(C) PLOS One
This story was originally published by PLOS One and is unaltered.
 . . . . . . . . . .



Dopamine and acetylcholine have distinct roles in delay- and effort-based decision-making in humans [1]

['Mani Erfanian Abdoust', 'Biological Psychology Of Decision Making', 'Institute Of Experimental Psychology', 'Heinrich Heine University Düsseldorf', 'Düsseldorf', 'Monja Isabel Froböse', 'Alfons Schnitzler', 'Institute Of Clinical Neuroscience', 'Medical Psychology', 'Medical Faculty']

Date: 2024-07 

In everyday life, we encounter situations that require tradeoffs between potential rewards and associated costs, such as time and (physical) effort. The literature indicates a prominent role for dopamine in discounting of both delay and effort, with mixed findings for delay discounting in humans. Moreover, the reciprocal antagonistic interaction between dopaminergic and cholinergic transmission in the striatum suggests a potential opponent role of acetylcholine in these processes. We found opposing effects of dopamine D2 (haloperidol) and acetylcholine M1 receptor (biperiden) antagonism on specific components of effort-based decision-making in healthy humans: haloperidol decreased, whereas biperiden increased the willingness to exert physical effort. In contrast, delay discounting was reduced under haloperidol, but not affected by biperiden. Together, our data suggest that dopamine, acting at D2 receptors, modulates both effort and delay discounting, while acetylcholine, acting at M1 receptors, appears to exert a more specific influence on effort discounting only.

To our knowledge, there has been no study so far that tested the impact of dopaminergic and cholinergic manipulations on both aspects of cost-benefit decision-making in 1 single experiment. To fill this gap, we investigated the effects of 2 drugs, haloperidol and biperiden, that selectively block either dopamine D2-like or muscarinic M1 acetylcholine receptors in human participants performing 2 decision-making tasks involving effort- and delay-based decisions. The goal of our study was 3-fold. First, we aimed to conceptually replicate the finding that dopamine D2 antagonists increase discounting of physical effort. Second, we aimed to assess the contribution of acetylcholine to cost-benefit decision-making and conceptually contrast it with the effects of dopamine. In doing so, we investigated whether any effects of muscarinic or dopaminergic receptor antagonism would modulate the computation of both time and effort costs or have a more specific effect on one cost dimension. Third, we sought to contribute new evidence to the thus far conflicting literature on the role of dopamine in delay-based decision-making using a large sample size and a within-subjects design. Based on previous findings, we hypothesised that the administration of haloperidol will (1) increase effort discounting; and (2) reduce delay discounting. In contrast, we expected opposite effects of biperiden, in particular (3) a decrease in effort discounting; and (4) an increase in delay discounting. In brief, we found opposing effects of haloperidol and biperiden only on specific components of effort-based choices. Specifically, haloperidol reduced the willingness to invest physical effort, whereas biperiden increased it. Results for delay discounting were less consistent. While haloperidol decreased delay discounting, there was no credible modulation by biperiden.

Striatal dopamine has been suggested to play a central role in both effort and delay discounting. For effort-based decision-making, studies in both humans and rodents indicate that diminished dopamine transmission makes individuals less willing to exert physical effort in exchange for larger rewards [ 3 – 9 ]. Pharmacological studies consistently demonstrate that agents that enhance dopaminergic transmission in humans increase reward sensitivity and decrease effort sensitivity, while studies directly investigating the effects of selective D2-receptor antagonism in healthy humans are rare [ 10 – 12 ]. For delay discounting, the existing literature is somewhat more mixed. Some studies suggest that increased dopamine transmission decreases delay discounting [ 13 – 15 ], whereas others have found either no effect [ 16 , 17 ] or an increase in delay discounting [ 18 ]. Notably, more recent evidence in humans indicates decreased delay discounting after blockade of D2 receptors [ 19 , 20 ]. While dopamine has received much attention in cost-benefit decision-making, other neuromodulators may also play an important role [ 21 – 23 ]. One modulator that has not received much attention in this regard is acetylcholine, which is surprising given the literature on reciprocal antagonistic interactions between acetylcholine and dopamine signalling in the striatum [ 24 – 28 ]. This suggests that pharmacological blockade of M1 receptors will have effects on cost-benefit decision-making that are opposite to those of blocking dopamine D2 receptors. In line with this, animal studies indicate that muscarinic agonists induce behavioural changes in effort-based choices that are similar to those produced by dopamine antagonists [ 23 , 29 ], underscoring the potential interplay between dopamine and acetylcholine in modulating cost-benefit decision-making. Furthermore, research in animal models points to a role for acetylcholine in modulating delay-based choices through both muscarinic and nicotinic receptors, albeit with inconsistent findings [ 30 – 33 ]. Importantly, studies directly investigating these effects in human decision-making are lacking.

Consider the following scenarios: Would you prefer to order an average pizza that can be delivered within minutes or rather wait an extra hour for your most favourite pizza? Similarly, would you visit a highly rated pizza place that requires climbing up a steep hill or opting for a conveniently accessible pizza place right in front of the next bus stop? In scenarios like this, decision-making involves the balancing of potential rewards against distinct costs required to obtain them. The tendency to devalue rewards as a function of effort or delay costs are commonly described as effort and delay discounting, respectively [ 1 , 2 ].

Lastly, as an exploratory analysis, we asked whether there was a relationship between discounting of delays versus efforts across individuals. In other words, we tested whether people showing steep discounting of physical efforts would also show steep delay-based discounting. In line with previous findings [ 36 , 38 ], we found no relationship between individuals’ tendency to discount rewards based on efforts compared to delays (no credible correlation between the effort and delay discounting parameters κ under placebo; r = −0.17, HDI 95% = [−0.06; 0.39]).

Taken together, the results derived from hierarchical Bayesian modelling are consistent with findings obtained from the GLMMs. Diminishing dopaminergic D2 receptor activity decreases the willingness to invest effort for rewards, possibly by amplifying effort discounting. Conversely, the administration of a cholinergic M1 receptor antagonist produces opposite effects. However, this pattern of opponent dopaminergic and cholinergic effects on specific components of effort discounting was not present in the delay discounting task. Instead, the findings suggest a reduced impact of delay on decision-making under haloperidol, while biperiden affected only the choice stochasticity. This implies that contrasting effects of dopaminergic and cholinergic manipulations may reflect computationally specific, rather than universal effects.

Lastly, to test whether putatively confounding side effects of the pharmacological treatment account for our main effects of interest, we tested medication effects on several control measures, including mood (alertness, calmness, and contentedness) and basic physiological parameters (heart rate, systolic, and diastolic blood pressure), using Bayesian linear mixed models. In short, biperiden administration induced reductions in systolic blood pressure and heart rate at T 1 and at T 2 . Moreover, relative to placebo, both biperiden and haloperidol caused decreases in subjective alertness ratings at T 2 (all HDI 95% < 0). T 1 represents the measurement taken before participants began the effort discounting task, and T 2 represents the measurement taken after they completed the delay discounting task. Importantly, neither the changes in blood pressure or heart rate, nor reductions in alertness showed credible correlations with any shift parameter that was credibly modulated by haloperidol or biperiden in both tasks. This implies that the drug-induced changes in behaviour were not linked to drug effects on alertness, heart rate, or blood pressure. More detailed information can be found in the Supporting Results in S1 Text .

For the delay discounting task, computational modelling revealed credible evidence only for a dopaminergic, but not cholinergic modulation of delay discounting ( Fig 4D and Table 2 ). In line with the diminished delay sensitivity reported above, haloperidol reduced delay discounting, making participants more willing to wait for greater financial rewards with increasing levels of delay (s κ HAL : HDI Mean = −0.630, HDI 95% = [−1.083; −0.237]). For biperiden, we did not observe any shift of the discounting parameter (s κ BIP : HDI Mean = −0.125, HDI 95% = [−0.410; 0.148]). Next, we examined the effects of dopaminergic and muscarinic antagonism on choice stochasticity ( Fig 4E ). Biperiden credibly increased the choice stochasticity (s β BIP : HDI Mean = −0.043, HDI 95% = [−0.085; −0.000]), while we did not find credible evidence for a modulatory effect of haloperidol, as the 95% HDI overlapped with zero (s β HAL : HDI Mean = −0.055, HDI 95% = [−0.114; 0.006]).

Posterior distributions and changes in the SV from the hierarchical Bayesian models. ( a ) In the effort discounting task, the discounting parameter κ is modulated in opposite directions by the drugs, with haloperidol increasing and biperiden decreasing effort discounting. ( b ) Similarly, these opposite effects are also present in the modulation of the softmax inverse temperature β, reflecting choice stochasticity. Biperiden administration led to more deterministic choices, while haloperidol induced more stochastic choices. ( c ) Modelled discounting functions show steeper discounting under haloperidol and flatter discounting under biperiden. ( d ) In the delay discounting task, the discounting parameter κ is reduced by haloperidol, with no credible modulatory effect of biperiden. ( e ) In contrast, the softmax inverse temperature β is reduced by biperiden, indicating more stochastic choices. ( f ) Overall, participants showed flatter discounting of future rewards under haloperidol compared to placebo. In ( a ), ( b ), ( d ), and ( e ), bold (light) dots represent the group-level (participant-level) mean estimate. The horizontal bars represent the group-level 95% HDI. In ( c ) and ( f ), the SVs are displayed as a discount function of effort and delay. Parabolic (effort discounting task) and hyperbolic (delay discounting task) functions are fitted on group-level mean estimates for each drug (see Materials and methods ). HDI, highest density interval; SV, subjective value.

The results from the computational model of the effort discounting task align with and extend the regression-based results presented above. Specifically, again biperiden and haloperidol exerted opposing effects on both the discounting and inverse temperature parameter. Haloperidol increased effort discounting, while biperiden diminished it. Similarly, haloperidol induced more stochastic choices, while biperiden led to more deterministic decisions ( Table 1 ). For both drug-specific effects on the effort discounting parameter κ, we acknowledge that, strictly speaking, the 95% HDI of s κ BIP (HDI Mean = −0.012, HDI 95% = [−0.000; 0.027]) and s κ HAL (HDI Mean = 0.013, HDI 95% = [−0.025; 0.000]) did overlap with zero, albeit to a very small extent ( Fig 4A ). Notably, the density that did not overlap with zero still accounted for more than 94% of the posterior distribution (94.8% HDI > 0 for haloperidol and 94.4% HDI > 0 for biperiden). This provides strong evidence for a credible modulation of effort discounting by cholinergic M1 and dopaminergic D2 receptor manipulation, despite the slight overlap.

To obtain a more detailed understanding of the mechanism underlying the drug effects on behaviour described above, we used hierarchical Bayesian modelling. To begin with, we determined the best-fitting discounting models for both effort and delay, comparing 4 commonly used models (linear, parabolic, hyperbolic, and exponential). In line with previous findings [ 2 , 34 – 37 ], model comparisons revealed that effort discounting was best described by a parabolic model, suggesting a greater impact of changes in high rather than low-effort levels. In contrast, delay discounting was best described by a hyperbolic model, indicating a greater impact of changes in low rather than high-delay levels ( S3 Table ). For each model, we first calculated the SVs of all choice options, based on participant-specific weighing of reward magnitude and associated costs (i.e., effort and delay levels). By introducing condition-specific shift parameters, we captured potential drug effects on the effort and delay discounting parameter κ (denoted as s κ HAL for haloperidol and s κ BIP for biperiden), with positive/negative shift parameter values indicating increased/decreased effort and delay discounting, respectively. We then used a softmax function to transform the option values into choice probabilities, with choice stochasticity being modelled by an inverse temperature parameter β. Again, condition-specific shift parameters for haloperidol and biperiden captured potential drug effects on choice stochasticity (s β HAL for haloperidol and s β BIP for biperiden), with positive/negative shift parameters indicating more deterministic/more stochastic decision-making (see Materials and methods , and Supporting Materials and Methods in S1 Text ). Model validation and parameter recovery confirmed that both models accurately captured key features of the choice data (see Supporting Results in S1 Text ).

To summarise, regression-based results show that haloperidol reduced the overall propensity to choose the high-cost options in the effort domain. While haloperidol had no effect on the average rate of selecting the high-cost option in delay-based choices, it attenuated participants’ sensitivity to delay costs. In contrast, biperiden increased the overall propensity of selecting the high-cost option in the effort discounting task, opposite to the effect of haloperidol. Consistent with this, biperiden also increased sensitivity to reward magnitudes during effort-based choices.

Next, to investigate session and drug-order effects, we included session and the two-way interactions between session and drug as additional predictors. These analyses confirmed a main effect of session only for effort-based choices (HDI Mean = 0.49, HDI 95% = [0.13; 0.85]), suggesting increased preference for high-effort options in later sessions, likely reflecting task familiarity. Crucially, the interaction effects between either drug and session were not credible in both tasks ( S14 and S15 Tables). This indicates that neither dopaminergic nor cholinergic manipulations modulated learning effects across sessions. Moreover, this finding underscores that the observed drug-induced effects cannot be explained by drug order effects.

To test for possible fatigue effects, we added trial number as well as the two-way interaction with drug as additional predictors. This analysis revealed a credible main effect of trial number for both the effort and delay discounting task (trial number effect on effort discounting: HDI Mean = −0.60, HDI 95% = [−0.75; −0.46]; trial number effect on delay discounting: HDI Mean = −0.20, HDI 95% = [−0.33; −0.08]), suggesting that the tendency to choose high-cost options, irrespective of cost type, decreased over the course of the experiment. Notably, we found a credible interaction effect between trial number and haloperidol in the effort discounting task (HDI Mean = −0.43, HDI 95% = [−0.63; −0.23]), indicating that the fatigue effect was more pronounced under haloperidol compared to placebo. Importantly, even after accounting for these fatigue-related effects, the main effect of haloperidol on effort discounting remained credible (HDI Mean = −0.48, HDI 95% = [−0.90; −0.05]) ( S12 and S13 Tables).

In contrast, in the delay discounting task, both drugs had no effect on the average rate of choosing the high-cost option (placebo: 71.59% ± 2.25; haloperidol: 71.55% ± 1.73; biperiden: 70.79% ± 1.82; Fig 2A (right panel)), as indicated by the absence of credible main effects for haloperidol (HDI Mean = −0.046, HDI 95% = [−2.648; 1.884]) and biperiden (HDI Mean = −1.315, HDI 95% = [−4.147; 0.655]; Fig 3D ). Importantly, however, the analysis revealed a reduced sensitivity to delays under haloperidol, evidenced by a credible interaction effect between haloperidol and delay (HDI Mean = 1.332, HDI 95% = [0.328; 2.440]; Fig 3F ), while the other drug interactions did not reach credibility ( Fig 3 and S2 Table ). This indicates that diminished dopaminergic activity attenuates the impact of time costs on decision-making.

Posterior distributions and 95% HDI of the logistic Bayesian GLMMs depict the estimate of each effect on choosing the high-cost option. ( a ) Biperiden credibly increased the overall willingness to invest physical effort for a corresponding reward, while haloperidol had the opposite effect. ( b ) Biperiden increased reward sensitivity in the effort discounting task, as indicated by a biperiden-by-reward interaction, without a credible effect of haloperidol. ( c ) In contrast, neither drug affected effort sensitivity. ( d ) In the delay discounting task, the willingness to tolerate delays for rewards was not affected by either drug. ( e ) Likewise, neither drug modulated reward sensitivity in the delay discounting task. ( f ) However, haloperidol decreased delay sensitivity, with no credible effect of biperiden. Here, a positive estimate of the interaction effect between haloperidol and delay indicates a reduction of the negative parameter estimate associated with delay, suggesting an attenuation of the impact of delay on choice behaviour. Bold dots represent the mean group-level estimate of the posterior distribution. The horizontal bars represent the group-level 95% HDI. GLMM, generalized linear mixed model; HDI, highest density interval.

Notably, and in line with our predictions, we found that the 2 drugs had opposite effects on the tendency to choose high-effort options in the effort discounting task relative to placebo. Haloperidol reduced the willingness to invest higher effort for greater reward, while biperiden increased this willingness (placebo: 79.17% ± 1.76; haloperidol: 74.34% ± 1.85; biperiden: 81.99% ± 1.64; Fig 2A (left panel)). These effects were confirmed by credible main effects of both haloperidol (HDI Mean = −0.532, HDI 95% = [−0.943; −0.136]) and biperiden (HDI Mean = 0.620, HDI 95% = [0.230; 1.048]; Fig 3A ) on choosing the high-cost option relative to placebo. Moreover, this analysis revealed that biperiden increased reward sensitivity, as evidenced by a credible interaction effect between biperiden and reward (HDI Mean = 0.802, HDI 95% = [0.256; 1.409]; Fig 3B ). See S1 Table for the full results.

( a ) The overall proportions of high-cost choices were modulated by drug administration in the effort (left), but not in the delay discounting task (right). Biperiden increased and haloperidol decreased the willingness to invest physical effort in return for reward. ( b ) The probability of choosing the high-cost option increased as a function of increasing reward magnitude for both the effort and delay discounting tasks. In the effort discounting task (left), biperiden increased the impact of the reward level on choice. ( c ) Similarly, the tendency to choose the high-cost option decreased as a function of increasing levels of effort and delay. This effect is reduced by haloperidol in the delay discounting task (right). Values in ( a ) show group-level (single-subject) means represented by bold (light) dots. Values in ( b ) and ( c ) display averaged group-level means per reward and cost level, with error bars representing the standard error of the mean. In ( b ), reward levels are presented as the difference in magnitude between the high- and low-cost option in the effort discontinuing task and as the absolute reward value of the high-cost option in the delay discounting task. Likewise in ( c ), the effort level represents the difference between the proportions of the individually calibrated MVC of the high- versus low-cost option, while the delay levels indicate the delay of the high-cost option. The data underlying the effort discounting task (left panel) can be found in S1 Data , and the data underlying the delay discounting task (right panel) can be found in S2 Data . MVC, maximum voluntary contraction.

Participants consistently exhibited a preference for the high-cost option over the low-cost alternative in both tasks (effort discounting task: 78.50% ± 1.03; delay discounting task: 71.31% ± 1.12; Fig 2A ). For each of the 2 tasks, we used a logistic Bayesian generalized linear mixed model (GLMM) to investigate the impact of drug manipulation, changes in task parameters, and the interaction between both, on choosing the high-cost option. Specifically, to regress participants’ choices, we included the following fixed-effect predictors: drug condition, reward magnitude, cost level (i.e., effort or delay), and all possible interactions. In the effort discounting task, we used the difference terms (high-cost option minus low-cost option) for both reward and effort. For the delay discounting task, we used the absolute reward and delay levels of the varying high-cost option. To account for within-subject variability, we included random intercepts for each subject along with random slopes for all fixed-effect predictors (see Materials and methods ). We first confirmed that participants effectively discounted rewards based on costs and thereby adhered to the task requirements: As expected, in both discounting tasks higher reward magnitudes increased, while higher cost levels (effort or delay) decreased participants’ likelihood to select the high-cost option (reward effect on effort discounting: HDI Mean = 3.44, HDI 95% = [2.96; 3.95]; reward effect on delay discounting: HDI Mean = 55.64, HDI 95% = [41.76; 69.38], effort effect on effort discounting: HDI Mean = −1.64, HDI 95% = [−1.87; −1.40]; delay effect on delay discounting: HDI Mean = −2.29, HDI 95% = [−3.30; −1.25], Fig 2B and 2C ).

( a ) The experimental sessions were standardised across all sessions except for the drug treatment, which was counterbalanced. To account for the different times of haloperidol and biperiden to reach peak plasma levels, a dummy drug application was introduced, ensuring that the maximum concentration of both drugs was aligned to task execution. Approximately 180 min after the first capsule and approximately 60 min after the second capsule, participants engaged in the effort discounting task (b), followed by the delay discounting task (c). Physiological measures (including heart rate and blood pressure) and mood ratings (using the Bond and Lader Visual Analogue Scales) were collected at 3 distinct time points. Additionally, participants completed the trail-making test part A before task execution (see Materials and methods for more details). In both tasks, participants were presented with 2 alternative options, each providing information about a monetary reward in return for specific costs. ( b ) Effort discounting task. One option required less effort (indicated by the horizontal yellow line) and provided a smaller reward (indicated by the number of apples, low-cost option), while the other option required more effort and provided a higher reward (high-cost option). Participants then chose 1 option and exerted the required effort (adjusted to the MVC) for at least 1 s. ( c ) Delay discounting task. Similarly, participants were presented with 2 offers: a smaller but immediately available reward (low-cost option) or a larger reward available after the delay indicated (high-cost option). MVC, maximum voluntary contraction.

The 62 healthy participants performed 2 cost-benefit decision-making tasks aimed to quantify the extent to which the subjective value (SV) of a monetary reward is discounted as a function of either effort or delay costs. Participants completed both tasks during 3 sessions under the influence of either the D2 receptor antagonist haloperidol (2 mg), the M1 acetylcholine receptor antagonist biperiden (4 mg), or a placebo in a within-subjects design ( Fig 1A ). The effort-based decision-making task (effort discounting task, Fig 1B ) involved choices between options varying in reward magnitude and effort requirement (using handgrip force), while the delay-based decision-making task (delay discounting task, Fig 1C ) involved choices between one option varying in reward magnitude and delay versus a fixed one. Thus, both tasks required participants to choose between a high-reward/high-cost (high-cost option) and a low-reward/low-cost (low-cost option) alternative.

Discussion

In this study, we investigated the effects of pharmacologically manipulating dopaminergic and cholinergic neurotransmission on cost-benefit decision-making in healthy young adults. Specifically, we administered haloperidol or biperiden, 2 drugs that selectively block either dopamine D2-like or muscarinic M1 acetylcholine receptors, respectively, and tested the effects on discounting a monetary reward as a function of either effort or delay. In short, we found that reducing dopaminergic transmission at D2-like receptors decreased participants’ willingness to invest physical effort for monetary rewards and attenuated the impact of delay on decision-making. In contrast, cholinergic M1 receptor manipulations promoted the preference for high-effort options, without affecting delay-based choices.

The goal of our study was 3-fold. First, we sought to conceptually replicate the boosting effects of dopamine on motivation. Second, we examined the contribution of acetylcholine, acting at M1 receptors, on cost-benefit decision-making, as the influence of muscarinic acetylcholine receptors on human (value-based) decision-making has rarely been reported. Third, we aimed to address the conflicting literature on dopamine’s role in delay-based decision-making by providing data from a large sample using a within-subjects design.

Dopamine has been widely suggested to play a key role in various aspects of reward processing and cost-benefit decision-making [39–41]. It modulates choices based on expected rewards and costs associated with different options [42–44]. However, the direction of dopaminergic manipulations on different aspects of costs remains ambiguous. Traditionally, dopamine has been implicated in promoting behaviour that maximises reward outcomes, suggesting that increased dopaminergic activity energises behaviour to approach high-reward options despite associated costs [7,45–47]. According to this view, increasing dopaminergic transmission should bias decision-making toward high-reward options, even when they involve higher costs such as effort, delay, and risk. Conversely, decreasing dopaminergic transmission should have the opposite effect.

In line with this view, in our study, the administration of a D2-like receptor antagonist did in fact reduce the willingness to invest physical effort for reward. However, contrary to the notion that dopamine supports reward-maximising behaviour, our findings in the delay discounting task did not confirm the expected pattern. We observed a decrease in delay discounting under haloperidol, rather than an increase. While there is indeed a substantial body of evidence suggesting enhancing effects of dopamine on motivation, as demonstrated by studies highlighting its involvement in promoting reward-seeking behaviour [8,9,48,49], the impact of pharmacological manipulations targeting other cost-related factors, such as delay [13,15,17,50] and risk [51–53], has produced rather inconsistent findings. These studies showed either contradicting results regarding dopaminergic manipulations or no effect at all. These inconsistencies challenge traditional views of dopamine and raise questions about the specific mechanisms through which dopamine modulates choices that incorporate time-dependent costs. A more recent theory proposes that dopamine, acting at D2-like receptors, biases action selection by increasing the preference for options with a proximity advantage over more distant alternatives [54,55]. This theory is based on studies in rodents, where firing rates of neurons in the nucleus accumbens are increased by spatially proximal rewards, promoting a decision bias towards nearby low-reward options [56]. This notion of proximity, originally referring to spatial proximity, has been extended to the context of psychological proximity [55]. According to this view, dopamine not only favours options that are physically closer in space, but also options that are psychologically closer, as in our case, rewards that are available sooner in time. Consequently, the administration of D2 receptor antagonists, which, according to this theory, are believed to reduce the proximity bias, have been shown in some studies to increase the preference for delayed and risky reward options, which lack the proximity advantage [19,20,51,57]. In other words, this theoretical framework proposes that dopamine may play a dual role in (1) promoting choices towards options with a psychological proximity advantage; and (2) weighing reward magnitudes against associated costs. Our findings of decreased delay discounting under haloperidol align with this theory, potentially suggesting a diminished preference for more proximal options when D2-receptor activity is reduced, resulting in increased endurance for higher (time) costs. On the other hand, the reduction of dopaminergic neurotransmission had the opposite effect on effort discounting, leading to a diminished willingness to bear higher (effort) costs. This apparently conflicting finding between effort and time costs may be explained by the fact that neither the low-effort nor the high-effort option had a pronounced psychological proximity advantage. Thus, in this specific context of cost-benefit decision-making, the proximity factor appears to be less relevant, and dopamine instead may promote behaviour aimed at maximising rewards simply by weighing rewards against associated costs. Importantly, our experimental paradigm did not directly manipulate psychological proximity, nor did it distinguish between proximity effects and the impact of delay. Therefore, specific effects of dopamine on psychological proximity in human delay discounting remain highly speculative and warrant further investigation with experimental paradigms specifically tailored to this question.

Low doses of dopamine receptor antagonists may facilitate dopaminergic transmission by primarily blocking presynaptic autoreceptors, which may increase rather than decrease dopamine release [58,59]. We would however argue that such a presynaptic mechanism of action is unlikely to explain our findings. First, a PET study found that the same dose of haloperidol as used here led to high levels of D2 receptor occupancy in the striatum [60]. Second, our pattern of results aligns with studies in rodents showing motivational deficits following ventral striatal dopamine depletion [61–63]. Third, high doses of haloperidol have been shown to reduce alertness [64], whereas drugs like methylphenidate that increase synaptic dopamine levels, enhance subjective ratings of alertness [65,66]. Consistent with this, we observed a reduction in alertness ratings following haloperidol administration. Together, this suggests that our results are best explained by a blockade of postsynaptic D2 receptors by haloperidol.

Notably, prior research has highlighted the contribution of other neurotransmitters such as serotonin [22,67], adenosine [68,69], and acetylcholine [23,29,70], in the weighing of costs and rewards. However, the exact role of these neurotransmitters in reward processing and cost-benefit decision-making in humans has been rarely investigated. Understanding the involvement of acetylcholine, due to its reciprocal activity with dopamine in the striatum, is of particular interest [24,26,27,71]. At the functional level, these mutual interactions are evident from the fact that muscarinic receptor antagonist diminish the extrapyramidal side effects of dopamine antagonists, and, vice versa, muscarinic agonists display antipsychotic properties, resembling the action of D2 antagonists [72,73]. Additionally, blocking M1 receptors has been shown to reverse motivational impairments induced by dopaminergic antagonism, further emphasising its potential role in modulating dopamine-related processes [29]. At the cellular level, it has been shown that M1 receptor activation inhibits D2-receptor–mediated effects in the striatum [74]. These observations highlight an interplay between acetylcholine and dopamine signalling. Therefore, in addition to testing dopaminergic D2-like receptor manipulations, we also investigated the role of cholinergic M1 receptor neurotransmission on decision-making.

Indeed, we observed opposing effects of dopaminergic and cholinergic manipulations within specific components of effort-, but not delay-based decision-making. Specifically, we found evidence for biperiden increasing the willingness to invest effort for rewards and, in line with this, decreasing effort discounting. In contrast, haloperidol reduced the general willingness to invest effort and increased effort discounting. However, we also observed drug effects that were not in opposite directions between biperiden and haloperidol. For example, biperiden increased reward sensitivity, while haloperidol had no credible effect on the impact of rewards on effort-based choices. Furthermore, haloperidol modulated decision times in both experimental tasks, whereas biperiden did not affect them. These findings suggest partially opposing effects between both neurotransmitters, mostly evident within specific components of effort-based choices.

This partially opposing mechanistic relationship between dopaminergic and cholinergic neurotransmission during the effort discounting task is further reflected in changes in choice stochasticity. Previous studies have linked D2 receptor antagonism with increased choice stochasticity [75,76]. Consistent with these findings, our results revealed that haloperidol administration increased stochasticity and thereby reduced value dependency on choices, while biperiden had the opposite effect. After lowering cholinergic M1 receptor activity, participants were more likely to choose the option with the highest expected value compared to placebo. However, it is important to note that biperiden had the opposite effect in the delay discounting task, increasing, rather than reducing choice stochasticity. This discrepancy indicates that the role of cholinergic neurotransmission in balancing deterministic versus stochastic behaviour is more complex and needs further investigation.

Some limitations should be noted. First, our focus was primarily on striatal D2 receptor activity, as haloperidol predominantly targets D2 receptors in the striatum [77]. However, it is important to note that in the context of cost-benefit decision-making, dissociable roles of D1 versus D2 receptor activity have been reported [41,54,78], and thus the general role of dopamine beyond its selective activity on striatal D2 receptors remains unclear. Conversely, biperiden primarily targets M1 receptors in the cortex and striatum [79], making it challenging to determine the precise mechanisms underlying cortical and striatal cholinergic modulation and the reciprocal effects on dopaminergic activity. Second, previous research has suggested a U-shaped dose-response function for dopamine, indicating deleterious effects of both extremely high and extremely low levels of dopamine [80,81]. According to this idea, the same dopamine agent can produce opposing effects in different individuals. Therefore, it may be insightful to consider individual differences in baseline dopamine levels when studying the effects of dopaminergic manipulations. A recent study found evidence for the absence of a correlation between dopamine synthesis capacity and putative behavioural proxies of dopamine, such as working memory or trait impulsivity [82]. Consequently, investigating baseline dopamine levels require more costly and invasive techniques, such as positron emission tomography. Third, as participants were required to exert physical effort on each trial, fatigue effects could have developed in the effort discounting task. While our additional analysis confirmed a general fatigue effect and a dual role of haloperidol in both reducing the overall propensity to invest effort and in exacerbating the fatigue effect throughout the task, we acknowledge that recent studies revealed the existence of distinct states of fatigue [83,84]. Importantly, this was discovered by using task paradigms and computational models that were designed to distinguish between these different types of fatigue. Additionally, it is important to consider that potential motor effects, particularly in the context of dopaminergic manipulations, could also affect effort discounting behaviour. Future studies could extend our approach by incorporating measurements of force pulses to investigate potential motor-related effects and apply task designs and computational models specifically tailored to capture how drug manipulations might affect different forms of fatigue. Lastly, it is important to note that several factors preclude a direct quantitative comparison between the delay and effort discounting tasks. These include the distinct nature of rewards and costs (hypothetical versus real), varying reward magnitudes in both tasks (large versus small), and differences in task structure (fixed versus variable alternative option). Additionally, both tasks were consistently performed in a fixed order within and between participants (effort followed by delay). While this ensured consistent drug levels within each task, it is potentially introducing task order effects and possibly leading to differential drug concentrations between tasks. These methodological differences limit our ability to directly compare drug effects across the 2 cost domains. Future research would benefit from applying experimental paradigms that manipulate both delay and effort costs within the same task, allowing a more controlled and direct comparison of how pharmacological manipulations differentially influence sensitivity to these distinct cost types.

In conclusion, our findings support prior research indicating an invigorating effect of dopamine on motivation in an effort-based decision-making task. Moreover, our study contributed to understanding dopamine’s involvement in temporal cost-benefit tradeoffs by revealing decreased delay discounting following dopaminergic D2-like receptor antagonism. Further, we demonstrate that the administration of biperiden, a muscarinic M1 receptor antagonist, had contrasting effects to those of dopaminergic D2 receptor antagonism in the general willingness to choose high-cost options and the effort discounting parameter. This suggests that in the context of human cost-benefit decision-making, the previously reported reciprocal relationship between both neurotransmitters may be limited only to specific components of behaviour. Our findings indicate that, while D2 receptor activity plays a role in integrating both delay and effort costs, acetylcholine, acting at M1 receptors, may have a more specific role in effort processing.

[END]
---
[1] Url: https://journals.plos.org/plosbiology/article?id=10.1371/journal.pbio.3002714

Published and (C) by PLOS One
Content appears here under this condition or license: Creative Commons - Attribution BY 4.0.

via Magical.Fish Gopher News Feeds:
gopher://magical.fish/1/feeds/news/plosone/