Weng et al. (2021) — Interventions and Manipulations of Interoception
The intervention quadrant of the TICS 2021 special issue. The wiki now holds four of its arms: computational (Petzschner), clinical-nosological (Bonaz), functional (Quigley), and this one — what you can actually do to an interoceptive system.
It is the first source in this wiki that puts contemplative training, a breathing device and an implanted nerve stimulator on the same diagram and asks what they have in common.
The framework, and why it is the contribution
Every interoceptive intervention gets read through four slots: intervention type → target → potential mechanism → clinical outcome. Applied to respiration:
| intervention | target | proposed mechanism | level |
|---|---|---|---|
| VNS / taVNS / RAVANS | vagal afferents to NTS and other brainstem nuclei | NA, 5-HT, ACh projections; autonomic regulation; cortical neuroplasticity | neural |
| Slow breathing | respiratory rate and tidal volume | cardiopulmonary baroreflex modulation; parasympathetic regulation | behavioural |
| Mindfulness approaches | interoceptive signalling; respiratory sensations; reframing interpretation | neuroplasticity of interoceptive focus and representation; autonomic regulation | psychological |
The organizing claim is that these are three entry points into one pathway, which is why their clinical outcome lists overlap so heavily (blood pressure, cardiovascular risk, pain severity, negative affect, stress, anxiety/PTSD, craving and substance use all appear under two or three of them). The authors are explicit that the framework is meant to generalize to gut, cardiovascular and urinary systems, and equally explicit that respiration was chosen because it is the easy case.
Why respiration is the easy case, stated as a claim about the systems themselves: “Respiration is unique compared with other systems (e.g., gastrointestinal) insofar as conscious regulation can immediately impact respiratory processes and… respiratory processes can affect emotion and cognition.” This is a real asymmetry and the wiki should hold it, because most of the field’s intervention evidence comes from breathing and is then generalized. See respiratory-interoception.
The three levels
1. Neuromodulation, and the one genuinely novel mechanism
Standard ground first: ~80% of vagal fibres are afferent, terminating mainly in the NTS, which relays to LC and raphe — so VNS is stimulation of an interoceptive pathway, and the auricular branch (ABVN), purely afferent and reaching the auricle at the cymba conchae, makes it non-invasively accessible (taVNS). Preliminary imaging finds taVNS modulates brainstem and cortical areas (left PFC, bilateral postcentral gyrus, left posterior cingulate, left insula) similar to classical VNS. Typical VNS parameters given: 20–30 Hz, 0.25–3.5 mA, 30–60 s ON / 5 min OFF. This extends bioelectronic-medicine, which the wiki built from Bonaz et al.
The novel item is RAVANS (respiratory-gated auricular vagal afferent nerve stimulation), and it deserves its own statement because of its logic rather than its evidence. The dorsal medullary vagal system runs in phase with breathing. Second-order NTS relay neurons receive pulmonary-stretch and aortic-baroreceptor afference primarily during inhalation, and may receive inhibitory input during inhalation and facilitatory input during exhalation from the ventral respiratory column. So stimulation is gated to exhalation — a phasic window with less competing afference and a more receptive relay. Reported result: enhanced brainstem targeting and therapeutic response in chronic pelvic pain and migraine.
Convergence worth flagging, and explicitly not support. This is the second time in three ingests that the wiki has met an intervention whose function is to raise the precision of an afferent signal by exploiting the timing of an efferent act. Quigley et al. supplied the first: phasic external-urethral-sphincter bursting during voiding defeats afferent accommodation (see urinary-interoception). RAVANS is the engineered version of the same shape. Neither literature invokes precision-weighting or active-inference, and neither cites the other. Recorded as a structural rhyme the wiki is noticing, not as evidence for a predictive account.
2. Slow breathing — the wiki’s first non-contemplative body intervention
See slow-breathing. The physiology is straightforward and, unusually for this wiki, does not go through representation at all: lower respiratory rate → larger tidal volume → cardiopulmonary stretch-receptor activation → reflex reduction in sympathetic outflow; plus improved arterial baroreflex control of SNS activity. Device-guided breathing (DGB) paces to ~5 breaths/min with musical tones.
The clinical case is built on sympathetic overactivity in PTSD and chronic kidney disease, both of which carry elevated cardiovascular risk, and on the fact that the pharmacological alternative (β-blockers, α1-blockers, central α2 agonists) has side effects — hypotension, orthostasis, fatigue, metabolic consequences — that make a behavioural option attractive. Evidence: DGB acutely reduces BP and muscle sympathetic nerve activity and improves baroreflex sensitivity in veterans with PTSD; 8 weeks of daily DGB reduced sympathetic reactivity to mental stress in PTSD.
Note what this intervention is not: it does not require the patient to notice anything. It is the cleanest case in the wiki of a “mind-body” treatment that could work with the mind entirely out of the loop — which makes the next result the most interesting thing in the paper.
3. Mindfulness, and the dissociation that gives it a mechanism
The neuroimaging summary is largely material the wiki holds first-hand from Farb and Lutz: breath-focused meditation increases activation in executive-function regions (PFC, ACC, premotor) and in interoceptive cortex including the insula, while decreasing DMN activation associated with mind-wandering and self-referential processing. Two specifics are worth carrying:
- Mindfulness training increased posterior insula activation — “sensitive to the respiratory rate and putatively considered primary interoceptive cortex” — during an interoceptive breath-focused task, and enhanced insula connectivity with the posterior ventromedial thalamus, a relay of the lamina-i-spinothalamocortical-pathway (Farb et al. 2013).
- Brain decoding: participant-specific ML classifiers for interoception / mind-wandering / self-referential states can estimate the percentage of time spent in interoceptive focus during meditation, and orienting attention to different body locations (breath vs feet) is decodable from fMRI (Weng et al. 2020). The wiki’s first measure of interoceptive attention that neither asks the participant nor uses a heartbeat.
The dissociation. In CKD patients, a single mindfulness session with a major breathing-awareness component reduced MSNA and blood pressure against a health-education control. The meditation also lowered respiratory rate — but slow breathing alone did not produce the same reductions in SNS activity or blood pressure (Park et al. 2014). One study, single session, small; and yet it is the wiki’s first evidence that the contemplative component does something a matched physiological manipulation does not. See the tension section below.
MABT: the wiki’s secondhand clinical vignette, now sourced
interoceptive-training-clinical has carried MABT since the first Farb (2015) ingest, cited to Price 2005/2007 as a vignette about women with trauma histories. Cynthia Price co-authors this review, and MABT arrives with trial results: two studies in women in SUD treatment showing longitudinal reductions in substance use, craving and emotion dysregulation versus control, with respiratory sinus arrhythmia and self-reported interoceptive awareness improved and maintained at 12 months. See mabt.
Two things to keep straight. First, the outcome is self-reported interoceptive awareness, not accuracy — so this is not a counterexample to does-mindfulness-enhance-interoceptive-accuracy. Second, the RSA result is the interesting one: a psychological training with a physiological readout, which the authors read as conscious interoceptive skills feeding back onto vagal tone. That is the same causal arrow as the whole framework, running upward instead of down.
The cross-system move: treating addiction through hunger
Leggio’s section proposes treating substance use by manipulating a different interoceptive system, on the grounds that craving is itself an interoceptive feeling and that addiction and feeding circuitry overlap. The ghrelin system: (acyl-)ghrelin, a stomach-derived peptide acting at GHS-R1a. Rodent work — deleting ghrelin peptide or receptor reduces alcohol intake; ghrelin administration increases intake, preference, self-administration and conditioned place preference; GHS-R1a blockade reverses these. Human: ghrelin administration increases cue-induced craving for alcohol but not for juice, and increases progressive-ratio alcohol self-administration; PF-5190457, a GHS-R1a blocker, was safe with alcohol and may reduce cue-elicited craving for alcohol and food.
The alcohol-not-juice specificity is what makes this more than a general appetite story. Recorded on craving.
Tension with what the wiki already holds
Does the mindfulness component add anything specific? The wiki’s position, built from Farb et al. (2022), has been sceptical: MBCT and a Well-Being CT that trains no body awareness reduced the same prefrontal biomarker equally, with identical relapse rates. Park et al. (2014) points the other way — mindfulness beat the respiratory manipulation it contains.
These are not in conflict, and the reason is worth stating because it sharpens both. Farb’s comparator is another psychotherapy and the outcome is depressive relapse; Park’s comparator is a physiological manoeuvre with no psychological content and the outcome is sympathetic outflow. Together they suggest a specific shape: mindfulness may add nothing over other structured psychological therapies, while still adding something over the bodily manipulation it produces as a side effect. Nothing here is a hard contradiction, and no page was rewritten — the Park result was added alongside on mindfulness-interoceptive-training.
The vagus, again in its mainstream form. As with Bonaz et al., this paper carries substantial vagal material — afferent proportions, NTS relay, RSA, taVNS — with no reference to Porges. The note on polyvagal-theory that the wiki holds two separate vagal literatures now has a second independent instance.
What is missing, and it is the same gap every time
Nine outstanding questions close the paper, and two of them are the wiki’s own recurring hole: what mechanisms underlie clinical improvement, and what are the steps and barriers to implementation. Notably absent from the framework: any way to decide which level a given patient should be treated at. The paper’s own logic — one pathway, three entry points — makes stratification the obvious next question, and it is not asked. That is the same gap recorded on bioelectronic-medicine and interoceptive-training-clinical, where the only candidate answer in the wiki remains Van der Does et al.’s mismatch-to-treatment rule (see van-der-does-2000-heartbeat-perception-reanalysis).
The paper also gestures, in one closing paragraph, at social context as a determinant of interoceptive cues — citing Fotopoulou & Tsakiris (2017) and physiological-linkage work — without developing it. See social-origins-of-interoception.