Article Figures & Data
Figures
- Figure 1.
Incentive integration task paradigm and behavioral results. A, Incentive integration task paradigm. Participants performed letter-digit cued task-switching and could earn monetary rewards and liquid incentives for accurate and fast performance (below an individualized criterion threshold). Notably, as the receipt of both monetary reward and liquid feedback was performance-contingent, participants had to integrate the value of both types of motivational incentives when performing this cognitive task (see also Extended Data Figure 1-1). B, Reward rate by motivational incentive conditions. Participants performed better with trials with higher monetary reward. In terms of liquid effects, participants performed worse on saltwater compared with juice or neutral trials. Error bars indicate SEM. For visualization of RT and accuracy by incentive condition, see Extended Data Figure 1-2. For mixed model results, see Extended Data Figure 1-3. C, Self-report motivation ratings by motivational incentive conditions. Participants reported they were more motivated for higher monetary reward and more appetitive liquid incentives. Error bars indicate SEM. A hierarchical regression revealed that the inclusion of motivation ratings significantly predicted variance in reward rate beyond the experimental effects. For hierarchical regression results, see Extended Data Figure 1-4. D, Scatterplot of z-scored averaged reward rate and self-report motivation ratings by participant. Notably, although both measures are sensitive to the motivational incentive conditions, there was only a weak positive association between self-reported motivation and reward rate. These data suggest that the two measures reflect overlapping yet dissociable motivational components of the incentivized cognitive task.
- Figure 2.
RT, accuracy, and RT switch costs between baseline and incentive blocks. A, Comparison of RT in baseline and incentive task blocks demonstrates that motivational incentives are associated with a significant reduction in RTs (ms) between baseline and incentive blocks for younger adults (t(45) = 16.298, p < 0.001). In other words, younger adults are faster with incentives compared with without incentives. B, Younger adults showed a significant drop in accuracy between baseline and incentive blocks (t(45) = 7.582, p < 0.001). Together, these data demonstrate that the participants are both faster and more accurate with monetary and liquid motivational incentives. This shift down the speed-accuracy curve to increase reward rate demonstrates that participants increase their effort in accordance with the bundled incentives. All error bars in all plots indicate 95% CIs. C, RT switch costs were significantly reduced between baseline and incentive blocks, thus revealing that increased recruitment of cognitive control during the incentive blocks relative to the baseline blocks (t(45) = 2.956, p = 0.005). For mixed model results of RT switch costs by motivational incentives, see Extended Data Figure 2-1.
- Figure 3.
dACC encodes both monetary rewards and liquid incentives in bundled β estimates. A, Bilateral dACC ROI mask. This ROI encompasses the peak voxels of dACC based on a prior meta-analysis on motivated cognitive control (Parro et al., 2018). B, Mixed block/event-related design for incentive integration task. Participants performed eight task blocks total, with two baseline blocks and six incentive blocks. Each participant's reward criterion (40%) was calculated based on performance in the baseline run and used to determine the RT threshold by which fast and accurate performance would lead to earned monetary and liquid incentives. Monetary reward value randomly varied on a trial-wise basis. Liquid type was blocked and counterbalanced, and delivered as performance feedback for successful attainment of monetary reward (colored arrows). Participants did not receive money nor liquid for slow, incorrect, or abstained responses. A GLM was applied to extract the β estimates for the sustained liquid block conditions and transient event-related motivation conditions for each participant (Petersen and Dubis, 2012). C, Trial-wise β estimates are illustrated for each of the nine motivational incentive conditions. Darker colors represent higher monetary reward level. Gray rectangles represent cue-related activity, demonstrating a significant monetary reward effect 4-6 s after cue onset. D, Bundled β estimates in dACC are calculated by combining the β estimates for sustained liquid effects and cue-related monetary reward effects from the event-related hemodynamic response. We assume an additive relationship between the monetary and liquid effects in terms of BOLD signal representation of integrated incentive value. Reward rate was significantly predicted by dACC bundled β, revealing that dACC represented the aggregate motivational value of primary and secondary incentives and is associated with parametric modulation of motivated cognitive task performance. For mixed model results of dACC β estimates by motivational incentives, see Extended Data Figure 3-1.
- Figure 4.
dACC effects on reward rate mediated by self-report motivation ratings. A, Within-subjects mediation analysis of dACC. Three linear mixed-effects models were conducted to confirm the effects of incentive conditions in dACC, reward rate, and motivation ratings (see red dashed lines). The models revealed that incentive conditions significantly predicted dACC bundled β, reward rate, and motivation ratings. Next, a Bayesian multilevel within-subjects mediation analysis was conducted to test for a relationship between these three variables. The mediation analyses revealed that the relationship between within-subjects variability in dACC and reward rate was mediated by within-subjects variability in motivation ratings (see blue lines). Inclusion of motivation ratings partially mediated the association between dACC and reward rate proportion mediated effect = 0.42, CI90 = (0.06, 1.10); 90% credible intervals are used as the range of the posterior distribution for each of the path parameters. ***p < 0.001. For parameter estimates, see Extended Data Figures 4-1 and 4-2. B, Between-subjects correlations of dACC. Left, Scatterplot of z-scored averaged dACC β estimates and reward rate by subject revealed an association between dACC and reward rate (though beneath threshold for statistical significance). Right, Scatterplot of z-scored averaged dACC β estimates and self-report motivation ratings also revealed association between dACC and motivation ratings.
- Figure 5.
dACC selectively encodes subjective motivational value and modulates motivated task performance. A, To test for whether motivational incentive integration effects were present in other brain regions, bundled β estimates were calculated for a priori ROIs associated with value-based decision-making (striatum, vmPFC) and taste processing (anterior insula). Mixed models were implemented with selected ROIs and plotted to compare with dACC effects on reward rate and motivation ratings. Error bars indicate 95% CIs. Right, Only dACC significantly predicted reward rate. Left, Motivation ratings were also significantly predicted anterior insula, caudate, and putamen, in addition to dACC, suggesting that self-reported motivation of incentives is encoded across multiple valuation brain regions. These data reveal the specificity of the relationship between dACC and reward rate, demonstrating that dACC encodes the translation of motivational value and is transformed into effortful actions on a cognitively demanding task. *p < 0.05. **p < 0.01. ***p < 0.001. B, Visualization of valuation brain ROIs.
Tables
Questions Motivation How motivated were you on Juice $ trials?
How motivated were you on Juice $$ trials?
How motivated were you on Juice $$$$ trials?
How motivated were you on Neutral $ trials?
How motivated were you on Neutral $$ trials?
How motivated were you on Neutral $$$$ trials?
How motivated were you on Saltwater $ trials?
How motivated were you on Saltwater $$ trials?
How motivated were you on Saltwater $$$$ trials?Liking Please indicate on a scale of 1-7 how much you like or dislike this liquid. Intensity Please indicate on a scale of 1-7 how intense you find the taste of this liquid. ↵aFor all the questions, participants completed Likert ratings on a scale from 1 (lowest) to 7 (highest). For the motivation questions, there were nine questions: one for each of the motivational incentive conditions. For the liking and intensity questions, there were three questions: one for each liquid type. For each liquid, participants were asked to rate how much they liked the liquid from 1 (least liked) to 7 (most liked), as well as how intense they found the taste of each liquid from 1 (least intense) to 7 (most intense). For Likert ratings of liking and intensity for the liquid incentives, see Extended Data Table 1-1.
Parcel ID ParcelnNumber MNI coordinates (RAS) x y z 107 LH_SalVentAttn_Med_1 −6 22 32 108 LH_SalVentAttn_Med_2 −6 0 40 110 LH_SalVentAttn_Med_4 −6 10 48 311 RH_SalVentAttn_Med_1 8 18 26 312 RH_SalVentAttn_Med_2 8 2 42
Figure 1-1
1. Detailed Schematic of Single Trial from Incentive Integration Task. Adapted for the MRI scanner. Participants performed a consonant-vowel odd-even (CVOE) switching task, which entailed being presented with an ambiguous letter-number pair, and being asked to categorize the target symbol based on the task cue preceding the target (e.g., “Attend Number” or “Attend Letter). A reward cue was placed above and below each instruction cue, which indicated low ($), medium ($$), or high ($$$$) reward. Monetary reward cues were randomized across trials within each block. Importantly, the dollar signs associated with the cues indicated how much the subject could earn on each trial. If subjects were accurate and faster than a subject criterion response time (40% of fastest correct response times for all trials during the baseline block), then they received 1 ml of liquid as performance feedback at the end of the trial. If subjects answered incorrectly, too slowly, or not at all, they neither received monetary reward or liquid. Liquid type was manipulated in a blocked fashion, counterbalanced across subjects, and was positive (apple juice), neutral (isotonic tasteless solution), or negative (saltwater). Because the liquid delivery is symbolic, which means that it conveys the same information regardless of the type, any differences across liquids would suggest that subjects are incorporating the value of that liquid to influence their task performance. Thus, this task design enables us to test for the dissociable and integrated effects of monetary and liquid incentives on cognitive task performance. Download Figure 1-1, EPS file.
Figure 1-2
Response Time and Accuracy for each of the nine Experimental Task Conditions. A) Response Time by nine motivational incentive conditions. Participants were faster on trials with high monetary reward, but there were no differences in liquid incentive, and no significant interaction. B) Accuracy by nine motivational incentive conditions. Accuracy did not differ across monetary reward level, though there was weak effect of liquid incentive that did not meet the threshold for statistical significance. Errors bars indicate 95% CIs. Download Figure 1-2, EPS file.
Figure 1-3
Response times and accuracy mixed model by monetary reward and liquid incentives. Enhancements in reward rate were primarily driven by faster RT. Download Figure 1-3, DOCX file.
Figure 1-4
Hierarchical regression with self-report motivation ratings. Motivation ratings were predicted by monetary reward and liquid incentives. Additionally, these motivation ratings significant predicted variance in reward rate in a model including experimental fixed effects (e.g., money, liquid). Download Figure 1-4, DOCX file.
Table 1-1
Self-Reported Liking and intensity ratings for each liquid. A) Participants reported significantly different and transitive liquid preferences [b = 1.685, t(45)=11.68, p < 0.001], preferring juice to neutral [t(45)=3.859, p < 0.001], and neutral over saltwater [t(45)=5.138, p < 0.001]. B) Participants reported juice and saltwater as more intense than neutral [juice versus Neutral: t(45)=12.423, p < 0.001; saltwater versus Neutral: t(45)=6.170, p < 0.001]. Download Table 1-1, EPS file.
Figure 2-1
RT switch cost by monetary and liquid incentives. Notably, we did not observe any incentive effects on switch cots (beyond the general switch-cost reduction observed in incentive blocks relative to baseline in Fig. 2). This lack of effect is not surprising, given that switch costs were quite small under incentive conditions (∼20 ms), and likely did not have sufficient dynamic range to exhibit sensitive to the more subtle parametric incentive manipulations. Download Figure 2-1, DOCX file.
Figure 3-1
Dorsal Anterior Cingulate Cortex (dACC) β estimates predicted by monetary reward and liquid incentives. Download Figure 3-1, DOCX file.
Figure 4-1
Estimated parameters from Bayesian multilevel mediation analysis of motivation ratings mediating relationship between dACC and reward rate. Download Figure 4-1, DOCX file.
Figure 4-2
Estimated parameters from Bayesian multilevel mediation analysis of reward rate mediating the relationship between dACC and motivation ratings. Download Figure 4-2, DOCX file.
