The Interoceptive Brain: Confidence Ratings and Accuracy Scores are Independently and Differently Associated With Task‐Related Alpha Power During the Heartbeat Tracking Task

ABSTRACT

Cardiac interoception is important for health and can be assessed in terms of accuracy (IAcc) and sensibility (IS), at least. While IAcc measures the correspondence between recorded and perceived heartbeats, IS means the confidence in interoceptive perceptions during the task. The present study investigated if brain activity during the heartbeat tracking task is associated with IAcc as well as IS. Specifically, we were interested if task‐related power (TRP) in the alpha band (8‐12 Hz), known to indicate task‐specific cognitive functions such as semantic, attentional, and sensory processes, is associated with IAcc and IS, respectively. In a sample of 30 participants, we found relatively higher TRP in the alpha band over left temporal and parietal areas (vs. right) during the interoception task. Furthermore, we observed a negative association between TRP in the alpha band and IS. Lower TRP in the alpha band might indicate that more pronounced cognitive and sensory processes are linked to higher IS. Furthermore, we found a positive effect for IAcc (independent from IS), which might indicate that more internal attention during the interoception task is beneficial for IAcc. We further discuss the findings in the context of methodological issues of the heartbeat tracking task. Taken together, the pattern of findings favors the investigation of task‐related IS (i.e., confidence ratings) in combination with IAcc to gain a better access to interoceptive processes and to improve our understanding of the neural underpinnings of (cardiac) interoception.

Short abstract

This research has two main findings. First, it indicates the general brain activity pattern during cardiac interoception (assessed via the heartbeat tracking task). Second, it shows that interoceptive sensibility is linked with this TRP pattern in alpha band, which is independent from interoceptive accuracy. It indicates that interoceptive sensibility is less driven by internal directed attention per se but by a generally more active processing mode of sensory information.

1. Introduction

Interoception is the health relevant capacity to perceive internal body states such as temperature, pain, pulse, or breath (Ibanez and Northoff 2024; Khalsa and Lapidus 2016; Nord and Garfinkel 2022). Consequently, higher interoceptive skills are associated with better self‐regulation, self‐control, higher emotion regulation skills, empathy, well‐being, as well as parameters of cognition and health (see e.g., Ibanez and Northoff 2024; Rominger and Schwerdtfeger 2023; Zamariola et al. 2018). Reduced interoceptive skills, on the other hand, are linked with several psychiatric disorders like schizophrenia and depression (for review see Khalsa et al. 2018; Khalsa and Lapidus 2016). Beyond that, research has shown that health‐relevant behavior such as fasting, body‐scan, and breathing interventions can improve interoceptive skills (Herbert et al. 2012; Leganes‐Fonteneau et al. 2021; Rominger, Weber, et al. 2021; Schwerdtfeger and Rominger 2024; for null‐effect of a breathing intervention see e.g., Rominger, Graßmann, et al. 2021).

However, interoception is a multidimensional phenomenon. Garfinkel et al. (2015) suggested three main dimensions of interoception and differentiated between interoceptive accuracy (IAcc), the correspondence between body signals and subjective perception; interoceptive sensibility (IS; Desmedt et al. 2022), the subjective confidence in interoceptive perceptions and performances; as well as interoceptive metacognition, which is the correspondence between the IAcc and the IS (for more dimensions and other aspects of interoception such as interoceptive attention see e.g., Murphy et al. 2019; Suksasilp and Garfinkel 2022). All these dimensions have value and provide unique information about people's interoceptive skills and potentials. As many aspects of interoception are conscious and others are unconscious (Nord and Garfinkel 2022), IAcc as well as IS might capture both conscious and unconscious facets of interoception differently (Suksasilp and Garfinkel 2022). Therefore, confidence ratings assessed from trial to trial, which fall under the category of IS, might provide a better subjective access to the quality of momentary interoceptive processes. Confidence ratings might therefore capture more conscious interoceptive processes (as compared to IAcc).

Importantly, most empirical research is based on interoception in the cardiac domain as a proxy for interoception in other domains (such as stomach, respiration, body temperature, etc.). These cardiac interoception paradigms are heavily criticized (Ainley et al. 2020; Desmedt et al. 2020, 2018; Ferentzi et al. 2022; Jones 1994; Paulus et al. 2019; Schulz and Vögele 2023). However, most of these critics target the validity of the IAcc score (Desmedt and van den Bergh 2024), and to the best of our knowledge, research has not provided many concerns about the validity of task‐related trial‐to‐trial assessment of IS ratings—that are confidence ratings (Garfinkel et al. 2015; Murphy et al. 2020; Suksasilp and Garfinkel 2022; but see Desmedt et al. 2022 for critics of IS questionnaires in general). Besides the fact that confidence ratings were seldomly applied (but see e.g., Leganes‐Fonteneau et al. 2021; Murphy et al. 2020; Rominger, Weber, et al. 2021), subjective measures of interoceptive skills might serve their purpose in terms of a reliable assessment of subjectively felt confidence in task‐related interoceptive performance. In line with this, research indicated that people can express their confidence in their interoceptive performances well (Garfinkel et al. 2015; Murphy et al. 2020; Rominger and Schwerdtfeger 2023). However, these performance‐related interoceptive confidence ratings and the neurophysiological underpinnings of these subjective (as well as objectified) interoceptive processes are highly understudied (Mai et al. 2018). Therefore, the main aim of the present study was to investigate the task‐related brain activity pattern (in terms of the task‐related power [TRP]; Gerloff 1998) associated with these two distinct dimensions of interoception—that are IAcc and IS—independently from each other.

Most neurophysiological interoception research studied IAcc by means of fMRI and EEG (but see e.g., Candia‐Rivera et al. 2022 for using fNIRS). While fMRI studies often reported a link between the (anterior) insula, the ACC, and the amygdala with interoception (Craig 2009; Paulus et al. 2019), most EEG studies investigated event‐related potentials (ERPs) such as the heartbeat‐evoked potentials (HEPs), which are negative deflections observed between 200 ms and 500 ms after the R‐wave (Couto et al. 2014; Hodossy et al. 2021; Petzschner et al. 2019). People with higher IAcc show higher HEPs (Mai et al. 2018; Pollatos et al. 2005; Pollatos and Schandry 2004 for review see Coll et al. 2021). HEPs increased when IAcc increased (Richter et al. 2020). In agreement with this, Petzschner et al. (2019) found higher HEPs when attention was allocated to the own heartbeats (see also Fittipaldi et al. 2020; Simmons et al. 2013; Villena‐González et al. 2017; but see Schulz et al. 2015) and Couto et al. (2014) found reduced HEPs when attention was allocated to external information. To sum up, the positive association between HEPs and IAcc argues for the assumption that more internal attention (directed to the own heartbeats) benefits interoception, which strengthens the relevance of cardiac interoceptive attention (Murphy et al. 2019; Suksasilp and Garfinkel 2022).

Beyond the well‐documented meaning of HEPs for attentional processes during interoception, the gating by inhibition theory of Jensen and Mazaheri (2010) and similar theoretical frameworks developed by Fries and Klimesch (Fries 2015; Klimesch et al. 2007) suggest that brain oscillations in the lower frequency bands—such as the alpha band (8‐12 Hz)—are indices of cortical inhibition, which are important for (sustained) attention (Clayton et al. 2015; Hanna et al. 2024; Rihs et al. 2007; Rominger, Fink, et al. 2019; van Diepen et al. 2016). Jensen et al. (2002) showed that higher internal attentional demands were associated with increased alpha power at posterior and occipital areas (see also Benedek et al. 2014). This posterior increase of alpha (from a rest to a task) indicates a lower involvement of these specific brain regions during the task, arguing for less sensory input saving neurocognitive resources for internal (mental) processing. Neurophysiological studies investigating the execution of movements (Gallicchio and Ring 2019), creativity (Fink and Benedek 2014), working memory (Benedek et al. 2014), mental imagery (Fink et al. 2018; Rominger et al. 2018), motor imagery (Fink et al. 2018), humor comprehension (Perchtold‐Stefan et al. 2020), resting state conditions (Rominger et al. 2024), and emotion regulation (Fink et al. 2017) reported findings well in line with these theories and the functional meaning of TRP in the alpha band.

Therefore, we investigated the mechanisms of cardiac interoception by means of TRP (Gerloff 1998) in the alpha band (see also Petzschner et al. 2019). TRP in the alpha band can precisely assess the task‐specific brain activity pattern linked to relevant cognitive processes such as (internal) attention (Benedek et al. 2014). To the best of our knowledge, only one study applied a similar approach to investigate cardiac interoceptive attention (Murphy et al. 2019; Suksasilp and Garfinkel 2022) by means of alpha power changes. Villena‐González et al. (2017) found a positive association between IAcc and alpha power increases from an external to an internal attention task (at posterior sites). However, the authors did not apply the well‐established TRP approach, assessing the alpha power change from rest to task (Gerloff 1998). Furthermore, Villena‐González et al. (2017) used a modified version of the heartbeat tracking task (Schandry 1981) complicating a direct comparison with previous findings. Most important for the aim of the present study, however, the authors did not assess trial‐to‐trial confidence ratings during the interoception task and therefore did not study the unique brain activity (in terms of TRP in the alpha band) associated with both interoceptive dimensions—IAcc and IS.

Therefore, the present study aimed to conceptually replicate and extend the findings of Villena‐González et al. (2017). We assessed IAcc as well as IS (confidence ratings) during performing an interoception task (i.e., heartbeat tracking task; Schandry 1981; see also Murphy et al. 2020). Based on the literature, we assumed that both dimensions of interoception should be linked with TRP in the alpha band. Additionally, we explored if the effects would extend to TRP in the adjacent beta (12‐30 Hz) and the theta (4‐8 Hz) band. To sum up, we aimed to study (1) the general neurophysiological activation pattern of interoception under instruction (vs. habitual interoceptive tendencies; see Suksasilp and Garfinkel 2022) by means of the heartbeat tracking task (Schandry 1981) and (2) investigate if this specific activation pattern (in terms of TRP in the alpha band) is uniquely related to IAcc and IS reflecting two important interoceptive dimensions.

2. Methods

2.1. Participants

In total, 35 participants took part in this study. Four participants were excluded from analyses due to technical problems and the loss of interoceptive performance measures. Therefore, the final sample size was 30 (13 men), which shows a power of 0.80 with alpha = 0.05 to detect medium‐sized effects of d = 0.53 (t‐test for dependent means). Participants were 25.47 years on average (SD = 4.45). Individuals with a history of major psychiatric disorders and individuals who reported having a neurological disease or using psychoactive medication (assessed via self‐report) were not included in the study. All participants were right‐handed and were requested to refrain from alcohol for twelve hours and from coffee and other stimulating beverages for two hours prior to their lab appointment, and to come to the session well rested. All participants provided informed consent. The authorized ethics committee approved the study (GZ. 39/72/63 ex 2019/20). All data and syntax are openly available at https://osf.io/2kfsx/?view_only=a8c1e63da08744628a2095b646caf74f.https://doi.org/10.17605/OSF.IO/2KFSX

2.2. Cardiac Interoception Task (Heartbeat Tracking Task)

Following Schandry (1981), participants counted their perceived heartbeats (HB) during seven randomly presented time intervals (20, 25, 35, 45, 55, 65, and 75 s; see Hickman et al. 2020; Rominger and Schwerdtfeger 2023). The ECG was recorded with the Accusync172 ECG Trigger Monitor in combination with the Biopac MP150 amplifier system (1000 Hz) running AcqKnowledge 4.3 (standard lead II configuration). Participants should report only the truly experienced heartbeats (for a modified instruction see Desmedt et al. 2020) and were not allowed to feel their pulse or to do any physical manipulation, which was verified via cameras. After each trial, participants typed the number of perceived heartbeats and rated their confidence in task performance (IS) without time limit on a visual analog scale (VAS) from 0–100. Participants received no performance feedback. The IAcc score was calculated by the formula: $\text{IAcc}=\frac{1}{n}\sum \left(1-\frac{|{\mathit{HB}}_{\text{actual}}-{\mathit{HB}}_{\text{perceived}}|}{{\mathit{HB}}_{\text{actual}}}\right)$, with M = 0.56 (SD = 0.29). The mean IS was 42.93 (SD = 21.63; see Table 1).

TABLE 1.

Pearson correlations and descriptive statistics for all behavioral and demographic variables.

	IAcc	IS	diaBP	sysBP	BMI	Age	Gender	M	SD
IAcc		0.587 (0.001)	0.241 (0.200)	0.534 (0.002)	0.173 (0.361)	0.094 (0.622)	0.544 (0.002)	0.56	0.29
IS	0.587 (0.001)		0.181 (0.338)	0.494 (0.006)	0.184 (0.329)	0.087 (0.648)	0.453 (0.012)	42.92	21.63
diaBP	0.241 (0.200)	0.181 (0.338)		0.665 (< 0.001)	0.006 (0.974)	−0.052 (0.785)	0.019 (0.922)	68.50	9.11
sysBP	0.534 (0.002)	0.494 (0.006)	0.665 (< 0.001)		0.313 (0.092)	0.032 (0.868)	0.486 (0.006)	119.63	16.16
BMI	0.173 (0.361)	0.184 (0.329)	0.006 (0.974)	0.313 (0.092)		0.541 (0.002)	0.495 (0.005)	22.56	3.50
Age	0.094 (0.622)	0.087 (0.648)	−0.052 (0.785)	0.032 (0.868)	0.541 (0.002)		0.307 (0.099)	25.47	4.45

Note: p‐values are presented in parentheses. Gender: women = 1 and men = 2.

Abbreviations: BMI, body mass index; diaBP, diastolic blood pressure; IAcc, interoceptive accuracy; IS, interoceptive sensibility; sysBP, systolic blood pressure.

2.3. EEG Recordings and Analysis

We recorded participants' EEG via 19 electrodes (Fp1, Fp2, F3, F4, F7, F8, C3, C4, T7, T8, P3, P4, P7, P8, O1, O2, Fz, Cz, Pz), positioned according to the 10–20 system (Brainvision BrainAmp Research Amplifier, Brain Products; 500 Hz sampling rate) in a separate and quiet room. All participants should keep their eyes open during the task, which was monitored via a camera. The ground electrode was located on the forehead, and the reference electrode was on the nose. Vertical and horizontal electrooculograms were measured with two bipolar channels for horizontal and vertical eye movements. Electrode impedances were kept below 5 kΩ for all electrodes. The EEG signals were re‐referenced to an averaged ear reference (Yao et al. 2019). We preprocessed the data by removing drifts (0.1 Hz) and applying a Notch filter (50 Hz) by means of the Brainvision Analyzer (Brain Products). We applied a combination of visual and semiautomatic ICA‐based correction for artifacts. We calculated the ICA implemented in Brainvision Analyzer (Brain Products) to detect and correct artifacts by excluding the component representing eye movements and blinks, and we visually checked (by eyes) the signal for artifacts and excluded them if necessary.

After these corrections, we calculated the band power values (μV²) by squaring the filtered EEG signals (8–12 Hz; FFT‐filter with a window size of 250 samples and an overlap of 249 samples; for similar approaches see e.g., Jauk et al. 2012; Rominger et al. 2020) by means of the g.BSanalyze software (g.tec, Graz, Austria). Only band power values from artifact‐free time intervals were averaged by means of the median. For the TRP analyses, the interval from 500 ms after the onset of the fixation cross until 500 ms before its offset served as the reference interval, and the period starting 250 ms after interoception interval onset until 250 ms before the offset served as the activation interval (see Figure 1, Rominger, Papousek, et al. 2019). Only trials with at least two valid seconds entered statistical analyses.

FIGURE 1.

Schematic display of the computerized heartbeat tracking task. A 4 s time period out of the 5 s fixation cross period (“Reference”) served as the reference interval (leaving out the first and the last 500 ms). The time period starting 250 ms after the interval onset until 250 ms before the offset (“Interoception period”) served as the activation interval. After their answer, the participants rated the confidence in their interoceptive perceptions (IS) on a VAS from 0–100 (very low confidence—very high confidence). Then the next trial started.

TRP scores were quantified for alpha power (8–12 Hz) for an electrode i by subtracting the log‐transformed power of the reference period (Pow _i,reference) from that of the activation period (Pow _i,activation), according to the formula: TRP _i  = Median (log(Pow _i,activation) _j —log(Pow _i,reference) _j ).

Negative values indicate a relative decrease of task‐related alpha power from the reference to the activation period, while positive values express a relative power increase.

2.4. Procedure

After informed consent, we equipped all participants with the EEG and ECG systems. Before working on the heartbeat tracking task, participants' blood pressure, height, and weight were measured. Following the heartbeat tracking task, further tasks not relevant to this study were conducted. After these tasks, participants were debriefed and received course credits for participation.

2.5. Statistical Analysis

For behavioral results, we calculated Pearson correlations (r) to indicate the association between IAcc and IS, as well as further relevant characteristics of the sample, such as age, gender, BMI, and blood pressure.

For neurophysiological results, we first investigated the general neurophysiological activation pattern associated with interoception by means of an 8 × 2 analysis of variance with the within‐subjects factors AREA (eight electrode positions Fp12, F34, F78, C34, T78, P34, P78, O12) and HEMISPHERE (left, right). This indicates the general spatial patterns of brain activity associated with cardiac interoception during the heartbeat tracking task (for similar approaches in other contexts see e.g., Jauk et al. 2012; Jausovec et al. 2006). Second, we used the same analysis of variance design; however, we additionally considered the continuous between‐subjects factors IAcc and IS. This allows us to investigate the impact of individual differences in interoceptive skills along these two dimensions on the TRP in the alpha band. Before statistical analyses, we checked relevant assumptions (e.g., dependency of variables, outliers, sphericity violations). We used the Greenhouse–Geisser correction in case of sphericity violations. Estimates of effect size are reported using partial eta‐squared (${\eta }_p^2$). We used a significance level of p < 0.05 (two‐tailed) for all analyses, which were calculated with SPSS 28. Bonferroni corrected t‐tests served as post‐tests. Brain maps were built with the FieldTrip toolbox (Oostenveld et al. 2011).

3. Results

3.1. Behavioral Results

IS was significantly associated with IAcc. Furthermore, IAcc and IS were significantly associated with systolic blood pressure (sysBP). Both measures were not related to BMI and age (see Table 1). However, gender showed significant effects, with higher IAcc and IS obtained scores for men.

3.2. EEG Results

3.2.1. Alpha Power During Cardiac Interoception

The 8 × 2 analysis of variance revealed a significant interaction AREA × HEMISPHERE (F(4.621,133.997) = 4.565, p < 0.001, ${\eta }_p^2$ = 0.14). The Bonferroni corrected t‐tests indicated a significant difference between T7 and T8 (p = 0.018) as well as P7 and P8 (p = 0.002). The difference between P3 and P4 showed a trend for a difference in TRP in the alpha band (p = 0.057). All other differences between positions were non‐significant (p >0.057). As illustrated in Figure 2, the TRP in the alpha band at more posterior sites of the left hemisphere (i.e., T7, P7, P3) was relatively higher (and even positive) compared to the positions of the right hemisphere (T8, P8, P4), which showed negative TRP changes (see Figure 2).

FIGURE 2.

General pattern of TRP (with error bars representing standard errors) in the alpha band during the heartbeat tracking task.

The main effect HEMISPHERE slightly failed to reach significance (F(1,29) = 4.137, p = 0.051, ${\eta }_p^2$ = 0.13), with lower TRP scores for the right hemisphere (M = −0.037, SD = 0.024) compared to the left (M = −0.011, SE = 0.024). The main effect POSITION was not significant (F(3.457,100.252) = 1.604, p = 0.187, ${\eta }_p^2$ = 0.05).

The main analysis with IAcc and IS as continuous between‐subjects factors showed a significant main effect for IS (F(1,27) = 12.412, p = 0.002, ${\eta }_p^2$ = 0.32) as well as for IAcc (F(1,27) = 4.476, p = 0.044, ${\eta }_p^2$ = 0.14). As illustrated in Figure 3, people who indicated a higher confidence in their interoceptive performance showed stronger TRP decreases in the alpha band during the interoception trials (rs = −0.561, p = 0.002). Furthermore, higher IAcc was associated with increased alpha power during the task (rs = 0.377, p = 0.044) independent of IS (and vice versa; see Figure 3).

FIGURE 3.

Task‐related power (TRP) in the alpha band and as a function of IS and IAcc separately as well as the partial correlation.

No further effects involving IS or IAcc were significant.

3.2.2. Additional Analyses

When additionally controlling for systolic blood pressure, BMI, gender, and age, the main effect IAcc (F(1,23) = 4.878, p = 0.037, ${\eta }_p^2$ = 0.18) and IS remained significant (F(1,23) = 10.953, p = 0.003, ${\eta }_p^2$ = 0.32). When calculating separate ANOVAs for IS and IAcc as covariates, only IS reached significance (F(1,28) = 7.062, p = 0.013, ${\eta }_p^2$ = 0.20), while IAcc did not (F(1,28) = 0.02, p = 0.961, ${\eta }_p^2$ = 0.00; see Figure 3). This indicates that only the unique variance of IAcc (independent from IS) was linked with relatively higher TRP in the alpha band.

Furthermore, the main effects of IAcc and IS were not significant for the alpha activity during the reference period (IAcc: F(1,27) = 1.543, p = 0.225, ${\eta }_p^2$ = 0.05; IS: F(1,27) = 0.385, p = 0.540, ${\eta }_p^2$ = 0.01) as well as for the alpha activity during activation (IAcc: F(1,27) = 0.615, p = 0.440, ${\eta }_p^2$ = 0.02; IS: F(1,27) = 0.015, p = 0.905, ${\eta }_p^2$ = 0.00). This indicates that the phasic brain activity changes from reference to interoception task are the main driver of the observed association between TRP in the alpha band and IAcc as well as IS, but not the tonic activation during reference or activation per se.

The main effects IAcc (F(1,27) = 1.287, p = 0.267, ${\eta }_p^2$ = 0.05) and IS (F(1,27) = 0.263, p = 0.612, ${\eta }_p^2$ = 0.01) were not significant for the beta (12‐30 Hz) and the theta band (4‐8 Hz; IAcc: F(1,27) = 0.299, p = 0.589, ${\eta }_p^2$ = 0.01; IS: F(1,27) = 0.4997, p = 0.487, ${\eta }_p^2$ = 0.02), thus indicating the specificity of the functional findings for the alpha band (for alpha, beta, and theta power values see Figures S1–S3).

4. Discussion

The present study investigated the TRP in the alpha band during cardiac interoception, which provided two main findings. First, we observed a distinct brain activity pattern during cardiac interoception (under instruction) and second, we indicated that this specific brain activity pattern in the alpha band was functionally relevant for the two performance‐related dimensions of interoception—IS and IAcc (independent from each other; see also Murphy et al. 2019).

Regarding the general brain activity pattern during cardiac interoception assessed via the heartbeat tracking task, we found TRP decreases in the alpha band from a reference phase to the interoceptive interval under most areas—when participants were instructed to perceive their heartbeats. Although we observed TRP decreases in general, TRP in the alpha band was relatively higher over left posterior sites (i.e., temporal and parietal; P4, P8) compared to right sites (i.e., P3, P7), which showed negative TRP values. In line with the gating by inhibition theory (Jensen 2023; Jensen and Mazaheri 2010), the negative TRP values over the right parietal areas might indicate that the right (temporal and parietal) sites are more actively involved during cardiac interoception, while left temporal and parietal sites might be less activated (or even inhibited) when listening to their own heart. In line with Hanna et al. (2024), who reported a decrease of left parietal alpha outflow to visual cortices when processing external sensory information, the relatively higher alpha power over left parietal and temporal sites (vs. right) might indicate less semantic and auditory sensory processes during interoception, which consequently leaves more brain resources for interoceptive attention under instruction (Murphy et al. 2019; Petzschner et al. 2019; Suksasilp and Garfinkel 2022). The relative alpha power decrease at right posterior sites (vs. left parietal sites) supports this interpretation since attentional control processes might be located at right parietal areas (Mengotti et al. 2020; Szczepanski et al. 2010; for a critical review see Mengotti et al. 2020).

However, previous studies interpreted relatively higher TRP in the alpha band in right parietal areas (vs. left) as indicating less sensory‐driven processes and more internally directed attention (Benedek et al. 2014; Rominger et al. 2022). In contrast to these studies, which investigated the attentional shift to more mentally driven tasks (e.g., creativity), the present study asked participants to allocate attention to internal sensory processes (i.e., the heart). This might be one reason for relatively higher TRP in the alpha band over left vs. right parietal sites. Like visual imagination, which goes along with similar brain activities as sensory‐based visual processing (Kosslyn et al. 2001) or imagination of movements, which shows activation in movement‐related brain areas (Neuper and Pfurtscheller 2001), the allocation of attention to internal sensory processes (such as to the heartbeats) might go along with a relatively higher recruitment of the right parietal cortex also involved during attentional processes to external stimuli (e.g., visual; Behrmann et al. 2004; Szczepanski et al. 2010; for a review see Mengotti et al. 2020). This is in line with studies indicating the relevance of (right) parietal areas for self‐perception and introspection (Igelström and Graziano 2017). However, this interpretation of interoceptive attention‐related brain activity in the context of cardiac interoception needs further elaboration in future studies.

The second main finding of this EEG study is that the distinct brain activity pattern in terms of TRP in the alpha band assessed during cardiac interoception was associated with task‐specific performance measures—that are IAcc and IS. This underlines the functional meaning of the reported brain activity. Both dimensions of interoception provided unique information about the involved brain mechanism, and the joint use of two prominent behavioral dimensions of interoception increases the interpretability of our neurophysiological findings. Furthermore, the results strengthen the assumption that interoceptive attention under instruction (indicated by TRP pattern in the alpha band) was differentially associated with IS and IAcc (Murphy et al. 2019). Specifically, individuals who had a higher confidence in their interoceptive performance showed the general interoceptive brain activity pattern (i.e., higher posterior TRP at the left hemisphere vs. right); however, at a lower alpha power level (i.e., stronger TRP decreases). Accordingly, a higher subjective feeling of confidence went along with less task‐related alpha power and more brain activity during the interoception task. Consequently, higher subjective confidence seems less driven by internal directed attention per se but by a generally more active processing mode (i.e., global alpha power decreases). This argues for the assumption that higher efforts and working memory loads (Hjortkjaer et al. 2020) during interoception under instruction benefit confidence. This finding extends previous literature reporting no association between IS and neurophysiological processes (Mai et al. 2018).

Furthermore, higher TRP was accompanied by a better IAcc independent from IS. This nicely aligns with previous findings of Villena‐González et al. (2017), who reported better IAcc in people showing stronger alpha power increases (from an external to internal attention task). Exactly these posterior increases in alpha power might signal the metaphorical closing of the eyes (and ears) in order to reach a more accurate performance outcome (Gallicchio and Ring 2019). People who perceived their heartbeats better (i.e., higher IAcc, independent from IS, gender, age, BMI, and systolic blood pressure) showed relatively higher task‐related alpha power also at more (left) posterior sites. This might signal less active neuronal processing during perceiving the own heartbeats (especially in left temporal and parietal sites), which, as argued above, might suggest a shift of attention towards internal sensory processes and a shutdown of semantic and external sensory processes (Hanna et al. 2024).

Taking together, a relative TRP decrease of alpha power went along with higher confidence ratings, which might indicate (relatively) higher involvement of cognitive relevant brain areas during the interoception task. Cognitive processes such as executive control, attentional processes, and semantic processes might be more pronounced when people perceive their heartbeats with a higher confidence—while the relative pattern of activation remains unchanged. We assume that a more active and salient processing of internal sensations, and sensory perceptions, and the consequently higher neurophysiological feedback on these cognitive efforts put into the cardiac interoception task increases participants' confidence in the accuracy of the perceived heartbeats. The contrary seems true for IAcc (independent from confidence), which might capture the more unconscious aspects of interoception (Nord and Garfinkel 2022). IAcc (independent from IS) seems to be related to higher global alpha power, while the nuanced general interoception induced brain activity pattern (relatively more task related alpha power over left compared to right posterior sites) remains unaffected.

Although this study provides a well interpretable pattern of findings, and the heartbeat tracking task has face validity (Critchley and Garfinkel 2017), results should be discussed considering the criticisms of the task itself. The IAcc assessed via the heartbeat tracking task might not assess the accuracy dimension of interoception adequately (Desmedt et al. 2020, 2018; Ferentzi et al. 2022). One criticism is that we cannot observe the behavior of participants, who might simply estimate the number of beats but not necessarily perceive their heartbeats (Desmedt et al. 2018). However, we used the modified instruction asking participants to report only the perceived heartbeats within each interval. Furthermore, we assessed the subjectively felt confidence in heartbeat perceptions from trial to trial and found convincingly high correlations between the confidence ratings and IAcc (Candia‐Rivera et al. 2022; Garfinkel et al. 2015; Murphy et al. 2020). Additionally, the observation that men showed higher IAcc and IS than women is in line with the literature (Lischke et al. 2020; Pollatos and Fischer 2018).

Nonetheless, future studies should apply the heartbeat tracking task together with further approaches to assess interoception in other domains. Furthermore, applying control tasks allows for a more in‐depth investigation of which specific brain pattern is linked with heartbeat perception, breathing perception, or external perceptions (see e.g., Zaccaro et al. 2024). Future interoception studies investigating TRP in the alpha band might use more explicit control tasks. Although the reported findings might not perfectly capture the specific cardiac interoception effects (vs. other domains), the reported association between IAcc and IS with the brain activity (in terms of TRP) indicated a functional meaning of the observed brain activity pattern during the cardiac interoception task. Therefore, we can conclude that interoceptive attentional processes (indicated via TRP in the alpha band) are differentially linked to IAcc as well as IS, although both are strongly interconnected.

Furthermore, future studies examining TRP changes in the context of interoception would benefit from applying more refined analytical methods to separate aperiodic (1/f‐like) from periodic (oscillatory) activity, which is increasingly recognized as important in EEG research (Donoghue et al. 2022; for approaches see e.g., Donoghue et al. 2020; Kosciessa et al. 2020; Wen and Liu 2016). While the absence of effects in the theta and beta bands in this study suggests alpha‐specific task‐related differences, implementing these advanced techniques would allow one to more explicitly control for broadband components in future studies.

A further limitation of the present study's approach is the use of different time intervals for calculating TRP scores (i.e., 4‐s reference and 20–75 s for activation). Although using different time intervals considers all available information (for similar approaches see e.g., Jauk et al. 2012; Rominger, Papousek, et al. 2019), it might lead to differences in the estimated effects of TRP. We reanalyzed the data using only the first 4 s of activation to calculate TRP, which showed very similar findings for IS and IAcc.

5. Conclusion

Cardiac interoception research by means of the heartbeat tracking task has its own unique drawbacks (see e.g., Desmedt et al. 2020; Desmedt and van den Bergh 2024; Ferentzi et al. 2022). However, by applying a combination of behavioral and neurophysiological methods, we were able to indicate a specific brain activity pattern: During the heartbeat tracking task, we found relatively higher TRP in the alpha band over left posterior sites (compared to right), indicating the allocation of attention to internal sensory processes. Furthermore, we found that this brain activity pattern was associated with task‐specific behavioral performance measures indicating the functional meaning of the observed brain activity pattern. Higher IS was associated with lower TRP, indicating higher cortical activation linked to interoceptive confidence. Higher IAcc was linked to higher TRP, suggesting more unconscious and automatic processes. These findings outline the potential of the combination of behavioral and neurophysiological methods to shed more light on the phenomenon of cardiac interoception and the involved (cognitive) processes.

Author Contributions

Christian Rominger: conceptualization, formal analysis, methodology, visualization, writing – original draft, writing – review and editing. Andreas Fink: methodology, resources, supervision, writing – review and editing. Corinna M. Perchtold‐Stefan: validation, writing – review and editing. Laurenz Schlögl: data curation, formal analysis, investigation, visualization, writing – review and editing. Andreas R. Schwerdtfeger: conceptualization, resources, supervision, writing – review and editing.

Conflicts of Interest

The authors declare no conflicts of interest.

Supporting information

Figures S1‐S3.

Data Availability Statement

The data that support the findings of this study are openly available at https://doi.org/10.17605/OSF.IO/2KFSX.