Vagus nerve stimulation boosts the drive to work for rewards

The study investigates whether non-invasive stimulation of vagal afferents via transcutaneous auricular vagus nerve stimulation (taVNS) modulates human motivation to work for rewards. Grounded in economic and neurobiological theories, motivation reflects integration of benefits and costs. Preclinical work shows gut-derived vagal signals modulate dopaminergic circuits and reinforce behavior, implicating the nucleus tractus solitarii (NTS) and striatal systems in invigoration of action. However, causal evidence in humans, generalization beyond primary rewards (food) to secondary rewards (money), and potential lateralization of effects remain unclear. The authors hypothesize that taVNS increases invigoration (energization) of effort by boosting perceived benefits (potentially via dopamine and/or noradrenaline), with possible lateralization leading to food-specific effects for left-ear stimulation. They also test whether taVNS reduces effort costs affecting maintenance (potential serotonergic mechanism).

Prior animal and human evidence indicates vagal afferents regulate energy balance and reward processing via NTS projections to midbrain/striatal dopamine systems; nutrients in the gut evoke striatal dopamine release tracking caloric load. VNS can reinforce behavior and modulate learning/memory; chronic VNS reduces food intake and body weight, while acute taVNS reduces gastric myoelectric frequency. fMRI studies show taVNS engages NTS and dopaminergic regions. Dopamine is implicated in response vigor/invigoration; noradrenaline also facilitates energization. Serotonin has been linked to overcoming effort costs and cost-evidence accumulation. Lateralization of vagal effects has been reported in rodents (right nodose increasing dopamine more). Previous human taVNS work suggests effects on reinforcement learning, but effects on effort invigoration versus maintenance, generalization to money, and lateralization had not been tested directly.

Design: Randomized, single-blind, cross-over study comparing active taVNS vs sham within participants; stimulation applied to either left or right ear across participants (left n=41, right n=40). Participants: n=81 right-handed, healthy adults (47 women; mean age 25.3 ± 3.8 years; BMI 23.0 ± 2.95). Sessions: Two visits (∼2.5 h each) after overnight fast (≥8 h), between 7:00–10:15. Stimulation: NEMOS device; biphasic 25 Hz; default 30 s on/30 s off with off phases shortened to align with trial VAS periods; electrode at cymba conchae (taVNS) or earlobe (sham) on same ear; intensity titrated to mild pricking (≈5/10 VAS; taVNS mean 1.28 mA, sham 1.85 mA). Blinding check indicated chance-level guessing (53.4%). Task: Effort allocation task adapted from Meyniel et al.; participants repeatedly pressed a button with right index finger to keep a virtual ball above a threshold to accrue reward tokens. Manipulations: reward type (food vs money), reward magnitude (low 1 pt/s vs high 10 pt/s), difficulty (easy vs hard: threshold at 75% vs 85% of individual maximal press frequency). Each trial began with 1 s reward cue; participants could take within-trial breaks. After each effort phase, participants rated exertion and wanting (VAS). 48 trials per session. Tokens converted post-session (5 tokens = 1 kcal or 1 cent). Mean outcomes per session: 362.8 kcal and €3.78. Maximum button-press frequency estimated via practice and updated if exceeded. Primary outcomes: Invigoration (slope of ramp-up from rest to initial work plateau; %/s) and effort maintenance (average relative button-press frequency across trial, % of max). Work and rest segments were segmented algorithmically (MATLAB findpeaks). Statistical analysis: Mixed-effects (hierarchical linear) models predicted invigoration or maintenance with fixed effects: stimulation (taVNS vs sham), reward type, reward magnitude, difficulty, reward magnitude × difficulty, and interactions with stimulation; participant-level covariates: stimulation order and stimulation side; random intercepts and slopes. Associations of outcomes with wanting/exertion were also modeled. Bayesian t-tests computed Bayes factors from order-corrected OLS individual effects (default Cauchy prior r=0.707). Robust regression at group level assessed effort utility slope (invigoration vs wanting) with permutation testing (10,000 permutations). Computational modeling: Hierarchical Bayesian estimation (R/JAGS) of cost-evidence accumulation model (parameters: amplitude A, work accumulation slope SE, rest dissipation slope SR; A and SR modulated by reward magnitude, SE by difficulty). Additive taVNS effects estimated; evidence assessed via posterior credible intervals and Bayes factors. Procedures: Standardized session order; device alignment with trial onsets; concurrent tasks included food-cue reactivity and reinforcement learning (not primary here). Ethics: Approved by University of Tübingen IRB; written informed consent; compensation included fixed €32 or course credit plus task-contingent rewards.

• Task validation: Invigoration increased with reward magnitude (b=5.79, t(78)=4.69, p<0.001) and decreased with difficulty (b=−2.44, t(78)=−3.26, p=0.002). Invigoration associated with wanting (t(78)=6.14, p<0.001) but not exertion (t(78)=0.15, p=0.88); prior-trial exertion did not predict next-trial invigoration (b=0.017, p=0.414). Food vs money elicited comparable invigoration (b=−0.63, t(78)=−0.77, p=0.45).
• Effort maintenance increased with reward magnitude (b=9.18, t(78)=7.09, p<0.001), decreased with difficulty (b=−6.71, t(78)=−6.66, p<0.001), and showed reward magnitude × difficulty interaction (b=2.08, t(78)=3.88, p<0.001). Maintenance associated with both wanting and exertion (ts>8.08, ps<0.001). Food vs money comparable (b=−0.67, t(78)=−0.75, p=0.45). Prior-trial exertion negatively predicted next-trial maintenance (b=−0.062, p<0.001).
• taVNS effects on invigoration: Main effect increase during taVNS vs sham (b=2.93, 95% CI [0.98, 4.88], t(78)=2.943, p=0.004; BF10=7.34). Effect magnitude ≈5.30% of intercept (55.32), >50% of the 10× reward magnitude effect (10.45%). Stimulation × reward type interaction indicated stronger effects for food (b=1.33, t(78)=1.998, p=0.049). Stimulation side moderated this interaction (cross-level: b=−2.82, t(78)=−2.122, p=0.037); left-sided taVNS showed a food-specific effect (t(39)=3.172, p=0.003, BF10=11.80), right-sided taVNS did not (t(38)=−0.118, p=0.91, BF10=0.17). taVNS effects on invigoration remained significant controlling for maintenance (badj=1.99, p=0.025).
• taVNS effects on maintenance: Not significant (b=1.21, t(78)=1.715, p=0.090; BF10=0.51). No modulation by reward type (p=0.86) or stimulation side (ps>0.20). No effects on work segment duration (p=0.17).
• Subjective ratings: No taVNS effect on wanting (t(78)=0.488, p=0.63) or exertion (t(78)=0.704, p=0.48).
• Effort utility slope: Robust regression showed decreased slope (greater invigoration per unit wanting) under taVNS relative to sham; permutation tests significant overall and for right-sided taVNS; left-sided taVNS showed interaction with reward type (p=0.029), consistent with food-specific effect.
• Computational modeling: No evidence that taVNS altered cost-evidence accumulation parameters (posterior CIs included 0; BF10 between 0.002 and 0.050). Model recovered individual mean work/rest segment lengths well (R2=0.84).

Findings demonstrate that non-invasive taVNS acutely enhances the invigoration of effortful behavior in humans without increasing effort maintenance, aligning with a mechanism that boosts the perceived benefits or utility of acting rather than reducing costs. The pattern fits with dopaminergic theories of vigor, where increased tonic dopamine or increased expected reward rate elevates response invigoration, and is also compatible with noradrenergic contributions via locus coeruleus activation. Absence of effects on subjective wanting/exertion and on cost-evidence accumulation suggests taVNS modulates the drive to initiate work rather than sustained effort or perceived effort costs. Lateralization effects indicate left-sided taVNS preferentially augments invigoration for primary (food) rewards, echoing rodent findings of asymmetric vagal-dopamine pathways and supporting partial functional specificity between primary and secondary reinforcers. Clinically, enhancing invigoration may be relevant for motivational syndromes (apathy, anhedonia) and could complement serotonergic treatments that impact cost-related processes. The results underscore a role for interoceptive vagal signaling in tuning instrumental action according to metabolic needs.

Non-invasive taVNS increases invigoration to work for rewards and shifts the invigoration–wanting relationship to favor initiating effort, particularly for less-wanted rewards. Effects are stronger for food rewards with left-ear stimulation, indicating lateralized generalization across reinforcers. There is no conclusive effect on maintenance of effort or subjective wanting/exertion nor on cost-evidence accumulation. taVNS may offer a translational avenue to ameliorate deficits in vigor (e.g., apathy/anhedonia). Future work should directly test dopaminergic versus noradrenergic mechanisms (e.g., pharmacological challenges, PET imaging), delineate neural circuitry via concurrent neuroimaging, compare stimulation sides within subjects, examine other reinforcers and metabolic states, and optimize stimulation protocols for motivational outcomes and long-term clinical efficacy.

• Neurochemical mechanism not directly tested; taVNS may affect multiple neurotransmitter systems; dopamine/noradrenaline contributions remain to be dissociated (e.g., via pharmacology or PET).
• Exact mechanism for faster invigoration unresolved; no concurrent neuroimaging in the effort task.
• Lateralization not directly compared within-subject across sides; only between-subject side assignment.
• Reward space limited to food vs money; other reinforcers and depletion/metabolic states may modulate effects.
• Stimulation intensity/protocol fixed to mild pricking; different parameters may yield different effects.
• Only anecdotal Bayesian evidence for the null on maintenance; small effects cannot be excluded; additional data needed.