Embodied selfhood

The claim, long-standing (James’s “material me”; Damasio, Craig, Metzinger), that mental representations of self are ultimately grounded in representations of the body, with the internal physiological milieu as the primary reference. Selfhood is a “constellation concept” operating on multiple levels: physiological homeostasis, bodily integrity/morphology/position, first-person perspective, intention/agency, and the metacognitive/narrative “I.”

Seth’s predictive account

In Seth (2013), embodied selfhood is unified under predictive-coding: the organism maintains generative models of the signals “most likely to be me” across interoceptive and exteroceptive domains (Apps & Tsakiris, “the free-energy self”). Self-related predictive coding simultaneously engages all levels of self-representation. Emotions (via interoceptive-inference) and the experience-of-body-ownership are two faces of the same self-modeling process.

Craig’s anatomical grounding

The “material me” is not just metaphor for Craig: Craig (2002) locates it in the homeostatic afferent pathway and its re-representation in the right anterior insula, and Craig (2009) pushes it to a full model of the sentient self (the global-emotional-moment), with the mirror test and von-economo-neurons as phylogenetic markers of self-awareness. This is the anatomical bedrock Seth’s predictive account reinterprets rather than replaces.

Anatomy and pathology

The anterior insular cortex (right-lateralized) is emphasized as supporting conscious access to interoceptive state and its relation to the biological self (see ad-craig, insular-cortex). Disorders of selfhood — depersonalization/derealization, Cotard’s, alexithymia, and self-disturbances in psychosis — are framed (by Seth) as disrupted or imprecise interoceptive inference, and (by Craig) as damage to the insular re-representation (e.g., VEN loss in frontotemporal dementia), giving the concept clinical traction from both directions.

How the bodily self gets built (Oldroyd et al. 2019)

Craig, Seth and Farb et al. all describe the embodied self as an achieved structure; none says how it is assembled. Oldroyd et al. (2019), following Fotopoulou & Tsakiris (2017) and Stern (1985), argue it is assembled socially. The argument turns on infant helplessness: an agent normally resolves ambiguity about its own interoceptive causes by acting and observing the result, but an infant has no such actions available. The caregiver’s response supplies what the infant cannot — the fussy infant learns hunger only once a nipple is offered and eating resolves the discomfort.

Their extension of attachment theory is precise: Stern (1985) and Fonagy (2001) held that for a child to know their own mind they must see it reflected in a sensitive caregiver; Oldroyd et al. contend the same for the body. Insensitive caregiving — slow, intermittent, or rejecting of distress — is thus proposed to impair the child’s ability to form accurate representations of bodily sensation, yielding the distrust-of-body profile seen in avoidant attachment. See social-origins-of-interoception, and social-vs-biological-origins-of-interoception for how far this claim can be pushed before it collides with Craig’s anatomical grounding above.

The historical shape of the field (Quigley et al. 2021)

A framing that organizes what this page already holds. Quigley et al. (2021), in Tsakiris’s section, point out that early work on body awareness studied the body from the outside — mirror self-recognition as the “hallmark of self-awareness,” and the RHI as visuo-tactile integration. On that evidence self-awareness looked “highly malleable and subject to the perception of the body from the outside.”

The interoceptive turn is the claim that this picture was incomplete because it omitted the viscera: from the somatic-marker hypothesis and the insular focus through to predictive-coding and computational models, “current views on the bodily basis of selfhood instead consider the visceral body and its cortical representation to be a core element of the self.”

So the wiki’s two anchors are the two halves of one history: the exteroceptive route (mirror test — which Craig also uses, as a phylogenetic marker) and the interoceptive route (Craig’s material me, Seth’s generative model). And Tsakiris, who arrived from the exteroceptive side, is the author of the result that joined them — lower interoceptive-sensitivity → stronger illusion, read as a precision competition between inside and outside evidence. See multisensory-integration, experience-of-body-ownership.

One phenomenological note worth carrying to interoceptive-taxonomy: interoceptive signals are “typically in the background of phenomenal awareness” — one does not attend to one’s gut all day — but can be “more or less foregrounded in attentional awareness.” That is attention, not accuracy, and keeping the two apart is exactly what the taxonomy pages exist to do.

How much of the self actually shows up unbidden (Banellis et al. 2026)

Every account on this page is a claim about structure — the body grounds the self, the material me is the reference. None of them says how much of a person’s ordinary mental life is in fact occupied by the body. Banellis et al. (2026) measure it, in the closest thing to an unconstrained condition an fMRI study allows: 536 people, 14 minutes of rest, then asked what they had been thinking about.

The answer is a corrective in both directions.

Less than the theory implies. Body content is less prevalent than cognitive, affective and descriptive content (median 36.3 vs 47). If the visceral body were continuously constitutive of the felt self in the way the strong readings of Craig and Damasio suggest, one might expect it to be nearer the surface of the resting stream than it is. This is the empirical form of the point Quigley et al. make phenomenologically and this page already records: interoceptive signals are “typically in the background of phenomenal awareness.”

More than a background hum, and structured. But when it does surface it is not noise. Body content forms a stable, reproducible interindividual dimension — body-wandering — with a coherent affective signature (negative), a physiological one (high arousal), and a neural one (thalamo-somatomotor, bridging the unimodal-to-transmodal gradient). Some people’s resting selfhood is substantially more visceral than others’, reliably so.

And it trades against the social self. The factor analysis puts body content and social-cognitive content (self, other) at opposite poles of one axis. That is worth holding against the social-origins material above: Oldroyd et al. argue the bodily self is assembled socially, while Banellis et al. find that in the adult resting mind, bodily and socially-oriented thought compete. Not a contradiction — developmental origin and adult dispositional trade-off are different claims — but a tension worth watching if either is pushed harder.

Held with the page’s usual caution and one specific to the source: this is self-report about a 14-minute period spent motionless in a scanner, which is neither ordinary experience nor an ordinary body.

Two selves, one built from the body (Farb et al. 2007)

The accounts above treat embodied selfhood as one structure with the body at its base. Farb et al. (2007) split it in two along the axis of time, and only one half is bodily. Their narrative-vs-experiential dissociation maps James’s “me” and “I” onto distinct networks: the self extended across time onto cortical midline (mPFC) elaboration, and the self in the present moment onto a right-lateralized viscerosomatic network — insula, SII, inferior parietal lobule. The momentary self, on this evidence, is the one grounded in the body; the narrative self is a linguistic-conceptual construction layered over it.

This is the empirical form of the “material me”: the present-centred “I” is the interoceptive/exteroceptive body, which the authors place, following Damasio (1999) and Panksepp, as evolutionarily and developmentally prior to the narrative “me.” And it comes with a mechanism for how the two selves normally fuse — the right insula’s default coupling to the vmPFC (R = 0.61 in novices), which mindfulness training decouples. Embodied selfhood, then, is not a single thing that is more or less bodily; it is (at least) two representations that are habitually integrated and can come apart.

The embodied self as relapse protection (Wu et al. 2026)

The 2007 split gives the momentary “I” a bodily grounding; Wu et al. (2026) give that grounding a clinical stake. In remitted depressed patients performing a self-referential encoding task, self-affirmation recruited posterior somatosensory regions (postcentral, supramarginal, precuneus, superior parietal) while self-criticism recruited prefrontal/orbitofrontal elaboration — and it was the somatosensory pole that mattered prognostically: greater somatosensory deactivation forecast relapse, and treatment-related recovery of somatosensory (supramarginal) engagement protected against it, dominating the combined survival model over prefrontal markers. The authors’ conclusion is this page’s concept in a clinical register: relapse vulnerability reflects “insufficient recruitment of somatosensory systems supporting embodied self-experience.” The caution the whole Farb arc carries applies — the protective pole is somatosensory rather than interoceptive-insular proper, so “embodied self-experience” here is grounded in bodily/sensory representation broadly, with interoception the generalization rather than the measured signal. See narrative-and-experiential-self-reference, two-factor-model-of-relapse-vulnerability.

The self as a cardiac rhythm, not just an insular re-representation (Engelen et al. 2023)

Every account above grounds the self in a representation — Craig’s insular re-representation, Seth’s generative model, Farb’s viscerosomatic network. Engelen, Solcà & Tallon-Baudry (2023) add a claim about timing: the heartbeat-evoked response — the cortical response to each beat — indexes the subject of experience beat by beat. HERs track self vs other in mental imagery, scale with body ownership during bodily illusions, stabilize preference-based decisions, and predict conscious detection of near-threshold stimuli and residual consciousness in post-comatose patients. The self, on this view, is not only where (right anterior insula) but when (locked to the cardiac cycle).

The load-bearing clinical fact is a dissociation this page can use: depersonalization/derealization patients — the paradigm disorder of a detached bodily self — do not show the normal modulation of the HER by interoceptive attention. That is a neural correlate of the felt self coming apart, and it sits naturally beside Seth’s reading of the same disorders as imprecise interoceptive-inference and Craig’s as damaged insular re-representation. Held with the caution the HEP page insists on: the HER indexes attention, arousal, prediction error and the self at once, so “the HER is the self” overstates what a badly underdetermined measure can carry.

A different “embodiment”: motor simulation is not this construct (Moffat & Cross 2024)

Every account on this page grounds the self in the felt/regulated body — interoceptive and (in the Farb arc) somatosensory representation. That is not the only thing “embodiment” names in cognitive neuroscience, and the two are worth holding apart to avoid a silent conflation. Moffat & Cross (2024) use the motor-simulation sense from the neuroaesthetics tradition (Freedberg & Gallese; Calvo-Merino; Kirsch & Cross): you are “embodied” with respect to a movement when your own motor system (the action-observation network) has produced it, which then shapes how you perceive and enjoy watching it. Their fNIRS result — awareness of having mirrored a movement enhances enjoyment, engaging sensorimotor/action-observation cortex (STG, IFG, IPL) — is about the body one moves, not the body one feels from the inside.

The two share a word and a broad anti-Cartesian commitment and little else load-bearing. The one genuine point of contact is the somatosensory pole: Wu et al.’s relapse-protective “embodied self-experience” is grounded in somatosensory (not interoceptive-insular) cortex, and Moffat & Cross’s aesthetic effects are sensorimotor too — so both are “embodiment” in a bodily-sensory register, with interoception the generalisation rather than the measured signal. This is exactly the caution the Wu and Farb-2007 sections above already flag, arriving here from the motor side.

The embodied-cognition reading, and pathology as a translation deficit (Herbert & Pollatos 2012)

Most accounts on this page ground the self in a representation (Craig’s insular re-representation, Seth’s generative model) or a timing (Engelen’s HER). Herbert & Pollatos (2012) ground it in embodied cognition: on the perceptual-symbols view (Barsalou; Niedenthal 2007), higher cognition operates on reactivations of the sensory-motor states of experience, and interoceptive representations are exactly such states for the internal body — so the felt self, and individual differences in it, are the visceral case of embodiment. It is the wiki’s most explicit statement of that bridge, joining the interoception–emotion literature to embodied-cognition theory rather than to the Jamesian or Bayesian framings the other sections use. As theory it adds a vocabulary, not a mechanism, and it stays on the Craig/Damasio read-out side of feedforward-vs-predictive-interoception (reactivation, not inference).

Its more reusable move is clinical. Following Fuchs & Schlimme’s Leib/Körper (lived body / object body) distinction, Herbert & Pollatos classify disturbances of embodiment by which body is affected, and read two as translation deficits — a body that responds but whose signals do not reach conscious feeling: alexithymia (a difficulty translating bodily signals into felt emotion, with high alexithymia ↔ low heartbeat-perception IA in their own data) on the lived-body side, and eating disorders on the object-body side (disturbed body image, plus reduced perception of hunger/satiety/cardiac signals in anorexia). This is the framing the wiki’s later interoceptive-psychopathology sources (quadt-2018-interoception-health-disease, bonaz-2021-diseases-of-interoception) develop; recorded here because “the self as what translates body into feeling, and its disorders as failed translation” is a clean statement of this page’s concept read through its pathologies.

The material me when it goes wrong (Wiebking et al. 2010)

The accounts above describe the material me as an achieved, healthy structure — Craig’s anatomical bedrock, Seth’s generative model, Farb’s viscerosomatic network. Wiebking et al. (2010), from Northoff’s programme, is the wiki’s first study of the material me as a clinical variable — what the bodily self looks like in depression, whose somato-vegetative symptoms (bodily preoccupation, altered appetite, fatigue) the affective-cognitive account of depression leaves aside. It borrows the “material me” term from James via Craig but asks the question Craig does not: how the insula’s bodily self is disordered.

Two dissociations do the work, and both cut against a naive “the depressed body-self is impaired” reading. First, on the BPQ the depressed body is more present in report — higher perceived autonomic reactivity, stress response and stress style — while the awareness subscale is flat. The material me is more preoccupying, not more accurately sensed; this is the sensibility/awareness split (interoceptive-taxonomy) arriving in a clinical self. Second, and more striking, the healthy correlation between right-anterior-insula rest activity and body-perception scores is absent in patients — the felt body and its cortical substrate decouple. So depression’s material me is not a self that feels its body too little or too much; it is a self whose body-report has come unhooked from the interoceptive cortex that normally grounds it, while an under-deactivating insula at rest tracks how severe the illness is. Held with the source’s cautions (all patients medicated, no true resting state, and “material me” a frame rather than a measured self-relatedness). It sits beside the Farb (2010) result — where keeping the right insula online under sadness went with less depression — as the wiki’s two first-hand data points on the insular self in mood disorder, reached from opposite directions (Farb: engagement protective; Wiebking: rest-decoupling pathological).

Two complementary senses (Farb et al. 2015)

Farb et al. (2015), building on Seth, Suzuki & Critchley (2011), argue the embodied self is “more fully realized” through awareness of ongoing interoceptive interaction via two complementary senses — see presence-and-agency: presence (connection to the moment) and agency (ability to effect change). Both are proposed as downstream markers of successful prediction-error minimization, giving embodied selfhood a testable behavioral signature beyond neuroanatomy alone.