Age-related interoceptive decline

The developmental axis that runs downward. Where social-origins-of-interoception asks how interoception is built in infancy, this asks what happens to it afterwards.

The wiki now holds two independent self-report lines on the finding — one incidental (Volynets, where age fell out of a universality study) and one purpose-built (MacCormack, designed to find it) — plus a named theory for it (maturational-dualism) and, since the Khalsa et al. (2009) ingest, a first-hand objective-accuracy source. What the self-report pair measure is the felt or represented body; what Khalsa measures is a heartbeat-detection score. All three constructs decline with age — the convergence maturational-dualism predicts — but the biological mechanism is instrumented by none of them, which remains the standing limit of everything on this page (and, in Khalsa’s objective case, the construct where “quieter heart” and “worse perceiver” are hardest to separate).

The pathological continuation: neurodegeneration (via Bonaz et al. 2021)

Everything else on this page is about normal ageing. Bonaz et al. (2021) place that decline as the first step of a longer curve, citing Khalsa et al. (2009) for the normative slope and then continuing past it:

Sensitivity to internal bodily signals (specifically, performance accuracy on interoceptive tasks) decreases with age, yet interoceptive and autonomic brain networks are further compromised in neurodegenerative disorders, notably Alzheimer’s disease.

In Alzheimer’s, degeneration spans insula and ACC plus brainstem nuclei (with prolonged vagus somatosensory evoked-potential latencies, present in AD but not in major depression — a differentiating marker), and the cardiac-interoceptive deficit is four-way: impaired accuracy (imprecise fingertap tracking of own heartbeats), abnormal heartbeat-evoked potential modulation, impaired interoceptive learning (negligible gain after feedback), and impaired metacognitive awareness (confidence–accuracy mismatch). In behavioural-variant frontotemporal dementia, frontal-predominant degeneration hits anterior insula and both perceptual sensitivity and metacognitive insight are compromised. Across neurodegenerative disorders, loss of self-awareness tracks damage to cingulate, prefrontal and fronto-temporal cortices strongly connected to the insula.

Two things this adds. First, it is the wiki’s only source measuring all four taxonomy dimensions in one sample — the taxonomy operationalized rather than argued about, and in a clinical population. Second, it supplies a partial answer to the mechanism question this page keeps flagging as unaddressed: in the neurodegenerative case the mechanism is structural loss of the interoceptive cortices themselves, which is at least instrumented. Whether normal ageing’s decline is the mild end of the same process or a different phenomenon (cardiodynamics, peripheral signal quality) is exactly what none of the sources here resolve — and Bonaz et al.’s “yet… further compromised” phrasing quietly assumes continuity without arguing for it.

The regulatory question this page holds open takes on a sharper edge here. Whatever case can be made that reduced interoceptive contact aids emotion regulation in healthy ageing (maturational-dualism, and the contemplative tension noted below), it plainly does not extend to Alzheimer’s. Recorded as a boundary condition on the “less contact may be adaptive” reading. See is-more-interoceptive-awareness-better.

The instrument runs out exactly where the curve gets interesting (Chen et al. 2021)

A constraint that sits underneath everything above and was not stated on this page until now. Chen et al. (2021) report that lifespan changes in interoception “remain largely unexplored,” and name the obstacle as instrumentation rather than interest: there is no measure of interoceptive sensitivity suitable for non-verbal populations — infants, or older adults with dementia.

That is a direct problem for the section above it. The Alzheimer’s findings Bonaz et al. report — four taxonomy dimensions impaired in one sample — depend on tasks requiring a participant to track their own heartbeats and rate their confidence. Those are exactly the demands the NIH review says cannot be met in dementia populations. Both claims can be true (the AD samples are presumably mild and still verbal), but the wiki should hold them together: the measured decline may be sampled from the part of the disease course where measurement still works, which would bias the observed curve toward whatever is detectable rather than toward whatever is happening.

It also cuts the same way at the other end. The developmental claims on social-origins-of-interoception depend on infant interoception, and Chen et al. cite one attempt (Maister et al. 2017) against a background of acknowledged inadequacy — the same verdict Tsakiris gives first-hand in Quigley et al. Two of the wiki’s developmental and ageing threads are thin at their ends for one shared reason, and it is a methodological reason. See interoceptive-taxonomy.

The constructive item: Chen et al. call for longitudinal designs and for identifying milestones of interoceptive development, neither of which exists.

The purpose-built source: MacCormack et al. (2021)

MacCormack, Henry, Davis, Oosterwijk & Lindquist (2021) is the study the earlier version of this page was waiting for — the first to ask the question directly, across two methods and two samples (N=143 + N=198, ages 18–75), from the Lindquist/Barrett constructionist lab. It supplies three things Volynets could not:

• Interoceptive specificity. In a property-association task, age weakened the association between emotion categories and interoceptive properties (e.g. “heart racing” for anger) while leaving behavioral and situational properties untouched. The decline is not general semantic or cognitive aging; it is specific to the body’s contribution to emotion. (Volynets could not test this — its instrument only colours the felt body.)
• Arousal specificity. The decline is carried by high-arousal interoceptive sensations (“blood pumping”), not low-arousal ones (“drained”), in both studies. This localizes the aging effect on the arousal dimension of core-affect — the high-arousal, autonomically-driven pole fades. Volynets found dampening for every emotion; MacCormack shows which part of the body-signal is fading.
• A structural claim. In daily-life self-reports (Day Reconstruction Method), interoception decoupled from emotional situations with age but not from behaviors — older adults still feel the body when overtly acting, and lean relatively more on situational/appraised meaning. That is a concrete, testable shape for maturational-dualism and a bridge to the wiki’s appraisal thread.

Two cautions the study itself supplies. The curvilinear shape (rise to ~45, then decline) appeared in Study 1 but not Study 2 — the linear-vs-curvilinear question is unsettled, and Study 2’s sample had few older adults. And critically, MacCormack et al. measure conceptual knowledge and self-reported intensity, not interoceptive accuracy or physiology — so, exactly like Volynets below, the felt effect is real and the biological mechanism is proposed (peripheral demyelination, reduced autonomic reactivity, poorer objective interoception; Khalsa, Rudrauf & Tranel 2009; Murphy et al. 2018), not measured. The objective-accuracy counterpart is that Khalsa 2009 paper, now read first-hand (see the section below); it supplies the missing accuracy leg but, being confoundable with cardiodynamics, does not by itself turn the mechanism from imported to measured.

The first source (incidental): Volynets et al. (2020)

Volynets et al. (2020) is the wiki’s first source on it, and the effect is incidental to that paper’s main claim — which makes it more interesting, not less: age was the only background variable that substantially moved bodily sensation maps once culture, language, sex, education and BMI were all in the same model.

Across ages 18–90 (N = 3954), felt bodily sensation dampened for every emotion. Map intensity correlated negatively with age for anger, fear, disgust, happiness, surprise, anxiety, love, contempt, pride, shame and jealousy, and positively for depression and sadness — which is dampening too, since those two are carried by deactivation (negative map values), so a positive correlation means the felt loss of activity also shrinks. Subjectwise mean absolute intensity fell with age at rs = 0.11 (p < 0.001): reliable, and small.

Note the structure of the result: what changes with age is intensity, not topography. The maps stay the same shape and lose contrast. Emotions are still felt in the same places; they are felt less.

The lifespan arc

Combined with Hietanen et al. (2015), where children’s bodily maps become progressively more discrete toward adulthood, the group proposes: sharpen, then fade. Both legs are cross-sectional, so both are equally open to cohort explanations — a 70-year-old and a 20-year-old differ in more than age, including in how they use a colouring interface.

Why it might matter (proposed, not tested)

Volynets et al. offer a chain of speculation worth recording because it is testable and because it inverts a familiar finding:

• Basal metabolism and physiological reactivity slow with age (Schofield 1985); limbic reactivity falls, prefrontal activity rises, cardiovascular response slows.
• So there is less bodily signal — and perhaps weaker interoceptive access to it — to feel.
• This could explain why subjective and physiological emotional responses decouple in ageing (Scheibe & Carstensen 2010).
• And it could explain why older adults regulate emotion better (Urry & Gross 2010): “due to weakened interoceptive awareness, they may become less connected to the bodily components of emotions, and this shift of emotions towards ‘cognitive’ as opposed to embodied processes might make the emotions easier to regulate.”

That last inference deserves flagging as a tension with the rest of this wiki rather than an extension of it. The contemplative and clinical literature here treats more interoceptive contact as the route to wellbeing — see mindfulness-interoceptive-training and farb-2015-interoception-contemplative-health. Volynets et al. propose that less of it may underlie the increased life satisfaction and emotional wellbeing of old age. Both cannot be straightforwardly right, and the resolution probably lies in the interoceptive-taxonomy — attention and sensibility are not sensitivity, and “less signal” is not “better relationship to signal.”

Update (Lyons ingest). This tension now has a page of its own — is-more-interoceptive-awareness-better — and its first adjacent source. Lyons et al. (2021) ran the same instrument in depression, found reduced felt bodily emotion, and drew the opposite moral: that body awareness should be increased in treatment. Same measure, same direction of effect, opposite valuation.

Two things that does and does not settle:

• It does not refute the ageing proposal. Depression and healthy ageing are different states, and “less felt body is bad in depression” is consistent with “less felt body is fine at 75.” A defender of Volynets can say the quieting is benign when it tracks a genuinely quieter body and pathological when it tracks withdrawal from a body that is still signalling.
• It does sharpen what the ageing proposal owes us. If reduced interoceptive contact were generically regulation-friendly, depression and antidepressant blunting should look like emotional competence rather than complaints. They do not. So the beneficial-quieting story needs a reason why ageing’s version is the good kind — and Volynets et al. do not supply one, because the mechanism they invoke (less signal → emotion shifts toward the cognitive → easier to regulate) is not specific to ageing at all.

Update (MacCormack ingest). The tension is now sharper, not resolved. MacCormack et al. draw the same moral as Volynets — a quieter body makes negative, high-arousal emotion less frequent and easier to regulate, part of the mechanism behind the well-being of old age — and add its dark twin: the same fading interoceptive input should starve somatic-marker guidance of affect-based decisions, and older adults do make worse such decisions (scams, finances). One mechanism, two opposite valuations. So the beneficial-quieting story now has a first-hand advocate with a purpose-built design, and still owes the same debt: a reason why aging’s version of reduced interoceptive contact is the good kind when depression’s and blunting’s are not. See maturational-dualism, which is where this two-edged consequence is worked out.

The decision-making arm, and the middle term nobody has measured

Added with the Bechara, Damasio & Damasio (2000) ingest, which reports Denburg et al. (1999): adults above 64 perform poorly on the iowa-gambling-task relative to ages 26–56.

The shape is the interesting part, and it matches nothing else on this page. Performance in older adults is dichotomous — “some perform very well and some perform very poorly” — rather than the group shifting downward together. The authors ask why, and connect it to why some older adults are especially vulnerable to advertising fraud in real life.

Why this belongs here. Khalsa, Rudrauf & Tranel (2009) — same Iowa group, same decade — close their paper by tying declining interoceptive ability forward to age-related decision decline, citing Denburg, Tranel & Bechara (2005). This is the earlier abstract behind that line. So the wiki now holds both ends of the Iowa argument: interoceptive accuracy falls with age (first-hand, discrimination task, r ≈ −.47), and decision quality falls with age (secondhand, gambling task).

And the middle term is missing. Nobody in this wiki has measured interoceptive accuracy and decision quality in the same older sample. The inference “older adults decide worse because they feel their bodies less” is assembled from two studies, in the same laboratory, with no overlapping participants — and the studies pull in different directions on distribution: Khalsa reports a continuous linear decline in accuracy, Denburg a bimodal split in decision quality. A continuous decline in the proposed cause is an awkward fit for a dichotomy in the proposed effect. Either something else sorts older adults into the two decision groups, or the accuracy decline is not what produces the decision decline.

Worth noting what would settle it, since it is the same unrun study named on iowa-gambling-task and does-somatic-feedback-guide-decisions from a different angle: a heartbeat task and a decision task in one older sample, with the per-choice bodily signal recorded. Werner et al. (2009) ran the extreme-groups version in young adults; the lifespan version does not exist here.

The objective-accuracy source: Khalsa, Rudrauf & Tranel (2009)

Khalsa, Rudrauf & Tranel (2009) is the paper both self-report sources above lean on by citation, now read first-hand — and it delivers what they could not: an objective decline. In 59 adults aged 22–63, accuracy on a heartbeat-discrimination task (Brener lineage, not Schandry counting) fell steeply with age — age alone accounted for 30% of the variance (R²=.30), reliably across two sessions, with BMI and sex contributing nothing. So the felt/represented decline this page catalogues now sits beside first-hand evidence that the ability declines too.

Two things it settles, and one it does not.

• It settles that the decline is not only in the felt/represented body. MacCormack and Volynets measure knowledge and felt intensity; Khalsa measures a detection score. All three fall with age. That is three constructs moving together, which is what maturational-dualism predicts and what a purely semantic-aging or cohort-conceptual account would not.
• It settles the direction and rough size of the objective effect — large (30% of variance), and steep enough to appear across a middle-aged sample, not just at the extremes of the lifespan.
• It does not settle the mechanism, and in one way it sharpens the problem. Khalsa argues the decline is central (insular/somatosensory cortical thinning) and/or cutaneous (Pacinian decline), explicitly not cardiac-afferent (transplant patients detect normally). But he did not measure stroke volume, and aging dampens cardiac force — so the cardiodynamic confound predicts the same age decline with no loss of perceptual skill at all. On the objective task, “quieter heart” and “worse perceiver” are the hardest to tell apart, and this study cannot separate them. The biological mechanism the whole page hands off remains, even with the accuracy source in hand, argued rather than instrumented.

The decisive test — pharmacological afferent blockade across ages, watching emotional experience (the nadolol study MacCormack names), or stroke-volume-controlled accuracy across ages — has still not been run.