Interoception

Following Craig (2002), interoception is “the sense of the physiological condition of the (entire) body” — Craig’s crucial widening from the narrow “visceral sense” of Sherrington’s taxonomy to include temperature, pain, itch, sensual touch, muscular/visceral sensation, hunger, thirst, and air hunger. On this view interoception is the afferent side of homeostasis, carried by the lamina-i-spinothalamocortical-pathway and represented in the insular-cortex (a posterior→anterior gradient from objective primary mapping to subjective re-representation). See ad-craig.

The field’s own consensus definition, from the Khalsa et al. (2018) roadmap, keeps Craig’s breadth and adds the processing verbs: interoception is “the process by which the nervous system senses, interprets, and integrates signals originating from within the body, providing a moment-by-moment mapping of the body’s internal landscape across conscious and unconscious levels.” Two commitments in that sentence matter. First, interoception spans all homeostatic systems — cardiovascular, pulmonary, gastrointestinal, genitourinary, nociceptive, chemosensory, osmotic, thermoregulatory, immune, autonomic — not just the viscera Sherrington meant. Second, most of it is not conscious: the roadmap presses the distinction between sensation (the raw signal from bodily sensors) and perception (what the system makes of it), and insists the term “interoceptive awareness” has been stretched too thin to be useful. That taxonomy problem gets its own treatment on interoceptive-taxonomy; the transdiagnostic clinical face of interoception is on interoceptive-psychopathology.

A third definition, useful for its explicit level structure, comes from Quadt, Critchley & Garfinkel (2018), who cast interoception as an umbrella term spanning four things at once: (1) afferent signalling to the brain over distinct neural and humoral (including immune and endocrine) channels; (2) the neural encoding, representation, and integration of that information; (3) its influence on other perceptions, cognitions and behaviours (which need not be conscious); and (4) the psychological expression of these representations as consciously accessible sensations and feelings. The addition the wiki’s mostly-neural, mostly-cardiac material underweights is the humoral/immune channel in (1): circulating cytokines and hormones reach the brain through the circumventricular organs and the NTS, which is how the immune system enters interoception and drives sickness behaviours. Interoception, on this view, is not one process but a stack from blood-borne signal to felt symptom.

A fourth framing, and the one that stretches the term furthest, comes from Bonaz et al. (2021). Where the definitions above are about what interoception is, theirs is about how far the system extends: interoception is one bidirectional hierarchy running from receptor to belief, with an afferent limb (representation) and an efferent limb (the central-autonomic-network, which controls the body it senses), and “perturbed interoceptive representation, integration, and predictive control can originate at any level of the neuraxis.” Homeostatic reflexes are not autonomous loops the brain occasionally overrides; their set-points are continuously written from above by allostatic policy. Two consequences the wiki’s other definitions do not force: interoception becomes a subject in internal medicine as much as in psychology (see disorders-of-brain-gut-interaction, chronic-pain, joint-hypermobility), and its signals include some that do not originate in the organism at all (microbiota-gut-brain-axis).

A fifth framing asks neither what interoception is nor how far it extends, but what it is for. Quigley et al. (2021) adopt the roadmap’s definition verbatim and then organize the field by function: energy regulation (intake and waste removal), memory, affective and emotional experience, and the psychological sense of self — with all of it derived from “what is every organism’s ultimate concern; namely, the efficient regulation of energy to support life.” Two additions the wiki had nothing on: excretion as an interoceptive function (urinary-interoception) and memory as a non-affective one (gut-hippocampal-memory). And one structural claim that changes how the term relates to the rest of perception: interoception is a party to multisensory-integration, not a channel that terminates and then influences things — integration occurs across interoceptive modalities and across interoceptive-exteroceptive pairs, which means the felt world is assembled from both at once. affective-touch is where the dichotomy stops being clean altogether, and note that Craig’s 2002 widening had already placed sensual touch inside interoception on independent grounds.

A sixth framing is the only one that is not a definition at all but a process diagram. Berntson & Khalsa (2021) adopt the roadmap’s wording in their glossary and then lay interoception out as a five-stage continuum:

sensors → pathways → networks → circuits → awareness

— transducers (mechano-, thermo-, chemo-, osmo-, gluco-, humoral receptors, free nerve endings); pathways (vagal, cranial, sacral, spinothalamic, somatosensory); networks (autonomic, enteric, thalamocortical, hypothalamic, limbic, sensorimotor, salience, default); circuits (appetitive, affective, arousal, thermal, nociceptive, cognitive, social, threat); and awareness (detection, attention, insight, magnitude, discrimination, accuracy, sensibility).

Two corrections to how this wiki has been reading the word. First, its coverage begins at pathways: the transduction stage had no page until interoceptive-sensors, and several of those transducers were characterized after 2015, so the afferent inventory is not the settled classical anatomy the theoretical pages assume. Second, awareness is the last stage and most of the process never reaches it — humans “rarely have high-resolution access to what is happening in the body,” conscious interoception arises mainly on perturbation or threat, and some organs (liver, kidney) are never felt at all. Nearly every measure in this wiki operates on that thin conscious tip. The reciprocity claim attached to the diagram is Berntson’s own: afferent interoception cannot be described apart from the efferent systems that alter what is being sensed, so the unit is a circuit, not a channel (see central-autonomic-network).

A seventh framing is the only one that changes what the word denotes rather than how it is carved up. Chen et al. (2021) — the NIH Blueprint framing paper for the same special issue — propose that interoception is “the processes by which an organism senses, interprets, integrates, and regulates signals from within itself,” with regulating meaning descending communication from brain to body. Every other definition on this page is afferent; this one folds the efferent limb inside, and names its parts: regulatory signals of interoception, central regulators, and interoceptive effectors (the target organs, whose altered state changes what is next sensed).

Three things to keep with it. First, the authors know they are legislating: they say plainly that this “significantly expands the traditional scope” and “may differ from descriptions used in much of the current literature.” Second, the justification is pragmatic, not theoretical — including body regulation “enables the development of more innovative methods to probe the functional impact of interoceptive processes.” It is a definition chosen for what it makes measurable. Third, they concede the cost: because a regulatory signal’s impact is read off the target organ’s response, “it is sometimes impossible to distinguish regulatory interoceptive signals from body regulation.” The expansion buys reach and spends precision.

The same paper supplies the thing the definitions above all lack — a criterion for where interoception stops. Not the skin, but whether signals and responses represent, rather than originate from, the internal world. On that test proprioception and the vestibular system come inside and taste stays mixed. It is one of three incompatible-in-emphasis criteria proposed in this one journal issue; they are held together at interoception-exteroception-boundary rather than merged.

An eighth framing is the only one that reads the seven above as a symptom. Desmedt, Luminet, Maurage & Corneille (2023) argue that the field’s habitual move — defining interoception both as the processing of internal signals (phenomenon-based) and as the activation of homeostatic pathways (physiological) — is internally contradictory, because the same internal states are also processed by non-homeostatic Aβ somatosensory fibres. Since no physiological pathway is necessary and sufficient, they conclude physiology should not define interoception at all, and offer a deliberately phenomenon-based replacement: interoception is “the top-down and bottom-up processes by which an organism senses, interprets, and integrates signals from within itself and below the skin, across conscious and non-conscious levels.” Three things distinguish it from the roadmap definition it revises: it makes the skin the barrier and excludes the classical exteroceptive senses (hearing one’s heartbeat, seeing one’s chest move) to stop overlap; it builds in predictive processing by naming top-down and bottom-up limbs explicitly; and it excludes efferent regulation, narrowing Chen et al.’s “regulates” back out. It is the wiki’s clearest statement that the proliferation of definitions on this page is not richness but a construct-validity problem — see interoception-exteroception-boundary and interoceptive-taxonomy.

Interoception is the common substrate underlying every paper in this wiki:

• For 2009), interoception is the homeostatic afferent system whose cortical re-representation in the anterior insula is the foundation of feeling, emotion, and self-awareness — the anatomical bedrock the rest of the field builds on.
• For Seth (2013), interoception is the domain to which predictive coding is extended — emotions become interoceptive-inference.
• For Friedman (2010), the visceral/autonomic feedback that James held to be the emotion is the interoceptive channel, and the question of whether emotions have distinct visceral signatures is the autonomic-specificity debate.
• For the classic theories, James–Lange makes interoceptive (visceral) feedback causally primary; Cannon–Bard denies it sufficient resolution; Schachter–Singer treats interoceptive arousal as undifferentiated raw material awaiting cognitive labeling.
• For Oldroyd et al. (2019), interoception is something that develops — and develops socially, out of a caregiver’s responses to the infant’s bodily states. See social-origins-of-interoception.
• For Volynets et al. (2020), interoception is what makes emotion feel like something somewhere — the felt body has a topography, and it is emotion-specific and much the same worldwide. See bodily-sensation-maps, embody.
• For Payne et al. (2015), interoception (with proprioception) is a therapeutic lever — the bottom-up route into a dysregulated subcortical network, used to treat trauma without going through cognition or memory. The wiki’s first source to treat interoceptive attention as the active ingredient of a named therapy. See somatic-experiencing, core-response-network.

Key open question in this cluster

Is interoception feed-forward (bottom-up accumulation of visceral evidence, the classical assumption) or predictive/inferential (top-down generative models of visceral causes)? This is the crux of feedforward-vs-predictive-interoception. It is measured behaviorally via interoceptive-sensitivity (e.g., the heartbeat-detection-task).

A second axis: where does it come from?

Every source above treats interoceptive capacity synchronically — as anatomy (Craig), as a precision parameter (Seth), as a trainable skill (Farb et al.). None asks how it is built. Oldroyd et al. (2019) add the developmental axis, proposing that early caregiving shapes interoceptive functioning via hpa-axis regulation, the growth of the interoceptive network, and caregiver mirroring of the infant’s bodily states (attachment-style). The strong form of that claim — that the origins of core feelings such as hunger and warmth are “social not biological” — sits in real tension with Craig’s homeostatic-afferent account of those same feelings: see social-vs-biological-origins-of-interoception.

The developmental axis now has a descending limb as well: Volynets et al. (2020) find felt bodily emotion dampens steadily from 18 to 90, the only background variable to move their maps at all. Paired with Hietanen et al.’s (2015) child data, the arc is sharpen, then fade — see age-related-interoceptive-decline, which also flags an unresolved tension with this wiki’s contemplative material about whether less interoceptive contact is a loss or an aid to regulation.

A fourth axis: it is built into the word

Ferré et al. (2024) add a layer none of the above touches — the lexicon. Asked what makes a word count as naming an emotion, 651 Spanish speakers rating 1,286 words answer, above all else, whether it is about internal body sensations: interoception takes R² = .417 of prototypicality variance on its own, ahead of feeling and ahead of every other component of emotion, while body expression, appraisal and action tendency contribute nothing or contribute negatively. Convergent, independent evidence comes from Connell et al. (2018), who found interoception outranked all five exteroceptive modalities in separating emotion words from nonemotion abstract words.

So interoception is not only the substrate the field’s theories reach for; it is the content ordinary speakers reach for when they decide something is an emotion at all. See emotion-prototypicality.

The catch, and it is the same catch as everywhere else on this page: these are judgments about words, not measurements of bodies. Read one way the folk concept is interoceptive because emotions really are felt in the body; read the other, word-cued reports of felt bodily emotion may be echoing a concept rather than a body — which is precisely the objection facing bodily-sensation-maps. See embody.

A third axis: does it look the same everywhere?

Interoception’s content — not its accuracy, mechanism, or origin, but what the felt body actually contains — turns out to be describable and largely shared: emotion-specific topographies of felt activation and deactivation, concordant across 101 countries. This is the wiki’s first source on interoception as a cross-cultural phenomenon and the first to treat felt bodily structure as data in its own right. It is also the wiki’s clearest case of interoceptive sensibility (report) being measured with no accuracy term at all — see interoceptive-taxonomy and cultural-universality-of-emotion.