You are asked to sit quietly and count your heartbeats without touching your pulse.
Thirty seconds pass.
How many were there?
The experimenter has an ECG recording. Your report can be compared with the physiological signal. It looks like a beautifully simple test of how accurately you can sense your body.
Then the complications begin.
Did you actually feel each beat, or estimate from what you know about heart rate? Did the instruction encourage guessing? Did you count a pulse sensation in the chest, throat or fingertips? Would you perform the same way if asked to judge whether tones were synchronized with your heartbeat? Would the result generalize to breathing, fullness or bladder sensation?
These questions are why interoceptive accuracy deserves a page of its own.
The phrase sounds straightforward. The measurement science is not.
Definition
Interoceptive accuracy generally refers to objectively assessed performance in detecting, discriminating or tracking internal bodily signals.
In the influential 2015 framework proposed by Garfinkel and colleagues, interoceptive accuracy was distinguished from interoceptive sensibility, meaning self-reported beliefs or tendencies about bodily awareness, and interoceptive awareness, defined there as metacognitive correspondence between objective performance and confidence.
The distinction was important because a person can believe they are highly sensitive to bodily signals without performing especially well on a specific detection task, and can perform well without being well calibrated about that performance.
Objective performance, subjective belief and metacognitive insight are related questions.
They are not synonyms.
“Objective” does not mean theory-free
An objective task has an external scoring rule.
The ECG provides a physiological reference. A respiratory resistance is manipulated by the experimenter. A gastric stimulus is delivered at known intensity. The participant's response can be compared with a measurable event.
That is a major advantage over self-report when the research question concerns detection performance.
But every task still embeds assumptions.
What response counts as detection? Which signal is relevant? How much time is allowed? What information can the participant infer without directly sensing the body? How do attention, expectation, memory and decision bias affect the answer?
Objective scoring reduces some ambiguities.
It does not eliminate construct validity.
The heartbeat counting task
The heartbeat counting task became one of the most widely used measures of cardiac interoceptive accuracy.
Participants are instructed to count perceived heartbeats over several intervals while actual beats are recorded. A score reflects correspondence between counted and recorded heartbeats.
Its advantages are obvious. The task is simple, inexpensive and easy to compare across studies.
Its limitations are equally important.
Participants can use knowledge about typical heart rate or estimate elapsed time. Instructions differ across studies. Some people may report a plausible number without consciously detecting many beats. A high score can therefore arise partly from non-interoceptive strategies.
This does not render every heartbeat-counting result meaningless.
It means the task measures performance under a particular procedure, not pure access to the heart.
Discrimination tasks
Heartbeat discrimination tasks take a different approach.
Instead of counting, participants judge whether external tones or intervals correspond to their actual heartbeat. Some designs use forced-choice comparisons or manipulate timing relative to the ECG.
These tasks can reduce reliance on explicit heart-rate knowledge.
They also place demands on temporal discrimination, attention, decision criteria and integration between internal and external signals.
A person can therefore perform differently across counting and discrimination tasks for reasons that are scientifically meaningful.
The difference is not simply that one task is good and the other bad.
Signal detection approaches
Signal-detection methods try to separate sensitivity from response bias.
That distinction matters. Imagine two people who both report feeling many heartbeats. One truly discriminates cardiac signals well. The other has a liberal criterion and says "yes, I felt it" frequently.
If a measure only counts affirmative responses, the two can look similar.
Signal-detection analysis asks whether the person can distinguish signal from non-signal while separately estimating response tendency.
The cardiovascular signal detection task, or cvSDT, is one contemporary attempt to improve measurement in this direction.
A 2025 comparison exposes the problem
Ferentzi and colleagues directly compared three cardiac interoception tasks in 73 young adults: heartbeat counting, a multi-interval discrimination task and the cardiovascular signal detection task.
They expected the measures to relate if they captured overlapping cardiac interoceptive ability.
The result was only partly cooperative.
Heartbeat counting and cvSDT sensitivity were moderately related. The multi-interval task was unrelated to both. Classification of high performers also diverged. Under the study's criteria, only one participant qualified as a heartbeat perceiver across all three tasks.
The authors conclude that the tasks likely assess partly different abilities and that method should be chosen to fit the research question.
This is a central finding for the whole field.
If two measures do not agree strongly, researchers should not write as though “interoceptive accuracy” were one fixed number waiting to be discovered.
Accuracy of what?
Cardiac interoception dominates much of the measurement literature because the heart is accessible to laboratory monitoring.
But interoception spans multiple bodily channels.
A person who detects heartbeat timing well may not show equivalent accuracy for breathing resistance, gastric sensations, temperature or other internal signals.
Schoeller and colleagues' 2025 critique collects evidence for weak cross-modal relationships and argues that the field risks treating a family of abilities as one unitary construct.
The title of their paper, There is no such thing as interoception, is intentionally provocative.
The useful methodological question beneath it is quieter: what exactly is being claimed to generalize?
An accuracy score should ideally specify the bodily channel and task.
"Cardiac interoceptive accuracy on the cvSDT" is more precise than "good interoception."
Accuracy is not sensibility
Self-report measures such as the MAIA ask about patterns of attention, noticing, worry, self-regulation, body listening and trust.
They can be valuable.
They answer different questions.
Ferentzi's 2025 comparison found very limited correspondence between questionnaire measures and behavioral task performance. That is consistent with a wider literature in which self-reported interoceptive sensibility often correlates weakly with objective accuracy.
This should not be treated as proof that self-report is defective.
If two instruments measure different dimensions, weak correlation may be expected.
The error is naming both "interoception" without specifying which level is under discussion.
Accuracy is not awareness in the metacognitive sense
A person can perform accurately without knowing they are performing accurately.
Conversely, confidence can be high when performance is modest.
Garfinkel and colleagues used the term interoceptive awareness for the degree to which confidence tracks objective performance.
That is a metacognitive problem.
It parallels a broader distinction in judgment research: accuracy and calibration are not the same. A decision-maker can sometimes be correct for the wrong reasons or be unable to distinguish moments of strong knowledge from moments of weak knowledge.
Interoceptive research gives this distinction an embodied form.
What affects task performance?
Several variables can shape interoceptive-accuracy scores.
Physiology matters. Heart rate, body composition, signal strength and individual anatomy can affect perceivability.
Task structure matters. Forced choice, timing windows and instructions change what participants can use.
Cognition matters. Attention, expectation, prior knowledge and working-memory demands can contribute.
Decision bias matters. Some people use more liberal or conservative criteria.
Learning may matter. Repeated exposure to a task can alter strategy.
Clinical state, medication and anxiety can influence both physiology and interpretation.
This makes a one-number hierarchy of people from "low" to "high" interoception scientifically hazardous.
Does higher accuracy mean better functioning?
Not necessarily.
Greater ability to detect an internal signal can be useful. It may support early recognition of exertion, hunger, arousal or physiological change.
But more accurate detection does not guarantee adaptive interpretation.
A person can accurately notice a strong heartbeat and catastrophically interpret it. Another can detect bodily signals well and become hypervigilant. A third can have only moderate accuracy yet use contextual information effectively.
The functional outcome depends on what happens after detection.
This is why interoceptive accuracy should not be sold as a universal wellness score.
Clinical research needs dimension-specific claims
Interoceptive differences are studied across anxiety, depression, eating disorders, autism, addiction and other conditions.
The pattern is not simply "clinical groups have lower accuracy."
Findings vary by condition, bodily channel, task and age. The 2025 autism systematic review and meta-analysis, for example, reported no significant difference in adult cardiac accuracy in the subset of studies suitable for that comparison, while other interoceptive dimensions showed heterogeneous findings.
Clinical translation therefore needs specificity.
Which condition?
Which signal?
Which task?
Which dimension?
Which population?
Without those nouns, "interoceptive deficit" is often too broad to be useful.
Can accuracy be trained?
This is an attractive question and an easy place to overclaim.
If practice makes people feel more connected to their bodies, it is tempting to assume objective detection improved.
Some interventions may improve performance on specific tasks. But transfer from one practiced signal to another, or from a laboratory task to everyday functioning, needs direct evidence.
A change in MAIA self-report scores does not demonstrate improved cardiac detection accuracy. A change in cardiac-task performance does not automatically demonstrate better emotional judgment.
Training claims should name the outcome actually measured.
Accuracy can be local, temporary and trainable without becoming a trait
Another temptation is to treat an accuracy score as a stable personal property: I am good at interoception or I am bad at it.
Many perceptual abilities are conditional. Performance can vary with the strength of the signal, posture, exercise, attention, fatigue, stress, familiarity with the task and the bodily channel being tested. Even when a score has acceptable reliability, it remains a performance under defined conditions.
This has two consequences.
First, a low score should not become an identity. The task may be difficult, the signal may be weak, or the person's access may differ across modalities. Second, a training effect on one task needs to be tested for transfer. Practice can improve familiarity with a procedure without creating a broad change in bodily sensing.
The distinction mirrors learning research more generally. Improvement on the practiced test is not identical to a generalized capacity.
For interoceptive science, the strongest training claim therefore names the exact outcome: performance improved on this measure, for this signal, under these conditions, over this period. Anything broader needs additional evidence.
Accuracy, instruments and disagreement
External sensors introduce a further complication. A physiological device can provide a more precise measure of a signal than subjective perception, but the device itself has a measurement model. Consumer wearables estimate some states from proxies rather than measuring the target directly. Research-grade ECG is excellent for detecting cardiac electrical activity, yet even perfect cardiac measurement would not determine what a sensation means psychologically.
When subjective sensation and instrumentation disagree, the discrepancy is data. It may reveal perceptual limits, device error, temporal mismatch, or a difference between the physiological quantity being measured and the experience the person is reporting.
A mature concept of interoceptive accuracy leaves room for that disagreement instead of deciding in advance that either the body or the device must always win.
What the evidence does not show
Current evidence does not show that:
- there is one universally accepted gold-standard test of interoceptive accuracy;
- heartbeat counting measures pure bodily perception without cognitive contribution;
- strong performance on one cardiac task guarantees strong performance on another;
- cardiac accuracy generalizes automatically to other bodily channels;
- self-reported body awareness is a substitute for objective performance;
- higher interoceptive accuracy is always psychologically healthier;
- low accuracy on one task is a diagnosis;
- improved subjective body awareness proves improved objective accuracy;
- greater accuracy makes external-world intuitions more trustworthy;
- an individual should use an interoceptive task score as a medical or psychological verdict.
Institute synthesis: accuracy needs a name and a denominator
Institute synthesis: embodied intelligence becomes more credible when it resists vague superlatives.
Instead of saying, "This person has excellent interoception," ask:
Accurate at detecting what?
Under which task?
Against which physiological reference?
With what decision bias?
How stable is the performance?
Does the person know when they are accurate?
Does the information improve action outside the task?
This is not skepticism about embodiment.
It is what taking embodiment seriously looks like when measurement enters the room.
From laboratory score to technological future
Wearables and consumer sensors increasingly provide external estimates of internal state.
Heart rate, heart-rate variability, sleep staging, temperature and respiratory proxies appear on wrists and phones. These systems create a new relationship between interoceptive accuracy and external measurement.
A person may feel fine while the device reports strain. They may feel exhausted while a readiness score says "recovered." Which source should dominate?
The scientific answer cannot be "trust your body" or "trust the device" in the abstract.
Both have measurement error. Both require calibration to the decision at hand.
The future of embodied intelligence may therefore involve learning when internal sensation, behavioral performance and external instrumentation agree, and what to do when they do not.
That is a richer question than accuracy alone.
Editorial limitation note
Interoceptive-accuracy tasks are research measures, not diagnostic instruments for self-use. The page avoids telling readers to ignore symptoms because a laboratory score is low or to treat a high score as proof that bodily interpretation is correct.
Research references
- Garfinkel SN, Seth AK, Barrett AB, Suzuki K, Critchley HD. Knowing your own heart: distinguishing interoceptive accuracy from interoceptive awareness. Biological Psychology. 2015;104:65-74. https://doi.org/10.1016/j.biopsycho.2014.11.004
- Ferentzi E, Vig L, Gerlach AL, et al. Cardiac Interoceptive Accuracy: An Empirical Comparison of Three Ability Measures. Psychophysiology. 2025;62(6):e70078. https://doi.org/10.1111/psyp.70078
- Greenwood BM, Garfinkel SN. Interoceptive Mechanisms and Emotional Processing. Annual Review of Psychology. 2025;76:59-86. https://doi.org/10.1146/annurev-psych-020924-125202
- Khalsa SS, Adolphs R, Cameron OG, et al. Interoception and Mental Health: A Roadmap. Biol Psychiatry Cogn Neurosci Neuroimaging. 2018;3(6):501-513. https://doi.org/10.1016/j.bpsc.2017.12.004
- Schoeller FA, Zhang B, Garcia T, Reggente N. There is no such thing as interoception. Frontiers in Psychology. 2025;16:1488415. https://doi.org/10.3389/fpsyg.2025.1488415