If you don't remember your password, you can reset it by entering your email address and clicking the Reset Password button. You will then receive an email that contains a secure link for resetting your password
If the address matches a valid account an email will be sent to __email__ with instructions for resetting your password
University of Tübingen, Tübingen Center for Mental Health, Department of Psychiatry and Psychotherapy, 72076, Tübingen, GermanyTrinity College Institute of Neuroscience, Trinity College Dublin, Dublin 2, Ireland
University of Tübingen, Tübingen Center for Mental Health, Department of Medical Psychology and Behavioral Neurobiology, 72076, Tübingen, GermanyGerman Center for Diabetes Research (DZD), 85764, München-Neuherberg, GermanyInstitute for Diabetes Research and Metabolic Diseases of the Helmholtz Center Munich at the Eberhard Karls University Tübingen, 72076, Tübingen, Germany
University of Tübingen, Tübingen Center for Mental Health, Department of Psychiatry and Psychotherapy, 72076, Tübingen, GermanyUniversity of Bonn, Department of Psychiatry and Psychotherapy, 53127 Bonn, Germany
Transcutaneous auricular vagus nerve stimulation (taVNS) can emulate interoceptive signals.
•
taVNS boosts stomach-brain coupling in the brainstem, midbrain, and transmodal cortex.
•
taVNS-induced changes in stomach-brain coupling mirror changes in hunger ratings.
Abstract
Background
Maintaining energy homeostasis is vital and supported by vagal signaling between digestive organs and the brain. Previous research has established a gastric network in the brain that is phase synchronized with the rhythm of the stomach, but tools to perturb its function were lacking.
Objective
To evaluate whether stomach-brain coupling can be acutely increased by non-invasively stimulating vagal afferent projections to the brain.
Methods
Using a single-blind randomized crossover design, we investigated the effect of acute right-sided transcutaneous auricular vagus nerve stimulation (taVNS) versus sham stimulation on stomach-brain coupling.
Results
In line with preclinical research, taVNS increased stomach-brain coupling in the nucleus of the solitary tract (NTS) and the midbrain while boosting coupling across the brain. Crucially, in the cortex, taVNS-induced changes in coupling occurred primarily in transmodal regions and were associated with changes in hunger ratings as indicators of the subjective metabolic state.
Conclusions
taVNS increases stomach-brain coupling via an NTS-midbrain pathway that signals gut-induced reward, indicating that communication between the brain and the body is effectively modulated by vago-vagal signaling. Such insights may help us better understand the role of vagal afferents in orchestrating the recruitment of the gastric network which could pave the way for novel neuromodulatory treatments.
]. Interoceptive signals from peripheral organs involved in energy metabolism, such as the stomach, convey the physiological state of the body via vagal afferents to the nucleus of the solitary tract (NTS) in the brainstem [
]. Efferent vagal signaling modulates digestion by regulating the movement of the gastrointestinal smooth muscles directly or indirectly via the interstitial cells of Cajal which act as pacemakers [
Previous research on the stomach-brain axis has established a “gastric network” in the brain, whose activity is coupled to the 0.05 Hz myoelectrical rhythm intrinsically produced in the stomach that paces digestive contractions [
], suggesting a role for gastric network oscillations in neuromodulation and behavior. Furthermore, dysfunctions of this network might be associated with disorders such as Parkinson's disease [
However, to better characterize the role of the gastric network, causal manipulations are necessary. Such a technique might be provided by transcutaneous auricular vagus nerve stimulation (taVNS). taVNS has been shown to afferently increase activity in the NTS [
]. To assess the effect of vagal afferent activation on stomach-brain coupling in healthy individuals, we used taVNS with concurrent electrogastrography (EGG) and fMRI (Fig. 1A). We expected taVNS to increase gastric coupling in regions with known vagal afferent projections, such as the NTS. We further assessed if these stimulation-induced changes in stomach-brain coupling are correlated with changes in metabolic state as quantified by ratings of hunger. Our results show that stomach-brain coupling can be modulated with taVNS via a pathway that conveys gut-induced reward signals, and that this modulation may alter interoception about metabolic state.
Fig. 1Study design and corroboration of the gastric network during baseline scans (before stimulation). A: We measured resting-state fMRI and electrogastrography (EGG) in 31 healthy participants (20 women) for 10 min during baseline and 10 min during stimulation (taVNS vs. sham, collected on different days). The coupling between the stomach and the brain was expressed by the phase-locking value (PLV), indicating the synchrony between time series. B: We replicated the core regions of the canonical gastric network at baseline by observing coupling in the postcentral gyrus, the cingulate gyrus, the precuneus, the occipital cortex, the fusiform gyrus, the inferior frontal gyrus, the inferior and superior parietal lobe, the thalamus, and the inferior cerebellum. C: Within anatomically defined brain regions, we observed a moderate to high correlation between PLVs obtained in our sample and the independent sample of Rebollo et al. (2018), average t-values: r = 0.318, p < .001, maximal t-values: r = 0.57, p < .001, t-values available in Table S1.
For this study, we initially recruited 45 healthy participants. Due to dropout (n = 4) and pre-established indices of quality (see SI), the final sample for this analysis was 31 healthy participants. The final sample size provided high power (1-β = 0.90) to study moderate within-subject differences (dz ∼0.6) induced by taVNS versus sham relative to a baseline without stimulation. To ensure eligibility, all participants completed a telephone screening before participation. The study was approved by the ethics committee of the University of Tübingen (reference number 235/2017BO1) and was conducted in accordance with the Declaration of Helsinki. Participants provided written informed consent at the beginning of the first session and received either monetary compensation (56€) or course credit for participation. Furthermore, they received additional money, office supplies, and snacks depending on their performance in two additional tasks (for details, see Ref. [
Each participant completed two sessions of the same standardized protocol (randomized cross-over design, Fig. 1A). Participants received taVNS in one session, and sham stimulation in the other session with the order being randomized in advance. Participants were asked to enter the experimental session neither hungry nor full (no food or caloric beverages <1h prior to each session). To ensure comparable delays to the last meal, they were asked to eat ∼1.5h prior to each session. Next, we measured heart rate, weight, waist and hip circumference, and assessed information about diet, last food, and last drink intake. Afterward, we collected ratings of hunger, fullness, thirst, and mood using the PANAS [
Thereafter, participants were positioned in the fMRI scanner and all electrodes for EGG and electrocardiography (ECG) were placed according to Rebollo et al. [
], the position of the electrode for taVNS stimulation was located at the right cymba conchae, whereas the electrode was placed at the earlobe of the same ear for sham stimulation (see Figure S3). The skin was cleaned at the position of the electrode using a cotton swab and alcoholic disinfectant. Then, electrode pads were wetted with contact fluid and the electrode was placed at the right ear by the experimenter while surgical tape kept the electrode and cable in position. For both electrode positions, we used the same stimulation protocol except for the stimulation strength. We set the stimulation strength individually using pain VAS. We initialized the stimulation with 1 mA and increased stepwise in 0.1–0.5 mA increments until the strength matched the participants’ sensation of “mild pricking” (MtaVNS_strength = 3.56 mA ± 1.07 [1.5–5]). Then, the stimulation was turned off again.
Next, we started the measurements (ECG: 3 channels, EGG: 4 channels) using Brainvision Recorder (BrainProducts, Germany). First, we did an anatomical scan, followed by fieldmaps (∼15min). This was followed by 10min of baseline (no stimulation) rs-fMRI measurement. Then, we turned on the stimulation (taVNS or sham) and measured for another ∼10min. During rs-fMRI, we showed Inscapes videos without audio to improve compliance while minimizing cognitive load and motion [
]. After the onset of the stimulation, taVNS or sham remained active for the rest of the session.
After rs-fMRI measurements, participants completed tasks (∼55min). After exiting the MR scanner, participants completed ratings of hunger, fullness, thirst, and mood again. At the end of each session, we asked participants whether they believed they had received taVNS or sham stimulation in this session (61.3% identified the condition correctly in session 1, pbinomial = .075; 70.1% in session 2, pbinomial = .005). The second session followed the same procedure and was usually conducted 1-7 days later at the same time of the day.
2.3 taVNS device
To administer the auricular stimulation of the vagus nerve, we used the NEMOS® stimulation device (cerbomed GmbH, Erlangen, Germany). It uses a biphasic stimulation protocol with 30s on (25 Hz), followed by 30s off stimulation. The stimulation intensity was set individually (possible range 0.1–5 mA) for each session until participants reported a mild pricking. We placed the stimulation electrode at the cymba conchae (taVNS) or the earlobe (sham) of the right ear [
]. In the current study, we decided to stimulate the right auricular branch of the vagus nerve because previous research had shown generalizing motivational effects [
]. We reasoned that stimulating the right ear may provide further insights into crucial mechanisms of gut-brain communication that may support energy homeostasis.
2.4 EGG acquisition
The EGG data were recorded following the procedure published by Rebollo et al. [
] using Brain Vision Recorder (Brain Products, Germany). To prevent the occurrence of phase differences due to unequal distances to the reference electrode, we used four bipolar electrodes for EGG acquisition. The data were recorded with a sampling rate of 5000 Hz with a low-pass filter of 1000 Hz and no high-pass filter. We recorded EGG (and ECG) continuously throughout the whole session and marked the beginning and end of events using Psychtoolbox.
2.5 fMRI data acquisition and preprocessing
fMRI data were acquired on a Siemens 3T PRISMA scanner equipped with a 64-channel RF receiver head coil and preprocessed with fMRIprep (see SI; [
] (https://github.com/irebollo/stomach_brain_Scripts; see SI). Briefly, this pipeline identifies the spectral peak of the individual gastric rhythm within the slow frequency normogastric band (0.03–0.07 Hz) for each participant and session.
Next, we extracted BOLD time series from rs-fMRI measurements for each voxel inside the brain mask provided by SPM. We then normalized and z-scored these time series. We denoised the signal using regressors for movement, cerebrospinal fluid, white matter, and respiration using the TAPAS toolbox [
]. Then, we split the time series into the same experimental phases as the EGG data (baseline and stimulation) and bandpass filtered both parts around the corresponding individual EGG peak frequency. Next, we performed a Hilbert transformation to obtain the phase information of the BOLD time series.
To determine the phase coupling between the EGG and BOLD time series, we computed the phase-locking value (PLV) for each voxel [
]. The PLV describes the synchrony between two signals, for example, that they de-/accelerate simultaneously, and can take values from 0 (no synchrony) to 1 (perfect synchrony) (Fig. 1A).
To test whether observed phase coupling exceeds levels occurring by chance, we used a permutation approach. We generated permuted phase sequences by computing chance PLV (CPLV), by circularly shifting the EGG phase signal (signal was cut from the end and prepended) and computing the PLV for each shift (see SI [
Next, we analyzed the effect of taVNS on stomach-brain coupling using full-factorial models in SPM12 by computing the interaction effect of Stimulation [taVNS-sham] × Time [stimulation-baseline] and added stimulation order as covariate. Moreover, we analyzed whether taVNS increases the variance in PLV changes using F-tests (taVNS/sham). We performed an exploratory analysis of the correlation between PLV changes (mean over region) and changes in gastric myoelectric frequency, BMI, and taVNS strength to evaluate potential influential factors that could explain part of the variance. To investigate if metabolic state mediates this effect of taVNS on stomach-brain coupling, we calculated the correlation between changes in coupling due to taVNS (ΔPLVtaVNS [stimulation-baseline] - ΔPLVsham [stimulation-baseline]) and changes in metabolic state over the session (post [hunger-satiety] - pre [hunger-satiety]).
To analyze the subjective metabolic state (VAS before and after measurement), we used linear mixed-effects models in lme4 v1.1-21 [
]: rating ∼ stimulation ∗ time+ (1 + stimulation | ID) (BIC = 1292.5; for the gradient analysis, see SI).
2.6.2 Statistical threshold and software
Our analyses were conducted with MATLAB v2019-2020 and SPM12 (for a full list, see SI). We considered α < .05 as significant, except for FWE cluster-corrected t-maps, where we used p < .001 (k ≥ 40) as cluster-forming threshold. The cluster-extent threshold was estimated with AlphaSim [
] based on the smoothness of the maps (αcorrected < .05, 2000 iterations). For baseline, we thresholded the empirical vs. chance PLV maps using a Monte-Carlo permutation test (cluster-α = .005, α = .025, 10,000 permutations).
3. Results
We analyzed phase coupling between the fMRI BOLD signal (providing high spatial resolution in deep brain regions) and the myoelectric signal of the stomach using PLV, a measure reflecting the synchrony between time series. First, we successfully replicated the topography of the gastric network as described before [
To evaluate whether stomach-brain coupling can be enhanced by taVNS, we compared PLV maps during taVNS and sham stimulation (relative to baseline) using a full-factorial model (i.e., Stimulation × Time interaction). Within our a priori ROIs, taVNS increased stomach-brain coupling in the NTS (Fig. 2A; psvc = .002, tmax = 4.33). On the individual level, 67.8% (21/31; pbinom = .015) of the participants showed taVNS-induced increases in PLV in the NTS (vs. 45.2% or 14/31 for sham, pbinom = .64), and 64.5% (20/31) participants showed relative increases in PLV after taVNS compared to sham. Moreover, we observed increases in the dopaminergic midbrain (Fig. 2A; combined mask VTA&SN: psvc = .037, peak in VTA: tmax = 3.40, psvc = .006). At a whole-brain level (corrected for multiple comparisons), taVNS increased stomach-brain coupling in transmodal cortical regions, namely in the mid frontal gyrus (Fig. 2B; tmax = 4.97, k = 63) and the precuneus (tmax = 4.25, k = 54). Also, we found an increased inter-individual variance in stomach-brain coupling during taVNS versus sham in several regions (p < .01, F(30,30) >2.39), such as the anterior cingulate cortex (Fig. 2B; F(30,30) = 2.96, p = .002), the left secondary somatosensory cortex (F(30,30) = 2.94, p = .002), the right insular cortex (F(30,30) = 2.66, p = .005), and the right supramarginal gyrus (F(30,30) = 2.52, p = .007), indicating taVNS-induced perturbation.
Fig. 2Transcutaneous vagus nerve stimulation (taVNS) increases stomach-brain coupling. A: taVNS boosts coupling in key target regions of vagal afferents: the NTS (tmax = 4.33, psvc = .002) and the dopaminergic midbrain (peak in the VTA/SN: psvc = .037, tmax = 3.40). B: taVNS (vs. sham) increases coupling in the precuneus/superior parietal lobe (tmax = 4.25, pFWE = .041) and the mid frontal gyrus (tmax = 4.97, pFWE = .019), and increases variance in PLV changes in several regions such as in both insular cortices and the anterior cingulate (AC) cortex. See also Figure S1. C: taVNS boosts coupling in regions that are intrinsically coupled with the stomach at baseline as indicated by Pearson correlations between baseline and stimulation interaction effects: (mean dz over each region: r = 0.21, p = .009; maximum dz within each region: r = 0.44, p < .001). D: Selected regions with the highest voxel-wise effect sizes (dots and error bars show M ± SD). VTA: ventral tegmental area, Cereb: cerebellum, NR: nucleus. ruber, PC: posterior cingulate, STN: subthalamic ncl., NTS: nucleus tractus solitarius, Ver: vermis, CO: central opercular cortex, HG: Heschl's gyrus, PCu: precuneus, MidFG: mid frontal gyrus, sLOC: superior lateral occipital cortex, TP: temporal pole, FOrb: frontal orbital cortex.
To assess potential influences of taVNS strength on stomach-brain coupling, we calculated the correlation between taVNS strength and changes in PLV (taVNS-sham). We found strong correlations in both insular cortices (Fig. S2, maps on NeuroVault; right: tmax = 4.52, k = 102; left: tmax = 4.10, k = 190) and the right supramarginal gyrus (tmax = 4.61, k = 74). Crucially, these regions showed an increased variance during taVNS, indicating that inter-individual differences in stimulation strength (matched according to sensory aspects) might contribute to the observed variance. Still, gastric myoelectric frequency did not change during taVNS in our sample [
To evaluate whether taVNS increases coupling in regions that are intrinsically coupled at baseline, we computed the correlation between the effect sizes (dz) of each region for the baseline and for taVNS-induced changes (taVNS–sham) in PLVs. Both mean (r = 0.21, p = .009), and maximum effect sizes within each region (r = 0.44, p < .001) were correlated, demonstrating that taVNS boosted stomach-brain coupling in regions with higher intrinsic coupling at baseline (Fig. 2C, Table S2). However, the largest taVNS effects on PLVs were observed in regions outside the baseline gastric network, notably in the frontoparietal network, the default-mode network (DMN), the midbrain, and the brainstem (Fig. 2D, networks according to Ref. [
]. These gradients characterize rs-networks in terms of a cortical functional hierarchy, where the first gradient extends from unimodal to transmodal (with two peaks at the extremes) and the second gradient from visual to somatomotor-auditory regions (Fig. 3A). In line with Rebollo et al. (2021), the baseline gastric network was primarily associated with unimodal cortical regions (Fig. 3B; p < .001 deviation from the standard rs-gradient, Kolmogorov-Smirnov test). Intriguingly, taVNS shifted this distribution towards transmodal cortical regions, such as the DMN and the frontoparietal network (Fig. 3C; p < .001).
Fig. 3Non-invasive vagus nerve stimulation (taVNS) primarily boosts coupling in transmodal cortical regions. A: Distribution of the cortical gradients described by Margulies, Ghosh, Goulas, Falkiewicz, Huntenburg, Langs, Bezgin, Eickhoff, Castellanos, Petrides, Jefferies, Smallwood [
]. Each dot depicts a bin of voxels (100 bins per gradient) and was color-coded according to the predominant Yeo network in each bin. Gradient 1 extends from uni-to transmodal regions while gradient 2 extends from visual to somato-motor-auditory regions. B: Comparison of the distribution of voxels that show significant taVNS-induced changes in coupling (highlighted) with the standard cortical gradients described by Margulies, Ghosh, Goulas, Falkiewicz, Huntenburg, Langs, Bezgin, Eickhoff, Castellanos, Petrides, Jefferies, Smallwood [
] depicted in grey. The colors indicate the numbers of voxels per bin. In line with Rebollo et al. (2021), the voxels of the baseline gastric network belonged primarily to unimodal regions, such as visual and somato-motor areas. C: In contrast, taVNS increased stomach-brain coupling primarily in transmodal regions, such as the default mode network and the frontoparietal network, which is in line with an increase in parasympathetic modulation. D: Distribution of voxels that show a significant correlation between taVNS-induced changes and sensed changes in hunger (highlighted) compared to the standard cortical gradients described by Margulies, Ghosh, Goulas, Falkiewicz, Huntenburg, Langs, Bezgin, Eickhoff, Castellanos, Petrides, Jefferies, Smallwood [
] depicted in grey. Intriguingly, taVNS-induced changes in stomach-brain coupling were correlated with subjectively sensed changes in hunger mostly in transmodal cortical regions as well, suggesting that changes in stomach-brain coupling in transmodal cortical regions may play a role in sensing interoceptive signals related to hunger (i.e., metabolic state). (For interpretation of the references to color in this figure legend, the reader is referred to the Web version of this article.)
3.2 taVNS strengthens the link between changes in stomach-brain coupling and changes in hunger
After demonstrating that taVNS boosts stomach-brain coupling, we analyzed whether changes in subjective ratings of metabolic state (assessed before and after the measurement) were associated with individual differences in taVNS-induced changes. These state ratings showed an increase in hunger during the session (p < .001) with inter-individual variability but did not differ depending on the stimulation condition (Fig. 4A; plme = .482). However, we found that taVNS increased the correlation of changes in stomach-brain coupling with changes in hunger ratings in regions across the cortex (Fig. 4B–E), such as the frontal poles (right: tmax = 6.49, k = 40; left: tmax = 5.16, k = 137), the cerebellum (tmax = 5.14, k = 200), the fusiform gyrus (tmax = 4.63, k = 92), the precuneus (left: tmax = 4.54, k = 100; right: tmax = 4.49, k = 67), the inferior parietal lobe (tmax = 4.25, k = 60), and the declive (tmax = 4.41, k = 55, Table S3). In line with the main effect of taVNS, the correlation of taVNS-induced changes in PLV with hunger ratings showed a shift towards transmodal cortical regions (Fig. 3D; p < .001 compared to the cortical gradients by Ref. [
]). Collectively, these findings support the idea that taVNS-induced changes in stomach-brain coupling can be sensed, indicating that emulated input via vagal afferents may act as interoceptive signal.
Fig. 4taVNS increases the correspondence of changes in stomach-brain coupling with subjective changes in metabolic state. A: Participants became hungrier during the scan (p < .001), but there was no interaction with stimulation (p = .482). B: taVNS increased the correlation between increases in hunger and changes in stomach-brain coupling across the cortex (density plot of the change in correlation (taVNS vs sham) of all cortical voxels, positive values indicate higher correlation). C: Clusters of voxels showing significant increases in correlation with subjective changes in metabolic state after whole-brain correction for multiple comparisons: frontal poles (tmax = 6.49), fusiform gyrus (tmax = 4.63), precuneus (tmax = 4.54), and inferior parietal lobe (tmax = 4.25; Table S3, maps on NeuroVault). D: taVNS increases the correlation between PLV changes and metabolic state ratings in the frontal poles (regions from Harvard-Oxford brain atlas). E: During taVNS, several regions showed an increased correlation between changes in coupling and metabolic state and the strongest effects were found in the DMN, the visual network, the frontoparietal network, and the dorsal attention network. Correlation was calculated voxelwise across each ROI, bars show the maximum r, dots and error bars indicate mean ± standard deviation. [FP: frontal pole, Cereb: cerebellum, TOFusC: temporal occipital fusiform gyrus, sLOC: superior lateral occipital cortex, iLOC: inferior lateral occipital cortex, toITG: temporooccipital inferior temporal gyrus, MidFG: mid frontal gyrus, pMTG: posterior middle temporal gyrus, PostCG: postcentral gyrus, PC: posterior cingulate, PreCG: precentral gyrus].
], but the lack of causal manipulations has so far hampered progress in delineating its function in humans. Here, we used non-invasive taVNS to emulate the activation of vagal afferents that are known to regulate the intrinsic rhythm of the stomach via a vago-vagal pathway [
]. For the first time in humans, we show that taVNS applied at the right ear boosts stomach-brain coupling in the NTS and the dopaminergic midbrain, a pattern that strongly resembles the pathway for gut-induced reward via the right nodose ganglion previously described in rodents [
]. Furthermore, we demonstrate increased coupling in transmodal cortical regions, as well as in regions that display high intrinsic coupling under baseline conditions. Crucially, we also show that taVNS increases the correlation between changes in stomach-brain coupling and in subjectively sensed hunger, again primarily in transmodal cortical regions. Consequently, we conclude that taVNS modulates stomach-brain coupling via an NTS-midbrain pathway, while changes in cortical coupling might reflect stronger integration of interoceptive signals encoding the current metabolic state of the body. Collectively, these results point to a vital role of the gastric network in supporting the maintenance of energy homeostasis opening new avenues for neuromodulatory treatments.
4.1 Anatomical pathways for the taVNS effect on the gastric network
Since the NTS is the main entry point of vagal afferents to the brain and is robustly activated by taVNS [
], we expected taVNS to increase stomach-brain coupling, which was confirmed. The NTS is a crucial part of the gastric-vagal reflex: vagal afferents delivering interoceptive information from the stomach reach the NTS, get integrated and projected to the dorsal vagal motor nucleus, which then sends efferent commands back to the stomach [
]. Previous findings also suggest the involvement of this pathway in the efferent modulation of gastric frequency following vagal stimulation in humans [
], it is known that the NTS projects to the parabrachial nuclei and the thalamus which acts as a relay station to cortical regions of the gastric network [
]. In line with previous work, we found that taVNS boosts stomach-brain coupling in the VTA/SN, which adds to the growing literature suggesting a role for dopamine in the control of digestion [
]. From a clinical perspective, increased stomach-brain coupling in the VTA/SN provides potential applications to prevalent gastrointestinal symptoms in several disorders, such as mood and anxiety disorders [
]. To summarize, our results suggest that a vagal-midbrain pathway may facilitate the integration of metabolic signaling into larger brain networks potentially shaping goal-directed behavior to ensure that metabolic demands are met.
4.2 The role of the gastric network in maintaining metabolic balance
Since feeding and digestion are vital functions of the digestive system, the gastric network likely supports the long-term maintenance of energy homeostasis [
]. Our finding that taVNS increases the correlation between changes in stomach-brain coupling and subjectively sensed changes in metabolic state provides intriguing support for the alleged link between the gastric network and interoception. This is in line with work showing that activity in the gastric network is related to future weight loss [
]. We found that transmodal regions showed the highest increase in correlation with sensed hunger during taVNS. This possibly reflects the hierarchically overarching integration of metabolic feedback, which is buffered in unimodal cortical areas, into larger transmodal brain networks after vagal afferent activation. The strong involvement of visual regions might reflect increased visual attention [
]. Relatedly, the highest increase in correlation was found in the frontal poles, which are part of the salience network and DMN, and are involved in goal-directed behavior and metacognition [
]. This supports the idea of a relationship between hunger-dependent alterations in stomach-brain coupling and food-seeking behavior. We conclude that vagal afferent stimulation facilitates the integration of metabolic information from unimodal representations into a transmodally-extended gastric network, which may help coordinate behavioral adaptation to ensure long-term energy homeostasis.
4.3 Gastric signaling as a common reference point for intero- and exteroception
The constant rhythmical input from the stomach may act as a common reference point for the alignment and coordination of both interoceptive and exteroceptive brain networks [
]. Our finding of increased coupling in transmodal cortical regions during taVNS (vs. unimodal cortical regions during baseline) provides the first causal support for this theory. As we emulated interoceptive signals by afferently stimulating the vagus nerve, the stronger involvement of higher-order transmodal regions, which integrate activity across sensory-motor/unimodal regions [
], might mirror the coordinated hierarchical integration of altered visceral inputs across brain networks. Our results therefore suggest that the baseline gastric network reflects ongoing integration of low-level interoceptive information. Once vagal afferents are activated, the gastric network expands to reflect heightened integration across modalities, leading to changes in the awareness of the metabolic state of the body which may facilitate adaptations of goal-directed behavior, such as food-seeking. Crucially, the higher-order integration of the transmodal cortex during taVNS is in line with a boost in parasympathetic activity, whereas the baseline gastric network is affected by sympathetic regulation [
Strikingly, several cortical regions also show increased variance in stomach-brain coupling during taVNS (vs. sham), such as the insula and the ACC. The insula displays a viscerotopic organization [
]. While BMI or changes in gastric frequency did not explain an increased variance during taVNS, we found strong associations between PLV changes and taVNS strength in the bilateral insula and the right supramarginal gyrus. These findings pose the intriguing question whether stomach-brain coupling could be parametrically modulated by adjusting the stimulation strength, which requires extended designs in the future. Relatedly, examining spatiotemporal dynamics of digestive signals during taVNS beyond gastric myoelectric frequency (a gross spatial mean) could provide more mechanistic insights into the intricate progression of signals along the stomach curvature [
]. We conclude that taVNS boosts the integration of interoceptive signaling in transmodal cortical regions, potentially modulating arousal in a parasympathetic network.
4.4 Limitations
Despite the novel insights provided by our unique investigation combining taVNS with concurrent fMRI and EGG recordings, there are limitations that will need to be addressed in future work. First, it is not known whether the duration of taVNS alters effects on stomach-brain coupling. Our design used a duration of 10min for each experimental phase, analogous to baseline measurements of previous work [
]. However, longer taVNS periods might elicit different changes in stomach-brain coupling as they may evoke effects on gastric frequency and motility which have been demonstrated after longer stimulation periods [
]. Second, it is not known whether left-vs. right-sided stimulation leads to comparable effects on the gastric network, as lateralization differences between visceral fibers of the vagus nerve might play a role [
] may also boost effects and future work may explore how taVNS-induced effects can be optimized. Third, we invited participants in an intermediate metabolic state (neither hungry, nor full). Future studies should include objective measures (e.g., glucose levels) and systematically investigate different states using more extensive interoceptive measures (such as interoceptive awareness and accuracy [
]). Fourth, potential interactions of the gastric network with exteroceptive input (e.g., visual or olfactory food cues) should be examined to gain insights into the alleged link between the gastric network and food-seeking behavior. Fifth, we provide the first evidence that taVNS increases stomach-brain coupling, but larger studies are necessary to investigate potential inter-individual differences with sufficient power.
4.5 Summary and conclusion
Procuring sufficient energy for survival is crucial and the recently discovered gastric network may play a vital role in ensuring long-term energy homeostasis. In support of this idea and in accordance with the previously established pathway for gut-induced reward [
], we showed that non-invasive stimulation of vagal afferents at the right ear via taVNS increases stomach-brain coupling in the NTS and the dopaminergic midbrain. Furthermore, we demonstrated increased stomach-brain coupling in transmodal cortical regions, including an increased correlation with changes in ratings of hunger. We conclude that taVNS is an effective method to modulate stomach-brain coupling, providing an important technique to causally study the function of the gastric network. Consequently, taVNS may open an avenue to future treatments of gastrointestinal, or more broadly, somatic symptoms of a broad array of disorders. In addition, work on vagal afferent signaling using the experimental control provided by taVNS could help us better understand inter-individual differences in non-homeostatic eating, such as emotional or disinhibited eating.
5. Data and code availability
The analyses are based on code published by Rebollo, Devauchelle, Beranger, Tallon-Baudry [
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.
Acknowledgment
We thank Sandra Neubert, Vinzent Wolf, Franziska Müller, Corinna Schulz, Wy Ming Lin, Monja Neuser, and Franziska Kräutlein for help with data acquisition. The study was supported by the University of Tübingen, Faculty of Medicine fortune grant #2453-0-0. Additional support was provided by a postdoctoral scholarship 32-04/19 by the Daimler and Benz Foundation as well as the grant KR 4555/7-1 by the Deutsche Forschungsgemeinschaft (DFG) awarded to NBK. We acknowledge support by the Open Access Publishing Fund of the University of Tübingen.
Appendix A. Supplementary data
The following is the Supplementary data to this article:
00197-8&id=ga1.jpg){kind=link}
00197-8&id=gr1.jpg){kind=link}
00197-8&id=gr2.jpg){kind=link}
00197-8&id=gr3.jpg){kind=link}
00197-8&id=gr4.jpg){kind=link}