Micah Allen

Enters the wiki through one instrument paper — which is itself characteristic, and the reason this page is worth having beyond a single citation.

The tooling stance

Most researchers on this wiki enter as theorists (Craig, Barrett, Seth, Damasio, LeDoux), as conveners (Khalsa), or as clinicians (Ehlers, Craske, van der Kolk). Allen enters as a builder, and the papers here are instruments: the Respiratory Resistance Sensitivity Task (2022) and respyra (2026), both automated, both given away.

That matters for how the wiki reads him. A theory paper is an argument to agree or disagree with. An instrument is a claim about what the field is capable of asking, and its argument is made by what it makes measurable. respyra’s argument — stated in its first sentence — is that interoception research has been a science of perception in a domain whose theories are all about control. See interoceptive-control.

Respiration over cardiac

Both instruments are respiratory, and the choice is programmatic rather than incidental. The wiki’s standing complaint is that the field is bottlenecked on a cardiac task that under 10% of healthy adults can perform, that 95% produce a number for anyway, and that is confoundable by stroke volume, arousal and belief. respiratory-interoception records that a respiratory answer is available and under-used: breathing is objectively and continuously measurable without asking the participant anything, and it can be manipulated parametrically.

Allen has spent a decade building that answer. The tension the wiki should keep in view is the one recorded on interoceptive-control: respiration is the most tractable channel because it is the least representative one.

The decorrelation result

Allen is senior author on [[banellis-2026-cross-modal-decorrelation|Banellis et al. (2026, Communications Psychology)]], now read first-hand: interoceptive ability is uncorrelated across respiratory and cardiac axes in 241 participants, with moderate Bayesian evidence for the null across sensitivity, precision and metacognitive efficiency. The forecast recorded here at the respyra ingest — that this would be the most consequential single finding for the wiki’s cross-study reasoning — largely holds: “interoceptive accuracy” has no demonstrated person-level referent, and claims comparing a cardiac score to an interoceptive construct measured some other way are comparing quantities shown not to covary. See is-interoception-domain-general.

Two qualifications the forecast missed. The evidence is moderate, not decisive (BF01 ≈ 5–7), and confined to two channels at rest in healthy Danish students — Allen’s own paper leaves open that coherence emerges under arousal, which is Khalsa’s standing methodological position arriving as the leading defence of the view the study attacks. And the study’s positive result turns out to matter as much as its null: mean confidence correlates strongly across cardiac, respiratory and auditory tasks alike, so the one thing that generalizes is not interoceptive.

There is an irony worth recording: the same author whose large-scale study decorrelates the channels also builds the instruments that would let each channel be measured properly. Those are two halves of one position — if the channels do not generalize, then the discipline needs good instruments per channel, not a better general one. Both of the instruments used to demonstrate the decorrelation are his group’s (HRDT, Legrand et al. 2022; RRST, Nikolova et al. 2022), which is the sharpest version of the tooling stance above: the null is only credible because the instruments are matched, and they are matched because he built them both.

The unresolved tension between his two 2026 papers, worth watching: the decorrelation result says perceptual interoception is organ-specific, while interoceptive-control — his constructive proposal — is measurable in exactly one organ. If control is channel-specific in the way perception has now been shown to be, the construct he is building the field toward is a fact about breathing. His own data cannot yet say.

Where he sits relative to the others

• sarah-garfinkel / hugo-critchley — shares the psychometric seriousness and the accuracy/sensibility/awareness apparatus (Garfinkel et al. 2015 is respyra’s opening citation), but pushes it onto a channel their programme does not centre, and toward control rather than metacognition.
• frederike-petzschner — the closest match in temperament: computationally committed, deflationary about what the computational programme has actually shown. Her three-arc loop gives Allen’s control construct its formal address.
• sahib-khalsa — shares the conviction that resting self-report is inadequate and that interoception must be perturbed to be measured. Khalsa perturbs pharmacologically (isoproterenol); Allen perturbs the sensorimotor mapping. Same instinct, opposite ends of the loop.
• helen-weng — both treat respiration as the entry point for intervening on interoception, from opposite directions: Weng asks what breathing does to the body and brain, Allen asks how well the person can drive it.