Ventura-Bort, Wendt & Weymar (2021) — Interoceptive sensibility and emotional conceptualization
A Potsdam individual-differences study framed as a test of two components of the theory-of-constructed-emotion: core-affect (indexed by interoceptive sensibility questionnaires) and conceptualization (indexed by emotion-clarity/alexithymia questionnaires). It asks how those individual differences moderate three aspects of emotional experience — intensity, arousal, and granularity.
The wiki should read it for something other than its stated aim. Its most consequential result is a psychometric one, and it lands on the construct this wiki uses most loosely.
The result the paper is framed around, and the result that matters here
What they set out to test. If core affect and conceptualization are separate ingredients, then measures of interoceptive sensibility and measures of emotional conceptualization should be separable, and each should predict emotional experience in its own way.
What they found. The measures did not separate that way. Across eight questionnaires (ICQ, IAS, MAIA-2, TAS-20, MAS, RDEES, TMMS, plus W-BQ12 as an outcome), the correlations between the bodily and the emotional instruments are pervasive and substantial — ICQ↔Difficulty Identifying Feelings r = .55, IAS↔TMMS Clarity r = .49, ICQ↔TMMS Clarity r = −.51. A principal component analysis forced to two factors split them along an entirely different seam:
| Factor 1 — “Sensibility” (38.9% var.) | Factor 2 — “Monitoring” (10% var.) | |
|---|---|---|
| what it is | belief that one accurately detects and can label internal states | tendency to deploy attention inward |
| bodily instruments | IAS (+.79), MAIA Attention regulation, MAIA Trusting, ICQ (−) | MAIA Noticing, MAIA Emotional awareness |
| emotional instruments | TAS-20 all three subscales (−), MAS Labeling (+.85), RDEES Differentiation, TMMS Clarity (+.84), TMMS Repair | MAS Monitoring (+.72), RDEES Range, TMMS Attention |
| loads on both | — | MAIA Body listening (equally) |
Read the rows, not the columns. Every cell in the “bodily instruments” row has an emotional counterpart in the same factor. A questionnaire asking whether you can tell when you are thirsty (IAS) and a questionnaire asking whether you can tell what you are feeling (TMMS Clarity) load on the same component; a questionnaire asking how often you notice your body (MAIA Noticing) and one asking how often you think about your mood (MAS Monitoring) load on the other. The split is believed accuracy vs deployed attention, and it runs orthogonally to the body/emotion distinction the study was designed around.
That is the finding this wiki should carry, and it is not the one in the title.
Why the split matters for the wiki’s own vocabulary
Every taxonomy on interoceptive-taxonomy contains sensibility as a single entry — Farb et al.’s “personal account of one’s internal-sensation experience, including confidence in their own interoceptive ability and feelings of engagement”; Khalsa et al.’s “self-perceived tendency to focus on interoceptive stimuli”; Garfinkel’s “subjective belief about one’s own ability to perceive bodily signals.”
Read those three definitions again against the table above. Farb’s definition contains both factors (confidence and engagement). Khalsa’s is Monitoring. Garfinkel’s is Sensibility. The taxonomies were not merely giving one thing three names; on this evidence two of them are naming different things with the same word, and the third is naming both at once. This is the same species of problem the taxonomies were written to fix, one level down, and it is now measured rather than argued.
The distinction is not new to the primary literature — Murphy, Catmur & Bird (2019) proposed classifying self-reported interoception into accuracy-belief and attention, and Gabriele et al. (2020) built the Interoceptive Attention Scale on it, both cited here. What this paper adds is that the split survives adding the emotion questionnaires, i.e. the two dimensions are dimensions of internal self-knowledge in general and not of interoception in particular.
The part with teeth: the two halves point in opposite directions
In the DRM — up to 30 recalled episodes rated on 19 emotion terms across two days:
| outcome | Sensibility | Monitoring |
|---|---|---|
| granularity, negative emotions | + β = 0.27, p = .011 | − β = −0.31, p = .004 |
| granularity, positive emotions | ns (β = −0.12) | − β = −0.21, p = .049 |
| emotional intensity | − β = −0.20, p = .05 (trend) | − β = −0.31, p = .003 |
| well-being (W-BQ12 total) | r = .58* | r = .12 (Z = 3.94***) |
| negative well-being | r = −.54* | r = .07 (Z = −5.23***) |
| MAIA Not-worrying | r = .40* | r = −.20* (Z = 4.94***) |
No interaction between the factors reached significance anywhere, so these are two independent dispositions with different — in the Not-worrying and negative-well-being rows, opposite-signed — relationships to how emotion is experienced and how well the person is doing.
The person who believes they read their insides accurately differentiates their negative emotions better and is doing better. The person who attends inward a lot differentiates worse, feels less intensely, and worries more. Both would score as “high interoceptive awareness” on a composite.
That profile is not new to this wiki — it is Oldroyd et al.’s attachment-anxious participant (high MAIA Noticing, low Not-worrying) arriving from a completely different literature, without attachment, and with the two halves separated by factor analysis rather than by subscale inspection. See is-more-interoceptive-awareness-better.
What the null half of the study is worth
Every effect above comes from the DRM. The emotion-differentiation task — 40 IAPS pictures, SAM valence/arousal ratings, then eight emotion labels per picture — produced nothing: not intensity, not arousal, not granularity, not a single interaction, in 127 participants.
The authors’ explanation is about labels and stimuli (eight labels may be too few; standardized pictures may evoke valence and arousal without evoking specific emotions). Probably right. But the null is worth more than that here, because of which paradigm it is.
affective-picture-viewing with SAM arousal ratings is the exact paradigm behind this wiki’s sharpest live disagreement — Pollatos et al. (2005) and Dunn et al. (2010) running nearly the same experiment and disagreeing about whether interoceptive accuracy predicts felt arousal. Here the same paradigm, in a larger sample than either, finds that interoceptive sensibility predicts felt arousal not at all: β = 0.012, t(84) = 0.13, p = .89.
So on the one paradigm where the wiki has both, accuracy has a contested relationship to felt arousal and sensibility has a flat one. That is a small, clean addition to the accuracy/sensibility dissociation recorded across interoceptive-taxonomy, maia and interoceptive-sensitivity — and it is a dissociation by outcome rather than the usual “the two don’t correlate,” which is a different and slightly stronger form.
Held with the obvious brake: it is a null, the DRM nulls nothing and the ED task nulls everything, and a design that fails to find its own hypothesized effects in one task is not a clean instrument for adjudicating anyone else’s.
What it does to the theory it was testing
The paper is an individual-differences test of TCE’s ingredients, and it is honest about the outcome: “the factor scores did not differentiate between interoceptive sensibility and emotional conceptualization scales.”
Two readings, and the paper takes the softer one.
The authors’ reading. Sensibility maps onto individual differences in conceptualization, Monitoring onto individual differences in attention — and attention is itself one of TCE’s four components (Barrett 2017), so the theory absorbs the result. That is available, and it is not ad hoc: attention is in the theory.
The reading the wiki should also record. The measures chosen as correlates of core affect (interoceptive sensibility) and of conceptualization (emotion clarity) turned out to load on the same two components as each other, and the components sort by response style rather than by construct. A theory whose ingredients cannot be recovered from the measures nominated as their indices has not been tested. This is the falsifiability worry already recorded on that page arriving in psychometric form: the result was absorbed, and it is not clear what result would not have been.
Neither reading is settled by the data. The discriminating study is the one the authors name: run these questionnaires alongside an objective interoceptive measure and a physiological one.
Two smaller things worth keeping
Positive and negative granularity came apart. They were uncorrelated in the DRM (r = .13, ns) and negative granularity was much higher (d = 1.23). Barrett (1998) reports them associated; the authors flag the divergence themselves and offer no account. Only negative granularity carried the Sensibility effect, converging with the outside literature (Barrett et al. 2001; Demiralp et al. 2012; Kalokerinos et al. 2019) that negative granularity is the half that predicts things. See emotional-granularity.
Alexithymia appears first-hand. alexithymia has been carried in this wiki entirely at one remove, through Quadt et al. and Bonaz et al.. Here the TAS-20 is administered directly (α = 0.83) alongside a bodily-sensibility scale, and the two load on one factor — the Brewer/Bird “alexithymia is a general deficit of interoception” claim in its weakest and most testable form: at the level of beliefs, difficulty identifying feelings and difficulty identifying bodily states are one dimension. That is a shared-method-variance result before it is a substantive one, and it should be held that way.