Someone sits quietly in a laboratory and is asked a question that sounds almost childlike: How many times did your heart beat?
No fingers on the wrist. No watch. No pulse oximeter. Just attention turned inward.
For decades, tasks built around this instruction were treated as a practical window into interoceptive ability. The appeal is obvious. A heartbeat is recurrent, measurable from the outside with electrocardiography, and sometimes felt from the inside. Compare reported beats with recorded beats and, in principle, the distance between the two looks like perceptual accuracy.
The difficulty is that a person can arrive at the right number without actually sensing every beat.
Heartbeat perception tasks are therefore one of the clearest examples of why measurement is not a technical afterthought. The task helps define the construct that researchers later claim to have measured.
What heartbeat perception tasks are
Heartbeat perception tasks are experimental procedures intended to assess some aspect of a person's access to cardiac signals. They include heartbeat counting or tracking tasks, discrimination tasks that ask whether external stimuli are synchronous with the heartbeat, confidence ratings, and newer procedures that manipulate or model cardiac signals more directly.
They do not all measure the same thing.
A counting task may combine bodily sensation with knowledge of one's usual heart rate, beliefs about how long an interval lasted, expectations about what the experimenter wants, time estimation and arithmetic. A discrimination task may depend on the timing of pulse-wave arrival at a body location, sensory integration, learning the task and tolerance for uncertainty. Confidence adds a metacognitive layer rather than another measure of raw sensitivity.
The phrase "heartbeat perception" should therefore be followed by a second question: By which task?
The Schandry heartbeat counting task
Rainer Schandry's 1981 work helped establish the heartbeat perception/counting paradigm. Participants were asked to count perceived heartbeats across time intervals without manually taking their pulse. The counted number could then be compared with recorded cardiac activity.
The task became popular because it is inexpensive, simple to administer and produces an intuitively meaningful score.
That simplicity contributed to its influence.
It also concealed several inferential problems.
If someone knows that their resting heart rate is around 60 beats per minute, a 30-second interval already supplies a plausible answer. If they estimate the duration of the interval, they can improve that guess without detecting individual heartbeats. People can also use vague bodily sensations, assumptions about fitness or age, and prior feedback.
A high score can therefore reflect a mixture of cardiac sensation and non-interoceptive knowledge.
Why the criticism became serious
Methodological work in the 2010s made the contamination problem difficult to ignore. Studies comparing the original counting instructions with instructions that required participants to report only heartbeats they actually felt found much lower performance under the stricter condition. Participants also described strategies such as estimating heart rate or relying on general bodily knowledge.
The implication was not that nobody can feel their heartbeat.
It was that the standard score does not cleanly tell us who can.
This distinction matters because heartbeat counting scores have been correlated with anxiety, emotion, meditation experience, body awareness and clinical symptoms. If the measure partly captures beliefs or knowledge, an association may be incorrectly interpreted as sensory accuracy.
A construct-validity problem can travel through an entire literature.
The debate is more nuanced than 'the task is invalid'
Critiques of heartbeat counting have themselves been debated. Statistical analyses of the scoring formula can create artifacts. Some criticisms depend on how heart rate and error ratios are modeled. Commentaries in 2020 argued over which objections were genuinely diagnostic and which were consequences of ratio variables or analytical choices.
This is scientifically healthy.
A method should not be discarded because a criticism is fashionable, and it should not be retained because it is traditional. The relevant question is which inferences survive direct tests of alternative explanations.
The strongest current position is not that heartbeat counting measures nothing. It is that its score is multiply determined and should not be described as a pure readout of interoceptive accuracy.
Heartbeat discrimination tasks
Another family of paradigms asks participants to judge whether external tones, lights or taps occur in synchrony with their heartbeat. These tasks reduce some opportunities for simple rate estimation, but introduce different problems.
The mechanical pulse reaches different body sites after the electrical R-wave recorded on ECG. People may feel the heartbeat in the chest, neck, fingertips or diffusely, each with different delays. A tone that is physiologically synchronous by one definition may not feel synchronous at the location a participant uses.
Performance also depends on task parameters, number of trials and how the experiment models chance.
Discrimination is therefore not an automatic gold standard replacing counting.
It is another operationalization with its own sensory and inferential demands.
Accuracy, sensibility and awareness
Modern interoception research often distinguishes multiple dimensions. Interoceptive accuracy refers to performance on an objective or performance-based task. Interoceptive sensibility refers to self-reported beliefs or tendencies concerning internal sensation. Metacognitive interoceptive awareness concerns how well confidence tracks performance.
Heartbeat tasks are useful partly because they made these distinctions necessary.
Someone can believe they are exceptionally attuned to their heart and perform poorly on a cardiac task. Another person can perform relatively well while reporting little confidence. These are different profiles.
The gap between confidence and performance may itself be more informative than either number alone.
Cardiac signals are only one interoceptive channel
Perhaps the most important boundary is easy to miss: heartbeat perception is not interoception in general.
Interoception includes information related to respiration, gastrointestinal state, temperature, pain, bladder fullness, metabolic need and other internal processes. Ability can differ across modalities. A person who is accurate at one cardiac task is not thereby demonstrated to have high global body awareness.
Historically, cardiac tasks became proxies for an entire field because they were convenient and physiologically recordable.
Convenience is not ontology.
The internal body is not one sense with one score.
Signal-detection and perturbation approaches
Newer work increasingly asks whether interoceptive tasks can separate sensitivity from response bias, manipulate signal strength, or perturb a bodily system rather than relying on passive detection.
Respiratory-load tasks, pharmacological perturbations, gastric procedures and computational models can provide different leverage. In cardiac work, adaptive psychophysics and signal-detection approaches aim to estimate thresholds and uncertainty more explicitly.
These approaches are not free from assumptions. They can be more expensive, less scalable and harder to compare across sites.
Their value is conceptual: they make the measurement model visible.
A task should state what sensory event is available, what decision the participant makes, what sources of prior knowledge remain, and how performance is separated from bias.
When self-report is the right measure
Not every research question requires a performance task.
If the question is how often someone notices bodily sensations, whether they trust them, whether attention becomes worried or avoidant, or how they describe their relationship with internal experience, self-report may be the correct method.
The mistake is not using self-report.
The mistake is calling a self-report scale objective sensory accuracy.
Likewise, the mistake is not using heartbeat counting when historical comparability matters. The mistake is interpreting the resulting score more broadly than its validity allows.
Good method selection begins with the intended claim.
Human-capacity research needs calibration, not a winner
The Institute is interested in whether people can notice internal change, distinguish signal from interpretation, calibrate confidence and use bodily information without becoming captive to it.
No single heartbeat task measures that complete capacity.
A defensible programme would combine methods: physiological recording to establish the signal, a performance task where appropriate, confidence ratings, self-report about attentional style, and contextual observation of how information is used.
This is more cumbersome than one score.
It is also closer to the phenomenon.
Reliability and validity are different questions
A task can produce similar scores on repeated occasions and still measure the wrong construct. That is reliability without sufficient validity. The reverse can also occur: a task may have a defensible connection to a sensory process but poor repeatability because the signal is weak, state dependent or sampled with too few trials.
Heartbeat paradigms face both problems. Resting cardiac sensations vary with posture, fitness, hydration, anxiety, medication and recent activity. A short laboratory session can therefore mix trait-like ability with current state. Researchers who want to compare individuals need enough trials and clear reporting of these conditions.
This is why a single impressive correlation does not validate a measure. Validation accumulates through converging evidence: known-groups comparisons, experimental manipulation, test-retest estimates, relations to other measures, and demonstrations that obvious alternative strategies do not explain performance.
Feedback changes the task
Some experiments provide participants with information about actual heartbeat timing or performance. Feedback can be useful for training and for testing whether participants can learn a mapping between sensation and the cardiac cycle. But once feedback is introduced, the study no longer measures only pre-existing perception. It also measures learning.
That distinction is valuable rather than inconvenient. If the aim is human-capacity development, learnability may matter more than baseline rank. The research question then becomes whether calibration improves and whether that improvement transfers beyond the practiced task.
A person who learns the laboratory timing rule without gaining useful access to ordinary bodily signals has improved at the task, not necessarily at everyday interoception.
Individual ranking requires a higher bar
Many heartbeat studies are designed to detect average differences between groups, not to classify a particular person. A measure can be adequate for a group-level experiment while being too noisy for individual diagnosis or selection.
This boundary is especially important when commercial systems turn research tasks into scores for wellness, recruitment or coaching. Before a task is used to tell someone who they are, its individual-level reliability, reference ranges, confounds and consequences of error need to be established.
Research convenience should not become a personality test by accident.
Reporting standards are part of validity
A paper should report the exact heartbeat task, interval lengths, instructions, feedback, exclusion rules, scoring formula, cardiac recording method and whether participants were asked about estimation strategies. Without those details, two studies carrying the same task name may not be comparable.
Researchers should also report heart rate itself. Because many counting formulas mathematically relate error to the number of recorded beats, group differences in resting rate can distort interpretation if ignored. Sensitivity analyses using alternative scoring or model-based methods can reveal whether a conclusion depends on one formula.
Transparent task reporting is not bureaucratic overhead. It is the minimum needed to know what a heartbeat score means.
Institute synthesis: measurement should preserve the chain
The Institute uses the phrase signal-to-claim chain for the sequence from physiological event, to sensory access, to task response, to score, to psychological interpretation, to practical claim.
This is Institute synthesis, not a standard term in heartbeat-perception science.
The point is simple. Error can enter at every link. An ECG may record the heart accurately while a behavioral score poorly isolates sensation. A behavioral score may be reliable while the research paper overgeneralizes from cardiac performance to whole-body awareness. A valid group effect may still be useless for judging an individual.
Preserving the chain makes overclaiming easier to see.
What the evidence does not show
Heartbeat counting does not provide a pure measure of interoceptive accuracy. A high score does not prove that a person consciously felt each beat. Heartbeat discrimination is not a universally uncontaminated replacement. Performance on one cardiac task does not establish general interoceptive ability. Self-reported body awareness cannot be substituted for objective cardiac performance, and objective performance cannot tell us whether attention to the body is helpful, distressing or wisely used.
The literature also does not support ranking people as globally "embodied" or "disembodied" from a heartbeat score.
These tasks are instruments, not character judgments.
Open questions
Can cardiac paradigms achieve better test-retest reliability without becoming detached from natural sensation? Which task features best separate sensory evidence from prior knowledge? How should pulse transit time be modeled in synchrony judgments? Do measures generalize across age, cardiovascular fitness and medication? What is the relationship between laboratory cardiac performance and the use of bodily information in real decisions?
A field matures when its favorite measure becomes an object of research rather than an unquestioned window.
Research references
- Schandry R. Heart beat perception and emotional experience. Psychophysiology. 1981;18(4):483-488. DOI: 10.1111/j.1469-8986.1981.tb02486.x.
- Desmedt O et al. The heartbeat counting task largely involves non-interoceptive processes: Evidence from both the original and an adapted counting task. Biological Psychology. 2018. PMID: 30218689.
- Corneille O et al. A heartfelt response to Zimprich et al.: Reconsidering the validity of the heartbeat counting task. Biological Psychology. 2020;152:107869. DOI: 10.1016/j.biopsycho.2020.107869.
- Zimprich D et al. Commentary on statistical and validity criticisms of the heartbeat counting task. Biological Psychology. 2020. PMID: 32061687.
- Garfinkel SN et al. Knowing your own heart: Distinguishing interoceptive accuracy from interoceptive awareness. Biological Psychology. 2015;104:65-74. PMID: 25451381.
- Khalsa SS et al. Interoception and Mental Health: A Roadmap. Biological Psychiatry: Cognitive Neuroscience and Neuroimaging. 2018;3(6):501-513. DOI: 10.1016/j.bpsc.2017.12.004.