Interoception is broadly defined as the processing of internal bodily states at multiple conscious and preconscious levels (Brewer et al., 2021). Because of the breadth of this construct, many measures exist to probe individual differences in interoception (Desmedt et al., 2023; Khalsa et al., 2018). Measures of interoception can be separated into different domains (e.g., cardiac, respiratory, gastric) and dimensions (e.g., accuracy, attention). Some of the most assessed domains of interoception, and associated measures, are reviewed below. We focus on interoceptive accuracy, as well as self-reported, implicit and neural measures of interoception, however, it should be noted that there are many other dimensions (Khalsa et al., 2018). Specific considerations for the measurement of stress alongside interoception are provided.

​

Background

A large body of work indicates an association between physical symptoms (including interoceptive signals) and stress. However, studies using objective measures of interoceptive accuracy (the ability to accurately perceive internal bodily states) report mixed findings concerning associations with stress, and results are difficult to interpret due to the limitations of existing measures of interoceptive accuracy (Schulz & Vögele, 2024; Irigoras Izagirre et al., in Prep). Objective measures of interoceptive accuracy – particularly cardiac interoceptive accuracy – do not separate perceptual changes from stimulus strength and rarely employ control tasks (Desmedt et al., 2023; Murphy, 2023). It is therefore difficult to determine the mechanism underlying changes following stress, given that they may be driven by physiological, perceptual, or cognitive (e.g., attentional) changes. Much work of this type, as well as work that has related objective measures of interoception to self-report measures of stress, has utilised estimation tasks. These tasks are limited as they are influenced by prior knowledge and beliefs regarding physiology (e.g., stomach capacity, typical heart rate etc.; Desmedt et al., 2023). The relationship between interoception and stress therefore remains unclear.

Measurement

Interoceptive accuracy

Interoceptive accuracy is defined as the ability to accurately perceive internal bodily states as measured by behavioural tests. For a detailed review of new measures, see Desmedt et al., (2023). A general limitation of all measures of interoceptive accuracy is the difficulty separating changes in stimulus strength from perception. This is particularly important for the measurement of stress-related changes in interoception. Where this is not possible, detailed physiological assessment in the domain of interest is required to ensure that changes in performance can be attributed to perceptual changes (see Murphy & Bird, 2025). We recommend that estimation tasks (e.g., water-load test, heartbeat counting task) are avoided. Tasks that involve the perturbation of the body’s state (typically respiratory tasks, but also gastric and cardiac; e.g., Garfinkel et al., 2016; Khalsa et al., 2009; Khalsa et al., 2022; Mayeli et al., 2021) provide some control over stimulus strength, but individual differences (e.g., in respiratory behaviour: rate, depth etc.) still require consideration, and these measures are not always appropriate due to invasiveness (Desmedt et al., 2023). For multi-sensory integration tasks that require participants to match an external stimulus to an internal bodily signal (e.g., a tone to one’s heartbeat), it is critical that such tasks are assumption free (Brener & Ring, 2016); for example, assumptions regarding the timing at which an individual perceives a tone to be synchronous with their heartbeat (e.g., predefining a “correct” answer) are best avoided (for a task that overcomes this issue see Palmer et al., 2025). For all measures, control tasks are essential for ensuring that any changes or individual differences in performance can be attributed to interoception (Murphy, 2023). Differences in task requirements - such as general cognitive demands - can strongly influence results in tasks of interoceptive accuracy, making the inclusion of control measures essential. Beyond this, evidence of state effects on interoception highlights the need for assessment on multiple occasions to adequately quantify individual differences and identify stable “trait” aspects (Murphy, 2023; Wittkamp et al., 2018). This is particularly important when measuring stress in parallel, which itself fluctuates over time.

​

Self-reported interoception

There are a huge number of self-report measures of interoception. Importantly, these measures are generally uncorrelated and thus careful selection and interpretation is required (Desmedt et al., 2022). Measures vary in a several ways – for example, in the domains they reference (e.g., cardiac vs. gastric), whether statements are positive or negative (“I hate to be too hot or too cold”), and the examples provided (e.g., “Feeling a knot in your stomach”, “I feel a burning sensation in my stomach”), all of which can influence individual differences (Clemente et al., 2024; Trevisan et al., 2021). Certain terms (e.g., “awareness of” or “sensitivity to” bodily sensations) can also be interpreted differently across participants, and thus self-report measures using such terms are best avoided (Gabriele et al., 2020). Researchers are recommended to avoid self-report measures that conflate multiple dimensions of interoception (e.g., attention vs. accuracy vs. evaluation), as such measures make it difficult to determine which specific facet of interoception is driving observed effects (Clemente et al., 2024; Gabriele et al., 2020). When utilising multiple measures, ensuring that the physiological states referred to are matched is ideal (see, for example, the interoceptive accuracy and attention questionnaires; Gabriele et al., 2020; Murphy et al., 2020). If relating a self-report interoception measure to a self-report measure of stress, or if assessing interoceptive interventions, careful selection of measures is required as many stress questionnaires also include statements referring to bodily states (for a similar discussion related to anxiety, see Clemente et al., 2024). To ensure independence of measurement, overlapping item content across measures designed to assess different constructs should ideally be avoided. Where this is not possible, at a minimum researchers should conduct sensitivity analyses to ensure that results are not driven by a lack of independence of measurement and/or to ensure changes are not solely driven by changes in physiological stress variables (if examining interventions). For studies seeking to examine changes in self-reported aspects of interoception over time, Experience Sampling measures may be more appropriate as they are generally more sensitive to change (Poerio et al., 2024). These daily sampling measures are generally uncorrelated with retrospective measures, in the same way that self-report measures are generally uncorrelated with measures of behavioural accuracy.

​

Implicit measures

Within the field there are several other measures thought to assess more implicit (or preconscious) aspects of interoception. These include neural measures such as the heartbeat evoked potential (HEP) and cardiac timing measures. However, such measures remain the subject of ongoing debate due to wide variation in the implementation, pre-processing, time windows and spatial locations selected etc., making them difficult to review (for relevant papers see Caparco et al., 2024; Coll et al., 2021). For the HEP, the extent to which changes and/or individual differences reflect variation in cognition (i.e. perception, attention) rather than changes in cardiac dynamics remains an unanswered question. Beyond this there are also some fMRI measures (see Desmedt et al., 2023; Suksasilp & Garfinkel, 2022, for discussion). These generally involve either 1) the presentation of an external stimulus that is synchronous or asynchronous with the body’s state (e.g., heartbeats; but caution is required given assumptions regarding cardiac timing as previously discussed) or 2) attention towards or away from the body (but this can be difficult to interpret given that attention in such circumstances will be influenced by perception for those who can perceive such bodily signals).