We tend to describe our emotions through physical sensations, as illustrated by expressions such as “having a lump in your throat” when experiencing anxiety, or “butterflies in the stomach” when in love. This emphasizes the idea that interpretation of physical sensations is closely linked to emergence, detection, and processing of emotions. The idea that emotions depend on bodily changes has been around for well over a century. It was first proposed by William James [1], and has shaped emotion research significantly ever since. Of course, emotion theories have evolved since the controversial James-Lange theory of emotion [2]. Although more recent theories like Schachter and Singer’s two-factor emotion-theory [3] or Damasio’s somatic marker theory [4] integrate other factors such as cognition and appraisal into emotion processing, physical changes are still considered crucial.

The ability to process and integrate such bodily states, or interoception, has also been found to be essential to a wide range of other psychological and physiological processes [5, 6]. Interoceptive processes are responsible for integrating afferent information about the body into conscious physical sensations through mental representation [6, 7]. Humans rely on interoception for the detection and understanding of physical sensations such as hunger or pain, most of the time without awareness of the process [5]. However, we are able to consciously perceive interoceptive processes, for instance by paying attention to the sensation of our heart beating. Three distinct interoceptive dimensions have been postulated and supported with empirical evidence [8, 9]. As described by Garfinkel et al. [9], these dimensions are (1) interoceptive accuracy: the ability to accurately detect physical sensations, measurable with objective tests; (2) interoceptive awareness: a metacognitive ability for insight on one’s interoceptive accuracy, measurable as confidence regarding accuracy; and (3) interoceptive sensibility: the subjectively perceived extent of being internally self-focused and detecting internal bodily signals, measurable through self-report. The most commonly researched interoceptive ability is heartbeat perception, usually measured with a heartbeat-perception task [10]. More recently interoceptive abilities in other modalities, such as respiratory or gastrointestinal sensations, have been investigated, suggesting a connection of interoceptive abilities across several physical systems [11, 12, 13].

Interoceptive abilities have proven essential for maintaining mental health [7, 14, 15, 16]. Interoceptive abilities differ individually, but alterations in interoceptive processes have been found to be associated with lower cognitive function, anxiety, depression, eating disorders, addictive disorders, and psychosomatic disorders [6, 16, 17, 18, 19, 20, 21, 22]. Fortunately, interoceptive abilities have also been found to improve with training [23]. For instance, integrating mindfulness into depression treatments is a concept that is growing in popularity, and a connection between mindfulness treatment success and concurrent increasing interoceptive awareness has been observed [24]. In an overview evaluating interoceptive aspects for mental health interventions, the authors reported finding symptom reductions in anxiety disorders, eating disorders, as well as psychosomatic and addictive disorders [25]. Improvements in eating-disorder treatment [20] and a reduction of anxiety symptoms in autists [26] could also be observed after interoceptive training. Thus, strong evidence suggests that interoceptive abilities are amenable to improvement.

Empirical evidence also supports the commonly postulated role that physical changes, and therefore interoceptive processes, play in emotion, whether in perceiving an emotion or in emotional-language processing [14, 27, 28]. In individuals with alexithymia, a disorder affecting detection and expressing of emotions, there is often also a decrease in interoceptive abilities [29]. Furthermore, higher interoceptive abilities are strongly associated with a higher intensity of emotional experience, as well as higher neural activity, during emotional paradigms [14, 30, 31, 32].

In line with this, recent research has attempted to identify and understand the neural substrates of interoception and emotion [6, 33]. Evidence widely suggests that the insular cortex is part of a neural network involving both interoceptive and emotional processing [6, 34, 35, 36, 37, 38]. The insula is a highly networked brain structure with connections to the prefrontal and orbitofrontal cortex, thalamus, amygdala, anterior cingulate cortex, and somatosensory cortex [6]. Cortical activity during interoceptive processing has already been documented in somatosensory cortex [34, 35, 39, 40, 41, 42], so interoceptive pathways that include both insular and somatosensory activity have been postulated [33, 43, 44]. A growing body of work additionally supports the assumption that due to its role in physiological perception, the somatosensory cortex is also strongly involved in emotional processing [45]. Empirical evidence has shown somatosensory activation in visual emotion recognition, the actual experience of emotions, and emotion regulation [45, 46, 47]. The primary somatosensory cortex (S1), in particular, seems to be involved in interoceptive cardiac perception, as well as in the processing of aversive emotional stimuli [38, 43]. The S1 comprises Brodman Areas 1, 2, and 3 and is located on the postcentral gyrus, representing the entire body and processing afferent information in somatotopically organized neurons [48, 49]. The main function of the S1 is responding to somatosensory sensation [48].

In addition to cortical activity being found in the S1 during interoceptive and emotional processing, neurostimulation techniques targeting this brain region have been found to affect bodily perception [50, 51, 52]. Changes in cardiac-interoceptive accuracy could be produced using transcranial magnetic stimulation (TMS) over the right somatosensory cortex, [52], highlighting the role of the somatosensory cortex in interoception. Changes in pain perception, temperature detection, and tactile spatial perception have also been observed after transcranial direct-current stimulation (tDCS) over the S1 [53, 54, 55]. TMS over interoceptive networks has also been found to alter emotional perception [56]. These findings illustrate both the interconnectivity of interoceptive and emotional processing, and their susceptibility to being changed by neurostimulation.

tDCS is a non-invasive neurostimulation method with growing popularity that can be used to modulate cortical excitability in humans [57, 58, 59]. Different effects can be observed with positive current (anodal tDCS) and negative current (cathodal tDCS). Anodal tDCS has been found to facilitate spontaneous and evoked cortical potentials, whereas cathodal tDCS robustly inhibits cortical excitability [51, 57, 60]. Conventionally, tDCS is accomplished by applying one target electrode and one reference electrode to the scalp of a participant [57]. Electrodes then emit a weak electrical direct current into the participant’s brain, creating electric fields that modulate the somas and axon terminals of affected neurons [61]. Traditionally, sponge electrodes are used for tDCS application [51]. However, over the past 15 years, studies have been evaluating the use of ring electrodes, observing that they offer a higher degree of spatial focality than do traditional sponge or disc electrodes [62, 63]. More specifically, so called High-Definition tDCS (HD-tDCS) using such ring electrodes has been the direct-current-stimulation method of choice recently, with advantages over conventional tDCS including greater spatial precision and longer lasting after-effects [60, 64, 65, 66]. HD-tDCS provides a safe, effective, non-invasive neurostimulation method, and has proven to be effective in randomized controlled trials [66, 67, 68].

Interoceptive abilities and emotional perception are evidently modifiable and linked, as well as is specific, equally modifiable, cortical activity; because transcranial magnetic stimulation has been successfully used to modify both interoception and emotion perception [52, 56], the question arises as to whether tDCS over the associated cortical regions can also achieve modulation of interoceptive abilities. Recently, effective modulation of hunger was achieved using cathodal tDCS over the tongue motor-cortex area [69]. Given the connection of the tongue motor cortex to the interoceptive neural network [33, 70], and the importance of interoception in hunger [71], this is evidence for the potential of modulating interoception with tDCS. Although the insula is commonly considered the interoceptive center of the brain, the S1 is linked to both interoception and emotion processing. As the S1 is located closer to the scalp than is the insula [6], the S1 also provides a more easily accessible target for brain stimulation than the insular cortex. To date, no studies evaluating an attempt to affect interoceptive abilities and emotional perception using HD-tDCS over the somatosensory cortex are known to us.

In the present study, we attempted to enhance interoceptive abilities and emotional perception by applying anodal HD-tDCS over the right S1. We hypothesized that interoceptive accuracy, interoceptive awareness, and interoceptive sensibility would increase after active tDCS. We also applied a physical pre-activation task, consisting of ergometer cycling, in order to activate bodily signal processing from the bottom up. Based on the established connection between interoceptive and emotional processing, and the overlapping neural substrates, we further expected emotional experience to intensify after active tDCS, but not after sham tDCS.

The present study found diverse effects of HD-tDCS over the right S1 on various dimensions of interoceptive and emotional processing. Contrary to expectations, we did not observe improvement in either cardiac or respiratory interoceptive accuracy after active tDCS. For interoceptive sensibility, we observed an increase in both active and sham tDCS. This finding can be easily explained: participants completed the questionnaire shortly after undergoing ergometric pre-activation. The questionnaire assessed the participant’s current awareness levels of various bodily sensations, including cardiac sensations, respiratory sensations, trembling, and temperature/sweating, all of which can be influenced by physical exercise. It is possible that the interval between ergometric activation and the response to the state interoceptive awareness questionnaire was too short to capture the effects elicited by tDCS. To address this, future studies may need to lengthen the interval between pre-activation and the completion of the questionnaire in order to observe potential tDCS effects more accurately. This adjustment could provide valuable insights into the impact of tDCS on interoceptive accuracy and sensibility, and further our understanding of the potential benefits of tDCS in this context. Regarding interoceptive awareness, we noticed an improvement in the heartbeat perception task from before to after measurements for both tDCS conditions. However, we found statistical significance only for the increase after active tDCS. Unfortunately, the expected interaction of measurement time and condition was not statistically significant, and we did not observe a significant increase in active tDCS for respiratory awareness. The robust pre-to-post improvement in both the cardiac and respiratory tasks may be attributed to the participants’ awareness that they had previously performed the interoceptive tasks in the pre-assessment. This knowledge could have increased their confidence in their performance during the post-measurement. Although this phenomenon could have influenced the results, it is essential to consider it as a potential confounding factor in the interpretation of the findings. To gain a comprehensive understanding of the impact of tDCS on interoceptive awareness, future studies may consider incorporating control measures to account for the effect of previous task experience. This way, we can further investigate the true effects of tDCS on interoceptive accuracy and ensure more robust and accurate conclusions. This is consistent with evidence of experienced meditators who rate confidence regarding cardiac interoceptive accuracy higher than do people with no experience in attending to body signals, even with no increase in actual cardiac-perception accuracy [99]. Ergometric pre-activation may also contribute to pre-to-post improvements, affecting confidence ratings by increasing awareness of physical sensations [100] or through the expectation that heart rate must be elevated after physical exercise [101]. However, these influential factors should be present equally in both conditions and therefore do not sufficiently explain the significant increase of cardiac interoceptive awareness in the active tDCS condition only. This finding provides support, albeit limited, for our hypothesis that HD-tDCS over the right somatosensory cortex can improve interoceptive awareness for the heartbeat-perception task. Regarding emotional perception, again contradicting our initial hypothesis, we observed no changes after active tDCS in valence or arousal ratings for either perception of interoceptive tasks, or rating of emotional pictures. Overall, we were therefore unable to produce consistent enhancement of interoceptive abilities or emotional experience in this study.

Possible explanations for these findings range from methodological aspects referring to interoceptive-accuracy measurements both in the cardiac as well as respiratory domain, to the design of our tDCS protocol, and to HD-tDCS effectiveness itself.

Methodological aspects of the heartbeat-perception task mainly relate to factors affecting heartbeat perception other than interoceptive accuracy itself, thereby threatening the validity of the heartbeat-perception task. For instance, we did not account for participants’ beliefs about their own heart rate, as well as cardiac processes such as blood pressure, which have both been found to be influential in heartbeat perception [101, 102]. If differences between participants occurred in these factors, the factors potentially influenced cardiac interoceptive-accuracy and awareness scores and thereby, could have overpowered effects elicited by HD-tDCS. Taking heartbeat beliefs and cardiodynamics into account may strengthen further investigations on the effects of HD-tDCS over the S1 on cardiac interoception.

Compared to cardiac-interoception-measurement methods, there are few methods available for measuring respiratory interoception, and no standardized method has yet been established. We designed our paradigm based on previously used protocols, asking participants to rate which of several inspiratory loads at different levels they are presented with, and calculating accuracy scores [13, 52]. However, there are newly developed paradigms for evaluating respiratory interoception that might contribute to a higher reliability and standardization of respiratory-interoceptive measurements. For instance, variation of the length of inspiratory occlusion rather, than of the intensity of inspiratory load, has been proposed [103]. A respiratory-resistance-sensitivity task has also been presented in which subjects are presented with two inhales at different inspiratory loads in each trial, and asked to indicate which of the two inhales required more effort, using a two-interval forced-choice protocol [104]. A custom 3D-printed apparatus and adaptive algorithm for this task allows for automated and standardized measurement of the minimal stimulus an individual can reliably discriminate from a regular inhale [104]. Future studies assessing tDCS effects on interoception might implement such novel tasks to improve respiratory interoceptive assessment.

Regarding study design, future investigations may benefit from alterations to our ergometric pre-activation. With this paradigm, we aimed to add a bottom-up activation in interoceptive neural networks to the expected top-down activation from active HD-tDCS. The pre-to-post increase for both active and sham tDCS conditions that we observed for cardiac- and respiratory-interoceptive awareness supports the notion that this bottom-up activation was successful. We chose a target heart rate of 150 bpm for our pre-activation, as it has been found to elicit higher cortical activity in interoception related brain structures than did exercise with a lower intensity [96]. However, a moderate level of physical strain may be more effective for improvement in interoceptive perception than a high level of physical strain [100]. Additionally, a fixed target of 150 bpm does not necessarily result in equal levels of physical strain for every participant. Exercise intensity depends more on the relation of exercise heart rate to maximum heart rate, rather than absolute exercise heart rate itself [105]. As maximum heart rate differs individually depending on factors such age, sex, and body mass index, a fixed heart rate of 150 bpm likely results in a different exercise intensity for different people [105, 106]. It might therefore be preferable to calculate maximum heart rate for each participant and set a fixed percentage of maximum heart rate as the target heart rate, to ensure that all participants reach a comparable level of exercise intensity. Taking these considerations into account, adapted ergometric pre-activation may be better suited for supporting tDCS effects with a lower risk of overpowering them.

Future studies should also examine the suitability of different aspects of tDCS application itself. The high variability in methods of existing tDCS research creates a wide range of possible application conditions. Reviews and meta-analyses [50, 60, 107, 108] show that there is great variability in many factors of tDCS application: for instance, there is great variety in stimulation duration, which can last from under 10 min to a full 30 min, and current intensity, which varies between 1 mA and 2 mA. In our study, we applied the maximum current intensity (2 mA) for the most commonly used duration (20 min) [50]. Our post-assessment was designed to begin 12 min after stimulation onset. Previous studies began assessment of stimulation effects as early as stimulation onset, and many authors, like us, conducted post-assessment during active stimulation, with a varying amount of time passing between studies [50, 107]. Although this procedure has led to significant tDCS effects in the past, several studies targeting the S1 opted to measure post-assessment offline instead, yet still observed effects [50]. Perhaps the most significant difference in tDCS methods lies in the number of administered-stimulation sessions. With more recent work, it seems that administering multiple, more precisely 5 to 10 sessions, is increasingly considered the appropriate approach [60, 108]. Therefore, a wide range of possible stimulation designs is offered, and it might be beneficial to explore tDCS effects on interoceptive abilities and emotional experience with the application of multiple sessions and assessments conducted after the full duration of stimulation. It is important that meta-analyses show great variability not only in tDCS administration protocols, but also in effectiveness [60, 107, 109, 110]. If future investigations continue to fail to observe consistent tDCS-induced changes in interoception and emotion, even with adjusted stimulation protocols as discussed, it may be necessary to discuss the utility and effectiveness of HD-tDCS in this area altogether. tDCS, as a neurostimulation method, delivers a weak electric current to the scalp, creating weak electric fields in the brain that can alter neuronal transmembrane potentials and thereby influence cortical excitability [111]. In contrast, TMS administers magnetic pulses that elicit focused currents in the brain, directly affecting and depolarizing neurons [111]. For altering interoceptive abilities and emotional processing, there is evidence emphasizing the utility of repetitive theta-burst TMS protocol [52, 56]. Whether this approach is generally more suitable for stronger effects in neuromodulation will need to be evaluated continuously in the future.

The most critical point to be discussed is the possibility that our tDCS stimulation was ineffective. Since we observed no significant changes in somatosensory perception as a result of active tDCS, the question arises whether the stimulation protocol achieved the expected neuromodulation of the S1. Explanations for a possibly unsuccessful neuromodulation of the S1 are several. For one, HD-Targets™ electrode configurations [90], which are based on male head models, may only be inappropriate for our 60% female sample, as brain size differences in respect to sex are consistently found [112]. Beyond sex differences, neuroimaging research from past decades has found robust interindividual structural differences in size and location of structures in human brains [113, 114]. Interindividual neuroanatomic variability at the stimulation target point is an even bigger risk using HD-tDCS, as it provides higher focality than traditional tDCS [115]. In future studies, using brain-imaging methods like an functional magnetic resonance imaging (fMRI) to determine the exact location of a target area may be useful to achieve higher precision in electrode placement and to ensure successful stimulation. Additionally, as the S1 comprises all of Brodman Area 1, 2, and 3 [48], and the HD-Targets™ software offers multiple configurations for each area, a different configuration may prove more effective for observing the expected changes in interoceptive abilities.

The lack of changes in our somatosensory-perception measurement is not necessarily rooted in an unsuccessful neuromodulation. Although tactile stimulation has been found to elicit cortical activity [86], evidence for a causal relation between cortical activity and somatosensory perception is lacking. Animal studies on the relationship between neural activity and behavior illustrate that there is variability in behavior that cannot be conclusively explained solely through neural activity [116]. Furthermore, there is no established measuring instrument for somatosensory perception, and a self-report measure, as used by us, may be susceptible to numerous biases and response sets [117]. Tactile stimulation in this study was also applied manually, which may have caused slight differences in the application of the tactile stimuli, therefore possibly threatening the validity of our somatosensory measurement. All of these aspects lend credibility to the conclusion that despite not observing alterations in the somatosensory perception, the targeted neuromodulation in our study may very well have succeeded.

All aspects discussed are important when explaining why we did not observe a significant change in emotion processing. As many theories emphasize the role of interoception for emotional perception [1, 2, 3, 4], a failure to improve interoceptive processing by neuromodulation may also account for the lack of effects on emotion processing. Empirically, reduced interoceptive accuracy, after TMS inhibition of interoceptive networks surrounding the frontotemporal anterior insula and somatosensory cortex, was found to be associated with a flattened emotional experience [56]. This indicates a causal relationship between interoceptive and emotion processing, which supports the idea that it is unlikely that changes in emotion processing can be observed, as long as there is no alteration in interoceptive processing, as we observed in this study. Furthermore, induction of emotion may not have been entirely successful using the IAPS, as it consists of pictures that are several decades old, including violent and erotic pictures that may not produce the same effect on emotional arousal nowadays. The fact that the arousal measurements elicited by positive and negative pictures in our study were lower than reference values of the IAPS [118], supports this notion. Weaker emotional induction may limit the chance to observe tDCS-induced changes in emotional perception. Future studies may need to assess suitability of the IAPS for emotion induction 20 years after measurement of reference ratings.

For the first time, we attempted to enhance interoceptive abilities and emotional experience using HD-tDCS over the right S1, and our study design has several strengths. For instance, conducting a randomized trial with repeated measurements so that each participant would be exposed to both tDCS conditions eliminated any possible confounding of group-specific variables with the observed effects. We met criticism of varying sham protocols in previous tDCS studies [92] by assessing the physical sensations elicited by tDCS. This allowed us to conclude that participants experienced comparable physical sensations in both active tDCS and sham tDCS, and were unable to distinguish between the two. Even though investigators were not blind to the order in which participants received tDCS, risk of bias in stimulation effects in our study was very low, as our main outcome of interoceptive accuracy was assessed by objective measurement [119]. Another strength of our study lies in our meaningful group size, in comparison to many other tDCS studies in which sample sizes were seldom as large as 30 subjects [50, 60].

Based on the insights our study provides, opportunities for improvement in future studies are created, as previously discussed. To extend these, we should mention that we did not include a subjective measure of interoceptive awareness such as the Multidimensional Assessment of Interoceptive Awareness (MAIA-2 [120]), which could have provided important insight into the awareness effect we observed. Furthermore, although the size of our sample is satisfactory, sample heterogeneity is not. We conducted this study with a very young, highly educated, physically active group of participants. Our results might therefore not be generalizable to populations with lower education levels or older age. Further studies with more diverse samples will be necessary to continue reliable examination of the potential for somatosensory tDCS stimulation on interoception and emotion.

The present study was the first to evaluate the effects of HD-tDCS over the right S1 on interoceptive abilities and emotional experience. Although we attempted to design and conduct the experiment with the utmost care and consideration of previous findings, we were unable to provide convincing evidence to support our initial hypotheses. We observed no changes attributable to active HD-tDCS in interoceptive accuracy, interoceptive sensibility, or emotional experience. Only cardiac interoceptive awareness increased significantly after active HD-tDCS. Despite the findings largely not supporting our assumptions, the present study contributes to the growing body of research on the association of interoception and emotion, and opens up new avenues for their potential modulation by non-invasive brain stimulation. Future studies may benefit by increasing the number of stimulation sessions, adjusting the electrode configuration, lowering the ergometric pre-activation intensity, and assessing additional variables with the potential to influence HD-tDCS effects. Overall, the present study provides an exciting novel approach to modulation of interoceptive and emotional processes, sheds light on potential areas of improvement, and thus creates meaningful grounds for future investigations.

HD-tDCS, high-definition transcranial direct current stimulation; IAPS, International Affective Picture System; S1, primary somatosensory cortex; tDCS, transcranial direct current stimulation; TMS, transcranial magnetic stimulation.

The data analyzed in this article is available upon request to the corresponding author.

JS, SAH, SML and OP designed the research study. JS and SAH performed the data assessment. SML provided help with participant recruitment and technical support during the experiment. JS, SAH, SML and OP analyzed and discussed the data. All authors contributed to editorial changes in the manuscript. All authors read and approved the final manuscript. All authors have participated sufficiently in the work and agreed to be accountable for all aspects of the work.

The study protocol was approved by the ethics committee of Ulm University (study number 207/22). All participants were informed about the study and gave written consent.

We thank all of the study participants for their time. We especially thank Prof. Dr. Thomas Kammer from the Section for Neurostimulation at the Department of Psychiatry, Ulm University, for his expertise on tDCS and his meaningful contribution to our neurostimulation paradigm.

This research received no external funding.

The authors declare no conflict of interest.