The majority of the models of emotional processing attribute subjective emotional feelings to physiological changes in the internal milieu, including the skeletomuscular, neuroendocrine, and autonomic nervous systems (1–3). These changes are often triggered by external events as well as the homeostatic changes of inner bodily status, and can affect body-mind interactions (4–6). Indeed, affective experiences initially give rise to specific physiological changes in the body, which are bidirectionally associated with the perception of one's emotions (7, 8). Central representations of the bodily responses caused by external emotive stimuli are perceived as emotions; i.e., automatically generated bodily responses via the autonomic nervous system can result in subjective emotional feelings (9). Derived from this theory is the notion that the relative ability to perceive internal body status influences measures of subjective affective experience, which highlights the fundamental importance of the body in emotional phenomena.

The ability to self-monitor one's own internal body status is a key feature of the body-mind interaction in emotional awareness. Interoception, which is the sensitivity to visceral sensations (10), refers to the current homeostatic conditions of the entire body (11) and the ability of this information to reach conscious awareness (12). Interoception derives not only from musculoskeletal, circulating, and humoral signals within the body, but also from afferent information via autonomic nervous reflexes and higher levels of autonomic regulation originating from visceral organs and vascular systems (13). Two distinct neural pathways convey interoceptive information. The lamina I spinothalamocortical pathway carries information regarding the internal milieu, such as muscle contractions in vessel walls, peripheral blood flow, temperature, pain, tissue injury, and levels of O2 and CO2, while the vagal nerve carries visceral signals regarding the cardiovascular, respiratory, gastrointestinal, and genito-urinary systems (14). The crosstalk between these two pathways within the projections of the brainstem to the insular and somatosensory cortices facilitates the organization of topographical maps of bodily status within structures (3). Namely, interoceptive information originates from multiple organs via corresponding conduits and is gathered and successively processed in the structures of the brain, providing topographically organized somatic maps. Among these, the sensitivity to one's own heartbeat is closely associated with interoceptive awareness (14, 15) and the intensity of experienced emotions (16–18). The accuracy of heartbeat counts has been found to be positively correlated with the sensitivity of emotional traits, such as tendencies for general anxiety (19, 20).

To investigate the cortical activation related to interoception, both heartbeat evoked potentials (HEPs) and heartbeat evoked magnetic fields (HEFs), which are measured using electroencephalography (EEG) and magnetoencephalography (MEG), respectively, serve as useful indexes of interoceptive processing, and reflect the sensitivity to ones' own internal body states (21–23). HEPs/HEFs are computed by averaging the neural signals with time-locked triggers of individual R peaks of electrocardiography (ECG), which emerge after 200 to 300 ms and are mainly derived from fronto-central regions of the brain (24–29). Thus far, HEPs/HEFs have been associated with cardiovascular arousal (30), heartbeat awareness (31–33), visual awareness (34), self-consciousness (35, 36), self-face recognition (37), pain perception (38), empathy (39), sleep states (40), and psychiatric disorders, including depression (41), depersonalization (42), and borderline personality disorder (43).

Aside from this accumulating evidence, the precise mechanisms of HEPs/HEFs on emotional experiences are still unclear. As most studies have used EEG to measure the effects on interoceptive awareness, poor spatial resolution has obscured elucidation of these physiological features. We hypothesized that the subjective experience of emotions would elicit the peripheral physiological responses initially, then affect the emotion related cortical activities. Particularly, we anticipated that disgust feeling of emotion will impact either myocardial or gastric functions, thus give rise to the alteration of HEFs reflecting these changes of bodily status. To assess our hypothesis that spontaneous heartbeat movements evoke cortical activities and connectivity changes related to emotional processing, we conducted MEG, which has a high temporal and spatial resolution. In this way, we investigated alterations in HEFs that are mediated by disgusting experiences to clarify the precise cortical activity, connectivity, and frequency properties to better understand brain-mind interactions.

The current study revealed that listening to sounds of disgust induces changes in HEFs. Behavioral data indicated that the sound stimuli affected the subjective emotional state by increasing disgust emotion as well as decreasing happiness. As the source current estimation by subtracting disgust minus control conditions revealed the activation, not the suppression, of bilateral insular cortices and inferior frontal cortices, the main effect provoked by the current stimuli was associated with the evoked disgust emotions than the suppressed happiness. Since several studies have reported there to be a relationship between disgust and insular cortical activity (65, 66), the most prominent locus affecting the alteration of HEFs was insular activities evoked by the subjective experience of disgust. In addition, the physiological result on the heart rate variability could not demonstrate the statistical differences in the peripheral cardiac effects such as heart rate, SDNN, and RMSSD on the contrary to our hypothesis. Since our study did not utilize the other indices, such as gastrointestinal information, it would be difficult to further evaluate the initial peripheral effect by means of these physiological data alone. Taken together, the psychophysiological results indicated the alteration of HEFs found in the current study were affected by subjective feeling of disgust, without significant changes in tonic myocardial activities detected by the heart rate variability analysis.

Besides the neural transfer relating to the myocardial activities, there assumed to be alternative conduit such as the humoral pathways involving circulating cytokines, and direct involvement via immune cells from the periphery. In addition, interoceptive processing is modulated by respiratory (67), gastrointestinal (68), and genito-urinary information, in addition to the cardiovascular status. Given the nature of interoception, alternative analysis by measuring spontaneous activity towards other organs may reveal the different characteristics of interoceptive processing; however, the temporal advantages of utilizing MEG would be the most efficient for analyzing relatively highly frequent and periodic responses, i.e., myocardial activity, based on rather accumulating evidence. Since the current study utilized disgust emotion mainly affecting the insular cortices, the afferent information was delivered by either autonomic or somatosensory pathway. The former was mainly carried via the parasympathetic vagal nerve, which projects mainly to the anterior ventral insular cortices, whereas the latter partially projected directly through the posterior insular cortices. In order to investigate the evoked responses as a carrier signal from periphery, and representing cortical/subcortical patterns of activity, we have estimated the source-evoked time courses and ITCs by dividing the insular cortices in each hemisphere into three sub-regions (anterior dorsal, anterior ventral, and posterior insular cortices). No significant between-condition changes in source-evoked time courses were found, however, the significant differences in ITC—increasing at 19 and 23 Hz and decreasing at 14 Hz—were found only at the right anterior ventral insular sub-region. In addition, statistical analysis of PLIs using the Mann-Whitney U test revealed a significant connectivity change between the right anterior ventral insular cortex and left VMPFC. Since the current results were calculated by averaging the MEG signals triggered by spontaneously delivered bioelectrical cues directly from living organs (individual heartbeats), not by externally supplied cues, the averaged responses reflect more indigenous and induced activities. The current results support our hypothesis that spontaneous heartbeats evoke cortical activity that is associated with the perceptive consequences of inner bodily states according to emotional states. In particular, when subjects felt disgust, brain activity in the right insular cortices and connectivity of beta frequency between the right anterior ventral insular cortices and VMPFC increased. These results suggest that the insular cortices are core regions that mediate the physiological changes modulated by emotional state, and closely associated with the VMPFC in the processing and propagation of information to other brain regions. Although previous studies (69–71) have suggested that the processing of somatic information was performed with low beta frequency (14 to 24 Hz), our results provide additional information of the frequency property within the low beta frequency range both for suppression of 14 Hz and amplification of 19 and 23 Hz when people experience emotions as a consequence of their inner bodily state.

Beyond cortico-cortical networks in the brain, Harrison et al. (68) have reported there to be a relationship between cerebral blood flow and peripheral bodily status. The authors evaluated this correlation using both cardiac and gastric response, which were measured using a combination of functional magnetic resonance imaging (fMRI) and simultaneous recording of multiorgan physiological responses while observing motion pictures of disgust. The authors used two distinct forms of disgust that are defined using functionally different domains (72), i.e., “core disgust” and “body-boundary-violation (BBV) disgust” (73), which elicited two independent patterns of peripheral physiological responses as well as cortical activation. “Core disgust” refers to the feeling related to protecting the body from disease or infection (74–77), is usually provoked by images of the ingestion of spoiled food or vomit, and elicits changes in gastric and right dorsal insular activities. In contrast, “BBV disgust” is related to the denial of mortality, is usually induced by images of body mutilation, blood, and venesection, and evokes changes in cardiac power and left dorsal anterior insular cortex activity. Indeed, “core disgust” induces feelings of nausea, retching, and rapid-gastric activities (78–80), whereas “BBV disgust” evokes feelings of dizziness and lightheadedness, affects cardiovascular responses, and in extreme forms can manifest as vasovagal syncope for 3%–4% of the population (81). Interestingly, both types of disgust stimuli have been found to evoke similar activity in the anterior insula; “core disgust” elicited more activity in the ventral mid-anterior region, while “BBV disgust” activated the primary sensorimotor cortices and superior parietal region (82). Thus, the insular cortices may be the regions in which emotions are embodied, whereby inner bodily states are projected to the brain to perceive and regulate the subjective experience of emotions. Interoceptive afferent signals are conveyed to the dorsal posterior insular cortices, where a highly resolved and somatotopically organized cortical representation of the physiological condition of all bodily tissues is provided (83). Emotionally salient information is replicated repeatedly along regions spanning the anterior to posterior insular cortices, which enables progressive incorporation with multimodal information. The anterior insular cortices are connected to the amygdala, cingulate, ventral striatum, prefrontal, and multimodal sensory regions (84). These connections integrate homeostatic conditions (85) according to motivational, hedonic, cognitive, and social signals, which ultimately form the corresponding emotional state. Our results support these concepts and suggest that listening to sounds of disgust affects activity in the right anterior insular cortices, thereby forming the subjective experience of emotion.

The insular cortices have been identified as the main cortical target of the interoceptive information by organizing topographical bodily maps documented in contemporary neuroscience (5, 14, 83, 84, 86). Afferent feedback from bodily states will eventually be represented in the specific brain network that supply neural substrates of visceral feeling and interoceptive status. The VMPFC is also involved in this network as the trigger region to re-activate somatic patterns, and also as the endpoint where bodily changes in response to brain activity are referred back via the somatosensory and insular cortices (87). In addition, recent functional neuroimaging studies have also implicated the insular cortices in both interoceptive processing and emotional feelings (8, 9, 66) in collaborating with somatosensory cortices, such as SI and SII (29, 33), as well as the specific cortical network of default mode (34). Abnormal activity of cortical network implicated in these regions, i.e. high resting-state activity in the insular and limbic/subcortical regions as well as the hypoactivity in higher-level cortical regions, are associated to the clinical conditions of depressive state, including anhedonia, hopelessness, apathy, and particularly dysphoria (88), in extremely forms manifesting, e.g., Cotard syndrome (89). The insular cortices play a key role in executing these functions by representing and projecting information from the sensory feedback of inner bodily status (3, 90).

This study has several limitations that should be mentioned. First, it is unclear whether R or T of ECG triggering were suitable for averaging the HEFs. Although several reports have used T trigger averaging (34, 91), our alternative analysis using T peaks did not clearly identify major components as reported in previous studies. Given the nature of ECG characteristics, the R peak is more prominent than T triggers; averaging using R enables more robust, stable, and technically easier applications without any jittering. Therefore, the HEF components in our study may have been contaminated by ECG T wave components, as the T wave follows the R peak by approximately 400 ms. More convenient and reliable methods for averaging with the T peak should be evaluated. Future studies should examine whether utilizing R or T in individual ECG is superior for obtaining HEFs. The second limitation is related to individual differences in the strength of HEFs. The variations of interoceptive awareness mean that the strength of HEFs thereby vary considerably (18). These variations could be influential distracters in grand-averaging across subjects. Given the premise that one subject may have a greater influence in grand-averaging, an individual with greater responses would more strongly affect the tendency. As our results may contain individual differences, grand-averaged data could reflect unequal contributions among subjects. The third limitation is that our stimuli induce “core” disgust, which mainly results in a feeling of dysphoria in the stomach, without any quantifications of gastrointestinal activity. By excluding BBV disgust, such as bleeding or body mutilation, the stimuli may still have been insufficient for induction of pure “core” disgust due to the limited selection from IADS. As we could not fully evaluate the peripheral effects on interoceptive processing, including cardiovascular as well as gastrointestinal activity, the whole picture of interoceptive carrier and the interpretations on cortical or subcortical reflections still remain uncertain. In fact, gastrointestinal information was also carried regarding the chemical and inflammatory environment as well as the composition of gut bacteria and endocrine responses, is known to modify emotional function (92, 93). Indeed, the responses evoked by the current study may contain all the consequences resulting from presentation of the stimuli, thus future study is required to clarify the characteristics of interoceptive afferent information not only within the regional carrier but also beyond the cortical or subcortical activity constituting brain network, by supporting with the reinforced findings both on behavioral and physiological data; here, the nature of information propagation may help our understanding of further forms of interoception associating to the clinical conditions with aberrant interoceptive processing.

Despite these limitations, the neuromagnetic dynamics demonstrated in the current study revealed HEF-evoked alterations in both activity and connectivity that were affected by emotional experiences. These results provide insights into the frequency properties of information propagation that occurs in the specific beta frequency. This novel method used heartbeat rhythms originating internally, i.e., spontaneous activity of a living organ and not using external stimuli, to determine how indigenous and induced brain activity develop. Moreover, this made it possible to adopt a block design, which is frequently used in fMRI and other brain imaging studies in translational research. Emotional experiences evoked by passive listening to sounds of disgust modulate the autonomic system, which can potentially give rise to alterations in HEFs via changes in activity of the right insular cortices and connectivity changes to the VMPFC.

The alteration of HEFs was investigated in 39 healthy volunteers in association with the subjective feeling of disgust, which evoked right insular cortical activity and the enhancement of cortical connectivity, within the beta frequency range, between the right anterior ventral insular cortex and the left ventromedial prefrontal cortex. Furthermore, the inter-trial coherence in the right anterior insular cortices increased at 19 Hz and 23 Hz, and decreased at 14 Hz, which highlights the involvement of low beta oscillations in emotional processing. The corticolimbic network, particularly involving the insular cortices, plays an important role in emotional regulation and perception. The current results provide a novel insight into the frequency properties of emotional processing, and suggest that emotional arousal impacts the autonomic nervous system, thus giving rise to the alteration of HEFs via connectivity changes in the anterior ventral insular cortex and ventromedial prefrontal cortex.

The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.

The studies involving human participants were reviewed and approved by Ethics Committee of Gunma University Graduate School of Medicine. The patients/participants provided their written informed consent to participate in this study.

YK and YT conceptualized, designed, and conducted the study. SU supported data analysis and interpretation. MM and MF supported the interpretations. YK and all co-authors corrected the manuscript. All authors contributed to the article and approved the submitted version.

This work was supported by the Japan Society for the Promotion of Science (Grant-in-Aid for Scientific Research on Innovative Areas; Science of personalized value development through adolescence: integration of brain, real-world, and life-course approaches), area B01 (16H06397), and Grants-in-Aid for Scientific Research B (15H04890, 16K19748). This work was also partially supported by the Japan Agency for Medical Research and Development, AMED (Strategic Research Program for Brain Sciences, Integrative Research on Depression, Dementia and Development Disorders [IR3D]; Development of diagnostic and therapeutic methods for depressive symptoms based on mesocorticolimbic system).

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.