Interoceptive exposure
Originating in panic-disorder treatment (Craske et al. 1997) and extended to chronic pain (Boswell et al. 2013; Zaman et al. 2015) in Farb et al. (2015). The technique rests on the idea that some psychosomatic pathology (e.g., chronic pain) involves fear-conditioned interoceptive avoidance — a conditioned withdrawal association that attentional distraction would reinforce rather than challenge, since distraction never disconfirms the feared interoceptive meaning.
The panic case, which this page had been asserting without a source
This page’s frontmatter claims “established efficacy in panic disorder,” and until the Van der Does et al. (2000) ingest the wiki had read nothing on panic at all — the claim came via Farb et al. citing Craske et al. (1997). The gap is still open (no efficacy trial is in raw/), but the theory the treatment comes from now has a page: cognitive-model-of-panic.
The field’s own assessment, which is more sober than this page’s frontmatter
The Khalsa et al. (2018) roadmap — a 24-author consensus statement — supplies the field-level verdict the wiki had been missing, and it downgrades the confident “established efficacy” this page inherited. On the one hand, interoceptive exposure is, they say, “the most common application of interoceptive evaluation in current clinical practice” for panic: patients self-induce symptoms (hyperventilation, jumping jacks, spinning, straw-breathing) while a clinician monitors distress. On the other hand, the roadmap’s candid catalogue of its limits:
- The manipulations often fail to reproduce the fear. “Possibly because the patient retains full control over the stimulation (the patient can quit at any time) and the perturbation remains predictable with minimal uncertainty.” This is a mechanistic critique — exposure may under-work because the two things that make interoceptive threat frightening (uncontrollability, unpredictability) are exactly what the clinical procedure removes. It motivates the roadmap’s interest in Floatation-REST and pharmacological probes, which take the control away.
- Pharmacological interoceptive exposure (CO₂, yohimbine) can reduce anxiety symptoms as monotherapy or augmentation, but “there are few studies of these procedures to date, the impact… on longer term outcomes… are unknown, and none of these approaches has translated into clinical practice.”
- A verification test exists — a standardized behavioral avoidance paradigm (tolerance of a small dark chamber for 10 minutes) — but it is a proxy for treatment success, not an efficacy trial.
So the wiki’s frontmatter claim survives only in weakened form: interoceptive exposure is standard practice and has some trial support, but by the field’s own account it frequently fails to engage the fear it targets, and the strong “established efficacy” phrasing overstates what the consensus is willing to assert. Recorded so the page stops resting on a claim its own supporting literature qualifies.
What the 2000 reanalysis adds is a complication and a proposal.
The complication. Ehlers (1995) found that good heartbeat perception predicts worse treatment outcome and panic recurrence. If interoceptive training raises interoceptive contact, and high contact predicts poor outcome, exposure looks contraindicated in exactly the population it was invented for. (Though see anxiety-sensitivity: accurate perceivers differ from inaccurate ones on the ASI and nothing else, so the predictor may be a catastrophic belief rather than perceptual contact — in which case exposure targets it directly and the paradox dissolves.)
The proposal, which the authors call counter-intuitive and speculative themselves. The patients most likely to benefit are the schema-guided ones: those whose attacks come with no real heart-rate increase but who report one anyway. They are reading a prior, not a heart. For them, interoceptive training — or simply “being informed that what they feel is not their heart” — could reconnect report to physiology or discredit the schema.
Note the clinical prerequisite, which is the interesting part: identifying the subgroup requires ambulatory HR monitoring, because the whole point is that the patient’s own report is uninformative about their heart. This is a stratification hypothesis whose stratifying variable no clinician has by default, and nothing in this wiki tests it.
The therapy that defines itself against this page
Somatic Experiencing (SE) is worth recording here because it is built as the negation of interoceptive exposure, and the contrast sharpens what exposure is committed to.
| interoceptive exposure | Somatic Experiencing | |
|---|---|---|
| contact with the feared sensation | sustained, non-avoidant | titrated — “drop by drop”, explicitly avoiding sustained aversive contact |
| memory of the trauma | approached / evoked | approached indirectly, evocation avoided |
| proposed mechanism | fear extinction (prior updates toward safety) | attractor shift — completing a thwarted defensive movement, said not to be extinction |
| relapse | conditioned fear easily reinstated by cue re-exposure (McNally 2007) | claimed robust to re-evocation |
SE’s argument is that extinction only overlays a competing association on an intact fear memory, so it is fragile — whereas completion changes the network’s dynamics discontinuously. Whether this is a genuine mechanistic difference or the same clinical outcome redescribed is untested — SE offers no comparison, only clinical impression that it works “more quickly and with much less distress.” The wiki keeps both on the table: exposure has efficacy trials it can point to (in panic; see above) and a clean predictive-coding reading below; SE has a sharper mechanism story and no data. See somatic-experiencing, biological-completion.
Note that the “fragile extinction over an intact trace” picture SE derides is, more or less, van der Kolk’s considered position — he holds the subcortical trace indelible and treatment as the restoration of cortical inhibition over it. So exposure/extinction has a theoretical ally here: if the trace really is permanent, an inhibition-strengthening treatment is the honest one, and SE’s claim to resolve rather than inhibit is the load-bearing bet. That bet is the are-traumatic-memories-indelible debate.
The other kind of alternative: treating the pathway instead
Somatic Experiencing defines itself against this method while sharing its premise — that the route into the problem is through attention to sensation. Bonaz et al. (2021) introduce a class of intervention that does not share it. Their translational close puts interoceptive exposure (“where patients learn to accustom themselves to sensations of physiological arousal”) and autonomic biofeedback on one side, and bioelectronic-medicine — VNS, spinal cord stimulation, TMS, tDCS, DBS — on the other, as two ways of targeting brain-body communication.
The contrast is worth keeping because it locates the disagreement precisely. Exposure bets the problem is in the relationship to the signal (a catastrophic prior that non-avoidant contact will update). Bioelectronic medicine bets it is in the signal and its conduction. Bonaz et al.’s own framing licenses both — dysfunction can strike at any level of the neuraxis, so an intervention at any level might be the right one — and offers nothing that would tell a clinician which level a given patient’s problem is on.
That is the same stratification gap this page already records twice: the roadmap’s finding that exposure often fails to reproduce the fear, and Van der Does et al.’s proposal that the treatable subgroup can only be identified by ambulatory HR monitoring the clinician does not have. Three sources, three different reasons the treatment cannot yet be matched to the patient.
Relation to the wiki’s predictive-coding framework
Interoceptive exposure can be read as a structured, clinically supervised form of perceptual-inference: sustained, non-avoidant contact with the feared sensation allows priors (e.g., “this heart-rate sensation means danger”) to update toward the actual, safer outcome, rather than being protected via avoidance (a form of active-inference that entrenches rather than resolves the prediction error). Yoga, which incorporates interoceptive training and is associated with decreased prefrontal activity, is cited as showing similar pain-management benefits (Villemure et al. 2013).