The effect of negative feedback hierarchy on creativity: the mediating role of positive affect and strategy modification Authors: Mengrong Liu, Jingwen Zhao, Weiguo Pang Published: 2025 ABSTRACT AbstractBackgroundNegative feedback, aimed at bridging the gap between current performance and external standards, can either enhance or diminish creative outcomes. On the basis of feedback intervention theory, this study proposes that the negative feedback hierarchy can influence individuals’ creativity through affective (positive and negative) and cognitive (strategy modification) pathways.MethodsA total sample of 264 university students participated across three experiments. Experiment 1 randomly assigned 95 participants to personal negative feedback, task negative feedback and no feedback groups and measured their pre- and post- affects and creativity. Experiment 2 randomly assigned 72 participants to personal negative feedback and task negative feedback groups and measured their strategy modification, pre- and post- motivation and creativity. Experiment 3 randomly assigned 97 participants to the cognitive reappraisal, strategy modification and dual intervention groups and measured their creativity before and after receiving personal negative feedback.ResultsCompared with those receiving personal negative feedback, participants receiving task negative feedback presented increased fluency and originality of Alternative Uses Tasks, reported less reduction in positive affect and displayed greater strategy modification. The relationship between the negative feedback hierarchy and creativity is partially mediated by positive affect and fully mediated by strategy modification.The strategy modification group outperformed the cognitive reappraisal group with respect to improvements in fluency of Realistic Presented Problems. The group receiving both interventions exhibited significant enhancements in fluency of both tasks and flexibility of Realistic Presented Problems, surpassing the other groups.ConclusionsCompared with personal negative feedback, task negative feedback leads to less decline in positive affect and greater strategy modification, thus promoting creativity. Supportive interventions targeting both emotional and cognitive pathways effectively mitigate the adverse effects of personal negative feedback on creativity.Supplementary InformationThe online version contains supplementary material available at 10.1186/s40359-025-03642-8.Keywords: Negative feedback, Feedback hierarchy, Creativity, Affect, Strategy modification Introduction Creativity has become increasingly essential in today’s rapidly evolving world. According to the World Economic Forum's Future of Jobs Report 2025 [77], creativity ranked fourth among both the core skills required in 2025 and the skills expected to gain greater significance by 2030. Given this rising prominence, it is crucial to deepen the understanding of how creativity develops and the factors that shape it. Creativity arises from the interaction between individual or group abilities and the creative process, producing novel and useful outcomes within a social context [58]. In this interactive process, feedback — information from agents such as teachers or peers — plays a crucial role [33, 49], since creators depend on feedback, especially from key figures in their field, to evaluate their work and make decisions about promotion, modification, or abandonment [14, 15, 30]. Feedback can be either positive or negative, indicating whether the creative work meets established standards [35, 81]. While positive feedback has been consistently shown to enhance creativity (e.g., [19, 50, 61]), the effects of negative feedback are less consistent. Some studies suggest that it can stimulate creativity, particularly when compared to receiving no feedback (e.g., [36, 51]), whereas others report inhibitory effects (e.g., [50, 81]). To account for these divergent findings, recent research has examined boundary conditions, showing that factors such as emotional intelligence, goal orientation, and feedback flow can moderate the relationship between negative feedback and creativity [1, 21, 42]. These studies underscore the contextual complexity of negative feedback and highlight when it is more or less likely to foster creativity. However, they shed less light on how negative feedback exerts its influence, making it necessary to investigate the psychological processes that may mediate its effects. In addition to boundary conditions, scholars have examined psychological mechanisms that may mediate the effects of feedback on creativity, focusing on affective, cognitive, and motivational pathways. First, affective processes have received considerable attention. de Rooij et al. [19] designed an interactive system that manipulated feedback about the originality of participants’ ideas during alternative uses tasks. They found that positive feedback on originality increased positive affect (satisfaction), whereas negative feedback increased negative affect (frustration). Crucially, the effect of feedback on originality was fully mediated by these emotional responses, providing causal evidence that positive affect augments creativity during idea generation. However, their assessment of creativity relied solely on originality, without considering fluency or flexibility, and the study primarily emphasized positive rather than negative feedback. Thus, it remains unclear whether and how negative feedback influences creativity through emotional pathways. Second, cognitive mechanisms have been explored through the concept of cognitive load. Redifer et al. [61] examined the joint effects of creative self-efficacy and feedback valence on divergent thinking performance. Across unusual uses and metaphor tasks, they found that positive feedback and higher self-efficacy reduced cognitive load, whereas negative feedback increased it. Elevated cognitive load, in turn, impaired fluency but was unrelated to originality, indicating that cognitive load partially mediated the effects of feedback and self-efficacy on creative performance. This study demonstrates the importance of cognitive resources in linking feedback to creativity, yet also shows that cognitive load is insufficient to fully account for creative outcomes, particularly those involving idea quality. Third, motivational factors have been considered as potential mediators. Drawing on regulatory focus theory, Van Dijk and Kluger [73] conducted a series of scenario experiments and a laboratory study to test how task type moderates the impact of positive versus negative feedback on motivation and performance. They found that creative, promotion-focused tasks elicited higher motivation and better performance following positive feedback, whereas accuracy-based, prevention-focused tasks elicited higher motivation and performance following negative feedback. These findings highlight the role of motivational fit in explaining variability in feedback effects. Nevertheless, motivation in their studies was primarily measured through single-item self-reports. More importantly, the findings explain variability in motivation but do not directly clarify how negative feedback operates through motivational pathways to affect creativity. Taken together, prior research underscores the value of examining mediating mechanisms but also reveals important limitations. First, many studies directly compare negative feedback to positive feedback, which may not be the most appropriate approach. In real-world settings, evaluators often replace negative feedback with silence to avoid defensiveness and interpersonal conflict, and may even alter the feedback [31, 80]. This practice highlights the limitations of using positive feedback as the benchmark for examining the effects of negative feedback. Second, although prior studies have proposed affective, cognitive, and motivational pathways, these investigations remain fragmented and limited. Specifically, some studies have examined only particular outcome indicators of creativity (e.g., fluency but not originality), others have tested mechanisms primarily in positive feedback contexts, and still others have relied on indirect inferences. As a result, the precise processes through which negative feedback shapes creativity remain unclear. Finally, research has rarely addressed intervention mechanisms for individuals already receiving negative feedback, underscoring the need to identify strategies that can mitigate its adverse effects on creativity. To address these limitations, a more systematic framework is needed to clarify the mechanisms through which negative feedback influences creativity. Feedback Intervention Theory (FIT; [44]) provides such a framework. FIT is based on five key arguments: (1) behavioral regulation occurs by comparing the gap between feedback and the goal/standard,(2) goals/standards are organized hierarchically; (3) individuals regulate their behavior only when they perceive a gap between their current state and the goal/standard; (4) limited attentional resources are typically allocated to a moderate level of the hierarchy; and (5) feedback directs attentional resources to different hierarchical levels, thereby influencing individual behavior. Drawing from these arguments, FIT categorizes goals into three hierarchical levels—personal, focal task, and task detail goals. Typically, attention is focused on the medium level (focal task) level, while feedback cues redirect attention to different levels. As illustrated in Fig. 1, personal (self) negative feedback (e.g., ability ranking) shifts attention from the task to personal goals (e.g., maintaining self-esteem, or gaining a sense of control), often triggering strong affective reactions that deplete cognitive resources and undermine performance. Conversely, task negative feedback (e.g., task speed) keeps attention focused on task-level goals (e.g., improving task performance), motivating individuals to adjust strategies and sustain effort to improve performance. Negative feedback at the task-detail level (e.g., correcting an error) redirects attention to very specific actions and is more relevant in domains with clear right-wrong answers [22], but has limited implications for originality and usefulness in creativity tasks [63]. Therefore, personal and task negative feedback are more relevant for understanding creativity-related processes. Schematic overview of feedback intervention theory [44] The present study Based on FIT, the primary factor influencing the varying effects of negative feedback on creativity is the hierarchical level to which negative feedback directs an individual's attentional resources. This factor is referred to as the negative feedback hierarchy (NFH) in this study. The differential effects of NFH arise because personal negative feedback (PNF) tends to elicit stronger emotional reactions and deplete cognitive resources, whereas task negative feedback (TNF) is more effective in enhancing task motivation. Notably, the affective, cognitive, and motivational pathways highlighted by FIT correspond closely to the mechanisms that prior research has proposed in explaining feedback effects. Building on this convergence, the present study investigates these three mediators in the context of negative feedback and further seeks to answer three key questions: (1) What is the effect of NFH on creativity? (2) Does NFH influence creativity through affect, cognition, and motivation? (3) How can the adverse effects of personal negative feedback be mitigated? To operationalize the concept of cognitive pathway, this study uses strategy modification—defined as the extent to which an individual develops and refines task-specific strategies to complete a task. Strategy modification is closely linked to cognitive resources and serves as a representative measure of an individual’s cognitive processes [55]. The research is organized into of three experiments: Experiment 1 examines the impact of NFH on creativity and the mediating role of affect. Experiment 2 verifies the effect of NFH on creativity and explores the mediating roles of motivation and strategy modification. Experiment 3 implements interventions targeting the identified mediators to determine effective strategies for reducing the negative effects of PNF. By clarifying these mechanisms, the present research contributes both theoretically and practically. Theoretically, it enriches feedback intervention theory by identifying affective and cognitive pathways through which different levels of negative feedback shape creativity. Practically, it provides evidence-based insights to help evaluators deliver negative feedback in ways that foster, rather than hinder, creative performance [11, 49], and to enable creators to respond more effectively. These findings also offer actionable guidance for organizations and educators seeking to cultivate and protect creativity in environments where negative feedback is prevalent [16, 71, 82]. Experiment 1 Hypotheses Although relatively few studies have directly examined the role of NFH on creativity, several lines of evidence provide indirect support. For example, Kim and Kim [42] found that bottom-up feedback was more conducive to creativity than top-down or lateral feedback, as it was less likely to challenge recipients’ sense of competence. Zhou [81] similarly reported that informational negative feedback enhanced creative performance compared with controlling feedback. Zhang [79] further demonstrated that negative feedback on creative outcomes enhanced creativity. Although these findings appear heterogeneous, they can be coherently interpreted within the framework of FIT. According to FIT, the impact of negative feedback on performance depends on the hierarchical level at which attention is directed. When feedback targets personal abilities or traits (PNF), it tends to shift limited attentional resources away from the focal task and toward higher-level self-goals, such as protecting self-esteem. Because PNF threatens one’s self-image, it often elicits strong negative affect, thereby impairing creativity. In contrast, task-focused negative feedback (TNF) highlights deficiencies in effort or strategy rather than in personal competence. By directing attention to task-level goals, TNF encourages problem-solving and adaptation, which can facilitate creative performance. Thus, FIT offers a theoretical account for why some forms of negative feedback hinder creativity, whereas others promote it. Thus, we hypothesize: Hypothesis 1: NFH affects individuals’ creativity. Specifically, PNF decreases creativity, while TNF increases creativity. Hypothesis 1: NFH affects individuals’ creativity. Specifically, PNF decreases creativity, while TNF increases creativity. Research consistently shows that negative feedback decreases positive affect and increases negative affect [13, 29, 35, 43, 53]. Although relatively few studies have directly compared the effects of NFH on affect, insights can be drawn from related research on constructive versus destructive criticism. Constructive criticism typically emphasizes task improvement and provides actionable guidance, whereas destructive criticism targets personal faults and often provokes anger and avoidance behaviors [6, 33]. There distinctions align closely with the contrast between TNF and PFN. According to FIT, PNF threatens self-related goals such as self-esteem, thereby redirecting attention toward the self rather than the task [44]. This shift is likely to evoke stronger negative affect and attenuate positive affect, particularly among individuals with low self-esteem, who have been shown to exhibit greater affective reactivity to negative feedback [7, 34, 57]. In contrast, TNF highlights performance strategies or effort, which is less likely to induce self-threat and therefore associated with weaker affective disruption. Thus, we hypothesize: Hypothesis 2: TNF leads to a smaller decline in positive affect and a lesser increase in negative affect than does PNF. Hypothesis 2: TNF leads to a smaller decline in positive affect and a lesser increase in negative affect than does PNF. Positive affect reliably enhance creative performance because it broadens attentional scope [26] and supports associative and flexible thinking [28]. Consistent with this account, individuals experiencing higher positive affect tend to outperform peers on divergent thinking tasks [9, 23]. By contrast, the effect of negative affect on creativity appears mixed across the literature. While some theories posit that negative affect narrows attention and impedes creativity [27], others, such as Baas et al. [3], indicate that short-lived anger can temporarily increase creativity. A theoretically coherent explanation is that boundary conditions determine the direction of the effect. First, when negative feedback is framed at the person level (PNF), it is more likely to elicit threat appraisals and self-focus, which in turn activate avoidance/prevention orientations detrimental to idea generation; recent evidence shows that threat vs. challenge appraisals systematically shape downstream prevention focus and creativity under negative supervisory feedback [46]. Second, negative affect often induces attentional narrowing and executive-resource costs (e.g., via rumination), which are especially harmful for tasks that require broad search and flexible switching,state rumination has been shown to impair concurrent executive functions, consistent with a resource-depletion account [62], and diary studies in work settings commonly observe negative affect as an antagonist of creativity [70]. Relatedly, neurocognitive models emphasize that creative performance depends on attentional control dynamics that enable strategic shifting between exploration and evaluation [78],affective states that constrict attentional breadth or consume control resources should therefore impede divergent thinking fluency and flexibility. Taken together, in contexts in this study—divergent thinking tasks following evaluative feedback—negative affect is expected, on average, to reduce creative output, even if specific high-arousal states (e.g., short-lived anger) may benefit other task types or phases. Integrating these insights, we propose: Hypothesis 3: Changes in positive and negative affect mediate the effects of NFH on creativity. Hypothesis 3: Changes in positive and negative affect mediate the effects of NFH on creativity. Methods Participants A priori power analysis via G*Power 3.1 indicated a minimum sample size of 90 for the experiment’s repeated measures mixed-design ANOVA (2 measurements, 3 groups, medium effect size of 0.25, alpha = 0.05). A total of 95 undergraduate students from a university in East China were initially recruited through online advertisements from a participant pool. Participants were required to have no prior participation in creativity studies. After excluding five participants for noncompliance (two guessed the study’s purpose, two attempted to use a search engine during the posttest, and one exhibited an extreme response), the final sample consisted of 90 participants (45 females, 45 males; Mage = 21.55, SDage = 1.78). All participants provided informed consent, were assured of confidentiality, and received 50 yuan (~ $7) as compensation. Design A 3 × 2 repeated-measures ANOVA was employed, where feedback type (personal negative, task negative, no feedback) served as the between-subjects factor and task time (pre-feedback, post-feedback) served as the within-subjects factor. The participants were randomly assigned to feedback conditions. Materials Affect was measured via the Positive and Negative Affects Scale (PANAS; [75], which includes two subscales: positive affect (PA) and negative affect (NA), each with 10 adjectives (e.g., interested, nervous). The participants rated their feelings on a 5-point Likert scale ranging from 1 (not at all) to 5 (very much). The internal consistency coefficients were as follows: PAPre α = 0.85, NAPre α = 0.88, PAPost α = 0.86, and NAPost α = 0.89. Creativity can be measured in various ways, with one of the most widely used approaches being divergent thinking tests [5], which primarily assess an individual’s creative potential through the creative thinking process, thereby predicting future creative performance [65]. In this study, we employ two commonly used divergent thinking tests to measure creativity: the Alternative Uses Task (AUT) and the Realistic Presented Problems (RPP) task. The AUT requires participants to generate as many novel uses as possible for a common object, such as an umbrella (e.g., “spinning an umbrella and flinging paint from it to create large-scale art”). Although the AUT has been demonstrated to be an effective measure of creative potential, it has been critiqued for being decontextualized [59]. Thus, we also used a more realistic measurement, the Realistic Presented Problem (RPP [67],), which asks participants to propose solutions to real-life scenarios, such as, “What should you do if a classmate interrupts you during class?” In our study, AUTs were limited to 5 min, whereas RPPs had no time limit. Objective ratings were employed to assess fluency, originality, and flexibility in both tasks [64, 66]. Fluency refers to the number of responses provided by participants. Each response was awarded one point, while no points were given for answers stating the object’s primary function (e.g., “to keep the rain off” for “umbrella”). Originality measures the rarity of a given response. For each question, all responses were pooled, and similar expressions were grouped together. Each answer was then scored according to its proportion in the pool: answers representing ≤ 1% of all responses received 2 points, answers between 1 and 5% received 1 point, and answers representing more than 5% received 0 points. A participant’s originality score for each question was the sum of their novelty scores for all responses. Flexibility refers to the number of distinct categories into which a participant’s answers can be classified. To calculate this, all answers in the pool were categorized, and the flexibility score for each participant was based on the number of categories into which their responses were grouped. Given the subjectivity involved in categorization, two randomly selected tasks were independently rated by two raters using the snapshot method as outlined by Forthmann et al. [24]. The inter-rater reliability for categorizing flexibility was found to be strong, with an intraclass correlation coefficient of 0.76 for the umbrella task and 0.89 for the newspaper task. The categorization of other tasks was conducted by the first author. Although the experiment counterbalanced the pre- and post-tests, several participants reported differences in difficulty between the tasks in post-experiment interviews. So all metrics were standardized for data analysis, yielding pre-test and post-test scores for creative performance. The negative feedback materials were adapted from the feedback material of Zhou [81]. It was tailored to personal and task hierarchies, ensuring clarity in focus (see Additional File—A). Manipulation Check Participants responded to four items measuring the task process [42] on a 7-point Likert scale from 1 (not at all) to 7 (very much), e.g., “the review helped me to focus my attention on completing the task”, achieving a Cronbach’s alpha of 0.73. Procedure The participants began by providing demographic information (age, sex) and their email addresses. They then completed a pretest assessing affect and creative performance, which included two AUTs and one RPP task. Afterward, the participants were informed of a 5-min break; those in the experimental groups were told that an anonymous reviewer would provide feedback on their pretest responses via email during this time (Fig. 2). Illustration of the experimental procedure for PNF and TNF groups in experiment 1 At the end of the break, the experimental groups received feedback from the “reviewer” via email. After reading this feedback, they completed manipulation test questions, whereas the control group had a break without feedback or additional tasks. All the groups subsequently completed posttests measuring positive/negative affect and creativity and then submitted questionnaires. The task order was counterbalanced across all groups. After completing the tasks, the participants in the experimental groups were debriefed and informed that the “feedback” was randomly generated. They had a brief interview about their perceptions of the feedback and any changes they noticed in their pre- and post-test performance. Finally, the participants were asked to keep the study’s details confidential to avoid influencing others who had not participated. Results Manipulation check An independent samples t-test revealed a significant difference between the two experimental groups on the manipulation check questions. The task NF group scored higher (M = 19.90, SD = 3.88) than the personal NF group (M = 17.16, SD = 5.62), t(60) = 2.22, p = 0.03, d = 0.57, confirming the validity of the manipulation. Additionally, the three groups did not significantly differ in age (F(2, 87) = 1.60, p = 0.209) or sex (F(2, 87) = 0.04, p = 0.962), so these factors were not included as covariates in the analysis. Descriptive statistics Table 1 presents descriptive statistics for positive and negative affect (raw scores) and AUT and RPP fluency, originality, and flexibility (standardized scores) for the three groups in both pre- and post-tests. Table 1Descriptive statistics (M ± SD) of experiment 1Personal negative feedback (n = 32)Task negative feedback (n = 30)Control group (n = 28)Pre-testPost-testPre-testPost-testPre-testPost-testPositive affect34.22 ± 6.0823.84 ± 3.8932.37 ± 5.7625.43 ± 4.2434.36 ± 4.7925.61 ± 4.80Negative affect20.69 ± 6.0627.28 ± 6.1119.33 ± 6.7127.23 ± 9.0418.61 ± 3.9728.21 ± 5.63AUT-Fluency0.01 ± 2.00−0.19 ± 2.00−0.39 ± 1.380.50 ± 2.08−0.23 ± 1.46−0.30 ± 1.29AUT-Originality0.04 ± 2.080.02 ± 2.19−0.46 ± 1.460.45 ± 2.21−0.25 ± 1.37−0.18 ± 1.33AUT-Flexibility0.01 ± 1.54−0.22 ± 1.67−0.03 ± 1.500.05 ± 1.560.01 ± 1.620.15 ± 1.66RPP-Fluency0.30 ± 0.93−0.20 ± 0.960.18 ± 1.12−0.09 ± 1.120.11 ± 0.97−0.24 ± 0.83RPP-Originality0.29 ± 1.09−0.22 ± 0.930.15 ± 1.150.06 ± 1.110.09 ± 0.86−0.28 ± 0.77RPP-Flexibility0.35 ± 0.80−0.28 ± 0.920.26 ± 0.99−0.17 ± 1.150.15 ± 1.07−0.27 ± 0.87AUT Alternative Uses Task, RPP Realistic Presented Problem Descriptive statistics (M ± SD) of experiment 1 AUT Alternative Uses Task, RPP Realistic Presented Problem Effect of NFH on affect A repeated measures ANOVA was conducted with NFH as the between-groups variable and pre- and post-affect as the repeated measures (Table 2). For positive affect (PA), there was a significant interaction effect between time and NFH (F(2, 87) = 3.31, p = 0.041, = 0.07), indicating a significant difference in PA change among the three groups (Fig. 3-A). To explore these differences, we calculated PA change by subtracting pretest scores from posttest scores and performed a one-way ANOVA on the change amounts, yielding the same results. Subsequently, post hoc comparisons were conducted using Scheffé’s test, which was selected for its robustness in controlling Type I error across multiple comparisons and its suitability for slightly unequal group sizes. The results (Fig. 3-B) showed that PA decreased significantly in the PNF group compared with the TNF group (p = 0.041, 95% CI [−6.78, −0.11]), but no significant differences were found between the two experimental groups and the control group (ps > 0.05). Additionally, there was no significant interaction effect between time and NFH for negative affect (NA) (F(2, 87) = 1.46, p = 0.238). Table 2The results of all repeated measures mixed-deign ANOVAs in experiment 1FdfpPositive Affect Time243.94***1< 0.0010.74 NFH0.5620.5740.01 Time × NFH3.31*2, 870.0410.07Negative Affect Time124.62***1< 0.0010.59 NFH0.1420.8660.00 Time × NFH1.462, 870.2380.03AUT-Fluency Time2.1310.1480.02 NFH0.2920.7490.01 Time × NFH6.11**2, 870.0030.12AUT-Originality Time4.08*10.0470.05 NFH0.1920.8310.00 Time × NFH3.64*2, 870.0300.08AUT-Flexibility Time0.0010.9910.00 NFH0.1220.8920.00 Time × NFH0.912, 870.4070.02RPP-Fluency Time16.37***1< 0.0010.16 NFH0.5120.8530.00 Time × NFH0.512, 870.6010.01RPP-Originality Time11.14***10.0010.11 NFH0.3820.6850.01 Time × NFH1.642, 870.1990.04RPP-Flexibility Time19.29***1< 0.0010.18 NFH0.1420.8660.00 Time × NFH0.372, 870.6890.01NFH Negative Feedback Hierarchy, AUT Alternative Uses Task, RPP Realistic Presented Problem. ***p<0.001, **p<0.01, *p<0.05 The results of all repeated measures mixed-deign ANOVAs in experiment 1 NFH Negative Feedback Hierarchy, AUT Alternative Uses Task, RPP Realistic Presented Problem. ***p<0.001, **p<0.01, *p<0.05 Positive affect changes in the three groups in experiment 1. Note. ***p < 0.001, **p < 0.01, *p < 0.05, †p < 0.1 Effect of NFH on creativity A repeated measures ANOVA was conducted with NFH as the between-groups variable and pre- and post-AUT/RPP performance as the repeated measures (Table 2). The results revealed significant interactions between time and NFH for AUT fluency (F(2, 87) = 6.11, p = 0.003, = 0.12; Fig. 4-A) and AUT originality (F(2, 87) = 3.64, p = 0.030, = 0.08; Fig. 5-A). Scheffé’s post hoc tests indicated that the TNF group showed significantly greater gains in AUT fluency than both the PNF group (p = 0.007, 95% CI [0.25, 1.93]) and the control group (p = 0.026, 95% CI [0.09, 1.82]), with no significant difference between the PNF and control groups (p = 0.930; Fig. 4-B). For AUT originality, the gain in the TNF group was greater than that in the PNF group, although this difference did not reach conventional significance (p = 0.052, 95% CI [−0.01, 1.86). No significant differences were found between the control group and either the TNF group (p = 0.102) or the PNF group (p = 0.972; Fig. 5-B). AUT fluency changes in the three groups in experiment 1 AUT originality changes in the three groups in experiment 1 No significant interaction effect was found for AUT flexibility (F(2, 87) = 0.91, p = 0.407) or RPP indicators (RPP fluency: F(2, 87) = 0.51, p = 0.601; RPP originality: F(2, 87) = 1.62, p = 0.202; RPP flexibility: F(2, 87) = 0.37, p = 0.689). The mediating role of affect in the effect of NFH on creativity Because group differences in negative affect did not reach significance, NA was excluded from the mediation analysis. The mediation analysis was therefore conducted using the bootstrap method with 5,000 samples to test whether different NFHs influenced creative performance through changes in positive affect. Only the PNF and TNF groups were included, with group classification represented by a dummy variable (PNF = 0, TNF = 1). The mediator was the change in PA, and the dependent variables were changes in AUT fluency and originality. Results are presented in Table 3. Table 3Bootstrap results of the mediating role of positive affect changeEffectsβBoot SE95%CI LL95%CI ULRatio of direct to total effectAUT fluency gain Total effect1.000.340.311.69 Direct effect0.800.350.091.50 Indirect effect0.210.18−0.030.4220.51%AUT originality gain Total effect1.110.350.401.82 Direct effect0.740.330.081.41 Indirect effect0.370.270.011.0033.20%AUT Alternative Uses Task Bootstrap results of the mediating role of positive affect change AUT Alternative Uses Task NFH significantly and positively predicts changes in PA. When both the NFH and PA changes are included in the regression equation, they significantly predict the gain of AUT fluency and originality. These findings suggest that PA change partially mediates the effect of NFH on the gain of AUT fluency and originality. The mediating effects account for 20.51% and 33.20% of the total effect, respectively. The path modeling of these effects is illustrated in Fig. 6. PA change mediates the effects of NFH on AUT fluency and originality gain Discussion Hypothesis 1 was partially supported: PNF did not significantly differ from no-feedback on any creativity indicator, whereas TNF produced a significant increase in AUT fluency compared with both PNF and the control group. For AUT originality, the TNF group scored higher than the PNF group, although this difference did not reach statistical significance, and neither experimental group differed from the control group. No other creativity indicators showed significant differences. Hypothesis 2 was also partially supported. Differences between TNF and PNF were reflected primarily in changes in positive affect, with TNF associated with a smaller decline in PA, rather thn with difference in NA. Hypothesis 3 received partial support as well, as changes in PA partially mediated the effect of NFH on AUT fluency and originality. Taken together, these findings are consistent with the assumptions of FIT [44], which posits that negative feedback motivates creators to recognize and address the gap between their current performance and desired goals [48]. Importantly, different hierarchies of negative feedback exert distinct effects on both creativity and affective responses. Consistent with Baron’s [6] research, PNF was associated with a stronger decline in PA, which in turn partially mediated the effects of NFH on creativity. This finding resonates with recent evidence from de Chantal and Organisciak [17], who showed in three studies on divergent thinking that higher levels of positive emotion were associated with greater originality when receiving feedback, underscoring the importance of positive affect as a pathway through which feedback influences creative performance. Post experimental interviews provided convergent qualitative evidence. Participants who received negative feedback frequently reported increased conscientiousness, greater effort, and a heightened desire to improve their performance. Many reflected on whether their responses satisfied the “novelty” criterion, indicating that they sought to close the performance gap by enhancing effort, increasing motivation, or modifying strategies to foster creativity. These self-reports echo Hattie and Timperley’s [33] account of feedback as a mechanism for directing attention and motivating adaptive change. Experiment 2 Hypotheses Experiment 1 confirmed differences between PNF and TNF in their effects on creativity, with TNF enhancing AUT fluency relative to PNF, and changes in PA partially mediating this effect. However, no significant differences were observed between either feedback condition and the control group. Building on these findings, Experiment 2 shifts the focus from affective processes to the cognitive and motivational pathways emphasized in FIT. Since Experiment 1 revealed no significant deviations from the control condition, the control group was removed in Experiment 2 to allow closer examination of the contrasts between PNF and TNF. According to FIT, the effects of negative feedback depend on the hierarchical level to which attention is directed [44]. TNF directs attention to task-level goals and highlights specific areas for improvement, which can increase engagement and strengthen motivation. In contrast, PNF shifts attention to higher-level self-goals, and often elicits defensive responses that undermine motivation. These defensive reactions deplete cognitive resources, limiting the individuals’ capacity to reflect on tasks or modify strategies, and ultimately reducing creative performance. Empirical evidence supports these distinctions. For example, Kamins and Dweck [41] found that children receiving task-related criticism demonstrated greater persistence than those who received person-related criticism. Similarly, autonomy-supportive feedback, which parallels TNF, has been shown to promote motivation and well-being more effectively than controlling feedback [10]. Burgers et al. [8] also reported that TNF encouraged sustained engagement, whereas PNF reduced it. Together, these findings suggest that TNF is more effective than PNF in sustaining motivation, consistent with FIT’s proposition that attention directed to task goals facilitates adaptive regulation. Hypothesis 4: TNF enhances motivation more than PNF does. Hypothesis 4: TNF enhances motivation more than PNF does. According to FIT, motivational differences are not only outcomes of feedback but also mechanisms that help explain its impact on creativity. Motivation determines how much effort individuals devote to a task, how persistently they pursue goals, and how flexibly they explore alternative approaches—all processes that are central to creative performance [68]. Recent evidence supports this view: Wang et al. [74] showed that individuals with lower avoidance-motivation intensity performed better on divergent thinking tasks than those with higher avoidance-motivation intensity, with neurocognitive data indicating that elevated avoidance motivation depleted executive resources necessary for idea generation. These findings suggest that defensive motivational states constrain creativity by consuming cognitive resources, whereas adaptive, task-focused motivation facilitates divergent thinking. Within the FIT framework, PNF is likely to undermine autonomy and elicit avoidance-oriented motivation, thereby weakening engagement, whereas TNF directs attention to task-level goals and fosters sustained, task-focused motivation. Thus, we propose: Hypothesis 5: Motivation mediates the effects of NFH on creativity. Hypothesis 5: Motivation mediates the effects of NFH on creativity. Individuals often modify their strategies in response to feedback. Negative feedback has been shown to stimulate reflection and strategy modification, prompting individuals to refine their approaches to meet performance standards [4, 52]. Strategy modification is particularly important for creativity, as it fosters cognitive flexibility and supports exploration during the creative process [38]. According to FIT, TNF directs attention to task-level discrepancies and supports adaptive regulation, whereas PNF diverts resources toward self-protection, thereby hindering effective strategy modification [44]. Hypothesis 6: TNF enhances strategy modification more than PNF does. Hypothesis 6: TNF enhances strategy modification more than PNF does. Strategy modification plays a critical role in the creative process because it fosters cognitive flexibility and supports exploratory thinking [38]. Individuals who are able to adapt their strategies not only demonstrate greater openness to novelty but also report lower stress in demanding performance contexts [37]. Importantly, experimental evidence indicates that explicit induction of strategies can directly enhance divergent thinking performance. Forthmann et al. [25] found that providing participants with strategy instructions, combined with a creativity prompt, significantly improved performance in divergent thinking tasks, underscoring the functional role of strategy adjustment in fostering originality and fluency. Within the framework of FIT, TNF is expected to facilitate such adaptive strategy modification, whereas PNF is more likely to undermine it by redirecting attention toward self-protective concerns. Building on Hypothesis 6, we further propose that: Hypothesis 7: Strategy modification mediates the effects of NFH on creativity. Hypothesis 7: Strategy modification mediates the effects of NFH on creativity. Method Participants Seventy-two undergraduate students from the same university were recruited using the same criteria as in Experiment 1. Six participants were excluded for noncompliance (four guessed the study’s purpose, and two provided invalid responses), leaving a final sample of 66 participants (55 females, 11 males; Mage = 20.90, SDage = 2.37). Informed consent was obtained, confidentiality was assured, and each participant was compensated with 30 yuan (~ $4.20). Design A 2 × 2 repeated measures mixed-design ANOVA was employed. NFH served as the between-subjects factor (PNF vs. TNF), whereas task time was the within-subjects factor (before feedback vs. after feedback). The participants were randomly assigned to one of the two feedback conditions. Materials Strategy modification was assessed via two questions adapted from Diefendorff and Lord [20]: “I tried to find ways to perform better on the second task” and “I tried to improve my performance by using/modifying strategies.” Participants rated their agreement on a 7-point scale (1 = not at all,7 = completely). If participants scored 4 or higher on the second question, they were prompted to specify the strategies they employed. Motivation was measured via both self-reported and objective methods. The self-reported motivation scale consists of five questions adapted from Burgers et al. [8], for example, “I put more effort into the second task than the first”. The participants rated these items on a 7-point scale, yielding a Cronbach’s alpha of 0.71. Additionally, participants’ time spent on each task served as an objective measure of motivation. The creativity tasks were the same as those in Experiment 1. To minimize performance degradation due to extended task durations, the maximum time allowed for each AUT was reduced to 3 min, and the maximum for each RPP was set at 5 min. Manipulation Check: The manipulation check used the same items as in Experiment 1, with a Cronbach’s alpha of 0.83. Procedure Experiment 2’s procedure mirrored that of Experiment 1, except for the deletion of the affect measure and the addition of a self-reported motivation scale and a strategy modification scale after the post-test. Results Manipulation check An independent samples t-test revealed significant differences between the two groups on the manipulation test questions, with the task process scores significantly higher in the TNF group (M = 15.28, SD = 2.64) than in the PNF group (M = 13.12, SD = 3.26), t(64) = 2.95, p = 0.004, d = 0.73. This indicates that the experimental manipulation was valid. Additionally, there were no significant differences between the groups in terms of age (t(64) = 1.33, p = 0.188) or sex (t(64) = 1.56, p = 0.123), so these variables were not included as covariates in the analysis. Descriptive statistics Descriptive statistics for self-reported motivation, strategy modification, time on task (in seconds) during the pre- and posttests, and creative performance (standardized scores) for the two groups are presented in Table 4. Due to a malfunction of the screenshot software, time on task data was missing for five participants. For AUT and RPP time, the TNF group had 28 participants, whereas the PNF group had 33. Table 4Descriptive statistics of experiment 2Task Negative Feedback(n = 32)Personal Negative Feedback(n = 34)Pre-testPost-testPre-testPost-testSelf-reported motivation-25.56 ± 3.07-24.03 ± 3.97Strategy modification-10.53 ± 1.52-0.09 ± 1.87AUT time297.39 ± 75.53324.36 ± 57.10314.30 ± 74.96325.91 ± 50.18RPP time244.57 ± 61.13241.61 ± 66.88255.09 ± 65.00243.85 ± 67.37AUT-fluency−0.20 ± 1.950.08 ± 1.640.17 ± 1.85−0.19 ± 1.96AUT-originality−0.46 ± 2.000.11 ± 1.580.28 ± 1.820.02 ± 1.94AUT-flexibility−0.22 ± 1.78−0.08 ± 1.610.11 ± 1.57−0.03 ± 1.90RPP-fluency−0.11 ± 0.770.06 ± 0.830.08 ± 1.28−0.06 ± 1.05RPP-originality−0.23 ± 0.870.06 ± 0.740.03 ± 1.300.12 ± 1.05RPP-flexibility−0.04 ± 0.83−0.01 ± 0.930.09 ± 1.14−0.04 ± 1.15AUT Alternative Uses Task, RPP Realistic Presented Problem Descriptive statistics of experiment 2 AUT Alternative Uses Task, RPP Realistic Presented Problem Effects of NFH on motivation and strategy modification Independent samples t-tests of self-reported motivation and strategy modification scores revealed no significant difference in self-reported motivation between the two groups, t(64) = 1.32, p = 0.190. However, there was a significant difference in strategy modification, with scores significantly higher in the task NF group than in the personal NF group, t(64) = 3.67, p = 0.001, d = 0.90. A repeated measures ANOVA was conducted with NFH as the between-group variable and time spent on the task (before and after) as the repeated variable. The results are presented in Table 5. No significant interaction effect was found for AUT time (F(1, 59) = 1.30, p = 0.259) or RPP time (F(1, 59) = 0.36, p = 0.553). Table 5Results of all repeated measures mixed-design ANOVAs in experiment 2FdfpAUT time Time8.18**10.0060.12 NFH0.3610.5500.01 Time × NFH1.301, 590.2590.02RPP time Time1.0510.3110.02 NFH0.1810.6770.00 Time × NFH0.361, 590.5530.01AUT-Fluency Time0.0610.8030.00 NFH0.0110.9160.00 Time × NFH4.45*1, 640.0390.07AUT-Originality Time0.9210.3410.01 NFH2.3810.1280.04 Time × NFH6.51*1, 640.0130.09AUT-Flexibility Time0.0010.9740.00 NFH0.2410.6260.00 Time × NFH0.631, 640.4300.01RPP-Fluency Time0.0310.8690.00 NFH0.0310.8630.00 Time × NFH1.941, 640.1690.03RPP-Originality Time1.6410.2040.03 NFH0.6410.4280.01 Time × NFH0.411, 640.5250.01RPP-Flexibility Time0.1410.7050.00 NFH0.0510.8210.00 Time × NFH0.461, 640.4990.01NFH Negative Feedback Hierarchy, AUT Alternative Uses Task, RPP Realistic Presented Problem. ***p<0.001, **p<0.01, *p<0.05 Results of all repeated measures mixed-design ANOVAs in experiment 2 NFH Negative Feedback Hierarchy, AUT Alternative Uses Task, RPP Realistic Presented Problem. ***p<0.001, **p<0.01, *p<0.05 Effect of NFH on creativity A repeated measures ANOVA was conducted with NFH as the between-groups variable and pre- and post- creative performance as the repeated variable. The results, shown in Table 5, indicated a significant interaction effect between time and the NFH for both AUT fluency (F(1, 64) = 4.45, p = 0.039, = 0.07; Fig. 7-A) and AUT originality (F(1, 64) = 6.51, p = 0.013, = 0.09; Fig. 7-B). AUT fluency and originality changes in the two groups in experiment 2 No significant interaction effect was found for AUT flexibility (F(1, 64) = 0.63, p = 0.430) or RPP indicators: RPP fluency (F(1, 64) = 1.94, p = 0.169), RPP originality (F(1, 64) = 0.41, p = 0.525), and RPP flexibility (F(1, 64) = 0.46, p = 0.499). The mediating role of strategy modification in the effect of the NFH on creativity The mediating role of strategy modification in the gains in AUT fluency and originality due to NFH was examined. The independent variable was NFH (personal NF group = 0, task NF group = 1), whereas the dependent variables were gains in AUT fluency and originality. The bootstrap method with 5,000 samples was used to test for mediating effects, and the results are presented in Table 6. Table 6Bootstrap results of the mediating role of strategy modificationEffectsβBoot SE95%CI LL95%CI ULRatio of direct to total effectAUT fluency gain Total effect0.550.260.031.07 Direct effect0.300.28−0.250.86 Indirect effect0.250.120.030.5044.61%AUT originality gain Total effect0.620.290.051.19 Direct effect0.310.31−0.300.93 Indirect effect0.310.140.010.5649.33%AUT Alternative Uses Task Bootstrap results of the mediating role of strategy modification AUT Alternative Uses Task NFH significantly and positively predicts strategy modification. The 95% CIs for the mediating effect did not contain 0 (AUT fluency gain: 95% CI [0.03, 0.50]; AUT originality gain: 95% CI [0.01, 0.56]), indicating a significant mediating effect. The mediating effects accounted for 44.61% and 49.33% of the total effects, respectively. When both NFH and strategy modification were included in the regression equation, only strategy modification significantly predicted AUT fluency and originality gain, whereas the direct effect of NFH was not significant. This demonstrates that strategy modification fully mediates the effect of NFH on AUT fluency and originality gain (Fig. 8). The mediation role of strategy modification in the effects of negative feedback hierarchy (PNF vs. TNF) on AUT fluency and originality gain Discussion The results from Experiment 2 robustly supported Hypothesis H1, demonstrating that compared with PNF, TNF led to significantly greater AUT fluency and originality. The shorter task du ration likely mitigated the substantial decrease in RPP performance observed in the PNF group during Experiment 1, suggesting that task characteristics influence the effects of negative feedback on creativity. Hypotheses H4 and H5 were not supported, as NFH had no significant effect on the motivational indicators of self-reported motivation and task duration. Both groups reported high motivation levels (TNF: 25.56 ± 3.07; PNF: 24.03 ± 3.97). However, only the TNF group spent significantly more time on the AUT posttest than on the pretest, which aligns with findings from Kamins and Dweck [41] regarding increased perseverance following negative feedback. The lack of statistical significance may be attributed to the imposed time limits, which some participants felt were insufficient for optimal performance. Support for Hypotheses H6 and H7 was evident, as strategy modification was significantly greater in the TNF group than in the PNF group. This pattern is consistent with FIT, which posits that TNF directs attention toward performance improvement, whereas PNF elicits defensive self-focus and reduces cognitive resources for adaptive regulation. Empirical findings echo this interpretation: Ball et al. [4] and Mesch et al. [52] demonstrated that negative feedback encourages strategic modification when it highlights task-relevant shortcomings. Extending these insights, our results suggest that TNF facilitates strategy modification, while PNF hinders such modification by triggering self-protective responses. Overall, the efficacy of negative feedback in enhancing creativity appears contingent upon its ability to facilitate cognitive resource management and reflective strategy regulation, in line with FIT. Strategies developed through this regulation—such as analyzing potential uses based on properties or reconsidering thought directions—were crucial for improving creative performance. TNFs fostered greater persistence and broader perspectives in problem solving, enhancing fluency and originality. Experiment 3 Hypotheses Experiments 1 and 2 demonstrated that PNF led to a stronger decline in positive affect and reduced strategy modification, which in turn impaired creative performance. These findings highlight two key pathways through which PNF exerts its adverse effects: an affective pathway, in which diminished positive affect constrains attentional breadth and associative thinking, and a cognitive pathway, in which reduced strategy modification limits adaptive reflection and problem solving. Building on these insights, Experiment 3 examined whether interventions targeting affect or cognition could buffer the negative impact of PNF. Cognitive reappraisal (CR) was chosen as the affective intervention because it is one of the most effective and well-studied strategies for regulating emotional responses to negative events [32, 40]. By reframing the meaning of negative feedback, CR is expected to mitigate defensive reactions and sustain positive affect, thereby preserving the cognitive resources necessary for creativity. Strategy modification (SM) was included as the cognitive intervention because prior research shows that prompting individuals to reconsider and refine their task strategies enhances performance in challenging contexts [76]. Results from Experiments 1 and 2 further indicated that positive affect and strategy modification mediated the effects of NFH on creativity, underscoring their central role in creative processes. Creativity itself requires the interplay of cognitive flexibility and persistence [18]. Although affect has been shown to influence creativity [2], these effects typically unfold through their impact on cognitive processes, as posited by FIT. Thus, we hypothesized: Hypothesis 8: The effectiveness of the strategy modification intervention will exceed that of the cognitive reappraisal group. Additionally, the dual intervention group will be more effective than either intervention alone. Hypothesis 8: The effectiveness of the strategy modification intervention will exceed that of the cognitive reappraisal group. Additionally, the dual intervention group will be more effective than either intervention alone. Methods Participants A total of 108 undergraduate students were recruited through the same participant pool and inclusion criteria used in Experiments 1 and 2. Eleven participants were excluded for noncompliance (seven guessed the study’s purpose, two provided invalid responses, and two exhibited extreme responses), resulting in a final sample of 97 participants (76 females, 21 males; Mage = 21.91, SDage = 2.29). All participants provided informed consent, were assured of confidentiality, and received 30 yuan (~ $4.20) as compensation. Design A 3 × 2 repeated measures ANOVA was conducted, with intervention as the between-subjects factor (cognitive reappraisal, CR; strategy modification, SM; dual intervention, DI) and task time as the within-subjects factor (pre-feedback, post-feedback). The participants were randomly assigned to three intervention conditions after receiving personal negative feedback. The personal negative feedback material used in this experiment was identical to that in Experiments 1 and 2. Materials The intervention materials were adapted from the strategy reported by the participants in Experiment 2 (see Additional File—B). The creativity tasks were the same as those in Experiments 1 and 2. Procedure Experiment 3 replicated the procedure of Experiment 2 but removed the motivation and strategy modification scales. An intervention phase was added before the post-test of creativity. Results Descriptive statistics Descriptive statistics for pre- and post-creative performance (standardized scores) across the three groups were analyzed, and the results are presented in Table 7. Table 7Descriptive statistics of experiment 3Cognitive reappraisal(n = 33)Strategy modification(n = 32)Dual Intervention(n = 32)Pre-testPost-testPre-testPost-testPre-testPost-testAUT-fluency−0.20 ± 1.56−0.75 ± 1.28−0.02 ± 1.77−0.01 ± 1.70−0.08 ± 1.780.16 ± 2.09AUT-originality−0.56 ± 1.54−0.81 ± 1.33−0.07 ± 1.800.08 ± 1.600.03 ± 1.730.44 ± 1.97AUT-flexibility0.40 ± 1.86−0.22 ± 1.31−0.06 ± 1.78−0.33 ± 1.570.32 ± 1.78−0.50 ± 1.79RPP-fluency0.12 ± 0.76−0.01 ± 1.00−0.29 ± 0.810.12 ± 1.09−0.32 ± 1.030.26 ± 1.25RPP-originality0.03 ± 0.88−0.03 ± 0.92−0.34 ± 0.820.08 ± 0.92−0.30 ± 1.010.30 ± 1.31RPP-flexibility0.10 ± 0.80−0.07 ± 0.74−0.13 ± 0.960.07 ± 1.10−0.32 ± 1.040.28 ± 1.29AUT Alternative Uses Task, RPP Realistic Presented Problem Descriptive statistics of experiment 3 AUT Alternative Uses Task, RPP Realistic Presented Problem Effects of the intervention on creative performance A repeated measures ANOVA was conducted with intervention type as the between-group variable and pre- and post- creative performance as the repeated variable. The results are shown in Table 8. Table 8Results of all repeated measures mixed-design ANOVAs in experiment 3FdfpAUT-Fluency Time0.7910.3770.01 Intervention type1.0120.3680.02 Time × Intervention type4.27*2, 940.0170.08AUT-Originality Time0.6710.4150.01 Intervention type3.07†20.0510.06 Time × Intervention type2.50†2, 940.0880.05AUT-Flexibility Time12.93***10.0010.12 Intervention type0.3220.7290.01 Time × Intervention type1.012, 940.3690.02RPP-Fluency Time10.34**10.0020.10 Intervention type0.2020.8220.00 Time × Intervention type5.79**2, 940.0040.11RPP-Originality Time8.39**10.0050.08 Intervention type0.2620.7720.01 Time × Intervention type3.11*2, 940.0490.06RPP-Flexibility Time4.13*10.0450.04 Intervention type0.0320.9700.00 Time × Intervention type4.52*2, 940.0130.08AUT Alternative Uses Task, RPP Realistic Presented Problem.***p<0.001, **p<0.01, *p<0.05, †p<0.1. Results of all repeated measures mixed-design ANOVAs in experiment 3 AUT Alternative Uses Task, RPP Realistic Presented Problem.***p<0.001, **p<0.01, *p<0.05, †p<0.1. The analysis indicated a significant interaction effect between time and intervention type for AUT fluency (F(2, 94) = 4.27, p = 0.017, = 0.08), RPP fluency (F(2, 94) = 5.79, p = 0.004, = 0.11), RPP originality (F(2, 94) = 3.11, p = 0.049, = 0.06), and RPP flexibility (F(2, 94) = 4.53, p = 0.013, = 0.09). However, no significant interaction effect was found for AUT flexibility (F(2, 94) = 1.01, p = 0.369). Tamhane’s T2 post hoc test results indicated that AUT fluency gains in the CR group were lower than those in the SM group, although this difference did not reach significance (p = 0.082, 95% CI [−1.16, 0.05]), and were significantly lower than those in the DI group (p = 0.020, 95% CI [−1.50, −0.10]). No significant difference was found between the SM and DI groups (p = 0.832). AUT originality gain in the CR group was also lower than that in the DI group, although this difference did not reach statistical significance (p = 0.092, 95% CI [−1.40, 0.08),and npo significant differences were observed between the SM group and either the CR or DI groups (ps > 0.1; Fig. 9). AUT changes in the three groups in experiment 3 Tamhane’s T2 post hoc test results revealed that RPP fluency gains in the CR group were significantly lower than in the SM group (p = 0.049, 95% CI [−1.08, 0.00]) and the DI group (p = 0.007, 95% CI [−1.24, −0.16]). No significant difference was observed between the SM and DI groups (p = 0.760). RPP originality gain in the CR group was also lower than that in the DI group, although this difference did not reach significance (p = 0.062, 95% CI [−1.16, 0.19]), and no significant differences were observed between the SM group and either the CR or DI group (ps > 0.1). RPP flexibility gain in the CR group was significantly lower than that in the DI group (p = 0.020, 95% CI [−1.45, −0.10]), and no significant differences were observed between the SM group and either the CR or DI groups (ps > 0.1; Fig. 10). RPP changes in the three groups in experiment 3 Discussion In total, H8 was supported. The SM group demonstrated significantly greater improvements in RPP fluency than did the CR group, and AUT fluency also tended to be higher in the SM group, although this difference did not reach statistical significance. This reinforces the findings from Experiments 1 and 2, emphasizing the critical role of strategy modification in mediating the effects of NFH on creativity. This outcome aligns with earlier research on formative assessment [69], which underscores the importance of strategic interventions in enhancing creative performance. Pairwise comparisons revealed significant decreases in AUT and RPP fluency, as well as RPP flexibility, in the CR group compared with the DI group. AUT fluency was also lower in the CR group than in the SM group, although this difference was not statistically significant. Similarly, AUT originality and RPP originality in the CR group were lower than in the DI group, but these differences did not reach significance. These results suggest that the CA intervention did not effectively assist participants in managing the impacts of negative feedback, which contrasts with the findings of previous studies [54, 72]. The high cognitive demands of creativity tasks [61] may explain this limitation, while CA might help moderate emotional responses, it does not translate into the improved cognitive functioning necessary for creative tasks. In contrast, the SM group achieved significantly greater RPP fluency than the CR group, but no significant changes were observed in originality or flexibility. This indicates that strategic modification can mitigate some of the adverse effects associated with PNF on creative performance. This finding is consistent with prior studies [25, 56], suggesting that integrating feedback with strategy interventions can enhance individuals’ creative performance. Additionally, the DI group presented significantly greater improvements in AUT fluency, RPP fluency, and RPP flexibility than the CR group, whereas the differences in AUT and RPP originality did not reach significance. No significant differences were found between the DI and SM groups. This finding further supports FIT, which asserts that PNF can provoke stronger emotional responses that detract from cognitive resources, ultimately impairing performance. Overall, the study demonstrated that interventions targeting both emotional and cognitive processes can alleviate the negative consequences of PNF, enhancing creative performance. While strategy modification proved to be an effective predictor of improved creativity, cognitive reappraisal also played a supportive role. This highlights the potential of a “two-pronged” approach in helping individuals navigate the challenges posed by negative feedback, ultimately fostering greater creativity. General discussion Experiments 1 and 2 revealed consistent patterns regarding the effects of NFH on creative performance. Compared with PNF, TNF significantly enhanced AUT fluency and originality. In this process, a change in positive affect partially mediated the effect, whereas strategy modification fully mediated it. Overall, the results align with feedback intervention theory, which posits that negative feedback directs individuals’ attentional resources to different hierarchical targets [44]. Since attention is a limited resource [39], focusing on personal-level goals—such as self-esteem and impression management—leads to stronger negative emotional responses [57]. These strong emotions can overwhelm cognitive resources, making it difficult for individuals to reflect deeply on tasks and adjust their strategies, ultimately impairing performance. Conversely, those receiving task negative feedback focus more on task performance, which is less threatening to their self-concept. This focus allows for better reflection and strategy modification, resulting in improved performance. These findings corroborate earlier research indicating that negative feedback can facilitate strategy modification [4] and that performance is optimized when individuals generate more task-specific strategies [76]. However, FIT’s hypothesis that task negative feedback enhances performance through motivational boosts was not confirmed. In Experiment 2, motivational indicators, both of which reported high levels of motivation, were not significantly different between the PNF and TNF groups. This aligns with goal-setting theory, which suggests that the need to close the gap between current performance and external standards drives motivation [47]. Given the time constraints imposed in Experiment 2, further exploration of this finding is warranted. With respect to strategies to mitigate the negative effects of PNF, the cognitive reappraisal intervention for individuals receiving PNF was ineffective in alleviating its adverse impacts. This finding diverges from previous research on cognitive reappraisal [54, 72] but is consistent with that of Experiment 1, where affect changes only partially mediated the effects of negative feedback on creative performance. Although cognitive reappraisal may reduce negative emotions, it does not redirect attention away from personal-level goals. Creative tasks require substantial cognitive effort [61], indicating that emotional intervention alone is insufficient for improved performance. In contrast, the strategy modification intervention was effective, with the SM group showing significantly greater AUT and RPP fluency than the CA group did. This finding supports previous findings [25, 56] and suggests that awareness of the performance gap, combined with task-specific strategies, enhances task outcomes. Furthermore, all the creativity indicators, except for AUT flexibility, were significantly greater in the dual-intervention group than in the CA group. The dual-intervention group outperformed the SM group in terms of AUT originality and RPP flexibility, suggesting that combining interventions offers advantages over a singular approach. This finding supports FIT’s assertion that personal negative feedback affects affect, cognitive resources, and task performance [44]. Regarding the two types of creative tasks, the AUT outperformed the RPP task in this study. Several factors may explain this outcome. First, the tasks differ fundamentally in their nature. The AUT involves generating novel uses for an object, engaging simpler cognitive processes, while the RPP task requires navigating complex real-world scenarios and interpersonal dynamics, demanding broader cognitive skills. Research shows that the AUT activates different neural networks from problem-solving tasks [60], reflecting these distinct cognitive processes. Additionally, the RPP task's complexity may have made it more difficult, explaining the lack of significant performance changes from pretest to posttest in Experiments 1 and 2. In contrast, the AUT's simplicity likely made it more responsive to feedback, leading to greater performance improvements. These findings highlight the need to consider task type and complexity when assessing creativity, especially in the context of negative feedback, and suggest further research into how these factors impact creativity outcomes. Limitations and future directions While this study enhances our understanding of the effects of negative feedback on creative performance, it has several limitations. First, the role of motivation in the influence of NFH on creativity requires further exploration. Although this study did not yield significant results in this area, future research could better control experimental conditions or differentiate types of motivation to gain deeper insights. Second, the ecological validity of the findings needs to be established. This study primarily examined the effects of negative feedback from (simulated) reviewers in a laboratory setting. However, in real-life scenarios, creativity-related feedback typically comes from familiar sources such as teachers, peers, and supervisors [45]. Some creators seek negative feedback in online communities to refine their work [12]. These real-world feedback situations are more complex, which may limit the generalizability of the study’s findings. Future research could employ longitudinal studies and diary methods to investigate the effects of authentic negative feedback on individual creativity and the underlying mechanisms in everyday contexts. Third, the initial study design did not include a priori power analysis for Experiment 2 and 3. Although the sample size exceeded the typical threshold of 30 participants per group, post-hoc power analysis revealed that it did not meet the optimal power level required to detect smaller effects (0.71 for Experiment 2 and 0.78 for Experiment 3). While the study was sufficiently powered to detect medium to large effects, the lower power for detecting smaller effects may limit the generalizability of some findings. To address this limitation, future research will incorporate an a priori power analysis to better align the sample size with the expected effect sizes, ensuring more reliable and generalizable results. Another limitation concerns the assessment of creativity tasks, particularly the calculation of flexibility. In Study 1, two AUT tasks were double-coded and demonstrated strong inter-rater reliability, whereas the remaining AUT tasks and the RPP were coded by the first author alone. This approach was adopted because the AUT and RPP tasks were identical across Studies 1–3 and participants were drawn from similar populations, making the inter-rater reliability established in Study 1 reasonably representative of the reliability across all studies. Even so, reliance on a single rater for most tasks inevitably reduces objectivity. At the same time, it is noteworthy that only the RPP flexibility in Study 3 revealed significant group differences. This pattern may be attributable to the scoring method, but it also suggests that the approach did not artificially inflate false positives. Taken together, while the present coding procedure provides a reasonable estimate of reliability, future studies should employ multiple independent raters for all tasks to further enhance validity. In this study, we examined several psychological mechanisms—affect, strategy modification, and motivation—as potential mediators of how negative feedback hierarchy influences creativity. However, we did not test all three mediators within a single experiment, resulting in differing experimental models. While this approach enabled the independent examination of each mediator, it also meant the models were not directly comparable, with only the direct effects being consistent across experiments. This limitation hindered the development of a cohesive, integrated model. Future research would benefit from a unified mediator modeling approach, which would ensure consistency across experiments and allow for more robust comparisons. Theoretical and practical implications This study extends Feedback Intervention Theory by demonstrating how the NFH influences creativity through both affective and cognitive pathways. The findings highlight the importance of distinguishing between PNF and TNF, with TNF having a less detrimental effect on affective states and promoting more effective cognitive strategy modification. These results contribute to the growing literature on the impact of negative feedback on creativity and deepen our understanding of the roles that affect and strategy modification play in this process. Furthermore, the study offers insights into the boundary conditions of personal negative feedback, suggesting that interventions targeting affective and cognitive pathways can mitigate its harmful effects. From a practical standpoint, the study emphasizes that TNF is more beneficial than PNF in creative environments, as it minimizes emotional disruption and fosters more adaptive cognitive strategies. The findings suggest that feedback providers should focus on task-related issues rather than personal characteristics to maintain or enhance creativity. Additionally, interventions that focus on cognitive reappraisal and strategy modification can effectively mitigate the negative effects of feedback. These strategies provide creativity fosters (such as administrators, educators, and mentors) with actionable tools to better support creative individuals, helping them not only cope with negative feedback but also enhance their creative performance by refocusing cognitive efforts. Conclusion In the context of creative work, negative feedback can be a double-edged sword—it can either motivate the creator to enhance their product or discourage them. This study explores the impact of negative feedback on individual creativity and the underlying psychological mechanisms from the perspective of feedback hierarchy. Potential intervention strategies based on these mechanisms are also proposed. The key conclusions are as follows: Differential Impact of the Feedback Hierarchy: There is a significant difference in how the negative feedback hierarchy affects individual creative performance. Compared with personal negative feedback, task negative feedback tends to reduce positive affect less and promote strategy modification more, leading to enhanced creativity. Differential Impact of the Feedback Hierarchy: There is a significant difference in how the negative feedback hierarchy affects individual creative performance. Compared with personal negative feedback, task negative feedback tends to reduce positive affect less and promote strategy modification more, leading to enhanced creativity. Dual intervention strategies: The simultaneous application of cognitive reappraisal and strategy modification in the context of personal negative feedback can effectively mitigate its adverse effects on individual creativity. Dual intervention strategies: The simultaneous application of cognitive reappraisal and strategy modification in the context of personal negative feedback can effectively mitigate its adverse effects on individual creativity. These findings underscore the nuanced role of negative feedback in creativity and highlight the potential for targeted interventions to foster creative performance. Supplementary Information Supplementary Material 1. Acknowledgements We thank all participants who consented to participate in this study. Abbreviations Feedback intervention theory Negative feedback hierarchy Personal negative feedback Task negative feedback Positive affect Negative affect Cognitive reappraisal Strategy modification Dual intervention Authors’ contributions M.L. wrote the initial draft of the article. J.Z. edited the original draft and provided suggestions for improvements. W.P. supervised the research and contributed theoretical insights based on his expertise in the field. All authors revised the manuscript. Funding This work was supported by the Ministry of Education Humanities and Social Sciences Research Project [grant number 23YJCZH133]. Data availability The experimental materials used in this study can be found in the additional file, and the data for this study can be obtained from the first author upon request. Declarations Ethics approval and consent to participate We received ethical approval for the study from the Human Subjects Protection Committee of the East China Normal University (Approval number: HR517-2020). Our study complies with the Declaration of Helsinki. Informed consent was obtained from all participants. Consent for publication All participants provided written informed consent for publication of the data and results in this study, except for identifying information, i.e. their email addresses. Competing interests The authors declare no competing interests. Footnotes Publisher’s Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. Contributor Information Mengrong Liu, Email: mrliu@zisu.edu.cn. Weiguo Pang, Email: wgpang@psy.ecnu.edu.cn. References Associated Data Supplementary Materials Supplementary Material 1. Data Availability Statement The experimental materials used in this study can be found in the additional file, and the data for this study can be obtained from the first author upon request.