Cholinergic modulation of dopamine release drives effortful behaviour

Gavin C Touponse; Matthew B Pomrenze; Teema Yassine; Nicholas Denomme; May Wang; Viraj Mehta; Zihui Zhang; Robert C Malenka; Neir Eshel

doi:10.1038/s41586-025-10046-6

. 2026 Jan 28;651(8107):1020–1029. doi: 10.1038/s41586-025-10046-6

Cholinergic modulation of dopamine release drives effortful behaviour

Gavin C Touponse ^1,^#, Matthew B Pomrenze ^2,^#, Teema Yassine ¹, Nicholas Denomme ², May Wang ¹, Viraj Mehta ¹, Zihui Zhang ^1,², Robert C Malenka ², Neir Eshel ^1,^✉

PMCID: PMC13017521 NIHMSID: NIHMS2148277 PMID: 41606339

Abstract

Effort is costly: given a choice, we tend to avoid it¹. However, in many cases, effort adds value to the ensuing rewards². From ants³ to humans⁴, individuals prefer rewards that had been harder to achieve. This counterintuitive process may promote reward seeking even in resource-poor environments, thus enhancing evolutionary fitness⁵. Despite its ubiquity, the neural mechanisms supporting this behavioural effect are poorly understood. Here we show that effort amplifies the dopamine response to an otherwise identical reward, and this amplification depends on local modulation of dopamine axons by acetylcholine. High-effort rewards evoke rapid acetylcholine release from local interneurons in the nucleus accumbens. Acetylcholine then binds to nicotinic receptors on dopamine axon terminals to augment dopamine release when reward is delivered. Blocking the cholinergic modulation blunts dopamine release selectively in high-effort contexts, impairing effortful behaviour while leaving low-effort reward consumption intact. These results reconcile in vitro studies, which have long demonstrated that acetylcholine can trigger dopamine release directly through dopamine axons^6–11, with in vivo studies that failed to observe such modulation^12–14, but did not examine high-effort contexts. Our findings uncover a mechanism that drives effortful behaviour through context-dependent local interactions between acetylcholine and dopamine axons.

Subject terms: Motivation, Neural circuits, Reward

In the nucleus accumbens, acetylcholine boosts dopamine release to promote effortful behaviour.

Main

Reward delivery evokes a burst of dopamine (DA) release in the nucleus accumbens (NAc), helping to promote reward-seeking behaviours^15,16. The amplitude of this DA burst integrates multiple attributes of the reward, including both the size of the reward and how much effort went into it^17–19. Although DA release is largely driven by the firing of DA neurons in the midbrain, studies have proposed a key role for local modulation of DA axon terminals in the striatum as well^20–23. Which patterns of DA release are behaviourally relevant, what inputs determine these patterns and whether these inputs dissociate DA cell body activity from striatal DA release are all subject to debate^24–26.

Recently, we have found that DA release scales with preceding effort even if DA axon terminals in the NAc are stimulated directly via optogenetics¹⁷. When mice work for an identical optogenetic stimulation, more effort leads to more DA release. We reasoned that this variability in DA release might result from local modulation of DA axon terminals, a phenomenon that has been studied at length in vitro^20–23 but has proven more difficult to isolate in vivo. Of the many modulators with potential to calibrate DA release, acetylcholine (ACh) consistently emerges as a potent effector, although the nature of this interaction has been contested^6–14,27. In particular, there is ongoing debate over whether cholinergic interneurons in the striatum, which are capable of eliciting axonal DA release independent of DA cell body activity^6–11, actually do so in any behavioural context^12–14.

Reward-evoked DA release encodes expended effort

To discover how reward-evoked DA release is modulated by effort, we used a behavioural task¹⁷ that varies effort requirements for the same reward (Fig. 1a,b). Mice were first trained to nose poke during fixed ratio 1 (FR1) and FR5 schedules of reinforcement for sucrose reward. Once they achieved accurate and stable responding, mice proceeded to a task in which effort was varied through descending 10-min FR blocks. Mice performed the task well, with very few inactive pokes (Extended Data Fig. 1a) and consistent behaviour from day to day (Extended Data Fig. 1b,c). As the effort requirement increased, mice modulated their performance in the expected ways, initially escalating their poking rates to maintain a high level of reward consumption before reducing their poking rates and reward earnings (Extended Data Fig. 1d,e). Compared with low-effort rewards, rewards delivered after high effort were retrieved more quickly and consistently, suggesting high task engagement (Extended Data Fig. 1f,g).

Extended Data Fig. 1 — Mice readily perform the sucrose and ChRmine effort tasks. a. Number of active and inactive nosepokes in the sucrose effort task. Paired two-sided t-test, t = 9.98, df = 28, ****p < 0.0001, n = 32 sessions from 6 mice. b. Nosepoking consistency on the last two days of training. Spearman correlation, r = 1.00, **p = 0.0028, n = 6 (12 sessions from 6 mice). c. Reward earning consistency on the last two days of training. Spearman correlation, r = 0.85, *p = 0.044, n = 6 (12 sessions from 6 mice). d. Number of active and inactive nosepokes by FR in the sucrose effort task. Active: Friedman’s test, Friedman statistic = 53.23, ****p < 0.0001; n = 6 mice. Inactive: Friedman’s test, Friedman statistic = 6.26, p = 0.18; n = 6 mice. e. Number of rewards by FR in the sucrose effort task. Friedman’s test, Friedman statistic = 83.82, ****p < 0.0001; n = 6 mice. f. Magazine latency by FR in the sucrose effort task. Mixed effects model, F(4,1540) = 4.80, ***p = 0.0008; n = 6 mice. g. SEM of magazine latency by FR in the sucrose effort task. **h-l**. Similar to A-G except for the self-stimulation task. Statistics for h: Paired two-sided t-test, t = 6.73, df = 47, ****p < 0.0001, n = 48 sessions from 8 mice. Statistics for i: Spearman correlation, r = 0.88, **p = 0.0072, n = 8 (16 sessions from 8 mice). Statistics for j: Spearman correlation, r = 0.88, **p = 0.0072, n = 8 (16 sessions from 8 mice). Statistics for k: Active: Friedman’s test, Friedman statistic = 47.36, ****p < 0.0001; n = 8 mice. Inactive: Friedman’s test, Friedman statistic = 8.55, p = 0.073; n = 8. Statistics for l: Friedman’s test, Friedman statistic = 168.1, ****p < 0.0001; n = 8 mice. Data are mean ± s.e.m.

Source Data

As mice worked for sucrose rewards, we recorded DA release in the NAc with GRAB DA (Fig. 1a). We observed robust DA release time locked to reward delivery and consumption, but not during nose pokes (Fig. 1c). Consistent with our previous findings¹⁷, DA release at the time of reward delivery scaled with FR, such that high-effort sucrose rewards evoked more DA release (Fig. 1d–f). This increase in DA release at higher FRs was not due entirely to longer intervals between rewards, because we observed similar modulation regardless of inter-trial interval (ITI; Extended Data Fig. 2a–h).

Extended Data Fig. 2 — a. GRAB DA activity in the sucrose effort task for trials separated by an inter-trial interval (ITI) of >15 s. b. Area under the curve (0-4 s) for each FR for trials separated by ITI of >15 s. Mixed-effects model, main effect of FR, F(4,331) = 6.48, ****p < 0.0001, n = 6 mice. c. GRAB DA activity in the sucrose effort task for trials separated by an ITI of >30 s. d. Area under the curve (0-4 s) for each FR for trials separated by ITI of >30 s. Mixed-effects model, main effect of FR, F(4,321) = 6.25, ****p < 0.0001, n = 6 mice. e. GRAB DA activity in the sucrose effort task for trials separated by an ITI of >45 s. f. Area under the curve (0-4 s) for each FR for trials separated by ITI of >45 s. Mixed-effects model, main effect of FR, F(4,305) = 4.82, ***p = 0.0009, n = 6 mice. g. GRAB DA activity in the sucrose effort task for trials separated by an ITI of >60 s. h. Area under the curve (0-4 s) for each FR for trials separated by ITI of >60 s. Mixed-effects model, main effect of FR, F(4,238) = 6.48, **p = 0.010, n = 6 mice. **i-p**. Similar to **a-h** except for self-stimulation task. Statistics for j: Mixed-effects model, main effect of FR, F(4,802) = 60.36, ****p < 0.0001, n = 8 DAT-Cre mice. Statistics for l: Mixed-effects model, main effect of FR, F(4,388) = 47.22, ****p < 0.0001, n = 8 DAT-Cre mice. Statistics for n: Mixed-effects model, main effect of FR, F(4,231) = 18.85, ****p < 0.0001, n = 8 DAT-Cre mice. Statistics for p: Mixed-effects model, main effect of FR, F(4,153) = 14.01, ****p < 0.0001, n = 8 DAT-Cre mice. q. Schematic of the homecage ChRmine stimulation control. r. GRAB DA activity in the homecage stimulation control split by 10-minute blocks. s. Area under the curve (0-6 s) for each block. Friedman’s test, Friedman statistic = 4.20, p = 0.41; n = 4 DAT-Cre mice. t. GRAB DA activity in the homecage control split by first versus second-half stimulations. u. Area under the curve (0-6 s) for early versus late stimulations. Wilcoxon matched-pairs two-sided signed rank test, sum of signed ranks = −10.00, p = 0.13; n = 4 DAT-Cre mice. Data are mean ± s.e.m.

Source Data

To more precisely determine what aspects of the task were encoded by DA, we trained a general linear model on various behaviours and task characteristics (kernels; Fig. 1g). The general linear model revealed a major contribution of reward delivery, but not nose pokes or magazine entries, to DA release (Fig. 1h). The model accurately predicted DA dynamics at both low and high efforts (Fig. 1i), with FR being a stronger predictor of DA release than ITI (Fig. 1j,k). We conclude that increased effort augments DA release for sucrose reward, recognizing that ‘effort’ includes not only physical exertion but also action repetition and behavioural persistence despite temporal delays.

To determine whether the effort encoding that we observed for sucrose reward generalized to other types of reinforcers, we used the same task but with optogenetic stimulation of DA axons in the NAc substituted for sucrose. DAT–Cre mice were prepared with Cre-dependent red-shifted ChRmine (rsChRmine) in the ventral tegmental area (VTA) and GRAB DA and an optical fibre in the NAc (Fig. 1l). These mice were subjected to the same task structure as for sucrose reward, except now for brain stimulation reward (Fig. 1m). Mice performed accurately (Extended Data Fig. 1h) and consistently (Extended Data Fig. 1i,j), with similar effort dependence as in the sucrose task (Extended Data Fig. 1k,l). When mice nose poked for optical stimulation of DA release (5 s of 625-nm light at 20 Hz, 6 mW), DA release was again time locked to reward delivery (Fig. 1n). Our previous work¹⁷ demonstrated that mice will exert effort for a spectrum of optical stimulation parameters, ranging from 1 s to 10 s. We chose 5 s here because it evokes reliable and robust behavioural responding, but note that this stimulation does not perfectly mimic DA responses to sucrose reward.

As with sucrose, we observed that DA release scaled with effort, such that the same optogenetic stimulation led to more DA at higher FRs (Fig. 1o–q). This effect was not due to the time elapsed between rewards (Extended Data Fig. 2i–p). In addition, changes in DA release were not due to opsin desensitization or ‘run down’ of optically evoked release over time, as 50 min of regular stimulations (independent of mouse behaviour) showed similar DA release across the entire session (Extended Data Fig. 2q–u). Just as for the sucrose data, we trained a general linear model on the photometry results (Fig. 1r) and identified the same patterns: the reward delivery kernel was the major contributor to model performance (Fig. 1s), the model accurately predicted release dynamics across FRs (Fig. 1t) and FR was a stronger predictor of DA release than ITI (Fig. 1u,v). Collectively, these data replicate and extend our previous work¹⁷ and support the hypothesis that DA release encodes expended effort.

Effort encoding by NAc DA is cell body-independent

We next sought to characterize the mechanism underlying the effect of effort on DA release. Modulation of DA release could involve alterations at the level of cell bodies in the VTA, axon terminals in the NAc or a combination of the two²². Indeed, recent work has suggested that cell body and terminal activity can be dissociated in reward contexts²⁸, although this theory remains contested²⁹. Consistent with DA neuron recordings in monkeys¹⁸, we hypothesized that as mice exerted effort for reward, DA cell bodies would become more excitable, leading to synchronized activity and augmented DA release at the time of reward delivery. It was unclear, however, whether cell body activity would fully mimic the effort encoding that we observed in NAc DA release. To find out, we conducted simultaneous recordings of DA cell bodies, their axons in the NAc and NAc DA release, all while mice worked for sucrose reward (Fig. 2a). In all three recordings, we detected enhanced reward-evoked activity at higher FR schedules (Fig. 2b–f). However, the enhancement of DA cell body Ca²⁺ activity (Fig. 2c) appeared to asymptote at an earlier FR than either axon terminal Ca²⁺ activity (Fig. 2e) or NAc DA release (Fig. 2f) recorded in the same mice at the same time. This plateau was visible in two independent cohorts of mice (Extended Data Fig. 3a–d) and was not due to GCaMP sensor saturation in the VTA DA cell bodies because social stimuli were able to evoke substantially higher cell body activity in the same mice (Extended Data Fig. 3e–j).

Fig. 2 — a, Schematic of the triple recordings of VTA DA cell bodies, DA axons and DA release in the NAc during the effort task for sucrose reward. Schematic was adapted with permission from ref. ⁶³, Elsevier. b, Representative image of VTA DA cells expressing GCaMP8m and fibre above. Scale bar, 100 µm. c, Average VTA DA cell body GCaMP activity aligned to reward delivery (left; n = 6 DAT–Cre mice), and AUC (0–4 s) for each FR (right). Friedman test, Friedman statistic = 15.07, **P = 0.0046, n = 6 DAT–Cre mice. d, Representative image of VTA DA axons expressing GCaMP8m and GRAB rDA (red DA) in the NAc. Scale bar, 100 µm. e, VTA DA axon GCaMP activity in the NAc aligned to reward delivery (left), and AUC (0–4 s) for each FR (right). Mixed-effects model, F(4,19) = 28.1, ****P < 0.0001; n = 6 DAT–Cre mice. f, DA release aligned to reward delivery (left), and AUC (0–4 s) for each FR (right). Mixed-effects model, F(4,19) = 22.93, ****P < 0.0001; n = 6 DAT–Cre mice. g, Schematic of the AAV injection and cannula placement for inhibition of VTA DA cell bodies during the self-stimulation task (left). Images of GRAB DA and rsChRmine expression in the NAc and VTA (top right). The experimental schedule (bottom right). Scale bars, 100 µm. IPN, interpeduncular nucleus. Schematic was adapted with permission from ref. ⁶³, Elsevier. h, DA release dynamics after microinjection of PBS into the VTA (n = 8 DAT–Cre mice). i, DA release dynamics after microinjection of muscimol into the VTA (n = 8 DAT–Cre mice). j, Peak DA (0–6 s) response for each FR for PBS and muscimol sessions. Mixed-effects model, fixed effect of FR: F(4,28) = 45.26, ****P < 0.0001; fixed effect of drug: F(1,7) = 0.24, P = 64; interaction of FR and drug: F(4,28) = 0.20, P = 0.93. Sidak-corrected multiple comparisons: P = 0.99 for FR1 PBS versus muscimol; P = 0.97 for FR5 PBS versus muscimol; P > 0.99 for FR10 PBS versus muscimol; P > 0.99 for FR21 PBS versus muscimol; and P = 0.99 for FR46 PBS versus muscimol. n = 8 DAT–Cre mice. k, Schematic of optogenetic inhibition of VTA DA cell bodies during the self-stimulation task. Schematic was adapted with permission from ref. ⁶³, Elsevier. l, DA release dynamics during control trials with no light in the VTA (n = 6 DAT–Cre mice). m, DA release dynamics during trials with blue light in the VTA to inhibit DA cell bodies (n = 6 DAT–Cre mice). n, Peak DA (0–6 s) response for each FR for trials with and without blue light in the VTA. Mixed-effects model, fixed effect of FR: F(4,37) = 26.10, ****P < 0.0001; fixed effect of light in the VTA: F(1,10) = 0.05, P = 0.82; and interaction of FR and light in the VTA: F(4,37) = 0.66, P = 0.62. Sidak-corrected multiple comparisons: P > 0.99 for FR1 no light versus blue light; P > 0.99 for FR5 no light versus blue light; P > 0.99 for FR10 no light versus blue light; P > 0.99 for FR21 no light versus blue light; and P = 0.77 for FR46 no light versus blue light. n = 6 DAT–Cre mice. Data are mean ± s.e.m.

Source Data

Extended Data Fig. 3 — a. Schematic of AAV injection and fiber implantation for VTA DA cell body recordings. The schematic was adapted from ref. ⁶³, Elsevier. b. Cohort 1: DA release area under the curve for each FR. Friedman test, Friedman statistic = 7.04, p = 0.134, n = 5 DAT-Cre mice. c. Cohort 2: DA release area under the curve for each FR (same data presented in Fig. 2c). Friedman test, Friedman statistic = 15.07, **p = 0.0046, n = 6 DAT-Cre mice. d. Both cohorts pooled: DA release area under the curve for each FR. Friedman test, Friedman statistic = 20.58, ***p = 0.0004, n = 11 DAT-Cre mice. e. Schematic of sucrose effort task. f. Average VTA DA cell body GCaMP activity at FR21 and FR46 aligned to reward delivery (same data presented in Fig. 2c). g. Schematic of recording during social interaction. h. Average VTA DA cell body GCaMP activity in response to social contact. Note these are the same mice tested in the effort task presented. i. Overlay of average VTA DA cell body GCaMP activity in response to FR21 and FR46 rewards and social contact. j. Left, DA release area under the curve for high FRs and social contact. Mixed effects model, F(2,10) = 5.637, *p < 0.022; n = 6. Right, DA release peak for high FRs and social contact. Mixed effects model, F(2,10) = 4.47, *p < 0.041; n = 6 DAT-Cre mice. Data are presented as mean.

Source Data

These data suggest that modifications occurring in DA cell bodies may only partially mediate the effort-based modulation of DA release in the NAc, as the dynamic range achieved in the NAc appears to be higher than that in the cell bodies. To directly examine the necessity of DA cell body activity for the changes that we observed in the NAc, we suppressed VTA DA cell body activity while recording NAc DA release as mice performed our task. In this case, mice worked for rsChRmine stimulation of DA axon terminals, because DA cell body inactivation caused motor impairments that precluded performance of the sucrose task. DAT–Cre mice were injected with GRAB DA in the NAc and DIO-rsChRmine in the VTA, and implanted with a fibre in the NAc and a unilateral guide cannula in the VTA for microinfusions of the GABA_A agonist muscimol to silence VTA neural activity (Fig. 2g). After training, mice were infused with muscimol or PBS into the VTA 5 min before an effort task session (Fig. 2g). Although unilateral muscimol successfully elicited the expected locomotor effects (Extended Data Fig. 4a–d) and strongly suppressed both spontaneous and reward-evoked VTA DA neuron activity in control recordings (Extended Data Fig. 4e–j), the drug had no effect on DA effort encoding in the NAc (Fig. 2h–j).

Extended Data Fig. 4 — a. Left, number of contralateral rotations mice made in the first 5 min of the ChRmine effort task when either PBS or muscimol was microinjected into the VTA. Wilcoxon matched-pairs two-sided signed rank test, sum of signed ranks = 28.00, *p = 0.016, n = 8 DAT-Cre mice. Right, number of ipsilateral rotations mice made in the same time. Wilcoxon matched-pairs two-sided signed rank test, sum of signed ranks = 2.00, p > 0.99, n = 8 DAT-Cre mice. b. Rewards earned for each FR after VTA microinjection with PBS. Friedman test, Friedman statistic = 31.12, ****p < 0.0001; n = 8 DAT-Cre mice. c. Rewards earned for each FR after VTA microinjection with muscimol. Friedman test, Friedman statistic = 25.83, ****p < 0.0001; n = 8 DAT-Cre mice. d. Rewards earned for each FR with pair-wise comparisons between PBS and muscimol. Mixed-effects model, fixed effect of FR, F(4,28) = 19.82, ****p < 0.0001; fixed effect of drug, F(1,7) = 1.86, p = 0.21; interaction of FR and drug; F(4,28) = 2.46; p = 0.068. Sidak-corrected multiple comparisons: FR1 PBS vs muscimol, *p = 0.037; FR5 PBS vs muscimol, p = 0.55; FR10 PBS vs muscimol, p > 0.99; FR21 PBS vs muscimol, p > 0.99; FR46 PBS vs muscimol, p > 0.99; n = 8 DAT-Cre mice. e. Left, schematic of recording and drug infusion configuration in the VTA of DAT-Cre mice. Right, schematic of setup for drug infusions during a head-fixed recording. The schematic was adapted from ref. ⁶³, Elsevier. f. Raw GCaMP8m traces from head-fixed recordings during PBS or muscimol infusions. F = raw fluorescent units. g. Expanded recording traces from the insert outlined in panel f. h. Top, behavioral setup of VTA DA cell body recordings during a Pavlovian rewards task. Bottom, experimental time course. i. Left, example traces with identified calcium transients from baseline GCaMP recordings after PBS (top) or muscimol (bottom) infusions into the VTA. Right, frequency of spontaneous calcium transients after PBS or muscimol infusions. Wilcoxon matched-pairs two-sided signed rank test, sum of signed ranks = −28.0, *p = 0.016; n = 7 DAT-Cre mice. j. Left, average VTA DA cell-body GCaMP response to reward cues. Right, area under the curve (0-4 s) of reward cue-evoked activity. Wilcoxon matched-pairs two-sided signed rank test, sum of signed ranks = −28.0, *p = 0.016; n = 7 DAT-Cre mice. Data are mean ± s.e.m.

Source Data

Because muscimol inhibited the entire VTA, not only the DA cell bodies, we next took an optogenetic approach to selectively silence DA cell bodies. In DAT–Cre mice, we expressed GRAB DA in the NAc and infused a cocktail of Cre-dependent rsChRmine and the soma-targeted blue-light-sensitive anion channel GtACR2 (stGtACR2)³⁰ in the VTA (Fig. 2k). We then implanted a fibre over the NAc for GRAB DA recordings and rsChRmine stimulation, and another fibre over the VTA for GtACR2-mediated inhibition of DA cell bodies. During the effort task, 50% of red-light rewards in the NAc were randomly paired with simultaneous blue-light exposure (5 s of 465-nm continuous light, 6 mW) in the VTA (Fig. 2k). Similar to muscimol, optical silencing of VTA DA cell body activity had minimal effect on effort encoding in the NAc (Fig. 2l–n). This null effect could not be explained by non-functional opsins, because in the same mice, we found that red light alone triggered robust DA release (Extended Data Fig. 5a–d), whereas blue light alone potently inhibited DA release in the NAc (Extended Data Fig. 5e–g; note that technical limitations precluded the measurement of DA cell body activity during stGtACR2 stimulation). Although we cannot prove that this inhibition fully blocked all effort-related signals from DA cell bodies to the NAc, the results are consistent with the muscimol data and suggest that although effort encoding can be detected in VTA DA cell body activity, this activity is not necessary for the scaling of NAc DA release by effort.

Extended Data Fig. 5 — a. Schematic of AAV injection, fiber implantation, and experimental design for the validation of both ChRmine stimulation and stGtACR2 inhibition. The schematic was adapted from ref. ⁶³, Elsevier. b. Left, representative trace of GRAB DA activity with ChRmine stimulation. Right, heatmap of GRAB DA activity with ChRmine stimulation. c. GRAB DA signal during ChRmine stimulation averaged across all mice (n = 6). d. Area under the curve of GRAB DA signal before (−5-0 s) and during (0-5 s) ChRmine stimulation. Wilcoxon matched-pairs two-sided signed rank test, sum of signed ranks = 21.00, *p = 0.031, n = 6 DAT-Cre mice. e. Left, representative trace of GRAB DA activity with stGtACR2 inhibition. Right, heatmap of GRAB DA activity with stGtACR2 inhibition. f. GRAB DA signal during stGtACR2 inhibition averaged across all mice (n = 6). g. Area under the curve of GRAB DA signal before (−10-0 s) and during (0-10 s) stGtACR2 inhibition. Wilcoxon matched-pairs two-sided signed rank test, sum of signed ranks = −21.00, *p = 0.031, n = 6 DAT-Cre mice. Data are mean ± s.e.m.

Source Data

Nicotinic signalling controls effort-related DA release

As VTA DA cell body activity appears dispensable for effort encoding in the NAc, we reasoned that physiological changes in DA axon terminals might support this phenomenon instead. As DA terminals in the striatum are sensitive to various neuromodulators²², we turned to pharmacology to test the role of different modulators and GPCRs in DA effort encoding. We prepared new DAT–Cre mice with rsChRmine in the VTA, GRAB DA in the NAc and an optical fibre-coupled cannula implant in the NAc for drug microinfusions into the recording site (Fig. 3a,b and Extended Data Fig. 6a). We tested whether nine different antagonists could disrupt effort encoding of optogenetically evoked DA release. Mice were trained in the task, habituated to microinjection procedures, and then tested for effort encoding 5 min after microinfusions (Extended Data Fig. 6b). Intra-NAc blockade of α1 adrenergic, muscarinic (pan and M5 specific) or neurokinin 3 receptors failed to affect DA effort encoding, as did systemic blockade of adenosine A₁, adenosine A_2A, neurokinin 1 or neurotensin 1 receptors (Extended Data Fig. 6c–l). By contrast, the nicotinic receptor antagonist dihydro-β-erythroidine (DHβE), which targets α4-containing and α6-containing nicotinic receptors³¹ (Extended Data Fig. 7), entirely blocked the enhanced DA release in high-FR blocks (Fig. 3c–f). DHβE did not abolish light-evoked DA release altogether; rather, it prevented the increased release during high-effort rewards (Fig. 3e,f). Furthermore, DHβE was similarly effective at blunting DA effort encoding when mice worked for sucrose instead of optogenetic reward (Fig. 3g–l). Together, these data reveal ACh signalling at α4-containing and α6-containing nicotinic receptors as a key and selective component of effort encoding by DA.

Fig. 3 — a, Joint optic fibre–cannula implants were used to locally infuse drugs before the self-stimulation task. Schematics was adapted with permission from ref. ⁶³ Elsevier. b, Representative images of viral expression and fibre placement in the NAc and VTA. Scale bars, 100 µm. c, DA release across FR blocks during the self-stimulation task after microinjection of PBS into the NAc (n = 6 DAT–Cre mice). d, DA release across FR blocks during the self-stimulation task after microinjection of DHβE into the NAc (n = 6 DAT–Cre mice). e, AUC (0–6 s) for each FR for PBS and DHβE sessions. For the mixed-effects model, fixed effect of FR: F(4,20) = 20.63, ****P < 0.0001; fixed effect of drug: F(1,5) = 37.91, **P = 0.0016; and interaction of FR and drug: F(4,16) = 24.43, ****P < 0.0001. Sidak-corrected multiple comparisons: P > 0.99 for FR1 PBS versus DHβE; P = 0.78 for FR5 PBS versus DHβE; P = 0.25 for FR10 PBS versus DHβE; ****P < 0.0001 for FR21 PBS versus DHβE; and ****P < 0.0001 for FR46 PBS versus DHβE. n = 6 DAT–Cre mice. f, Peak DA response (0–6 s) for each FR for PBS and DHβE sessions. For the mixed-effects model, fixed effect of FR: F(4,20) = 39.33, ****P < 0.0001; fixed effect of drug: F(1,5) = 52.20, ***P = 0.0008; and interaction of FR and drug: F(4,16) = 17.68; ****P < 0.0001. Sidak-corrected multiple comparisons: P > 0.99 for FR1 PBS versus DHβE; P = 0.77 for FR5 PBS versus DHβE; P = 0.36 for FR10 PBS versus DHβE; ****P < 0.0001 for FR21 PBS versus DHβE; and ****P < 0.0001 for FR46 PBS versus DHβE. n = 6 DAT–Cre mice. g–l, Similar to a–f, except for the sucrose task. Schematic in g was adapted with permission from ref. ⁶³, Elsevier. Statistics for panel k: for the mixed-effects model, fixed effect of FR: F(4,20) = 6.54, **P = 0.0016; fixed effect of drug: F(1,5) = 2.85, P = 0.15; and interaction of FR and drug: F(4,14) = 3.83; *P = 0.027. Sidak-corrected multiple comparisons: P = 0.94 for FR1 PBS versus DHβE; P > 0.99 for FR5 PBS versus DHβE; P = 0.98 for FR10 PBS versus DHβE; P = 0.062 for FR21 PBS versus DHβE; and P = 0.11 for FR46 PBS versus DHβE. n = 6. Statistics for panel l: for the mixed-effects model, fixed effect of FR: F(4,20) = 11.75, ****P < 0.0001; fixed effect of drug: F(1,5) = 14.71, *P = 0.012; and interaction of FR and drug: F(4,14) = 6.56, **P = 0.0034. Sidak-corrected multiple comparisons: P > 0.99 for FR1 PBS versus DHβE; P = 0.75 for FR5 PBS versus DHβE; P = 0.80 for FR10 PBS versus DHβE; **P = 0.0058 for FR21 PBS versus DHβE; and **P = 0.0035 for FR46 PBS versus DHβE. n = 6 DAT–Cre mice. m, Schematic of AAV injection and electrophysiological recording configuration. Schematic was adapted with permission from ref. ⁶³, Elsevier. n, Representative brain slice image. Scale bar, 50 µm. o, Representative axonal action potential traces from an eYFP⁺ VTA DA axon bleb evoked by rsChRmine stimulation of cholinergic interneurons at baseline and in the presence of 1 µM DHβE. p, Time course of average axonal action potentials (axAPs) evoked by rsChRmine stimulation before and during acute perfusion of 1 µM DHβE. n = 3 axons, n = 3 mice. Dashed line, average baseline amplitude. q, Average opto-evoked axAP (o-axAP) amplitude over 10 sweeps in baseline and 1 µM DHβE conditions. Significance was determined using a paired two-sided t-test, t = 12.54, d.f. = 2, **P = 0.0063, n = 3 axons, n = 3 mice. Data are mean ± s.e.m.

Source Data

Extended Data Fig. 6 — a. Top, schematic of AAV injection and fiber-cannula implant for the self-stimulation task with injection of receptor antagonist. Bottom, representative images of NAc fiber placement and viral expression of ChRmine. The schematic was adapted from ref. ⁶³, Elsevier. b. Schematic of effort task and experimental schedules.c-g. Average GRAB DA activity, area under the curve (AUC, 0-6 s), and peak DA release for each FR during the self-stimulation task with microinjection of PBS (c, AUC: Friedman’s test, Friedman statistic = 16.00, ****p < 0.0001; n = 4 DAT-Cre mice; Peak: Friedman’s test, Friedman statistic = 16.00, ****p < 0.0001; n = 4 DAT-Cre mice), terazosin (d, AUC: Mixed effects model, F(4,11) = 21.96, ****p < 0.0001; n = 4 DAT-Cre mice; Peak: Mixed effects model, F(4,11) = 34.15, ****p < 0.0001; n = 4 DAT-Cre mice), scopolamine (e, AUC: Friedman’s test, Friedman statistic = 16.00, ****p < 0.0001; n = 4 DAT-Cre mice; peak: Friedman’s test, Friedman statistic = 16.00, ****p < 0.0001; n = 4 DAT-Cre mice), ML375 (f, AUC: Mixed effects model, F(4,8) = 34.06, ****p < 0.0001; n = 4 DAT-Cre mice; peak: One-way ANOVA, F(4,8) = 30.32, ****p < 0.0001; n = 4 DAT-Cre mice), and osanetant (g, AUC: Mixed effects model, F(4,7) = 5.28, *p = 0.028; n = 4 DAT-Cre mice; peak: Mixed effects model, F(4,7) = 7.18, *p = 0.013; n = 4 DAT-Cre mice). **h-l**. Average GRAB DA activity, AUC (0-6 s), and peak DA release for each FR during the self-stimulation task with systemic injection of vehicle corn oil/saline (h, AUC: Mixed effects model, F(4,11) = 12.92, ***p = 0.0004; n = 4 DAT-Cre mice; peak: Mixed effects model, F(4,11) = 18.29, ****p < 0.0001; n = 4 DAT-Cre mice), DPCPX (i, AUC: Mixed effects model, F(4,10) = 16.03, ***p = 0.0002; n = 4 DAT-Cre mice; peak: Mixed effects model, F(4,10) = 24.40, ****p < 0.0001; n = 4 DAT-Cre mice), istradefylline (j, AUC: One-way ANOVA, F(4,12) = 10.01, ***p = 0.0008; n = 4 DAT-Cre mice; peak: One-way ANOVA, F(4,12) = 15.20, ***p = 0.0001; n = 4 DAT-Cre mice), aprepitant (k, AUC: Mixed effects model, F(4,10) = 14.16, ***p = 0.0004; n = 4 DAT-Cre mice; peak: Mixed effects model, F(4,10) = 16.52, ***p = 0.0002; n = 4 DAT-Cre mice), and SR 48692 (l, AUC: Mixed effects model, F(3,9) = 13.21, **p = 0.0012; n = 4 DAT-Cre mice; peak: Mixed effects model, F(3,9) = 20.79, ***p = 0.0002; n = 4 DAT-Cre mice). Data are presented as mean.

Source Data

Extended Data Fig. 7 — a. Nicotinic α subunit gene expression in major cell-types of the VTA. Note enriched expression of α4 (*Chrna4*) and α6 (*Chrna6*) in VTA DA neurons. b. α subunit gene expression in major cell-types of the NAc. c. α subunit gene expression in layer-specific cell-types of the prelimbic-infralimbic-orbitofrontal region (PL-ILA-ORB). Note that Layer 2/3 IT and Layer 5 IT cells are predicted to project to the NAc. d. α subunit gene expression in layer-specific cell-types of the anterior cingulate cortex (ACC). Note that Layer 5 IT cells are predicted to project to the NAc. e. α subunit gene expression in relevant cell-types of the regions (CA1-ProS) where cells are predicted to project to the NAc. f. α subunit gene expression in relevant cell-types of the thalamus (TH). Note that analysis was restricted to the paraventricular region (PVT) where cells are predicted to project to the NAc. g. α subunit gene expression in relevant cell-types of the cortical subplate (CTXsp). Note that analysis was restricted to the basolateral amygdala (BLA) where cells are predicted to project to the NAc. h. Left, gene expression (log2(CPM + 1)). Middle, percentage of cells expressing gene of interest. Right, green boxes around cell-types predicted to project to the NAc. All data analysis performed on publicly available 10x scRNAseq data sets⁷⁵.

Interactions between DA and ACh are well established^32–36, with substantial evidence in vitro that ACh can trigger DA release through nicotinic receptors in both the dorsal and ventral striatum^6–11. However, direct nicotinic-dependent excitation of DA axons has only been shown in the dorsal striatum^7,11. To determine whether these dorsal striatal findings generalize to the NAc, we crossed DAT–Flp mice to ChAT–Cre mice and injected Flp-dependent eYFP into the VTA and Cre-dependent rsChRmine into the NAc. Acute slices were prepared, and perforated patch recordings of DA axons within the NAc were performed while stimulating cholinergic interneurons with red light (Fig. 3m,n). We detected robust excitation of DA axons in response to optogenetic stimulation of local cholinergic interneurons, including excitatory postsynaptic potentials, single action potentials, compound action potential–excitatory postsynaptic potentials and double action potentials (Fig. 3o and Extended Data Fig. 8) that were blocked by 1 µM DHβE (Fig. 3p,q and Extended Data Fig. 8). These data confirm that cholinergic interneurons can excite DA axons locally in the NAc.

Extended Data Fig. 8 — a. Schematic of AAV injection and electrophysiological recording configuration. The schematic was adapted from ref. ⁶³, Elsevier. b. Representative sweeps from a single recording of an eYFP+ VTA DA axon bleb in response to rsChRmine stimulation of cholinergic interneurons at baseline and in the presence of 1 µM DHβE. c. Time course of average axonal action potential (axAP) amplitude evoked by rsChRmine stimulation before and during acute perfusion of 1 µM DHβE; n = 3 axons, N = 3 mice. d. Representative traces showing VTA DA axon responses to rsChRmine stimulation of cholinergic interneurons, including axEPSPs, axAPs, compound axAP-EPSPs, and axAP bursts. e. Representative DA axon bleb recording showing a relationship between the irradiance of cholinergic rsChRmine stimulation and axonal response amplitude. Data are mean ± s.e.m.

Source Data

NAc ACh release encodes expended effort

If ACh gates DA release to encode effort, then ACh release dynamics should also reflect exerted effort. To measure these dynamics, we prepared new wild-type mice with GRAB ACh in the NAc and ran them through the same behavioural procedures for sucrose reward (Fig. 4a,b). Consistent with previous work^7,37–39, we observed complex ACh release waveforms during different phases of the task (Fig. 4c). Focusing on the response around reward delivery, we observed a subtle ramping of activity directly before reward, followed by a triphasic ACh waveform with a sharp peak right after delivery, a rapid dip and a second peak around the time of reward consumption (Fig. 4c). All of these dynamics were robustly modulated by FR (Fig. 4d–f), even after controlling for the effect of ITI (Extended Data Fig. 9a–i). Bleaching of the photometry signal over the course of the session cannot explain our observations, as we found the same effort encoding in a control session with ascending FR blocks (Extended Data Fig. 9j–m). Finally, ACh appeared to track effort but not reward size, because in a separate task that kept effort constant but varied sucrose concentration between 5% and 32%, ACh release was nearly identical for both rewards (Extended Data Fig. 9n–w). These data suggest that, contrary to DA release, which tracks both reward size and effort¹⁷, ACh release in the NAc tracks effort alone.

Fig. 4 — a, Schematic of ACh recordings during the sucrose task, with a representative image of GRAB ACh expression. Scale bar, 100 µm. Schematic was adapted with permission from ref. ⁶³, Elsevier. b, Trial structure (left), and within-session schedule of effort blocks (right). c, ACh release (green) and UV control signal (black) in the NAc during the sucrose task aligned to the first nose poke, reward delivery and reward consumption. n = 8 mice. d, Traces and heatmaps of reward-evoked ACh release from an example session. e, ACh release across FRs time locked to reward delivery, averaged across mice (n = 8). f, Expanded view of individual components of the ACh release waveform. The pre-reward ramp (top left) and the AUC of the pre-reward period (−5 to 0 s) for each FR (top right) are shown. For the mixed-effect model, fixed effect of FR: F(4,34) = 4.47, **P = 0.0052. The first peak (middle left) and the AUC of the first peak (0.25–0.75 s; middle right) are shown. For the mixed-effect model, fixed effect of FR: F(4,27) = 19.64, ****P < 0.0001. The second peak (bottom left) and the AUC for the second peak (1–2.5 s; bottom right) are shown. For the mixed-effect model, fixed effect of FR: F(4,27) = 6.17, **P = 0.0012. n = 8 mice. g, Schematic of AAV injections and representative images for the sucrose effort task with recording of GRAB ACh (top) and GRAB DA (bottom). Scale bars, 100 µm. Schematic was adapted with permission from ref. ⁶³, Elsevier. h, ACh (top; n = 8 mice) and DA (bottom; n = 6 mice) release averaged across all trials during the sucrose effort task. i, Enlarged plot of ACh (top) and DA (bottom) release from panel h, showing the first second after reward delivery. j, Overlay of the FR5 and FR21 ACh and DA signals immediately after reward delivery. k, Histogram of the latency to peak signal for all sucrose reward deliveries in the GRAB ACh and GRAB DA recordings (left), and a bar graph comparing the average latency to peak for GRAB ACh and GRAB DA (right). Significance was determined using an unpaired two-sided t-test, t = 58.18, d.f. = 4,225, ****P < 0.0001, n = 8 mice for GRAB ACh and 6 mice for GRAB DA. Data are mean ± s.e.m.

Source Data

Extended Data Fig. 9 — a. Top, heatmap and bottom, averaged traces of GRAB ACh activity during the sucrose effort task indicating the pre-reward ramp, first peak, dip, and second peak (n = 8 mice). b. GRAB ACh activity during the pre-reward ramp for trials separated by >30 s. c. Area under the curve (AUC, −5 to 0 s) for each FR during the pre-reward period. One-way ANOVA, F(4,497) = 9.68, ****p < 0.0001; n = 8 mice. d. GRAB ACh activity during the dip for trials separated by >30 s. e. AUC (0.75-1.25 s) during the dip. One-way ANOVA, F(4,497) = 2.74, *p = 0.028; n = 8 mice. f. GRAB ACh activity during the first peak for trials separated by >30 s. g. AUC (0.25-0.75 s) for each FR during the first peak. One-way ANOVA, F(4,497) = 5.39, ***p = 0.0003; n = 8 mice. h. GRAB ACh activity during the second peak for trials separated by >30 s. i. AUC (1-2.5 s) for each FR during the second peak. One-way ANOVA, F(4,497) = 7.21, ****p < 0.0001; n = 8 mice. j. Task design for the ascending FR sucrose effort task. k. GRAB ACh activity during the ascending sucrose effort task indicating the first peak. l. Closer view of the first peak of GRAB ACh activity during the ascending effort task. m. AUC (0.25-0.75 s) for each FR during the first peak. Mixed effects model, F(4,23) = 9.90, ****p < 0.0001; n = 8 mice. n. Task design for changing the concentration of sucrose reward in an FR5 task. o. GRAB ACh activity at reward delivery for 5% and 32% sucrose earned at FR5 (n = 8 mice). p. GRAB ACh activity during the pre-reward ramp for 5% and 32% sucrose rewards. q. AUC (−5-0 s) for each FR pre-reward. One-way ANOVA, F(4,497) = 9.68; p < 0.0001. Wilcoxon matched-pairs two-sided signed rank test, sum of signed ranks = −2.00, p = 0.95; n = 8 mice. r. GRAB ACh activity during the dip for 5% and 32% sucrose rewards. s. AUC (0.75-1.25 s) for each FR during the dip. Wilcoxon matched-pairs two-sided signed rank test, sum of signed ranks = 20.00, p = 0.20; n = 8 mice. t. GRAB ACh activity during the first peak for 5% and 32% sucrose rewards. u. AUC (0.25-0.75 s) for each FR during the first peak. Wilcoxon matched-pairs two-sided signed rank test, sum of signed ranks = 24.00, p = 0.11; n = 8 mice. v. GRAB ACh activity during the second peak for 5% and 32% sucrose rewards. w. AUC (1-2.5 s) for each FR during the second peak. Wilcoxon matched-pairs two-sided signed rank test, sum of signed ranks = 22.00, p = 0.15; n = 8 mice. Data are mean ± s.e.m.

Source Data

Although the NAc receives ACh inputs from multiple sources, local interneurons are thought to predominate when it comes to modulation of DA⁴⁰. We recorded the activity of cholinergic interneurons in the NAc during the effort task in ChAT–Cre mice expressing Cre-dependent GCaMP8m (Extended Data Fig. 10a,b). We observed a strikingly similar set of dynamics to our GRAB ACh recordings (Extended Data Fig. 10c–f), again with effort encoding independent from ITI (Extended Data Fig. 10g–o). These data indicate that cholinergic interneurons in the NAc are activated during states of high effort, putting them in a position to potentiate DA release upon reward delivery.

Extended Data Fig. 10 — a. Schematic of AAV injection and representative image for the recording of ChAT GCaMP activity in the sucrose effort task. Scale = 100 µm. The schematic was adapted from ref. ⁶³, Elsevier. b. Task structure for recording ChAT GCaMP activity during the sucrose effort task. c. ChAT GCaMP activity aligned to the first nosepoke, reward delivery, and reward consumption (n = 10 ChAT-Cre mice). d. ChAT GCaMP average traces and heat maps aligned to reward delivery from a single example session. e. ChAT GCaMP activity averaged across all mice and sessions during the sucrose effort task indicating the pre-reward ramp, first peak, dip, and second peak. f. ChAT GCaMP activity during the pre-reward, first peak and second peak of ChAT GCaMP response to reward delivery. Area under the curve for each FR during the pre-reward (−5-0 s) period, Friedman test, Friedman statistic = 24.16, ****p < 0.0001; during first peak (0.25-0.75 s), Friedman test, Friedman statistic = 28.48, ****p < 0.0001; and during second peak (1-2.5 s), Friedman test, Friedman statistic = 37.36, ****p < 0.0001; n = 10 ChAT-Cre mice. g. Top, heatmap and bottom, averaged traces of ChAT GCaMP activity across all mice (n = 10) during the sucrose effort task, restricted to trials with inter-trial interval (ITI) > 30 s. h. ChAT GCaMP activity during the pre-reward period of the reward delivery response for trials with ITI > 30 s. i. Area under the curve of the ChAT GCaMP activity during the pre-reward period (−5-0 s) for trials with ITI > 30 s. One-way ANOVA, F(4,1199) = 10.89, ****p < 0.0001, n = 10 ChAT-Cre mice. j. GRAB ACh activity during the dip for trials with ITI > 30 s. k. Area under the curve of the GRAB ACh activity during the dip (0.75-1.25 s) for trials with ITI > 30 s. One-way ANOVA, F(4,1199) = 5.23, ***p = 0.0004, n = 10 ChAT-Cre mice. l. ChAT GCaMP activity during the first peak of the reward delivery response for trials with ITI > 30 s. m. Area under the curve of the ChAT GCaMP activity during the first peak (0.25-0.75 s) for trials with ITI > 30 s. One-way ANOVA, F(4,1199) = 19.45, ****p < 0.0001, n = 10 ChAT-Cre mice. n. ChAT GCaMP activity during the second peak of the reward delivery response for trials with ITI > 30 s. o. Area under the curve of the ChAT GCaMP activity during the second peak (1-2.5 s) for trials with ITI > 30 s. One-way ANOVA, F(4,1199) = 12.50, ****p < 0.0001, n = 10 ChAT-Cre mice. Data are mean ± s.e.m.

Source Data

Acetylcholine regulates DA effort encoding and behaviour

To examine how ACh might augment DA release in this task, we first inspected the timing of ACh and DA release relative to reward delivery (Fig. 4g–i). Direct comparison suggested that reward-evoked ACh release precedes DA release by roughly 400 ms, a lag that holds true across all FR blocks (Fig. 4j,k), even though the on-kinetics of the GRAB DA sensor are faster than those of GRAB ACh (Y. Li, personal communication).

This temporal offset supports our hypothesis that an initial burst of ACh gates DA release to high-effort rewards, but the evidence is correlational. To directly test the role of ACh in DA effort encoding, we injected the NAc of ChAT–Cre mice with both GRAB DA and Cre-dependent tetanus toxin (TetTox), a construct that chronically silences neurons by preventing vesicular transmitter release (Fig. 5a). We trained the mice on the sucrose task and compared DA release patterns to control mice expressing eGFP instead of TetTox (Fig. 5b). Compared with control mice, TetTox mice displayed severely disrupted DA effort encoding (Fig. 5c,d). Similar to blockade of nicotinic receptors, we found a progressive disruption of DA release as effort increased, with a significant blockade of DA release already at FR5 (Fig. 5e,f).

Fig. 5 — a, Schematic and representative image of DA recordings during the sucrose effort task after chronic cholinergic neuron silencing via TetTox. Scale bar, 100 µm. Schematic was adapted with permission from ref. ⁶³, Elsevier. b, Experimental design and task structure. c, DA release during the sucrose effort task for mice expressing eGFP control (left) or TetTox in cholinergic neurons (right). d, AUC (0–4 s) of DA release for each FR from mice expressing eGFP (left; Friedman test, Friedman statistic = 22.93, ***P = 0.0001, n = 6) or TetTox (right; Friedman test, Friedman statistic = 15.87, **P = 0.0032, n = 6). e, DA release in FR5 (left) or FR21 (right) trials in mice with eGFP versus TetTox in ChAT neurons. f, AUC (0–4 s) of DA release for each FR in TetTox versus control mice. For the mixed-effects model, fixed effect of FR: F(4,20) = 42.55, ****P < 0.0001; fixed effect of TetTox: F(1,5) = 27.57, **P = 0.0033; and interaction of FR and TetTox: F(4,20) = 16.06, ****P < 0.0001. Sidak-corrected multiple comparisons: P = 0.23 for FR1 eGFP versus TetTox; **P = 0.29 for FR5 eGFP versus TetTox; ***P = 0.0001 for FR10 eGFP versus TetTox; ****P < 0.0001 for FR21 eGFP versus TetTox; and ****P < 0.0001 for FR46 eGFP versus TetTox. n = 6 ChAT–Cre mice expressing TetTox and 6 mice expressing eGFP. g, Schematic and representative image of DA recordings with cholinergic neuron inhibition during the sucrose task. Scale bar, 100 µm. Schematic was adapted with permission from ref. ⁶³, Elsevier. h, Experimental design and task structure with alternating days of red light to inhibit cholinergic neurons. i, DA release dynamics during the sucrose task for control sessions with no red light (left) and sessions with red light inhibition of cholinergic neurons (right). n = 8 ChAT–Cre mice. j, AUC (0–4 s) for each FR during no-light sessions (left; Friedman test, Friedman statistic = 25.10, ****P < 0.0001), and the AUC during red light sessions (right; Friedman test, Friedman statistic = 16.40, **P = 0.0025). n = 8 ChAT–Cre mice. k, Reward-evoked DA release in sessions with red light off or on for FR5 trials (left) and FR21 trials (right). n = 8 ChAT–Cre mice. l, AUC (0–4 s) for each FR with or without optical inhibition. For the mixed-effects model, fixed effect of FR: F(4,28) = 21.94, ****P < 0.0001; fixed effect of inhibition: F(1,7) = 23.05, **P = 0.0020; and interaction of FR and inhibition: F(4,28) = 3.52, *P = 0.019. Sidak-corrected multiple comparisons: P = 0.99 for FR1 no light versus inhibition; P = 0.96 for FR5 no light versus inhibition; *P = 0.024 for FR10 no light versus inhibition; **P = 0.0012 for FR21 no light versus inhibition; and **P = 0.0043 for FR46 no light versus inhibition. n = 8 ChAT–Cre mice. m, Schematic of the sucrose task and bilateral cannula implant in the NAc for PBS or DHβE infusion. Schematic was adapted with permission from ref. ⁶³, Elsevier. n, Rewards earned at each FR after microinjection of PBS. Friedman test, Friedman statistic = 21.25, ****P < 0.0001, n = 8 mice. o, Rewards earned at each FR after microinjection of DHβE. Friedman test, Friedman statistic = 25.33, ****P < 0.0001, n = 8 mice. p, Rewards earned for each FR after PBS or DHβE infusion. For the mixed-effects model, fixed effect of FR: F(4,28) = 36.36, ****P< 0.0001; fixed effect of drug: F(1,7) = 5.26, P = 0.056; and interaction of FR and drug: F(4,28) = 4.85, **P = 0.0042. Sidak-corrected multiple comparisons: P = 0.95 for FR1 PBS versus DHβE; P = 0.90 for FR5 PBS versus DHβE; *P = 0.016 for FR10 PBS versus DHβE; **P = 0.0069 for FR21 PBS versus DHβE; and P = 0.79 for FR46 PBS versus DHβE. n = 8 ChAT–Cre mice. Data are mean ± s.e.m.

Source Data

Although TetTox has the advantage that it completely abolishes transmitter release, it can also lead to chronic compensations in network activity, and requires between-subject comparisons. To further test the acute role of ACh release, we used optogenetics to inhibit cholinergic interneuron activity specifically during reward consumption and compare DA signals within the same subjects. New ChAT–Cre mice were prepared with GRAB DA and Cre-dependent NpHR, an inhibitory opsin sensitive to red light, in the NAc (Fig. 5g). These mice were trained to perform the sucrose task and then subjected to alternating days of red-light exposure for 4 s after each reward delivery (Fig. 5h). In trials with no red light, we observed the characteristic effort encoding by DA. When red light was delivered, however, effort encoding was significantly blunted (Fig. 5i,j). Similar to blockade of nicotinic receptors and TetTox, we found a progressive disruption of DA release as effort increased, in which low-effort DA release was no different than control conditions (Fig. 5k,l). We found no such disruption of DA effort encoding in control mice expressing mCherry rather than NpHR in cholinergic neurons (Extended Data Fig. 11a–f).

Extended Data Fig. 11 — a. Schematic of AAV injections and representative image for mCherry red light control of cholinergic interneurons with GRAB DA recording. Scale bar = 100 µm. b. Task design for red light control of cholinergic neuron inhibition sucrose effort task. c. GRAB DA activity during the sucrose task during no light and red-light sessions (n = 7 ChAT-Cre mice). d. Area under the curve (0-4 s) for each FR in no light sessions and red-light sessions. No light, Friedman test, Friedman statistic = 20.46, ****p < 0.0001; n = 7 ChAT-Cre mice. Red light: mixed-effects model, F(4,23) = 9.90, ****p < 0.0001; n = 7 ChAT-Cre mice. e. GRAB DA activity during the sucrose effort task for no light and red-light sessions comparing FR5 and FR21. f. Area under the curve (0-4 s) for each FR comparing no light to red-light. Mixed-effects model, fixed effect of FR, F(4,24) = 14.55, ****p < 0.0001; fixed effect of red-light, F(1,6) = 0.36, p = 0.57; fixed effect of FR and red-light interaction; F(4,23) = 0.23; p = 0.92. Sidak-corrected multiple comparisons: FR1 no light vs red-light, p = 0.99; FR5 no light vs red-light, p > 0.99; FR10 no light vs red-light, p = 0.99; FR21 no light vs red-light, p = 0.80; FR46 no light vs red-light, p = 0.99; n = 7 ChAT-Cre mice. g. Latency to initiate the next trial after reward delivery with PBS vs DHβE microinjection during the sucrose effort task, comparing low effort (FR1 and FR5) vs high effort (FR10, 21, and 46) trials. Mixed-effects model, fixed effect of effort, F(1,26) = 2.69, p = 0.11; fixed effect of drug, F(1,26) = 5.54, *p = 0.026; interaction of effort and drug; F(1,26) = 4.37; *p = 0.047. Sidak-corrected multiple comparisons: low vs high effort PBS, p = 0.93, low vs high effort DHβE, *p = 0.034; n = 8 mice. h. Schematic of cannula implant for the sucrose control task with PBS or DHβE microinjection. The schematics in panels a,h were adapted from ref. ⁶³, Elsevier. i. Task design for sucrose control task in which mice were injected with PBS or DHβE before performing an entire session at either FR1 or FR21. j. Rewards earned after PBS vs DHβE microinjection in FR1 sessions. Wilcoxon matched-pairs two-sided signed rank test, W = 11.00, p = 0.57; n = 9 mice. k. Rewards earned after PBS vs DHβE microinjection in FR21 sessions. Wilcoxon matched-pairs two-sided signed rank test, W = −45.00, ***p = 0.0039; n = 9 mice. Data are mean ± s.e.m.

Source Data

One possible consequence of increased DA release to high-effort rewards is the maintenance of effortful responding. Thus, we tested whether ACh–DA interactions in the NAc promote effortful behaviour. We prepared new mice with bilateral guide cannulas targeting the NAc for bilateral DHβE infusions (Fig. 5m). Compared with PBS infusions, DHβE treatment reduced sucrose reward responding and increased latency to initiate trials at high-FR blocks but not low-FR blocks (Fig. 5n–p and Extended Data Fig. 11g). To control for drug metabolism across the session, we tested the same mice in a counterbalanced, across-session schedule in which mice worked for sucrose at either FR1 or FR21 on alternating days (Extended Data Fig. 11h,i). Infusion of DHβE had no effect at FR1 but strongly reduced reward consumption at FR21 (Extended Data Fig. 11j,k). These findings support the proposition that when rewards are earned through high effort, ACh signalling potentiates DA release in the NAc, promoting continued effort exertion.

Discussion

We found that in high-effort conditions, ACh levels rise quickly in response to reward delivery, binding to nicotinic receptors on DA axons to augment DA release. Three different manipulations that prevented the release or action of ACh in the NAc dramatically reduced high-effort DA release. By contrast, two different manipulations that reduced VTA DA neuron activity had no detectable effects on the increases in NAc DA release after increased effort. This locally driven increase in DA release appears to motivate continued effort expenditure, because blocking the nicotinic receptors on NAc DA axons reduced reward earnings specifically when high effort was required.

Our results help to explain an observation across the animal kingdom that individuals value rewards more when they have worked harder for them^{2–5,41–44}. Such a process may be adaptive in promoting sustained reward seeking even when costs are high or rewards are scarce. We found that this evolutionarily conserved psychological effect appears to depend on the same receptors responsible for the reinforcing effects of nicotine⁴⁵. Future mechanistic models of addiction, therefore, may benefit from considering how the added value of effort, mediated by local interactions between ACh and DA axons, may reinforce drug-seeking behaviour.

We designed two versions of the effort task that utilized different forms of reinforcement: natural sucrose reward and ChRmine-mediated stimulation of DA axons. Although both paradigms elicited NAc DA release, they did so with different temporal dynamics and through different circuit mechanisms. Sucrose recruits a distributed set of brain regions⁴⁶, including VTA DA cell bodies, whereas axon stimulation is a much more spatially limited manipulation. Nonetheless, we found that local cholinergic signalling in the NAc is essential in both cases to augment DA release at high efforts. Thus, cholinergic modulation of DA release may be a generalized mechanism underlying effortful reward-seeking behaviour.

In vitro studies have long demonstrated robust cholinergic regulation of DA release^6–11. In slice, the stimulation of even a single cholinergic interneuron can elicit DA release in a wide swath of the striatum⁶. In vivo studies, however, have mostly failed to find large effects of ACh on DA release, at least in baseline conditions or during simple motor or reward tasks^12–14. However, these in vivo studies did not examine ACh–DA interactions in high-effort contexts. Only by modulating effort requirements were we able to unmask a functionally relevant effect of ACh on DA release. Our findings imply that that the interactions between neuromodulators are also highly contingent on behavioural context.

The findings are consistent with studies showing decreased motivation²⁷ and DA release^7,47 with blockade or deletion of nicotinic receptors. It is important to note, however, that the role of ACh in modulating DA and motivation is complex, with influential work demonstrating that cholinergic interneurons can oppose, rather than enhance, the motivational influence of appetitive cues^48–52. Our results here are temporally precise, at the time of reward delivery, and limited to the contribution of nicotinic receptors. It is likely that ACh–DA interactions vary depending on the ongoing behaviour⁵³, the preceding activity of DA and ACh neurons^22,23,52, the region of the striatum^38,52 and the type of ACh receptor examined⁵¹.

Although DHβE infusion in the NAc is predicted to largely affect DA axons (which are enriched with α4-containing and α6-containing nicotinic receptors; Extended Data Fig. 7a) while sparing other NAc circuit elements (which are mostly not; Extended Data Fig. 7b), it is possible that the behavioural effects of DHβE could be explained by other, non-dopaminergic mechanisms. For example, cortical inputs to the NAc may express α4-containing nicotinic receptors (Extended Data Fig. 7c–g), and thus DHβE could exert its behavioural effects by modulating the activity of these inputs, in addition to DA axons. Future work can test this possibility by specifically removing nicotinic receptors only on DA axons.

The precise ways that DA release modulates effortful behaviour in our task are complex. In our previous work, we inhibited DA release at each reward delivery, regardless of FR, and found that the animals tended to work harder¹⁷. Here, by inhibiting cholinergic signalling, we selectively inhibited DA release at higher FRs, and found that the animals worked less hard. Perhaps the relative balance of DA release at different FRs is a crucial feature that determines effort allocation. Future work should test how the specific timing of DA inhibition affects motivation, and explore what downstream circuit mechanisms could account for these behavioural effects.

The exact mechanism by which nicotinic channels modulate DA release, especially during optogenetic manipulation of DA axons, remains unknown. Our slice recordings demonstrate that ACh release reliably triggers action potentials or excitatory events in DA axons, yet multiple mechanisms probably work in concert to translate this excitation into enhanced release⁷. For example, during ChRmine stimulation of DA axons, it is possible that ACh recruits additional DA axons that either do not express ChRmine or are inaccessible to the light. Alternatively, by enhancing DA axon excitability, ACh may counteract any propagation failures from ChRmine-induced depolarization, especially by recruiting DA axon branches distal to the propagation failure. Future ex vivo work pairing optogenetic stimulation of DA axons with stimulation of nearby cholinergic neurons will allow for a greater mechanistic understanding of how ACh modulates DA release in this context.

In both ACh release and cholinergic interneuron activity, we observed a triphasic waveform at the time of reward delivery, consistent with previous work^37–39. Each phase of the waveform appeared to scale with effort. The initial peak is probably driven by glutamatergic inputs to the NAc^54–57, which could originate from cortical or thalamic regions that track the effort state of the animal. The rapid dip in cholinergic activity also scaled with effort, suggesting that it may be mediated by DA release. Indeed, cholinergic interneurons express D2 receptors and are modulated by DA under some circumstances^13,58. However, cholinergic pausing may also arise from VTA GABAergic neurons^59,60, which powerfully modulate DA neuron activity in reward tasks⁶¹. The final cholinergic peak may reflect rebound activation in response to the dip or a second phase of excitatory input^55,62. Future efforts should identify the specific inputs to cholinergic interneurons that determine each phase of the response and how these inputs modulate effortful behaviour.

Besides shedding light on effort valuation and the interactions between DA and ACh, our work speaks to an ongoing debate about whether the activity of DA cell bodies can be dissociated from the release of DA in striatal targets^28,29. We found that in the case of sucrose rewards, VTA DA neurons encode expended effort, probably accounting for at least some of the effect that we observed on DA release in the NAc. However, the cell body activity appeared to plateau at a lower level of effort than simultaneous recordings of VTA DA axons in the NAc as well as DA release itself. In addition, inhibiting VTA DA neurons did not reduce effort encoding in the NAc, whereas blocking local cholinergic signalling in the NAc dramatically suppressed effort encoding. Thus, we have shown that in the right context, local ACh can expand the dynamic range of DA release beyond what is driven by cell body activity, with an important role in boosting effortful behaviour.

Methods

Subjects

Male and female mice of C57BL/6J background (Jackson Laboratories, strain 000664) were used in approximately equal numbers for all experiments. Wild-type mice were crossed with DAT:IRES–Cre mice (Jackson Laboratory, strain 006660) or ChAT:IRES–Cre mice (Jackson Laboratory, strain 006410). For slice electrophysiology experiments, ChAT:IRES–Cre mice were crossed with DAT–Flp mice (Jackson Laboratories, strain 035436). All transgenic animals used in experiments were genotyped and found to be heterozygous for Cre-recombinase or Flp-recombinase. Mice were separated by sex and group housed after weaning before surgical procedures or behavioural assays. All behavioural experiments were conducted during the dark cycle (12 h light–dark) when mice were 10–24 weeks old. Mice were housed at approximately 21 °C in 30–70% humidity, and they were food restricted to 85% of ad libitum body weight for all behavioural experiments to facilitate motivated behaviour. All procedures complied with the animal care standards set forth by the National Institutes of Health (NIH) and were approved by the Stanford University Administrative Panel on Laboratory Animal Care.

Behavioural training

Food restriction and behavioural training began at a minimum of 2 weeks after virus injection and fibre implantation. Before starting behavioural experiments, mice were habituated for 2 days, including handling, tethering to the patch cords and allowing exploration of the operant boxes (Med Associates) for 30 min. After habituation, mice were exposed to 1 day of a Pavlovian task in which rewards were delivered at random intervals spanning 25–35 s for 30 min. Operant chambers were equipped with a speaker for white noise and three identical nose-poke ports, each with an associated cue light and an infrared emitter sensor to measure port entry times. Rewards were delivered in conjunction with a 2-s cue comprising white noise and central port light so that mice would associate reward delivery with the light–sound cue. For sucrose rewards, 10 µl of 32% (w/v) sucrose was used, apart from one experiment in which 5% sucrose was used (Extended Data Fig. 9n–w). After 1 day of Pavlovian training, mice progressed to the operant task. Mice trained on FR1 for a minimum of five sessions. The active nose-poke port (left or right) was counterbalanced between mice; each mouse continued with the same active port for all experiments. On days 1–3, the active food port was baited with a crushed portion of fruit loop to encourage exploration of that port. After earning at least 10 rewards at FR1, with an accuracy (% active nose pokes) of more than 80%, mice progressed to FR5. Mice continued FR5 for at least five sessions and until the number of rewards earned per session remained within 20% for 3 consecutive days. Once this criterion was met, mice progressed to the sucrose effort task. We did not attempt to equalize reward consumption between mice, but rather aimed to ensure that all mice had accurate and internally consistent performance. Training for optogenetic self-stimulation was achieved in the same manner as for sucrose, except the sucrose reward was replaced with 5-s optogenetic stimulation paired with the 2-s light and white-noise cue. All behavioural tasks were coded in Med-PC V (Med Associates).

Sucrose effort task

The sucrose effort task comprised five 10-min blocks of FRs including FR46, FR21, FR10, FR5 and FR1. After mice poked in the active port for the required number of times, sucrose reward (10 µl 32% (w/v)) was delivered in a central magazine, accompanied by a 2-s light and white-noise cue. Once the cue stopped and mice entered the magazine to consume the reward, they were free to start the next trial at their own pace by poking in the active port. As in our previous work¹⁷, the FRs were presented in descending order to prevent mice from achieving early satiety under low-cost conditions, except for one control experiment with ascending presentation of blocks (Extended Data Fig. 9j–m). Block transitions were not signalled to the mice.

Modified sucrose effort task

For microinjection experiments with PBS or DHβE, due to concern that the behavioural effect of the drug was a result of the drug wearing off over the course of the session, we used a modified sucrose effort task. In the modified task, after PBS or DHβE injection, mice underwent a 30-min session in which the poking requirement was kept constant for the entire session at either FR1 or FR21. The order of the 4 experimental days (PBS versus DHβE; FR1 versus FR21) was counterbalanced between mice.

ChRmine effort task

The ChRmine effort task was structured identically to the sucrose effort task, except that mice worked for optogenetic DA stimulation. ChRmine stimulation was paired with the same 2-s light and white-noise cue in the magazine port, and upon cue cessation, the mice were free to start the next trial at their own pace.

Stereotaxic injections and viruses

Mice (8–12 weeks old) were anaesthetized with isoflurane (5% induction, 1–2% maintenance). Subjects were fixed on a stereotaxic frame (Kopf Instruments), a small incision was made and burr holes were drilled in the skull over the sites of injection or fibre implantation. The following coordinates relative to bregma were used: VTA, −3.1 mm anteroposterior, ±0.4 mm mediolateral and 4.2 mm dorsoventral; NAc core, 1.5 mm anteroposterior, ±0.9 mm mediolateral and 4.1 mm dorsoventral from the skull surface. Microliter syringes (Hamilton) were lowered to the specified depth from the skull and used to inject 500 nl of virus solution at a flow rate of 0.1–0.25 ml min⁻¹. Borosilicate optic fibres for photometry and/or optogenetic stimulation with 200–400 µm diameter and numerical aperture 0.66 (Doric) were implanted directly above the striatal or midbrain virus injection site and secured to the skull using screws (Antrin Miniature Specialties) and light-cured dental adhesive cement (DenMat, Geristore A&B paste). For the cohorts with recordings in the NAc and cannula or optogenetic fibre placement in the midbrain (Fig. 2 and Extended Data Figs. 3–5), the VTA implant was cemented first and then the NAc fibre was implanted in the same hemisphere. For cohorts with drug microinjections and recordings, a dual optical fibre–cannula (multiple fluid injection cannula, 400 µm diameter, 0.66 numerical aperture; OmFC, Doric) was implanted in the NAc or midbrain. Cannulas for drug microinfusion were implanted 1.5 mm above the target site with the injector extending to the site. For all cohorts, the hemisphere targeted for recordings was counterbalanced between mice.

The adeno-associated viruses (AAVs) used for stereotaxic injections were AAV9-hSyn-GRAB_DA DA2m (DA4.4, WZ Biosciences), red-shifted AAV9-hSyn-GRAB_DA rDA2m (WZ Biosciences), AAV2/9-hSyn-gACh4m (GRAB ACh, brain VTA), AAV5-hSyn1-SIO-stGtACR2-FusionRed (Addgene), AAV9-syn-FLEX-jGCaMP8m-WPRE (Addgene), AAV-8-EF1a-DIO-rsChRmine-oScarlet-WPRE (Stanford Gene Vector and Virus Core), AAV-dj-CMV-DIO-eGFP-2A-TetTox (Stanford Gene Vector and Virus Core), AAV-dj-CMV-DIO-eGFP (Stanford Gene Vector and Virus Core), AAV-dj-Ef1a-DIO-NpHR3.0-mCherry (Stanford Gene Vector and Virus Core), AAV-dj-Ef1a-DIO-mCherry (Stanford Gene Vector and Virus Core) and AAV-dj-Ef1a-fDIO-eYFP (Stanford Gene Vector and Virus Core). AAV titres ranged from 1 × 10¹² to 2 × 10¹³ gc ml⁻¹.

Fibre photometry

Mice implanted with optical implants (400 µm diameter, 0.66 numerical aperture, Doric lenses) were connected via a ceramic sleeve (Precision Fiber Products) to low-autofluorescence patch cords of matching diameter and numerical aperture (Doric). Signals passed through a fibre optic rotary joint (Doric) before filtering through a fluorescence mini cube (Doric) and reaching a femtowatt photodetector (2151, Newport). For triple recordings (Fig. 2a–f), the signals from the VTA and NAc fibres were passed through different mini cubes to different photodetectors, thus ensuring full separation between the cell body and axon GCaMP recordings. LEDs (Doric) emitted frequency-modulated ultraviolet (405 nm) and blue (465 nm) light to stimulate control and either GRAB or GCaMP signals through the same fibre. Blue LED power was adjusted to approximately 35 µW at the fibre tip and lowered as needed if the signal saturated the photodetector. Ultraviolet LED power was adjusted to approximately 10 µW at the fibre tip. Digital signals were sampled at 1.0173 kHz, demodulated, lock-in amplified and recorded by a real-time signal processor (RZ5P, Tucker-Davis Technologies) using Synapse software (Tucker-Davis Technologies). Synchronized behavioural events from the Med Associates boxes were collected directly into the RZ5P digital input ports. To reduce any confounds from photobleaching, animals were recorded for about 5 min before behavioural testing began.

Optogenetic manipulations

Optogenetic manipulations with rsChRmine and NpHR were conducted simultaneously as fibre photometry recordings through the same optical implants. Optogenetic manipulation with stGtACR was conducted through optical implants of 200 µm with 0.66 numerical aperture without simultaneous fibre photometry recording. Optogenetic manipulation with rsChRmine or NpHR was conducted by connecting a 625-nm LED light source (Prizmatix) via a plastic fibre (1 mm diameter, NA 0.63) and a fibre optic rotary joint (Doric). rsChRmine stimulation was performed at a stimulation of 5 s, 20 Hz and 6 mW with a pulse width of 10 ms. Optogenetic inhibition with NpHR or control mCherry was conducted with a constant 4 s, approximately 6-mW light. Optogenetic manipulation with stGtACR2 was conducted with a 450-nm LED light source (Prizmatix), using 5–10 s of constant approximately 10-mW light.

Drug administration

For systemic injection experiments, mice were administered intraperitoneal injections of aprepitant 10 mg kg⁻¹ (Axon Medchem), DPCPX 2 mg kg⁻¹ (Tocris), istradeffyline 2 mg kg⁻¹ (Tocris) or SR 48692 5 mg kg⁻¹ (Tocris). DPCPX and instradeffyline were dissolved in DMSO (Sigma-Aldrich) and diluted to 20% DMSO in 0.9% saline to prevent precipitation. Aprepitant and SR 48692 were dissolved in DMSO and diluted to 5% DMSO in corn oil (Sigma-Aldrich) to prevent precipitation. Mice were injected with a volume of 10 ml kg⁻¹. After injection, mice were placed in their home cages for 10 min, then placed in the operant chambers for 5 min of habituation before beginning the rsChRmine effort task. For these agents, systemic injection was chosen over intracranial injection due to risk of brain injury from the necessary solvents.

For local injection experiments, mice were administered an intra-VTA infusion of muscimol 1 mM (Tocris) or intra-NAc infusion of DHβE hydrobromide 30 µg per hemisphere (Tocris), terazosin 3 µg (Tocris), osanetant 375 ng (Axon Medchem), scopolamine 22 nmol (Tocris) or ML375 105 pmol (Axon Medchem). Osanetant was dissolved in Tween 20% until homogeneous, then diluted to 1% Tween in filtered molecular-quality PBS to prevent precipitation. All other antagonists were dissolved in filtered molecular-quality PBS only. We also infused the GABA_A antagonist bicuculline (Tocris) and the GABA_B antagonist CGP 35348 (Tocris) into the NAc, but this treatment evoked severe motor deficits, including seizures in some cases, preventing mice from performing the task. Thus, we could not collect time-locked DA recordings during these infusions.

Drugs were infused through an injector cannula connected to a 5-µl Hamilton syringe using a microinfusion pump (Harvard Apparatus) at a continuous rate of 150 nl per minute to a total volume of 0.3 µl unilaterally or bilaterally. For unilateral infusions with fibre photometry recording, multiple fluid injection cannulas (400 µm diameter, 0.66 numerical aperture; OmFC, Doric) were used. For bilateral infusions, bilateral infusion cannulas (P1 Technologies) were used. Injector cannulas were removed 2 min after infusions were complete. After infusion, mice were placed in the operant chambers and allowed to habituate for 5 min before beginning the sucrose or rsChRmine effort task.

Validation of VTA DA cell body inhibition

To validate the effect of muscimol on VTA DA cell bodies, a cohort of DAT:IRES–Cre mice was injected with AAV9-syn-FLEX-jGCaMP8m into the VTA and implanted with multiple fluid injection cannulas (OmFC, Doric). To visualize the effect of muscimol in real time, one of these mice was prepared with a steel cross-bar for head-fixed recording during PBS or muscimol microinjection (Extended Data Fig. 4e–g). This mouse was habituated to a cylindrical enclosure and bilateral clamps were fixed to each side of the cross-bar. The pre-loaded injector was then inserted into the cannula and the recording began. After a 5-min baseline, the drug was injected and the recording continued for an additional 10 min. The injector was removed after the recording ended. Although this procedure allowed us to record photometry signals before, during and after the muscimol infusion, the head fixation prevented the animal from seeking or consuming liquid rewards in our behavioural setup. Thus, the remainder of the mice were not head fixed. Instead, they were microinjected with either muscimol or PBS immediately before a photometry recording that consisted of a 10-min baseline and then a 30-min, freely moving Pavlovian reward delivery session (the sucrose reward was delivered along with a sound or light cue every 30 s on average). Neural activity during both the baseline and Pavlovian reward task were then compared between muscimol and PBS sessions (Extended Data Fig. 4h–j). We chose the Pavlovian task because it minimized the motor requirements for these mice. Owing to the rotations elicited by muscimol, the mice could reliably enter one port (the magazine to consume sucrose), but not two (the active nose poke and the magazine), and thus they could not perform the sucrose task after muscimol infusion.

It was not technically possible to use fibre photometry to directly validate the effect of stGtACR on DA cell body activity because of saturation or photoswitching artefacts when this opsin is co-expressed with either green-shifted or red-shifted calcium sensors^14,64–66. Instead, stGtACR was validated through GRAB DA recordings in the NAc (Extended Data Fig. 5e–g).

Slice electrophysiology

Acute brain slices were prepared from adult ChAT–Cre:DAT–Flp mice between postnatal days 120–180. ChAT–Cre:DAT–Flp mice were sterotaxically injected with 500 nl of AAVdj-eF1a-fDIO-eYFP into the VTA (−3.1 mm anteroposterior, +0.4 mm mediolateral and −4.2 dorsolateral) and AAV8-Ef1a-DIO-rsChRmine-oScarlet into the NAc (+1.5 mm anteroposterior, +0.9 mm mediolateral and −4.1 mm dorsolateral) at postnatal days 90–120 and were used for electrophysiology experiments 4–8 weeks post-injection. Mice were anaesthetized with isoflurane anaesthesia and transcardially perfused before decapitation. Brains were carefully extracted and 250-µm thick coronal slices were cut using a vibratome (Leica VT1200 S) in ice-cold slice solution containing 110 mM sucrose, 62.5 mM NaCl, 2.5 mM KCl, 6 mM MgCl₂, 1.25 mM KH₂PO₄, 26 mM NaHCO₃, 0.5 mM CaCl₂ and 20 mM d-glucose, pH 7.35–7.40. Slices were incubated in slice solution for 20 min at 33 °C, then transferred to a room-temperature holding chamber with artificial cerebrospinal fluid (ACSF; 125 mM NaCl, 2.5 mM KCl, 1 mM MgCl₂, 1.25 mM KH₂PO₄, 26 mM NaHCO₃, 2 mM CaCl₂ and 20 mM d-glucose, pH 7.35–7.40). Brain slices were stored in room temperature ACSF and used for recording 30–300 min later. All solutions were saturated with carbogen.

Individual brain slices were placed in an RC-27 recording chamber (Warner Instruments) and superfused with carbogen-saturated ACSF (33–36 °C) at a flow rate of 2–3 ml min⁻¹. eYFP-positive cut ends of axons (blebs) originating from the VTA present at the slice surface in the NAc were visualized with a BX51WI upright microscope (Olympus) equipped with epifluorescence, IR-DIC optics, 470 nm (ET-GFP 49002, Chroma) and 596/83-nm single-band bandpass (FF01-596/83-25, Semrock) filter cubes, and a ×40 0.8 NA water immersion objective (LUMPLFLN, Olympus). Perforated-patch recordings were made using parafilm-wrapped borosilicate pipettes filled with internal solution containing 135 mM KCl, 10 mM NaCl, 2 mM MgCl₂, 10 mM HEPES, 0.5 mM EGTA and 0.1 mM CaCl₂, 280 mOsm, pH 7.4 adjusted with KOH. Recording electrodes (6–10 MΩ) were back filled with internal solution, then filled with internal containing 100 µg ml⁻¹ gramicidin (Sigma). Pipette capacitance neutralization and bridge balance were manually adjusted before current clamp recordings. ChAT–Cre⁺ neurons expressing rsChRmine were optically stimulated with 10-ms transistor-transistor logic (TTL) pulses every 10 s using orange (596 nm) light ranging from 0.2 to 3 mW delivered through the microscope objective by an LED driver (Thorlabs). Baseline axonal responses to rsChRmine stimulation in ACSF were measured for 3 min, and the average response from the first 10 sweeps at baseline was compared with the average response from the final 10 sweeps during a 3-min period in the presence of 1 µM DHβE (Tocris). Acute application of DHβE was achieved via gravity-driven bath perfusion. Current clamp recordings were sampled at 20 kHz and filtered at 10 kHz using a MultiClamp 700B and Digidata 1550B and analysed using pClamp11 (Molecular Devices). Data are presented as mean ± s.e.m., where n or N represent the number of cells and animals, respectively.

Immunohistochemistry

Mice were transcardially perfused with 4% (w/v) paraformaldehyde in PBS and brains were removed and post-fixed overnight at 4 °C. A vibratome (Leica Biosystems) was used to prepare 50-µm coronal sections. After three 10-min washes in PBS on a shaker, the slices were incubated with blocking solution (10% normal goat serum, 0.2% bovine serum albumin and 0.5% Triton X-100 in PBS) for 1 h. After one 10-min wash in PBS, slices were incubated in primary antibodies using a concentration of 1:1,000 for 12–20 h at 4 °C on a shaker. Primary antibodies included rat mCherry monoclonal antibody (M11217, Invitrogen), chicken anti-GFP (GFP-1020, Aves Labs), mouse anti-tyrosine hydroxylase (MAB318, Millipore), and, to visualize TetTox, the mouse anti-2A peptide (MABS2005, Millipore). After three washes of 10 min in PBS, secondary antibodies were added at a concentration of 1:750 and incubated for 2 h at room temperature on a shaker. Secondary antibodies included goat anti-rat Alexa Fluor 594 (A11007, Invitrogen), goat anti-chicken Alexa Fluor 488 (AB150169, Abcam) and goat anti-mouse Alexa Fluor 647 (A21235, Invitrogen). Both primary and secondary antibodies were incubated with slices in a carrier solution (1% normal goat serum, 0.2% bovine serum albumin and 0.5% Triton X-100 in PBS). After three more washes, the slices were mounted on SuperFrost Plus glass slides with DAPI Fluoromount-G mounting medium (Southern Biotech) for visualization using a Nikon A1 confocal microscope with NIS Elements AR software or a Keyence BZ-X800 microscope.

Data analysis

Subject mice were excluded (less than 5%) from the analysis if they did not reach behavioural criteria or on the basis of histology if they had (1) inaccurate implant placement, or (2) off-target or minimal transgene expression. MATLAB (MathWorks) scripts from Tucker-Davis Technologies were used for signal processing. Signals were downsampled by a factor of 10, underwent LOESS smoothing (window size of 30 ms) to reduce high-frequency noise and analysed in 15-s windows around the following timestamps: the first nose poke of each trial (excluding FR1 trials, where the first nose poke triggers reward delivery), reward delivery (which is coincident with a light–sound cue) and reward consumption (defined as when the mice first entered the magazine port after sucrose reward delivery). Entire sessions were debleached according to a previously published iterative method⁶⁷ which calculates the dF/F in short moving windows, centres and normalizes these windows, and then repeats these calculations for 100 temporally offset windows in the same session. Z-scores were then calculated using the mean and standard deviation of the signal spanning the entire session to minimize confounds in behaviour contaminating the local baseline periods for trial-based methods of analysis. As a control, the analyses were all repeated in a different manner, using local baselines to calculate Z-scores, and the results were qualitatively identical.

To separate the contribution of individual behavioural events that occurred close in time, we (1) fit regression models to predict photometry data based on multiple variables (see below), and (2) examined neural activity on the subset of trials with large ITIs (for example, more than 30 or 60 s), which revealed similar results (for example, Extended Data Figs. 2, 9 and 10). All trials within a session were first combined into a single session average, and then sessions were averaged together for each mouse. All photometry figures in the article show the mean and standard error of the photometry signal across mice, which is a more conservative approach than using the trial or session average. To quantify neural activity, we used AUC — calculated using the trapezoidal numerical integration of the Z-scores for the windows defined in the figure legends — or calculated the maximum value in the windows denoted in the figure legends. Windows for quantification were chosen based on visual inspection of the traces and applied consistently throughout the analysis for any direct comparison between traces. To minimize bleaching confounds, we removed the first 5 min of each recording, when the steepest bleaching was likely to occur. We also limited our interpretations to short windows of data, avoiding any analysis of longer-timescale changes, which are more likely to be confounded by bleaching or other gradual changes (for example, slight adjustments in the connection between the implant and the patch cord), and more likely to vary depending on the exact debleaching strategy used. In addition, for GRAB DA recordings, we took the approach reported by Sych et al.⁶⁸ and examined simultaneously recorded control signals (405 nm), which we found to be flat or slightly negative traces with substantially lower amplitude than what we observed with the experimental excitation (465 nm). Although this analysis is imperfect because 405 nm is not the isosbestic point for GRAB DA, the result implies that motion artefacts, bleaching or other intrinsic, non-DA-dependent signal changes could not have made a major contribution to our results.

To analyse spontaneous activity in VTA DA cell bodies after muscimol versus saline (Extended Data Fig. 4i), we debleached the signal by fitting a double-exponential curve and subtracting the fit from the raw data, and then used the MATLAB function findpeaks to identify all local peaks that were at least 1.5 Z-scores in prominence and 150 ms in width at half peak, as previously described⁶⁹.

Kernel regression analysis

To quantify the contribution of behavioural variables to neural activity, linear mixed effects models were used as in previous work^13,70–73. First, task-relevant behavioural predictors (for example, active nose poke, inactive nose poke, reward delivery, magazine entry, FR and ITI) were aligned to the photometry signal^13,70,74. Behavioural events were assigned as fixed effects, represented in a predictor matrix X of dimension N × p, where N is the number of timestamps in the photometry recording and p is the number of predictor variables¹³. Variables were represented in binary form, indicating whether a behavioural event occurred at a given timestamp. The mouse from which the data were collected was treated as a random effect, stored in a sparse design matrix Z of size N × m, where m is the number of mice from which data were collected. The predictor matrix X was time shifted by T₁ (T₁ = 60 timeshifts per 1.5 s for sucrose; T₁ = 100 timeshifts per 2.5 s for ChRmine) timestamps backwards and T₂ (T₂ = 125 timeshifts per 4 s for sucrose; T₂ = 400 timeshifts per 10 s for ChRmine) timestamps forwards to encapsulate effects of behavioural variables on the signal at previous and future timestamps¹³. The new, time-shifted predictor matrix was of dimension N × p(2(T_1 + T₂) + 1). Data for timestamps at which no behavioural event of interest occurred, or which did not fall within the [−T₁, T₂] timestamps of a behavioural event, were excluded. The fixed effects matrix X and random effects matrix Z were used to train a linear mixed effects model of the form y = Xβ + Zµ + ε, where y is the N × 1 dimensional response vector consisting of the photometry signal values at each timestamp, β is a p(2(T_1 + T₂) + 1) × 1 vector of β-coefficients for each time-shifted predictor variable, µ is an m × 1 vector of coefficients for each random effect, and ε is an N × 1 vector representing the residual variance that cannot be explained by the fixed or random effects. Response kernels were generated and plotted over the time window for each behavioural variable representing the contribution of each behavioral variable to the predicted fibre photometry signal^13,70,74. The observed signal was plotted against the predicted signal to demonstrate the efficacy of the model fit. The model was trained using tenfold cross-validation; reported R² values and plotted β-coefficients are the average across model runs for all tenfolds. The relative contribution of each behavioural variable to the total photometry signal was determined by performing a leave-one-out analysis where behavioural variables were sequentially removed from the model to observe the change in the overall predictive ability of the model as quantified by the R² value. The relative contribution of a variable was then defined as the proportional change in R² when that variable was removed from the model⁷¹.

Single-cell RNA sequencing analysis

We accessed the whole mouse brain 10X single-cell RNA sequencing datasets made publicly available by the Allen Institute⁷⁵. We analysed neurons that belong to cell classes already identified by the Allen Institute for each brain region. We restricted our analysis to the midbrain, striatal ventral region, anterior cingulate region, hippocampal region, prelimbic/infralimbic/orbital areas, thalamus and cortical subplate. Within each region outside of the NAc and midbrain, we examined cell types predicted to project to the NAc. We extracted expression levels for major cell types local to the region of interest, focusing on the following nicotinic α-subunit genes: Chrna1, Chrna2, Chrna3, Chrna4, Chrna5, Chrna6 and Chrna7. Transcript expression was quantified by calculating the log₂(CPM + 1), where CPM is counts per million. The percentage of cells expressing each transcript was also quantified. These analyses were performed in Python with custom scripts and notebooks from the Allen Institute.

Statistics and reproducibility

Investigators were blinded to the manipulations that experimental subjects had undergone during the behavioural testing, recordings and data analysis. All photometry analysis and behavioural analysis was conducted in MATLAB. Quantifications of the photometry and behavioural results were graphed and analysed using GraphPad. All data were tested for normality of sample distributions using the Shapiro–Wilk normality test, and when violated, non-parametric statistical tests were used. Normally distributed data were analysed by paired, two-tailed t-tests, or one-factor or two-factor repeated measures analysis of variance. Non-normally distributed data were analysed using the Wilcoxon matched-pairs signed rank test or Friedman’s test. When data were not paired, unpaired t-tests or the Mann–Whitney rank test was performed for normally and non-normally distributed data, respectively. If individual data points were missing from these matched comparisons, mixed-effects models were used instead. Mixed-effects models were also used when examining the effects of multiple fixed effects (for example, FR and drug), accounting for the random effect of subject. In these cases, the significance of the fixed effects was reported in the figure legends and if there were significant fixed effects, Sidak-corrected post-hoc comparisons were reported with asterisks in the figures. Kruskal–Wallis tests were used to compare three or more unmatched groups (for example, DA responses from trials with different ITIs). Spearman correlations were used to measure the association between two independently measured observations (for example, rewards on consecutive days). Fluorescent micrographs are representative of six mice (Fig. 1a), eight mice (Fig. 1l), six mice (Fig. 2b), six mice (Fig. 2d), eight mice (Fig. 2g), six mice (Fig. 3b), six mice (Fig. 3h), three mice (Fig. 3n), eight mice (Fig. 4a), eight mice (Fig. 4g, top), six mice (Fig. 4g, bottom), six mice (Fig. 5a), eight mice (Fig. 5g), six mice (Extended Data Fig. 6a), nine mice (Extended Data Fig. 10a) and seven mice (Extended Data Fig. 11a). For experiments involving multiple groups, subjects were randomly assigned to experimental conditions. No statistical methods were used to predetermine sample size. All statistical tests were two-sided and performed in MATLAB (Mathworks) or Prism (GraphPad). NS, not significant. *P < 0.05, **P < 0.01, ***P < 0.001 and ****P < 0.0001. In all figures, data are shown as mean ± s.e.m.

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.

Online content

Any methods, additional references, Nature Portfolio reporting summaries, source data, extended data, supplementary information, acknowledgements, peer review information; details of author contributions and competing interests; and statements of data and code availability are available at 10.1038/s41586-025-10046-6.

Supplementary information

Reporting Summary^{(470.2KB, pdf)}

Peer Review File^{(1.3MB, pdf)}

Source data

Source Data Fig. 1^{(26.3KB, xlsx)}

Source Data Fig. 2^{(27.7KB, xlsx)}

Source Data Fig. 3^{(30.5KB, xlsx)}

Source Data Fig. 4^{(218KB, xlsx)}

Source Data Fig. 5^{(36.8KB, xlsx)}

Source Data Extended Data Fig. 1^{(38.3KB, xlsx)}

Source Data Extended Data Fig. 2^{(100.3KB, xlsx)}

Source Data Extended Data Fig. 3^{(24.1KB, xlsx)}

Source Data Extended Data Fig. 4^{(404.5KB, xlsx)}

Source Data Extended Data Fig. 5^{(20.9KB, xlsx)}

Source Data Extended Data Fig. 6^{(45.2KB, xlsx)}

Source Data Extended Data Fig. 8^{(21.1KB, xlsx)}

Source Data Extended Data Fig. 9^{(80.5KB, xlsx)}

Source Data Extended Data Fig. 10^{(104.6KB, xlsx)}

Source Data Extended Data Fig. 11^{(55.1KB, xlsx)}

Acknowledgements

We thank the Eshel laboratory, Malenka laboratory and A. Shenhav for critical discussion; P. Kramer for technical assistance with DA axon recordings; J. Baek for computer-coding assistance; Y. Li for providing the gACh4m virus; and the Stanford Gene Vector and Virus core for reagents. This work was supported by philanthropic funds donated to the Nancy Pritzker Laboratory at Stanford University. G.C.T. was supported by the Stanford Medicine Berg Scholars Program. M.B.P. was supported by NIH grant K99 DA056573. Z.Z. was supported by the Stanford Wu Tsai Neurosciences Institute. R.C.M. was supported by a grant from the Stanford Wu Tsai Neurosciences Institute, a grant from the UCSF Dolby Family Center for Mood Disorders and NIH grant P50 DA042012. N.E. was supported by NIH grant K08 MH123791, a Brain & Behavior Research Foundation Young Investigator Grant, a Burroughs Wellcome Fund Career Award for Medical Scientists and a Simons Foundation Bridge to Independence Award.

Extended data figures and tables

Author contributions

M.B.P., R.C.M. and N.E. acquired funding. R.C.M. and N.E. provided key resources. G.C.T., M.B.P., T.Y., M.W. and N.E. performed fibre photometry, optogenetics, pharmacology and behavioural experiments. V.M. and Z.Z. performed the computational modelling and data analysis. N.D. performed the slice electrophysiology. M.B.P. performed the RNA sequencing analyses. M.B.P., G.C.T. and N.E. visualized the data and designed the figures. The manuscript was written by M.B.P., G.C.T. and N.E., and then reviewed and edited by all authors.

Peer review

Peer review information

Nature thanks Arif Hamid, Margaret Rice and the other, anonymous, reviewer(s) for their contribution to the peer review of this work. Peer reviewer reports are available.

Data availability

Source data for all datasets generated and analysed in this study are provided with this article. All other data, including raw photometry data, are available on request from the corresponding authors. Publicly available single-cell RNA sequencing datasets from the Allen Brain Institute can be accessed from ref. ⁷⁵. Source data are provided with this paper.

Code availability

The code used for data processing and analysis is available from the corresponding authors on reasonable request.

Competing interests

R.C.M. is an advisor to Bayshore Global Management and on the scientific advisory boards of MapLight Therapeutics, MindMed and Aelis Farma. N.E. is a consultant for Boehringer Ingelheim. The other authors declare no competing interests.

Footnotes

Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

These authors contributed equally: Gavin C. Touponse, Matthew B. Pomrenze

Extended data

is available for this paper at 10.1038/s41586-025-10046-6.

Supplementary information

The online version contains supplementary material available at 10.1038/s41586-025-10046-6.

References

1.Hull, C. L. Principles of Behavior: an Introduction to Behavior Theory (Appleton-Century, 1943).
2.Inzlicht, M., Shenhav, A. & Olivola, C. Y. The effort paradox: effort is both costly and valued. Trends Cogn. Sci.22, 337–349 (2018). [DOI] [PMC free article] [PubMed] [Google Scholar]
3.Czaczkes, T. J., Brandstetter, B., di Stefano, I. & Heinze, J. Greater effort increases perceived value in an invertebrate. J. Comp. Psychol.132, 200–209 (2018). [DOI] [PubMed] [Google Scholar]
4.Norton, M. I., Mochon, D. & Ariely, D. The IKEA effect: when labor leads to love. J. Consum. Psychol.22, 453–460 (2012). [Google Scholar]
5.Kacelnik, A. & Marsh, B. Cost can increase preference in starlings. Anim. Behav.63, 245–250 (2002). [Google Scholar]
6.Matityahu, L. et al. Acetylcholine waves and dopamine release in the striatum. Nat. Commun.14, 6852 (2023). [DOI] [PMC free article] [PubMed] [Google Scholar]
7.Liu, C. et al. An action potential initiation mechanism in distal axons for the control of dopamine release. Science375, 1378–1385 (2022). [DOI] [PMC free article] [PubMed] [Google Scholar]
8.Cachope, R. et al. Selective activation of cholinergic interneurons enhances accumbal phasic dopamine release: setting the tone for reward processing. Cell Rep.2, 33–41 (2012). [DOI] [PMC free article] [PubMed] [Google Scholar]
9.Threlfell, S. et al. Striatal dopamine release is triggered by synchronized activity in cholinergic interneurons. Neuron75, 58–64 (2012). [DOI] [PubMed] [Google Scholar]
10.Kosillo, P., Zhang, Y.-F., Threlfell, S. & Cragg, S. J. Cortical control of striatal dopamine transmission via striatal cholinergic interneurons. Cereb. Cortex26, 4160–4169 (2016). [DOI] [PMC free article] [PubMed] [Google Scholar]
11.Kramer, P. F. et al. Synaptic-like axo-axonal transmission from striatal cholinergic interneurons onto dopaminergic fibers. Neuron110, 2949–2960.e4 (2022). [DOI] [PMC free article] [PubMed] [Google Scholar]
12.Krok, A. C. et al. Intrinsic dopamine and acetylcholine dynamics in the striatum of mice. Nature621, 543–549 (2023). [DOI] [PMC free article] [PubMed] [Google Scholar]
13.Chantranupong, L. et al. Dopamine and glutamate regulate striatal acetylcholine in decision-making. Nature621, 577–585 (2023). [DOI] [PMC free article] [PubMed] [Google Scholar]
14.Taniguchi, J. et al. Comment on ‘Accumbens cholinergic interneurons dynamically promote dopamine release and enable motivation’. eLife13, e95694 (2024). [DOI] [PMC free article] [PubMed] [Google Scholar]
15.Schultz, W. Updating dopamine reward signals. Curr. Opin. Neurobiol.23, 229–238 (2013). [DOI] [PMC free article] [PubMed] [Google Scholar]
16.Berke, J. D. What does dopamine mean? Nat. Neurosci.21, 787–793 (2018). [DOI] [PMC free article] [PubMed] [Google Scholar]
17.Eshel, N. et al. Striatal dopamine integrates cost, benefit, and motivation. Neuron112, 500–514.e5 (2024). [DOI] [PMC free article] [PubMed] [Google Scholar]
18.Tanaka, S., O’Doherty, J. P. & Sakagami, M. The cost of obtaining rewards enhances the reward prediction error signal of midbrain dopamine neurons. Nat. Commun.10, 3674 (2019). [DOI] [PMC free article] [PubMed] [Google Scholar]
19.Covey, D. P., Hernandez, E., Luján, M. Á & Cheer, J. F. Chronic augmentation of endocannabinoid levels persistently increases dopaminergic encoding of reward cost and motivation. J. Neurosci.41, 6946–6953 (2021). [DOI] [PMC free article] [PubMed] [Google Scholar]
20.Holly, E. N., Galanaugh, J. & Fuccillo, M. V. Local regulation of striatal dopamine: a diversity of circuit mechanisms for a diversity of behavioral functions? Curr. Opin. Neurobiol.85, 102839 (2024). [DOI] [PMC free article] [PubMed] [Google Scholar]
21.Cachope, R. & Cheer, J. F. Local control of striatal dopamine release. Front. Behav. Neurosci.8, 188 (2014). [DOI] [PMC free article] [PubMed] [Google Scholar]
22.Sulzer, D., Cragg, S. J. & Rice, M. E. Striatal dopamine neurotransmission: regulation of release and uptake. Basal Ganglia6, 123–148 (2016). [DOI] [PMC free article] [PubMed] [Google Scholar]
23.Threlfell, S. & Cragg, S. J. Dopamine signaling in dorsal versus ventral striatum: the dynamic role of cholinergic interneurons. Front. Syst. Neurosci.5, 11 (2011). [DOI] [PMC free article] [PubMed] [Google Scholar]
24.Cai, X. et al. Dopamine dynamics are dispensable for movement but promote reward responses. Nature635, 406–414 (2024). [DOI] [PMC free article] [PubMed] [Google Scholar]
25.Liu, C., Goel, P. & Kaeser, P. S. Spatial and temporal scales of dopamine transmission. Nat. Rev. Neurosci.22, 345–358 (2021). [DOI] [PMC free article] [PubMed] [Google Scholar]
26.Gershman, S. J. et al. Explaining dopamine through prediction errors and beyond. Nat. Neurosci.27, 1645–1655 (2024). [DOI] [PubMed] [Google Scholar]
27.Mohebi, A., Collins, V. L. & Berke, J. D. Accumbens cholinergic interneurons dynamically promote dopamine release and enable motivation. eLife12, e85011 (2023). [DOI] [PMC free article] [PubMed] [Google Scholar]
28.Mohebi, A. et al. Dissociable dopamine dynamics for learning and motivation. Nature570, 65–70 (2019). [DOI] [PMC free article] [PubMed] [Google Scholar]
29.de Jong, J. W., Liang, Y., Verharen, J. P. H., Fraser, K. M. & Lammel, S. State and rate-of-change encoding in parallel mesoaccumbal dopamine pathways. Nat. Neurosci.27, 309–318 (2024). [DOI] [PMC free article] [PubMed] [Google Scholar]
30.Mahn, M. et al. High-efficiency optogenetic silencing with soma-targeted anion-conducting channelrhodopsins. Nat. Commun.9, 4125 (2018). [DOI] [PMC free article] [PubMed] [Google Scholar]
31.Zoli, M. et al. Identification of the nicotinic receptor subtypes expressed on dopaminergic terminals in the rat striatum. J. Neurosci.22, 8785–8789 (2002). [DOI] [PMC free article] [PubMed] [Google Scholar]
32.Leyrer-Jackson, J. M. et al. Accumbens cholinergic interneurons mediate cue-induced nicotine seeking and associated glutamatergic plasticity. eNeuro8, ENEURO.0276-20.2020 (2021). [DOI] [PMC free article] [PubMed] [Google Scholar]
33.Witten, I. B. et al. Cholinergic interneurons control local circuit activity and cocaine conditioning. Science330, 1677–1681 (2010). [DOI] [PMC free article] [PubMed] [Google Scholar]
34.Crespo, J. A., Sturm, K., Saria, A. & Zernig, G. Activation of muscarinic and nicotinic acetylcholine receptors in the nucleus accumbens core is necessary for the acquisition of drug reinforcement. J. Neurosci.26, 6004–6010 (2006). [DOI] [PMC free article] [PubMed] [Google Scholar]
35.Yee, J. et al. Muscarinic acetylcholine receptors in the nucleus accumbens core and shell contribute to cocaine priming-induced reinstatement of drug seeking. Eur. J. Pharmacol.650, 596–604 (2011). [DOI] [PMC free article] [PubMed] [Google Scholar]
36.Nunes, E. J., Kebede, N., Bagdas, D. & Addy, N. A. Cholinergic and dopaminergic-mediated motivated behavior in healthy states and in substance use and mood disorders. J. Exp. Anal. Behav.117, 404–419 (2022). [DOI] [PMC free article] [PubMed] [Google Scholar]
37.Morris, G., Arkadir, D., Nevet, A., Vaadia, E. & Bergman, H. Coincident but distinct messages of midbrain dopamine and striatal tonically active neurons. Neuron43, 133–143 (2004). [DOI] [PubMed] [Google Scholar]
38.Duhne, M., Mohebi, A., Kim, K., Pelattini, L. & Berke, J. D. A mismatch between striatal cholinergic pauses and dopaminergic reward prediction errors. Proc. Natl Acad. Sci. USA121, e2410828121 (2024). [DOI] [PMC free article] [PubMed] [Google Scholar]
39.Aosaki, T. et al. Responses of tonically active neurons in the primate’s striatum undergo systematic changes during behavioral sensorimotor conditioning. J. Neurosci.14, 3969–3984 (1994). [DOI] [PMC free article] [PubMed] [Google Scholar]
40.Brimblecombe, K. R. et al. Targeted activation of cholinergic interneurons accounts for the modulation of dopamine by striatal nicotinic receptors. eNeuro5, ENEURO.0397-17 (2018). [Google Scholar]
41.Lydall, E. S., Gilmour, G. & Dwyer, D. M. Rats place greater value on rewards produced by high effort: an animal analogue of the “effort justification” effect. J. Exp. Soc. Psychol.46, 1134–1137 (2010). [Google Scholar]
42.Johnson, A. W. & Gallagher, M. Greater effort boosts the affective taste properties of food. Proc. Biol. Sci.278, 1450–1456 (2011). [DOI] [PMC free article] [PubMed] [Google Scholar]
43.Clement, T. S., Feltus, J. R., Kaiser, D. H. & Zentall, T. R. Work ethic’ in pigeons: reward value is directly related to the effort or time required to obtain the reward. Psychon. Bull. Rev.7, 100–106 (2000). [DOI] [PubMed] [Google Scholar]
44.Pompilio, L., Kacelnik, A. & Behmer, S. T. State-dependent learned valuation drives choice in an invertebrate. Science311, 1613–1615 (2006). [DOI] [PubMed] [Google Scholar]
45.Maskos, U. et al. Nicotine reinforcement and cognition restored by targeted expression of nicotinic receptors. Nature436, 103–107 (2005). [DOI] [PubMed] [Google Scholar]
46.Bijoch, Ł et al. Whole-brain tracking of cocaine and sugar rewards processing. Transl. Psychiatry13, 20 (2023). [DOI] [PMC free article] [PubMed] [Google Scholar]
47.Koranda, J. L. et al. Nicotinic receptors regulate the dynamic range of dopamine release in vivo. J. Neurophysiol.111, 103–111 (2014). [DOI] [PMC free article] [PubMed] [Google Scholar]
48.Collins, A. L. et al. Nucleus accumbens cholinergic interneurons oppose cue-motivated behavior. Biol. Psychiatry86, 388–396 (2019). [DOI] [PMC free article] [PubMed] [Google Scholar]
49.Franklin, N. T. & Frank, M. J. A cholinergic feedback circuit to regulate striatal population uncertainty and optimize reinforcement learning. eLife4, e12029 (2015). [DOI] [PMC free article] [PubMed] [Google Scholar]
50.Cox, J. & Witten, I. B. Striatal circuits for reward learning and decision-making. Nat. Rev. Neurosci.20, 482–494 (2019). [DOI] [PMC free article] [PubMed] [Google Scholar]
51.Collins, A. L., Aitken, T. J., Greenfield, V. Y., Ostlund, S. B. & Wassum, K. M. Nucleus accumbens acetylcholine receptors modulate dopamine and motivation. Neuropsychopharmacology41, 2830–2838 (2016). [DOI] [PMC free article] [PubMed] [Google Scholar]
52.Zhang, Y.-F. et al. An axonal brake on striatal dopamine output by cholinergic interneurons. Nat. Neurosci.28, 783–794 (2025). [DOI] [PMC free article] [PubMed] [Google Scholar]
53.Costa, K. M. et al. Dopamine and acetylcholine correlations in the nucleus accumbens depend on behavioral task states. Curr. Biol.35, 1400–1407.e3 (2025). [DOI] [PMC free article] [PubMed] [Google Scholar]
54.Klug, J. R. et al. Differential inputs to striatal cholinergic and parvalbumin interneurons imply functional distinctions. eLife7, e35657 (2018). [DOI] [PMC free article] [PubMed] [Google Scholar]
55.Ratna, D. D. & Francis, T. C. Extrinsic and intrinsic control of striatal cholinergic interneuron activity. Front. Mol. Neurosci.18, 1528419 (2025). [DOI] [PMC free article] [PubMed] [Google Scholar]
56.Guo, Q. et al. Whole-brain mapping of inputs to projection neurons and cholinergic interneurons in the dorsal striatum. PLoS ONE10, e0123381 (2015). [DOI] [PMC free article] [PubMed] [Google Scholar]
57.Ding, J. B., Guzman, J. N., Peterson, J. D., Goldberg, J. A. & Surmeier, D. J. Thalamic gating of corticostriatal signaling by cholinergic interneurons. Neuron67, 294–307 (2010). [DOI] [PMC free article] [PubMed] [Google Scholar]
58.Gallo, E. F. et al. Dopamine D2 receptors modulate the cholinergic pause and inhibitory learning. Mol. Psychiatry27, 1502–1514 (2022). [DOI] [PMC free article] [PubMed] [Google Scholar]
59.Brown, M. T. C. et al. Ventral tegmental area GABA projections pause accumbal cholinergic interneurons to enhance associative learning. Nature492, 452–456 (2012). [DOI] [PubMed] [Google Scholar]
60.Al-Hasani, R. et al. Ventral tegmental area GABAergic inhibition of cholinergic interneurons in the ventral nucleus accumbens shell promotes reward reinforcement. Nat. Neurosci.24, 1414–1428 (2021). [DOI] [PMC free article] [PubMed] [Google Scholar]
61.Eshel, N. et al. Arithmetic and local circuitry underlying dopamine prediction errors. Nature525, 243–246 (2015). [DOI] [PMC free article] [PubMed] [Google Scholar]
62.Schulz, J. M. & Reynolds, J. N. J. Pause and rebound: sensory control of cholinergic signaling in the striatum. Trends Neurosci.36, 41–50 (2013). [DOI] [PubMed] [Google Scholar]
63.Pomrenze, M. B. et al. Modulation of 5-HT release by dynorphin mediates social deficits during opioid withdrawal. Neuron110, 4125–4143.e6 (2022). [DOI] [PMC free article] [PubMed] [Google Scholar]
64.Akerboom, J. et al. Genetically encoded calcium indicators for multi-color neural activity imaging and combination with optogenetics. Front. Mol. Neurosci.6, 2 (2013). [DOI] [PMC free article] [PubMed] [Google Scholar]
65.Dana, H. et al. Sensitive red protein calcium indicators for imaging neural activity. eLife5, e12727 (2016). [DOI] [PMC free article] [PubMed] [Google Scholar]
66.Wu, J. et al. Improved orange and red Ca²⁺ indicators and photophysical considerations for optogenetic applications. ACS Chem. Neurosci.4, 963–972 (2013). [DOI] [PMC free article] [PubMed] [Google Scholar]
67.Bruno, C. A. et al. pMAT: an open-source software suite for the analysis of fiber photometry data. Pharmacol. Biochem. Behav.201, 173093 (2021). [DOI] [PMC free article] [PubMed] [Google Scholar]
68.Sych, Y., Chernysheva, M., Sumanovski, L. T. & Helmchen, F. High-density multi-fiber photometry for studying large-scale brain circuit dynamics. Nat. Methods16, 553–560 (2019). [DOI] [PubMed] [Google Scholar]
69.Higginbotham, J. A. et al. Estradiol protects against pain-facilitated fentanyl use via suppression of opioid-evoked dopamine activity in males. Neuron113, 1413–1429.e5 (2025). [DOI] [PMC free article] [PubMed] [Google Scholar]
70.Blanco-Pozo, M., Akam, T. & Walton, M. E. Dopamine-independent effect of rewards on choices through hidden-state inference. Nat. Neurosci.27, 286–297 (2024). [DOI] [PMC free article] [PubMed] [Google Scholar]
71.Elum, J. E. et al. Distinct dynamics and intrinsic properties in ventral tegmental area populations mediate reward association and motivation. Cell Rep.43, 114668 (2024). [DOI] [PMC free article] [PubMed] [Google Scholar]
72.Engelhard, B. et al. Specialized coding of sensory, motor and cognitive variables in VTA dopamine neurons. Nature570, 509–513 (2019). [DOI] [PMC free article] [PubMed] [Google Scholar]
73.Parker, N. F. et al. Choice-selective sequences dominate in cortical relative to thalamic inputs to NAc to support reinforcement learning. Cell Rep.39, 110756 (2022). [DOI] [PMC free article] [PubMed] [Google Scholar]
74.Parker, N. F. et al. Reward and choice encoding in terminals of midbrain dopamine neurons depends on striatal target. Nat. Neurosci.19, 845–854 (2016). [DOI] [PMC free article] [PubMed] [Google Scholar]
75.Yao, Z. et al. A high-resolution transcriptomic and spatial atlas of cell types in the whole mouse brain. Nature624, 317–332 (2023). [DOI] [PMC free article] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

Reporting Summary^{(470.2KB, pdf)}

Peer Review File^{(1.3MB, pdf)}

Source Data Fig. 1^{(26.3KB, xlsx)}

Source Data Fig. 2^{(27.7KB, xlsx)}

Source Data Fig. 3^{(30.5KB, xlsx)}

Source Data Fig. 4^{(218KB, xlsx)}

Source Data Fig. 5^{(36.8KB, xlsx)}

Source Data Extended Data Fig. 1^{(38.3KB, xlsx)}

Source Data Extended Data Fig. 2^{(100.3KB, xlsx)}

Source Data Extended Data Fig. 3^{(24.1KB, xlsx)}

Source Data Extended Data Fig. 4^{(404.5KB, xlsx)}

Source Data Extended Data Fig. 5^{(20.9KB, xlsx)}

Source Data Extended Data Fig. 6^{(45.2KB, xlsx)}

Source Data Extended Data Fig. 8^{(21.1KB, xlsx)}

Source Data Extended Data Fig. 9^{(80.5KB, xlsx)}

Source Data Extended Data Fig. 10^{(104.6KB, xlsx)}

Source Data Extended Data Fig. 11^{(55.1KB, xlsx)}

Data Availability Statement

The code used for data processing and analysis is available from the corresponding authors on reasonable request.

[CR1] 1.Hull, C. L. Principles of Behavior: an Introduction to Behavior Theory (Appleton-Century, 1943).

[CR2] 2.Inzlicht, M., Shenhav, A. & Olivola, C. Y. The effort paradox: effort is both costly and valued. Trends Cogn. Sci.22, 337–349 (2018). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR3] 3.Czaczkes, T. J., Brandstetter, B., di Stefano, I. & Heinze, J. Greater effort increases perceived value in an invertebrate. J. Comp. Psychol.132, 200–209 (2018). [DOI] [PubMed] [Google Scholar]

[CR4] 4.Norton, M. I., Mochon, D. & Ariely, D. The IKEA effect: when labor leads to love. J. Consum. Psychol.22, 453–460 (2012). [Google Scholar]

[CR5] 5.Kacelnik, A. & Marsh, B. Cost can increase preference in starlings. Anim. Behav.63, 245–250 (2002). [Google Scholar]

[CR6] 6.Matityahu, L. et al. Acetylcholine waves and dopamine release in the striatum. Nat. Commun.14, 6852 (2023). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR7] 7.Liu, C. et al. An action potential initiation mechanism in distal axons for the control of dopamine release. Science375, 1378–1385 (2022). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR8] 8.Cachope, R. et al. Selective activation of cholinergic interneurons enhances accumbal phasic dopamine release: setting the tone for reward processing. Cell Rep.2, 33–41 (2012). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR9] 9.Threlfell, S. et al. Striatal dopamine release is triggered by synchronized activity in cholinergic interneurons. Neuron75, 58–64 (2012). [DOI] [PubMed] [Google Scholar]

[CR10] 10.Kosillo, P., Zhang, Y.-F., Threlfell, S. & Cragg, S. J. Cortical control of striatal dopamine transmission via striatal cholinergic interneurons. Cereb. Cortex26, 4160–4169 (2016). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR11] 11.Kramer, P. F. et al. Synaptic-like axo-axonal transmission from striatal cholinergic interneurons onto dopaminergic fibers. Neuron110, 2949–2960.e4 (2022). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR12] 12.Krok, A. C. et al. Intrinsic dopamine and acetylcholine dynamics in the striatum of mice. Nature621, 543–549 (2023). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR13] 13.Chantranupong, L. et al. Dopamine and glutamate regulate striatal acetylcholine in decision-making. Nature621, 577–585 (2023). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR14] 14.Taniguchi, J. et al. Comment on ‘Accumbens cholinergic interneurons dynamically promote dopamine release and enable motivation’. eLife13, e95694 (2024). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR15] 15.Schultz, W. Updating dopamine reward signals. Curr. Opin. Neurobiol.23, 229–238 (2013). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR16] 16.Berke, J. D. What does dopamine mean? Nat. Neurosci.21, 787–793 (2018). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR17] 17.Eshel, N. et al. Striatal dopamine integrates cost, benefit, and motivation. Neuron112, 500–514.e5 (2024). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR18] 18.Tanaka, S., O’Doherty, J. P. & Sakagami, M. The cost of obtaining rewards enhances the reward prediction error signal of midbrain dopamine neurons. Nat. Commun.10, 3674 (2019). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR19] 19.Covey, D. P., Hernandez, E., Luján, M. Á & Cheer, J. F. Chronic augmentation of endocannabinoid levels persistently increases dopaminergic encoding of reward cost and motivation. J. Neurosci.41, 6946–6953 (2021). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR20] 20.Holly, E. N., Galanaugh, J. & Fuccillo, M. V. Local regulation of striatal dopamine: a diversity of circuit mechanisms for a diversity of behavioral functions? Curr. Opin. Neurobiol.85, 102839 (2024). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR21] 21.Cachope, R. & Cheer, J. F. Local control of striatal dopamine release. Front. Behav. Neurosci.8, 188 (2014). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR22] 22.Sulzer, D., Cragg, S. J. & Rice, M. E. Striatal dopamine neurotransmission: regulation of release and uptake. Basal Ganglia6, 123–148 (2016). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR23] 23.Threlfell, S. & Cragg, S. J. Dopamine signaling in dorsal versus ventral striatum: the dynamic role of cholinergic interneurons. Front. Syst. Neurosci.5, 11 (2011). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR24] 24.Cai, X. et al. Dopamine dynamics are dispensable for movement but promote reward responses. Nature635, 406–414 (2024). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR25] 25.Liu, C., Goel, P. & Kaeser, P. S. Spatial and temporal scales of dopamine transmission. Nat. Rev. Neurosci.22, 345–358 (2021). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR26] 26.Gershman, S. J. et al. Explaining dopamine through prediction errors and beyond. Nat. Neurosci.27, 1645–1655 (2024). [DOI] [PubMed] [Google Scholar]

[CR27] 27.Mohebi, A., Collins, V. L. & Berke, J. D. Accumbens cholinergic interneurons dynamically promote dopamine release and enable motivation. eLife12, e85011 (2023). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR28] 28.Mohebi, A. et al. Dissociable dopamine dynamics for learning and motivation. Nature570, 65–70 (2019). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR29] 29.de Jong, J. W., Liang, Y., Verharen, J. P. H., Fraser, K. M. & Lammel, S. State and rate-of-change encoding in parallel mesoaccumbal dopamine pathways. Nat. Neurosci.27, 309–318 (2024). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR30] 30.Mahn, M. et al. High-efficiency optogenetic silencing with soma-targeted anion-conducting channelrhodopsins. Nat. Commun.9, 4125 (2018). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR31] 31.Zoli, M. et al. Identification of the nicotinic receptor subtypes expressed on dopaminergic terminals in the rat striatum. J. Neurosci.22, 8785–8789 (2002). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR32] 32.Leyrer-Jackson, J. M. et al. Accumbens cholinergic interneurons mediate cue-induced nicotine seeking and associated glutamatergic plasticity. eNeuro8, ENEURO.0276-20.2020 (2021). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR33] 33.Witten, I. B. et al. Cholinergic interneurons control local circuit activity and cocaine conditioning. Science330, 1677–1681 (2010). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR34] 34.Crespo, J. A., Sturm, K., Saria, A. & Zernig, G. Activation of muscarinic and nicotinic acetylcholine receptors in the nucleus accumbens core is necessary for the acquisition of drug reinforcement. J. Neurosci.26, 6004–6010 (2006). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR35] 35.Yee, J. et al. Muscarinic acetylcholine receptors in the nucleus accumbens core and shell contribute to cocaine priming-induced reinstatement of drug seeking. Eur. J. Pharmacol.650, 596–604 (2011). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR36] 36.Nunes, E. J., Kebede, N., Bagdas, D. & Addy, N. A. Cholinergic and dopaminergic-mediated motivated behavior in healthy states and in substance use and mood disorders. J. Exp. Anal. Behav.117, 404–419 (2022). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR37] 37.Morris, G., Arkadir, D., Nevet, A., Vaadia, E. & Bergman, H. Coincident but distinct messages of midbrain dopamine and striatal tonically active neurons. Neuron43, 133–143 (2004). [DOI] [PubMed] [Google Scholar]

[CR38] 38.Duhne, M., Mohebi, A., Kim, K., Pelattini, L. & Berke, J. D. A mismatch between striatal cholinergic pauses and dopaminergic reward prediction errors. Proc. Natl Acad. Sci. USA121, e2410828121 (2024). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR39] 39.Aosaki, T. et al. Responses of tonically active neurons in the primate’s striatum undergo systematic changes during behavioral sensorimotor conditioning. J. Neurosci.14, 3969–3984 (1994). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR40] 40.Brimblecombe, K. R. et al. Targeted activation of cholinergic interneurons accounts for the modulation of dopamine by striatal nicotinic receptors. eNeuro5, ENEURO.0397-17 (2018). [Google Scholar]

[CR41] 41.Lydall, E. S., Gilmour, G. & Dwyer, D. M. Rats place greater value on rewards produced by high effort: an animal analogue of the “effort justification” effect. J. Exp. Soc. Psychol.46, 1134–1137 (2010). [Google Scholar]

[CR42] 42.Johnson, A. W. & Gallagher, M. Greater effort boosts the affective taste properties of food. Proc. Biol. Sci.278, 1450–1456 (2011). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR43] 43.Clement, T. S., Feltus, J. R., Kaiser, D. H. & Zentall, T. R. Work ethic’ in pigeons: reward value is directly related to the effort or time required to obtain the reward. Psychon. Bull. Rev.7, 100–106 (2000). [DOI] [PubMed] [Google Scholar]

[CR44] 44.Pompilio, L., Kacelnik, A. & Behmer, S. T. State-dependent learned valuation drives choice in an invertebrate. Science311, 1613–1615 (2006). [DOI] [PubMed] [Google Scholar]

[CR45] 45.Maskos, U. et al. Nicotine reinforcement and cognition restored by targeted expression of nicotinic receptors. Nature436, 103–107 (2005). [DOI] [PubMed] [Google Scholar]

[CR46] 46.Bijoch, Ł et al. Whole-brain tracking of cocaine and sugar rewards processing. Transl. Psychiatry13, 20 (2023). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR47] 47.Koranda, J. L. et al. Nicotinic receptors regulate the dynamic range of dopamine release in vivo. J. Neurophysiol.111, 103–111 (2014). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR48] 48.Collins, A. L. et al. Nucleus accumbens cholinergic interneurons oppose cue-motivated behavior. Biol. Psychiatry86, 388–396 (2019). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR49] 49.Franklin, N. T. & Frank, M. J. A cholinergic feedback circuit to regulate striatal population uncertainty and optimize reinforcement learning. eLife4, e12029 (2015). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR50] 50.Cox, J. & Witten, I. B. Striatal circuits for reward learning and decision-making. Nat. Rev. Neurosci.20, 482–494 (2019). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR51] 51.Collins, A. L., Aitken, T. J., Greenfield, V. Y., Ostlund, S. B. & Wassum, K. M. Nucleus accumbens acetylcholine receptors modulate dopamine and motivation. Neuropsychopharmacology41, 2830–2838 (2016). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR52] 52.Zhang, Y.-F. et al. An axonal brake on striatal dopamine output by cholinergic interneurons. Nat. Neurosci.28, 783–794 (2025). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR53] 53.Costa, K. M. et al. Dopamine and acetylcholine correlations in the nucleus accumbens depend on behavioral task states. Curr. Biol.35, 1400–1407.e3 (2025). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR54] 54.Klug, J. R. et al. Differential inputs to striatal cholinergic and parvalbumin interneurons imply functional distinctions. eLife7, e35657 (2018). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR55] 55.Ratna, D. D. & Francis, T. C. Extrinsic and intrinsic control of striatal cholinergic interneuron activity. Front. Mol. Neurosci.18, 1528419 (2025). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR56] 56.Guo, Q. et al. Whole-brain mapping of inputs to projection neurons and cholinergic interneurons in the dorsal striatum. PLoS ONE10, e0123381 (2015). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR57] 57.Ding, J. B., Guzman, J. N., Peterson, J. D., Goldberg, J. A. & Surmeier, D. J. Thalamic gating of corticostriatal signaling by cholinergic interneurons. Neuron67, 294–307 (2010). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR58] 58.Gallo, E. F. et al. Dopamine D2 receptors modulate the cholinergic pause and inhibitory learning. Mol. Psychiatry27, 1502–1514 (2022). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR59] 59.Brown, M. T. C. et al. Ventral tegmental area GABA projections pause accumbal cholinergic interneurons to enhance associative learning. Nature492, 452–456 (2012). [DOI] [PubMed] [Google Scholar]

[CR60] 60.Al-Hasani, R. et al. Ventral tegmental area GABAergic inhibition of cholinergic interneurons in the ventral nucleus accumbens shell promotes reward reinforcement. Nat. Neurosci.24, 1414–1428 (2021). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR61] 61.Eshel, N. et al. Arithmetic and local circuitry underlying dopamine prediction errors. Nature525, 243–246 (2015). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR62] 62.Schulz, J. M. & Reynolds, J. N. J. Pause and rebound: sensory control of cholinergic signaling in the striatum. Trends Neurosci.36, 41–50 (2013). [DOI] [PubMed] [Google Scholar]

[CR63] 63.Pomrenze, M. B. et al. Modulation of 5-HT release by dynorphin mediates social deficits during opioid withdrawal. Neuron110, 4125–4143.e6 (2022). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR64] 64.Akerboom, J. et al. Genetically encoded calcium indicators for multi-color neural activity imaging and combination with optogenetics. Front. Mol. Neurosci.6, 2 (2013). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR65] 65.Dana, H. et al. Sensitive red protein calcium indicators for imaging neural activity. eLife5, e12727 (2016). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR66] 66.Wu, J. et al. Improved orange and red Ca²⁺ indicators and photophysical considerations for optogenetic applications. ACS Chem. Neurosci.4, 963–972 (2013). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR67] 67.Bruno, C. A. et al. pMAT: an open-source software suite for the analysis of fiber photometry data. Pharmacol. Biochem. Behav.201, 173093 (2021). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR68] 68.Sych, Y., Chernysheva, M., Sumanovski, L. T. & Helmchen, F. High-density multi-fiber photometry for studying large-scale brain circuit dynamics. Nat. Methods16, 553–560 (2019). [DOI] [PubMed] [Google Scholar]

[CR69] 69.Higginbotham, J. A. et al. Estradiol protects against pain-facilitated fentanyl use via suppression of opioid-evoked dopamine activity in males. Neuron113, 1413–1429.e5 (2025). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR70] 70.Blanco-Pozo, M., Akam, T. & Walton, M. E. Dopamine-independent effect of rewards on choices through hidden-state inference. Nat. Neurosci.27, 286–297 (2024). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR71] 71.Elum, J. E. et al. Distinct dynamics and intrinsic properties in ventral tegmental area populations mediate reward association and motivation. Cell Rep.43, 114668 (2024). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR72] 72.Engelhard, B. et al. Specialized coding of sensory, motor and cognitive variables in VTA dopamine neurons. Nature570, 509–513 (2019). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR73] 73.Parker, N. F. et al. Choice-selective sequences dominate in cortical relative to thalamic inputs to NAc to support reinforcement learning. Cell Rep.39, 110756 (2022). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR74] 74.Parker, N. F. et al. Reward and choice encoding in terminals of midbrain dopamine neurons depends on striatal target. Nat. Neurosci.19, 845–854 (2016). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR75] 75.Yao, Z. et al. A high-resolution transcriptomic and spatial atlas of cell types in the whole mouse brain. Nature624, 317–332 (2023). [DOI] [PMC free article] [PubMed] [Google Scholar]

PERMALINK

Cholinergic modulation of dopamine release drives effortful behaviour

Gavin C Touponse

Matthew B Pomrenze

Teema Yassine

Nicholas Denomme

May Wang

Viraj Mehta

Zihui Zhang

Robert C Malenka

Neir Eshel

Abstract

Main

Reward-evoked DA release encodes expended effort

Fig. 1. Reward-evoked DA release encodes expended effort.

Extended Data Fig. 1.

Extended Data Fig. 2. Inter-trial interval does not account for increased DA release after high effort expenditure.

Effort encoding by NAc DA is cell body-independent

Fig. 2. Effort encoding does not require VTA DA cell body activity.

Extended Data Fig. 3. Additional analysis of VTA DA cell body activity.

Extended Data Fig. 4. Validation of muscimol silencing of VTA DA cell body activity.

Extended Data Fig. 5. Validation of optogenetic silencing of VTA DA cell body activity.

Nicotinic signalling controls effort-related DA release

Fig. 3. NAc nicotinic receptor signalling gates DA release during high effort.

Extended Data Fig. 6. Effects of neuromodulator blockade on DA encoding of effort.

Extended Data Fig. 7. Analysis of single-cell RNA seq data highlighting α4 and α6 expression across relevant brain regions and cell-types.

Extended Data Fig. 8. Additional electrophysiology data from DA axon recordings.

NAc ACh release encodes expended effort

Fig. 4. ACh release encodes expended effort.

Extended Data Fig. 9. Inter-trial interval and task structure do not account for increased ACh release after high effort expenditure.

Extended Data Fig. 10. NAc cholinergic interneuron activity encodes effort and closely resembles ACh release.

Acetylcholine regulates DA effort encoding and behaviour

Fig. 5. Local cholinergic signalling is essential for DA effort encoding and effortful behaviour.

Extended Data Fig. 11. Behavioral controls for ACh manipulations.

Discussion

Methods

Subjects

Behavioural training

Sucrose effort task

Modified sucrose effort task

ChRmine effort task

Stereotaxic injections and viruses

Fibre photometry

Optogenetic manipulations

Drug administration

Validation of VTA DA cell body inhibition

Slice electrophysiology

Immunohistochemistry

Data analysis

Kernel regression analysis

Single-cell RNA sequencing analysis

Statistics and reproducibility

Reporting summary

Online content

Supplementary information

Source data

Acknowledgements

Extended data figures and tables

Author contributions

Peer review

Peer review information

Data availability

Code availability

Competing interests

Footnotes

Extended data

Supplementary information

References

Associated Data

Supplementary Materials

Data Availability Statement

ACTIONS

PERMALINK

RESOURCES

Similar articles

Cited by other articles

Links to NCBI Databases