Interoception, at the conscious level, is the perception of the body’s internal signals in response to various internal and external stimuli. Outside our consciousness, interoception describes the brain’s automatic formation of probabilities for sensory experiences within the body, given past experiences (Friston, 2010; Clark, 2013). As interoception research grows, new definitions arise to consider the multidimensionality of interoceptive abilities, from perceiving hunger and temperature to heart rate (HR) and metabolism (Khalsa and Lapidus, 2016). While Vaitl (1996) describes interoception as generally relating to proprioception and visceroception, Paulus (2013) defines interoception as the integration of information in the body with the central nervous system. There is no consensus on whether interoception should be measured in terms of attendance to internal signals (attention or awareness), perceived intensity of internal signals (magnitude), how well an individual interprets these signals (accuracy), believes they interpret sensations (sensibility), or views perceptions of the body as helpful or harmful (beliefs; Schandry et al., 1993; Simmons et al., 2012; Khalsa et al., 2015; Forkmann et al., 2016; MacCormack et al., 2022).

One approach utilized to explain interoception is allostasis, which describes the brain as a predictive organ that anticipates the body’s needs (Barrett and Simmons, 2015). Predictions based on probabilities from past experiences serve as hypotheses that the brain refines as sensory input is received. The brain does not merely react but anticipates information from the world and within the body, allowing for refined energy allocation. Prediction errors allow for the refinement of active inference, resulting in the production of more accurate visceromotor predictions. The body’s visceral experience is a combination of what is remembered of past and expected sensory experiences.

Changes in the body and brain influence emotion, yet we are not always perceptive of these physiological signals. Anger because of hunger, for example, is not always recognized by the “hangry” individual, and perception of emotion may not be correctly attributed to the body but other noise (MacCormack and Lindquist, 2016). One person may be sensitive to the somatic markers that assist emotion perception; however, another individual may experience elevated HR and sweaty palms yet not pay attention to these biomarkers (Bechara and Damasio, 2005). Somatic markers create unconscious changes in physiology that influence emotion and consequent decisions, reflective of interoceptive processes that are not perceived despite the body’s attempt to tell us (Bechara et al., 2005). Attention to these signals is associated with more intense emotional experiences that influence cognitive processes for better or worse, whether those be biases, judgments, or behaviors (Damasio et al., 1991; Garfinkel and Critchley, 2013).

Some individuals are not acute interoceptors and attribute decisions to “a gut feeling,” whereas others flag changes in physiology to explain attitudes. Being consciously interoceptive, termed interoceptive ability, allows one to interpret and regulate their body’s needs in response to hindrances to functioning. Possessing interoceptive ability has been examined in relation to various physiological and psychological experiences, such as those involving the gastrointestinal tract, hunger, eating disorders, stress, and cravings (Young et al., 2017; MacCormack and Lindquist, 2019; Khalsa et al., 2022). Increasing links to health illustrate the benefits of conscious interoceptive insight (Khalsa et al., 2018; Adams et al., 2022). Poor interoception, related to poor emotion regulation, contributes to the development of mental disorders, and individuals with depression or anxiety score lower on measures of interoceptive accuracy (IAcc) in comparison to non-clinical controls, such as the Heartbeat Counting Task (HCT) and Heartbeat Detection Task (HDT; Paulus and Stein, 2010; Herbert and Pollatos, 2014; Mata et al., 2015; Quadt et al., 2018; Martin et al., 2019).

About a third of individuals are cardiac interoceptive, perpetuating the need for methods that accurately measure HR detection (Khalsa and Lapidus, 2016). Various methods attempt to measure cardiac interoception, such as the HCT, which requires participants to count their heartbeats in timed intervals, or the HDT, which asks participants to synchronize a tone to their HR. Self-reports are then compared to an objective measure of HR to determine accuracy (Leganes-Fonteneau et al., 2021; Desmedt et al., 2022; Legrand et al., 2022). However, these tasks are flawed and likely invalid (Zamariola et al., 2018). The HDT fails to discriminate between interoceptive awareness (IAw) and IAcc. Murphy et al. (2019) describe IAcc as the ability to interpret signals correctly yet IAw as a metacognitive capability determined by correspondence between IAcc and how well an individual believes they perceive their body’s signals (i.e., interoceptive insight). Second, these tasks are vulnerable to bias, such as prior knowledge of resting heart rate (RHR; Windmann et al., 1999; Brener and Ring, 2016). Further, these tasks measure cardiac interoception during rest. Understanding interoception in contexts of discomfort (e.g., anxiety, hunger, dietary fasting) is significant for understanding the relationship between interoception and health (Khalsa et al., 2018; Rominger et al., 2021). Another limitation is distinguishing between subjective and objective measurement. Murphy et al. (2019) present a 2 × 2 model of interoceptive abilities to clarify if a measure is objective or subjective, as well as if it measures accuracy or attention (awareness). While some subjective measures of interoception specify which facet of interoception they measure (e.g., the Body Perception Questionnaire measures awareness), many do not.

Inspired by the CARdiac Elevation Detection (CARED) Task (Ponzo et al., 2021), the present study compares subjective ratings of IAcc to an objective measure of HR—though in the presence of art. Research examining the influence of art on cognition and perception (i.e., neuroaesthetics) has continued to grow since the 1960s when visual arts were the most common medium used to study aesthetics (Arnheim, 1966; Berlyne, 1971). Various factors contribute to aesthetic experiences, one being aesthetic appreciation. Curiosity for and enjoyment of music are related to individual differences in musical training, for example (Gerstgrasser et al., 2023). Although no research directly examines aesthetic appreciation and interoception, those trained in the arts are more interoceptive, suggesting a relationship between aesthetic interest and interoception (Christensen et al., 2017). Other research not specifically examining interoception finds that individuals considered experts in music experience more intense and diverse emotion when listening to music in comparison to non-experts (Gerstgrasser et al., 2023). Similarly, individuals with arts education experience heightened feelings of liveliness when viewing art (Miu et al., 2016). Given that emotion influences and is influenced by internal perceptions, aesthetic appreciation may moderate the relationship between emotion reactivity and interoceptive ability in the presence of art. Overall, research suggests that individuals with art training are more interoceptive and experience greater emotional responses during aesthetic experiences, which relate to greater physiological arousal. Further, greater physiological intensity is associated with heightened ability to notice changes in physiology (Ring et al., 2015; Körmendi et al., 2021), supporting our hypothesis that individuals reporting greater aesthetic appreciation are more interoceptive.

A contributor to increased interoceptive ability in those engaged in the arts may be reduction in the stress hormone cortisol. Previous research finds that gallery visits as short as 35 min significantly reduce cortisol and reports of stress, identifying cortisol as a primary contributor to dysregulated interoception (Clow and Fredhoi, 2006; Schulz and Vögele, 2015). Potentially, those with greater aesthetic appreciation are inclined to engage in aesthetic experiences and, therefore, experience reduced cortisol that improves bodily perceptions. Therefore, it may be that individuals that report greater aesthetic appreciation will perform better in a task that objectively measures their IAcc. Given art reduces cortisol, which inhibits interoception, we hypothesize that as time progresses in the presence of art, the odds of correctly perceiving one’s HR will improve.

When participants explore galleries unrestricted, self-reports of aesthetic-emotional experience significantly relate to changes in HR intensity, providing evidence of the power aesthetics have on the body and emotion (Tschacher et al., 2012). In-museum data collection is a new and promising method for capturing physiological change under ecological conditions where greater physiological arousal occurs in contrast to viewing artwork in the lab (Mastandrea et al., 2018; Krauss et al., 2021). Recent research examining aesthetic experience has measured physiological responses during live concerts, for example (Wald-Fuhrmann et al., 2021; Czepiel et al., 2023; Wald-Fuhrmann et al., 2023). Although experience sampling introduces a lack of controls, museums present a multi-sensory experience that extends the applicability of results to where they are most helpful (i.e., outside the lab).

In addition to examining the relationship between interoception and aesthetic appreciation, this study introduces a novel measure of interoceptive ability to further understanding of differences between subjective and objective measures of interoception, as well as between IAw and IAcc. The current study examines self-reports of physiological responses to an aesthetic experience in a naturalistic setting, a growing concept of data collection (Bannister and Eerola, 2023; Merrill et al., 2023).

Using the current study’s novel method, we are first interested in whether results obtained from the CARED Task are replicable. The CARED Task is a recent measure of interoceptive ability by asking participants to report their HR in the moment and outside the lab. We expect to replicate results from Ponzo et al. (2021) and find no significant relationship between objective IAcc (measured by the number of correct responses submitted by the participants during the task) and subjective IAcc [measured with the Interoceptive Accuracy Scale (IAS)]. If the current study obtains this result, there is further evidence that subjective measures of interoception are not measuring objective interoceptive ability. This result, as found by Ponzo et al. (2021), contrasts with Garfinkel et al. (2015) and Murphy et al. (2019), who obtained significant positive correlations between the IAS and objective IAcc with the HCT.

Similarly, we expect no relationship between objective IAcc and subjective IAw, measured by the Body Awareness (BA) subscale of the Body Perception Questionnaire Short Form (BPQ-SF). Our hypothesis is supported by current literature, and we expect our novel method to obtain a similar result, given argued reliability of past measures (Critchley et al., 2004; Garfinkel et al., 2015; Ferentzi et al., 2018). If the current study’s method replicates these results, there will be further evidence of significant differences in subjective and objective measures yet also awareness and accuracy. Regarding the BPQ and IAS, previous literature finds no relationship between these subjective measures of IAw and IAcc, and therefore we expect the same result (Murphy et al., 2020; Ponzo et al., 2021; Gabriele et al., 2022). However, we will utilize the BA subscale of the BPQ-SF to measure attention to physiological signals. The BPQ is a general measure of subjective IAw, yet there is internal reliability between its subscales (Cabrera et al., 2017).

Previous literature argues that participants’ confidence, or interoceptive insight, predicts objective IAcc (Khalsa et al., 2008). This is seen with the CARED Task and HCT (Murphy et al., 2020). We hypothesize a similar result where confidence will significantly predict the odds of submitting a correct response and positively relate to the number of correct responses. Given that the presence of art reduces cortisol, therefore improving interoception, we expect confidence to increase.

Participants were recruited from the university’s research participation system and were at least 18 years old. The present study recruited 65 participants, and this sample size is based on previous literature (Ponzo et al., 2021; Schuette et al., 2021; Legrand et al., 2022). Of the 65 participants, four were excluded from analyses due to excessive noise or loss of data, resulting in a final sample of 61 (34 female, age range: 18–22, M = 18.98, SD = 0.892).

The study occurred in the DeWitt Wallace Arts Museum. Once the Shimmer was attached, the researcher instructed participants to freely explore the gallery for 40 min. The participants held a smartphone with the application that randomly prompted the two questions. The Shimmer measured participants’ HR during the gallery exploration. At the end of the exploration, the application paused data collection and notified participants to return to the researcher; participants then completed the questionnaires on a tablet. While participants were at rest, the Shimmer resumed data collection for 15 min to determine RHR. If the questionnaires did not fill the entire 15-min period, participants read a copy of the university’s magazine.

To clean and analyze data, researchers used IBM SPSS Statistics for Windows 29.0.0.0. HR data was processed by band-passing with a Butterworth filter between 40 and 150 beats per minute (BPM) to remove low/high-frequency noise. To convert photoplethysmography (PPG) data to BPM, researchers used the window approach in the HeartPy framework, which ran peak detection over slices of time 1 min in length (van Gent et al., 2019a,b). Participants’ RHR was then calculated by averaging the final 15 min of HR data. To determine participants’ responses to the first question as correct, the times at which participants submitted responses were matched to the times that the Shimmer recorded their HR; responses were recorded as 0 for incorrect (false alarm/miss) or 1 for correct (correct rejection/hit). A participant’s HR was identified as higher than usual if their HR exceeded their RHR by 5 BPM. We chose a 5 BPM cutoff since this exceeds the average inter-individual variability in RHR for adults and is described as non-normal (Quer et al., 2020).