Cold pressor stress induces opposite effects on cardioceptive accuracy dependent on assessment paradigm

A between-subjects experiment (Schulz, Lass-Hennemann, Sütterlin, Schächinger & Vögele) that does something rare in the wiki’s cardiac literature: it takes one acute manipulation and runs it through two paradigms in the same people, and finds the manipulation reverses sign between them. Forty-two healthy undergraduates were randomised to a socially-evaluated cold pressor (n=21) or a warm-water control (n=21); before and after, each completed a Schandry heartbeat-counting task and auditory and visual versions of the Whitehead heartbeat-discrimination task, with cardiovascular measures throughout.

Metadata note: the copy in raw/ is the accepted author manuscript (no journal line printed on it); journal, year and DOI are recorded from the published version, Biological Psychology 93(1):167–174.

The one result that matters

Acute stress raised counting accuracy and lowered discrimination accuracy. The Schandry score improved after the cold pressor (time×stress interaction, F[1,40]=4.17, p<.05; CP group pre→post p<.01, controls flat). The visual Whitehead score moved the other way (three-way time×modality×stress interaction, F[1,40]=5.85, p=.02; visual CA fell in the CP group, rose in controls). The auditory Whitehead task was at chance and moved for no one.

So “does stress improve cardiac interoception?” has no answer independent of the instrument. This is the paper’s contribution, and it is why the study is filed against is-the-heartbeat-counting-task-valid rather than treated as a result about stress.

Two readings of the dissociation, and the paper endorses both

Attentional (the title’s reading). The Schandry task asks only for attention to visceral sensation; the Whitehead task demands attention to visceral sensation and an external stimulus (a tone or a flash) at once. If acute stress narrows attention onto task-relevant, salient cues (Chajut & Algom 2003) — and if under stress the salient cue is the body — then stress helps the purely-internal task and hurts the divided-attention one. The authors frame this via Pennebaker’s competition-of-cues model: internal and external signals compete for one limited processing capacity, and stress reallocates it inward. On this reading the study is evidence that stress effects on interoception are gated by where attention is deployed, not a fixed property of “stress.”

Cardiodynamic (the reading this wiki weights more heavily). The cold pressor raised heart rate by 8.2 bpm and systolic pressure by 20.6 mmHg — sympathetic activation, which increases cardiac contractility and stroke volume, which makes each heartbeat a bigger mechanical signal. A louder heart is easier to count, so the Schandry increase is exactly what a signal-amplitude account predicts with no change in perceptual skill (see heartbeat-detection-task, the cardiodynamic-confound sections). The authors say this themselves and flag it as speculative because they did not measure stroke volume. The same sympathetic shift plausibly hurts the Whitehead task: the fixed 230 ms R-wave-to-stimulus delay is calibrated to a resting cardiac cycle, and stress-shortened pre-ejection period / pulse transit time move the felt beat earlier, so the previously-”synchronous” stimulus no longer feels synchronous. Elevated heart rate is also independently associated with worse discrimination performance (and here correlated r=.38 with visual CA across participants).

These are not rivals so much as two mechanisms the design cannot separate, and the honest verdict is that the study demonstrates the dissociation cleanly and adjudicates its cause not at all.

Why the wiki keeps this study

Three of its incidental results are load-bearing elsewhere:

• Counting and discrimination did not intercorrelate (r=.22 and .08, both ns) while the two discrimination modalities did (r=.63). This is a first-hand instance of the divergence Quadt, Critchley & Garfinkel (2018) cite (Ring & Brener: “heartbeat counting is unrelated to heartbeat detection”) and the heartbeat-detection-task page already carries — two tasks with the same name and modality measuring different things, in one sample.
• A within-session state manipulation of accuracy. Almost everything on the wiki treats a heartbeat score as a trait measured once; state-vs-trait-interoception notes that nobody has manipulated accuracy within a session and watched it move. This study does — stress moves it, in opposite directions by task — which is the strongest form of “the number is not a fixed personal attribute” the accuracy literature offers, even granting that the mover may be the heart rather than the perceiver.
• Self-report vs performance, once more mostly decoupled. Only one of many correlations tied a questionnaire (BPQ ‘Awareness’) to a task score (visual Whitehead, r=.31, flagged by the authors as fragile given the number of comparisons), and chronic stress (TICS) predicted no task at all — consistent with the wiki’s recurring finding that believed and measured interoception come apart (see interoceptive-taxonomy).

What it does not show

It does not show that stress improves cardiac perception — the Schandry increase is confounded with signal amplitude, unmeasured here. It does not show HPA involvement — no cortisol. And its auditory task being at chance means one of its three instruments contributed a floor rather than a datum. The authors’ closing suggestion is the right one for this wiki too: reactive counting/discrimination tasks may be the wrong tools for studying stress and interoception, and non-reactive measures (HEPs, cardiac modulation of startle) should be tried instead.