Advertisement
Browse Subject Areas
?

Click through the PLOS taxonomy to find articles in your field.

For more information about PLOS Subject Areas, click here.

  • Loading metrics

Open Access

Peer-reviewed

Research Article

Children and adults differ in how primary and secondary incentives modulate valuation, effort, and cognitive control

  • Sebastijan Veselic,

    Roles Conceptualization, Data curation, Formal analysis, Investigation, Methodology, Project administration, Software, Visualization, Writing – original draft, Writing – review & editing

    Affiliation Department of Experimental Psychology, University of Oxford, United Kingdom

    ⨯
  • Claire Rosalie Smid,

    Roles Data curation, Investigation, Writing – original draft, Writing – review & editing

    Affiliation Division of Psychology and Language Sciences, University College London, United Kingdom

    ⨯
  • Francis Beveridge,

    Roles Data curation, Investigation, Project administration, Resources, Writing – review & editing

    Affiliation Division of Psychology and Language Sciences, University College London, United Kingdom

    ⨯
  • Nikolaus Steinbeis

    Roles Conceptualization, Funding acquisition, Project administration, Supervision, Writing – original draft, Writing – review & editing

    n.steinbeis@ucl.ac.uk

    Affiliation Division of Psychology and Language Sciences, University College London, United Kingdom

    ⨯

Abstract

Rewards have a profound impact on human motivation, cognition, affect, and behaviour. The study of reward processing and incentive effects therefore occupies a central place in psychology and cognitive neuroscience. One common assumption when comparing groups or individuals is that different reward types are valued similarly. Here we examined this assumption in a sample of 51 adults and 39 children (7–12 years) using both primary and secondary rewards. Across three tasks – subjective valuation, willingness to exert cognitive effort, and reward-related modulation of cognitive control – adults showed stronger effects of secondary relative to primary reinforcers, whereas children showed comparatively similar responses across reward types. While we interpret our findings as consistent with age-group differences in the value assigned to secondary reinforcers, larger longitudinal studies using more closely matched incentives will be required to determine how such differences emerge across development. More broadly, our work highlights the importance of carefully considering incentive value when comparing different groups on reward-related processes.

Citation: Veselic S, Smid CR, Beveridge F, Steinbeis N (2026) Children and adults differ in how primary and secondary incentives modulate valuation, effort, and cognitive control. PLoS One 21(6): e0351143. https://doi.org/10.1371/journal.pone.0351143

Editor: Poppy Watson, University of Technology Sydney, AUSTRALIA

Received: October 18, 2025; Accepted: May 24, 2026; Published: June 15, 2026

Copyright: © 2026 Veselic et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.

Data Availability: All data and analysis code required to reproduce the findings reported in this study are available at OSF: https://osf.io/8bgeu/. DOI: https://doi.org/10.17605/OSF.IO/8BGEU.

Funding: SV was supported by the Leverhulme Award DS-2017-026. CRS was supported by the European Research Council (European Research Council (ERC) grant agreement no. 715282, project DEVBRAINTRAIN). NS was supported by the European Research Council (European Research Council (ERC) grant agreement no. 715282, project DEVBRAINTRAIN) and a Jacobs Foundation Research Fellowship. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.

Competing interests: The authors have declared that no competing interests exist.

Introduction

Like many other species, humans are highly sensitive to rewards [1,2]. One core function of rewards is to energize and motivate actions directed towards goals. Rewards therefore play a central role in shaping motivation, cognition, affect, and behaviour, influencing processes as diverse as learning, decision-making, and cognitive control [2–4]. A common assumption in studies of reward processing is that the subjective value of the reward under study is comparable across individuals or groups. This assumption is particularly important in developmental psychology, where many studies implicitly assume that the same incentives are similarly valued across age groups, such that group differences are attributed to changes in the underlying processes of interest rather than to differences in the subjective value assigned to the rewards themselves [5,6]. Here, we compared children and adults to examine whether age-group differences in reward-guided behaviour varied across primary versus secondary reinforcers, using convergent measures spanning explicit ratings, effort allocation, and motivated cognitive control.

Age-related differences in reward-guided behaviour have been documented extensively, including differences in reward sensitivity [7], risk-taking [8,9], intertemporal choice [10–12], prosocial behaviour [13–15], model-based decision-making [16], reward exploration [17], and cognitive control [18]. These behavioural differences are often linked to maturational changes in reward-related neural systems, including developmental changes in dopamine receptor density in striatum and prefrontal cortex [19–22], changes in resting-state connectivity of striatal regions [23], and changes in corticostriatal connectivity from childhood into adulthood [24–28]. For example, adolescents show stronger ventral striatal activation than adults during reward processing even when behaviour is matched across groups [29]. At the same time, recent work suggests that adolescents and adults may differ less in the value they assign to money than in how strategically they allocate effort to obtain it [30]. Although there is little doubt that reward processing changes with age [31,32], some observed age effects may partly reflect the particular incentives used, especially when studies rely heavily on secondary reinforcers such as money; for a few exceptions see [33–35].

The value of secondary reinforcers such as money is acquired through experience [36,37]. It is therefore plausible that their value changes with age. Alternatively, their novelty may make them particularly attractive to younger participants [38]. Under either account, secondary reinforcers may not be valued uniformly across development, which complicates the interpretation of age-group differences in reward-guided behaviour. This is especially relevant given that the goals humans pursue change with development [39], which may shape the types of rewards they value and the degree of value assigned to them [5]. At the same time, preferences for primary reinforcers such as pleasant tastes are also not fixed across development, but are shaped by maturation and exposure [40–42; see also 35]. Thus, if the value of primary and secondary reinforcers changes differently across age groups, then reward-guided behaviour may differ not only because of changes in reward processing itself, but also because of differences in the subjective value of the incentives used.

Previous work investigating similar questions [29,30] focused on adolescents. Here, we focused on middle childhood (7–12 years) because this period is marked by substantial development in executive function, rule use, and motivated behaviour, while also corresponding to increasing everyday exposure to symbolic and other secondary reinforcers, but preceding the broader social and pubertal changes associated with adolescence [5,8,34].

To test whether age-group differences in subjective value and reward-guided behaviour depend on reward type, we directly compared primary reinforcers (pleasant tastes) and age-appropriate secondary reinforcers (money for adults; monetary points exchangeable for prizes for children) [47]. We assessed three aspects of reward-guided behaviour: explicit valuation, willingness to exert cognitive effort, and reward-related modulation of cognitive control. In the present study, we use valuation in a pragmatic sense, indexing it through explicit preference reports, willingness to exert cognitive effort, and reward-related modulation of cognitive control. Across the three paradigms, adults showed stronger effects of secondary relative to primary reinforcers, whereas children showed more similar responses across reward types.

Materials and methods

Participants

Fifty-one adults (22 male; mean age = 24.61 years, range = 18.7–35.3 years) were recruited through university listings using a subject pool (SONA) with an advertisement indicating that the experiment involved several cognitive tasks performed for rewards. The adult sample primarily comprised undergraduate and graduate students. Thirty-nine children (21 male; mean age = 9.63 years, range = 7.1–12.0 years) were recruited from schools in the Greater London area using an existing lab database and school/community outreach. Parents were initially called and asked whether they would like their child to participate in an experiment with cognitive tasks for rewards. The child sample included 15 White (38.46%), 5 Mixed (12.82%), 2 Black (5.13%), 2 South Asian (5.13%), 1 Southeast Asian (2.6%), 2 participants who would describe their ethnicity as not listed (5.13%), and 12 participants whose parents did not provide information regarding their child’s ethnicity (30.77%) (further details in Table S1 in S1 File). One participant was excluded from all analyses because they failed to respond on most trials. Participants were recruited between 1 February 2019 and 18 December 2022.

Our sample size was comparable to previously published work [43,44], and we increased the number of trials per participant to support robust subject-level estimation of the effects. This included approximately 550 trials per participant in the cognitive control task and approximately 140 trials per participant in the willingness-to-work task. A sensitivity analysis for the achieved sample (children n = 39; adults n = 51; total N = 90) indicated ~80% power at α = .05 to detect a between-group main effect of approximately Cohen’s d ≈ 0.60–0.65 in a two-sample comparison.

Participants were pre-screened for juice-related allergies. All participants provided written informed consent prior to participation. For child participants, written informed consent was obtained from a parent or legal guardian. The study, including all consent procedures, was approved by the UCL Research Ethics Committee in accordance with national regulations (project ID: 12271/003). We report how we determined our sample size, all data exclusions, all manipulations, and all measures in the study.

Tasks

Matrix Reasoning: We administered the Matrix Reasoning subtest from the Wechsler Abbreviated Scale of Intelligence, Second Edition (WASI-II). On each trial, participants viewed an incomplete visual matrix or visual sequence and selected the option that best completed the pattern. Items (30 in total) were presented in order of increasing difficulty, following standard WASI-II administration procedures: at the beginning of administration, three sample items were presented, and administration was discontinued once participants provided three incorrect answers in a row. Furthermore, in line with administration procedures, only items up to number 24 were presented for ages 6–8. Administration typically took approximately five minutes. Thus, before completing the willingness-to-work task, participants had direct experience with multiple matrix items spanning increasing difficulty, although the exact number of items varied across individuals according to standard administration rules.

Explicit valuation: In this task, participants were shown either a picture of a juice carton indicating the primary reinforcer or a coin indicating the secondary reinforcer. Crucially, children were told that the coin image corresponds to the points they will be able to earn and exchange for a reward and not actual money. Participants answered three questions on a scale from 1 (least) to 6 (most): “How pleasant do you find this reward?”, “How much would you like to receive this reward?”, and “How happy would you feel right now about this reward?” These questions were modelled on previous work in the field [48]. Each question appeared only once per reinforcer, yielding an overall preference rating independent of reward size.

Willingness to work: We developed a novel task for measuring willingness to exert cognitive effort, adapted from prior work on temporal and cognitive effort discounting [49–51]. Participants made binary yes/no choices about whether they were willing to exert varying levels of cognitive effort (4 levels) on the matrix reasoning subtest of the WASI-II in order to obtain rewards that varied in size (6 levels) and type (primary, secondary). Trial combinations were repeated three times in pseudorandom order, resulting in 144 trials, with an additional four familiarization trials at the start of the task. Decisions were not timed, and stimuli remained on screen until a response was made. Participants received a prompt after 2 s if they had not yet responded.

A critical component of the willingness to work task was that effort levels were anchored to each participant’s performance and subjective experience on the matrix reasoning subtest. Because the test is standardized, difficulty could be expressed relative to age-normed performance (t-score distances within an age group). That is, based on participants’ performance and age, we could titrate them into a percentile rank and then use that for determining the cognitive effort levels. These effort levels corresponded to: items solved correctly without effort (“easiest”), items solved correctly but requiring thought (“easy”), items solved with uncertainty (“medium”), and items participants believed they solved incorrectly (“hard”). Participants were introduced to these four effort levels during the instruction phase of the willingness to work task. Note that this “titration process” was not explained to participants. Participants were merely told about the existence of the four cognitive effort levels.

In the willingness to work task, participants were instructed to decide whether they were willing to exert a given amount of cognitive effort for a given reward size. Crucially, they were instructed that their choices would relate to hypothetical items from the matrix reasoning subtest to ensure their responses were not contaminated by their belief about having to provide answers to items that were identical to the ones they had just experienced. They were further told that, from all choices, two would be selected at random at the end of the experiment, one for each reinforcer type. If a “no” trial was selected, they would forfeit any chance of obtaining that reward. If a “yes” trial was selected, they would obtain the reward for a correct answer. If their answer was incorrect, they were told there is a small chance of obtaining the reward (a residual 20% probability). This residual probability was included so that “yes” responses remained meaningful even for higher effort offers where, by definition, they might have considered it impossible to obtain a reward otherwise.

To validate whether participants perceived the four effort levels as increasingly effortful, they completed a control task at the end of the experiment. On each trial, participants saw a cue indicating the upcoming effort level, followed by a matrix reasoning item corresponding to that level. They were given unlimited time to solve the item and then answered questions from the NASA Task Load Index [52]. We used the cognitive load item as an indicator of perceived cognitive effort (range 0–20) and report these ratings in Fig. 3. Importantly, participants did not see the explicit mapping between effort levels and specific items during the willingness-to-work task itself.

Cognitive control: We used a standard go/no-go task based on previous work [53,54], with the instruction procedure modelled after [46]. Participants were required to press “f” or “k” depending on the direction of a fish stimulus (left = “f”, right = “k”) or withhold a response depending on stimulus colour (red = withhold response on 33% of trials; blue = respond on 66% of trials). The task combined a go/no-go structure with a congruent flanker-like condition where trial type was cued by fish colour (blue = go; red = no-go) facing the left or right direction. The imbalance in trial frequency was used to induce a prepotent tendency to respond, thereby increasing inhibitory demands on no-go trials. Participants were informed of the colour-response mapping but were not explicitly told the exact go/no-go probabilities. Importantly, reward outcomes depended deterministically on response correctness and reward condition.

The task included three main factors: reward type (primary, secondary), difficulty (easy, difficult), and reward size (low, high) all of which were signalled by visual stimuli. Notably, the visual stimuli used for high and low reward size differed from the visual stimulus indicating reinforcer type. Participants were explicitly instructed how the visual cues mapped onto reward type and reward size and understood these mappings.

On correct trials, adults earned either 2 pence (low) or 6 pence (high) in the secondary-reinforcer condition, whereas children saw the equivalent point increment, which they had been told would later be exchanged for prizes of comparable value. In the primary reinforcer condition, correct responses were rewarded with small squirts of juice delivered using a gustatory stimulation device (0.02 mL for low reward; 0.06 mL for high reward).

Participants first completed 10 familiarization trials, which were also used to initialize an adaptive response threshold. On each trial, participants first saw the condition combination (difficulty and reward size), followed by stimulus presentation. Fish were displayed for 0.4 s on easy trials and 0.2 s on difficult trials. Responses were recorded for up to 1.0 s from stimulus onset. The response threshold used for scoring was updated online as a function of the participant’s recent RT distribution (mean of last 10 trials). On easy trials, 2*SD was added to the mean, on hard trials, 0.5*SD was added. Go responses were scored as correct only if the appropriate keypress occurred before this adaptive threshold; correct keypresses after the deadline were coded as “too slow” and omissions as misses. On no-go trials, withholding a response for the full response window was scored as a correct rejection. Feedback was shown for 1000 ms after each response. Randomized trial sequences were generated for each participant, and there were approximately 550 trials per participant, with around 30 trials per individual condition.

Procedure

At the beginning of the session, we collected information about participants’ preferred juice flavours (e.g., pineapple, apple, orange), along with self-reported thirst and tiredness. Participants then completed the Matrix Reasoning subtest of the WASI-II [45], which was used to derive age-normed effort levels for the willingness-to-work task.

Participants next completed three tasks under two reward-type conditions (primary and secondary reinforcers): explicit valuation, willingness to work, and cognitive control. Task order and reward-type order were counterbalanced across participants. Instructions were automated and presented on screen, and an experimenter remained present to ensure that participants understood the task requirements. In the child sample, the experimenter conducted verbal comprehension checks before the main tasks began.

Before the tasks, participants were introduced to both reward types. Both groups were told that primary rewards involved receiving varying amounts of their preferred juice. For secondary rewards, adults were told that rewards would be paid out in pounds after the experiment. Children were told that they would earn (monetary) points that could later be exchanged for prizes varying in attractiveness that were comparable in pound value to that of adult rewards. Children were explicitly shown the available prizes. In the child sample, points were used as a secondary reinforcer because money could not be used due to ethical constraints. As a substitute, we chose monetary points because they approximated the key characteristic of money (i.e., allowing goods to be attained). Furthermore, this approach replicated the set-up of previous studies using monetary rewards in children. Notably, for both groups, the visual stimuli they saw related to rewards were identical.

After the three main tasks, participants completed additional measures used to assess possible confounds. These included a need-for-cognition scale, a judgement of the perceived monetary worth of 20 mL of juice, the control task for perceived cognitive effort described above, and post-task preference ratings for low versus high rewards in both the primary and secondary reinforcer conditions.

Adults were asked to avoid drinking fluids for four hours before the experiment in order to standardize thirst levels, following previous work [46,48]. This procedure was not applied to children for ethical reasons. This may have created a difference in “wanting” between the two groups in relation to the primary reinforcer. We therefore collected self-reported thirst measures and repeated key analyses including thirst as a covariate (see S1 File). At the end of the experiment, we also collected data indicating when they had last consumed any liquids.

Analysis

Data were analyzed using R, version 4.0.0 [55] and the package ggplot2, version 3.3.2 [56]. This study’s design and its analysis were not pre-registered.

Explicit valuation: We used a linear mixed-effects model to test whether participants’ valuation of the two reward types differed by group. Subjective value was computed as the mean of the three ratings for each reinforcer. Task order was included as a nuisance covariate because, for some participants, the valuation task occurred after other tasks in which reward contingencies had already been experienced. Owing to technical issues, only one reinforcer condition was recorded for 18 participants. Additional control analyses incorporated perceived monetary worth of juice and self-reported thirst measures (S1 File). We additionally ran a control model that included a further nuisance regressor for the perceived monetary worth of juice and assessed group differences in different aspects of thirst (S1 File).

Willingness to work: We quantified effort discounting using an exponential effort-discounting model adapted from prior discounting work [57] using hierarchical Bayesian model fitting in STAN [58]. Briefly, on each trial, the subjective value (SV) of the work option was modelled as:

where is reward magnitude, is cognitive effort level, is the participant-specific effort discounting parameter, and is the participant-specific effort sensitivity parameter. Reward magnitude was coded as integer values from 1–6 units. Cognitive effort was coded using t-value distance to define four effort levels: easiest = 1, easy = 7, medium = 13, and hard = 19. A non-zero value was used for the easiest level to allow the model to capture response variability on these trials. Choices were modelled at the trial level with a Bernoulli likelihood ( logistic choice rule (. The probability of choosing the work option was:

where is a participant-specific choice sensitivity (inverse temperature) and denotes the subjective value of the default option. Parameters were estimated using hierarchical Bayesian inference in STAN. Participant-level parameters were drawn from group-level normal distributions on an unconstrained scale and transformed to enforce bounds:

So that ∊ (0, 1), and βi ∊ (0, 5) with representing the standard normal CDF. We used weakly informative priors:

We ran 8 MCMC chains (NUTS) with 10,000 iterations per chain (2,000 warmup; 8 cores). Convergence and sampling quality were assessed using trace plots, , effective sample sizes, and checks for divergent transitions. Participant-level parameter estimates used in subsequent statistical analyses were derived from posterior summaries.

We tested whether effort discounting differed by reward type and age group (Group x reinforcer) using mixed-effects modelling on participant-level parameter estimates. In each group, we further used a general linear model to test for an effect of cognitive effort level. This was done to establish that different cognitive effort levels from the willingness to work task were perceived as more cognitively effortful. Because the mapping between cognitive effort level and item was not shown to participants prior to the control task, their responses on the willingness to work task were uncontaminated by these mappings and required an explicit control test.

To assess within-participant reliability in effort-based preferences, we computed a choice-consistency metric defined as the proportion of consistent yes/no responses across the three repetitions of each decision context. This metric provided an index of within-participant choice consistency independent of mean willingness to exert effort.

Cognitive control: We used generalized linear mixed-effects modelling (binomial family with logit link) to investigate whether the probability of a correct response in the go/no-go task varied by reinforcer and group. Analyses were conducted on trial-level responses (correct = 1, incorrect = 0) in two separate models: one fitted to go trials and one fitted to no-go trials, allowing us to dissociate reward effects on action initiation (go) from response inhibition (no-go). The fixed effects included group, reinforcer, and their interaction. To account for repeated measures, models included participant-level random intercepts and random slopes for within-subject predictors. In addition, we included random slopes for trial-level nuisance predictors capturing systematic changes across trials (e.g., trial index to model learning and fatigue effects) and response tendencies (e.g., direction biases). The maximal model candidate of interest for our main hypothesis was a response ~ group * reinforcer + (1 + reinforcer + trial_index + direction | sub_id). This was compared to a small set of theoretically motivated candidate models differing in random-effects complexity and nuisance covariates. Models that did not converge or had singular fits were not considered. We performed additional exploratory analyses with more complex models and higher-order interactions that included reward size and difficulty (Tables S2–S6 in S1 File). Due to the limited sample size, these models should be interpreted with caution.

Results

Reward-specific valuation differs by age group

To test our main hypothesis, we first assessed whether participants’ explicit valuation of the two reward types (Fig 1a) interacted with age. We found an interaction (Fig 2a) between group and reward type ( = 7.20, p < .01) where adults preferred secondary (M = 5.08 ± 0.20) over primary (M = 4.38 ± 0.16) rewards (t38 = 3.15, 95% CI [0.27, 1.22], p < .01) whereas children did not (Msecondary = 4.40 ± 0.24, Mprimary = 4.75 ± 0.19; t28 = 0.76, 95% CI [−.41, 0.89], p = 0.46), with moderate evidence for the null hypothesis (BF01 = 3.89). Furthermore, while both groups equally valued the primary reinforcer (t67.57 = 1.51, 95% CI [−0.86, 0.12], p = .14), adults valued the secondary reinforcer more strongly compared to children (t68.02 = 2.19, 95% CI [0.06, 1.29], p = .03). The observed interaction was not explained by group differences in their perception of how much juice is worth in pounds (t82.2 = 0.76, p = 0.45, Fig 2b), with the interaction remaining significant when this information was added as a covariate. Furthermore, this interaction remained significant and was not explained by confounds, such as how thirsty participants were in the beginning of the experiment (see S1 File).

thumbnail
Download:
Fig 1. Task descriptions.

The experiment consisted of one session and three tasks. a) An explicit valuation task (top panel) to establish the subjective value of both reward types. b) A willingness to work task (middle panel) to measure the willingness to exert cognitive effort as a function of reward type where reward size (6 levels) and cognitive effort (4 levels) were parametrically varied. c) A cognitive control task (bottom panel) to measure the invigorating effect of reward type on cognitive control. In this task, participants were rewarded on a trial-by-trial basis for correct responses either by observing an increase in their cumulative rewards for secondary rewards or by receiving small squirts of juice using a custom-built gustometer for correct responses, as commonly done in rodent or non-human primate research. In addition to reward type, reward size (2 levels), and difficulty (2 levels) were used in this task.

https://doi.org/10.1371/journal.pone.0351143.g001

thumbnail
Download:
Fig 2. Subjective value of reward differs across age groups.

Individual dots depict participants, the dot plot represents the group mean with whiskers representing 95% bootstrapped confidence intervals. The green dotted line represents the grand mean across groups and conditions. a) Subjective valuation of reward types in both groups. b) Perceived worth of juice in log (pence) across both groups. ** p < .01.

https://doi.org/10.1371/journal.pone.0351143.g002

In post-task preference ratings, we further replicated these results (significant Group x Reinforcer interaction (p = .005): adults preferred secondary over primary rewards at both reward sizes (both p < .001), while children showed no preference (both p = ns). Importantly, while both groups preferred large over small rewards (both p < .001), adults rated small secondary rewards more highly compared to children (p < .001).

Age-group differences in the effect of rewards on cognitive effort are reward-specific

Next, we asked whether group differences in explicit valuation were also reflected in their willingness to exert cognitive effort for such rewards. To test this, we used a novel and age-adapted cognitive effort task in which we parametrically varied reward size (values from 1 to 6 corresponding to one to six images of a given reinforcer) and difficulty level (values from 1 to 4 corresponding to coloured circle quadrants, also denoted as “easiest”, “easy”, “medium”, “hard”, see Fig 1b. See also Methods and S1 File for more details) to examine participants’ willingness to exert cognitive effort.

Participants discounted the value of rewards as a function of cognitive difficulty (Fig 3a) in a similar fashion as shown previously in the domain of physical [61], cognitive [50,59] and time [60] discounting. To quantify and statistically evaluate these differences, we investigated whether participants’ discounting rate (r) parameters for each reward type interacted with age. The parameters were obtained by fitting an effort-sensitive model previously used to model time discounting [57,59,60]. Qualitatively, participants with high r-parameters discount subjective value of a reward more strongly as more cognitive effort must be exerted to receive it, compared to participants with a low r-parameter value. As in the explicit valuation task, we observed a significant interaction between group and reward type on the discounting rate (r) parameter ( = 79.82, p < .001, Fig. 3b). This effect remained after controlling for participants’ choice consistency and self-reported thirst (both p < .001 for the interaction effect, see Methods and S1 File for more details). Adults discounted value more strongly for primary compared to secondary rewards (t44 = 4.94, 95% CI [0.026, 0.06], p < .001). However, this difference was reversed in children who discounted value more strongly for the secondary compared to the primary rewards (t28 = 10.68, 95% CI [0.04, 0.05], p < .001). Finally, adults more strongly discounted value for primary rewards compared to children (t56.328 = 4.03, 95% CI [0.02, 0.07], p < .001). These results demonstrate that adults were more willing to exert cognitive effort for secondary compared to primary rewards, while the opposite was true for children, indicating different types of reward may invigorate both groups when cognitive effort needs to be exerted in cognitive tasks. One interesting pattern observed in children’s discounting rates was reduced variance in the secondary reinforcer condition, suggesting a higher degree of response similarity in the child group as a whole.

thumbnail
Download:
Fig 3. Reward-specific effects on exertion of cognitive effort across age groups.

a) Average probability of accepting an offer in the cognitive effort task across cognitive effort (i.e., difficulty) levels, reward sizes and reward types for both groups. b) Discounting rate (r-parameter) from the cognitive effort model across both groups and conditions. c) Estimated cognitive load from the NASA Task Load Index for individual difficulty levels of the control task. *** p < .001.

https://doi.org/10.1371/journal.pone.0351143.g003

As a control, participants were required to rate the experienced cognitive load corresponding to individual cognitive effort levels they were making choices about in the willingness to work task (Fig 1b). To do this, participants were required to solve items of the matrix reasoning subtest at each difficulty level at the end of the experiment. Both groups rated more difficult items as cognitively more effortful on the NASA TLX (children: F(3, 139) = 11.46, p < .001; adults: F(3, 191) = 3.32, p = .02, Fig 3c).

Age-group differences in the invigorating effect of rewards on cognitive control are reward-specific

Finally, we examined whether reward type influenced the allocation of cognitive control differently across groups (Fig 1c). Given that we observed a reward type x group interaction on two separate tasks, one measuring explicit valuation and another willingness to exert cognitive effort, one would further predict a similar reward-type-specific pattern in a task requiring direct allocation of cognitive control [18,28,62–64].

Using generalized linear mixed-effects modelling, we found a similar interaction as in the other two tasks ( = 12.76, p < .001): adults performed better on go trials when rewarded with the secondary compared to the primary reinforcer (t50 = 7.79, p < .001, 95% CI [0.02, 0.04]), while children’s performance was comparable across both reward types (t36 = 0.27, p = .79; Fig 4a. See also S1 File for additional analyses involving reaction time data, models examining effects of reward size, and confound controls for self-reported thirst), with moderate evidence for the null hypothesis (BF01 = 5.47).

thumbnail
Download:
Fig 4. Cognitive control is invigorated in a reward-specific way across age groups.

a) Participants’ linear mixed model coefficients across group and reward type on go trials. b) Correlation between participants’ model coefficients and self-reported subjective value for each reinforcer. c) Participants’ linear mixed model coefficients across group and reward type on no-go trials. *** p < .001.

https://doi.org/10.1371/journal.pone.0351143.g004

Considering the pattern observed on this task resembled the one observed in the explicit valuation task, we predicted the self-reported explicit value ratings would be correlated with their performance estimates on go trials from Fig 4a. Indeed, adults’ subjective value of secondary rewards was correlated with interindividual differences in the go trials for secondary (r = .39, p = .009) but not primary rewards (r = .26, p = .09). In children, these correlations had a reversed sign but were not significant for either the primary (r = −.16, p = .4) or secondary (r = −.2, p = .26) reinforcer. Despite this correlation, we observed no correlation with r-parameters from the willingness to work task (p > .16). This suggests that the pattern across tasks is unlikely to be explained solely by general factors such as greater attentiveness across tasks.

Finally, when we investigated no-go trials for the same effect, we observed no such interaction ( = 0.85, p = 0.36, see Table S4 in S1 File). Notably, Adults were marginally better at inhibiting the prepotent response for secondary compared to primary reinforcers (t50 = 1.76, p = .08) while children performed equally well (t36 = 0.62, p = 0.54).

Discussion

Secondary incentives such as money or monetary points are widely used to study motivation and cognitive control across development [10,12,15–18,28,50,65–68,72,73]. Thus, it has been found that compared to adults, children or adolescents take greater risks for secondary reinforcers such as money [72,73], are less generous [67], and more impatient [15,65], while at the same time being less affected by secondary reinforcers in boosting their executive functions [28,66]. These studies are often interpreted under the assumption that such incentives have comparable value across age groups, such that observed differences reflect changes in the process under study rather than differences in the subjective value of the rewards themselves. The present findings challenge that assumption. Across three behavioural tasks: explicit valuation, cognitive effort allocation, and reward-motivated cognitive control, we observed systematic differences between children and adults in how primary and secondary incentives influenced behaviour. Adults showed stronger behavioural modulation for secondary incentives, whereas children responded more similarly for both primary and secondary incentives. These findings are consistent with the possibility that age-group differences in reward-guided behaviour partly reflect differences in the value assigned to the incentives used. However, because of potential differences in motivation and how rewards were introduced, these effects cannot be attributed uniquely to age-related changes in valuation.

The present results relate to an adult literature comparing secondary reinforcers such as money with primary reinforcers such as liquids. Prior work has shown that both types of reinforcers can modulate cognition and behaviour, sometimes to a similar behavioural degree, while relying on partially dissociable neural mechanisms. For example, Beck et al. [46] reported comparable behavioural improvements under monetary and liquid reinforcers during working memory, whereas Yee et al. [74] showed that monetary incentive level and liquid feedback valence can jointly shape motivated cognitive control depending on task structure. In our adult sample, behavioural modulation was stronger under secondary than primary reinforcement, suggesting that the relative impact of different reinforcer types may vary across contexts and populations. Notably, there are clear and genuine reward-related changes through development, as evidenced by structural and functional maturation of reward circuitry [24,26,27]. However, such differences can only be properly understood when behavioural and neural responses are interpreted within the ecology of the developing individual [70].

More broadly, the findings highlight the importance of considering subjective value when interpreting age-group differences in motivated behaviour. Secondary reinforcers such as money acquire value through experience and repeated associations with other desirable outcomes [36,37]. Adults encounter money as a central medium of exchange in daily life, whereas children’s exposure is typically more limited. Reinforcement-learning accounts would therefore predict that secondary reinforcers like money become more influential with repeated experience. This interpretation is corroborated by work showing adolescents aged 12 and above already are invigorated by secondary reinforcers such as money for exerting physical effort [30]. In contrast, the value of pleasant tastes may be shaped differently [2]. Differences in everyday goals and experiences across development may thus alter which incentives are most effective in guiding behaviour and how the relative value difference between them modulates this [5,39]. Our post-task preference ratings support this interpretation: both groups preferred larger over smaller rewards, but the clearest age-group divergence emerged for low-value secondary rewards, which adults preferred more than children. Post-task preference ratings, as used here and in previous work [75], provide a useful way to assess subjective reward value.

One intriguing difference is that we found a reinforcer effect in the cognitive control task only on go trials. We believe this may have arisen for several reasons. First, no-go trials comprised only 1/3 of total trials, which may have limited power to detect more subtle interaction effects. Second, reward may preferentially enhance response initiation or action invigoration rather than inhibitory control itself. Third, because children showed lower overall performance on no-go trials, floor effects may have further reduced sensitivity to detect reward-related differences in inhibition. Our findings also raise a broader question about how the observed differences interact with intrinsic reward [69] or findings showing reward learning remaining stable [71].

Our study has several limitations. First, the study used a cross-sectional design with a modest sample size, which limits our ability to make strong claims about developmental trajectories. While we used a Bayesian approach to estimate evidence for null hypotheses in our data, larger longitudinal samples are needed in future work. Second, the secondary rewards differed across groups: adults received money, whereas children earned monetary points they could exchange for prizes of comparable monetary value. While this allowed us to use developmentally appropriate incentives, it prevented direct comparison of identical reward types across groups. Third, the physiological state differed across groups because adults were asked to avoid fluids before the experiment, whereas children were not. Although the main results remained after controlling for self-reported thirst, this difference remains an interpretational limitation. However, it is worth noting that this may have led to an underestimation of the effects we observed rather than an over-estimation: if children had been subjected to the same procedure, they would have likely rated the primary reinforcer as more valuable. Fourth, we did not calibrate juice volume to individual physiology or track satiety across the session, so group differences in the motivational impact of the primary reinforcer cannot be ruled out (e.g., children may have viewed juice as a treat, whereas adults may have avoided sugary drinks). Relatedly, while the cognitive control task used a brief stimulus presentation window, adaptive response cutoffs meant that there was some degree of individualised calibration also in this task. Fifth, the willingness-to-work task required participants to interpret hypothetical and probabilistic offers; the clustered response patterns observed in children, particularly in the secondary reinforcer condition, may therefore partly reflect simplified strategies or age-group differences in task understanding rather than valuation alone. Sixth, the cognitive control task used brief stimulus presentation and adaptive response deadlines, meaning that age-group differences in processing speed and inhibitory control may have contributed to the observed reward effects. Seventh, the child sample spanned a relatively broad age range but was analysed as a single group. Thus, the present study cannot characterize variation within childhood itself. Finally, demographic comparability across groups was limited because adults were recruited primarily from a university student population, whereas children were recruited through community and school-based outreach.

Despite these limitations, our results show a consistent pattern across three different tasks, including post-task ratings that allowed within-subject assessment of relative value differences across reinforcer types. We believe our work highlights that future studies should carefully calibrate incentives across developmental samples, include reward-free baseline conditions, and consider both motivational and ecological relevance when selecting rewards. Doing so will help ensure that observed behavioural and neural differences reflect genuine differences in reward processing rather than differences in how incentives are valued.

Supporting information

S1 File. Supplementary materials.

This file contains the demographic breakdown of child participants and supplementary mixed-model results for go-trial accuracy, go-trial reaction times, no-go-trial accuracy, and reward size and difficulty effects.

https://doi.org/10.1371/journal.pone.0351143.s001

(DOCX)

S1 Fig. Willingness-to-work choice consistency by group and reinforcer.

https://doi.org/10.1371/journal.pone.0351143.s002

(PDF)

S2 Fig. Post-task preference ratings across adults and children for primary and secondary reinforcers.

https://doi.org/10.1371/journal.pone.0351143.s003

(PDF)

Acknowledgments

We would like to thank Somya Iqbal, Joshua Spowage, and Harriet Phillips for their help with organizing the data collection of the child sample. We would like to thank Marie Mueller and Roser Canigueral for their feedback on a previous version of the manuscript.

References

  1. 1. Dayan P, Niv Y. Reinforcement learning: the good, the bad and the ugly. Curr Opin Neurobiol. 2008;18(2):185–96. pmid:18708140
  2. 2. Lee D, Seo H, Jung MW. Neural basis of reinforcement learning and decision making. Annu Rev Neurosci. 2012;35:287–308. pmid:22462543
  3. 3. O’Doherty JP, Cockburn J, Pauli WM. Learning, Reward, and Decision Making. Annu Rev Psychol. 2017;68:73–100. pmid:27687119
  4. 4. Rangel A, Camerer C, Montague PR. A framework for studying the neurobiology of value-based decision making. Nat Rev Neurosci. 2008;9(7):545–56. pmid:18545266
  5. 5. Davidow JY, Insel C, Somerville LH. Adolescent Development of Value-Guided Goal Pursuit. Trends Cogn Sci. 2018;22(8):725–36. pmid:29880333
  6. 6. Kray J, Schmitt H, Lorenz C, Ferdinand NK. The influence of different kinds of incentives on decision-making and cognitive control in adolescent development: a review of behavioral and neuroscientific studies. Front Psychol. 2018;9:768. pmid:29875720
  7. 7. van Duijvenvoorde ACK, Huizenga HM, Somerville LH, Delgado MR, Powers A, Weeda WD, et al. Neural correlates of expected risks and returns in risky choice across development. J Neurosci. 2015;35(4):1549–60. pmid:25632132
  8. 8. Braams BR, van Duijvenvoorde ACK, Peper JS, Crone EA. Longitudinal changes in adolescent risk-taking: a comprehensive study of neural responses to rewards, pubertal development, and risk-taking behavior. J Neurosci. 2015;35(18):7226–38. pmid:25948271
  9. 9. Galvan A, Hare T, Voss H, Glover G, Casey BJ. Risk-taking and the adolescent brain: who is at risk?. Developmental Science. 2007;10(2):8–14.
  10. 10. de Water E, Cillessen AHN, Scheres A. Distinct age-related differences in temporal discounting and risk taking in adolescents and young adults. Child Dev. 2014;85(5):1881–97. pmid:24749521
  11. 11. Green L, Fry AF, Myerson J. Discounting of Delayed Rewards: A Life-Span Comparison. Psychol Sci. 1994;5(1):33–6.
  12. 12. van den Bos W, Rodriguez CA, Schweitzer JB, McClure SM. Adolescent impatience decreases with increased frontostriatal connectivity. Proc Natl Acad Sci U S A. 2015;112(29):E3765-74. pmid:26100897
  13. 13. Blake PR, Rand DG. Currency value moderates equity preference among young children. Evolution and Human Behavior. 2010;31(3):210–8.
  14. 14. Fehr E, Bernhard H, Rockenbach B. Egalitarianism in young children. Nature. 2008;454(7208):1079–83. pmid:18756249
  15. 15. Steinbeis N, Haushofer J, Fehr E, Singer T. Development of Behavioral Control and Associated vmPFC-DLPFC Connectivity Explains Children’s Increased Resistance to Temptation in Intertemporal Choice. Cereb Cortex. 2016;26(1):32–42. pmid:25100855
  16. 16. Decker JH, Otto AR, Daw ND, Hartley CA. From Creatures of Habit to Goal-Directed Learners: Tracking the Developmental Emergence of Model-Based Reinforcement Learning. Psychol Sci. 2016;27(6):848–58. pmid:27084852
  17. 17. Schulz E, Wu CM, Ruggeri A, Meder B. Searching for Rewards Like a Child Means Less Generalization and More Directed Exploration. Psychol Sci. 2019;30(11):1561–72. pmid:31652093
  18. 18. Hardin MG, Schroth E, Pine DS, Ernst M. Incentive-related modulation of cognitive control in healthy, anxious, and depressed adolescents: development and psychopathology related differences. J Child Psychol Psychiatry. 2007;48(5):446–54. pmid:17501725
  19. 19. Andersen SL, Thompson AP, Krenzel E, Teicher MH. Pubertal changes in gonadal hormones do not underlie adolescent dopamine receptor overproduction. Psychoneuroendocrinology. 2002;27(6):683–91. pmid:12084661
  20. 20. Seeman P, Bzowej NH, Guan HC, Bergeron C, Becker LE, Reynolds GP, et al. Human brain dopamine receptors in children and aging adults. Synapse. 1987;1(5):399–404. pmid:3505371
  21. 21. Simon NW, Moghaddam B. Neural processing of reward in adolescent rodents. Dev Cogn Neurosci. 2015;11:145–54. pmid:25524828
  22. 22. Tarazi FI, Tomasini EC, Baldessarini RJ. Postnatal development of dopamine D4-like receptors in rat forebrain regions: comparison with D2-like receptors. Brain Res Dev Brain Res. 1998;110(2):227–33. pmid:9748595
  23. 23. Fareri DS, Gabard-Durnam L, Goff B, Flannery J, Gee DG, Lumian DS, et al. Normative development of ventral striatal resting state connectivity in humans. Neuroimage. 2015;118:422–37. pmid:26087377
  24. 24. Asato MR, Terwilliger R, Woo J, Luna B. White matter development in adolescence: a DTI study. Cereb Cortex. 2010;20(9):2122–31. pmid:20051363
  25. 25. Barnea-Goraly N, Menon V, Eckert M, Tamm L, Bammer R, Karchemskiy A, et al. White matter development during childhood and adolescence: a cross-sectional diffusion tensor imaging study. Cereb Cortex. 2005;15(12):1848–54. pmid:15758200
  26. 26. Caballero A, Granberg R, Tseng KY. Mechanisms contributing to prefrontal cortex maturation during adolescence. Neurosci Biobehav Rev. 2016;70:4–12. pmid:27235076
  27. 27. Gogtay N, Giedd JN, Lusk L, Hayashi KM, Greenstein D, Vaituzis AC, et al. Dynamic mapping of human cortical development during childhood through early adulthood. Proc Natl Acad Sci U S A. 2004;101(21):8174–9. pmid:15148381
  28. 28. Insel C, Kastman EK, Glenn CR, Somerville LH. Development of corticostriatal connectivity constrains goal-directed behavior during adolescence. Nat Commun. 2017;8(1):1605. pmid:29184096
  29. 29. Barkley-Levenson E, Galván A. Neural representation of expected value in the adolescent brain. Proc Natl Acad Sci U S A. 2014;111(4):1646–51. pmid:24474790
  30. 30. Rodman AM, Powers KE, Insel C, Kastman EK, Kabotyanski KE, Stark AM, et al. How adolescents and adults translate motivational value to action: Age-related shifts in strategic physical effort exertion for monetary rewards. J Exp Psychol Gen. 2021;150(1):103–13. pmid:32496090
  31. 31. Gelman SA, Echelbarger ME. Children, Object Value, and Persuasion. J Consum Psychol. 2019;29(2):309–27.
  32. 32. Williams J, Ashill N, Thirkell P. How is value perceived by children?. Journal of Business Research. 2016;69(12):5875–85.
  33. 33. Csikszentmihalyi M, Larson R, Prescott S. The ecology of adolescent activity and experience. J Youth Adolesc. 1977;6(3):281–94. pmid:24408457
  34. 34. Guyer AE, McClure-Tone EB, Shiffrin ND, Pine DS, Nelson EE. Probing the neural correlates of anticipated peer evaluation in adolescence. Child Dev. 2009;80(4):1000–15. pmid:19630890
  35. 35. Galván A, McGlennen KM. Enhanced striatal sensitivity to aversive reinforcement in adolescents versus adults. J Cogn Neurosci. 2013;25(2):284–96. pmid:23163417
  36. 36. Brandstaetter E, Brandstaetter H. What’s money worth? Determinants of the subjective value of money. Journal of Economic Psychology. 1996;17:443–64.
  37. 37. Searle J. The construction of social reality. New York: Free Press; 1995.
    • 38. Bevins RA, Besheer J. Novelty reward as a measure of anhedonia. Neurosci Biobehav Rev. 2005;29(4–5):707–14. pmid:15876456
    • 39. Hofferth SL, Sandberg JF. How american children spend their time. J of Marriage and Family. 2001;63(2):295–308.
    • 40. Ventura AK, Worobey J. Early influences on the development of food preferences. Curr Biol. 2013;23(9):R401-8. pmid:23660363
    • 41. Mennella JA, Bobowski NK. The sweetness and bitterness of childhood: Insights from basic research on taste preferences. Physiol Behav. 2015;152(Pt B):502–7. pmid:26002822
    • 42. Petty S, Salame C, Mennella JA, Pepino MY. Relationship between Sucrose Taste Detection Thresholds and Preferences in Children, Adolescents, and Adults. Nutrients. 2020;12(7):1918. pmid:32610456
    • 43. Cohen AO, Phaneuf CV, Rosenbaum GM, Glover MM, Avallone KN, Shen X, et al. Reward-motivated memories influence new learning across development. Learn Mem. 2022;29(11):421–9. pmid:36253009
    • 44. Saragosa-Harris NM, Cohen AO, Reneau TR, Villano WJ, Heller AS, Hartley CA. Real-World Exploration Increases Across Adolescence and Relates to Affect, Risk Taking, and Social Connectivity. Psychol Sci. 2022;33(10):1664–79. pmid:36219573
    • 45. Wechsler D. Wechsler Abbreviated Scale of Intelligence—Second Edition (WASI-II) Manual. San Antonio (TX): NCS Pearson. 2011.
      • 46. Beck SM, Locke HS, Savine AC, Jimura K, Braver TS. Primary and secondary rewards differentially modulate neural activity dynamics during working memory. PLoS One. 2010;5(2):e9251. pmid:20169080
      • 47. Levy DJ, Glimcher PW. Comparing apples and oranges: using reward-specific and reward-general subjective value representation in the brain. J Neurosci. 2011;31(41):14693–707. pmid:21994386
      • 48. Yousuf M, Heldmann M, Göttlich M, Münte TF, Doñamayor N. Neural processing of food and monetary rewards is modulated by metabolic state. Brain Imaging Behav. 2018;12(5):1379–92. pmid:29243121
      • 49. Odum AL. Delay discounting: I’m a k, you’re a k. J Exp Anal Behav. 2011;96(3):427–39. pmid:22084499
      • 50. Chevalier N. Willing to Think Hard? The Subjective Value of Cognitive Effort in Children. Child Dev. 2018;89(4):1283–95. pmid:28397991
      • 51. Westbrook A, Kester D, Braver TS. What is the subjective cost of cognitive effort? Load, trait, and aging effects revealed by economic preference. PLoS One. 2013;8(7):e68210. pmid:23894295
      • 52. Hart SG, Staveland LE. Development of NASA-TLX (Task Load Index): results of empirical and theoretical research. Adv Psychol. 1988;52:139–83.
      • 53. Brydges CR, Anderson M, Reid CL, Fox AM. Maturation of cognitive control: delineating response inhibition and interference suppression. PLoS One. 2013;8(7):e69826. pmid:23894548
      • 54. Bunge SA, Dudukovic NM, Thomason ME, Vaidya CJ, Gabrieli JDE. Immature frontal lobe contributions to cognitive control in children: evidence from fMRI. Neuron. 2002;33(2):301–11. pmid:11804576
      • 55. R Core Team. R: A language and environment for statistical computing. Vienna: R Foundation for Statistical Computing. 2020.
        • 56. Wickham H. ggplot2: Elegant graphics for data analysis. 2nd ed. New York: Springer. 2016.
        • 57. Ebert JEJ, Prelec D. The Fragility of Time: Time-Insensitivity and Valuation of the Near and Far Future. Management Science. 2007;53(9):1423–38.
        • 58. Ahn W-Y, Haines N, Zhang L. Revealing Neurocomputational Mechanisms of Reinforcement Learning and Decision-Making With the hBayesDM Package. Comput Psychiatr. 2017;1:24–57. pmid:29601060
        • 59. Chong TT-J, Apps M, Giehl K, Sillence A, Grima LL, Husain M. Neurocomputational mechanisms underlying subjective valuation of effort costs. PLoS Biol. 2017;15(2):e1002598. pmid:28234892
        • 60. Johnson MW, Bickel WK. Within-subject comparison of real and hypothetical money rewards in delay discounting. J Exp Anal Behav. 2002;77(2):129–46. pmid:11936247
        • 61. Bonnelle V, Manohar S, Behrens T, Husain M. Individual Differences in Premotor Brain Systems Underlie Behavioral Apathy. Cereb Cortex. 2016;26(2):807–19. pmid:26564255
        • 62. Geier CF, Terwilliger R, Teslovich T, Velanova K, Luna B. Immaturities in reward processing and its influence on inhibitory control in adolescence. Cereb Cortex. 2010;20(7):1613–29. pmid:19875675
        • 63. Jazbec S, Hardin MG, Schroth E, McClure E, Pine DS, Ernst M. Age-related influence of contingencies on a saccade task. Exp Brain Res. 2006;174(4):754–62. pmid:16733706
        • 64. Padmanabhan A, Geier CF, Ordaz SJ, Teslovich T, Luna B. Developmental changes in brain function underlying the influence of reward processing on inhibitory control. Dev Cogn Neurosci. 2011;1(4):517–29. pmid:21966352
        • 65. Achterberg M, Peper JS, van Duijvenvoorde ACK, Mandl RCW, Crone EA. Frontostriatal White Matter Integrity Predicts Development of Delay of Gratification: A Longitudinal Study. J Neurosci. 2016;36(6):1954–61. pmid:26865618
        • 66. Niebaum JC, Chevalier N, Guild RM, Munakata Y. Adaptive control and the avoidance of cognitive control demands across development. Neuropsychologia. 2019;123:152–8. pmid:29723599
        • 67. Steinbeis N, Bernhardt BC, Singer T. Impulse control and underlying functions of the left DLPFC mediate age-related and age-independent individual differences in strategic social behavior. Neuron. 2012;73(5):1040–51. pmid:22405212
        • 68. Strang NM, Pollak SD. Developmental continuity in reward-related enhancement of cognitive control. Dev Cogn Neurosci. 2014;10:34–43. pmid:25160678
        • 69. Bains A, Barber A, Nell T, Ripollés P, Krishnan S. The role of intrinsic reward in adolescent word learning. Dev Sci. 2024;27(5):e13513. pmid:38685611
        • 70. Mobbs D, Trimmer PC, Blumstein DT, Dayan P. Foraging for foundations in decision neuroscience: insights from ethology. Nat Rev Neurosci. 2018;19(7):419–27. pmid:29752468
        • 71. Pauli R, Brazil IA, Kohls G, Klein-Flügge MC, Rogers JC, Dikeos D, et al. Action initiation and punishment learning differ from childhood to adolescence while reward learning remains stable. Nat Commun. 2023;14(1):5689. pmid:37709750
        • 72. Defoe IN, Dubas JS, Figner B, van Aken MAG. A meta-analysis on age differences in risky decision making: adolescents versus children and adults. Psychol Bull. 2015;141(1):48–84. pmid:25365761
        • 73. Levin IP, Hart SS, Weller JA, Harshman LA. Stability of choices in a risky decision‐making task: a 3‐year longitudinal study with children and adults. Behavioral Decision Making. 2006;20(3):241–52.
        • 74. Yee DM, Krug MK, Allen AZ, Braver TS. Humans integrate monetary and liquid incentives to motivate cognitive task performance. Frontiers in Psychology. 2016;6:2037.
        • 75. Phaneuf-Hadd CV, Jacques IM, Insel C, Otto AR, Somerville LH. Characterizing age-related change in learning the value of cognitive effort. J Exp Psychol Gen. 2025;154(6):1667–80. pmid:40146600