Right vagus nerve stimulation promotes motor cortical plasticity Authors: Hailey F Welch, Meena Rajendran, Catherine A Thorn Published: 2026 ABSTRACT AbstractStimulation of the left cervical vagus nerve (l-VNS) paired with motor training is FDA-approved to improve upper limb function after stroke. However, recent preclinical studies have revealed a striking asymmetry in engagement of the midbrain dopaminergic reward system by right versus left VNS. Unlike l-VNS, right VNS (r-VNS) reinforces behavior and drives activation of midbrain dopaminergic neurons. These findings raise the possibility that r-VNS may provide therapeutic advantages by engaging plasticity- or motivation-enhancing dopaminergic signaling. In the current study, we tested whether r-VNS enhances VNS-driven motor cortical plasticity or behavioral performance compared to traditional l-VNS. Healthy rats were trained on a stimulation-paired lever-pressing task previously shown to predict the efficacy of l-VNS treatment post-stroke. Both l-VNS and r-VNS significantly expanded the representation of task-relevant musculature in the motor cortex, with no differences observed between stimulation groups. Like l-VNS, r-VNS was not accompanied by increases in task engagement or percent correct performance in healthy subjects. Combined, our findings do not support the hypothesis that r-VNS provides significant benefit over l-VNS through differential engagement of motor learning or motivational circuits. However, our results suggest that r-VNS, like l-VNS, is highly effective at inducing cortical neuroplasticity, which is associated with enhanced stroke recovery.Keywords: Vagus nerve stimulation, Lateralization, Asymmetry, Stroke, Motivation, Motor learningSubject terms: Neuroscience, Physiology Introduction Stimulation of the left vagus nerve (l-VNS) paired with motor training was approved by the United States Food and Drug Administration in 2021 as a therapy to improve upper limb motor function in chronic stroke. When paired with post-stroke rehabilitative training in both preclinical and clinical settings, l-VNS produces robust, long-lasting improvements in impaired motor function1–7. In healthy rats, training paired l-VNS has repeatedly been shown to enhance the somatotopic representation of task-relevant musculature in the motor cortex (M1)8–15. The stimulation parameters that induce M1 plasticity in healthy rats are highly correlated with those found to be effective at enhancing upper limb function following stroke in both rats1–4 and humans5,6,16, providing a highly translational assay for predicting VNS efficacy. Traditional l-VNS is thought to work by driving plasticity-promoting neuromodulation throughout the central nervous system. When paired with a motor task, VNS-induced enhancement of noradrenergic, cholinergic, and serotonergic signaling facilitates task-relevant cortical neuroplasticity9,10,14,16. Interestingly, while dopamine-mediated reinforcement is well known to be central to motor learning, neurotoxic lesions to motor cortex-projecting dopaminergic neurons do not block l-VNS induced neuroplasticity15. Moreover, stimulation of the left vagus nerve fails to promote conditioned place preferences, reinforce operant behaviors, or drive early gene expression in the ventral tegmental area (VTA) or the substantia nigra pars reticulata (SNc)17–20. Thus, l-VNS does not appear to significantly activate midbrain dopaminergic circuits to drive M1 neuroplasticity or functional motor recovery. However, recent studies have revealed a striking asymmetry in the engagement of the midbrain dopaminergic reward system by right versus left VNS. Unlike l-VNS, r-VNS has been shown to reinforce operant behaviors and increase activation of dopaminergic neurons in the VTA and SNc17,20. Thus, while l-VNS recruits multiple plasticity promoting neuromodulatory systems, r-VNS uniquely activates the dopaminergic reward system. L-VNS is the clinical standard due to concerns arising largely from anatomical studies that r-VNS may carry a higher risk of adverse cardiac effects21–23. However, more recent preclinical and clinical studies indicate that r-VNS is safe and well-tolerated24–29. Combined, these findings raise the question of whether r-VNS could provide therapeutic benefits over l-VNS during stroke recovery by additionally engaging dopaminergic neuromodulation. Cortical and subcortical dopaminergic signaling is known to contribute to mood and motivation, as well as movement quality and motor learning30–32. Dopamine release driven by r-VNS could perhaps speed motor recovery following injuries such as stroke or increase motivation to complete difficult rehabilitation exercises. To directly compare the effects of r-VNS and l-VNS, we used a well-established translational rat assay that has previously been shown to predict the efficacy of VNS in preclinical and clinical stroke studies1,8,11,13. Healthy rats received l-VNS, r-VNS, or sham stimulation paired with a skilled reaching lever press task to determine the extent to which VNS treatments induced task-relevant motor map plasticity, improved motor performance, or increased task engagement. We found that training-paired r-VNS was as effective as l-VNS at driving cortical map plasticity, and that neither r-VNS nor l-VNS significantly improved behavioral performance. Our results suggest that training-paired r-VNS is unlikely to drive behaviorally relevant DA-mediated improvements to mood, motivation, or motor performance that enhance its efficacy over traditional l-VNS therapy. However, in subjects for which l-VNS is not feasible, due to mechanical damage, for example, training-paired r-VNS may offer a potential alternative approach that is likely to be effective at enhancing functional recovery. Results To examine the extent to which r-VNS and l-VNS produce differential effects on cortical plasticity and motor behavior, we leveraged a skilled reaching lever press assay (Fig. 1A, B) that has repeatedly been shown to enhance the somatotopic representation of task-relevant musculature in the motor cortex (M1) when paired with l-VNS8–11,13–15. Once criterial behavioral performance was reached on the lever press task, animals were randomly assigned to r-VNS (n = 8), l-VNS (n = 8), or sham (n = 7) treatment groups and implanted with a VNS cuff electrode around the right or left cervical vagus nerve (Fig. 1C). Prior to Treatment, group characteristics and behavioral performance were equivalent across treatment groups (Table 1). After surgical and behavioral recovery, rats received 5 treatment-paired lever press sessions prior to undergoing ICMS motor mapping. Experimental design. (A) Rats were implanted with VNS cuff electrodes and were trained to perform a skilled reaching lever press task with their right forelimb. (B) Example traces illustrating lever deflection during hit (yellow line) versus miss (brown line) trials. Rats received a food reward pellet immediately upon successful completion of trials in which they fully depressed (deflection > 9.5 deg; Hit Threshold) and released (deflection returns to < 4.75 deg; Release/Reward Threshold) the lever within a 2-sec window (grey box). (C) Experimental timeline. After criterial performance was reached during Acquisition, rats were implanted with a VNS cuff electrode around the left or right cervical vagus nerve. Rats then continued to train until criterial performance was re-established. During the last 5 sessions of training, r-VNS, l-VNS, or sham simulation was paired with each correct lever press. The Sham group included animals with left (n = 3) and right (n = 4) implants. M1 mapping occurred ca. 24 h after the final training-paired treatment session. Table 1Pre-Treatment group characteristics and behavioral performance.ShamMean ± SEML-VNSMean ± SEMR-VNSMean ± SEMANOVAp-value [F]Age (weeks), Pre-treatment53.3 ± 4.552.2 ± 4.459.4 ± 6.40.577 [0.57]Weight (g), Pre-treatment261.2 ± 10.0273.2 ± 8.9260.6 ± 11.40.613 [0.50]Sessions to criteria, pre-surgery29.7 ± 5.340.8 ± 9.231.4 ± 8.00.572 [0.58]Sessions to criteria, post-surgical recovery10.1 ± 2.910.4 ± 2.117.1 ± 4.20.241 [1.53]Percent correct, Pre-treatment82.7 ± 3.787.0 ± 3.287.8 ± 1.60.442 [0.85]Trials per session, Pre-treatment147.7 ± 13.0146.0 ± 15.1153.2 ± 10.90.920 [0.08]Trial duration (s), Pre-treatment0.42 ± 0.070.43 ± 0.070.40 ± 0.060.953 [0.05]Lever press speed (deg/s), Pre-treatment256.3 ± 26.1217.6 ± 13.3237.0 ± 24.40.472 [0.78] Pre-Treatment group characteristics and behavioral performance. r-VNS drives task-relevant reorganization of cortical motor maps Motor maps were compared across treatment groups to determine whether training-paired r-VNS or l-VNS resulted in quantifiable changes in the somatotopic map representations in M1 (Fig. 2A). Consistent with prior studies, we found that l-VNS paired with lever press training significantly increased the cortical motor map representation of the task-relevant proximal forelimb (PFL) musculature (Fig. 2B; Tables 2 and 3). Importantly, r-VNS treatment also induced a highly significant expansion of the PFL representation, with no differences detected relative to l-VNS. Both l-VNS and r-VNS treatments drove a simultaneous decrease in the map representation of distal forelimb (DFL) musculature (Fig. 2C; Tables 2 and 3). Anterior (head, vibrissa, and neck) and posterior (trunk, hindlimb, and tail) body representations did not differ across treatment groups (Table 2), nor did the overall size of the cortical maps (Fig. 2D; Table 2). Combined, our findings demonstrate that, when paired with motor training, both l-VNS and r-VNS are effective at inducing task-relevant map reorganization in M1. Both l-VNS and r-VNS induce task-relevant M1 plasticity. (A) M1 motor maps from representative sham, left VNS (Left), and right VNS (Right) animals. PFL: proximal forelimb, DFL: distal forelimb, Ant. Body: anterior body representations (jaw, vibrissa, and neck), Post. Body: posterior body representations (trunk, hindlimb, and tail), NR: no response. (B-C) Rats in l-VNS and r-VNS treatment groups exhibited a significant increase in the size of the PFL representation in M1 (B) and a decrease in the size of the DFL representation (C) compared to sham treated rats. **p < 0.01, ***p < 0.001, ****p < 0.0001, Bonferroni corrected t-tests. (D) Overall map size was similar across treatment groups. Black circles in B-D denote data from individual animals; Error bars denote SEM. Table 2Effects of treatment on motor cortical map representations.Map representation (mm2)Shammean ± SEMl-VNSmean ± SEMr-VNSmean ± SEM1-way ANOVAp-value [F]Proximal forelimb1.0 ± 0.13.7 ± 0.63.5 ± 0.40.001 [10.02]Distal forelimb4.3 ± 0.41.5 ± 0.22.1 ± 0.3< 0.0001 [24.39]Anterior body4.1 ± 0.43.6 ± 0.54.5 ± 0.60.507 [0.7]Posterior body2.2 ± 0.22.2 ± 0.42.6 ± 0.30.635 [0.46]Total map size11.6 ± 0.510.9 ± 1.412.6 ± 0.90.507 [0.7]Significant values are in bold. Effects of treatment on motor cortical map representations. Significant values are in bold. Table 3Pairwise between-group comparisons of proximal forelimb (PFL) and distal forelimb (DFL) representations.ComparisonPFLDFL p Effect size p Effect sizeSham vs. l-VNS0.0020− 1.88 < 0.0001 3.23Sham vs. r-VNS0.0034− 2.65 0.0001 2.19l-VNS vs. r-VNS1.00− 0.100.46450.87Reported p-values are from posthoc Bonferroni-corrected t-tests, and effect sizes are Hedge’s g. Significant values are in bold. Pairwise between-group comparisons of proximal forelimb (PFL) and distal forelimb (DFL) representations. Reported p-values are from posthoc Bonferroni-corrected t-tests, and effect sizes are Hedge’s g. Significant values are in bold. r-VNS did not improve behavioral performance Prior studies demonstrate that training-paired l-VNS drives cortical map plasticity that is unaccompanied by significant changes in behavioral performance in healthy, well-trained rats8–11,14,15. To assess the impacts of r-VNS on lever-press behavior, we first asked whether any treatment-specific effects on percent correct performance could be observed in our study. We found no differences between l-VNS, r-VNS, and sham treatment groups in percent correct performance during the Pre-Treatment (Table 1) or Treatment (Table 4) periods, nor was there a significant treatment-dependent change in percent correct performance within any group (Fig. 3A, B; Table 4). These results suggest that, similar to l-VNS, training-paired r-VNS does not significantly enhance or impair successful task performance in healthy, well-trained rats. Table 4Effects of treatment on behavioral performance.Behavioral parameterShammean ± SEMl-VNSmean ± SEMr-VNSmean ± SEM1-way ANOVAp-value[F value]Weight (g)261.7 ± 11.0278.9 ± 9.1262.4 ± 12.10.452 [0.83]Percent correct performance82.6 ± 4.486.9 ± 3.585.1 ± 2.40.687 [0.38]Trials per Session143.3 ± 8.1153.3 ± 15.4124.2 ± 12.90.273 [1.39]Trial duration (s)0.40 ± 0.070.43 ± 0.060.40 ± 0.050.907 [0.10]Lever pressing speed (deg/s)251.5 ± 25.2218.9 ± 12.0231.9 ± 22.70.544 [0.63]Δ Weight (% Pre-Treatment)100.2 ± 0.9102.1 ± 0.5100.7 ± 0.90.213 [1.67]Δ Percent Correct (% Pre-Treatment)99.6 ± 2.099.9 ± 1.797.2 ± 2.90.659 [0.43]Δ Trials per session (% Pre-Treatment)100.4. ± 8.8107.9 ± 10.381.7 ± 7.20.116 [2.40]Δ Trial duration (% Pre-Treatment)97.9 ± 5.8103.7 ± 8.3108.1 ± 12.50.762 [0.28]Δ Lever pressing speed (% Pre-Treatment)98.3 ± 1.9101.2 ± 3.399.2 ± 5.20.864 [0.15] Effects of treatment on behavioral performance. VNS did not alter task execution. (A–F) No significant differences were observed across treatment groups in percent correct performance (A), average trial duration (C), or average lever pressing speed (E) during Pre-Treatment (-) or Treatment (+) periods, and no treatment produced a significant change in percent correct (A, B), trial duration (C, D), or lever pressing speed (E, F). Black circles in B, D, and F represent data from individual animals; Error bars represent SEM. Increases in tonic or phasic dopamine release have also been correlated with improved movement coordination and increased motivation30–32, which may not be associated with an overall change in percent correct performance in our lever press assay. To gain further insight into whether r-VNS might be impacting movement execution or motivation, we analyzed the effect of stimulation on trial durations and lever pressing speeds. Both trial durations (Fig. 3C, D) and pressing speeds (Fig. 3E, F) were similar between groups prior to Treatment, and neither measure was altered by any treatment (Table 4). We next analyzed the number of trials performed per session to assess the animals’ motivation to engage in the lever press task. Both prior to Treatment (Table 1), and during Treatment (Table 4), there were no differences between groups in the number of trials performed per session. A one-way ANOVA revealed no significant differences across the 3 treatment groups in stimulation-driven changes in the number of trials performed (Fig. 4A, B). However, within-group comparisons did reveal a statistically significant reduction in the number of trials performed during r-VNS Treatment sessions compared to the Pre-Treatment period (Table 5). No treatment-dependent changes in task engagement were observed in either the sham or l-VNS stimulation groups (Table 5). These results suggest that training-paired r-VNS does not drive a significant increase in motivation to perform the behavioral task. Rather, our analyses raise the possibility that r-VNS may be associated with an unexpected reduction in task engagement. r-VNS treatment was associated with a decrease in task engagement. (A) A significant decrease in number of trials performed was observed between the Pre-Treatment (-) and Treatment (+) periods in the r-VNS group, but not in the sham or l-VNS groups; *p < 0.05, paired t-test. (B) Across treatment groups, there was no significant difference in the treatment-dependent change in trials performed, despite the reduction seen within the r-VNS group. (C) Average pellets rewarded across session time during Treatment (solid lines) and Pre-Treatment (dashed lines) periods. r-VNS Treatment was associated with a significant reduction in pellets received across all time bins, compared to Pre-Treatment; *all FDR q < 0.05, time-resolved contrast analysis. Shading denotes SEM. (D) Across all rats, the relationship between average weight and average number of pellets rewarded per session trended toward significance (Pearson’s linear correlation); dots represent average data across all 10 Pre-Treatment and Treatment sessions for each rat in the study. (E) Variations in weight around each animals’ average are anticorrelated with variation in pellets received, suggesting task engagement is sensitive to caloric need. Dots represent data from each Pre-Treatment and Treatment session for each rat in sham (black), l-VNS (blue), and r-VNS (red) groups; text insets are Pearson’s correlation coefficient (r) and p-value for all sessions from all rats (black) and from the r-VNS treated group only (red); dashed lines and shading denote linear fit and 95% confidence intervals for all sessions from all rats (gray) and from the r-VNS treated group (red). Table 5Within-group comparisons of average trials performed per session during pre-treatment vs. treatment periods.Treatment groupPre-treatment(mean ± SEM)Treatment(mean ± SEM)Paired t-test(p)Effect size(Cohen’s dz)Sham147.7 ± 13.0143.3 ± 8.10.743− 0.13l-VNS146.0 ± 15.1153.3 ± 15.40.6710.16r-VNS1.00− 0.10 0.039 − 0.90Significant values are in bold. Within-group comparisons of average trials performed per session during pre-treatment vs. treatment periods. Significant values are in bold. Consistent with the decrease in total number of trials performed, we also observed that the average number of pellets rewarded during the 60-minute sessions was significantly lower during Treatment in the r-VNS group (r-VNS group, number of hits performed, Pre-Treatment: 132.2 ± 9.9, Treatment: 105.4 ± 11.8, mean ± SEM; p = 0.032, paired t-test; effect size = − 0.94, Cohen’s dz), but not in the Sham or l-VNS groups (Fig. 4C; Number of hits performed, Sham group, Pre-Treatment: 117.0 ± 10.1, Treatment: 115.8 ± 11.7, mean ± SEM; p = 0.915, paired t-test; effect size = − 0.04, Cohen’s dz; l-VNS group, Pre-Treatment: 126.4 ± 16.7, Treatment: 126.9 ± 13.77, mean ± SEM; p = 0.968, paired t-test; effect size = 0.01, Cohen’s dz). To analyze this effect in more detail, we compared the average time course of hits performed across session during Pre-Treatment vs. Treatment periods using a linear mixed effects model (Fig. 4C). Modeling confirmed an overall effect of treatment (Treatment period vs. mean effect: p = 0.004, β = − 0.111, CI = [− 0.185, − 0.036]), which was significant in the r-VNS group (Group x Treatment, r-VNS deviation: p = 0.001, β = − 0.169, CI = [− 0.272, − 0.065]); l-VNS deviation: p = 0.078, β = 0.093, CI = [− 0.011, 0.197]). A significant decline in hits performed across session-time was observed for all groups (overall Time effect: p < 0.001, β = − 0.053, CI = [− 0.055, − 0.051]), and we observed a significant Group x Time deviation for the l-VNS, but not the r-VNS group (Group x Time effect, r-VNS deviation: p = 0.271, β = − 0.002, CI = [− 0.005, 0.001]; l-VNS deviation: p < 0.001, β = − 0.006, CI = [− 0.009, − 0.003]). Time-resolved contrast analysis revealed that for the r-VNS group, the reduction in hits performed during the stimulation period was significant at all time points (all FDR q < 0.05). As no significant Treatment x Time interactions were observed (overall Treatment x Time effect: p = 0.296; r-VNS x Treatment x Time, p = 0.637; l-VNS x Treatment x Time: p = 0.789), these analyses suggest that overall task engagement was reduced throughout the session during the r-VNS treatment period, but that r-VNS did not significantly alter the rate of responding across session-time. We have previously shown that task engagement and pellet consumption in the lever pressing task is significantly correlated with weight and caloric need13. We therefore examined whether the decrease in presses and hits seen in the r-VNS group during the Treatment period could be explained by factors unrelated to VNS delivery. Consistent with our prior work, we saw that, in the current study, the relationship between animals’ weights and the number of pellets they received during the task trended toward statistical significance, with larger rats tending to earn more pellets than smaller rats (Fig. 4D; r = 0.3684, p = 0.0836, Pearson’s linear correlation). Daily fluctuations in weight also significantly impacted the number of pellets received: as animals’ weight increased from their average weight, they performed fewer trials and earned fewer reward pellets during task performance (Fig. 4E; All rats: Change in weight vs. Change in trials: r = − 0.2633, p = 0.0001; Change in weight vs. Change in hits: r = − 0.2233, p = 0.0006; Pearson’s linear correlation). These correlations were especially strong in the r-VNS group throughout the Pre-Treatment and Treatment periods (r-VNS group: Change in weight vs. Change in trials: r = − 0.4361, p = 0.0001; Change in weight vs. Change in hits: r = − 0.4397, p = 0.0000, Pearson’s linear correlation), however, the decline in pellets received was not accompanied by a significant increase in average weight during the Treatment period (r-VNS group average weight, Pre-Treatment: 260.6 ± 11.4, Treatment: 262.4 ± 12.1, p = 0.4905, paired t-test). Though we cannot rule out a direct effect of r-VNS on behavioral motivation, our analyses suggest that engagement in the lever press task is strongly modulated by the hunger/satiety state of the rats, and that r-VNS did not dramatically alter this relationship. Discussion Our results demonstrate that both r-VNS and l-VNS paired with motor training produced significant reorganization of motor cortical maps, suggesting that stimulation of either vagus nerve can engage the central neuromodulatory systems that promote task-relevant plasticity. L-VNS has been well established to enhance motor cortical plasticity through recruitment of cholinergic, serotonergic, and noradrenergic systems that facilitate experience-dependent remodeling7,9,10,14,16. Unlike l-VNS, r-VNS is thought to additionally engage the midbrain dopamine system17,20, which could have implications for improving motor rehabilitation or motivation during stroke recovery. The present study extends this prior work by showing that, like l-VNS, training-paired r-VNS drives significant task-relevant cortical reorganization in healthy rats. As the assay used here is highly predictive of VNS efficacy in models of stroke recovery, our results suggest that training-paired r-VNS is also likely to be effective at driving cortical plasticity and functional recovery. Learning-related improvements in motor skills and sensory discrimination are typically accompanied by reorganization of cortical maps, with expansion of task-relevant map regions correlating with gains in performance during early stages of training33–36. However, as tasks become well learned, these cortical representations tend to return to a macro-scale organization that resembles that of naïve animals, even as high levels of behavioral proficiency are preserved8,33,37. This dissociation suggests that while cortical map reorganization reflects plasticity processes that are important for the acquisition of new skills, the long-term maintenance of expert performance does not depend on sustained map expansion. Indeed, map reorganization induced by training-paired l-VNS does not enhance performance of a well-learned task in healthy rats8–11,13–15. In light of this published work, it is unsurprising that the r-VNS-induced M1 map plasticity observed in the current study was unaccompanied by significant improvements in motor performance. After stroke or other neural injuries, however, expansion of task-relevant cortical representations is again correlated with functional improvement7,38–42. Disrupting local circuit function both prevents M1 reorganization and impairs behavioral improvement, suggesting that map reorganization reflects cortical plasticity mechanisms that are critical for rehabilitation after injury7,40,41,43. Of particular translational relevance, the efficacy of l-VNS at inducing M1 map plasticity in healthy rats is highly correlated with the efficacy of the same parameters during VNS-paired stroke rehabilitation4,11. Thus, our finding that r-VNS drives the expansion of the task-relevant musculature representation in M1 suggests that training-paired r-VNS, like l-VNS, should enhance post-injury recovery. We previously reported that r-VNS is sufficient to reinforce lever-pressing and activate midbrain dopamine neurons in a self-administration assay17. We were thus surprised to find that r-VNS was not associated with increased task engagement in the current study, though several key differences in the training paradigms may contribute to these disparate results. Although animals in both studies were food deprived to increase motivation to perform the task, animals in the self-administration paradigm did not receive a food reward during the training-paired VNS sessions, whereas rats in the current study were rewarded with a food pellet after each successful VNS-paired lever press. Acute food deprivation increases incentive motivation and dopamine system responsivity across reward domains. Fasting elevates somatodendritic dopamine release in the VTA44 and enhances phasic dopamine firing evoked by food rewards45. In parallel, fasting-associated elevations in ghrelin increase dopamine neuron activity and synaptogenesis in the VTA and promote dopamine release in the nucleus accumbens46,47. These effects extend beyond food, as fasting increases incentive salience for non-food rewards such as drugs or electrical brain stimulation47,48. We hypothesize that food deprivation may similarly enhance the reward value of r-VNS in the self-administration task, leading animals to continue to lever-press throughout the session when food is withheld. Dopamine efflux in the striatum scales with the caloric density of intragastric infusates49,50, and is modulated by hunger and satiety state51. These findings suggest that in the presence of food rewards, dopamine functions, in part, to match behavioral motivation to caloric need. Consistent with this, satiety-associated peptides can also suppress motivation for non-food rewards. For example, leptin reduces heroin seeking even after food deprivation52, suggesting that signals of energy sufficiency can broadly suppress dopamine-mediated motivation. Thus, in the current study, the receipt of food rewards during r-VNS-paired training may have enhanced satiety signaling to drive an unexpected decrease in motivation throughout the session, unlike in the self-administration task. Dopamine activity also reflects cost-based decision-making in combination with motivational state53,54, such that increased effort requirements or satiety reduce dopamine release for the same reward55. The effort demands of the two tasks also differ, which may additionally reduce dopamine signaling and behavioral motivation to work for r-VNS in the current study. The self-administration task required only a small deflection (> 1.5 degrees) of an easy-to-reach lever, whereas in the present study, we aimed to ensure engagement of the proximal forelimb musculature during task performance by requiring the rats to reach outside of the training booth to fully depress (> 9.5 degrees) and release the lever. Under these conditions, if r-VNS enhances dopaminergic signaling, it may amplify the influence of satiety and effort-related costs on value. Our finding that r-VNS treatment was associated with an unexpected decrease in task engagement is consistent with the possibility that r-VNS may enhance the sensitivity of dopaminergic networks to metabolic need, satiety, and/or task demands, though additional work is needed confirm this result and investigate the mechanisms involved. In light of the prior literature, we hypothesize that the reward value of r-VNS itself may be strongly modulated by energy state and/or effort demands, which may limit the effectiveness of r-VNS as a tool to sustain increased behavioral motivation during motor training. Under canonical reward prediction error (RPE) frameworks, phasic dopamine responses by neurons in the ventral tegmental area (VTA) and substantia nigra pars compacta (SNc) signal the discrepancy between expected and received outcomes30–32,56,57. Early in training, RPEs are largest around the time of reward delivery, but as learning proceeds, these signals shift to the delivery of the earliest reward-predicting cues. The precise timing of RPEs is thought to support motor learning by reinforcing actions that lead to successful outcomes. More recent work highlights additional roles for phasic dopamine release, particularly in the prefrontal cortex, in cognitive flexibility and strategy selection58–60. While r-VNS can activate the dopaminergic neurons of the VTA and SNc17, and the VNS parameters used in the current study are thought to evoke phasic neuromodulator release61,62, the precise temporal dynamics of r-VNS-evoked dopamine release are unknown. Repeated stimulation may also elevate tonic dopamine levels within or across sessions by driving repeated release events that increase baseline extracellular dopamine63. It is possible that disruption of the temporal specificity and context dependence of endogenous dopamine signaling by r-VNS could contribute to reduced task engagement in our r-VNS group, perhaps through impaired estimation of reward or action values, altered action or strategy selection thresholds, or disruption of attentional processes. Future research is needed to elucidate the real-time effects of r-VNS on dopaminergic activity in regions of the reward system downstream of the VTA and SNc, including prefrontal cortical and basal ganglia circuits. An important limitation of the present study is the exclusive use of female rats. Dopamine system function in females is known to fluctuate across the 4–5 day estrous cycle due to ovarian hormone modulation. Dopamine neuron excitability and basal dopamine levels in the prefrontal cortex and basal ganglia are elevated during estrus and are lower and more comparable to males during diestrus64. The efficacy of VNS has also been seen to vary across the estrous cycle, with greater effects observed when VNS is delivered during diestrus65. In the current study, VNS treatment was delivered across 5 days of training, ensuring that stimulation was delivered during all phases of the estrus cycle to minimize potential bias and variability arising from hormone cycling. In prior studies using this paradigm, robust l-VNS-driven cortical plasticity8–11,13,14 and stroke recovery1,2,4,7 have been consistently observed in female rats, with effect sizes that are comparable to those seen in male rats13,15. In light of this prior work, our main finding that training-paired l-VNS and r-VNS both produce cortical map reorganization is highly likely to generalize to males. Nonetheless, sex-specific mechanisms could influence the extent to which r-VNS engages the midbrain dopamine system and should be considered in future studies. Clinically, the left cervical vagus nerve is preferentially targeted due to concerns that anatomical asymmetry may increase the likelihood of adverse cardiac side effects with r-VNS. These concerns arise from anatomical observations indicating that efferent fibers of the right vagus nerve preferentially innervate the sinoatrial (SA) node, which regulates heart rate, whereas fibers within the left vagus nerve preferentially project to the atrioventricular (AV) node, influencing contractility and conduction velocity21–23. In practice, however, these anatomical asymmetries have not translated into consistent differences in side-effect severity between l-VNS and r-VNS. Preclinical studies demonstrate that r-VNS reduces seizure severity as effectively as l-VNS24, and case studies in patients with refractory epilepsy have shown r-VNS to be clinically effective and well tolerated when left-sided implantation was not feasible, with no reports of significant stimulation-induced cardiac complications25–27. More recently, r-VNS has been specifically evaluated as a therapy for heart failure and has been found to be safe and well tolerated, though reported benefits on primary efficacy endpoints are mixed28,29. Collectively, these studies indicate that, while l-VNS remains the clinical standard, r-VNS can be safely administered and engages therapeutic mechanisms similar to those of left-sided stimulation. In the current study, we report additional preclinical findings that likewise suggest that VNS therapy paired with motor exercises effectively promotes training-dependent cortical plasticity whether delivered on the left or right side. Methods All procedures were conducted in accordance with the National Institutes of Health Guide for the Care and Use of Laboratory Animals and were approved by the University of Texas Institutional Animal Care and Use Committee (Protocol #2024 − 0126). Experiments were designed, conducted, and reported in compliance with ARRIVE guidelines. Subjects: To determine appropriate animal numbers for these experiments, we performed a priori power analyses (α = 0.017; β = 0.1; power = 0.9) based on behavioral effect sizes observed in our published data13, estimating that a minimum of 7 rats per group would be needed to observe statistically meaningful effects. A total of fifty-one female Long-Evans rats, aged 8–14 weeks at the start of the experiments, were included in the current study. Twenty-six subjects were excluded from the study due to health complications prior to (n = 3, not assigned to group) or during recovery from VNS surgery (n = 7; Sham: n = 1; l-VNS: n = 3; r-VNS: n = 3), failure to meet initial behavioral acquisition criteria within 6 months (pre-surgery drop-out: n = 5, not assigned to group) or to recover criterial behavior post-surgery (n = 3; Sham: n = 2; r-VNS: n = 1), headcap failure (n = 5; Sham: n = 1; l-VNS: n = 2; r-VNS: n = 2), or death during ICMS surgery (n = 3; Sham: n = 1; l-VNS: n = 1; r-VNS: n = 1). Post group assignment exclusions were distributed across groups without evidence of systematic imbalance, reducing the likelihood of selection bias in the final sample. Inclusion and exclusion criteria were also determined a priori. A total of twenty-three rats completed the study and were included in the data analyses. All rats were pair-housed before surgery and single housed after surgery in a 12:12 h reversed light cycle room (lights on: 6:00 pm). Both before and after surgery, animals received homecage enrichment consisting of paper nesting material (Bed-r’Nest) and a 1.25 in. wooden gnawing block. Rats were monitored during the surgical and behavioral recovery period for signs of poor health or potential isolation-related stress. For all subjects included in the study, health recovered normally within 1 week of surgery, and no overt signs of a prolonged increase in stress were observed after this period (e.g., weight loss, poor grooming, poor nesting, fear of handling, failure to participate in behavioral task). Rats received behavioral training during their active cycle. Animals were food-restricted during training; weights were monitored 5 days per week and maintained above 90% of their free-feeding body weight. Female rats were used because published work has shown that there are no sex differences in the cortical map plasticity evoked by training-paired VNS13, and because we wished to be able to compare the results of the current study to prior experiments that demonstrated significant right-left asymmetry in VNS-driven midbrain dopamine neuron activation in female rats17. As in prior studies8–15, VNS treatment was delivered across 5 days of training, to ensure that stimulation was delivered during all phases of the estrus cycle and minimize variability arising from hormone cycling. Behavioral training Prior to the start of behavioral training, all animals received 5 daily 10-minute sessions to acclimate to experimental handling. Rats then underwent training on a skilled reaching lever press task requiring them to depress and release a lever located 2 cm outside a MotoTrak training booth (Vulintus Inc, Lafayette, CO) with their right forelimb to receive a 45 mg food pellet reward (#F0021, BioServ, Flemington, NJ). Training on the task occurred in 2 stages. During Habituation (Stage 1), the lever was located 1 cm inside the training booth and rats were rewarded for making progressively larger lever depressions until the rats pressed to a full depression (> 9.5 degrees from horizontal) for 100 trials/day for at least 2 days. During Acquisition (Stage 2), the lever position was progressively moved back (0.5 cm increments) every 30 successful trials until the final position (2 cm outside the booth) was reached. Rats were rewarded for trials in which they fully depressed (deflection > 9.5 degrees) and released (deflection returns to < 4.75 degrees) the lever within a 2-second window. The reward was delivered immediately upon detection of each successful lever press-and-release. Once the final lever position was reached, Acquisition continued until criterial performance was achieved (at least 60 successful trials per session and at least 60% correct performance in 4 out of 5 consecutive training sessions). Animals were then randomized into groups using dynamic allocation and implanted with a VNS cuff electrode and headcap connector. After an approximately 1-week surgical recovery period, rats were returned to behavioral training until criterial performance was re-established. During this behavioral recovery period, rats were connected to the stimulator to acclimate to the VNS cables (no stimulation was delivered). After 5 days of criterial performance (Pre-Treatment), rats received 5 Treatment sessions in which l-VNS (n = 8), r-VNS (n = 8), or sham stimulation (n = 7) was delivered immediately following each correct lever press-and-release. Within 24 h of the final treatment-paired training session, animals underwent the terminal ICMS motor mapping procedure. For all behavioral sessions, lever deflection was sampled continuously at 100 Hz, and timestamps for each trial initiation (lever deflection > 1 degree) and successful press-and-release (Hit) were recorded automatically by MotoTrak software. Rats typically received 5 training sessions per week. After each daily training session, rats received ca. 7–9 g of rodent chow (Labdiet Prolab RMH 1800) in their homecage. After the last session of the week, rats were given ad libitum access to food for ca. 36 h; food was removed ca. 18–24 h prior to the first training session of each week. VNS cuff electrode and headcap implantation: Vagus nerve cuff electrodes were implanted as previously described13–15,17. Subjects were anesthetized with isoflurane (2–3%, inhaled) and a neck incision was made 1 mm to the left or right of the subject’s midline. Electrode leads were tunneled subcutaneously from a second incision over the skull. The vagus nerve was exposed by blunt dissection of the overlying neck musculature and carotid sheath, and the isolated nerve was placed inside the cuff electrode. A 10-second train of biphasic pulses (pulse amplitude: 0.8 mA; pulse frequency: 30 Hz; phase width: 100 µs; total pulse width: 200 µs) was delivered to evoke brief breathing cessation and validate cuff functionality. The headcap connector was then fixed to the skull via surgical screws and acrylic and the wounds closed with sutures. VNS parameters: VNS parameters were identical to those previously shown to drive M1 plasticity8–15 and stroke recovery1,2,4,7 in rats when paired with motor training, and to improve upper limb function in chronic stroke patients5,6,16. During treatment-paired training sessions, a 0.5-second train of biphasic pulses (pulse amplitude: 0.8 mA; pulse frequency: 30 Hz; phase width: 100 µs; total pulse width: 200 µs) was delivered following each successful lever press. The sham treatment group underwent identical handling, training, and surgical procedures, but did not receive stimulation during the final 5 days of training. ICMS motor mapping: Intracranial microstimulation (ICMS) motor mapping was performed as previously described13–15 Click or tap here to enter text. Animals were anesthetized via intraperitoneal injections of ketamine (75 mg/kg) and xylazine (5 mg/kg) and received supplemental doses of ketamine (25 mg/kg) as necessary to maintain areflexia. Animals were placed in a stereotaxic frame, and a craniotomy and durotomy were made over the left motor cortex (ca. +4.0 to − 3.0 mm anterior/posterior and ca. − 0.2 to − 5 mm lateral to bregma). A low impedance (< 1 MΩ) tungsten electrode was randomly placed at sites within a grid (0.5 mm spacing) over M1. At each site, the electrode was lowered 1.8 mm into the brain and high-frequency microstimulation was delivered using a 40 ms train of monophasic pulses (pulse frequency: 300 Hz, pulse width: 200 µs). Stimulus amplitude was increased from 10 to 200 µA until motor movement was first evoked (e.g., muscle twitch) or maximal amplitude was reached. If no movement was observed at 1.8 mm, the electrode was subsequently moved to 2.0 mm for stimulation at maximal amplitude, then repositioned to 1.6 mm, where stimulation was repeated. If no movement was observed at any of the three depths, the site was classified as “no response”. Threshold-evoked movement responses were classified into one of the following categories: proximal forelimb (PFL), distal forelimb (DFL), vibrissa, jaw, neck, trunk, hindlimb, or tail. The boundaries of the map were defined by contiguous “no response” sites. To reduce bias, mapping was performed by 2 experimenters. The first experimenter placed the electrodes. The second experimenter, blind to the treatment condition and electrode position, delivered stimulation and determined the primary movement and threshold stimulation amplitude. Data analysis All statistical analyses were performed in MATLAB. Anterior body representations (vibrissa, head, and jaw sites) were grouped together for analyses, as were posterior body representations (trunk, hindlimb, and tail sites). ICMS mapping results for each body representation were compared across treatment groups using one-way analysis of variance (ANOVA). Representative maps were determined by calculating the Euclidean distance between each map and the group mean across the four body areas of interest (PFL, DFL, anterior body, posterior body). Where ANOVA detected significant treatment-specific effects, these were followed by post-hoc Bonferroni-corrected t-tests to determine differences between groups. Effect sizes were calculated using Hedge’s g, a small-sample corrected version of Cohen’s d. The average weight, percent correct, number of trials, average trial duration, and average lever press speed were calculated for each animal by computing the mean across the five sessions immediately prior to stimulation (Pre-Treatment) and across the five sessions in which stimulation was delivered (Treatment). One-way ANOVA was used to compare weight and performance parameters across treatment groups during the Pre-Treatment and Treatment periods. To assess differences in Treatment-dependent effects on behavior, the percent change between Treatment and Pre-Treatment periods was also calculated for each parameter and compared across treatment groups using one-way ANOVA. We tested for within-group changes in each behavioral parameter between the Pre-Treatment and Treatment periods using paired t-tests. Effect sizes for paired comparisons were calculated using Cohen’s dz. For analysis of pellets rewarded across Pre-Treatment vs. Treatment sessions, pellet delivery was quantified in 1-minute bins during each session and averaged across sessions for each animal. These data were analyzed using a linear mixed-effects model with treatment group, treatment period, and session time as fixed effects and rat-specific random effects. Categorical predictors were coded using sum-to-zero (effects) coding such that coefficients reflect deviations from the grand mean. Time-resolved contrast analyses were performed to compare pre-treatment and treatment conditions at each time point within each group. These contrasts were derived from the fitted linear mixed-effects model as linear combinations of fixed effects evaluated across time. To account for multiple comparisons across time bins, p-values were corrected using false discovery rate (FDR). Prior to parametric testing, normality of all data distributions was confirmed using Shapiro–Wilk tests. For all statistical testing, results are reported as significant for p-values less than 0.05. All summary statistics are reported as mean ± standard error of the mean (SEM). Author contributions H.F.W. participated in conceptualization, investigation, formal analysis, visualization, data curation, and writing – original draft; M.R. participated in investigation and writing – review & editing; C.A.T. participated in conceptualization, formal analysis, data curation, writing – original draft, project administration, and funding acquisition. Funding This work was funded by NIH R01 NS126816. Data availability The raw data used in this study are available from the corresponding author upon request. Declarations Competing interests The authors declare no competing interests. Footnotes Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. References Associated Data Data Availability Statement The raw data used in this study are available from the corresponding author upon request.