Interoceptive training in clinical practice

Four clinical vignettes assembled in Farb et al. (2015) as examples of dysfunctional interoceptive integration and its remediation — presented together because they share the common thread of restoring tolerated access to interoceptive signals under skillful guidance, rather than avoidance or catastrophic reactivity.

The frame this page sits inside. The Khalsa et al. (2018) roadmap gives the disorders treated here a common name — interoceptive psychopathology, a transdiagnostic dimension of dysfunction — and a common warning that maps onto this page’s whole argument: the roadmap’s own survey of treatments concedes that interoceptive interventions have thin trial support (interoceptive exposure “often fails to reproduce the fear response”; “none of these approaches has translated into clinical practice”), and that whether interoceptive dysfunction is a cause or consequence of a disorder is unresolved. That is the deficit-to-treatment problem below, stated at field level. The roadmap also formally lists MBSR, yoga and meditation/movement treatments among current interoceptive therapies (citing Farb et al. 2015), ratifying the contemplative rationale this page runs on.

Interpersonal/sexual trauma

Women with a history of interpersonal sexual violence, treated with Mindful Awareness in Body-oriented Therapy (MABT; Price 2005, 2007), learn to engage perceptual-inference: recognizing when sensory cues of dissociation are emotionally triggered, and maintaining awareness of the body rather than dissociating into habitual active-regulatory avoidance. Outcomes reported: bodies experienced as informative resources rather than threat, greater safety, improved intimate engagement without dissociation, and wholeness/empowerment. It remains undetermined whether this reflects a direct increase in interoceptive accuracy or other taxonomy elements (sensitivity, regulation) — see interoceptive-taxonomy.

Chronic pain

The standard approach (cognitive reframing + attentional distraction) is questioned: chronic pain pathology may involve fear-conditioned interoceptive avoidance that distraction reinforces rather than challenges (Zaman et al. 2015). interoceptive-exposure is offered as an alternative, alongside yoga (associated with decreased prefrontal activity and improved pain tolerance; Villemure et al. 2013).

Substance use disorder

Women in substance-use treatment taught to connect physiological and emotional distress report increased capacity to attend to and negotiate emotional stress, facilitating sobriety (Bowen et al. 2007; Price et al. 2012b) — interoceptive awareness reframed as a self-care resource rather than a trigger to be numbed.

Upgraded from vignette to trial with the Weng et al. (2021) ingest. This case and the trauma case above are both MABT, and Cynthia Price co-authors that review: two studies of MABT as an adjunct to women’s SUD treatment now report longitudinal reductions in substance use, craving and emotion dysregulation versus control, with self-reported interoceptive awareness and respiratory sinus arrhythmia improved and maintained at 12-month follow-up. That makes this the best-evidenced case on the page — and the only one where a psychological interoceptive training shows a durable physiological change. See mabt, cynthia-price. The caveats are recorded there: the interoceptive outcome is self-report rather than accuracy, mediation is untested, and the wiki holds the trials only through their own author’s review section.

Anorexia nervosa — an explicit caution

Unlike the other three cases, anorexia is flagged as a case where naive application of interoceptive training can backfire: patients often hold powerful associations between bodily-state awareness (linked to starvation) and a “doing” mode focused on controlling eating, shape, and weight. Bringing awareness to the body is initially met with an increase, not decrease, in maladaptive behavior. The paper is explicit that skillful, condition-specific guidance is necessary before advocating body-awareness interventions in this population — a direct counter-example to any assumption that “more interoceptive awareness is always better.”

Major depression — a fifth case, and a boundary dispute

Lyons et al. (2021) add depression to this page and, in doing so, walk straight into the caution above.

Their argument: depressed patients report weaker felt bodily emotion on emBODY, therefore “increasing body awareness might be an important target in the treatment of patients with major depression,” supported via MBCT (Segal et al. 2018), emotion-focused therapy (Greenberg 2004), or exercise (Mead et al. 2008).

The boundary problem. Farb et al. (2015) name severely depressed and suicidal patients as a population where enhancing interoceptive awareness without a regulatory framework can be maladaptive. Lyons et al. excluded current suicidality — along with somatoform disorder, chronic pain, and chronic physical disease. So their sample is precisely the population Farb et al.’s warning does not cover, and their data are silent on the patients the warning is about. The two papers are not in conflict; they are describing adjacent, non-overlapping populations, and the recommendation transfers from one to the other only if the exclusion criteria are doing no work. That is exactly the assumption Farb et al. deny.

What the evidence actually supports. Weaker on inspection than the recommendation implies, for two reasons:

1. The group difference is fragile. The “less activation in depression” result rests almost entirely on one fear cell, in the emotion the authors report they failed to induce, and disappears when fear is excluded — see the arithmetic note on lyons-2021-body-maps-depression. The better-supported finding is localised reduction in sadness (stomach, chest) from the difference maps.
2. The inference is deficit-to-treatment, which this wiki has already watched fail. “Depressed patients have less X, therefore raise X” is the reasoning anorexia falsifies one section above: those patients also have a disordered relationship to bodily awareness, and raising it directly makes things worse before better. A cross-sectional deficit does not license a treatment target, because it does not say whether the deficit is the illness, a defence against it, or a consequence of it.

What survives, and is genuinely new. The finding that depressed patients’ maps remain fully differentiated — as accurate as controls, and readable by a classifier trained on controls. If body-awareness work in depression helps, this is evidence it will not be by teaching patients to tell their emotions apart in the body, because that capacity is intact. Something else is wrong. That is a more useful constraint on mechanism than the deficit claim it sits next to.

What the “something else” might be — from inside this wiki’s own author. Farb et al. (2011) followed remitted depressed patients for 18 months and found the relapse predictor was not how much they felt but how they processed it: elaborative, self-referential (mPFC) reactivity to a sad film predicted relapse and tracked rumination, while sensory reactivity protected and tracked acceptance — with felt sadness predicting nothing. That reframes the depression case on this page. The target of a body-awareness intervention in depression is plausibly not raising felt-body intensity (Lyons shows differentiation is intact, and Farb shows intensity is not the predictor) but shifting the mode — from ruminative elaboration toward decentered, perceptual observation. This is precisely the regulation construct (relationship to signal), not the intensity construct, and it is the mechanism MBCT (Segal) was built to move. The honest limit stays visible: Farb’s protective sensory pole was visual, not interoceptive, so this is the rationale for a body-focused MBCT-style approach, not evidence that interoceptive attention specifically is the active ingredient — see cognitive-reactivity.

A sobering update from the RCT sequel. Farb et al. (2022) scanned 85 remitted patients before and after prophylactic therapy and split relapse vulnerability into a two-factor model: a dynamic prefrontal-reactivity factor that therapy did reduce, and a static sensory-deactivation factor that therapy did not move at all — unchanged by either MBCT or Well-Being CT. For this page that cuts two ways. Against: the sensory/bodily deactivation these interventions would most want to reverse is exactly the part current therapies left untouched, so a body-awareness rationale that promises to restore sensory integration is, so far, unproven at the level of the biomarker. For: the prefrontal factor that did respond is the rumination target, and it responded equally to a therapy that trains body awareness (MBCT) and one that does not (WB-CT) — so whatever helps depressed patients here is trans-therapeutic, not specific to interoceptive training. Both readings argue for humility about “raise body awareness” as the mechanism.

See is-more-interoceptive-awareness-better, where this disagreement is tracked as a live debate rather than resolved here.

Panic disorder — a sixth case, and the wiki’s only stratification hypothesis

Added with the Van der Does et al. (2000) ingest. It is unlike the five cases above in that the proposal comes with a rule for who gets the treatment, and the rule is not a diagnosis.

The problem. Ehlers (1995): good heartbeat perception predicts poor treatment outcome and recurrence of panic after remission. Prospective, one-year. On the face of it that is a flat contraindication for raising interoceptive contact in panic — and this page’s whole mechanism is raising interoceptive contact.

The proposed dissolution, in two moves. First, the predictor may not be perception. Across 709 pooled participants, accurate heartbeat perceivers differ from inaccurate ones on anxiety-sensitivity and on nothing else — not trait anxiety, not state anxiety, not somatosensory amplification. So “good perceivers do worse” may be “patients who believe bodily sensations are dangerous do worse,” which is a belief this page’s interventions are meant to target rather than a reason not to run them.

Second, and more interesting: the patients likely to benefit are identifiable in advance. Those whose panic attacks come with no real heart-rate increase but who report one are reading a prior rather than a heart. Van der Does et al. argue these patients “may be helped by interoceptive training or by being informed that what they feel is not their heart” — the training reconnects report to physiology, or discredits the schema.

Why this is worth more than its evidence. There is no evidence: the authors label it counter-intuitive and speculative, and no trial in this wiki tests it. What it supplies is a shape the rest of this page lacks. Every other case here reasons deficit-to-treatment — patients have less X, therefore raise X — which the anorexia section falsifies and the depression section is criticized for. This one reasons mismatch-to-treatment: the intervention is indicated not by how much interoceptive contact a patient has but by whether their report and their physiology have come apart, and in which direction. That is a criterion you can measure before treating.

And it names its own prerequisite. Identifying the subgroup needs ambulatory HR monitoring, precisely because the patient’s report is the thing under suspicion. A stratification rule whose stratifying variable is unavailable in ordinary practice is not yet a clinical proposal — but it is the only candidate on this page that would tell a clinician who not to treat, which is what Farb et al.’s cautions keep asking for and never operationalize. See cognitive-model-of-panic, interoceptive-exposure, is-more-interoceptive-awareness-better.

The proposal is older than the reanalysis, and pointed the other way. Ehlers (1993) — now the first-hand source for this whole panic thread — had already proposed interoceptive training, but for a different subgroup and a different reason. Her rationale: panic self-control programmes (paced breathing, applied relaxation, vagal-tone stimulation) require the patient to act “when they feel the first signs of a panic attack,” and “the use of these coping techniques requires accurate interoception,” so “patients who do not show adequate awareness of their physiological state may require interoceptive training before using self-control strategies.” That targets under-perceivers (train them up so coping can be deployed in time), whereas Van der Does et al.’s later stratification targets schema-guided over-reporters (train them to reconnect report to physiology, or discredit the schema). Same intervention, opposite indication — a real, if unresolved, disagreement about who interoceptive training in panic is for, now sourced first-hand from both sides. Note Ehlers’s own closing caution that the group differences her programme found “tended to be small compared to the error variance,” so even her prerequisite-for-coping rationale is offered tentatively.

Trauma — a named therapy, and the one case here with no evidence at all

The interpersonal-trauma case above (MABT) sits inside this page’s shared logic — restore tolerated interoceptive access under guidance. The Payne et al. (2015) ingest adds a second, more sharply theorized trauma approach — Somatic Experiencing (SE) — and it is worth a separate entry for two reasons.

It operationalizes the “skillful guidance” this page keeps invoking. Every case above ends at “under skilled, condition-specific guidance” without saying what the guidance is. SE names it: establish embodied safety via positive somatic markers first (resourcing), approach the trauma “drop by drop” (titration), oscillate between activation and discharge (pendulation), and complete the frozen defensive movement (biological-completion). Whether or not SE’s theory is right, this is the most explicit protocol for the sequenced, safety-first interoceptive contact the rest of the page gestures at — and it answers the page’s recurring worry (anorexia, severe depression) about raising bodily awareness too fast, by making titration the whole method.

But it is the weakest-evidenced entry on the page, and that should not be blurred by its theoretical richness. The other cases here cite feasibility studies and group comparisons. SE’s originating paper reports no peer-reviewed studies of the therapy and offers a composite (invented) case, and its authors have a financial stake in it. So SE enters this page as the best-specified and least-evidenced of the clinical applications — a protocol worth knowing precisely because it says what “skilled guidance” means, held with the caveat that no one has shown it works. See somatic-experiencing, interoceptive-exposure (which SE defines itself against), and is-more-interoceptive-awareness-better.

Where these presentations may come from

Farb et al. (2015) describe what to do about maladaptive interoceptive profiles; Oldroyd et al. (2019) propose where they originate. The avoidant profile they describe — attention diverted from bodily cues, low trust in the body as a source of reliable signal, and self-reported calm alongside elevated physiological arousal — is close to the experiential-avoidance presentation this page treats, and their Study 2 finding gives it a developmental aetiology: youth whose mothers reported rejecting their negative emotions showed lower correspondence between reported distress and sympathetic arousal (self-report-physiology-congruence).

Two implications, both tentative:

• Interoceptive training may sometimes be remedial rather than additive. If a patient’s relationship to their body was shaped by a caregiver who denied their bodily reports, then “learning to feel the body” is partly unlearning an acquired distrust — closer to relational repair than to skill acquisition. This is consistent with, and may partly explain, the cautions above about enhancing awareness without a regulatory framework.
• Profile, not deficit, should guide the intervention. The anxious and avoidant profiles are opposite: hypervigilant worried monitoring vs. disengaged distrust. Both would score as “impaired interoception” on a single composite, but they plausibly need different treatment — an argument for assessing with the interoceptive-taxonomy or a multidimensional instrument like the maia rather than a single score. The maia Trusting subscale in particular may be the one that tracks what needs repair.

Caveat: Oldroyd et al. studied non-clinical samples with small cross-sectional effects, and no source in this wiki tests whether attachment history moderates response to interoceptive training. The link above is a hypothesis worth holding, not an evidence base — see social-origins-of-interoception.