. It does not establish that the anterior insula exclusively represents every bodily state or that greater activation always indicates superior interoception. Hemodynamic measurement, correlated task demands, and individual differences constrain the generalization.
Craig proposed that posterior-to-anterior insular organization contributes to a progressive integration of bodily representation with affective and subjective significance Craig (2009). This account remains influential because it links neuroanatomy to the phenomenology of feeling. Yet the relevant system includes subcortical structures, brainstem nuclei, hypothalamic mechanisms, anterior cingulate cortex, somatomotor regions, and distributed cortical networks. A single localization claim cannot adequately explain autonomic control, awareness, body ownership, and valuation together. The research program must distinguish anatomical representation, functional participation, necessity, and sufficiency. These are empirically different claims, requiring complementary methods including lesion studies, autonomic recordings, imaging, and targeted experimental interventions.
Kleckner and colleagues described a large-scale intrinsic connectivity arrangement associated with allostasis and interoception, supporting distributed rather than narrowly localized inquiry Kleckner et al. (2017). Functional connectivity identifies correlated activity and candidate network architecture; it does not directly demonstrate the content of a representation or a causal pathway from organ to subjective feeling. Network analyses also depend on preprocessing, parcellation, statistical thresholding, and the behavioral tasks used for interpretation. The strongest current conclusion is that internal-state processing recruits multiple coordinated systems whose functional contributions differ by bodily channel and task. An integrated explanation requires convergence among physiological perturbations, neural response, reports, and behavior across independent designs, not a simple equation between an image-derived network and experienced feeling.
Allostatic Regulation and Anticipatory Bodily Control
Allostasis introduces the idea that regulatory systems act in anticipation of demands and adjust bodily resources as circumstances change. In psychological research, this emphasizes that the brain's representation of internal state supports preparation as well as correction. Energy allocation, thermoregulation, cardiovascular response, and endocrine regulation can be shaped by predicted activity and environmental conditions. Barrett and Simmons interpret interoceptive processing within a framework of descending predictions and ascending sensory constraints Barrett & Simmons (2015). This interpretation gives internal sensing a place in models of adaptive control, but its computational language does not automatically identify a unique neural implementation. The degree of prediction, the temporal horizon, and the response variable require explicit specification in each experiment.
The evidence for distributed regulatory organization includes the intrinsic connectivity analysis by Kleckner and colleagues Kleckner et al. (2017). Their work connects networks implicated in interoception with a larger account of physiological regulation. It is especially valuable as a bridge between interoceptive sensations and regulatory functions that are often unconscious. Nevertheless, correlated resting activity provides a description of network organization under the measurement conditions, and alternative models can explain some of the same correlations. A compelling causal account also needs interventions that vary physiological demands, track relevant neural changes, and show corresponding consequences for autonomic output or adaptive behavior. Functional significance becomes more secure when observed across multiple types of challenge and independent samples.
A further complication is that adaptive regulation may reduce conscious salience precisely when it works efficiently. A person can maintain blood pressure and respiratory function while attending to an external task; heightened awareness may emerge with unexpected load, uncertainty, or learned threat associations. High self-reported interoceptive attention is therefore not a universal marker of successful regulation. The multidimensional research synthesis in Khalsa et al. (2018) urges interpretation of bodily signal detection alongside context, clinical state, and the regulation being attempted. Allostatic theory provides an explanatory framework for these observations, while the empirical task is to measure how anticipatory responses, conscious experience, and behavioral policies are coupled across changing physiological demands.
Predictive Processing and Interoceptive Inference
Predictive-processing approaches describe perception as the ongoing interaction of incoming signals with expectations about their likely causes. Seth applied this framework to interoception by proposing that experienced bodily feeling emerges through inference about the causes of afferent activity Seth (2013). Expectations may affect the weighting of ambiguous bodily evidence, while unexpected signals may revise beliefs or initiate regulatory action. This account unifies several lines of research on emotion, embodiment, and conscious selfhood. Its scientific value depends on specifying falsifiable contrasts with alternative theories, including direct sensory accounts, attention-based explanations, conditioning, and general arousal. An elegant computational narrative cannot substitute for evidence discriminating among mechanistic models.
Barrett and Simmons distinguish ascending sensory constraints from descending visceromotor predictions and discuss how the two might jointly shape internal experience Barrett & Simmons (2015). Seth and Friston extend this perspective toward active inference, where actions and autonomic changes participate in reducing discrepancy between predicted and sensed states Seth & Friston (2016). The central research question is how confidence or precision is assigned to competing signals and expectations. Precision is a formal model parameter, not simply a synonym for attention or intensity. It must be estimated from behavior, neural measures, or experimental manipulation under independently testable assumptions. Otherwise the same observation can be attributed after the fact to stronger predictions, weaker afferents, or different precision weighting.
Computational models deserve comparison at the level of observable predictions. A model may predict different effects of cue reliability, physiological perturbation, prior learning, or attention on detection and confidence. Experimental paradigms can independently change signal strength and expected probability, then assess whether responses fit model-specific patterns. Hierarchical estimation may also explain why context changes the appraisal of identical physiological events. Yet predictive processing encompasses a family of models rather than one universally confirmed algorithm. Research should report parameter recoverability, predictive fit on held-out data, and robustness to model alternatives. The conceptual critiques of Desmedt et al. (2025) make clear why inferential interpretation needs measures that genuinely isolate the psychological dimensions under investigation.
Dimensions of Interoceptive Function
Interoception has often been described through three dimensions: interoceptive accuracy, interoceptive sensibility, and interoceptive awareness. In the influential framework of Garfinkel and colleagues, accuracy refers to objective performance on a bodily signal task, sensibility to self-reported beliefs about bodily sensing, and awareness to correspondence between performance and confidence Garfinkel et al. (2015). The distinctions matter because a participant may report strong bodily attention while performing poorly on a particular detection task. Likewise, a person who performs well can remain uncertain about their performance. Findings from one dimension cannot be substituted for those from another, and the constructs should be labeled according to the actual measurement procedure used.
The tripartite account was an important step toward separating performance from belief and metacognitive evaluation, but it creates problems when each construct is treated as unitary. Murphy, Catmur, and Bird argued for a more differentiated classification, including the distinction between attention and accuracy Murphy et al. (2019). A participant's ability to perceive heartbeats might not predict sensitivity to respiratory obstruction or gastric stimulation, and a questionnaire about body trust can reflect appraisal rather than sensory discrimination. Interoceptive attention can be adaptive during some tasks and maladaptive when threat monitoring becomes self-reinforcing. Construct interpretation therefore requires bodily domain, conscious access, affective value, attentional context, and measurement method to be reported separately.
Desmedt and colleagues' critical analysis emphasizes that familiar dimensions often have been operationalized by tasks whose empirical relationships are weak or uncertain Desmedt et al. (2025). The 2023 systematic review of newer accuracy measures also documents the challenges of establishing valid performance measures across cardiac, respiratory, and gastrointestinal domains Desmedt et al. (2023). These critiques warrant a research design that measures convergent validity, discriminant validity, test-retest stability, and sensitivity to known changes in stimulus conditions. A comprehensive interoception theory requires evidence that presumed common factors actually generalize across channels. Until such evidence is present, domain-specific capacities and specific measurement outcomes provide a more defensible scientific vocabulary than a global ranking of people as good or bad interoceptors.
Heartbeat Counting and the Validity Problem
Heartbeat counting asks a participant to estimate the number of heartbeats occurring during specified intervals without directly measuring every beat through an overt response. The procedure gained influence after Schandry's early study of heartbeat perception and emotional experience Schandry (1981). Researchers often compare reported counts with objectively recorded beats and transform the discrepancy into an accuracy score. This method is inexpensive and easy to combine with questionnaires and other psychological procedures. Yet a favorable score can arise from an informed guess based on typical heart rate, familiarity with one's body, interval duration, or counting strategy. The crucial issue is whether the score reflects detection of actual afferent cardiac events or indirectly acquired knowledge.
Zamariola and colleagues examined the heartbeat counting score and found patterns challenging its interpretation as a straightforward measure of perceptual accuracy, including strong associations with actual heart rate and pervasive underreporting Zamariola et al. (2018). Such evidence is particularly important because measurement error can bias correlations between bodily sensing and emotion, anxiety, or other outcomes. A statistically significant relationship with a heartbeat counting score cannot by itself establish that a psychological trait is associated with accurate cardiac sensation. Alternative sources of score variance must be evaluated empirically, including resting pulse, prior knowledge, instruction wording, estimation bias, and demand characteristics. Physiological recording solves the problem of knowing the actual heartbeat count but does not validate what participants used to produce their reports.
The systematic review by Desmedt and colleagues documents why development of new measures has become an active research priority Desmedt et al. (2023). Better designs can record cardiovascular signals continuously, include explicit controls for heart-rate knowledge, vary intervals, and compare the counting task against discrimination-based methods. Performance should be analyzed using a model of decision strategy rather than treating a single transformed score as direct access to perception. Reported limitations do not erase the historical value of Schandry's paradigm, but they do change what one may infer from earlier literature. The defensible conclusion is that heartbeat counting is a behavioral task with multiple determinants, whose construct validity must be demonstrated within the particular study before it is used as an index of interoceptive acuity.
Heartbeat Discrimination, Confidence, and Signal Detection
Heartbeat discrimination tasks ask participants to judge whether external events coincide with their heartbeats or to compare temporal relations between cardiac and external signals. Compared with counting, such paradigms can reduce reliance on simple estimates of average pulse, but they introduce other perceptual and cognitive demands. Timing acuity, attention, sensory synchrony, prior knowledge, response criteria, and the placement of delays influence accuracy. In the experiment of Garfinkel and colleagues, distinct task-based measures and confidence judgments revealed separability among performance, reported sensing, and metacognitive correspondence Garfinkel et al. (2015). A high score in one paradigm should therefore be understood as performance under specified instructions and stimulus conditions, with no automatic generalization to every bodily sense.
Signal detection theory distinguishes discrimination sensitivity from response criterion, a distinction well suited to bodily-signal experiments. Researchers can vary true and false signal trials, characterize the resulting decisions, and estimate whether a participant distinguishes changes in physiological events better than chance. Confidence reports permit separate evaluation of metacognitive calibration, provided that task difficulty and the dependence between first-order and second-order performance are handled appropriately. The methodological proposals of Murphy et al. (2019) encourage distinguishing attentional deployment from perceptual success. Good measurement also requires appropriate trial counts and repeated testing because unstable individual scores can create spurious relationships with clinical questionnaires or neural activity.
Newer interoceptive accuracy paradigms increasingly include adaptive thresholds, controlled perturbations, and modality-specific signal detection procedures. The systematic review by Desmedt et al. (2023) examines their advantages and unresolved limits across body systems. Improved task validity may increase experimental control but can also narrow the phenomenon being studied; performance during externally induced stimulation need not equal sensitivity to spontaneous endogenous signals. The choice of paradigm should match the inference being sought: detection thresholds for a physical change, accuracy of heartbeat timing judgments, metacognitive confidence, or everyday bodily monitoring. Convergent designs with simultaneous physiology and self-report are preferable to treating agreement between two conceptually different methods as an assumption.
Respiratory Interoception and Experimental Perturbation
Respiration provides a particularly useful domain for studying interoception because the breathing cycle is simultaneously automatic, voluntarily modifiable, and phenomenologically accessible. Experimental work can manipulate inspiratory resistance, air hunger, respiratory timing, or gas composition while recording physiological responses and subjective reports. These manipulations create identifiable bodily changes against which perceptual sensitivity and unpleasantness can be assessed. Their interpretation requires care: effort, anxiety, motor command, chemoreceptor response, and the perception of breathlessness can vary independently. The roadmap developed by Khalsa et al. (2018) emphasizes the need to combine respiratory physiology with explicit psychological task design and avoid reducing respiratory perception to a single universal internal-sensing parameter.
Respiratory sensation is shaped by prediction and prior learning. A participant who expects breathing difficulty may allocate more attention to inspiratory effort and interpret a mild load as threatening, while an experienced athlete may give a comparable sensation a different significance. Experimental manipulations must distinguish perceived intensity from emotional unpleasantness, detection threshold, and willingness to persist. These outcomes address related but separable questions. Interoceptive models such as those discussed by Seth (2013) propose mechanisms by which expected respiratory states can influence experience; clinical research must test their implications independently rather than presuming that reporting intense dyspnea proves altered prediction-error processing.
The review of new interoceptive accuracy measures by Desmedt et al. (2023) includes respiratory tasks alongside cardiac and gastrointestinal paradigms. This broadening is methodologically valuable because it can test whether performance is modality specific. The strongest evidence would demonstrate task reliability, genuine sensitivity to graded physiological manipulation, and discrimination from ordinary attention or suggestibility. Ideally experiments combine load thresholds, autonomic responses, confidence judgments, and multiple sessions. Respiratory tasks may be ecologically meaningful for exertion and respiratory distress, but generalization to chronic illness, panic vulnerability, and individual differences requires representative populations and sensitivity to medical status. Ethical experimental procedures also need limits appropriate to physiological challenge and the recruitment population.
Gastrointestinal Signals, Hunger, and Satiety
Gastrointestinal interoception encompasses sensations related to gastric distension, intestinal movement, nausea, hunger, fullness, and the changing significance of nutrients and energy state. Unlike cardiac-phase paradigms, these processes commonly unfold over minutes or longer and involve complex peripheral, hormonal, and learned cues. Subjective appetite can be affected by meal timing, expectations, social context, emotional state, and prior conditioning. Therefore a rating of fullness cannot be interpreted as a direct or exclusive measure of gastric afferent accuracy. Researchers need a controlled perturbation or physiological reference, together with attention to the multiple routes through which bodily needs guide food-related decisions. The broad interoception framework in Craig (2002) places visceral afference within a wider representation of bodily condition.
Studies of gastric sensing can use controlled distension or other carefully specified physiological challenges to assess perceptual thresholds, discomfort, and appraisal. The procedure must separate physical stimulus intensity from the emotional meaning of symptoms, and measurements should distinguish perception from regulatory response. Gastrointestinal sensations are also modulated by prior exposure and expectation: identical levels of distension may be interpreted differently depending on anticipated consequences. New measurement paradigms summarized by Desmedt et al. (2023) address some limitations of traditional self-report, although reliability and ecological transfer remain under development. Cross-domain validation is essential because a person who detects a cardiac event accurately may show a different profile for slow or spatially diffuse visceral changes.
The psychological significance of gastrointestinal interoception is apparent in appetite regulation and symptom experience, yet evidence is often complicated by physiological heterogeneity. Gastric motility, inflammatory states, medication, sleep, and metabolic conditions can alter both the input signal and the interpretation of that signal. Correlational relationships between self-reported body awareness and eating behavior do not by themselves identify a perceptual mechanism. A rigorous synthesis treats gastrointestinal interoception as a distinct experimental system, while recognizing the broader clinical and cognitive relevance of learning, regulation, and affect. The construct warnings of Desmedt et al. (2025) apply directly: measures labeled interoceptive should make transparent whether they index sensation, appraisal, attention, expectation, or a combination of these processes.
Self-Report Instruments and Metacognitive Interpretation
Self-report questionnaires capture aspects of how people notice, interpret, trust, and respond to bodily sensations in everyday life. The Multidimensional Assessment of Interoceptive Awareness developed by Mehling and colleagues assesses several domains of self-reported body awareness rather than a single perceptual threshold Mehling et al. (2012). Its multidimensional organization recognizes that noticing discomfort, managing attention, using bodily signals for self-regulation, and trusting the body can vary independently. These questionnaires can be useful for studying habitual attitudes and experience when direct laboratory perturbation is impractical. They remain reports of beliefs, dispositions, and remembered experience, subject to interpretation and response-style differences. Their scores should be named as self-reported interoceptive dimensions and should not be called objective sensory accuracy.
The MAIA-2 revision introduced changes to address weaknesses in the original instrument and was evaluated psychometrically in an adult sample drawn from museum visitors Mehling et al. (2018). Such validation supports a defined questionnaire structure under the tested conditions, while leaving open questions about measurement invariance across age, culture, language, clinical status, and context. Psychometric evidence should include factor structure, internal consistency, temporal stability, known-group behavior, and the relationship to nearby constructs. A scale can distinguish self-reported tendencies reliably yet still be weakly related to the ability to detect a physical bodily event. The study design therefore determines whether a questionnaire is used appropriately as a description of experience or inappropriately as proof of perceptual performance.
Metacognitive judgment introduces another layer: how accurately people know when their interoceptive decisions are correct. Performance and confidence can be compared within a task, as illustrated by Garfinkel et al. (2015), but confidence itself may be affected by general response preferences, task understanding, and incentives. Murphy and colleagues suggest that classification must also take account of interoceptive attention and the domain being tested Murphy et al. (2019). An individual who persistently monitors bodily signals can be highly confident and still make inaccurate judgments under uncertainty. Conversely, occasional attention may support effective behavior without producing strong questionnaire endorsement. Scientific synthesis benefits from preserving these distinctions rather than combining behavioral accuracy, self-description, and metacognition into an uninterpretable overall score.
Cardiac Phase, Threat Detection, and Neural Dynamics
The cardiac cycle creates a naturally recurring physiological event with a definable temporal relation to sensory stimulation. During systole, arterial baroreceptor signaling differs from the period between beats, providing an opportunity to ask whether ongoing bodily activity influences perception and evaluation. Garfinkel and colleagues presented emotional faces at different cardiac phases and reported enhanced detection and intensity judgments for fearful faces at systole, with corresponding neural findings Garfinkel et al. (2014). This study illustrates a stronger causal leverage than simply correlating questionnaire scores with anxiety because stimulus timing is experimentally controlled. The result is specific to the experimental conditions and should be tested across other tasks, populations, and emotional categories before broad generalization.
Neural measurements timed to the heartbeat can also relate endogenous bodily processing to externally oriented attention. Using magnetoencephalography, Park and colleagues showed that spontaneous fluctuations in neural responses to heartbeats before a visual stimulus predicted whether a faint visual stimulus would be detected Park et al. (2014). The finding links naturally occurring cardiac-related cortical activity to perception, but its observational component and complex neural sources require careful interpretation. Heartbeat-evoked signals can be influenced by cardiac electrical artifacts, pulse-related movement, attention, and data-processing choices. Claims about a distinctly interoceptive neural mechanism are strengthened by appropriate controls for these sources and by replication with different physiological and imaging methods.
These findings help correct a simplified view of perception as a purely outward-facing operation. Bodily timing can modulate attention and the experience of external events, but the direction and size of modulation depend on the signal, task, and state. Enhanced fear detection at a particular cardiac phase does not show that all interoceptive sensitivity increases anxiety, nor does a heartbeat-locked neural association establish general access to internal bodily information. A coherent experimental program separately examines physiological timing, neural encoding, emotional categorization, and subjective interpretation. The combined evidence of Garfinkel et al. (2014) and Park et al. (2014) makes this cross-modal research area valuable precisely because it offers testable designs rather than unqualified claims about body–mind unity.
Interoception and the Organization of Emotion
Emotional experience frequently includes identifiable bodily changes: changes in breathing, heart rate, muscular readiness, visceral feeling, warmth, and fatigue. Interoception offers one route through which these physiological changes become affectively meaningful, while contextual interpretation, social learning, memory, and goals also contribute. Craig's synthesis linked the representation of the body's condition to subjective feeling Craig (2002), and contemporary accounts extend the relationship toward prediction and regulation Barrett & Simmons (2015). The evidence supports substantial involvement of body-related information in emotional functioning. It does not support the proposition that every emotion is caused solely by awareness of a distinctive peripheral signature, because similar physiological changes can accompany multiple emotional and nonemotional states.
Experiments provide more precise claims than global statements about bodily feeling. Critchley and colleagues identified neural associations during a heartbeat timing task Critchley et al. (2004). Garfinkel and colleagues observed changes in fear perception with cardiac phase Garfinkel et al. (2014). These studies interrogate different mechanisms: one concerns task-linked neural recruitment and another evaluates temporal influences of bodily signals on emotional stimulus processing. They should not be merged into a single effect estimate or interpreted as proving one causal pathway for all emotional categories. Their combined significance is that bodily signaling can be integrated with the systems used to evaluate events, while the pattern depends on the operational context.
Predictive and construction-oriented accounts describe emotion as an inference involving bodily input, learned conceptual categories, and anticipated regulatory needs Seth (2013). Their explanatory scope is broad, but independent experiments must determine when prediction, sensory precision, or contextual learning accounts best for an effect. Self-report measures further complicate interpretation because anxiety, bodily attention, and emotional vocabulary can influence how participants describe sensations. The conceptual analysis of Desmedt et al. (2025) thus has direct relevance for emotion research: an association with body awareness cannot be interpreted as evidence about objective sensory sensitivity until the measured construct is specified. Interoception contributes to emotion through multiple pathways whose evidence base varies by task and theory.
Anxiety, Panic, and Threat Interpretations of Bodily Signals
Anxiety is a major clinical context for studying interoception because autonomic arousal, respiratory sensations, and heightened monitoring of the body can become associated with anticipated harm. A racing heart may be interpreted as a normal consequence of exertion in one setting and as a threat in another. This pattern has encouraged models in which biased interpretation, selective attention, learned fear, and bodily prediction contribute to anxiety symptoms. The scientific question is which component explains the observed association. Heightened vigilance is not equivalent to superior perceptual accuracy, and frightening appraisal is not necessarily a consequence of stronger physiological input. The clinical roadmap of Khalsa et al. (2018) identifies the need for multidimensional tasks, mechanistic specificity, and disorder-sensitive assessment.
A systematic review and meta-analysis of seventy-one studies by Clemente, Amanda Murphy, and Jennifer Murphy found associations between anxiety and several self-reported aspects of interoception, particularly negative evaluations and attention to bodily signals Clemente et al. (2024). The results concern questionnaire-defined dispositions, with heterogeneity among scales and possible overlap between anxiety symptoms and the wording of interoception items. They must not be translated into a conclusion that anxious people necessarily detect their heartbeats more accurately. Some tasks depend on heart-rate knowledge and estimation, while other tasks assess discrimination or attention under explicit instructions. Measurement overlap can amplify apparent associations even when underlying perceptual mechanisms are not changing.
In panic-related processes, an unpleasant bodily event can acquire threat value through experience and contextual learning. A clinically relevant experimental program would independently manipulate bodily stimulation, expectation, and feedback while recording intensity ratings, fear, avoidance, and subsequent learning. Such designs can test whether treatment-related changes involve sensory thresholds, appraisal, attentional patterns, or action tendencies. Evidence on perceived bodily cues can inform psychological intervention research, but a research dossier cannot substitute for individually assessed treatment or licensed clinical guidance. Distinguishing the mechanisms also clarifies why interventions targeting tolerance or interpretation might alter distress without necessarily improving physiological discrimination. The multidimensional framework in Desmedt et al. (2025) helps prevent explanatory confusion between these outcomes.
Depression and Transdiagnostic Mental Health Research
Links between interoception and depression have been investigated through cardiac tasks, bodily awareness scales, and theoretical models of altered energy regulation and affect. These strands often produce different results because they measure different phenomena. A person with depressive symptoms may report diminished vitality, increased bodily discomfort, or changed attention to internal states without exhibiting any uniform shift in objectively measured signal detection. Theoretical interpretations involving impaired integration of bodily information require evidence that separates symptom-related motivation and cognition from perceptual sensitivity. The roadmap of Khalsa et al. (2018) encourages using several modalities and levels of measurement so that transdiagnostic hypotheses can be evaluated in clinically meaningful rather than merely nominal terms.
A 2026 systematic review and meta-analysis by Tan and colleagues reported a small negative association between depressive symptoms and cardiac interoceptive accuracy as assessed by heartbeat counting, while the heartbeat discrimination results did not show the same relationship Tan et al. (2026). This contrast is scientifically important precisely because the counting task has well-documented validity problems Zamariola et al. (2018). A small pooled association cannot therefore be taken as decisive evidence that depression impairs internal sensing; the finding is conditional on the selected studies and measurement procedure. The meta-analysis synthesizes the available samples and reports heterogeneity, but causal direction remains unresolved in correlational evidence. Sleep, medication, medical illness, and symptom duration may also change the measured inputs.
The proposition that interoception is a transdiagnostic mechanism should be evaluated construct by construct and symptom domain by symptom domain. A self-report measure of dysfunctional attention may correlate with anxiety, depression, and somatic complaints partly because it captures a general negative appraisal style. Distinguishing shared vulnerability from overlapping questionnaire content requires longitudinal and experimental data, with clinical groups defined independently of the predictor. Tests should ask whether bodily sensing prospectively predicts changes in symptoms, whether changing the proposed mechanism alters outcomes, and whether the effect replicates across measurement paradigms. A multidisciplinary synthesis can identify plausible relations without declaring a unitary interoceptive deficit present across mental disorders. The careful clinical review in Clemente et al. (2024) illustrates the need to interpret subdimensions independently.
Bodily Self-Consciousness, Ownership, and Embodiment
The sense of having a body integrates tactile, visual, proprioceptive, vestibular, and internal physiological information. Interoceptive theories propose that ongoing bodily regulation contributes to the continuity and affective significance of embodied self-experience. Seth's inferential account connects predictions about internal state to experienced selfhood Seth (2013), while Craig's work associates anterior insular integration with subjective feeling Craig (2009). These are explanatory proposals that intersect with distinct experimental paradigms for body ownership and multisensory integration. The relationship between an internal signal and a reported feeling of ownership must be established experimentally rather than inferred from the fact that both processes are bodily in a broad sense.
Tsakiris, Tajadura-Jiménez, and Costantini studied individual differences in heartbeat-related sensitivity alongside the malleability of body representations, using an experimental approach related to body-ownership illusions Tsakiris et al. (2011). Their results support a relationship between measures of interoception and multisensory body representation in the studied sample and procedure. However, interpretations depend on the sensitivity measure and on the degree to which laboratory illusions capture everyday bodily self-experience. Subsequent research must test whether the association survives changes in cardiac measurement, whether it extends beyond healthy adults, and how it interacts with proprioception, visual synchrony, and prior beliefs. Such findings are a research invitation rather than a global measurement of identity or consciousness.
An embodied-self explanation can be organized into several testable levels: physiological continuity, sensory integration, perceptual ownership, autobiographical interpretation, and social recognition. Evidence may establish a relationship at one level without proving a unified mechanism connecting all levels. Heartbeat-evoked neural signals offer a possible bridge between ongoing internal processing and outward perceptual operations Park et al. (2014), but neural association alone does not demonstrate that a particular conscious self-representation depends on the observed activity. Research gains explanatory precision when it combines bodily perturbation, self-report, objective task performance, and alternative multisensory controls. The most productive conclusion is that internal signals contribute to embodied experience in interaction with other senses, expectations, and context, with several causal links remaining open.
Attention, Perception, and Cognitive Performance
Attention changes which bodily signals become salient and how they influence competing cognitive tasks. Directing attention toward breathing or heartbeat can alter reports of sensation, but the effect may arise through increased monitoring, a different decision criterion, or improved discrimination. Cognitive resources also determine whether bodily events interrupt external tasks or remain largely outside awareness. Murphy and colleagues argue that attention to interoceptive information should be distinguished from accuracy of perception Murphy et al. (2019). This distinction is crucial in experiments comparing people who habitually notice their body with people who simply endorse fewer bodily statements. The two groups may differ in self-description without differing in sensitivity to a physiological stimulus.
Studies integrating cardiac physiology with perceptual tasks show how internal and external information can interact. In the experiment of Park and colleagues, ongoing neural responses related to heartbeats predicted visual detection Park et al. (2014). Garfinkel and colleagues showed that the cardiac phase modulated detection and evaluation of fearful expressions Garfinkel et al. (2014). These designs reveal experimentally tractable links between physiological timing and outward-oriented cognition. Their results should remain tied to task, timing, and stimulus type; enhancement of a particular fear response cannot be generalized to superior attention in every domain. Independent replication across laboratories and control for heartbeat-locked artifacts are essential to distinguish robust functional coupling from analysis-specific associations.
Judgment and decision-making may also involve bodily signals through arousal, valuation, effort, and anticipated consequences. Yet effects attributed to interoception can be confounded with broader constructs such as emotional intensity or threat monitoring unless signal sensing is measured independently. Predictive theories offer candidate explanations for the interaction of current physiology and expectation Barrett & Simmons (2015), but additional data are required to specify which parameters change in a given decision. A strong design separates the bodily perturbation, cognitive task, conscious report, and behavioral outcome, then tests whether the sensory measure mediates any observed effect. This approach preserves the scientific importance of embodiment while preventing the term interoception from functioning as a catchall label for any influence of the body on cognition.
Developmental, Individual, and Cultural Variation
Interoceptive functions develop within a changing physiological and social environment. Childhood and adolescence involve maturational changes in cardiovascular, respiratory, and autonomic systems, as well as learning how to recognize and name bodily experiences. Adults differ in fitness, age, disease history, attention, and the vocabulary available for describing sensations. These differences can affect task performance and self-report independently. It is therefore inadequate to compare mean heartbeat-counting scores across groups without considering resting heart rate, instructions, numeracy, and estimation strategies. The measurement critique of Zamariola et al. (2018) and the multidimensional classification of Murphy et al. (2019) provide foundations for more careful developmental inference, even though neither alone establishes a complete developmental trajectory.
Culture shapes expectations about whether bodily sensations should be noticed, discussed, minimized, or interpreted as signs of illness, social excitement, exertion, or spiritual experience. Questionnaire items can carry different implications across languages and populations. Establishing measurement invariance is especially important before comparing self-report scores internationally. The MAIA and MAIA-2 are useful examples of instruments for which a researcher must check translation quality, factor structure, and population-specific reliability rather than assuming identical meaning across groups Mehling et al. (2012) Mehling et al. (2018). The available evidence base remains uneven across regions, ages, and clinical populations, with a substantial concentration of laboratory data in relatively accessible adult samples. This imbalance limits claims about universal patterns.
Individual differences should ultimately be understood as profiles across bodily channels, attention, evaluation, and metacognition. One person may accurately detect respiratory load while rarely reporting heartbeat sensations; another may monitor heartbeat intensely but rely on nonperceptual estimation. Such variation becomes theoretically informative when a study measures several domains with reliable methods and examines stability over time. The review by Desmedt et al. (2023) highlights the difficulties and opportunities of comparing modalities with new tasks. Longitudinal developmental research, culturally validated instruments, and explicit physiological covariates are needed to discover whether there are general interoceptive capacities alongside domain-specific skills. The strongest current scientific stance is to report the profile measured rather than infer an unobserved global trait.
Training, Mindfulness, and Intervention Research
Interoception has attracted interest as a potential target for training, especially through breath-focused practices, attention to bodily sensation, and psychological interventions involving exposure to physical cues. Such practices can change how people attend to and appraise physiological events, and some may alter task performance under certain conditions. Their outcomes should be separated into sensory discrimination, affective tolerance, trust in bodily cues, emotional regulation, and clinical symptoms. The theoretical review by Farb and colleagues links contemplative practice to interoceptive attention and health while also identifying competing explanatory possibilities Farb et al. (2015). Its hypotheses provide a rationale for controlled study rather than direct proof that mindfulness uniformly increases objective physiological accuracy.
Randomized designs are needed to isolate changes attributable to intervention from expectancy, repeated testing, nonspecific therapist contact, and regression to the mean. Physiological thresholds should be evaluated with tasks designed for the relevant modality, while validated questionnaires can assess changes in habitual appraisal. Improvement on the same task used during training may reflect practice and strategy rather than general sensory transfer. Studies should preregister their primary outcomes, use credible comparison conditions, assess retention, and investigate whether improvements generalize from cardiac awareness to respiratory or gastrointestinal signals. The methodological review of Desmedt et al. (2023) shows why this question requires substantially better outcome measures than have often been available.
Intervention research is also clinically nuanced. A person with distressing bodily hypervigilance may benefit from a different attentional strategy than someone who experiences difficulty describing bodily needs. Increased noticing can support self-regulation in one context and intensify worry in another. Changes in subjective body trust can be meaningful even without improved perceptual discrimination, but the two outcomes should not be conflated. The multidimensional questionnaire developed by Mehling et al. (2012) provides a way to characterize some experiential changes, while the mental-health roadmap of Khalsa et al. (2018) emphasizes the need to establish treatment-relevant causal mechanisms. This dossier evaluates intervention evidence as part of psychological science and does not supply individual clinical treatment instructions.
Methodological Quality, Causal Inference, and Replication
Interoception research combines different levels of evidence, including physiological perturbation, subjective rating, accuracy tasks, cross-sectional surveys, functional connectivity, and computational modeling. Each answers a different question. Experimental manipulation of bodily stimulation can support conclusions about the effects of that manipulation under the tested conditions; correlations between self-reported awareness and symptoms chiefly establish association. Functional imaging locates task-related signals without independently demonstrating their necessity. A critical synthesis must track design features rather than treating all studies as equivalent endorsements of a general interoception theory. The explicit criticism of definitions and measures in Desmedt et al. (2025) is essential because construct ambiguity can undermine causal interpretation before statistical modeling even begins.
The longstanding heartbeat counting literature illustrates how easily an attractive construct can become dependent on an imperfect operational indicator. Associations may arise from heart-rate knowledge, estimation strategies, interval timing, and the mathematical properties of a derived score Zamariola et al. (2018). Newer tasks improve some controls while raising questions about stimulation intensity, external cue use, physiological calibration, fatigue, and between-session reliability Desmedt et al. (2023). Replication should therefore reproduce the operational details and evaluate alternative scoring models. Sufficient trial counts, preregistration, transparent exclusions, confidence intervals, and sensitivity analyses are especially important when researchers interpret small differences between clinical groups or individual differences.
Research synthesis also requires attention to selective publication and heterogeneity. Questionnaire studies can be influenced by shared item content, scale translation, and sample characteristics; experimental effects can be task-specific and dependent on physiological state. The anxiety meta-analysis by Clemente et al. (2024) and the depression meta-analysis by Tan et al. (2026) illustrate how conclusions differ across measures and symptom domains. Pooled associations should be interpreted with their heterogeneity and prediction intervals when available. Convergence is strongest when independent investigators obtain compatible results using measures with different error structures. An adequate claim of an interoceptive mechanism requires a chain from defined physiological signal through measured process to a predicted outcome, with credible alternatives considered at every link.
A Critical Synthesis Across Theories and Measures
The historical and current evidence supports a multilevel account of bodily-state processing. Peripheral signals inform brainstem and higher neural systems; regulatory circuits integrate internal conditions with environmental demands; attention and learned prediction influence conscious experience; and bodily information can alter emotional evaluation and action. Craig's anatomical framework Craig (2002), the insular findings of Critchley et al. (2004), and distributed allostatic network evidence from Kleckner et al. (2017) provide complementary pieces of this picture. Their explanatory claims remain distinct: an anatomical pathway identifies possible signal transmission, task-related activity identifies neural participation, and connectivity identifies covariation among regions. None by itself explains every dimension of interoceptive awareness or adaptive regulation.
The strongest methodological insight is that interoception must be matched to its operational definition. An objective perturbation measure, confidence report, questionnaire scale, and clinical symptom inventory can all concern bodily experience without measuring the same variable. Garfinkel and colleagues' distinction among accuracy, sensibility, and awareness helped expose these differences Garfinkel et al. (2015), and subsequent reviews sharpened the problem of task validity Murphy et al. (2019) Desmedt et al. (2025). A useful synthesis therefore reports each method's sensory channel, target process, reliability, confounds, and supported inference. This standard also reduces apparent theoretical contradictions that originate from comparisons among non-equivalent measures rather than disagreements about a common experimental observation.
The evidence for psychological relevance is substantial but asymmetric. Cardiac-phase experiments provide direct examples of physiologically timed modulation of perception Garfinkel et al. (2014), and neuroimaging demonstrates associations between interoceptive tasks and distributed circuitry. The relationship between questionnaire-defined attention and anxiety is comparatively well synthesized Clemente et al. (2024), whereas generalized claims about a global accuracy deficit across psychopathology remain limited by task diversity and causality problems. Predictive models clarify potential mechanisms but must compete with attention, conditioning, and sensory-discrimination explanations. A mature theory should specify the circumstances under which one account improves prediction over its alternatives, while recognizing that physiological regulation, bodily meaning, and conscious feeling may be coupled through several interacting processes.
Human–Artificial Research and Future Psychological Methods
Psychological research in the Artificial Era introduces an additional question: how external computational systems may affect the way humans recognize and interpret bodily states. A wearable sensor, symptom-tracking interface, conversational system, or adaptive feedback device can supply a representation of heart rate, respiration, sleep, or activity that changes the user's expectations. Such information can help some people understand physiological patterns, but its precision, framing, and timing can also amplify monitoring or bias interpretation. The relevant scientific object remains the human user's sensing and appraisal of bodily conditions. The broader models of prediction and bodily inference developed by Barrett & Simmons (2015) and Seth (2013) provide a basis for asking how external feedback modifies expectation, confidence, and self-report.
Future human–artificial experiments should randomly vary accuracy, uncertainty communication, and presentation of physiological feedback while recording objective bodily signals and responses. They can test whether repeated reliance on a device improves discrimination, replaces it with learned heuristics, or changes emotional interpretation. A useful design would compare correctly calibrated physiological feedback with misleading or delayed information under ethical consent, assessing the ability to distinguish genuine bodily variation after feedback is removed. This approach requires careful attention to privacy, health risks, and unintended anxiety. It also treats artificial systems as part of a human–artifact relationship, rather than assuming that the presence of a biological sensor automatically creates an interoceptive capacity in software.
The conceptual horizon of Artificial Sapiens raises separate questions about artificial self-monitoring, regulation, and persistence. One may investigate functional analogues of internal-state estimation in an artificial system, such as resource monitoring, error detection, or anticipatory control, using explicit operational definitions. Such work should identify the computational variables and observable consequences while keeping claims about subjective bodily feeling tied to their own evidentiary requirements. The interoception literature itself cannot establish that an artificial system has human visceral sensations. The productive cross-disciplinary inquiry is comparative: which principles of prediction, monitoring, and regulation travel across embodiments, and which depend on specifically biological afference? The criteria developed by Khalsa et al. (2018) for clear constructs and convergent methods offer a model for disciplined investigation.
Conclusion
Interoception is a fundamental field of psychological research into the processing of internal bodily conditions. Its subject spans peripheral sensation, nonconscious regulation, neural integration, conscious feeling, attention, learning, and the behavioral use of physiological information. Its history runs from early physiological classifications through psychophysiological heartbeat tasks to contemporary neuroimaging, predictive modeling, and multidimensional measurement. The literature establishes that bodily processes participate in emotion and cognition, and experiments demonstrate specific forms of coupling between cardiac events and perceptual evaluation. These findings are best interpreted through their actual task designs and measurement conditions rather than as confirmation of one universally applicable internal sense. The historical and anatomical syntheses of Ceunen et al. (2016) and Craig (2002) locate the field's enduring questions.
The greatest current obstacle to cumulative inference is inconsistency between theoretical constructs and what commonly used instruments measure. Heartbeat counting, discrimination, questionnaire-defined attention, metacognitive confidence, and physiological perturbation assess overlapping but distinct properties. Evidence from Zamariola et al. (2018), Desmedt et al. (2023), and Desmedt et al. (2025) warrants greater restraint in claims about a general interoceptive accuracy trait or an undifferentiated clinical deficit. The strongest contemporary evidence connects precise physiological manipulations with specific perceptual outcomes, integrates convergent neural and behavioral data, and interprets survey measures as psychological reports with their own validity requirements. Clinical associations are scientifically meaningful when their effect sizes, design limitations, and competing interpretations remain explicit.
A durable research program will combine multimodal physiological measurements, psychophysics, reliable metacognitive assessment, transparent computational comparison, longitudinal development, culturally appropriate measures, and intervention studies that test causal mechanisms. The scientific task is to explain when internal signals become salient, how they guide action, and under which conditions attention or prediction changes their meaning. Artificial Era research may extend these questions to technologically mediated human awareness and functionally comparable artificial monitoring, with each claim established through independently appropriate evidence. Interoception thus provides a rich field for understanding bodily regulation and psychological organization while demanding unusually careful coordination among concepts, methods, and levels of explanation. The evidence agenda described in Khalsa et al. (2018) remains central to this progression.
References
- Barrett, L. F., & Simmons, W. K. (2015). Interoceptive predictions in the brain. Nature Reviews Neuroscience, 16(7), 419–429. https://doi.org/10.1038/nrn3950
- Ceunen, E., Vlaeyen, J. W. S., & Van Diest, I. (2016). On the origin of interoception. Frontiers in Psychology, 7, 743. https://doi.org/10.3389/fpsyg.2016.00743
- Clemente, R., Murphy, A., & Murphy, J. (2024). The relationship between self-reported interoception and anxiety: A systematic review and meta-analysis. Neuroscience & Biobehavioral Reviews, 167, 105923. https://doi.org/10.1016/j.neubiorev.2024.105923
- Craig, A. D. (2002). How do you feel? Interoception: The sense of the physiological condition of the body. Nature Reviews Neuroscience, 3, 655–666. https://doi.org/10.1038/nrn894
- Craig, A. D. (2009). How do you feel—now? The anterior insula and human awareness. Nature Reviews Neuroscience, 10, 59–70. https://doi.org/10.1038/nrn2555
- Critchley, H. D., Wiens, S., Rotshtein, P., Öhman, A., & Dolan, R. J. (2004). Neural systems supporting interoceptive awareness. Nature Neuroscience, 7(2), 189–195. https://doi.org/10.1038/nn1176
- Desmedt, O., Luminet, O., Walentynowicz, M., & Corneille, O. (2023). The new measures of interoceptive accuracy: A systematic review and assessment. Neuroscience & Biobehavioral Reviews, 153, 105388. https://doi.org/10.1016/j.neubiorev.2023.105388
- Desmedt, O., Luminet, O., Maurage, P., & Corneille, O. (2025). Discrepancies in the definition and measurement of human interoception: A comprehensive discussion and suggested ways forward. Perspectives on Psychological Science, 20(1), 76–98. https://doi.org/10.1177/17456916231191537
- Farb, N., Daubenmier, J., Price, C. J., Gard, T., Kerr, C., Dunn, B. D., Klein, A. C., Paulus, M. P., & Mehling, W. E. (2015). Interoception, contemplative practice, and health. Frontiers in Psychology, 6, 763. https://doi.org/10.3389/fpsyg.2015.00763
- Garfinkel, S. N., Minati, L., Gray, M. A., Seth, A. K., Dolan, R. J., & Critchley, H. D. (2014). Fear from the heart: Sensitivity to fear stimuli depends on individual heartbeats. Journal of Neuroscience, 34(19), 6573–6582. https://doi.org/10.1523/JNEUROSCI.3507-13.2014
- Garfinkel, S. N., Seth, A. K., Barrett, A. B., Suzuki, K., & Critchley, H. D. (2015). Knowing your own heart: Distinguishing interoceptive accuracy from interoceptive awareness. Biological Psychology, 104, 65–74. https://doi.org/10.1016/j.biopsycho.2014.11.004
- Khalsa, S. S., Adolphs, R., Cameron, O. G., Critchley, H. D., Davenport, P. W., Feinstein, J. S., Feusner, J. D., Garfinkel, S. N., Lane, R. D., Mehling, W. E., Meuret, A. E., Nemeroff, C. B., Oppenheimer, S., Petzschner, F. H., Pollatos, O., Rhudy, J. L., Schramm, L. P., Simmons, W. K., Stein, M. B., ... Paulus, M. P. (2018). Interoception and mental health: A roadmap. Biological Psychiatry: Cognitive Neuroscience and Neuroimaging, 3(6), 501–513. https://doi.org/10.1016/j.bpsc.2017.12.004
- Kleckner, I. R., Zhang, J., Touroutoglou, A., Chanes, L., Xia, C., Simmons, W. K., Quigley, K. S., Dickerson, B. C., & Barrett, L. F. (2017). Evidence for a large-scale brain system supporting allostasis and interoception in humans. Nature Human Behaviour, 1, 0069. https://doi.org/10.1038/s41562-017-0069
- Mehling, W. E., Price, C., Daubenmier, J., Acree, M., Bartmess, E., & Stewart, A. (2012). The Multidimensional Assessment of Interoceptive Awareness (MAIA). PLOS ONE, 7(11), e48230. https://doi.org/10.1371/journal.pone.0048230
- Mehling, W. E., Acree, M., Stewart, A., Silas, J., & Jones, A. (2018). The Multidimensional Assessment of Interoceptive Awareness, Version 2 (MAIA-2). PLOS ONE, 13(12), e0208034. https://doi.org/10.1371/journal.pone.0208034
- Murphy, J., Catmur, C., & Bird, G. (2019). Classifying individual differences in interoception: Implications for the measurement of interoceptive awareness. Psychonomic Bulletin & Review, 26, 1467–1471. https://doi.org/10.3758/s13423-019-01632-7
- Park, H.-D., Correia, S., Ducorps, A., & Tallon-Baudry, C. (2014). Spontaneous fluctuations in neural responses to heartbeats predict visual detection. Nature Neuroscience, 17, 612–618. https://doi.org/10.1038/nn.3671
- Schandry, R. (1981). Heart beat perception and emotional experience. Psychophysiology, 18(4), 483–488. https://doi.org/10.1111/j.1469-8986.1981.tb02486.x
- Seth, A. K. (2013). Interoceptive inference, emotion, and the embodied self. Trends in Cognitive Sciences, 17(11), 565–573. https://doi.org/10.1016/j.tics.2013.09.007
- Seth, A. K., & Friston, K. J. (2016). Active interoceptive inference and the emotional brain. Philosophical Transactions of the Royal Society B: Biological Sciences, 371(1708), 20160007. https://doi.org/10.1098/rstb.2016.0007
- Tan, G., Yuan, M., He, W., & Zhang, W. (2026). The association between depressive symptoms and cardiac interoceptive accuracy: A systematic review and meta-analysis. Journal of Psychiatric Research, 198, 111–122. https://doi.org/10.1016/j.jpsychires.2026.03.034
- Tsakiris, M., Tajadura-Jiménez, A., & Costantini, M. (2011). Just a heartbeat away from one's body: Interoceptive sensitivity predicts malleability of body-representations. Proceedings of the Royal Society B: Biological Sciences, 278(1717), 2470–2476. https://doi.org/10.1098/rspb.2010.2547
- Zamariola, G., Maurage, P., Luminet, O., & Corneille, O. (2018). Interoceptive accuracy scores from the heartbeat counting task are problematic: Evidence from simple bivariate correlations. Biological Psychology, 137, 12–17. https://doi.org/10.1016/j.biopsycho.2018.06.006