Research gap at a glance
- Question: When stroke volume is measured and statistically controlled, do heartbeat-task scores still predict the outcomes, age effects and questionnaire scores that are attributed to interoceptive accuracy?
- Why it matters: If a louder heart accounts for the associations, a large share of the wiki’s accuracy findings are facts about cardiodynamics rather than about perception. If the associations survive, the strongest standing objection to cardiac accuracy tasks is answered.
- Theme: Measurement
- Status: open
What the wiki already knows
- Stroke volume predicts heartbeat-detection performance (Schandry, Bestler and Montoya 1993), so people with a more forceful heart have an easier signal to detect (Heartbeat detection task, Rainer Schandry).
- Manipulations that make the heart louder raise counting scores with no opportunity for perceptual learning: exercise produces accurate perceivers above about 100 bpm who disappear by about 95 bpm, and a cold pressor raised counting accuracy in the same session (Does et al. (2000), Schulz et al. (2013)).
- Counting accuracy correlated with the Interoceptive Accuracy Scale (IAS) at r = .27 to .34; a loud-heart account predicts this selectivity too, and predicts it should weaken when stroke volume is controlled, but no study on the debate page measured stroke volume (Murphy et al. (2020), Is the heartbeat-counting task a valid measure of cardiac perception?).
- Heartbeat discrimination accuracy fell steeply with age (R squared = .30) without stroke volume being measured, although ageing dampens cardiac force (Khalsa et al. (2009), Age-related interoceptive decline).
- Discrimination tasks remove the guessing and arousal confounds of counting but not the loudness confound, since a forceful beat is also easier to time (Heartbeat detection task).
What would answer it
- In a sample spanning a wide age range, record beat-by-beat stroke volume with a non-invasive cardiac output measure during both counting and discrimination versions of the Heartbeat detection task, collect the Interoceptive Accuracy Scale (IAS) and a standard outcome, and test whether the task-outcome, task-questionnaire and age-task associations survive partialling out stroke volume; repeat after a cardiac-loudening manipulation such as the Cold pressor test.
Connections
- Raised on: Heartbeat detection task, Is the heartbeat-counting task a valid measure of cardiac perception?, Age-related interoceptive decline, Maturational dualism, Dunn et al. (2010)
- Related concepts: Interoceptive sensitivity, Age-related interoceptive decline, State vs trait interoception
- Related debates: Is the heartbeat-counting task a valid measure of cardiac perception?, Is more interoceptive awareness better?
The cardiodynamic confound is the objection to heartbeat tasks that the wiki meets on more pages than any other, and it is also the one that has never been tested directly. Every source that raises it, from Schulz et al. (2013) to Khalsa et al. (2009), concedes that stroke volume was not measured. The problem is not that the confound has been tested and found; it is that nobody has looked.
It matters because the confound can generate many of the wiki’s headline results without any perception in them. A forceful heart is a bigger signal to count and to time, so it could produce the age decline in accuracy, the gain in accuracy after exercise or stress, and the selective correlation between counting and believed accuracy in Murphy et al. (2020), since someone with a loud heart may also learn that they can feel it racing. In Dunn et al. (2010) the confound is uncomfortably close to the effect itself: interoceptive accuracy moderated the coupling between heart-rate response and felt arousal, and a larger cardiac signal is also a larger signal to couple.
Is the heartbeat-counting task a valid measure of cardiac perception? notes a point that keeps this from being simply a nuisance variable. Ehlers read the stroke-volume finding as evidence the task tracks real cardiac events, while the sceptical reading takes it as evidence the score indexes signal amplitude. Both can be partly right, which is why the analysis needs both a between-person partial correlation and a within-person manipulation.