The neurocomputational mechanisms underlying the impact of social comparison on effort investment

The neurocomputational mechanisms underlying the impact of social comparison on effort investment | Research Square window.SnipcartSettings = { analytics: { enabled: false } }; (function() { var accessVector = localStorage.getItem('access_vector') || ''; window.dataLayer = window.dataLayer || []; if (accessVector) { window.dataLayer.push({ user: { profile: { profileInfo: { snid: accessVector } } } }); } })(); (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-K279D39R'); Browse Preprints In Review Journals COVID-19 Preprints AJE Video Bytes Research Tools Research Promotion AJE Professional Editing AJE Rubriq About Preprint Platform In Review Editorial Policies Our Team Advisory Board Help Center Sign In Submit a Preprint Cite Share Download PDF Article The neurocomputational mechanisms underlying the impact of social comparison on effort investment Ping Wei, Jiarui Dong, Yachao Rong, Shengjie Ma, Yang Xu This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-7749786/v1 This work is licensed under a CC BY 4.0 License Status: Under Review Version 1 posted You are reading this latest preprint version Abstract Previous studies have documented stronger neural responses when participants receive better monetary outcome than others, but how performance-based comparisons shape self-efficacy and subsequent effort behavior remains unclear. We conducted behavioral (Experiment 1, N = 32) and electrophysiological (Experiment 2, N = 34) experiments where participants performed an effortful task and received downward/upward feedback relative to a purported competitor. Computational modeling and mediation analysis revealed dynamic self-efficacy updates, with downward feedback enhancing subsequent effort via full mediation by self-efficacy. Electrophysiological results revealed stronger neural activities in response to downward compared to upward feedback for the current round, and enhanced contingent negative variation (CNV) and cue-beta power, reflecting better task preparation, for the upcoming round. Notably, CNV amplitudes were modulated by efficacy, with higher efficacy predicting more negative-going CNV. These findings demonstrate that performance-based comparison dynamically regulates self-efficacy, thereby shaping both neural preparatory processes and subsequent effort allocation in goal-directed behavior. Biological sciences/Psychology/Human behaviour Biological sciences/Neuroscience/Social neuroscience social comparison effort self-efficacy Reward Positive CNV Figures Figure 1 Figure 2 Figure 3 Figure 4 Figure 5 Figure 6 Introduction Social comparison is a pervasive psychological phenomenon that shapes individual’s judgments, emotions, and goal-dircted behaviors 1 . Rooted in social comparison theory, people inherently evaluate their abilities and self-worth against others, particularly when objective assessment criteria are absent 2 . Among its various effects, social comparison may exert the greatest influence on individual’s self-perception of ability, providing an immediate and efficient mechanism for assessing their abilities relative to others 3 , 4 . These self-perceptions of ability, in turn, underpin individual’s willingness to invest effort in behaviors needed to achieve desired outcomes 5 – 7 . Theoretically, social comparison influences goal-directed behavior via a dynamic, multi-stage process: eveluating comparison outcomes, updating self-evaluations of abilities, and guiding subsequent effort allocation. Social comparison initiates a process where individuals first evaluate discrepancies between themselves and others, perceiving either superiority (downward comparison) or inferiority (upward comparison). Electrophysiological studies consistently link this evaluation to reward- and feedback-sensitive neural signals, including reward-related positivity (RewP), feedback-related potentials (fb-P3), and late positive potential (LPP). These studies show stronger neural responses to downward compared to upward social comparison feedback 8 – 10 , suggesting that individuals perceive downward comparisons as rewarding and experience positive emotions when affirming their superiority. Critically, most of these studies have employed gambling tasks with monetary payoffs as the comparison dimension 11 – 16 , introducing confounding between social comparison evaluation and the valuation of monetary rewards. Fewer studies have used performance-based comparision (e.g., profession-related knowledge quizzes 17 or stopwatch tasks 18 , 19 ). This leave a gap that limits conclusions about how social comparison shapes ability perceptions in real-world task contexts. Following comparison outcomes, individuals update their self-efficacy. Bandura’s (1977) social cognitive theory defines self-efficacy as beliefs in one’s ability to execute goal-directed actions, with performance experiences (successes/failures) driving increases/decreases in efficacy 5 . In social contexts, downward and upward comparisons act as proxies for success and failure, directly shape self-efficacy 5 , 21 . Neurocomputational work further shows that self-efficacy updates are stronger in response to positive (vs. negative) performance feedback 7 , 20 , 22 . However, how these efficacy updates translate to effort investment, a key outcome of social comparison, remains debated. Bandura’s framework predicts social comparison feedback and updated self-efficacy jointly regulate effort (Bandura, 1977), but empirical evidence is limited. Few studies have examined post-comparison risk-taking behaviors 15 , 23 – 25 , and recent research using ecological momentary assessment has reported findings contradicts Bandura’s prediction (Diel et al., 2021, 2024). For example, Diel and colleagues reported that upward (not downward) social comparison increased self-improvement motivation, and led to stronger effort intentions and increased effort investment. By contrast, downward social comparisons foster complacency, prompting individuals to disregard improvement strategies and enter a “coasting” state marked by reduced effort. These effects are particularly pronounced in contexts of moderate social comparison 3 , 26 , 27 . However, these studies rely on self-report measures of motivation and effort, which may be susceptible to social desirability bias or distorted by inaccurate self-perceptions. To address these gaps, we combined behavioral (Experiment 1) and electrophysiological (Experiment 2) experiments to dissect the neurocognitive mechanisms linking performance-based social comparison, self-efficacy updates, and subsequent effort allocation—free from monetary reward confounds. Both experiments used essentially identical designs (Fig. 1 ). Participants performed a Stroop task, with effort investment quantified by the number of responses within random time windows (Experiment 1: 8s, 10s, 12s; Experiment 2: 6s, 8s, 10s, Leng et al., 2021). After each trial, participants received trial-by-trial social comparison feedback, signaling whether they performed better (downward comparison) or worse (upward comparison) than a purported competitor. To align with the prior studies (Diel et al., 2021, 2024), we created a moderate comparison context where participants received balanced downward/upward feedback (Fig. 1 a). Every 3–5 trials, participants rated their likelihood of winning the next round (Fig. 1 b) to index self-efficacy. Behaviorally, according to Bandura’s theory 5 , we predict that downward social comparison feedback would enhance self-efficacy and drive better task performance compared to upward feedback. However, if our design mirrors Diel et al.’s (2021, 2024) contexts, upward comparison feedback may instead yield better task performance than downward comparison. Using reinforcement learning model and linear mixed models, we aim to quantify trial-by-trial self-efficacy updates, and to test how social comparison feedback and self-efficacy jointly influence subsequent effort allocation. Moreover, we evaluate whether self-efficacy mediates the effect of social comparison on effort investment using multilevel mediation models. Electrophysiologically (Experiment 2), we investigated neural correlates of feedback evaluation and task preparation. We expected enhanced RewP, fb-P3, and LPP amplitudes for downward compared to upward feedback during the evaluation phase, as downward comparison is primarily experienced as rewarding. In line with the behavioral results, we hypothesized that social comparison feedback and updated self-efficacy would modulate task preparatory responses, such as contingent negative variation (CNV) amplitudes and cue-beta band power in the subsequent round. Our experimental design served three key purposes: it allowed us to assess the neural signatures associated with the evaluation of social comparison based on performance, to track the dynamic changes in efficacy induced by social comparison, and to investigate how these two factors (i.e., feedback and efficacy) shaped effort investment in the subsequent round. Our findings revealed two main results: first, relative to upward comparison, downward social comparison elicited stronger neural responses and higher levels of efficacy; second, both downward social comparison and higher efficacy positively contributed to effort investment in the subsequent round. Results Experiment 1 Adjusted Efficacy Expectations Based on Different Social Comparison Feedback We first examined how different types of social comparison feedback shape efficacy expectations (i.e., the ratings of the probability of winning in the next round). The LMM results revealed that participants' ratings were significantly affected by the direction of the recent social comparison feedback, F (1, 32.47) = 33.82, b = 0.190, p < .001, 95% CI [0.126, 0.255], with higher ratings of wining probability after downward feedback compared to upward feedback (62.91 vs. 59.54, Fig. 2 a). Next, we used the two-learning-rate model to calculate the efficacy estimates for each round. We then examined how the model-based efficacy estimates related to participants’ rating scores to verify the reliability. The LMM results revealed that model-based efficacy estimates significantly predicted rating scores, F (1, 31.64) = 111.36, b = 0.547, p < .001, 95% CI [0.445, 0.648] (Fig. 2 b), confirming that the model-based efficacy estimates accurately reflect participants’ efficacy expectations. Finally, we analyzed the impact of social comparison feedback on model-based efficacy estimates. The LMM results revealed that social comparison feedback significantly predicted the model-based efficacy estimates, F (1, 32.00) = 64.31, b = 0.316, p < .001, 95% CI [0.239, 0.393], with higher model-based efficacy estimations following downward social comparison compared to upward comparison (0.62 vs. 0.47, Fig. 2 c). These results provide evidence that participants adjusted their efficacy expectations based on different social comparison feedbacks, with downward feedback leading to increased efficacy estimates. Our findings not only demonstrate that the two-learning-rate model successfully captures trial-by-trial changes in efficacy expectations, but also confirm that downward social comparison feedback functions as a positive performance-related experience, thereby effectively enhancing efficacy expectations. Improved Performance Following Downward Social Comparison Feedback and Higher Efficacy Expectations To clarify the impact of social comparison feedback on effort investment, we first performed LMMs to assess how social comparison feedback and model-based efficacy expectations influenced correct responses per second (CRPS) and mean reaction times (RTs), respectively. The LMM results for CRPS revealed significant main effects of social comparison feedback (Fig. 3 a), F (1, 103.10) = 5.08, b = 0.028, p = .026, 95% CI [0.004, 0.052], and model-based efficacy estimates (Fig. 3 b), F (1, 14.42) = 5.56, b = 0.049, p = .026, 95% CI [0.011, 0.088]. Specifically, CRPS was higher in the upcoming round following downward social comparison feedback compared to upward social comparison feedback (0.971 vs. 0.961). Similarly, higher model-based efficacy estimates were associated with higher CRPS for the upcoming round (Fig. 3 b). However, no significant main effects were found for mean RTs (658 vs. 664 ms, Fig. 3 a) being affected by social comparison feedback, F (1, 32.01) = 2.48, b = -0.017, p = .124, 95% CI [-0.038, 0.004], or model-based efficacy estimates (Fig. 3 b), F (1, 12.07) = 4.45, b = -0.098, p = .056, 95% CI [-0.190, -0.007]. Subsequently, we fitted the Hierarchical Drift Diffusion Model (HDDM) to trial-level data to investigate the latent computational processes underlying these performance patterns. We first performed HDDM to explore how social comparison feedback influence drift rates and response threshold. The results revealed that the best-fitting model was the one included Stroop congruency as a control variable (see Hierarchical Drift Diffusion Model in Method). The main effect of social comparison feedback on drift rates was significant, with faster drift rates observed following downward social comparison feedback compared to upward social comparison feedback ( b = 0.05, 95% CI [0.01, 0.08], P b>0 = 0.97, Fig. 3 c). However, there were no significant main effect of social comparison on response threshold ( b = -0.00, 95% CI [-0.02, 0.01], P b<0 = 0.67). Next, we performed HDDM to explore the effect of model-based efficacy estimates on drift rates and response threshold. The best-fitting model was also the one controlled for Stroop congruency. The results revealed a significant main effect of model-based efficacy estimates on drift rates, with higher model-based efficacy estimates being associated with faster drift rates ( b = 0.17, 95% CI [0.01, 0.35], P b>0 =0.96, Fig. 3 c). However, there was no effect of model-based efficacy on response threshold ( b = -0.05, 95% CI [-0.40, 0.31], P b<0 = 0.61) . These results indicate that both downward social comparison feedback and higher efficacy expectations are associated with improved performance, as evidenced by higher CRPS and faster drift rates. Moreover, this improved performance reflects greater cognitive control investment, with increased attentional allocation to targets rather than a shift in the speed-accuracy tradeoff. Efficacy Expectations Mediated the Relationship Between Social Comparison and Effort Investment These findings demonstrate that participants adjusted their efficacy expectations in response to different social comparison feedback, which subsequently influenced their effort investment in the upcoming round. To further validate this relationship and elucidate the mechanism of social comparison and effort investment, we conducted multilevel mediation analyses to examine whether efficacy expectations mediate the effect of social comparison feedback on effort, as indexed by CRPS and mean RTs. The results revealed that model-based efficacy estimates fully mediated this relationship (Fig. 3 d). This was supported by significant indirect effects for both CRPS ( ab = 0.025, p = .041, 95% CI [0.001, 0.048]) and mean RTs ( ab = -0.025, p = .012, 95% CI [-0.045, -0.005]). Specifically, downward social comparison feedback enhanced efficacy expectations, which then led to greater effort investment in the next round. Experiment 2 Behavioral Results The results of Experiment 1 demonstrated that participants updated their efficacy expectations based on social comparison feedback, which subsequently guided their effort investment at the behavioral level. Building on these findings, Experiment 2 aimed to identify the neural signatures associated with the evaluation of social comparison feedback and to reveal how social comparison feedback and efficacy expectations modulate subsequent effort investment at the neural level. Consistent with Experiment 1, we observed a significant effect of social comparison feedback on participants’ self-reported rating scores, with higher scores following downward versus upward comparison (62.13 vs. 59.60), F (1, 34.21) = 18.03, b = 0.140, p < .001, 95% [0.075, 0.204]. In addition, model-based efficacy estimates significantly predicted self-reported rating scores, F (1, 29.25) = 125.25, b = 0.465, p < .001, 95% [0.384, 0.547]. Social comparison feedback also significantly influenced efficacy estimates, F (1, 34.00) = 32.81, b = 0.245, p < .001, 95% [0.161, 0.329], with higher estimates following downward versus upward feedback (0.58 vs. 0.47). These findings replicate those of Experiment 1 and strengthen support for the influence of social comparison feedback on both self-reported ratings and efficacy expectations. LMM analysis of CRPS revealed neither the significant main effect of social comparison feedback (1.083 vs. 1.078), F (1, 106.68) = 0.89, b = 0.013, p = .348, 95% [-0.013, 0.038], nor of model-based efficacy estimates, F (1, 24.27) = 1.29, b = 0.034, p = .266, 95% [-0.030, 0.094]. While non-significant, these results exhibited a similar trend to those in Experiment 1. Similarly, LMM analysis of mean RTs revealed neither the significant effect of social comparison feedback (597 vs. 600 ms), F (1, 62.46) = 2.00, b = -0.017, p = .161, 95% [-0.040, 0.006], nor of model-based efficacy estimates, F (1, 25.80) = 0.78, b = -0.035, p = .382, 95% [-0111, 0.042]. Consistent with Experiment 1, the best-fitting HDDM models were those controlled for the effect of Stroop trial congruency. The results showed that model-based efficacy estimates exerted a significant main effect on drift rate ( b = 0.19, 95% CI [0.01, 0.37], P b>0 = 0.96) but not on the threshold ( b = 0.10, 95% CI [-0.07, 0.26], P b0 = 0.68) nor on threshold ( b = -0.00, 95% CI [-0.02, 0.01], P b<0 = 0.60). As in Experiment 1, model-based efficacy estimations fully mediated the influence of social comparison on effort investment, as indicated by significant indirect effects on both CRPS ( ab = 0.033, p < .001, 95% CI [0.015, 0.051]) and mean RTs ( ab = -0.043, p < .001, 95% CI [-0.059, -0.028]). Overall, the behavioral results replicated those of Experiment 1. Neural Signatures for Social Comparison Outcomes Evaluation To investigate the neural signatures underlying the evaluation of different social comparison feedback types, we first focused on event-related potential (ERP) components during the feedback phase: specifically, the RewP, fb-P3, and LPP. Our analysis revealed significant main effects of social comparison on the RewP, F (1, 33.76) = 5.08, b = 0.034, p = .031, 95% [0.004, 0.065], fb-P3, F (1, 34.12) = 4.85, b = 0.052, p = .034, 95% [0.006, 0.098], and LPP, F (1, 268.96) = 7.94, b = 0.044, p = .005, 95% [0.014, 0.074]. Each component displayed a consistent pattern, with more positive-going waveforms following downward versus upward social comparison feedback. Specifically, the RewP (13.57 ± 14.48 vs. 12.60 ± 13.72 µV, Fig. 4 a and 4 b), fb-P3 (14.35 ± 13.37 vs. 12.92 ± 12.78 µV, Fig. 4 c and 4 d), and LPP (3.24 ± 11.39 vs. 2.23 ± 11.25 µV, Fig. 4 e and 4 f) all exhibited this effect. Collectively, these ERP components exhibited heightened sensitivity to downward versus upward social comparison feedback. These findings align with previous studies 8 , 10 , 43 , which suggest that downward social comparison feedback, perceived as a form of social reward, elicits stronger neural activities associated with reward evaluation and emotional arousal. Neural Signatures of Effort Preparation Are Influenced by Social Comparison and Efficacy Expectations Our behavioral results demonstrated that participants adjusted their effort investment in round t + 1 based on social comparison feedback and efficacy expectations derived from round t. To explore the neural underpinnings of this process, we focused on the CNV and Cue-beta band activity observed in round t + 1. Both of these components are linked to task preparation and cognitive/anticipatory processes for the upcoming task 37 , 42 We first analyzed the impact of round t social comparison feedback on these two neural signatures in round t + 1. Significant main effects of social comparison were observed for both the CNV, F (1, 114.41) = 4.15, b = -0.034, p = .044, 95% [-0.066, -0.001], and Cue-beta band power, F (1, 1905.17) = 4.63, b = -0.034, p = .031, 95% [-0.065, -0.003]. Specifically, downward versus upward social comparison feedback from round t induced more negative-going CNV waveforms (-1.27 ± 13.19 vs. -0.37 ± 13.40 µV, Fig. 5 a and Fig. 5 b) and higher Cue-beta band power (-0.41 ± 3.82 vs. -0.15 ± 3.85 dB, Fig. 6 a and Fig. 6 b) in round t + 1. Next, we analyzed the impact of model-based efficacy estimates (from round t) on round t + 1 CNV and cue-beta power. The CNV was significantly affected by model-based efficacy estimates, F (1, 40.22) = 7.90, b = -0.056, p = .007, 95% [-0.097, -0.015], with more negative-going waveforms in round t + 1 associated with higher round t model-based efficacy estimates (Fig. 5 c). However, model-based efficacy estimates had no significant effect on Cue-beta band power (Fig. 6 c), F (1, 46.49) = 0.31, b = -0.009, p = .583, 95% [-0.050, 0.033]. These results suggest that participants engaged in enhanced task preparation for upcoming targets following both downward social comparison feedback and higher model-based efficacy estimates. Collectively, these findings underscore the role of social comparison dynamics and self-efficacy in shaping the neural mechanisms associated with preparatory engagement and cognitive readiness for subsequent tasks. Discussion By manipulating a social comparison context, we investigated how evaluating one’s performance relative to others modulates effortful behavior. Participants exhibited enhanced neural responses to downward versus upward social comparison feedback, indexed by larger RewP, fb-P3, and LPP components, reflecting the motivational and emotional significance of the superior social comparisons. Participants then dynamically updated their self-efficacy based on the direction of social comparison feedback. Finally, both downward social comparison and higher self-efficacy improved the preparatory engagement of cognitive control, as indicated by enhanced CNV and cue-beta power, and facilitated the subsequent effort investment, as indicated by improved performance in the next round. We observed more positive RewP, fb-P3, and LPP components in the downward versus upward feedback condition, aligning with previous studies using gambling tasks and monetary payoffs to manipulate social comparison direction 10 , 16 , 44 , 45 . The RewP and fb-P3 are typical components for reward outcome evaluation, with larger amplitudes elicited by more favorable outcomes (see Glazer et al., 2018 for a review). Additionally, the LPP component reflects later-stage affective evaluation and the motivational relevance of the feedback (Hu et al., 2017; Luo et al., 2015; Wu et al., 2012). These stronger responses to downward versus upward feedback support the notion that downward social comparison is experienced as rewarding, associated with more positive emotions and stronger motivation 17 , 44 , 46 . Importantly, we manipulated performance as social comparison dimension and explicitly informed participants that their monetary compensation was independent of task performance. Thus, the heightened electrophysiological responses to downward versus upward feedback clearly demonstrate that these components are sensitive to social comparison itself, rather than monetary reward contingencies. Participants then updated their self-perceptions of ability based on different types of social comparison feedback. Previous studies have indicated that positive social feedback (e.g., high-ability feedback or better performance feedback in competitive contexts) enhances participants’ confidence and improves self-evaluations of ability 4 , 7 , 20 , 21 . Our results confirm that downward social comparison can be conceptualized as a form of performance accomplishment that significantly improves participants’ efficacy perceptions, whereas upward comparisons undermines them 5 . Based on social comparison direction and updated efficacy expectations, participants adjusted their effort investment. More negative-going CNV waveforms and stronger cue-beta power following downward comparison indicate enhanced control preparation and motor readiness for the upcoming round (Frömer et al., 2021; Glazer et al., 2018; Rong et al., 2022; Y. Zhang et al., 2023; Doyle, Yarrow, & Brown, 2005; Gable et al., 2016), which then led to better behavioral performance, evidenced by higher CRPS and faster drift rates across both experiments. HDDM analyses further revealed that participants actively adjusted their attention in response to downward feedback and elevated self-efficacy, characterized by accelerated evidence accumulation without compromising the speed-accuracy trade-off. Downward social comparison functions as rewarding feedback: it fosters positive emotions, boosts self-efficacy, and increases the perceived likelihood of goal attainment. These factors collectively motivated participants to invest greater effort to maintain this favorable state. Our multilevel mediation model further supported this process, confirming that self-efficacy fully mediated the relationship between social comparison and effort investment. However, as noted in the Introduction, recent studies used ecological momentary assessment have found that upward comparisons promote greater self-reported effort investment intentions 3 , 27 , which is inconsistent with our results. This discrepancy may stem from methodological differences in effort assessment. Retrospective self-reports are susceptible to social desirability and memory biases 47 , and even accurately-reported intentions may not translate to actual behavior due to the well-known intention–behavior gap 48 . In contrast, our study used a real-time effort task to directly measure how social comparison feedback shapes actual effort investment. The current study’s behavioral and neurocognitive findings thus provide direct evidence of social comparison’s impact on effort behavior. Prior research has manipulated perceived control efficacy in non-social context, where different cues indicated whether performance feedback (i.e., monetary reward) was contingent on participants’ performance (high efficacy) or not (low efficacy, Cao et al., 2022; Frömer et al., 2021). These studies reported better behavioral performance under high-efficacy conditions. In contrast, our study employed a fixed, non-meaningful cue for task preparation (see also Grahek et al., 2022). The enhanced brain responses and improved behavioral performance following downward social comparison suggest effort adjustments are driven by past comparison outcomes and internal self-efficacy updates, highlighting the critical role of self-efficacy in shaping goal-directed behavior (Bandura, 1977). These findings converge with animal studies showing that “winning” mice exhibit heightened initiative and sustained effortful behaviors in competitive contexts compared to “losing” mice 50 , reflecting the “winner effect”, where success history boosts the likelihood of future achievement in both humans and animals 50 – 52 . Nonetheless, our findings should be interpreted with caution. We observed a stronger motivational effect of downward versus upward social comparison in moderate competitive contexts (50% downward/50% upward comparisons), but this relationship may vary with the ratio of downward to upward feedback. Our results do not discount the adaptive effects of upward comparison; rather, both comparison directions can foster adaptive motivational responses 3 . This is partly supported by the behavioral results of our Experiment 2: shortening the task time window from 8–12 to 6–8 seconds reduced fatigue and narrowed the motivational discrepancy between upward and downward comparisons relative to Experiment 1. The motivational impacts of downward and upward feedback may thus compete, with downward feedback exerting greater influence under conditions of heightened participant fatigue. Future research should explore performance dynamics in more complex competitive environments. In conclusion, our study illustrates the cognitive and neural mechanisms through which downward and upward social comparisons influence effort investment, tracing the process from initial feedback evaluation to self-efficacy updates and subsequent effort allocation. Findings reveal that, compared to upward comparison, downward comparison feedback elicits enhanced neural activity and stronger self-efficacy expectations, both of which positively correlate with improvements in behavioral performance. Critically, self-efficacy fully mediates the relationship between social comparison and effort investment, underscoring its pivotal role as a motivational driver of goal-directed effort. These findings provide empirical support for a dynamic, multi-stage framework in which social comparison shapes efficacy expectations and guides effort allocation over time. Methods Experiment 1 Participants We recruited 32 healthy participants (10 male; mean age: 22.75 ± 2.03 years) with non-psychology backgrounds. All participants had normal or corrected-to-normal vision and normal color perception. We conducted a post hoc power analysis using G*Power (Faul et al., 2007) for this sample size (n = 32), using a paired-sample t-test (to test the difference between two dependent means). With an α level of 0.05, the analysis yielded a statistical power of 0.85. This study was approved by the Ethics Committee of University, and all participants gave informed consent prior to the experiment. Experimental Design We used a within-participant design with two social comparison conditions: downward social comparison and upward social comparison. PsychoPy3 was used to present experimental stimuli and record participant response data ( https://www.psychopy.org ). Before the formal experiment, participants completed a practice session to familiarize themselves with the procedure. The practice consisted of two parts. The first part included 80 trials in which Stroop stimulus was presented at the center of the screen. Participants were required to discriminate the color of the Stroop target by pressing the corresponding keys (red-D, green-F, yellow-J, blue-K) with their left and right index and middle fingers. Feedback was presented after each response, with a checkmark (“√”) indicating a correct response and a cross (“×”) indicating an incorrect one. Participants needed to achieve at least 80% accuracy in this part before progressing to the next. The second part of the practice mirrored the formal experiment. A triangle cue was displayed for 1500 ms, followed by a series of Stroop targets. Participants responded to the color of each Stroop target one after another within a randomly selected time window (8 s, 10 s, and 12 s). At the end of this time window, a diamond feedback symbol was presented for 2000 ms to signal the end of that round. After completing 12 practice rounds, participants rated the difficulty of the practice on a scale from 0 to 100, with a time limit of 10 seconds to submit their response. Participants were then informed of the social comparison feedback rules (see below) and that they would receive a final-pay of 55 RMB, regardless of their task performance. The formal experiment contained 144 rounds. Each round began with a variable interval of 1400–1600 ms. A triangle cue was then displayed for 1500 ms, followed by a variable interval of 1000–1200 ms, during which participants were required to concentrate and prepare. The Stroop targets was then presented consecutively. Participants were instructed to respond to the color of each Stroop target by pressing the D, F, J, or K (red-D, green-F, yellow-J, blue-K) with their left and right index and middle fingers. There was no fixed presentation time for the Stroop target stimuli. Instead, a 200 ms fixation cross appeared after each response, and the next Stroop target stimulus was presented immediately after the fixation cross disappeared. Participants could respond to as many Stroop targets as they wished/could during each round of randomly selected time window of 8 s, 10 s, or 12 s. The variable time windows prevented participants from counting Stroop stimuli 28 . Subsequently, a fixation cross displayed for 500 ms, followed by a “calculating” screen displayed for 1500 ms to enhance the realism of the social comparison context. Finally, after a variable interval of 1400–1600 ms, feedback was presented for 2000 ms, indicating the social comparison feedback for that round (Fig. 1 a). To effectively induce social comparison, participants were explicitly informed before the experiment that they would engage in a real-time comparison with a genuine competitor. They were informed about how social comparison feedback presented. Correct responses per second (CRPS), serving as an indicator of participants' performance, was calculated at the end of each round. Social comparison feedback was then presented based on the CRPS values of the participant and the competitor. An upward arrow indicated worse performance (i.e., lower CRPS than the competitor in that round, upward social comparison). In contrast, a downward arrow indicated better performance, (i.e., higher CRPS than the competitor in that round, downward social comparison). A bidirectional arrow indicated similar performance between the participant and the competitor (lateral social comparison, Fig. 1 a). The lateral rounds served as fillers to enhance the realism of the social comparison context. Moreover, participants received upward comparison feedback in rounds where CRPS was ≤ 0.4. In other rounds, the order of the three types of social comparison feedback was pseudo-randomized, and the feedback was independent of participants’ actual performance. There were 66 rounds with upward comparison feedback, 66 rounds with downward comparison feedback, and 12 rounds with lateral feedback, divided into 6 blocks of 24 rounds, with a fixed two-minute break at the end of each block. Each type of feedback was limited to appearing continuously no more than three times. After every 3 to 5 rounds, participants were required to complete a rating based on their past performance to estimate the probability of winning in the subsequent round, with scores ranging from 0 to 100 (see Fig. 1 b). Participants had 10 seconds to submit their responses. A total of 36 ratings were gathered throughout the experiment. Behavioral Analyses To investigate how social comparison influences effort investment, we examined the effect of feedback from the current round on effort investment in the next round (see Fig. 3 a). Accordingly, we excluded data from the first round and from rounds following lateral feedback for each participant in the analyses of RTs and CRPS. Mean RTs and CRPS served as indices of effort investment. First, we excluded incorrect trials for each participant (7.91% of the total trials). Following this, we calculated mean RTs for each round. Rounds with mean RTs exceeding three standard deviations (1.91% of total rounds) and CRPS values beyond three standard deviations (1.28% of total rounds) were removed for each participant in each social comparison condition. Statistical Analysis Reinforcement Learning Models. To compute trial-by-trial model-based efficacy estimates, we used a two-learning-rate reinforcement learning model. This model updates the efficacy estimate for the next round ( \(\:{E}_{t+1}\) ) based on the efficacy estimate from the current round ( \(\:{E}_{t}\) ) and the prediction error ( \(\:{\delta\:}_{t}\) ), with the update weighted by a learning rate ( \(\:\alpha\:,\:\:\:0<\alpha\:<1\) ) specific to the direction of the prediction error. The prediction error ( \(\:{\delta\:}_{t}\) ) quantifies the discrepancy between the actual social comparison feedback from the current round ( \(\:{f}_{t}\) ) and the current round’s efficacy estimate ( \(\:{E}_{t}\) ). To capture asymmetric updating of efficacy expectations (i.e., differential learning from positive vs. negative prediction errors), we used two distinct learning rates: a positive learning rate ( \(\:{\alpha\:}_{pos}\) ) applied when \(\:{\delta\:}_{t}\) >0 (positive prediction errors, indicating feedback exceeded current efficacy expectations) and a negative learning rate ( \(\:{\alpha\:}_{neg}\) ) applied when \(\:{\delta\:}_{t}\) >0 (negative prediction errors, indicating feedback fell below current efficacy expectations). This relationship can be expressed as follows: $$\:{\delta\:}_{t}={f}_{t}-{E}_{t}$$ $$\:f\left(x\right)=\left\{\begin{array}{c}{E}_{t}+{\alpha\:}_{pos}\ast\:{\delta\:}_{t},\:\:{\delta\:}_{t}>0\\\:{E}_{t}+{\alpha\:}_{neg}\ast\:{\delta\:}_{t},\:\:{\delta\:}_{t}<0\end{array}\right.$$ Linear Mixed Models . We first conducted linear mixed models (LMMs) to explore three core questions: (1) how social comparison feedback influences participants’ self-reported rating scores; (2) how social comparison feedback influences model-based efficacy estimates; and (3) how model-based efficacy estimates predict participants’ rating scores. Subsequently, we used additional LMMs to examine the effects of social comparison feedback and model-based efficacy estimates on effort investment, measured by mean RTs and CRPS. Mean Stroop congruency (proportion of congruent vs. incongruent trials per round) was included as a level-1 control variable to account for task-related variance in performance. For all models, age and gender were entered as fixed effects to control for potential demographic influences. The model structure included random intercepts for participants (to account for individual differences in baseline responses) and random slopes for all predictor variables (to allow predictor effects to vary across participants). Hierarchical Drift Diffusion Model . To decompose performance into cognitive subprocesses, we fit Hierarchical Drift Diffusion Models (HDDMs) to trial-level data, including RTs and response accuracy (incorrect = 0 and correct = 1) for each Stroop target, across different social comparison conditions. HDDM estimates two key parameters linked to effort and decision-making: drift rate (v, representing the speed of evidence accumulation) and response threshold (a, indicating the degree of response caution) under different social comparison conditions (with upward social comparison set as the intercept). Additionally, we investigated how drift rate and threshold were influenced by model-based efficacy estimates. For each experiment, we fitted two models focused on social comparison feedback: one examined the effects of social comparison feedback on drift rate and response threshold, while the other additionally controlled for Stroop trial congruency (congruent vs. incongruent) in both parameters. Model comparison revealed that the model accounting for the effect of Stroop congruency was the best-fitting one for social comparison feedback. Similarly, we fitted two models for efficacy estimates: one assessed the effects of efficacy estimates on drift rate and response threshold, and the other likewise incorporated Stroop congruency as a control predictor for both parameters. Again, model comparison confirmed that the best-fitting model for efficacy estimates was the one that included the effect of Stroop congruency. Finally, we selected these best-fitting models for further analyses on how social comparison and efficacy estimates influence participants’ effort allocation. Multilevel Mediation Model. To test whether model-based efficacy estimates mediate the relationship between social comparison feedback and effort investment (indexed by mean RT and CRPS), we conducted two 1-1-1 multilevel mediation models (level-1: rounds; level-2: participants) using round-level data. These models included independent variable ( \(\:X\) , social comparison feedback, coded as 0 = upward, 1 = downward) at level-1, mediator ( \(\:M\) , model-based efficacy estimates at level-1), and two outcome variables ( \(\:{Y}_{1}\) , mean RTs, and \(\:{Y}_{2}\) , CRPS all at level-1), which allowed us estimate the mediation effect at the round level. Experiment 2 Participants We recruited another 42 healthy participants with non-psychology backgrounds. All participants had normal or corrected-to-normal vision and normal color perception. Two participants were excluded due to crash of computer program, and four participants were excluded for not completing the formal experiment. Additionally, two participants were excluded due to insufficient data per condition (< 30 rounds) after removing the eyeblinks or muscle artifacts (see below EEG acquisition and processing). Consequently, the final sample included 34 participants (9 males, 22.12 ± 1.91 years). We conducted a post hoc power analysis using G*Power 35 for this sample size (n = 34), employing a paired-samples t-test (to test the difference between two dependent means). With an α level of 0.05, the achieved statistical power was 0.88. This study was approved by the Ethics Committee of University, and all participants gave informed consent prior to the experiment. Experimental Design The experimental procedure in Experiment 2 was identical to that of Experiment 1, with one modification: the duration of target presentation was shortened to a randomly selected duration of 6 s, 7 s, and 8 s. This adjustment was made to reduce participant fatigue during the EEG experiment. Additionally, participants received a fixed payment of 95 RMB, independent of their task performance. EEG Acquisition and Processing Participants were seated 60 cm from the computer monitor in a dimly lit and sound-attenuated room. Electroencephalography (EEG) data were recorded by 64-channel Neuroscan equipment (NeuroScan 4.3.1, USA), following the international 10–20 system of EEG electrode placement. The left mastoid served as the reference electrode, and the center of the forehead served as the ground electrode. Vertical electrooculograms were placed above and below the right eye to record eye blinks, while horizontal electrooculograms were placed at the outer canthi of the eyes to record eye movements. All electrode impedance were kept below 5 kΩ during recording, and the sampling rate was set at 500 Hz. Offline EEG data were preprocessed using the EEGLAB toolbox ( http://sccn.ucsd.edu/eeglab ) in MATLAB2018b. The EEG data were re-referenced to the average of the left and right mastoid electrodes and filtered with a band-pass of 0.1–40 Hz. Independent components analysis (ICA) was performed to detect and remove eye blinks and movement artifacts. For ERP analysis, the cleaned EEG data were then divided into epochs for the cue phase (-200 to 1500 ms) and the feedback phase (-200 to 1600 ms). Baseline correction was applied using a pre-stimulus interval of 200 ms for each epoch. Any epoch with amplitudes exceeding ± 100 µV was excluded from further analysis, resulting in the exclusion of 10.24% of epochs from the cue phase and 11.27% from the feedback phase. Time-frequency analysis was performed on the preprocessed EEG data using MATLAB2018b . The ICA-cleaned EEG data were segmented into epochs from − 600 to 1600 ms for both the cue and feedback phases. Epochs with amplitudes exceeding ± 100 µV were removed, resulting in the exclusion of 10.96% of epochs from the cue phase and 12.15% from the feedback phase. The EEG data were then analyzed using short-time Fourier transform (STFT) with a fixed 400 ms Hanning-tapered window. A pre-stimulus time interval from − 400 ms to -200 ms was used for baseline correction, and power values were converted to a decibel scale (dB, 10 × log10). Behavioral Analyses As in Experiment 1, incorrect trials (7.96% of total trials) were first excluded from each round for each participant. Following this, we calculated the mean RTs for each round. Rounds with mean RTs exceeding three standard deviations (1.39% of total rounds) and CRPS values beyond three standard deviations (1.88% of total rounds) were subsequently removed from each social comparison condition for each participant. We used the same LMM, HDDM, and multilevel mediation models as those employed in Experiment 1. For the reinforcement learning model analysis, we also fitted three models (intercept, one-learning-rate, and two-learning-rate), with the two-learning-rate model providing the best fit, consistent with the findings from Experiment 1. EEG Analyses Based on the collapsed localizers approach 36 and previous studies, we selected the ERP and time-frequency component within the specific time windows and electrode sites. Following this, we extracted round-level data for all ERP components under different social comparison feedback conditions for each participant. For the feedback phase, we focused on the RewP component (280–340 ms, across C1, Cz, and C2 electrodes), the fb-P3 component (350–450 ms, across CP1, CPz, and CP2 electrodes) and the LPP component (600–1000 ms, across CP1, CPz and CP2 electrodes). The RewP component is a fronto-central positive deflection peaking at 250–350 ms post-feedback onset, which is sensitive to reward feedback 12 , 37 , 38 . The fb-P3 component is a centro-parietal positive deflection that peaks at 300–600 ms post-feedback onset, reflecting the evaluation of feedback valence and magnitudes 10 , 12 , 37 . The LPP component is a sustained positive deflection occurring post-feedback onset over centro-parietal electrodes, indicating emotional arousal levels elicited by feedback, with larger amplitudes corresponding to higher arousal 12 , 14 . For these feedback-locked EEG components, linear mixed models were employed to examine the impact of social comparison feedback on the RewP, fb-P3, LPP. All models included age and gender as fixed effects to control for potential demographic influences, with predictor variables treated as both fixed and random effects. For the cue phase, we focused on the CNV component and cue-beta band activity. The CNV component, characterized as a slowly developing negative potential over fronto-central electrodes following cue onset, was analyzed across FC1, FCz, and FC2 electrodes from 550 ms to 950 ms post-cue onset. This component reflects increased motor preparation for the upcoming task in response to incentive or effort cues 38 – 41 . Additionally, the cue-beta band was analyzed across F3, F1, and Fz electrodes from 150 ms to 600 ms post-cue onset, focusing on the frequency range of 15–21 Hz. Increased beta suppression during the cue-evaluation phase has been associated with the preparation and execution of upcoming responses 37 , 42 . Aligning with the behavioral analyses, we encoded the cue-code for the next round (t + 1) based on the feedback-code (t) of the current round. This approach allowed us to specifically analyze the effect of social comparison feedback (t) on preparatory engagement of effort investment, as indexed by the CNV and cue-beta activity in the subsequent round (t + 1). For cue-locked EEG components, linear mixed models were performed to explore the influence of social comparison on the CNV and cue-beta power, as well as the effect of model-based efficacy estimation on the CNV and cue-beta power. All models incorporated age and gender as fixed effects to control for potential demographic influences, while prediction variables were incorporated as both fixed and random effects simultaneously. Statistics and reproducibility Reinforcement Learning Models To fit the reinforcement learning model parameters, we used two inputs: (1) social comparison feedback from each round, and (2) participants’ actual efficacy estimates (operationalized as their self-reported probability-of-wining ratings, z-scored per participant to standardize across individuals). To ensure the coherence and completeness of the empirical data, we also included rounds that involved lateral social comparisons. The actual feedback ( \(\:{f}_{t}\) ) was assigned a value of 1 for downward social comparison feedback, 0 for upward social comparison feedback, and 0.5 for lateral social comparison feedback. The prediction error ( \(\:{\delta\:}_{t}\) ) defined as the discrepancy between the actual feedback ( \(\:{f}_{t}\) ) and the model’s current efficacy estimate ( \(\:{E}_{t}\) ). The learning rate ( \(\:\alpha\:\) ) scaled the impact of the prediction error ( \(\:{\delta\:}_{t}\) ) on updates to the next round’s efficacy estimate ( \(\:{E}_{t+1}\) ) 29 . We fitted three candidate models to estimate α for each participant, using maximum likelihood estimation (MLE) to minimize the difference between the model’s predicted estimates and participants’ observed efficacy estimates. The first model fitted a single learning rate for both positive and negative prediction errors, while the second model fitted two distinct learning rates: a positive learning rate for positive prediction errors and a negative learning rate for negative prediction errors. Additionally, we implemented an intercept model, which assumed the efficacy estimate remained constant throughout the experiment without any updates. After comparing Akaike information criterion (AIC) and Bayesian information criterion (BIC) values (where lower values indicate a better fit), we determined that the two-learning-rate model provided the best fit. To validate the two-learning-rate model, we conducted a parameter recovery analysis. We first generated data for 200 synthetic participants (144 rounds each) using empirical parameter distributions: a fixed noise parameter of 0.15, a fixed intercept of 0.5, and learning rates sampled from a uniform distribution ranging from 0.001 to 0.5. We then fitted the two-learning-rate model to the simulated dataset in an attempt to recover the learning rate parameters. Finally, we computed Pearson correlations between the recovered learning rates parameters and the input learning rates parameters 30 . The results indicated that we successfully recovered the parameters for the two-learning-rate model. Lastly, we calculated the model-based efficacy estimate of each round for each participant based on the parameters from the two-learning-rate model ( \(\:{\alpha\:}_{pos}\) , \(\:{\alpha\:}_{neg}\) , initial value). The model began with an initial value ( \(\:{E}_{1}\) ) and was updated using the learning rates ( \(\:{\alpha\:}_{pos}\) , \(\:{\alpha\:}_{neg}\) ) and the prediction error ( \(\:{\delta\:}_{t}\) ). Learning models were fitted, compared, and recovered by MATLAB2018b , while the model-based efficacy estimates were calculated in R-Studio. Linear Mixed Model LMMs were performed in R-Studio using the lme4 package for modeling 31 and the lmerTest package for fixed-effects testing and p- value estimation 32 . Standardized regression coefficients and 95% confidence intervals were derived using the model_parameters function from the parameters package. Statistical significance was set at p < 0.05. Hierarchical Drift Diffusion Model All models utilized 5 Markov Chain Monte Carlo (MCMC) chains, drawing 7,000 samples and discarding the first 4,000 samples as burn-in for each chain. Model convergence was assessed using the Gelman-Rubin statistic ( \(\:\widehat{\text{R}}\) ); all model \(\:\widehat{\text{R}}\) values close to 1 and below 1.1 indicated good convergence and minimal variation between chains 33 . Model comparison was performed using the deviance information criterion (DIC), where lower values indicate a better fit, to select the optimal model 34 . Posterior predictive checks were performed to evaluate the ability of the optimal model to reproduce the observed data. Specifically, 500 parameter samples were drawn from the posterior distribution of the fitted model, and these samples were used to simulate new datasets for comparison with the observed data. Across all optimal models, the statistics of observed data fell within the 95% credible intervals of the statistics derived from the simulated datasets, indicating that each model adequately captured the patterns in the observed data. Model fit, comparison and posterior predictive checks were performed using the HDDM package in Python ( https://secure.travis-ci.org/hddm-devs/hddm.png?branch=master ). Bayesian hypothesis tests were conducted separately in R-Studio using the brms package. Multilevel Mediation Model The multilevel mediation analyses were performed using the MLmed macro ( https://njrockwood.com/mlmed ) in SPSS 25.00. Model estimation was based on Restricted Maximum Likelihood (REML). Indirect effects were assessed via Monte Carlo confidence intervals with 10,000 simulated samples and considered statistically significant at p < .05 when the 95% Monte Carlo confidence interval (CI) did not include zero. Declarations Competing interests All authors declare no conflict of interests. Data a vailability All data used in this study are publicly available on the Open Science Framework (OSF): https://osf.io/vpr9e/. Code availability All analysis code used in this study are publicly available on the Open Science Framework (OSF): https://osf.io/vpr9e/. References Corcoran, K., Crusius, J. & Mussweiler, T. Social comparison: Motives, standards, and mechanisms. in Theories in Social Psychology 119–139 (Wiley-Blackwell, 2011). Festinger, L. A theory of social comparison processes. Hum. Relat. 7 , 117–140 (1954). Diel, K., Grelle, S. & Hofmann, W. A motivational framework of social comparison. J. Pers. Soc. Psychol. 120 , 1415–1430 (2021). Gerber, J. P., Wheeler, L. & Suls, J. A social comparison theory meta-analysis 60+ years on. Psychol. Bull. 144 , 177–197 (2018). Bandura, A. Self-efficacy: Toward a unifying theory of behavioral change. Psychol. Rev. 84 , 191–215 (1977). Bandura, A. Self Efficacy: The Exercise of Control . (W. H. Freeman, New York (N.Y.), 1997). Müller-Pinzler, L. et al. Neurocomputational mechanisms of affected beliefs. Commun. Biol. 5 , 1241 (2022). Boksem, M. A. S., Kostermans, E. & De Cremer, D. Failing where others have succeeded: Medial Frontal Negativity tracks failure in a social context. Psychophysiology 48, 973–979 (2011). Lin, H. & Liang, J. Working memory load reduces the processing of outcome evaluation involving others but not oneself: Event‐related potential evidence. Psychophysiology 58 , (2021). Wu, Y., Zhang, D., Elieson, B. & Zhou, X. Brain potentials in outcome evaluation: When social comparison takes effect. Int. J. Psychophysiol. 85 , 145–152 (2012). Du, X. et al. The neural circuitry of reward processing in complex social comparison: evidence from an event-related FMRI study. PLoS ONE 8 , e82534 (2013). Hu, X., Xu, Z. & Mai, X. Social value orientation modulates the processing of outcome evaluation involving others. Soc. Cogn. Affect. Neurosci. 12 , 1730–1739 (2017). Li, Q. et al. Social comparisons differentially affect fair and unfair processing in ultimatum bargaining. Neuropsychologia 174 , 108318 (2022). Luo, Y. et al. Social Comparison Manifests in Event-related Potentials. Sci. Rep. 5 , 12127 (2015). Peng, M. et al. Good luck or bad luck? The influence of social comparison on risk‐taking decision and the underlying neural mechanism. Psychophysiology e14730 (2024) doi:10.1111/psyp.14730. Zhang, H. et al. Interpersonal relationships modulate outcome evaluation in a social comparison context: The pain and pleasure of intimacy. Cogn. Affect. Behav. Neurosci. 20 , 115–127 (2020). Lindner, M. et al. Neural patterns underlying social comparisons of personal performance. Soc. Cogn. Affect. Neurosci. 10 , 569–576 (2015). Shen, Q. et al. To Reveal or Not to Reveal? Observation of Social Outcomes Facilitates Reward Processing. Front. Neurosci. 14 , 579702 (2021). Wang, L., Zhang, X., Li, L. & Meng, L. Taking Others as a Mirror: Contingent Social Comparison Promotes Task Engagement. Front. Hum. Neurosci. 12 , 476 (2018). Schröder, A. et al. Computational Modeling shows Confirmation Bias during Formation and Revision of Self-Beliefs. Preprint at https://doi.org/10.1101/2024.08.30.610443 (2024). Wittmann, M. K. et al. Self-Other Mergence in the Frontal Cortex during Cooperation and Competition. Neuron 91 , 482–493 (2016). Müller-Pinzler, L. et al. Negativity-bias in forming beliefs about own abilities. Sci. Rep. 9 , 14416 (2019). Bault, N., Joffily, M., Rustichini, A. & Coricelli, G. Medial prefrontal cortex and striatum mediate the influence of social comparison on the decision process. Proc. Natl. Acad. Sci. 108 , 16044–16049 (2011). Liu, Z. et al. Social comparison modulates the neural responses to regret and subsequent risk-taking behavior. Soc. Cogn. Affect. Neurosci. 13 , 1059–1070 (2018). Tor, A. & Garcia, S. M. The neuroscience of social comparison and competition. Cogn. Affect. Behav. Neurosci. 23 , 920–943 (2023). Buunk, B. P., Collins, R. L., Taylor, S. E., VanYperen, N. W. & Dakof, G. A. The affective consequences of social comparison: either direction has its ups and downs. J. Pers. Soc. Psychol. 59 , 1238–1249 (1990). Diel, K., Hofmann, W., Grelle, S., Boecker, L. & Friese, M. Prepare to Compare: Effects of an Intervention Involving Upward and Downward Social Comparisons on Goal Pursuit in Daily Life. Pers. Soc. Psychol. Bull. 01461672231219378 (2024) doi:10.1177/01461672231219378. Leng, X., Yee, D., Ritz, H. & Shenhav, A. Dissociable influences of reward and punishment on adaptive cognitive control. PLOS Comput. Biol. 17 , e1009737 (2021). Zhang, L., Lengersdorff, L., Mikus, N., Gläscher, J. & Lamm, C. Using reinforcement learning models in social neuroscience: frameworks, pitfalls and suggestions of best practices. Soc. Cogn. Affect. Neurosci. 15 , 695–707 (2020). Wilson, R. C. & Collins, A. G. Ten simple rules for the computational modeling of behavioral data. eLife 8 , e49547 (2019). Bates, D., Mächler, M., Bolker, B. & Walker, S. Fitting Linear Mixed-Effects Models Using lme4 . J. Stat. Softw. 67 , (2015). Kuznetsova, A., Brockhoff, P. B. & Christensen, R. H. B. lmerTest Package: Tests in Linear Mixed Effects Models. J. Stat. Softw. 82 , (2017). Pedersen, M. L. & Frank, M. J. Simultaneous Hierarchical Bayesian Parameter Estimation for Reinforcement Learning and Drift Diffusion Models: a Tutorial and Links to Neural Data. Comput. Brain Behav. 3 , 458–471 (2020). Spiegelhalter, D. J., Best, N. G., Carlin, B. P. & Van Der Linde, A. Bayesian Measures of Model Complexity and Fit. J. R. Stat. Soc. Ser. B Stat. Methodol. 64 , 583–639 (2002). Faul, F., Erdfelder, E., Lang, A.-G. & Buchner, A. G*Power 3: A flexible statistical power analysis program for the social, behavioral, and biomedical sciences. Behav. Res. Methods 39 , 175–191 (2007). Luck, S. J. & Gaspelin, N. How to get statistically significant effects in any ERP experiment (and why you shouldn’t). Psychophysiology 54 , 146–157 (2017). Glazer, J. E., Kelley, N. J., Pornpattananangkul, N., Mittal, V. A. & Nusslock, R. Beyond the FRN: Broadening the time-course of EEG and ERP components implicated in reward processing. Int. J. Psychophysiol. 132 , 184–202 (2018). Zhang, Y. et al. Comparing monetary gain and loss in the monetary incentive delay task: EEG evidence. Psychophysiology 60 , e14383 (2023). Frömer, R., Lin, H., Dean Wolf, C. K., Inzlicht, M. & Shenhav, A. Expectations of reward and efficacy guide cognitive control allocation. Nat. Commun. 12 , 1030 (2021). Rong, Y. et al. Expectations of immediate and delayed reward differentially affect cognitive task performance. NeuroImage 262 , 119582 (2022). Schevernels, H., Krebs, R. M., Santens, P., Woldorff, M. G. & Boehler, C. N. Task preparation processes related to reward prediction precede those related to task-difficulty expectation. NeuroImage 84 , 639–647 (2014). Gable, P. A., Threadgill, A. H. & Adams, D. L. Neural activity underlying motor-action preparation and cognitive narrowing in approach-motivated goal states. Cogn. Affect. Behav. Neurosci. 16 , 145–152 (2016). Lin, H. & Liang, J. Comparison with others influences encoding and recognition of their faces: Behavioural and ERP evidence. NeuroImage 288 , 120538 (2024). Fliessbach, K. et al. Social comparison affects reward-related brain activity in the human ventral striatum. Science 318 , 1305–1308 (2007). Zou, F. et al. P2 Manifests Subjective Evaluation of Reward Processing Under Social Comparison. Front. Psychol. 13 , 817529 (2022). Luo, Y., Eickhoff, S. B., Hétu, S. & Feng, C. Social comparison in the brain: A coordinate-based meta-analysis of functional brain imaging studies on the downward and upward comparisons. Hum. Brain Mapp. 39 , 440–458 (2018). Scollon, C. N., Kim-Prieto, C. & Scollon, C. N. Experience Sampling: Promises and Pitfalls, Strengths and Weaknesses. J. Happiness Stud. 4 , 5–34 (2003). Sheeran, P. Intention–Behavior Relations: A Conceptual and Empirical Review. in European Review of Social Psychology (eds Stroebe, W. & Hewstone, M.) 1–36 (Wiley, 2001). doi:10.1002/0470013478.ch1. Grahek, I., Frömer, R., Prater Fahey, M. & Shenhav, A. Learning when effort matters: neural dynamics underlying updating and adaptation to changes in performance efficacy. Cereb. Cortex bhac215 (2022) doi:10.1093/cercor/bhac215. Zhou, T. et al. History of winning remodels thalamo-PFC circuit to reinforce social dominance. Science 357 , 162–168 (2017). Bonabeau, E. Dominance Orders in Animal Societies: The Self-organization Hypothesis Revisited. Bull. Math. Biol. 61 , 727–757 (1999). Page, L. & Coates, J. Winner and loser effects in human competitions. Evidence from equally matched tennis players. Evol. Hum. Behav. 38 , 530–535 (2017). Additional Declarations There is NO Competing Interest. Supplementary Files RS1255.pdf Reporting Summary Cite Share Download PDF Status: Under Review Version 1 posted You are reading this latest preprint version Research Square lets you share your work early, gain feedback from the community, and start making changes to your manuscript prior to peer review in a journal. As a division of Research Square Company, we’re committed to making research communication faster, fairer, and more useful. We do this by developing innovative software and high quality services for the global research community. Our growing team is made up of researchers and industry professionals working together to solve the most critical problems facing scientific publishing. Also discoverable on Platform About Our Team In Review Editorial Policies Advisory Board Help Center Resources Author Services Accessibility API Access RSS feed Manage Cookie Preferences © Research Square 2026 | ISSN 2693-5015 (online) Privacy Policy Terms of Service Do Not Sell My Personal Information {"props":{"pageProps":{"initialData":{"identity":"rs-7749786","acceptedTermsAndConditions":true,"allowDirectSubmit":false,"archivedVersions":[],"articleType":"Article","associatedPublications":[],"authors":[{"id":537504996,"identity":"7ef93a10-e440-4845-906e-f80288ecb885","order_by":0,"name":"Ping Wei","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAA0klEQVRIiWNgGAWjYJCCAwwGDAz8IFZCASlaJBtAWgxIscrgAJgkQiX/jOSDhwsK7thtPr868cMDAwZ5frED+LVI3EhLODzD4FnythtvN0sAHWY4c3YCAWtu5xgc5jE4nGx24+wGkJYEg9sEtMjfzv8A1mI84+zmH0RpMbidwwDSYmfA37uNOFsM7z8zAPrlcILEDd5tFgkGEoT9Infm8OPPBX8O2/P3n91880eFjTy/NAEtIMAMxIkNEmCVEoSVw7TYM/AfIE71KBgFo2AUjDwAANZiSlgy7vrWAAAAAElFTkSuQmCC","orcid":"https://orcid.org/0000-0002-3781-3844","institution":"Capital Normal University","correspondingAuthor":true,"prefix":"","firstName":"Ping","middleName":"","lastName":"Wei","suffix":""},{"id":537504997,"identity":"35a48696-ad15-4891-8ac7-066977aec703","order_by":1,"name":"Jiarui Dong","email":"","orcid":"","institution":"","correspondingAuthor":false,"prefix":"","firstName":"Jiarui","middleName":"","lastName":"Dong","suffix":""},{"id":537504998,"identity":"a8851c5f-7ec9-4f63-a5b9-5ae544bb33e7","order_by":2,"name":"Yachao Rong","email":"","orcid":"","institution":"","correspondingAuthor":false,"prefix":"","firstName":"Yachao","middleName":"","lastName":"Rong","suffix":""},{"id":537504999,"identity":"d3a167f4-6a72-4bf9-9fbb-f7e60f09fc62","order_by":3,"name":"Shengjie Ma","email":"","orcid":"","institution":"","correspondingAuthor":false,"prefix":"","firstName":"Shengjie","middleName":"","lastName":"Ma","suffix":""},{"id":537505000,"identity":"b0575355-2193-48b1-b84b-cfc3354c46de","order_by":4,"name":"Yang Xu","email":"","orcid":"","institution":"","correspondingAuthor":false,"prefix":"","firstName":"Yang","middleName":"","lastName":"Xu","suffix":""}],"badges":[],"createdAt":"2025-09-30 09:27:38","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-7749786/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-7749786/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":95798415,"identity":"dfdf23ba-5e99-4b97-9f9f-d07afe4ffff9","added_by":"auto","created_at":"2025-11-13 08:16:44","extension":"docx","order_by":0,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":2677397,"visible":true,"origin":"","legend":"","description":"","filename":"Theneurocomputationalmechanismsunderlyingtheimpactofsocialcomparisononeffortinvestment.docx","url":"https://assets-eu.researchsquare.com/files/rs-7749786/v1/29878998dca4dbc229154f1a.docx"},{"id":95797332,"identity":"4ade589c-3255-4e72-bcc1-dd0df028a72e","added_by":"auto","created_at":"2025-11-13 08:03:38","extension":"tif","order_by":1,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":7876452,"visible":true,"origin":"","legend":"","description":"","filename":"Figure1.tif","url":"https://assets-eu.researchsquare.com/files/rs-7749786/v1/01215c0daf87749a66e19f60.tif"},{"id":95663047,"identity":"7564cb9b-4e09-4d12-a01d-625fa61ab7b5","added_by":"auto","created_at":"2025-11-11 16:38:22","extension":"tif","order_by":2,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":6712868,"visible":true,"origin":"","legend":"","description":"","filename":"Figure2.tif","url":"https://assets-eu.researchsquare.com/files/rs-7749786/v1/812befbdefc0fb7e4dae8ada.tif"},{"id":95663094,"identity":"27b680b6-c0ec-489d-a4da-77e5eb5c1c31","added_by":"auto","created_at":"2025-11-11 16:38:23","extension":"tif","order_by":3,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":17551072,"visible":true,"origin":"","legend":"","description":"","filename":"Figure3.tif","url":"https://assets-eu.researchsquare.com/files/rs-7749786/v1/de8821479f0c35d237207144.tif"},{"id":95663098,"identity":"42ff2bcf-5980-4965-9b74-c9396771b03b","added_by":"auto","created_at":"2025-11-11 16:38:23","extension":"tif","order_by":4,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":19818508,"visible":true,"origin":"","legend":"","description":"","filename":"Figure4.tif","url":"https://assets-eu.researchsquare.com/files/rs-7749786/v1/28943471481c3a8a24435220.tif"},{"id":95797336,"identity":"cbd01db7-8c0c-411b-912e-100e2e123750","added_by":"auto","created_at":"2025-11-13 08:03:41","extension":"tif","order_by":5,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":7098180,"visible":true,"origin":"","legend":"","description":"","filename":"Figure5.tif","url":"https://assets-eu.researchsquare.com/files/rs-7749786/v1/1da6c4f0dce3678b855ca892.tif"},{"id":95663090,"identity":"ce1a1cbd-7280-4b6a-8b2f-7293af19f7bb","added_by":"auto","created_at":"2025-11-11 16:38:23","extension":"tif","order_by":6,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":7344308,"visible":true,"origin":"","legend":"","description":"","filename":"Figure6.tif","url":"https://assets-eu.researchsquare.com/files/rs-7749786/v1/8f3aee96d7c92766891b2399.tif"},{"id":95797323,"identity":"c80288c6-7514-4adc-92e7-c0df2cb552bc","added_by":"auto","created_at":"2025-11-13 08:03:24","extension":"json","order_by":7,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":5927,"visible":true,"origin":"","legend":"","description":"","filename":"COMMSBIO259544.json","url":"https://assets-eu.researchsquare.com/files/rs-7749786/v1/3fe955688edd79a53f494a75.json"},{"id":95798537,"identity":"93572dd2-0955-4409-8e59-7d139b282f12","added_by":"auto","created_at":"2025-11-13 08:16:58","extension":"xml","order_by":8,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":143682,"visible":true,"origin":"","legend":"","description":"","filename":"COMMSBIO2595440enriched.xml","url":"https://assets-eu.researchsquare.com/files/rs-7749786/v1/32a145c7d8194b4f42f66f50.xml"},{"id":95663085,"identity":"4ba1c3c2-e5dc-4d94-95c8-067ff26d8a5e","added_by":"auto","created_at":"2025-11-11 16:38:22","extension":"tif","order_by":9,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":7876452,"visible":true,"origin":"","legend":"","description":"","filename":"Figure1.tif","url":"https://assets-eu.researchsquare.com/files/rs-7749786/v1/a40fa2eee64db764504eb50f.tif"},{"id":95662821,"identity":"f752664a-94df-4adc-a1a3-5e8d411e0f4e","added_by":"auto","created_at":"2025-11-11 16:38:15","extension":"tif","order_by":10,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":6712868,"visible":true,"origin":"","legend":"","description":"","filename":"Figure2.tif","url":"https://assets-eu.researchsquare.com/files/rs-7749786/v1/3475f53a47ae680caf47253c.tif"},{"id":95662807,"identity":"e9b4aa86-23bc-4971-9c2c-d5cfa045c2e0","added_by":"auto","created_at":"2025-11-11 16:38:13","extension":"tif","order_by":11,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":17551072,"visible":true,"origin":"","legend":"","description":"","filename":"Figure3.tif","url":"https://assets-eu.researchsquare.com/files/rs-7749786/v1/fb5bdb4f8dbf09669ed2f48b.tif"},{"id":95662990,"identity":"146067e8-d723-489f-8381-a94f54d1cce2","added_by":"auto","created_at":"2025-11-11 16:38:19","extension":"tif","order_by":12,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":19818508,"visible":true,"origin":"","legend":"","description":"","filename":"Figure4.tif","url":"https://assets-eu.researchsquare.com/files/rs-7749786/v1/037d136aa27712ecf69e0bf6.tif"},{"id":95663172,"identity":"b141cb0a-ac56-4340-9ecf-e43273512aeb","added_by":"auto","created_at":"2025-11-11 16:38:27","extension":"tif","order_by":13,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":7098180,"visible":true,"origin":"","legend":"","description":"","filename":"Figure5.tif","url":"https://assets-eu.researchsquare.com/files/rs-7749786/v1/fdd5b60b50f51711e843b5e5.tif"},{"id":95663014,"identity":"c913f88c-0061-4aa2-99e4-d5ab3224b60c","added_by":"auto","created_at":"2025-11-11 16:38:21","extension":"tif","order_by":14,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":7344308,"visible":true,"origin":"","legend":"","description":"","filename":"Figure6.tif","url":"https://assets-eu.researchsquare.com/files/rs-7749786/v1/11ad855f31b18b0371afcde9.tif"},{"id":95663096,"identity":"bfdcbce3-a185-486a-a256-556f52b7219e","added_by":"auto","created_at":"2025-11-11 16:38:23","extension":"jpeg","order_by":15,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":185248,"visible":true,"origin":"","legend":"","description":"","filename":"floatimage1.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-7749786/v1/de6c8053452017d940ec49ca.jpeg"},{"id":95662871,"identity":"31fd38f2-0142-4a6d-8645-53e1f063965a","added_by":"auto","created_at":"2025-11-11 16:38:16","extension":"jpeg","order_by":16,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":280072,"visible":true,"origin":"","legend":"","description":"","filename":"floatimage2.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-7749786/v1/4da24c489f2f712f2b08588c.jpeg"},{"id":95663093,"identity":"3ccb637d-0bb9-45de-b147-08bf264ca237","added_by":"auto","created_at":"2025-11-11 16:38:23","extension":"jpeg","order_by":17,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":637278,"visible":true,"origin":"","legend":"","description":"","filename":"floatimage3.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-7749786/v1/086f4837eea5b5ccbca0ef4d.jpeg"},{"id":95663115,"identity":"8427c55d-da44-4807-87cf-34a30cec9089","added_by":"auto","created_at":"2025-11-11 16:38:24","extension":"jpeg","order_by":18,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":1074310,"visible":true,"origin":"","legend":"","description":"","filename":"floatimage4.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-7749786/v1/01b34fb0d14a4cb8dadc8c0a.jpeg"},{"id":95663089,"identity":"a75fc768-6c42-4a9e-b2f9-78382c342639","added_by":"auto","created_at":"2025-11-11 16:38:23","extension":"jpeg","order_by":19,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":344366,"visible":true,"origin":"","legend":"","description":"","filename":"floatimage5.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-7749786/v1/8b8cf6735ef873d90088793c.jpeg"},{"id":95797459,"identity":"0a0f023b-c8c0-459d-947c-fc9d83368608","added_by":"auto","created_at":"2025-11-13 08:05:34","extension":"jpeg","order_by":20,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":435424,"visible":true,"origin":"","legend":"","description":"","filename":"floatimage6.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-7749786/v1/e80d3fd859242b17c75a5c78.jpeg"},{"id":95663045,"identity":"03494e43-f26e-4a71-a4b2-59a357def149","added_by":"auto","created_at":"2025-11-11 16:38:22","extension":"png","order_by":21,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":36799,"visible":true,"origin":"","legend":"","description":"","filename":"OnlineFigure1.png","url":"https://assets-eu.researchsquare.com/files/rs-7749786/v1/c41867d5e94287b0bbd159b1.png"},{"id":95662815,"identity":"d37b46d3-9927-4a3f-b282-eeb54b1f3fc0","added_by":"auto","created_at":"2025-11-11 16:38:14","extension":"png","order_by":22,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":230263,"visible":true,"origin":"","legend":"","description":"","filename":"OnlineFigure2.png","url":"https://assets-eu.researchsquare.com/files/rs-7749786/v1/c33e9f3b9cf61ee7b186ad3e.png"},{"id":95662917,"identity":"598180bc-81bc-4bf8-a768-0d2dca6abb0a","added_by":"auto","created_at":"2025-11-11 16:38:18","extension":"png","order_by":23,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":652903,"visible":true,"origin":"","legend":"","description":"","filename":"OnlineFigure3.png","url":"https://assets-eu.researchsquare.com/files/rs-7749786/v1/aa5f50679247401d864b3bcd.png"},{"id":95662991,"identity":"7cf0e328-7826-4a4a-afa9-6a2b93364851","added_by":"auto","created_at":"2025-11-11 16:38:20","extension":"png","order_by":24,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":1134565,"visible":true,"origin":"","legend":"","description":"","filename":"OnlineFigure4.png","url":"https://assets-eu.researchsquare.com/files/rs-7749786/v1/7f46f5cac44453cc6cc9dff1.png"},{"id":95663044,"identity":"3fa97079-ead5-4f2f-a670-3ad43e5e56f3","added_by":"auto","created_at":"2025-11-11 16:38:22","extension":"png","order_by":25,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":465785,"visible":true,"origin":"","legend":"","description":"","filename":"OnlineFigure5.png","url":"https://assets-eu.researchsquare.com/files/rs-7749786/v1/4703f2ca2c310f3b31484164.png"},{"id":95663179,"identity":"90007e00-bea9-4a45-b8d6-702865532879","added_by":"auto","created_at":"2025-11-11 16:38:28","extension":"png","order_by":26,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":510181,"visible":true,"origin":"","legend":"","description":"","filename":"OnlineFigure6.png","url":"https://assets-eu.researchsquare.com/files/rs-7749786/v1/e4f1b8235fd9e30c76504630.png"},{"id":95663088,"identity":"776fb829-d135-4616-a186-10a1027c8517","added_by":"auto","created_at":"2025-11-11 16:38:22","extension":"png","order_by":27,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":28157,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage1.png","url":"https://assets-eu.researchsquare.com/files/rs-7749786/v1/8bf6b2323c1361fcf6d456ae.png"},{"id":95663080,"identity":"5fa63666-54fa-4b10-acef-15ea1d3fd915","added_by":"auto","created_at":"2025-11-11 16:38:22","extension":"png","order_by":28,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":48560,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage2.png","url":"https://assets-eu.researchsquare.com/files/rs-7749786/v1/805d04b00f4b828c44d66525.png"},{"id":95663042,"identity":"9756ea2f-8cfd-4fc1-803a-fa0966d5139e","added_by":"auto","created_at":"2025-11-11 16:38:21","extension":"png","order_by":29,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":117059,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage3.png","url":"https://assets-eu.researchsquare.com/files/rs-7749786/v1/ce7f4dbd1f7dbeff8cef2aad.png"},{"id":95663043,"identity":"f11d3688-6f9e-42c1-b68e-20564db2372a","added_by":"auto","created_at":"2025-11-11 16:38:21","extension":"png","order_by":30,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":176220,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage4.png","url":"https://assets-eu.researchsquare.com/files/rs-7749786/v1/7c6b7f05949c7bc6ddcb9ff5.png"},{"id":95662808,"identity":"f58f700c-855f-4d08-976c-bf23f03b64fe","added_by":"auto","created_at":"2025-11-11 16:38:13","extension":"png","order_by":31,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":59968,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage5.png","url":"https://assets-eu.researchsquare.com/files/rs-7749786/v1/a123409d2ae4796c316e0932.png"},{"id":95663095,"identity":"57f4ba29-7bc9-4339-bd0e-5cba0e3c0d54","added_by":"auto","created_at":"2025-11-11 16:38:23","extension":"png","order_by":32,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":59153,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage6.png","url":"https://assets-eu.researchsquare.com/files/rs-7749786/v1/bb5908c554b7b300625ca31b.png"},{"id":95663146,"identity":"c962c1e8-3532-42d2-beeb-da163b8777cc","added_by":"auto","created_at":"2025-11-11 16:38:25","extension":"xml","order_by":33,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":140007,"visible":true,"origin":"","legend":"","description":"","filename":"COMMSBIO2595440structuring.xml","url":"https://assets-eu.researchsquare.com/files/rs-7749786/v1/d9ca8c6b948b51de1ec0ad7a.xml"},{"id":95663041,"identity":"f33f99e9-b54e-4057-bfe4-d3ec99286833","added_by":"auto","created_at":"2025-11-11 16:38:21","extension":"html","order_by":34,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":157314,"visible":true,"origin":"","legend":"","description":"","filename":"earlyproof.html","url":"https://assets-eu.researchsquare.com/files/rs-7749786/v1/559a7d44d662ba25025c9a80.html"},{"id":95663100,"identity":"197bb0cc-9f32-448f-aa74-54f7b844e982","added_by":"auto","created_at":"2025-11-11 16:38:23","extension":"png","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":278762,"visible":true,"origin":"","legend":"\u003cp\u003eExample stimuli and round structure. Each round began with a triangle cue stimulus, feedback was indicated by up arrows, bidirectional arrows, and down arrows, which indicated upward, lateral, and downward social comparisons (a), respectively. Within the designated time window, participants’ correct responses per second (CRPS) and reaction times (RTs) were used as indicators of their effort investment. After every 3-5 rounds, participants rated the probability of winning the next round based on their past performance (b).\u003c/p\u003e","description":"","filename":"Figure1.png","url":"https://assets-eu.researchsquare.com/files/rs-7749786/v1/f65919c6f73887089f35f8e8.png"},{"id":95662996,"identity":"e09d93ab-5308-437f-9e91-ed87d1044649","added_by":"auto","created_at":"2025-11-11 16:38:20","extension":"png","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":490968,"visible":true,"origin":"","legend":"\u003cp\u003eBehavioral results for Experiment 1. Mean values of rating scores (a) and efficacy estimates (c) are shown for the downward and upward social comparison conditions. Each dot represents a participant’s value in the two conditions. The fixed effect for efficacy estimates on rating scores is shown in (b). Error bars and the shaded areas indicate the standard error of the mean (SEM). *\u003cem\u003ep \u003c/em\u003e\u0026lt; .05, ** \u003cem\u003ep\u003c/em\u003e \u0026lt; .01, *** \u003cem\u003ep\u003c/em\u003e \u0026lt; .001.\u003c/p\u003e","description":"","filename":"Figure2.png","url":"https://assets-eu.researchsquare.com/files/rs-7749786/v1/c0ca8cb9733e81a1fafd1310.png"},{"id":95663063,"identity":"7113fec1-ba1c-4f72-819f-d76230890df6","added_by":"auto","created_at":"2025-11-11 16:38:22","extension":"png","order_by":3,"title":"Figure 3","display":"","copyAsset":false,"role":"figure","size":1550372,"visible":true,"origin":"","legend":"\u003cp\u003eBehavioral results for Experiment 1. Mean correct responses per second and reaction times fordownward and upward social comparison conditions are shown in (a). Each dot represents each participant’s average. Error bars represent the standard error of the mean (SEM). Fixed effects of efficacy estimates on correct responses per second and reaction times are shown in (b), with shaded areas indicating the SEM. Posterior probability density of drift rates and thresholds for downward social comparison and efficacy estimates are shown in (c). Results of 1-1-1 multilevel mediation models for correct responses per second and reaction times are displayed in (d). *\u003cem\u003ep \u003c/em\u003e\u0026lt; .05, ** \u003cem\u003ep\u003c/em\u003e \u0026lt; .01, *** \u003cem\u003ep\u003c/em\u003e \u0026lt; .001.\u003c/p\u003e","description":"","filename":"Figure3.png","url":"https://assets-eu.researchsquare.com/files/rs-7749786/v1/a2cd5ea5665d27dcf5267aa4.png"},{"id":95663097,"identity":"22acd3e6-8372-45bf-9be8-2ede43cbad76","added_by":"auto","created_at":"2025-11-11 16:38:23","extension":"png","order_by":4,"title":"Figure 4","display":"","copyAsset":false,"role":"figure","size":2570730,"visible":true,"origin":"","legend":"\u003cp\u003eFeedback-locked ERP results. Grand-averaged ERP waveforms and topographic maps of the RewP (a), fb-P3 (c), and LPP (e) components for the two social comparison conditions are shown in (a), (b), and (c), respectively. Shaded areas indicate the time window for the RewP (280-340 ms), fb-P3 (350-450 ms), and LPP (600-1000 ms) components. Mean amplitudes of RewP, fb-P3, and LPP components for the two social comparison conditions are shown in (b), (d), and (f), respectively. Error bars indicate standard error of the mean (SEM). *\u003cem\u003ep \u003c/em\u003e\u0026lt; .05, ** \u003cem\u003ep\u003c/em\u003e \u0026lt; .01, *** \u003cem\u003ep\u003c/em\u003e \u0026lt; .001.\u003c/p\u003e","description":"","filename":"Figure4.png","url":"https://assets-eu.researchsquare.com/files/rs-7749786/v1/fdfee689a1d4ac80309d6d31.png"},{"id":95662992,"identity":"6c4aa526-16bd-44c4-87a2-f1996605250b","added_by":"auto","created_at":"2025-11-11 16:38:20","extension":"png","order_by":5,"title":"Figure 5","display":"","copyAsset":false,"role":"figure","size":1082582,"visible":true,"origin":"","legend":"\u003cp\u003eCue ERP results. Grand-averaged ERP waveforms and topographic maps of the CNV components for the two social comparison conditions is shown in (a). The shaded area indicates the time window for the CNV (550-950 ms). Mean power of the CNV component in downward and upward social comparison conditions is shown in (b), error bars indicate the standard error of the mean (SEM). The fixed effect of efficacy on the CNV component is displayed in (c), with shaded areas indicating the standard error of the mean (SEM). *\u003cem\u003ep \u003c/em\u003e\u0026lt; .05, ** \u003cem\u003ep\u003c/em\u003e \u0026lt; .01, *** \u003cem\u003ep\u003c/em\u003e \u0026lt; .001.\u003c/p\u003e","description":"","filename":"Figure5.png","url":"https://assets-eu.researchsquare.com/files/rs-7749786/v1/24a26f7cb52d8c62d2f716c5.png"},{"id":95662812,"identity":"297b470b-02e3-43b8-9fe4-7a91ad4f1845","added_by":"auto","created_at":"2025-11-11 16:38:13","extension":"png","order_by":6,"title":"Figure 6","display":"","copyAsset":false,"role":"figure","size":1217301,"visible":true,"origin":"","legend":"\u003cp\u003eCue-locked ERSP results.\u003cem\u003e \u003c/em\u003eGrand-averaged ERSP power (extracted from F3/F1/Fz electrodes) and topographic maps of cue-beta (150-600 ms post-cue onsed; frequency range: 15-21 hz; indicated by black boxes) for downward and upward social comparison conditions are shown in (a). Mean cue-beta band power for the two conditions is displayed in (b); error bars indicate the standard error of the mean (SEM). Fixed effects of model-based efficacy estimates on cue-beta band power is displayed in (c); shaded areas indicate the SEM. *\u003cem\u003ep \u003c/em\u003e\u0026lt; .05, ** \u003cem\u003ep\u003c/em\u003e \u0026lt; .01, *** \u003cem\u003ep\u003c/em\u003e \u0026lt; .001.\u003c/p\u003e","description":"","filename":"Figure6.png","url":"https://assets-eu.researchsquare.com/files/rs-7749786/v1/3807e34e58839f20f0481e5e.png"},{"id":95804514,"identity":"f937aa72-279f-411e-8ff4-d43f8c9d9f0f","added_by":"auto","created_at":"2025-11-13 08:37:38","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":8169848,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-7749786/v1/08a4b1a9-7be7-40de-912f-65ae1af269c4.pdf"},{"id":95663147,"identity":"505e2d97-e84c-4404-9bf3-00b3f2af5b56","added_by":"auto","created_at":"2025-11-11 16:38:25","extension":"pdf","order_by":1,"title":"","display":"","copyAsset":false,"role":"supplement","size":2153489,"visible":true,"origin":"","legend":"Reporting Summary","description":"","filename":"RS1255.pdf","url":"https://assets-eu.researchsquare.com/files/rs-7749786/v1/9d790e617ea452ce3691fb97.pdf"}],"financialInterests":"There is \u003cb\u003eNO\u003c/b\u003e Competing Interest.","formattedTitle":"The neurocomputational mechanisms underlying the impact of social comparison on effort investment","fulltext":[{"header":"Introduction","content":"\u003cp\u003eSocial comparison is a pervasive psychological phenomenon that shapes individual\u0026rsquo;s judgments, emotions, and goal-dircted behaviors \u003csup\u003e\u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e1\u003c/span\u003e\u003c/sup\u003e. Rooted in social comparison theory, people inherently evaluate their abilities and self-worth against others, particularly when objective assessment criteria are absent \u003csup\u003e\u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2\u003c/span\u003e\u003c/sup\u003e. Among its various effects, social comparison may exert the greatest influence on individual\u0026rsquo;s self-perception of ability, providing an immediate and efficient mechanism for assessing their abilities relative to others \u003csup\u003e\u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e3\u003c/span\u003e,\u003cspan citationid=\"CR4\" class=\"CitationRef\"\u003e4\u003c/span\u003e\u003c/sup\u003e. These self-perceptions of ability, in turn, underpin individual\u0026rsquo;s willingness to invest effort in behaviors needed to achieve desired outcomes \u003csup\u003e\u003cspan additionalcitationids=\"CR6\" citationid=\"CR5\" class=\"CitationRef\"\u003e5\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR7\" class=\"CitationRef\"\u003e7\u003c/span\u003e\u003c/sup\u003e. Theoretically, social comparison influences goal-directed behavior via a dynamic, multi-stage process: eveluating comparison outcomes, updating self-evaluations of abilities, and guiding subsequent effort allocation.\u003c/p\u003e\u003cp\u003eSocial comparison initiates a process where individuals first evaluate discrepancies between themselves and others, perceiving either superiority (downward comparison) or inferiority (upward comparison). Electrophysiological studies consistently link this evaluation to reward- and feedback-sensitive neural signals, including reward-related positivity (RewP), feedback-related potentials (fb-P3), and late positive potential (LPP). These studies show stronger neural responses to downward compared to upward social comparison feedback \u003csup\u003e\u003cspan additionalcitationids=\"CR9\" citationid=\"CR8\" class=\"CitationRef\"\u003e8\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR10\" class=\"CitationRef\"\u003e10\u003c/span\u003e\u003c/sup\u003e, suggesting that individuals perceive downward comparisons as rewarding and experience positive emotions when affirming their superiority.\u003c/p\u003e\u003cp\u003eCritically, most of these studies have employed gambling tasks with monetary payoffs as the comparison dimension \u003csup\u003e\u003cspan additionalcitationids=\"CR12 CR13 CR14 CR15\" citationid=\"CR11\" class=\"CitationRef\"\u003e11\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR16\" class=\"CitationRef\"\u003e16\u003c/span\u003e\u003c/sup\u003e, introducing confounding between social comparison evaluation and the valuation of monetary rewards. Fewer studies have used performance-based comparision (e.g., profession-related knowledge quizzes \u003csup\u003e\u003cspan citationid=\"CR17\" class=\"CitationRef\"\u003e17\u003c/span\u003e\u003c/sup\u003e or stopwatch tasks \u003csup\u003e\u003cspan citationid=\"CR18\" class=\"CitationRef\"\u003e18\u003c/span\u003e,\u003cspan citationid=\"CR19\" class=\"CitationRef\"\u003e19\u003c/span\u003e\u003c/sup\u003e). This leave a gap that limits conclusions about how social comparison shapes ability perceptions in real-world task contexts.\u003c/p\u003e\u003cp\u003eFollowing comparison outcomes, individuals update their self-efficacy. Bandura\u0026rsquo;s (1977) social cognitive theory defines self-efficacy as beliefs in one\u0026rsquo;s ability to execute goal-directed actions, with performance experiences (successes/failures) driving increases/decreases in efficacy \u003csup\u003e\u003cspan citationid=\"CR5\" class=\"CitationRef\"\u003e5\u003c/span\u003e\u003c/sup\u003e. In social contexts, downward and upward comparisons act as proxies for success and failure, directly shape self-efficacy \u003csup\u003e\u003cspan citationid=\"CR5\" class=\"CitationRef\"\u003e5\u003c/span\u003e,\u003cspan citationid=\"CR21\" class=\"CitationRef\"\u003e21\u003c/span\u003e\u003c/sup\u003e. Neurocomputational work further shows that self-efficacy updates are stronger in response to positive (vs. negative) performance feedback \u003csup\u003e\u003cspan citationid=\"CR7\" class=\"CitationRef\"\u003e7\u003c/span\u003e,\u003cspan citationid=\"CR20\" class=\"CitationRef\"\u003e20\u003c/span\u003e,\u003cspan citationid=\"CR22\" class=\"CitationRef\"\u003e22\u003c/span\u003e\u003c/sup\u003e.\u003c/p\u003e\u003cp\u003eHowever, how these efficacy updates translate to effort investment, a key outcome of social comparison, remains debated. Bandura\u0026rsquo;s framework predicts social comparison feedback and updated self-efficacy jointly regulate effort (Bandura, 1977), but empirical evidence is limited. Few studies have examined post-comparison risk-taking behaviors \u003csup\u003e\u003cspan citationid=\"CR15\" class=\"CitationRef\"\u003e15\u003c/span\u003e,\u003cspan additionalcitationids=\"CR24\" citationid=\"CR23\" class=\"CitationRef\"\u003e23\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR25\" class=\"CitationRef\"\u003e25\u003c/span\u003e\u003c/sup\u003e, and recent research using ecological momentary assessment has reported findings contradicts Bandura\u0026rsquo;s prediction (Diel et al., 2021, 2024). For example, Diel and colleagues reported that upward (not downward) social comparison increased self-improvement motivation, and led to stronger effort intentions and increased effort investment. By contrast, downward social comparisons foster complacency, prompting individuals to disregard improvement strategies and enter a \u0026ldquo;coasting\u0026rdquo; state marked by reduced effort. These effects are particularly pronounced in contexts of moderate social comparison \u003csup\u003e\u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e3\u003c/span\u003e,\u003cspan citationid=\"CR26\" class=\"CitationRef\"\u003e26\u003c/span\u003e,\u003cspan citationid=\"CR27\" class=\"CitationRef\"\u003e27\u003c/span\u003e\u003c/sup\u003e. However, these studies rely on self-report measures of motivation and effort, which may be susceptible to social desirability bias or distorted by inaccurate self-perceptions.\u003c/p\u003e\u003cp\u003eTo address these gaps, we combined behavioral (Experiment 1) and electrophysiological (Experiment 2) experiments to dissect the neurocognitive mechanisms linking performance-based social comparison, self-efficacy updates, and subsequent effort allocation\u0026mdash;free from monetary reward confounds. Both experiments used essentially identical designs (Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003e). Participants performed a Stroop task, with effort investment quantified by the number of responses within random time windows (Experiment 1: 8s, 10s, 12s; Experiment 2: 6s, 8s, 10s, Leng et al., 2021). After each trial, participants received trial-by-trial social comparison feedback, signaling whether they performed better (downward comparison) or worse (upward comparison) than a purported competitor. To align with the prior studies (Diel et al., 2021, 2024), we created a moderate comparison context where participants received balanced downward/upward feedback (Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003ea). Every 3\u0026ndash;5 trials, participants rated their likelihood of winning the next round (Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003eb) to index self-efficacy.\u003c/p\u003e\u003cp\u003eBehaviorally, according to Bandura\u0026rsquo;s theory \u003csup\u003e\u003cspan citationid=\"CR5\" class=\"CitationRef\"\u003e5\u003c/span\u003e\u003c/sup\u003e, we predict that downward social comparison feedback would enhance self-efficacy and drive better task performance compared to upward feedback. However, if our design mirrors Diel et al.\u0026rsquo;s (2021, 2024) contexts, upward comparison feedback may instead yield better task performance than downward comparison. Using reinforcement learning model and linear mixed models, we aim to quantify trial-by-trial self-efficacy updates, and to test how social comparison feedback and self-efficacy jointly influence subsequent effort allocation. Moreover, we evaluate whether self-efficacy mediates the effect of social comparison on effort investment using multilevel mediation models.\u003c/p\u003e\u003cp\u003eElectrophysiologically (Experiment 2), we investigated neural correlates of feedback evaluation and task preparation. We expected enhanced RewP, fb-P3, and LPP amplitudes for downward compared to upward feedback during the evaluation phase, as downward comparison is primarily experienced as rewarding. In line with the behavioral results, we hypothesized that social comparison feedback and updated self-efficacy would modulate task preparatory responses, such as contingent negative variation (CNV) amplitudes and cue-beta band power in the subsequent round.\u003c/p\u003e\u003cp\u003eOur experimental design served three key purposes: it allowed us to assess the neural signatures associated with the evaluation of social comparison based on performance, to track the dynamic changes in efficacy induced by social comparison, and to investigate how these two factors (i.e., feedback and efficacy) shaped effort investment in the subsequent round. Our findings revealed two main results: first, relative to upward comparison, downward social comparison elicited stronger neural responses and higher levels of efficacy; second, both downward social comparison and higher efficacy positively contributed to effort investment in the subsequent round.\u003c/p\u003e\u003cp\u003e\u003c/p\u003e"},{"header":"Results","content":"\u003cdiv id=\"Sec3\" class=\"Section2\"\u003e\u003ch2\u003eExperiment 1\u003c/h2\u003e\u003cdiv id=\"Sec4\" class=\"Section3\"\u003e\u003ch2\u003eAdjusted Efficacy Expectations Based on Different Social Comparison Feedback\u003c/h2\u003e\u003cp\u003eWe first examined how different types of social comparison feedback shape efficacy expectations (i.e., the ratings of the probability of winning in the next round). The LMM results revealed that participants' ratings were significantly affected by the direction of the recent social comparison feedback, \u003cem\u003eF\u003c/em\u003e (1, 32.47)\u0026thinsp;=\u0026thinsp;33.82, \u003cem\u003eb\u003c/em\u003e\u0026thinsp;=\u0026thinsp;0.190, \u003cem\u003ep\u003c/em\u003e\u0026thinsp;\u0026lt;\u0026thinsp;.001, 95% CI [0.126, 0.255], with higher ratings of wining probability after downward feedback compared to upward feedback (62.91 vs. 59.54, Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003ea).\u003c/p\u003e\u003cp\u003eNext, we used the two-learning-rate model to calculate the efficacy estimates for each round. We then examined how the model-based efficacy estimates related to participants\u0026rsquo; rating scores to verify the reliability. The LMM results revealed that model-based efficacy estimates significantly predicted rating scores, \u003cem\u003eF\u003c/em\u003e (1, 31.64)\u0026thinsp;=\u0026thinsp;111.36, \u003cem\u003eb\u0026thinsp;=\u003c/em\u003e\u0026thinsp;0.547, \u003cem\u003ep\u003c/em\u003e\u0026thinsp;\u0026lt;\u0026thinsp;.001, 95% CI [0.445, 0.648] (Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003eb), confirming that the model-based efficacy estimates accurately reflect participants\u0026rsquo; efficacy expectations. Finally, we analyzed the impact of social comparison feedback on model-based efficacy estimates. The LMM results revealed that social comparison feedback significantly predicted the model-based efficacy estimates, \u003cem\u003eF\u003c/em\u003e (1, 32.00)\u0026thinsp;=\u0026thinsp;64.31, \u003cem\u003eb\u0026thinsp;=\u003c/em\u003e\u0026thinsp;0.316, \u003cem\u003ep\u003c/em\u003e\u0026thinsp;\u0026lt;\u0026thinsp;.001, 95% CI [0.239, 0.393], with higher model-based efficacy estimations following downward social comparison compared to upward comparison (0.62 vs. 0.47, Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003ec).\u003c/p\u003e\u003cp\u003eThese results provide evidence that participants adjusted their efficacy expectations based on different social comparison feedbacks, with downward feedback leading to increased efficacy estimates. Our findings not only demonstrate that the two-learning-rate model successfully captures trial-by-trial changes in efficacy expectations, but also confirm that downward social comparison feedback functions as a positive performance-related experience, thereby effectively enhancing efficacy expectations.\u003c/p\u003e\u003cp\u003e\u003c/p\u003e\u003c/div\u003e\u003c/div\u003e\n\u003ch3\u003eImproved Performance Following Downward Social Comparison Feedback and\u003c/h3\u003e\n\u003cdiv id=\"Sec6\" class=\"Section2\"\u003e\u003ch2\u003eHigher Efficacy Expectations\u003c/h2\u003e\u003cp\u003eTo clarify the impact of social comparison feedback on effort investment, we first performed LMMs to assess how social comparison feedback and model-based efficacy expectations influenced correct responses per second (CRPS) and mean reaction times (RTs), respectively. The LMM results for CRPS revealed significant main effects of social comparison feedback (Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003ea), \u003cem\u003eF\u003c/em\u003e (1, 103.10)\u0026thinsp;=\u0026thinsp;5.08, \u003cem\u003eb\u003c/em\u003e\u0026thinsp;=\u0026thinsp;0.028, \u003cem\u003ep\u003c/em\u003e\u0026thinsp;=\u0026thinsp;.026, 95% CI [0.004, 0.052], and model-based efficacy estimates (Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003eb), \u003cem\u003eF\u003c/em\u003e (1, 14.42)\u0026thinsp;=\u0026thinsp;5.56, \u003cem\u003eb\u003c/em\u003e\u0026thinsp;=\u0026thinsp;0.049, \u003cem\u003ep\u003c/em\u003e\u0026thinsp;=\u0026thinsp;.026, 95% CI [0.011, 0.088]. Specifically, CRPS was higher in the upcoming round following downward social comparison feedback compared to upward social comparison feedback (0.971 vs. 0.961). Similarly, higher model-based efficacy estimates were associated with higher CRPS for the upcoming round (Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003eb). However, no significant main effects were found for mean RTs (658 vs. 664 ms, Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003ea) being affected by social comparison feedback, \u003cem\u003eF\u003c/em\u003e (1, 32.01)\u0026thinsp;=\u0026thinsp;2.48, \u003cem\u003eb\u003c/em\u003e = -0.017, \u003cem\u003ep\u003c/em\u003e\u0026thinsp;=\u0026thinsp;.124, 95% CI [-0.038, 0.004], or model-based efficacy estimates (Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003eb), \u003cem\u003eF\u003c/em\u003e (1, 12.07)\u0026thinsp;=\u0026thinsp;4.45, \u003cem\u003eb\u003c/em\u003e = -0.098, \u003cem\u003ep\u003c/em\u003e\u0026thinsp;=\u0026thinsp;.056, 95% CI [-0.190, -0.007].\u003c/p\u003e\u003cp\u003eSubsequently, we fitted the Hierarchical Drift Diffusion Model (HDDM) to trial-level data to investigate the latent computational processes underlying these performance patterns. We first performed HDDM to explore how social comparison feedback influence drift rates and response threshold. The results revealed that the best-fitting model was the one included Stroop congruency as a control variable (see Hierarchical Drift Diffusion Model in Method). The main effect of social comparison feedback on drift rates was significant, with faster drift rates observed following downward social comparison feedback compared to upward social comparison feedback (\u003cem\u003eb\u003c/em\u003e\u0026thinsp;=\u0026thinsp;0.05, 95% CI [0.01, 0.08], \u003cem\u003eP\u003c/em\u003e\u003csub\u003eb\u0026gt;0\u003c/sub\u003e = 0.97, Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003ec). However, there were no significant main effect of social comparison on response threshold (\u003cem\u003eb\u003c/em\u003e = -0.00, 95% CI [-0.02, 0.01], \u003cem\u003eP\u003c/em\u003e\u003csub\u003eb\u0026lt;0\u003c/sub\u003e = 0.67). Next, we performed HDDM to explore the effect of model-based efficacy estimates on drift rates and response threshold. The best-fitting model was also the one controlled for Stroop congruency. The results revealed a significant main effect of model-based efficacy estimates on drift rates, with higher model-based efficacy estimates being associated with faster drift rates (\u003cem\u003eb\u003c/em\u003e\u0026thinsp;=\u0026thinsp;0.17, 95% CI [0.01, 0.35], \u003cem\u003eP\u003c/em\u003e\u003csub\u003eb\u0026gt;0\u003c/sub\u003e =0.96, Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003ec). However, there was no effect of model-based efficacy on response threshold (\u003cem\u003eb\u003c/em\u003e = -0.05, 95% CI [-0.40, 0.31], \u003cem\u003eP\u003c/em\u003e\u003csub\u003eb\u0026lt;0\u003c/sub\u003e = 0.61) .\u003c/p\u003e\u003cp\u003eThese results indicate that both downward social comparison feedback and higher efficacy expectations are associated with improved performance, as evidenced by higher CRPS and faster drift rates. Moreover, this improved performance reflects greater cognitive control investment, with increased attentional allocation to targets rather than a shift in the speed-accuracy tradeoff.\u003c/p\u003e\u003c/div\u003e\n\u003ch3\u003eEfficacy Expectations Mediated the Relationship Between Social Comparison and Effort Investment\u003c/h3\u003e\n\u003cp\u003e These findings demonstrate that participants adjusted their efficacy expectations in response to different social comparison feedback, which subsequently influenced their effort investment in the upcoming round. To further validate this relationship and elucidate the mechanism of social comparison and effort investment, we conducted multilevel mediation analyses to examine whether efficacy expectations mediate the effect of social comparison feedback on effort, as indexed by CRPS and mean RTs. The results revealed that model-based efficacy estimates fully mediated this relationship (Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003ed). This was supported by significant indirect effects for both CRPS (\u003cem\u003eab\u003c/em\u003e\u0026thinsp;=\u0026thinsp;0.025, \u003cem\u003ep\u003c/em\u003e\u0026thinsp;=\u0026thinsp;.041, 95% CI [0.001, 0.048]) and mean RTs (\u003cem\u003eab\u003c/em\u003e = -0.025, \u003cem\u003ep\u003c/em\u003e\u0026thinsp;=\u0026thinsp;.012, 95% CI [-0.045, -0.005]). Specifically, downward social comparison feedback enhanced efficacy expectations, which then led to greater effort investment in the next round.\u003c/p\u003e\u003cp\u003e\u003c/p\u003e\u003cdiv id=\"Sec8\" class=\"Section2\"\u003e\u003ch2\u003eExperiment 2\u003c/h2\u003e\u003cdiv id=\"Sec9\" class=\"Section3\"\u003e\u003ch2\u003eBehavioral Results\u003c/h2\u003e\u003cp\u003eThe results of Experiment 1 demonstrated that participants updated their efficacy expectations based on social comparison feedback, which subsequently guided their effort investment at the behavioral level. Building on these findings, Experiment 2 aimed to identify the neural signatures associated with the evaluation of social comparison feedback and to reveal how social comparison feedback and efficacy expectations modulate subsequent effort investment at the neural level.\u003c/p\u003e\u003cp\u003eConsistent with Experiment 1, we observed a significant effect of social comparison feedback on participants\u0026rsquo; self-reported rating scores, with higher scores following downward versus upward comparison (62.13 vs. 59.60), \u003cem\u003eF\u003c/em\u003e(1, 34.21)\u0026thinsp;=\u0026thinsp;18.03, \u003cem\u003eb\u003c/em\u003e\u0026thinsp;=\u0026thinsp;0.140, \u003cem\u003ep\u003c/em\u003e\u0026thinsp;\u0026lt;\u0026thinsp;.001, 95% [0.075, 0.204]. In addition, model-based efficacy estimates significantly predicted self-reported rating scores, \u003cem\u003eF\u003c/em\u003e(1, 29.25)\u0026thinsp;=\u0026thinsp;125.25, \u003cem\u003eb\u003c/em\u003e\u0026thinsp;=\u0026thinsp;0.465, \u003cem\u003ep\u003c/em\u003e\u0026thinsp;\u0026lt;\u0026thinsp;.001, 95% [0.384, 0.547]. Social comparison feedback also significantly influenced efficacy estimates, \u003cem\u003eF\u003c/em\u003e(1, 34.00)\u0026thinsp;=\u0026thinsp;32.81, \u003cem\u003eb\u003c/em\u003e\u0026thinsp;=\u0026thinsp;0.245, \u003cem\u003ep\u003c/em\u003e\u0026thinsp;\u0026lt;\u0026thinsp;.001, 95% [0.161, 0.329], with higher estimates following downward versus upward feedback (0.58 vs. 0.47). These findings replicate those of Experiment 1 and strengthen support for the influence of social comparison feedback on both self-reported ratings and efficacy expectations.\u003c/p\u003e\u003cp\u003eLMM analysis of CRPS revealed neither the significant main effect of social comparison feedback (1.083 vs. 1.078), \u003cem\u003eF\u003c/em\u003e(1, 106.68)\u0026thinsp;=\u0026thinsp;0.89, \u003cem\u003eb\u003c/em\u003e\u0026thinsp;=\u0026thinsp;0.013, \u003cem\u003ep\u003c/em\u003e\u0026thinsp;=\u0026thinsp;.348, 95% [-0.013, 0.038], nor of model-based efficacy estimates, \u003cem\u003eF\u003c/em\u003e(1, 24.27)\u0026thinsp;=\u0026thinsp;1.29, \u003cem\u003eb\u003c/em\u003e\u0026thinsp;=\u0026thinsp;0.034, \u003cem\u003ep\u003c/em\u003e\u0026thinsp;=\u0026thinsp;.266, 95% [-0.030, 0.094]. While non-significant, these results exhibited a similar trend to those in Experiment 1. Similarly, LMM analysis of mean RTs revealed neither the significant effect of social comparison feedback (597 vs. 600 ms), \u003cem\u003eF\u003c/em\u003e(1, 62.46)\u0026thinsp;=\u0026thinsp;2.00, \u003cem\u003eb\u003c/em\u003e = -0.017, \u003cem\u003ep\u003c/em\u003e\u0026thinsp;=\u0026thinsp;.161, 95% [-0.040, 0.006], nor of model-based efficacy estimates, \u003cem\u003eF\u003c/em\u003e(1, 25.80)\u0026thinsp;=\u0026thinsp;0.78, \u003cem\u003eb\u003c/em\u003e = -0.035, \u003cem\u003ep\u003c/em\u003e\u0026thinsp;=\u0026thinsp;.382, 95% [-0111, 0.042].\u003c/p\u003e\u003cp\u003eConsistent with Experiment 1, the best-fitting HDDM models were those controlled for the effect of Stroop trial congruency. The results showed that model-based efficacy estimates exerted a significant main effect on drift rate (\u003cem\u003eb\u003c/em\u003e\u0026thinsp;=\u0026thinsp;0.19, 95% CI [0.01, 0.37], \u003cem\u003eP\u003c/em\u003e\u003csub\u003eb\u0026gt;0\u003c/sub\u003e = 0.96) but not on the threshold (\u003cem\u003eb\u003c/em\u003e\u0026thinsp;=\u0026thinsp;0.10, 95% CI [-0.07, 0.26], \u003cem\u003eP\u003c/em\u003e\u003csub\u003eb\u0026lt;0\u003c/sub\u003e = 0.16). However, social comparison feedback had neither significant main effect on drift rate (\u003cem\u003eb\u003c/em\u003e\u0026thinsp;=\u0026thinsp;0.01, 95% CI [-0.03, 0.06], \u003cem\u003eP\u003c/em\u003e\u003csub\u003eb\u0026gt;0\u003c/sub\u003e = 0.68) nor on threshold (\u003cem\u003eb\u003c/em\u003e = -0.00, 95% CI [-0.02, 0.01], \u003cem\u003eP\u003c/em\u003e\u003csub\u003eb\u0026lt;0\u003c/sub\u003e = 0.60).\u003c/p\u003e\u003cp\u003eAs in Experiment 1, model-based efficacy estimations fully mediated the influence of social comparison on effort investment, as indicated by significant indirect effects on both CRPS (\u003cem\u003eab\u003c/em\u003e\u0026thinsp;=\u0026thinsp;0.033, \u003cem\u003ep\u003c/em\u003e\u0026thinsp;\u0026lt;\u0026thinsp;.001, 95% CI [0.015, 0.051]) and mean RTs (\u003cem\u003eab\u003c/em\u003e = -0.043, \u003cem\u003ep\u003c/em\u003e\u0026thinsp;\u0026lt;\u0026thinsp;.001, 95% CI [-0.059, -0.028]). Overall, the behavioral results replicated those of Experiment 1.\u003c/p\u003e\u003c/div\u003e\u003c/div\u003e\n\u003ch3\u003eNeural Signatures for Social Comparison Outcomes Evaluation\u003c/h3\u003e\n\u003cp\u003eTo investigate the neural signatures underlying the evaluation of different social comparison feedback types, we first focused on event-related potential (ERP) components during the feedback phase: specifically, the RewP, fb-P3, and LPP.\u003c/p\u003e\u003cp\u003eOur analysis revealed significant main effects of social comparison on the RewP, \u003cem\u003eF\u003c/em\u003e(1, 33.76)\u0026thinsp;=\u0026thinsp;5.08, \u003cem\u003eb\u003c/em\u003e\u0026thinsp;=\u0026thinsp;0.034, \u003cem\u003ep\u003c/em\u003e\u0026thinsp;=\u0026thinsp;.031, 95% [0.004, 0.065], fb-P3, \u003cem\u003eF\u003c/em\u003e(1, 34.12)\u0026thinsp;=\u0026thinsp;4.85, \u003cem\u003eb\u003c/em\u003e\u0026thinsp;=\u0026thinsp;0.052, \u003cem\u003ep\u003c/em\u003e\u0026thinsp;=\u0026thinsp;.034, 95% [0.006, 0.098], and LPP, \u003cem\u003eF\u003c/em\u003e(1, 268.96)\u0026thinsp;=\u0026thinsp;7.94, \u003cem\u003eb\u003c/em\u003e\u0026thinsp;=\u0026thinsp;0.044, \u003cem\u003ep\u003c/em\u003e\u0026thinsp;=\u0026thinsp;.005, 95% [0.014, 0.074]. Each component displayed a consistent pattern, with more positive-going waveforms following downward versus upward social comparison feedback. Specifically, the RewP (13.57\u0026thinsp;\u0026plusmn;\u0026thinsp;14.48 vs. 12.60\u0026thinsp;\u0026plusmn;\u0026thinsp;13.72 \u0026micro;V, Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003ea and \u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003eb), fb-P3 (14.35\u0026thinsp;\u0026plusmn;\u0026thinsp;13.37 vs. 12.92\u0026thinsp;\u0026plusmn;\u0026thinsp;12.78 \u0026micro;V, Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003ec and \u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003ed), and LPP (3.24\u0026thinsp;\u0026plusmn;\u0026thinsp;11.39 vs. 2.23\u0026thinsp;\u0026plusmn;\u0026thinsp;11.25 \u0026micro;V, Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003ee and \u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003ef) all exhibited this effect.\u003c/p\u003e\u003cp\u003eCollectively, these ERP components exhibited heightened sensitivity to downward versus upward social comparison feedback. These findings align with previous studies \u003csup\u003e\u003cspan citationid=\"CR8\" class=\"CitationRef\"\u003e8\u003c/span\u003e,\u003cspan citationid=\"CR10\" class=\"CitationRef\"\u003e10\u003c/span\u003e,\u003cspan citationid=\"CR43\" class=\"CitationRef\"\u003e43\u003c/span\u003e\u003c/sup\u003e, which suggest that downward social comparison feedback, perceived as a form of social reward, elicits stronger neural activities associated with reward evaluation and emotional arousal.\u003c/p\u003e\u003cp\u003e\u003c/p\u003e\u003cdiv id=\"Sec11\" class=\"Section2\"\u003e\u003ch2\u003eNeural Signatures of Effort Preparation Are Influenced by Social Comparison and Efficacy Expectations\u003c/h2\u003e\u003cp\u003e Our behavioral results demonstrated that participants adjusted their effort investment in round t\u0026thinsp;+\u0026thinsp;1 based on social comparison feedback and efficacy expectations derived from round t. To explore the neural underpinnings of this process, we focused on the CNV and Cue-beta band activity observed in round t\u0026thinsp;+\u0026thinsp;1. Both of these components are linked to task preparation and cognitive/anticipatory processes for the upcoming task \u003csup\u003e\u003cspan citationid=\"CR37\" class=\"CitationRef\"\u003e37\u003c/span\u003e,\u003cspan citationid=\"CR42\" class=\"CitationRef\"\u003e42\u003c/span\u003e\u003c/sup\u003e\u003c/p\u003e\u003cp\u003eWe first analyzed the impact of round t social comparison feedback on these two neural signatures in round t\u0026thinsp;+\u0026thinsp;1. Significant main effects of social comparison were observed for both the CNV, \u003cem\u003eF\u003c/em\u003e (1, 114.41)\u0026thinsp;=\u0026thinsp;4.15, \u003cem\u003eb\u003c/em\u003e = -0.034, \u003cem\u003ep\u003c/em\u003e\u0026thinsp;=\u0026thinsp;.044, 95% [-0.066, -0.001], and Cue-beta band power, \u003cem\u003eF\u003c/em\u003e (1, 1905.17)\u0026thinsp;=\u0026thinsp;4.63, \u003cem\u003eb\u003c/em\u003e = -0.034, \u003cem\u003ep\u003c/em\u003e\u0026thinsp;=\u0026thinsp;.031, 95% [-0.065, -0.003]. Specifically, downward versus upward social comparison feedback from round t induced more negative-going CNV waveforms (-1.27\u0026thinsp;\u0026plusmn;\u0026thinsp;13.19 vs. -0.37\u0026thinsp;\u0026plusmn;\u0026thinsp;13.40 \u0026micro;V, Fig.\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e5\u003c/span\u003ea and Fig.\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e5\u003c/span\u003eb) and higher Cue-beta band power (-0.41\u0026thinsp;\u0026plusmn;\u0026thinsp;3.82 vs. -0.15\u0026thinsp;\u0026plusmn;\u0026thinsp;3.85 dB, Fig.\u0026nbsp;\u003cspan refid=\"Fig6\" class=\"InternalRef\"\u003e6\u003c/span\u003ea and Fig.\u0026nbsp;\u003cspan refid=\"Fig6\" class=\"InternalRef\"\u003e6\u003c/span\u003eb) in round t\u0026thinsp;+\u0026thinsp;1.\u003c/p\u003e\u003cp\u003e\u003c/p\u003e\u003cp\u003eNext, we analyzed the impact of model-based efficacy estimates (from round t) on round t\u0026thinsp;+\u0026thinsp;1 CNV and cue-beta power. The CNV was significantly affected by model-based efficacy estimates, \u003cem\u003eF\u003c/em\u003e (1, 40.22)\u0026thinsp;=\u0026thinsp;7.90, \u003cem\u003eb\u003c/em\u003e = -0.056, \u003cem\u003ep\u003c/em\u003e\u0026thinsp;=\u0026thinsp;.007, 95% [-0.097, -0.015], with more negative-going waveforms in round t\u0026thinsp;+\u0026thinsp;1 associated with higher round t model-based efficacy estimates (Fig.\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e5\u003c/span\u003ec). However, model-based efficacy estimates had no significant effect on Cue-beta band power (Fig.\u0026nbsp;\u003cspan refid=\"Fig6\" class=\"InternalRef\"\u003e6\u003c/span\u003ec), \u003cem\u003eF\u003c/em\u003e (1, 46.49)\u0026thinsp;=\u0026thinsp;0.31, \u003cem\u003eb\u003c/em\u003e = -0.009, \u003cem\u003ep\u003c/em\u003e\u0026thinsp;=\u0026thinsp;.583, 95% [-0.050, 0.033].\u003c/p\u003e\u003cp\u003eThese results suggest that participants engaged in enhanced task preparation for upcoming targets following both downward social comparison feedback and higher model-based efficacy estimates. Collectively, these findings underscore the role of social comparison dynamics and self-efficacy in shaping the neural mechanisms associated with preparatory engagement and cognitive readiness for subsequent tasks.\u003c/p\u003e\u003cp\u003e\u003c/p\u003e\u003c/div\u003e"},{"header":"Discussion","content":"\u003cp\u003e By manipulating a social comparison context, we investigated how evaluating one’s performance relative to others modulates effortful behavior. Participants exhibited enhanced neural responses to downward versus upward social comparison feedback, indexed by larger RewP, fb-P3, and LPP components, reflecting the motivational and emotional significance of the superior social comparisons. Participants then dynamically updated their self-efficacy based on the direction of social comparison feedback. Finally, both downward social comparison and higher self-efficacy improved the preparatory engagement of cognitive control, as indicated by enhanced CNV and cue-beta power, and facilitated the subsequent effort investment, as indicated by improved performance in the next round.\u003c/p\u003e\u003cp\u003eWe observed more positive RewP, fb-P3, and LPP components in the downward versus upward feedback condition, aligning with previous studies using gambling tasks and monetary payoffs to manipulate social comparison direction \u003csup\u003e\u003cspan citationid=\"CR10\" class=\"CitationRef\"\u003e10\u003c/span\u003e,\u003cspan citationid=\"CR16\" class=\"CitationRef\"\u003e16\u003c/span\u003e,\u003cspan citationid=\"CR44\" class=\"CitationRef\"\u003e44\u003c/span\u003e,\u003cspan citationid=\"CR45\" class=\"CitationRef\"\u003e45\u003c/span\u003e\u003c/sup\u003e. The RewP and fb-P3 are typical components for reward outcome evaluation, with larger amplitudes elicited by more favorable outcomes (see Glazer et al., 2018 for a review). Additionally, the LPP component reflects later-stage affective evaluation and the motivational relevance of the feedback (Hu et al., 2017; Luo et al., 2015; Wu et al., 2012). These stronger responses to downward versus upward feedback support the notion that downward social comparison is experienced as rewarding, associated with more positive emotions and stronger motivation \u003csup\u003e\u003cspan citationid=\"CR17\" class=\"CitationRef\"\u003e17\u003c/span\u003e,\u003cspan citationid=\"CR44\" class=\"CitationRef\"\u003e44\u003c/span\u003e,\u003cspan citationid=\"CR46\" class=\"CitationRef\"\u003e46\u003c/span\u003e\u003c/sup\u003e. Importantly, we manipulated performance as social comparison dimension and explicitly informed participants that their monetary compensation was independent of task performance. Thus, the heightened electrophysiological responses to downward versus upward feedback clearly demonstrate that these components are sensitive to social comparison itself, rather than monetary reward contingencies.\u003c/p\u003e\u003cp\u003eParticipants then updated their self-perceptions of ability based on different types of social comparison feedback. Previous studies have indicated that positive social feedback (e.g., high-ability feedback or better performance feedback in competitive contexts) enhances participants’ confidence and improves self-evaluations of ability \u003csup\u003e\u003cspan citationid=\"CR4\" class=\"CitationRef\"\u003e4\u003c/span\u003e,\u003cspan citationid=\"CR7\" class=\"CitationRef\"\u003e7\u003c/span\u003e,\u003cspan citationid=\"CR20\" class=\"CitationRef\"\u003e20\u003c/span\u003e,\u003cspan citationid=\"CR21\" class=\"CitationRef\"\u003e21\u003c/span\u003e\u003c/sup\u003e. Our results confirm that downward social comparison can be conceptualized as a form of performance accomplishment that significantly improves participants’ efficacy perceptions, whereas upward comparisons undermines them \u003csup\u003e\u003cspan citationid=\"CR5\" class=\"CitationRef\"\u003e5\u003c/span\u003e\u003c/sup\u003e.\u003c/p\u003e\u003cp\u003eBased on social comparison direction and updated efficacy expectations, participants adjusted their effort investment. More negative-going CNV waveforms and stronger cue-beta power following downward comparison indicate enhanced control preparation and motor readiness for the upcoming round (Frömer et al., 2021; Glazer et al., 2018; Rong et al., 2022; Y. Zhang et al., 2023; Doyle, Yarrow, \u0026amp; Brown, 2005; Gable et al., 2016), which then led to better behavioral performance, evidenced by higher CRPS and faster drift rates across both experiments. HDDM analyses further revealed that participants actively adjusted their attention in response to downward feedback and elevated self-efficacy, characterized by accelerated evidence accumulation without compromising the speed-accuracy trade-off. Downward social comparison functions as rewarding feedback: it fosters positive emotions, boosts self-efficacy, and increases the perceived likelihood of goal attainment. These factors collectively motivated participants to invest greater effort to maintain this favorable state. Our multilevel mediation model further supported this process, confirming that self-efficacy fully mediated the relationship between social comparison and effort investment.\u003c/p\u003e\u003cp\u003eHowever, as noted in the Introduction, recent studies used ecological momentary assessment have found that upward comparisons promote greater self-reported effort investment intentions \u003csup\u003e\u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e3\u003c/span\u003e,\u003cspan citationid=\"CR27\" class=\"CitationRef\"\u003e27\u003c/span\u003e\u003c/sup\u003e, which is inconsistent with our results. This discrepancy may stem from methodological differences in effort assessment. Retrospective self-reports are susceptible to social desirability and memory biases \u003csup\u003e\u003cspan citationid=\"CR47\" class=\"CitationRef\"\u003e47\u003c/span\u003e\u003c/sup\u003e, and even accurately-reported intentions may not translate to actual behavior due to the well-known intention–behavior gap \u003csup\u003e\u003cspan citationid=\"CR48\" class=\"CitationRef\"\u003e48\u003c/span\u003e\u003c/sup\u003e. In contrast, our study used a real-time effort task to directly measure how social comparison feedback shapes actual effort investment. The current study’s behavioral and neurocognitive findings thus provide direct evidence of social comparison’s impact on effort behavior.\u003c/p\u003e\u003cp\u003ePrior research has manipulated perceived control efficacy in non-social context, where different cues indicated whether performance feedback (i.e., monetary reward) was contingent on participants’ performance (high efficacy) or not (low efficacy, Cao et al., 2022; Frömer et al., 2021). These studies reported better behavioral performance under high-efficacy conditions. In contrast, our study employed a fixed, non-meaningful cue for task preparation (see also Grahek et al., 2022). The enhanced brain responses and improved behavioral performance following downward social comparison suggest effort adjustments are driven by past comparison outcomes and internal self-efficacy updates, highlighting the critical role of self-efficacy in shaping goal-directed behavior (Bandura, 1977). These findings converge with animal studies showing that “winning” mice exhibit heightened initiative and sustained effortful behaviors in competitive contexts compared to “losing” mice \u003csup\u003e\u003cspan citationid=\"CR50\" class=\"CitationRef\"\u003e50\u003c/span\u003e\u003c/sup\u003e, reflecting the “winner effect”, where success history boosts the likelihood of future achievement in both humans and animals \u003csup\u003e\u003cspan additionalcitationids=\"CR51\" citationid=\"CR50\" class=\"CitationRef\"\u003e50\u003c/span\u003e–\u003cspan citationid=\"CR52\" class=\"CitationRef\"\u003e52\u003c/span\u003e\u003c/sup\u003e.\u003c/p\u003e\u003cp\u003eNonetheless, our findings should be interpreted with caution. We observed a stronger motivational effect of downward versus upward social comparison in moderate competitive contexts (50% downward/50% upward comparisons), but this relationship may vary with the ratio of downward to upward feedback. Our results do not discount the adaptive effects of upward comparison; rather, both comparison directions can foster adaptive motivational responses \u003csup\u003e\u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e3\u003c/span\u003e\u003c/sup\u003e. This is partly supported by the behavioral results of our Experiment 2: shortening the task time window from 8–12 to 6–8 seconds reduced fatigue and narrowed the motivational discrepancy between upward and downward comparisons relative to Experiment 1. The motivational impacts of downward and upward feedback may thus compete, with downward feedback exerting greater influence under conditions of heightened participant fatigue. Future research should explore performance dynamics in more complex competitive environments.\u003c/p\u003e\u003cp\u003eIn conclusion, our study illustrates the cognitive and neural mechanisms through which downward and upward social comparisons influence effort investment, tracing the process from initial feedback evaluation to self-efficacy updates and subsequent effort allocation. Findings reveal that, compared to upward comparison, downward comparison feedback elicits enhanced neural activity and stronger self-efficacy expectations, both of which positively correlate with improvements in behavioral performance. Critically, self-efficacy fully mediates the relationship between social comparison and effort investment, underscoring its pivotal role as a motivational driver of goal-directed effort. These findings provide empirical support for a dynamic, multi-stage framework in which social comparison shapes efficacy expectations and guides effort allocation over time.\u003c/p\u003e\u003cdiv id=\"Sec13\" class=\"Section2\"\u003e\u003cdiv id=\"Sec14\" class=\"Section3\"\u003e\u003cdiv id=\"Sec15\" class=\"Section4\"\u003e\u003c/div\u003e\u003c/div\u003e\u003c/div\u003e\u003cdiv id=\"Sec19\" class=\"Section2\"\u003e\u003cdiv id=\"Sec20\" class=\"Section3\"\u003e\u003c/div\u003e\u003c/div\u003e\u003cdiv id=\"Sec22\" class=\"Section2\"\u003e\u003cdiv id=\"Sec23\" class=\"Section3\"\u003e\u003c/div\u003e\u003c/div\u003e\u003cdiv id=\"Sec24\" class=\"Section2\"\u003e\u003cdiv id=\"Sec25\" class=\"Section3\"\u003e\u003cdiv id=\"Sec26\" class=\"Section4\"\u003e\u003c/div\u003e\u003c/div\u003e\u003cdiv id=\"Sec27\" class=\"Section3\"\u003e\u003c/div\u003e\u003c/div\u003e"},{"header":"Methods","content":"\u003ch2\u003eExperiment 1\u003c/h2\u003e\u003ch2\u003eParticipants\u003c/h2\u003e\u003cp\u003e We recruited 32 healthy participants (10 male; mean age: 22.75 ± 2.03 years) with non-psychology backgrounds. All participants had normal or corrected-to-normal vision and normal color perception. We conducted a post hoc power analysis using G*Power (Faul et al., 2007) for this sample size (n = 32), using a paired-sample t-test (to test the difference between two dependent means). With an α level of 0.05, the analysis yielded a statistical power of 0.85. This study was approved by the Ethics Committee of University, and all participants gave informed consent prior to the experiment.\u003c/p\u003e\u003ch2\u003eExperimental Design\u003c/h2\u003e\u003cp\u003e We used a within-participant design with two social comparison conditions: downward social comparison and upward social comparison. PsychoPy3 was used to present experimental stimuli and record participant response data (\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://www.psychopy.org\u003c/span\u003e\u003cspan address=\"https://www.psychopy.org\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e).\u003c/p\u003e\u003cp\u003eBefore the formal experiment, participants completed a practice session to familiarize themselves with the procedure. The practice consisted of two parts. The first part included 80 trials in which Stroop stimulus was presented at the center of the screen. Participants were required to discriminate the color of the Stroop target by pressing the corresponding keys (red-D, green-F, yellow-J, blue-K) with their left and right index and middle fingers. Feedback was presented after each response, with a checkmark (“√”) indicating a correct response and a cross (“×”) indicating an incorrect one. Participants needed to achieve at least 80% accuracy in this part before progressing to the next. The second part of the practice mirrored the formal experiment. A triangle cue was displayed for 1500 ms, followed by a series of Stroop targets. Participants responded to the color of each Stroop target one after another within a randomly selected time window (8 s, 10 s, and 12 s). At the end of this time window, a diamond feedback symbol was presented for 2000 ms to signal the end of that round. After completing 12 practice rounds, participants rated the difficulty of the practice on a scale from 0 to 100, with a time limit of 10 seconds to submit their response. Participants were then informed of the social comparison feedback rules (see below) and that they would receive a final-pay of 55 RMB, regardless of their task performance.\u003c/p\u003e\u003cp\u003eThe formal experiment contained 144 rounds. Each round began with a variable interval of 1400–1600 ms. A triangle cue was then displayed for 1500 ms, followed by a variable interval of 1000–1200 ms, during which participants were required to concentrate and prepare. The Stroop targets was then presented consecutively. Participants were instructed to respond to the color of each Stroop target by pressing the D, F, J, or K (red-D, green-F, yellow-J, blue-K) with their left and right index and middle fingers. There was no fixed presentation time for the Stroop target stimuli. Instead, a 200 ms fixation cross appeared after each response, and the next Stroop target stimulus was presented immediately after the fixation cross disappeared. Participants could respond to as many Stroop targets as they wished/could during each round of randomly selected time window of 8 s, 10 s, or 12 s. The variable time windows prevented participants from counting Stroop stimuli \u003csup\u003e\u003cspan citationid=\"CR28\" class=\"CitationRef\"\u003e28\u003c/span\u003e\u003c/sup\u003e. Subsequently, a fixation cross displayed for 500 ms, followed by a “calculating” screen displayed for 1500 ms to enhance the realism of the social comparison context. Finally, after a variable interval of 1400–1600 ms, feedback was presented for 2000 ms, indicating the social comparison feedback for that round (Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003ea).\u003c/p\u003e\u003cp\u003e To effectively induce social comparison, participants were explicitly informed before the experiment that they would engage in a real-time comparison with a genuine competitor. They were informed about how social comparison feedback presented. Correct responses per second (CRPS), serving as an indicator of participants' performance, was calculated at the end of each round. Social comparison feedback was then presented based on the CRPS values of the participant and the competitor. An upward arrow indicated worse performance (i.e., lower CRPS than the competitor in that round, upward social comparison). In contrast, a downward arrow indicated better performance, (i.e., higher CRPS than the competitor in that round, downward social comparison). A bidirectional arrow indicated similar performance between the participant and the competitor (lateral social comparison, Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003ea). The lateral rounds served as fillers to enhance the realism of the social comparison context. Moreover, participants received upward comparison feedback in rounds where CRPS was ≤ 0.4. In other rounds, the order of the three types of social comparison feedback was pseudo-randomized, and the feedback was independent of participants’ actual performance. There were 66 rounds with upward comparison feedback, 66 rounds with downward comparison feedback, and 12 rounds with lateral feedback, divided into 6 blocks of 24 rounds, with a fixed two-minute break at the end of each block. Each type of feedback was limited to appearing continuously no more than three times.\u003c/p\u003e\u003cp\u003eAfter every 3 to 5 rounds, participants were required to complete a rating based on their past performance to estimate the probability of winning in the subsequent round, with scores ranging from 0 to 100 (see Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003eb). Participants had 10 seconds to submit their responses. A total of 36 ratings were gathered throughout the experiment.\u003c/p\u003e\u003ch2\u003eBehavioral Analyses\u003c/h2\u003e\u003cp\u003eTo investigate how social comparison influences effort investment, we examined the effect of feedback from the current round on effort investment in the next round (see Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003ea). Accordingly, we excluded data from the first round and from rounds following lateral feedback for each participant in the analyses of RTs and CRPS.\u003c/p\u003e\u003cp\u003eMean RTs and CRPS served as indices of effort investment. First, we excluded incorrect trials for each participant (7.91% of the total trials). Following this, we calculated mean RTs for each round. Rounds with mean RTs exceeding three standard deviations (1.91% of total rounds) and CRPS values beyond three standard deviations (1.28% of total rounds) were removed for each participant in each social comparison condition.\u003c/p\u003e\u003ch2\u003eStatistical Analysis\u003c/h2\u003e\u003cp\u003e\u003cb\u003eReinforcement Learning Models.\u003c/b\u003e To compute trial-by-trial model-based efficacy estimates, we used a two-learning-rate reinforcement learning model. This model updates the efficacy estimate for the next round (\u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:{E}_{t+1}\\)\u003c/span\u003e\u003c/span\u003e) based on the efficacy estimate from the current round (\u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:{E}_{t}\\)\u003c/span\u003e\u003c/span\u003e) and the prediction error (\u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:{\\delta\\:}_{t}\\)\u003c/span\u003e\u003c/span\u003e), with the update weighted by a learning rate (\u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:\\alpha\\:,\\:\\:\\:0\u0026lt;\\alpha\\:\u0026lt;1\\)\u003c/span\u003e\u003c/span\u003e) specific to the direction of the prediction error. The prediction error (\u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:{\\delta\\:}_{t}\\)\u003c/span\u003e\u003c/span\u003e) quantifies the discrepancy between the actual social comparison feedback from the current round (\u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:{f}_{t}\\)\u003c/span\u003e\u003c/span\u003e) and the current round’s efficacy estimate (\u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:{E}_{t}\\)\u003c/span\u003e\u003c/span\u003e). To capture asymmetric updating of efficacy expectations (i.e., differential learning from positive vs. negative prediction errors), we used two distinct learning rates: a positive learning rate (\u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:{\\alpha\\:}_{pos}\\)\u003c/span\u003e\u003c/span\u003e) applied when \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:{\\delta\\:}_{t}\\)\u003c/span\u003e\u003c/span\u003e \u0026gt;0 (positive prediction errors, indicating feedback exceeded current efficacy expectations) and a negative learning rate (\u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:{\\alpha\\:}_{neg}\\)\u003c/span\u003e\u003c/span\u003e) applied when \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:{\\delta\\:}_{t}\\)\u003c/span\u003e\u003c/span\u003e \u0026gt;0 (negative prediction errors, indicating feedback fell below current efficacy expectations). This relationship can be expressed as follows:\u003c/p\u003e\u003cdiv id=\"Equa\" class=\"Equation\"\u003e\u003cdiv format=\"TEX\" class=\"mathdisplay\" id=\"FileID_Equa\" name=\"EquationSource\"\u003e\n$$\\:{\\delta\\:}_{t}={f}_{t}-{E}_{t}$$\u003c/div\u003e\u003c/div\u003e\u003cdiv id=\"Equb\" class=\"Equation\"\u003e\u003cdiv format=\"TEX\" class=\"mathdisplay\" id=\"FileID_Equb\" name=\"EquationSource\"\u003e\n$$\\:f\\left(x\\right)=\\left\\{\\begin{array}{c}{E}_{t}+{\\alpha\\:}_{pos}\\ast\\:{\\delta\\:}_{t},\\:\\:{\\delta\\:}_{t}\u0026gt;0\\\\\\:{E}_{t}+{\\alpha\\:}_{neg}\\ast\\:{\\delta\\:}_{t},\\:\\:{\\delta\\:}_{t}\u0026lt;0\\end{array}\\right.$$\u003c/div\u003e\u003c/div\u003e\u003cp\u003e\u003c/p\u003e\u003cp\u003e\u003cb\u003eLinear Mixed Models\u003c/b\u003e. We first conducted linear mixed models (LMMs) to explore three core questions: (1) how social comparison feedback influences participants’ self-reported rating scores; (2) how social comparison feedback influences model-based efficacy estimates; and (3) how model-based efficacy estimates predict participants’ rating scores. Subsequently, we used additional LMMs to examine the effects of social comparison feedback and model-based efficacy estimates on effort investment, measured by mean RTs and CRPS. Mean Stroop congruency (proportion of congruent vs. incongruent trials per round) was included as a level-1 control variable to account for task-related variance in performance.\u003c/p\u003e\u003cp\u003eFor all models, age and gender were entered as fixed effects to control for potential demographic influences. The model structure included random intercepts for participants (to account for individual differences in baseline responses) and random slopes for all predictor variables (to allow predictor effects to vary across participants).\u003c/p\u003e\u003cp\u003e\u003cb\u003eHierarchical Drift Diffusion Model\u003c/b\u003e. To decompose performance into cognitive subprocesses, we fit Hierarchical Drift Diffusion Models (HDDMs) to trial-level data, including RTs and response accuracy (incorrect = 0 and correct = 1) for each Stroop target, across different social comparison conditions. HDDM estimates two key parameters linked to effort and decision-making: drift rate (v, representing the speed of evidence accumulation) and response threshold (a, indicating the degree of response caution) under different social comparison conditions (with upward social comparison set as the intercept). Additionally, we investigated how drift rate and threshold were influenced by model-based efficacy estimates.\u003c/p\u003e\u003cp\u003eFor each experiment, we fitted two models focused on social comparison feedback: one examined the effects of social comparison feedback on drift rate and response threshold, while the other additionally controlled for Stroop trial congruency (congruent vs. incongruent) in both parameters. Model comparison revealed that the model accounting for the effect of Stroop congruency was the best-fitting one for social comparison feedback. Similarly, we fitted two models for efficacy estimates: one assessed the effects of efficacy estimates on drift rate and response threshold, and the other likewise incorporated Stroop congruency as a control predictor for both parameters. Again, model comparison confirmed that the best-fitting model for efficacy estimates was the one that included the effect of Stroop congruency. Finally, we selected these best-fitting models for further analyses on how social comparison and efficacy estimates influence participants’ effort allocation.\u003c/p\u003e\u003cp\u003e\u003cb\u003eMultilevel Mediation Model.\u003c/b\u003e To test whether model-based efficacy estimates mediate the relationship between social comparison feedback and effort investment (indexed by mean RT and CRPS), we conducted two 1-1-1 multilevel mediation models (level-1: rounds; level-2: participants) using round-level data. These models included independent variable (\u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:X\\)\u003c/span\u003e\u003c/span\u003e, social comparison feedback, coded as 0 = upward, 1 = downward) at level-1, mediator (\u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:M\\)\u003c/span\u003e\u003c/span\u003e, model-based efficacy estimates at level-1), and two outcome variables (\u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:{Y}_{1}\\)\u003c/span\u003e\u003c/span\u003e, mean RTs, and \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:{Y}_{2}\\)\u003c/span\u003e\u003c/span\u003e, CRPS all at level-1), which allowed us estimate the mediation effect at the round level.\u003c/p\u003e\u003ch2\u003eExperiment 2\u003c/h2\u003e\u003ch2\u003eParticipants\u003c/h2\u003e\u003cp\u003eWe recruited another 42 healthy participants with non-psychology backgrounds. All participants had normal or corrected-to-normal vision and normal color perception. Two participants were excluded due to crash of computer program, and four participants were excluded for not completing the formal experiment. Additionally, two participants were excluded due to insufficient data per condition (\u0026lt; 30 rounds) after removing the eyeblinks or muscle artifacts (see below EEG acquisition and processing). Consequently, the final sample included 34 participants (9 males, 22.12 ± 1.91 years). We conducted a post hoc power analysis using G*Power \u003csup\u003e\u003cspan citationid=\"CR35\" class=\"CitationRef\"\u003e35\u003c/span\u003e\u003c/sup\u003e for this sample size (n = 34), employing a paired-samples t-test (to test the difference between two dependent means). With an α level of 0.05, the achieved statistical power was 0.88. This study was approved by the Ethics Committee of University, and all participants gave informed consent prior to the experiment.\u003c/p\u003e\u003ch2\u003eExperimental Design\u003c/h2\u003e\u003cp\u003eThe experimental procedure in Experiment 2 was identical to that of Experiment 1, with one modification: the duration of target presentation was shortened to a randomly selected duration of 6 s, 7 s, and 8 s. This adjustment was made to reduce participant fatigue during the EEG experiment. Additionally, participants received a fixed payment of 95 RMB, independent of their task performance.\u003c/p\u003e\u003ch2\u003eEEG Acquisition and Processing\u003c/h2\u003e\u003cp\u003eParticipants were seated 60 cm from the computer monitor in a dimly lit and sound-attenuated room. Electroencephalography (EEG) data were recorded by 64-channel Neuroscan equipment (NeuroScan 4.3.1, USA), following the international 10–20 system of EEG electrode placement. The left mastoid served as the reference electrode, and the center of the forehead served as the ground electrode. Vertical electrooculograms were placed above and below the right eye to record eye blinks, while horizontal electrooculograms were placed at the outer canthi of the eyes to record eye movements. All electrode impedance were kept below 5 kΩ during recording, and the sampling rate was set at 500 Hz.\u003c/p\u003e\u003cp\u003eOffline EEG data were preprocessed using the \u003cem\u003eEEGLAB\u003c/em\u003e toolbox (\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttp://sccn.ucsd.edu/eeglab\u003c/span\u003e\u003cspan address=\"http://sccn.ucsd.edu/eeglab\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e) in \u003cem\u003eMATLAB2018b.\u003c/em\u003e The EEG data were re-referenced to the average of the left and right mastoid electrodes and filtered with a band-pass of 0.1–40 Hz. Independent components analysis (ICA) was performed to detect and remove eye blinks and movement artifacts. For ERP analysis, the cleaned EEG data were then divided into epochs for the cue phase (-200 to 1500 ms) and the feedback phase (-200 to 1600 ms). Baseline correction was applied using a pre-stimulus interval of 200 ms for each epoch. Any epoch with amplitudes exceeding ± 100 µV was excluded from further analysis, resulting in the exclusion of 10.24% of epochs from the cue phase and 11.27% from the feedback phase.\u003c/p\u003e\u003cp\u003eTime-frequency analysis was performed on the preprocessed EEG data using \u003cem\u003eMATLAB2018b\u003c/em\u003e. The ICA-cleaned EEG data were segmented into epochs from − 600 to 1600 ms for both the cue and feedback phases. Epochs with amplitudes exceeding ± 100 µV were removed, resulting in the exclusion of 10.96% of epochs from the cue phase and 12.15% from the feedback phase. The EEG data were then analyzed using short-time Fourier transform (STFT) with a fixed 400 ms Hanning-tapered window. A pre-stimulus time interval from − 400 ms to -200 ms was used for baseline correction, and power values were converted to a decibel scale (dB, 10 × log10).\u003c/p\u003e\u003ch2\u003eBehavioral Analyses\u003c/h2\u003e\u003cp\u003e As in Experiment 1, incorrect trials (7.96% of total trials) were first excluded from each round for each participant. Following this, we calculated the mean RTs for each round. Rounds with mean RTs exceeding three standard deviations (1.39% of total rounds) and CRPS values beyond three standard deviations (1.88% of total rounds) were subsequently removed from each social comparison condition for each participant.\u003c/p\u003e\u003cp\u003eWe used the same LMM, HDDM, and multilevel mediation models as those employed in Experiment 1. For the reinforcement learning model analysis, we also fitted three models (intercept, one-learning-rate, and two-learning-rate), with the two-learning-rate model providing the best fit, consistent with the findings from Experiment 1.\u003c/p\u003e\u003ch2\u003eEEG Analyses\u003c/h2\u003e\u003cp\u003eBased on the collapsed localizers approach \u003csup\u003e\u003cspan citationid=\"CR36\" class=\"CitationRef\"\u003e36\u003c/span\u003e\u003c/sup\u003e and previous studies, we selected the ERP and time-frequency component within the specific time windows and electrode sites. Following this, we extracted round-level data for all ERP components under different social comparison feedback conditions for each participant.\u003c/p\u003e\u003cp\u003eFor the feedback phase, we focused on the RewP component (280–340 ms, across C1, Cz, and C2 electrodes), the fb-P3 component (350–450 ms, across CP1, CPz, and CP2 electrodes) and the LPP component (600–1000 ms, across CP1, CPz and CP2 electrodes). The RewP component is a fronto-central positive deflection peaking at 250–350 ms post-feedback onset, which is sensitive to reward feedback \u003csup\u003e\u003cspan citationid=\"CR12\" class=\"CitationRef\"\u003e12\u003c/span\u003e,\u003cspan citationid=\"CR37\" class=\"CitationRef\"\u003e37\u003c/span\u003e,\u003cspan citationid=\"CR38\" class=\"CitationRef\"\u003e38\u003c/span\u003e\u003c/sup\u003e. The fb-P3 component is a centro-parietal positive deflection that peaks at 300–600 ms post-feedback onset, reflecting the evaluation of feedback valence and magnitudes \u003csup\u003e\u003cspan citationid=\"CR10\" class=\"CitationRef\"\u003e10\u003c/span\u003e,\u003cspan citationid=\"CR12\" class=\"CitationRef\"\u003e12\u003c/span\u003e,\u003cspan citationid=\"CR37\" class=\"CitationRef\"\u003e37\u003c/span\u003e\u003c/sup\u003e. The LPP component is a sustained positive deflection occurring post-feedback onset over centro-parietal electrodes, indicating emotional arousal levels elicited by feedback, with larger amplitudes corresponding to higher arousal \u003csup\u003e\u003cspan citationid=\"CR12\" class=\"CitationRef\"\u003e12\u003c/span\u003e,\u003cspan citationid=\"CR14\" class=\"CitationRef\"\u003e14\u003c/span\u003e\u003c/sup\u003e.\u003c/p\u003e\u003cp\u003eFor these feedback-locked EEG components, linear mixed models were employed to examine the impact of social comparison feedback on the RewP, fb-P3, LPP. All models included age and gender as fixed effects to control for potential demographic influences, with predictor variables treated as both fixed and random effects.\u003c/p\u003e\u003cp\u003eFor the cue phase, we focused on the CNV component and cue-beta band activity. The CNV component, characterized as a slowly developing negative potential over fronto-central electrodes following cue onset, was analyzed across FC1, FCz, and FC2 electrodes from 550 ms to 950 ms post-cue onset. This component reflects increased motor preparation for the upcoming task in response to incentive or effort cues \u003csup\u003e\u003cspan additionalcitationids=\"CR39 CR40\" citationid=\"CR38\" class=\"CitationRef\"\u003e38\u003c/span\u003e–\u003cspan citationid=\"CR41\" class=\"CitationRef\"\u003e41\u003c/span\u003e\u003c/sup\u003e. Additionally, the cue-beta band was analyzed across F3, F1, and Fz electrodes from 150 ms to 600 ms post-cue onset, focusing on the frequency range of 15–21 Hz. Increased beta suppression during the cue-evaluation phase has been associated with the preparation and execution of upcoming responses \u003csup\u003e\u003cspan citationid=\"CR37\" class=\"CitationRef\"\u003e37\u003c/span\u003e,\u003cspan citationid=\"CR42\" class=\"CitationRef\"\u003e42\u003c/span\u003e\u003c/sup\u003e.\u003c/p\u003e\u003cp\u003eAligning with the behavioral analyses, we encoded the cue-code for the next round (t + 1) based on the feedback-code (t) of the current round. This approach allowed us to specifically analyze the effect of social comparison feedback (t) on preparatory engagement of effort investment, as indexed by the CNV and cue-beta activity in the subsequent round (t + 1).\u003c/p\u003e\u003cp\u003eFor cue-locked EEG components, linear mixed models were performed to explore the influence of social comparison on the CNV and cue-beta power, as well as the effect of model-based efficacy estimation on the CNV and cue-beta power. All models incorporated age and gender as fixed effects to control for potential demographic influences, while prediction variables were incorporated as both fixed and random effects simultaneously.\u003c/p\u003e\u003ch2\u003eStatistics and reproducibility\u003c/h2\u003e\u003ch2\u003eReinforcement Learning Models\u003c/h2\u003e\u003cp\u003eTo fit the reinforcement learning model parameters, we used two inputs: (1) social comparison feedback from each round, and (2) participants’ actual efficacy estimates (operationalized as their self-reported probability-of-wining ratings, z-scored per participant to standardize across individuals). To ensure the coherence and completeness of the empirical data, we also included rounds that involved lateral social comparisons. The actual feedback (\u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:{f}_{t}\\)\u003c/span\u003e\u003c/span\u003e) was assigned a value of 1 for downward social comparison feedback, 0 for upward social comparison feedback, and 0.5 for lateral social comparison feedback. The prediction error (\u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:{\\delta\\:}_{t}\\)\u003c/span\u003e\u003c/span\u003e) defined as the discrepancy between the actual feedback (\u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:{f}_{t}\\)\u003c/span\u003e\u003c/span\u003e) and the model’s current efficacy estimate (\u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:{E}_{t}\\)\u003c/span\u003e\u003c/span\u003e). The learning rate (\u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:\\alpha\\:\\)\u003c/span\u003e\u003c/span\u003e) scaled the impact of the prediction error (\u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:{\\delta\\:}_{t}\\)\u003c/span\u003e\u003c/span\u003e) on updates to the next round’s efficacy estimate (\u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:{E}_{t+1}\\)\u003c/span\u003e\u003c/span\u003e) \u003csup\u003e29\u003c/sup\u003e.\u003c/p\u003e\u003cp\u003eWe fitted three candidate models to estimate α for each participant, using maximum likelihood estimation (MLE) to minimize the difference between the model’s predicted estimates and participants’ observed efficacy estimates. The first model fitted a single learning rate for both positive and negative prediction errors, while the second model fitted two distinct learning rates: a positive learning rate for positive prediction errors and a negative learning rate for negative prediction errors. Additionally, we implemented an intercept model, which assumed the efficacy estimate remained constant throughout the experiment without any updates. After comparing Akaike information criterion (AIC) and Bayesian information criterion (BIC) values (where lower values indicate a better fit), we determined that the two-learning-rate model provided the best fit.\u003c/p\u003e\u003cp\u003eTo validate the two-learning-rate model, we conducted a parameter recovery analysis. We first generated data for 200 synthetic participants (144 rounds each) using empirical parameter distributions: a fixed noise parameter of 0.15, a fixed intercept of 0.5, and learning rates sampled from a uniform distribution ranging from 0.001 to 0.5. We then fitted the two-learning-rate model to the simulated dataset in an attempt to recover the learning rate parameters. Finally, we computed Pearson correlations between the recovered learning rates parameters and the input learning rates parameters \u003csup\u003e\u003cspan citationid=\"CR30\" class=\"CitationRef\"\u003e30\u003c/span\u003e\u003c/sup\u003e. The results indicated that we successfully recovered the parameters for the two-learning-rate model.\u003c/p\u003e\u003cp\u003eLastly, we calculated the model-based efficacy estimate of each round for each participant based on the parameters from the two-learning-rate model (\u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:{\\alpha\\:}_{pos}\\)\u003c/span\u003e\u003c/span\u003e, \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:{\\alpha\\:}_{neg}\\)\u003c/span\u003e\u003c/span\u003e, initial value). The model began with an initial value (\u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:{E}_{1}\\)\u003c/span\u003e\u003c/span\u003e) and was updated using the learning rates (\u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:{\\alpha\\:}_{pos}\\)\u003c/span\u003e\u003c/span\u003e, \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:{\\alpha\\:}_{neg}\\)\u003c/span\u003e\u003c/span\u003e) and the prediction error (\u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:{\\delta\\:}_{t}\\)\u003c/span\u003e\u003c/span\u003e). Learning models were fitted, compared, and recovered by \u003cem\u003eMATLAB2018b\u003c/em\u003e, while the model-based efficacy estimates were calculated in R-Studio.\u003c/p\u003e\u003ch2\u003eLinear Mixed Model\u003c/h2\u003e\u003cp\u003eLMMs were performed in R-Studio using the lme4 package for modeling \u003csup\u003e\u003cspan citationid=\"CR31\" class=\"CitationRef\"\u003e31\u003c/span\u003e\u003c/sup\u003e and the lmerTest package for fixed-effects testing and \u003cem\u003ep-\u003c/em\u003evalue estimation \u003csup\u003e\u003cspan citationid=\"CR32\" class=\"CitationRef\"\u003e32\u003c/span\u003e\u003c/sup\u003e. Standardized regression coefficients and 95% confidence intervals were derived using the model_parameters function from the parameters package. Statistical significance was set at \u003cem\u003ep\u003c/em\u003e \u0026lt; 0.05.\u003c/p\u003e\u003ch2\u003eHierarchical Drift Diffusion Model\u003c/h2\u003e\u003cp\u003eAll models utilized 5 Markov Chain Monte Carlo (MCMC) chains, drawing 7,000 samples and discarding the first 4,000 samples as burn-in for each chain. Model convergence was assessed using the Gelman-Rubin statistic (\u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:\\widehat{\\text{R}}\\)\u003c/span\u003e\u003c/span\u003e); all model \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:\\widehat{\\text{R}}\\)\u003c/span\u003e\u003c/span\u003e values close to 1 and below 1.1 indicated good convergence and minimal variation between chains \u003csup\u003e\u003cspan citationid=\"CR33\" class=\"CitationRef\"\u003e33\u003c/span\u003e\u003c/sup\u003e. Model comparison was performed using the deviance information criterion (DIC), where lower values indicate a better fit, to select the optimal model \u003csup\u003e\u003cspan citationid=\"CR34\" class=\"CitationRef\"\u003e34\u003c/span\u003e\u003c/sup\u003e. Posterior predictive checks were performed to evaluate the ability of the optimal model to reproduce the observed data. Specifically, 500 parameter samples were drawn from the posterior distribution of the fitted model, and these samples were used to simulate new datasets for comparison with the observed data. Across all optimal models, the statistics of observed data fell within the 95% credible intervals of the statistics derived from the simulated datasets, indicating that each model adequately captured the patterns in the observed data.\u003c/p\u003e\u003cp\u003eModel fit, comparison and posterior predictive checks were performed using the HDDM package in Python (\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://secure.travis-ci.org/hddm-devs/hddm.png?branch=master\u003c/span\u003e\u003cspan address=\"https://secure.travis-ci.org/hddm-devs/hddm.png?branch=master\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e). Bayesian hypothesis tests were conducted separately in R-Studio using the brms package.\u003c/p\u003e\u003ch2\u003eMultilevel Mediation Model\u003c/h2\u003e\u003cp\u003eThe multilevel mediation analyses were performed using the MLmed macro (\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://njrockwood.com/mlmed\u003c/span\u003e\u003cspan address=\"https://njrockwood.com/mlmed\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e) in SPSS 25.00. Model estimation was based on Restricted Maximum Likelihood (REML). Indirect effects were assessed via Monte Carlo confidence intervals with 10,000 simulated samples and considered statistically significant at \u003cem\u003ep\u003c/em\u003e \u0026lt; .05 when the 95% Monte Carlo confidence interval (CI) did not include zero.\u003c/p\u003e"},{"header":"Declarations","content":"\u003cp\u003e\u003ch2\u003eCompeting interests\u003c/h2\u003e\u003cp\u003eAll authors declare no conflict of interests.\u003c/p\u003e\u003c/p\u003e\u003cp\u003e\u003cstrong\u003eData\u0026nbsp;\u003c/strong\u003e\u003cstrong\u003ea\u003c/strong\u003e\u003cstrong\u003evailability\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eAll data used in this study are publicly available on the Open Science Framework (OSF): https://osf.io/vpr9e/.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eCode availability\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eAll analysis code used in this study are publicly available on the Open Science Framework (OSF): https://osf.io/vpr9e/.\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\n\u003cli\u003eCorcoran, K., Crusius, J. \u0026amp; Mussweiler, T. Social comparison: Motives, standards, and mechanisms. in \u003cem\u003eTheories in Social Psychology\u003c/em\u003e 119\u0026ndash;139 (Wiley-Blackwell, 2011).\u003c/li\u003e\n\u003cli\u003eFestinger, L. A theory of social comparison processes. \u003cem\u003eHum. Relat.\u003c/em\u003e \u003cstrong\u003e7\u003c/strong\u003e, 117\u0026ndash;140 (1954).\u003c/li\u003e\n\u003cli\u003eDiel, K., Grelle, S. \u0026amp; Hofmann, W. A motivational framework of social comparison. \u003cem\u003eJ. Pers. Soc. Psychol.\u003c/em\u003e \u003cstrong\u003e120\u003c/strong\u003e, 1415\u0026ndash;1430 (2021).\u003c/li\u003e\n\u003cli\u003eGerber, J. P., Wheeler, L. \u0026amp; Suls, J. A social comparison theory meta-analysis 60+ years on. \u003cem\u003ePsychol. Bull.\u003c/em\u003e \u003cstrong\u003e144\u003c/strong\u003e, 177\u0026ndash;197 (2018).\u003c/li\u003e\n\u003cli\u003eBandura, A. Self-efficacy: Toward a unifying theory of behavioral change. \u003cem\u003ePsychol. Rev.\u003c/em\u003e \u003cstrong\u003e84\u003c/strong\u003e, 191\u0026ndash;215 (1977).\u003c/li\u003e\n\u003cli\u003eBandura, A. \u003cem\u003eSelf Efficacy: The Exercise of Control\u003c/em\u003e. (W. H. Freeman, New York (N.Y.), 1997).\u003c/li\u003e\n\u003cli\u003eM\u0026uuml;ller-Pinzler, L. \u003cem\u003eet al.\u003c/em\u003e Neurocomputational mechanisms of affected beliefs. \u003cem\u003eCommun. Biol.\u003c/em\u003e \u003cstrong\u003e5\u003c/strong\u003e, 1241 (2022).\u003c/li\u003e\n\u003cli\u003eBoksem, M. A. S., Kostermans, E. \u0026amp; De Cremer, D. Failing where others have succeeded: Medial Frontal Negativity tracks failure in a social context. Psychophysiology 48, 973\u0026ndash;979 (2011).\u003c/li\u003e\n\u003cli\u003eLin, H. \u0026amp; Liang, J. Working memory load reduces the processing of outcome evaluation involving others but not oneself: Event‐related potential evidence. \u003cem\u003ePsychophysiology\u003c/em\u003e \u003cstrong\u003e58\u003c/strong\u003e, (2021).\u003c/li\u003e\n\u003cli\u003eWu, Y., Zhang, D., Elieson, B. \u0026amp; Zhou, X. Brain potentials in outcome evaluation: When social comparison takes effect. \u003cem\u003eInt. J. Psychophysiol.\u003c/em\u003e \u003cstrong\u003e85\u003c/strong\u003e, 145\u0026ndash;152 (2012).\u003c/li\u003e\n\u003cli\u003eDu, X. \u003cem\u003eet al.\u003c/em\u003e The neural circuitry of reward processing in complex social comparison: evidence from an event-related FMRI study. \u003cem\u003ePLoS ONE\u003c/em\u003e \u003cstrong\u003e8\u003c/strong\u003e, e82534 (2013).\u003c/li\u003e\n\u003cli\u003eHu, X., Xu, Z. \u0026amp; Mai, X. Social value orientation modulates the processing of outcome evaluation involving others. \u003cem\u003eSoc. Cogn. Affect. Neurosci.\u003c/em\u003e \u003cstrong\u003e12\u003c/strong\u003e, 1730\u0026ndash;1739 (2017).\u003c/li\u003e\n\u003cli\u003eLi, Q. \u003cem\u003eet al.\u003c/em\u003e Social comparisons differentially affect fair and unfair processing in ultimatum bargaining. \u003cem\u003eNeuropsychologia\u003c/em\u003e \u003cstrong\u003e174\u003c/strong\u003e, 108318 (2022).\u003c/li\u003e\n\u003cli\u003eLuo, Y. \u003cem\u003eet al.\u003c/em\u003e Social Comparison Manifests in Event-related Potentials. \u003cem\u003eSci. Rep.\u003c/em\u003e \u003cstrong\u003e5\u003c/strong\u003e, 12127 (2015).\u003c/li\u003e\n\u003cli\u003ePeng, M. \u003cem\u003eet al.\u003c/em\u003e Good luck or bad luck? The influence of social comparison on risk‐taking decision and the underlying neural mechanism. \u003cem\u003ePsychophysiology\u003c/em\u003e e14730 (2024) doi:10.1111/psyp.14730.\u003c/li\u003e\n\u003cli\u003eZhang, H. \u003cem\u003eet al.\u003c/em\u003e Interpersonal relationships modulate outcome evaluation in a social comparison context: The pain and pleasure of intimacy. \u003cem\u003eCogn. Affect. Behav. Neurosci.\u003c/em\u003e \u003cstrong\u003e20\u003c/strong\u003e, 115\u0026ndash;127 (2020).\u003c/li\u003e\n\u003cli\u003eLindner, M. \u003cem\u003eet al.\u003c/em\u003e Neural patterns underlying social comparisons of personal performance. \u003cem\u003eSoc. Cogn. Affect. Neurosci.\u003c/em\u003e \u003cstrong\u003e10\u003c/strong\u003e, 569\u0026ndash;576 (2015).\u003c/li\u003e\n\u003cli\u003eShen, Q. \u003cem\u003eet al.\u003c/em\u003e To Reveal or Not to Reveal? Observation of Social Outcomes Facilitates Reward Processing. \u003cem\u003eFront. Neurosci.\u003c/em\u003e \u003cstrong\u003e14\u003c/strong\u003e, 579702 (2021).\u003c/li\u003e\n\u003cli\u003eWang, L., Zhang, X., Li, L. \u0026amp; Meng, L. Taking Others as a Mirror: Contingent Social Comparison Promotes Task Engagement. \u003cem\u003eFront. Hum. Neurosci.\u003c/em\u003e \u003cstrong\u003e12\u003c/strong\u003e, 476 (2018).\u003c/li\u003e\n\u003cli\u003eSchr\u0026ouml;der, A. \u003cem\u003eet al.\u003c/em\u003e Computational Modeling shows Confirmation Bias during Formation and Revision of Self-Beliefs. Preprint at https://doi.org/10.1101/2024.08.30.610443 (2024).\u003c/li\u003e\n\u003cli\u003eWittmann, M. K. \u003cem\u003eet al.\u003c/em\u003e Self-Other Mergence in the Frontal Cortex during Cooperation and Competition. \u003cem\u003eNeuron\u003c/em\u003e \u003cstrong\u003e91\u003c/strong\u003e, 482\u0026ndash;493 (2016).\u003c/li\u003e\n\u003cli\u003eM\u0026uuml;ller-Pinzler, L. \u003cem\u003eet al.\u003c/em\u003e Negativity-bias in forming beliefs about own abilities. \u003cem\u003eSci. Rep.\u003c/em\u003e \u003cstrong\u003e9\u003c/strong\u003e, 14416 (2019).\u003c/li\u003e\n\u003cli\u003eBault, N., Joffily, M., Rustichini, A. \u0026amp; Coricelli, G. Medial prefrontal cortex and striatum mediate the influence of social comparison on the decision process. \u003cem\u003eProc. Natl. Acad. Sci.\u003c/em\u003e \u003cstrong\u003e108\u003c/strong\u003e, 16044\u0026ndash;16049 (2011).\u003c/li\u003e\n\u003cli\u003eLiu, Z. \u003cem\u003eet al.\u003c/em\u003e Social comparison modulates the neural responses to regret and subsequent risk-taking behavior. \u003cem\u003eSoc. Cogn. Affect. Neurosci.\u003c/em\u003e \u003cstrong\u003e13\u003c/strong\u003e, 1059\u0026ndash;1070 (2018).\u003c/li\u003e\n\u003cli\u003eTor, A. \u0026amp; Garcia, S. M. The neuroscience of social comparison and competition. \u003cem\u003eCogn. Affect. Behav. Neurosci.\u003c/em\u003e \u003cstrong\u003e23\u003c/strong\u003e, 920\u0026ndash;943 (2023).\u003c/li\u003e\n\u003cli\u003eBuunk, B. P., Collins, R. L., Taylor, S. E., VanYperen, N. W. \u0026amp; Dakof, G. A. The affective consequences of social comparison: either direction has its ups and downs. \u003cem\u003eJ. Pers. Soc. Psychol.\u003c/em\u003e \u003cstrong\u003e59\u003c/strong\u003e, 1238\u0026ndash;1249 (1990).\u003c/li\u003e\n\u003cli\u003eDiel, K., Hofmann, W., Grelle, S., Boecker, L. \u0026amp; Friese, M. Prepare to Compare: Effects of an Intervention Involving Upward and Downward Social Comparisons on Goal Pursuit in Daily Life. \u003cem\u003ePers. Soc. Psychol. Bull.\u003c/em\u003e 01461672231219378 (2024) doi:10.1177/01461672231219378.\u003c/li\u003e\n\u003cli\u003eLeng, X., Yee, D., Ritz, H. \u0026amp; Shenhav, A. Dissociable influences of reward and punishment on adaptive cognitive control. \u003cem\u003ePLOS Comput. Biol.\u003c/em\u003e \u003cstrong\u003e17\u003c/strong\u003e, e1009737 (2021).\u003c/li\u003e\n\u003cli\u003eZhang, L., Lengersdorff, L., Mikus, N., Gl\u0026auml;scher, J. \u0026amp; Lamm, C. Using reinforcement learning models in social neuroscience: frameworks, pitfalls and suggestions of best practices. \u003cem\u003eSoc. Cogn. Affect. Neurosci.\u003c/em\u003e \u003cstrong\u003e15\u003c/strong\u003e, 695\u0026ndash;707 (2020).\u003c/li\u003e\n\u003cli\u003eWilson, R. C. \u0026amp; Collins, A. G. Ten simple rules for the computational modeling of behavioral data. \u003cem\u003eeLife\u003c/em\u003e \u003cstrong\u003e8\u003c/strong\u003e, e49547 (2019).\u003c/li\u003e\n\u003cli\u003eBates, D., M\u0026auml;chler, M., Bolker, B. \u0026amp; Walker, S. Fitting Linear Mixed-Effects Models Using \u003cstrong\u003elme4\u003c/strong\u003e. \u003cem\u003eJ. Stat. Softw.\u003c/em\u003e \u003cstrong\u003e67\u003c/strong\u003e, (2015).\u003c/li\u003e\n\u003cli\u003eKuznetsova, A., Brockhoff, P. B. \u0026amp; Christensen, R. H. B. \u003cstrong\u003elmerTest\u003c/strong\u003e Package: Tests in Linear Mixed Effects Models. \u003cem\u003eJ. Stat. Softw.\u003c/em\u003e \u003cstrong\u003e82\u003c/strong\u003e, (2017).\u003c/li\u003e\n\u003cli\u003ePedersen, M. L. \u0026amp; Frank, M. J. Simultaneous Hierarchical Bayesian Parameter Estimation for Reinforcement Learning and Drift Diffusion Models: a Tutorial and Links to Neural Data. \u003cem\u003eComput. Brain Behav.\u003c/em\u003e \u003cstrong\u003e3\u003c/strong\u003e, 458\u0026ndash;471 (2020).\u003c/li\u003e\n\u003cli\u003eSpiegelhalter, D. J., Best, N. G., Carlin, B. P. \u0026amp; Van Der Linde, A. Bayesian Measures of Model Complexity and Fit. \u003cem\u003eJ. R. Stat. Soc. Ser. B Stat. Methodol.\u003c/em\u003e \u003cstrong\u003e64\u003c/strong\u003e, 583\u0026ndash;639 (2002).\u003c/li\u003e\n\u003cli\u003eFaul, F., Erdfelder, E., Lang, A.-G. \u0026amp; Buchner, A. G*Power 3: A flexible statistical power analysis program for the social, behavioral, and biomedical sciences. \u003cem\u003eBehav. Res. Methods\u003c/em\u003e \u003cstrong\u003e39\u003c/strong\u003e, 175\u0026ndash;191 (2007).\u003c/li\u003e\n\u003cli\u003eLuck, S. J. \u0026amp; Gaspelin, N. How to get statistically significant effects in any ERP experiment (and why you shouldn\u0026rsquo;t). \u003cem\u003ePsychophysiology\u003c/em\u003e \u003cstrong\u003e54\u003c/strong\u003e, 146\u0026ndash;157 (2017).\u003c/li\u003e\n\u003cli\u003eGlazer, J. E., Kelley, N. J., Pornpattananangkul, N., Mittal, V. A. \u0026amp; Nusslock, R. Beyond the FRN: Broadening the time-course of EEG and ERP components implicated in reward processing. \u003cem\u003eInt. J. Psychophysiol.\u003c/em\u003e \u003cstrong\u003e132\u003c/strong\u003e, 184\u0026ndash;202 (2018).\u003c/li\u003e\n\u003cli\u003eZhang, Y. \u003cem\u003eet al.\u003c/em\u003e Comparing monetary gain and loss in the monetary incentive delay task: EEG evidence. \u003cem\u003ePsychophysiology\u003c/em\u003e \u003cstrong\u003e60\u003c/strong\u003e, e14383 (2023).\u003c/li\u003e\n\u003cli\u003eFr\u0026ouml;mer, R., Lin, H., Dean Wolf, C. K., Inzlicht, M. \u0026amp; Shenhav, A. Expectations of reward and efficacy guide cognitive control allocation. \u003cem\u003eNat. Commun.\u003c/em\u003e \u003cstrong\u003e12\u003c/strong\u003e, 1030 (2021).\u003c/li\u003e\n\u003cli\u003eRong, Y. \u003cem\u003eet al.\u003c/em\u003e Expectations of immediate and delayed reward differentially affect cognitive task performance. \u003cem\u003eNeuroImage\u003c/em\u003e \u003cstrong\u003e262\u003c/strong\u003e, 119582 (2022).\u003c/li\u003e\n\u003cli\u003eSchevernels, H., Krebs, R. M., Santens, P., Woldorff, M. G. \u0026amp; Boehler, C. N. Task preparation processes related to reward prediction precede those related to task-difficulty expectation. \u003cem\u003eNeuroImage\u003c/em\u003e \u003cstrong\u003e84\u003c/strong\u003e, 639\u0026ndash;647 (2014).\u003c/li\u003e\n\u003cli\u003eGable, P. A., Threadgill, A. H. \u0026amp; Adams, D. L. Neural activity underlying motor-action preparation and cognitive narrowing in approach-motivated goal states. \u003cem\u003eCogn. Affect. Behav. Neurosci.\u003c/em\u003e \u003cstrong\u003e16\u003c/strong\u003e, 145\u0026ndash;152 (2016).\u003c/li\u003e\n\u003cli\u003eLin, H. \u0026amp; Liang, J. Comparison with others influences encoding and recognition of their faces: Behavioural and ERP evidence. \u003cem\u003eNeuroImage\u003c/em\u003e \u003cstrong\u003e288\u003c/strong\u003e, 120538 (2024).\u003c/li\u003e\n\u003cli\u003eFliessbach, K. \u003cem\u003eet al.\u003c/em\u003e Social comparison affects reward-related brain activity in the human ventral striatum. \u003cem\u003eScience\u003c/em\u003e \u003cstrong\u003e318\u003c/strong\u003e, 1305\u0026ndash;1308 (2007).\u003c/li\u003e\n\u003cli\u003eZou, F. \u003cem\u003eet al.\u003c/em\u003e P2 Manifests Subjective Evaluation of Reward Processing Under Social Comparison. \u003cem\u003eFront. Psychol.\u003c/em\u003e \u003cstrong\u003e13\u003c/strong\u003e, 817529 (2022).\u003c/li\u003e\n\u003cli\u003eLuo, Y., Eickhoff, S. B., H\u0026eacute;tu, S. \u0026amp; Feng, C. Social comparison in the brain: A coordinate-based meta-analysis of functional brain imaging studies on the downward and upward comparisons. \u003cem\u003eHum. Brain Mapp.\u003c/em\u003e \u003cstrong\u003e39\u003c/strong\u003e, 440\u0026ndash;458 (2018).\u003c/li\u003e\n\u003cli\u003eScollon, C. N., Kim-Prieto, C. \u0026amp; Scollon, C. N. Experience Sampling: Promises and Pitfalls, Strengths and Weaknesses. \u003cem\u003eJ. Happiness Stud.\u003c/em\u003e \u003cstrong\u003e4\u003c/strong\u003e, 5\u0026ndash;34 (2003).\u003c/li\u003e\n\u003cli\u003eSheeran, P. Intention\u0026ndash;Behavior Relations: A Conceptual and Empirical Review. in \u003cem\u003eEuropean Review of Social Psychology\u003c/em\u003e (eds Stroebe, W. \u0026amp; Hewstone, M.) 1\u0026ndash;36 (Wiley, 2001). doi:10.1002/0470013478.ch1.\u003c/li\u003e\n\u003cli\u003eGrahek, I., Fr\u0026ouml;mer, R., Prater Fahey, M. \u0026amp; Shenhav, A. Learning when effort matters: neural dynamics underlying updating and adaptation to changes in performance efficacy. \u003cem\u003eCereb. Cortex\u003c/em\u003e bhac215 (2022) doi:10.1093/cercor/bhac215.\u003c/li\u003e\n\u003cli\u003eZhou, T. \u003cem\u003eet al.\u003c/em\u003e History of winning remodels thalamo-PFC circuit to reinforce social dominance. \u003cem\u003eScience\u003c/em\u003e \u003cstrong\u003e357\u003c/strong\u003e, 162\u0026ndash;168 (2017).\u003c/li\u003e\n\u003cli\u003eBonabeau, E. Dominance Orders in Animal Societies: The Self-organization Hypothesis Revisited. \u003cem\u003eBull. Math. Biol.\u003c/em\u003e \u003cstrong\u003e61\u003c/strong\u003e, 727\u0026ndash;757 (1999).\u003c/li\u003e\n\u003cli\u003ePage, L. \u0026amp; Coates, J. Winner and loser effects in human competitions. Evidence from equally matched tennis players. \u003cem\u003eEvol. Hum. Behav.\u003c/em\u003e \u003cstrong\u003e38\u003c/strong\u003e, 530\u0026ndash;535 (2017).\u003c/li\u003e\n\u003c/ol\u003e"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":true,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":false,"hideJournal":false,"highlight":"","institution":"","isAcceptedByJournal":true,"isAuthorSuppliedPdf":false,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":false,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"nature-portfolio","isNatureJournal":true,"hasQc":false,"allowDirectSubmit":false,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"","title":"Nature Portfolio","twitterHandle":"","acdcEnabled":false,"dfaEnabled":false,"editorialSystem":"ejp","reportingPortfolio":"","inReviewEnabled":true,"inReviewRevisionsEnabled":false},"keywords":"social comparison, effort, self-efficacy, Reward Positive, CNV","lastPublishedDoi":"10.21203/rs.3.rs-7749786/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-7749786/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003cp\u003ePrevious studies have documented stronger neural responses when participants receive better monetary outcome than others, but how performance-based comparisons shape self-efficacy and subsequent effort behavior remains unclear. We conducted behavioral (Experiment 1, \u003cem\u003eN\u003c/em\u003e\u0026thinsp;=\u0026thinsp;32) and electrophysiological (Experiment 2, \u003cem\u003eN\u003c/em\u003e\u0026thinsp;=\u0026thinsp;34) experiments where participants performed an effortful task and received downward/upward feedback relative to a purported competitor. Computational modeling and mediation analysis revealed dynamic self-efficacy updates, with downward feedback enhancing subsequent effort via full mediation by self-efficacy. Electrophysiological results revealed stronger neural activities in response to downward compared to upward feedback for the current round, and enhanced contingent negative variation (CNV) and cue-beta power, reflecting better task preparation, for the upcoming round. Notably, CNV amplitudes were modulated by efficacy, with higher efficacy predicting more negative-going CNV. These findings demonstrate that performance-based comparison dynamically regulates self-efficacy, thereby shaping both neural preparatory processes and subsequent effort allocation in goal-directed behavior.\u003c/p\u003e","manuscriptTitle":"The neurocomputational mechanisms underlying the impact of social comparison on effort investment","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2025-11-11 16:28:07","doi":"10.21203/rs.3.rs-7749786/v1","editorialEvents":[],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"communications-biology","isNatureJournal":true,"hasQc":false,"allowDirectSubmit":false,"externalIdentity":"commsbio","sideBox":"Learn more about [Communications Biology](http://www.nature.com/commsbio/)","snPcode":"","submissionUrl":"","title":"Communications Biology","twitterHandle":"","acdcEnabled":true,"dfaEnabled":true,"editorialSystem":"ejp","reportingPortfolio":"Communications Series","inReviewEnabled":true,"inReviewRevisionsEnabled":false}}],"origin":"","ownerIdentity":"23f24009-b0ff-431e-b3bd-f527bb37f7e0","owner":[],"postedDate":"November 11th, 2025","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"under-review","subjectAreas":[{"id":57163514,"name":"Biological sciences/Psychology/Human behaviour"},{"id":57163515,"name":"Biological sciences/Neuroscience/Social neuroscience"}],"tags":[],"updatedAt":"2026-04-28T21:07:40+00:00","versionOfRecord":[],"versionCreatedAt":"2025-11-11 16:28:07","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-7749786","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-7749786","identity":"rs-7749786","version":["v1"]},"buildId":"8U1c8b4HqxoKbykW_rLl7","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}