Murphy (2022): Propensity to use interoceptive signals
An invited two-page commentary in Biological Psychology, responding to Suksasilp & Garfinkel (2022)‘s eight-dimension interoception model — the scheme the wiki tracks as the fifth taxonomy. Murphy grants the model is “the most comprehensive to date” and then names one thing it does not contain: whether a person actually uses the interoceptive signals they can perceive.
The gap: ability is not use
Every dimension in the taxonomies is a property of perception — how strong the signal is, how accurately it is detected, how confidently it is rated, where attention points. Murphy’s point is that a perfectly perceived signal can still go unused, and that the degree of use is itself a stable, consequential individual difference the models omit. She splits it three ways:
- General interoceptive propensity — the tendency to gauge internal states from internal cues versus external ones (posture, situation, learned rules about how one “should” feel).
- Process-specific reliance — how far a person leans on interoceptive signals for a particular job (emotion, hunger, intuitive decision-making), which need not be uniform across jobs.
- Adaptiveness of use — whether deploying the signal helps or hurts, which depends on the person’s interoceptive insight and the task at hand.
This is the internal-vs-external axis turned into an individual-difference variable: not can you read the body, but do you, and should you.
The empirical wedge: sex differences
Murphy’s motivating anomaly is a dissociation the accuracy literature cannot absorb. Men outperform women on lab tasks of interoceptive accuracy, yet the difference vanishes in naturalistic settings where external cues are available for compensation (Cox et al. 1985; Pennebaker & Watson 1988) — and emotional processing relates to interoception more strongly in men (Pennebaker & Roberts 1992; Lischke et al. 2020), even though women generally outperform men on tests of emotion (Thompson & Voyer 2014). If interoceptive ability drove emotional ability, the group with better emotion should have better interoception. It does not. Murphy’s resolution: the groups differ in propensity — how much they weight internal versus external information — not (only) in ability, and for some processes weighting external cues is the better strategy.
That is the same logic Banellis et al. (2026) reach from psychophysics (what generalizes across the body is confidence, not ability) and Ventura-Bort et al. (2021) reach from factor analysis (self-report splits into believed accuracy vs deployed attention). Murphy names the disposition those results keep circling: use, not competence.
What makes use adaptive — and why it isn’t intrinsically good
Murphy’s sharpest move is to deny that using the body is a virtue. Use is adaptive only conditionally, and she offers two failure modes:
- Over-use with poor insight. Someone who wrongly believes their signals are reliable, or that they perceive them accurately, will deploy interoceptive cues where it does not help.
- Under-use with poor insight. Someone who wrongly believes their signals are unreliable will discard a genuine source of information.
And she cites the evidence that use is sometimes maladaptive by construction: Dunn et al. (2010) found interoceptive accuracy improved decisions only when bodily cues favoured advantageous options — the “gain term” result. Murphy generalizes Dunn’s finding from accuracy to use: perceiving the body more, or relying on it more, has no fixed valence; it amplifies whatever the body is signalling, right or wrong. Women’s superior emotion-recognition despite poorer lab accuracy is her second instance — for some processes, weighting non-interoceptive cues is the advantageous policy.
Trait or state — left open, and pointed at the right page
Murphy explicitly does not resolve whether propensity is a trait or a state: “if the use of interoceptive signals varies across situations, one’s ability to identify and learn when using interoceptive signals is adaptive — or not — may also be a meaningful individual difference.” If it is a trait, “the available evidence suggests it may only be adaptive in certain (likely individual-specific) situations.” This is the state-vs-trait-interoception question posed for a construct that does not yet have a measure — and it is posed for use rather than for perception, where the wiki’s one variance-partitioning estimate (Wallman-Jones’ ICC = 0.51) already lives.
How to measure it (proposed, not built)
Murphy sketches two routes, both about isolating cue-use rather than perception:
- Compare an individual’s performance in naturalistic versus lab-based assessment — where the naturalistic setting supplies external cues for compensation (Cox et al. 1985, at the individual level).
- Lab tasks that manipulate access to external cues (Murphy, Catmur & Bird 2018) — hold the internal signal fixed, vary whether an external cue is available, and read propensity off how much the person’s judgment shifts.
No such instrument is delivered here; the commentary’s contribution is to name the construct and argue it is missing, not to operationalise it.
Placement in the wiki
The study is the source page for interoceptive-propensity and a field position on is-more-interoceptive-awareness-better (use as a conditionally-adaptive gain term, generalizing the Dunn row). It bears on is-interoception-domain-general — propensity is process-specific by hypothesis, so a person may weight the body heavily for hunger and lightly for emotion — and it is the first source in raw/ from Murphy, whose Bird-lab work on alexithymia as a “multidomain, multidimensional failure of interoception” (Murphy, Catmur & Bird 2018) the wiki had carried only by citation. Held for exactly what it is: a well-argued conceptual proposal with no data behind it and no measure attached, whose value is diagnostic — it names the use dimension the field has theorised around without ever putting on a taxonomy.