Brief Growth Mindset and Mindfulness Inductions to Facilitate Task Persistence After Negative Feedback

Academic setbacks may emerge frequently in school and can curb students’ motivation and learning engagement (Kluger & DeNisi, 1996; Van Dijk & Kluger, 2011). Moreover, it may also boost the probability of dropping out (Vitaro et al., 2001). To reduce the potential negative consequences of academic setbacks, considering certain individual differences might be helpful. One such individual difference might be students` beliefs about the malleability of their intelligence (i.e., implicit theories of intelligence; Dweck & Leggett, 1988). Individuals with a fixed mindset believe that their intelligence is stable and individuals with a growth mindset believe that intelligence can be improved (Dweck & Leggett, 1988; Yeager & Dweck, 2020). Implicit theories of intelligence have been shown to influence whether students perceive challenging or negative situations, like academic setback experiences in adaptive or maladaptive ways (Aditomo, 2015; Dweck & Yeager, 2019). These different perceptions can then transfer to more constructive or destructive thoughts, feelings, and behaviors and may predict students’ learning and growth.

Different mechanisms have been shown to explain different patterns of the two belief systems. For instance, students with a fixed mindset tend to set performance-type goals in achievement situations (Davis et al., 2011): they are motivated to measure and demonstrate their intellectual capacities mostly based on comparing themselves to others. They avoid challenging situations where they do not feel like they have the capabilities to perform well (Mangels et al., 2006). In this mindset, setbacks are signs of low abilities (Blackwell et al., 2007). In contrast, a growth mindset can promote setting mastery-type goals aiming to learn in academic settings by embracing challenges, focusing on self-improvement, remaining persistent, and exerting effort on demanding activities (Davis et al., 2011; Dweck & Yeager, 2019; Elliot et al., 2011). For individuals having a growth mindset, setbacks are part of the learning process, and these experiences are not measures of their intellectual abilities. Moreover, a growth mindset is associated with more adaptive goal setting, like learning or mastery-oriented goals (Davis et al., 2011). Several studies have underpinned the theoretical implications of a growth mindset, concerning effort beliefs (e.g., Miele et al., 2013), challenge-seeking (Porter et al., 2020; Yeager et al., 2019), persistence (Porter et al., 2020) focus on learning (Mangels et al., 2006), and achievement in school (Dweck & Yeager, 2019; Lam & Zhou, 2019; A. J. Mrazek et al., 2018; Renaud-Dubé et al., 2015).

Beliefs about intelligence, however, are not set in stone and can be changed. Hong and colleagues (1999) described that the short-term psychological mechanism behind mindset induction is that although people might have persistent preferences for growth or fixed belief systems, both systems may represent familiar modes of thought to most individuals at some level. Thus, students who receive a growth mindset induction might be nurtured to attribute setbacks to the lack of effort invested in the tasks, while those who get a fixed mindset induction would attribute setbacks to the lack of ability (Blackwell et al., 2007; Dweck & Yeager, 2019). Yet, only a handful of studies demonstrated the beneficial effects of growth mindset induction in experimental settings related to behavioral outcomes (Niiya et al., 2010), and after getting negative feedback (Henderson & Dweck, 1990; Hong et al., 1999; Nussbaum & Dweck, 2008). Niiya and colleagues (2010) found that participants subjected to growth mindset induction practiced more on an alleged IQ test than those in the fixed mindset group. Similarly, Hong and colleagues (1999) reported that participants exposed to a growth mindset induction (as opposed to a fixed mindset induction) were more likely to take a remedial tutorial after experiencing setbacks. Furthermore, Nussbaum and Dweck (2008) found that after setbacks, participants in the growth mindset condition were more likely to choose to work on unmastered material (i.e., seek challenge), instead of already mastered material as opposed to those in the fixed mindset condition. These findings suggest that growth mindset inductions may facilitate people to demonstrate more self-improving behavior (e.g., being more persistent) on an optional behavioral task after facing negative feedback.

A recent study has shown that despite facing setbacks with a growth mindset, one may still sometimes experience persistent fixed-mindset thoughts and emotions and can doubt their intellectual abilities (Orosz et al., 2020). We argue that another individual difference that might be helpful in handling setbacks, in addition to a growth mindset, is mindfulness. Mindfulness refers to being aware of the present moment without judgment (Creswell, 2017; Kabat-Zinn, 2003). Dispositional mindfulness has been widely studied and shown to be helpful in handling negative emotions and ruminative thoughts (e.g., Gu et al., 2015). People with higher dispositional mindfulness show less rumination (Tomlinson et al., 2018), more positive academic reappraisal, and higher self-efficacy following failure (Hanley et al., 2015). Thus, in addition to a growth mindset, mindfulness might be helpful for people facing setback experiences. Mindfulness has shown to be helpful not only as an individual difference but also as a practice - in the form of mindfulness training (for a review, see Keng et al., 2011). Thus, as a complementary strategy in addition to implementing a growth mindset, mindfulness-promoting messages or basic mindfulness meditation may be useful to decrease counterproductive thoughts and facilitate coping with setbacks by reducing rumination and worry (Gu et al., 2015).

Mindfulness meditation-based interventions are increasingly implemented in educational settings (Schonert-Reichl & Lawlor, 2010; Schonert-Reichl & Roeser, 2016; Zenner et al., 2014). Studies show that these interventions may boost cognitive performance and resilience to stress (Takacs & Kassai, 2019; Zenner et al., 2014), in addition to nurturing well-being and mental health (Carsley et al., 2018). Brief mindfulness inductions also seem to offer small benefits to cognitive functioning, as they may facilitate coping with negative emotions and rumination (Gill et al., 2020), while evidence regarding other emotion-regulatory strategies like decentering is inconclusive (Leyland et al., 2019). Prior studies examined the interplay between beliefs of mindfulness and intelligence (Kong & Jolly, 2019; A. J. Mrazek et al., 2018; Orosz et al., 2020; Samuel & Warner, 2021), but they did not examine whether and how mindfulness meditation can contribute to the beneficial effects of growth mindset inductions in terms of taking remedial actions after setbacks.

Mindfulness meditation may have both a top-down and a bottom-up influence on task persistence. Top-down processes can include more efficient attention allocation (Malinowski, 2013; Norris et al., 2018), while bottom-up processes may include a decrease in physiological and subjective arousal (e.g. Shearer et al., 2016). In the present study, we used a brief and pure mindfulness meditation that focused on awareness and voluntary dismissal of experiences (e.g., bodily sensations, thoughts, feelings), and existing in the present moment (e.g., Lindsay & Creswell, 2017). We chose to use this core version of mindfulness induction to exclude the possible confounding mindfulness-related elements such as acceptance of emerging thoughts and feelings, self-compassion, and normality of having a negative experience. At the same time, decentering was highlighted as an optional coping strategy by mentioning that letting thoughts go can contribute to optimal learning. In sum, we expected a direct benefit of mindfulness manipulation on task persistence by the application of decentering as a coping mechanism in the context of negative feedback and indirect effects through optimal emotion regulation.

To sum up, both growth mindset and mindfulness inductions have been shown to promote mastery behaviors — such as task persistence — in the face of setbacks. Even though two prior studies (e.g. Orosz et al., 2020; Samuel & Warner, 2021) have combined growth mindset and mindfulness interventions to help students face difficulties, differentiating the effects of these interventions was not possible in these studies. As these studies combined the two interventions and did not test the two main effects (growth mindset and mindfulness induction) separately, we have limited knowledge about the underlying mechanisms. Thus, conducting laboratory and field experiments would be crucial to understand the causal effects and mechanisms of state mindfulness and mindset on mastery behaviors separately.

To identify some of the relevant mechanisms, it might be worth exploring secondary outcomes. We expect that the inductions will influence participants’ achievement goals — shifting from performance-oriented to mastery goals —, resulting in decreased boredom and anxiety, which, in turn, may lead to better performance. Growth mindset inductions have been shown to promote the adoption of mastery goals as they change the focus from performance-orientation to learning-orientation (Cury et al., 2006). Furthermore, it has been shown that people who are more mindful, set more adaptive goals (i.e., mastering new skills or tasks) in achievement situations (McCarthy, 2011). As people set more adaptive goals, they may experience decreased anxiety as the focus changes from fulfilling expectations to self-improvement. In fact, decreased anxiety has been found both after growth mindset (Schleider & Weisz, 2018) and mindfulness inductions (Zenner et al., 2014). Focusing on self-improvement may also result in less boredom (Karumbaiah et al., 2017). As a consequence of adaptive goal setting, and decreased anxiety and boredom, we expect people to attempt solving more optional tasks, which could result in higher cognitive performance (Blackwell et al., 2007; M. D. Mrazek et al., 2013).

Additionally, it is plausible to assume that individual differences in mindset beliefs and dispositional mindfulness play a role in the effects of mindset and mindfulness interventions. Emerging evidence suggests that individual differences can moderate the influence of interventions. For example, dispositional mindfulness is one of the most important individual differences that can moderate the effectiveness of mindfulness interventions (Tang & Braver, 2020). People with higher dispositional mindfulness were found to benefit more from a mindfulness-based stress reduction program with regard to well-being and stress as compared to individuals low on dispositional mindfulness (Shapiro et al., 2011). Similarly, people with higher dispositional mindfulness exhibited better stress regulation in a family quarrel after receiving a brief mindfulness induction (Laurent et al., 2015). Increased dispositional mindfulness may boost the effect of mindfulness induction, as they possibly have more practice with meditation and can get into a mindful state more easily (see Tang & Braver, 2020).

Based on results from other psychological constructs (e.g., mindfulness; Tang & Braver, 2020) and previous results on mindset manipulations, we can expect two different mechanisms in terms of the moderating effect of dispositional mindset. On the one hand, similar to the results of the moderating effect of dispositional mindfulness (Laurent et al., 2015; Shapiro et al., 2011), a growth mindset induction might be useful for those who already have a pre-existing growth mindset belief. As Hong and colleagues (1999) pointed out, people can be familiar with ideas of both growth and fixed mindsets to some extent. Messages that match people’s dispositional mindset beliefs may get reinforced. Thus, those who already think that intelligence can be improved might be more convinced by reading a persuasive article that is consistent with this belief. This can be particularly true in the context of IQ assessment where researchers — who are seen as authorities on the topic of intelligence — provide information about the malleability of intelligence. On the other hand, it may be possible that learning about the malleable nature of intelligence acts as a novel piece of information among people who think intelligence is fixed. This may result in shifting goals and changing expectations, thus impacting motivation and persistence. Accordingly, some studies found that growth mindset interventions were more effective in people with a fixed mindset at baseline (Blackwell et al., 2007; Török et al., 2022; Yeager et al., 2014). Thus, based on previous findings, we expect a moderating effect of pre-induction growth mindset belief on the effect of growth mindset induction. However, we cannot currently predict the direction of the moderation effect as both positive and negative moderation seems equally plausible for different reasons. For the reasons discussed above, people with both low and high pre-induction growth mindset beliefs can benefit from a growth mindset induction via different mechanisms.

In a pre-registered experiment, we aimed to test whether task persistence can be boosted after negative feedback. As the primary goal of this study, we tested if a brief growth mindset and/or mindfulness induction (independent variables) can facilitate participants to complete more optional practice tasks (dependent variable) after experiencing negative feedback. We also assumed that negative feedback can be regarded as less threatening if one views their intelligence as malleable (through adaptive growth mindset beliefs) and mindfulness meditation (through adaptive emotion regulation and decentering) would help participants keep an emotional distance from failure, resulting in a more positive attitude towards the task, less anxiety, and ultimately manifesting as increased task persistence. Thus, we hypothesize that a growth mindset induction can help boost task persistence and that augmenting a mindset induction with a brief mindfulness induction may further promote this effect.

The project was pre-registered on November 30, 2019, at the following link: https://osf.io/xf3uc).

The number of required participants was calculated using Monte Carlo simulation. We expected that participants in the control group would solve 4.0 extra tasks on average (out of the maximum of 17), whereas the mindset/mindfulness group was expected to solve 5.7 extra tasks on average (this is the equivalent of the effect size of 1.22 in rate ratio). We used these as lambda parameters to simulate datasets following Poisson distributions. We aggregated the results of generalized linear models (using log link for Poisson distributed outcome) on the simulated datasets. The p values associated with one-sided hypothesis tests were used to calculate the proportion of significant tests (statistical power).

We tested cell sizes up to 200 using ten thousand samples per cell. We investigated if any or both main effects had an associated p-value of less than .05 in the regression model. We calculated the exact number of participants required to reach 80%, 90%, and 95% power for both scenarios using linear interpolation. Based on this, the cell size of 98 yielded 80% power for detecting both main effects at the same time, while 95% power to detect at least one of the main effects. We expected that a 10% surplus was required to complement the participants who have to be excluded (e.g. due to suspicion about the negative feedback manipulation or inattention). Thus, we planned to recruit at least 432 participants.

Participants were recruited from the university participant pool of ELTE Eötvös Loránd University, Budapest. 441 participants completed the study. We excluded 58 participants due to technical malfunction or based on the pre-registered exclusion criteria (see Figure 1). The final sample included 383 participants (mean age = 22.3 years, SD = 4.2), of which 306 (80%) were female. The study was approved by the university IRB (Ref.: 2016/082) and adhered to the requirements of the Declaration of Helsinki. Participants provided consent and took part voluntarily in exchange for partial course credits.

The research procedure is shown in Figure 2. Data were collected in small groups of approximately 20 people in a lab at the university. Participants were told that they would be involved in the validation of an online IQ test, based on a reliable offline IQ test. We called the test “Stanford Raven test”. The research protocol was fully computerized, and all randomization was done by the Qualtrics questionnaire platform.

First, participants completed questionnaires about trait characteristics (see later), then they randomly received a fixed or a growth mindset induction. In the growth mindset condition, participants were shown a recorded audiovisual presentation about the malleability of intellectual skills, using analogies of neuroplasticity (Zatorre, 2013). In the fixed mindset condition, participants were shown another presentation about the genetic background of intellectual skills, demonstrated by twin experiments and illustrated by gray matter overlap in the brain (Thompson et al., 2001). After the mindset induction, participants completed a practice IQ test session (12 items) as a “warm-up”. They had 45 seconds time limit to complete each task, thus solving them was difficult. Irrespective of performance, they were given a negative evaluation (“We calculated your IQ score based on your responses. Compared to other participants, your IQ is: below average”).

Afterward, participants were randomly exposed to guided mindfulness meditation or a control condition. The mindfulness induction included a brief (nine-minute-long) mindfulness exercise based on the most popular online mindfulness meditation app introductory exercise that is freely available. The beneficial effects of these materials were examined by prior studies (e.g., Flett et al., 2019). It instructed participants to focus on the present moment and let go of negative thoughts and feelings. The material started with a brief breathing exercise, continued with relaxation and mindful listening, and ended with mindful thoughts. In the control condition, a brief (seven-minute-long) audio clip about healthy eating and the effects of diet on brain functioning was implemented.

After they listened to either the mindfulness or the control audio clip, we asked students about how much they were able to get into a focused, relaxed state; how attentive they were during the audio tapes; and answered a mindset induction check item. In the following block, participants were instructed to complete 12 items from the final version (i.e., not just the warm-up) of the offline IQ test with the instruction: “In this section, you can complete the tasks from the final version of the Stanford Raven test. The task will be the same as in the warm-up exercises, but now, for real.”

Subsequently, participants could choose to practice before the last section of the IQ test by completing alternative, extra practice items (a maximum of 17 items). The instruction was: “In the following section, you can complete some additional tasks that we may use in the final version of the test package. This can help us measure IQ as accurately as possible. Since you’ve worked a lot on previous tasks, you can choose how many of them you would like to complete. This block is optional, so complete these alternative tasks only if you want to”. Before each practice item, we asked participants whether they wanted to continue with a new practice item or proceed to the final IQ test section. Each item had a 45-second time limit, imitating real-life items from the main IQ blocks. We used the number of items completed in this block as a behavioral measure of task persistence. Participants were told previously that they would have to wait for all other participants with the debriefing, thus skipping more or all practice tasks did not allow them to leave sooner. We assumed that those who invested extra effort and time in this block chose to improve, instead of avoiding exerting extra effort.

In the next section, the final part of the IQ test (12 items) started, with the instruction: “In this last section, you will find the final test items, which will refine the result of your previous completion”. Participants had 45 seconds to complete each task.

After the IQ test exercises, participants completed questions regarding task boredom (three items), anxiety and stress (six items), and achievement goals (12 items) regarding the IQ test and the practice task session. Finally, all participants were probed for suspicion, debriefed (written and verbal) to alleviate the distress that may have arisen due to negative feedback, and dismissed.

Besides the main outcome measures, we assessed several psychological constructs that may serve as moderators or mediators between the inductions and task persistence. Scale values were calculated by taking the average of the items. Thus, the upper and lower class limits correspond to the minimum and maximum values of the response scales.

IQ task (Raven Progressive Matrices). The Raven Progressive Matrices assess “general cognitive ability” or “meaning-making” ability (J.C. Raven & Court, 1998; John C. Raven, 2000). We used items from two series: the basic one, Standard Progressive Matrices (or SPM); and the advanced one, Advanced Progressive Matrices (APM). We calculated the percentage of correct responses against all tasks for each test block.

Task persistence. The main outcome variable was the number of completed optional tasks between the warm-up and “final” IQ sessions. Items were also selected from the Raven Progressive Matrices; however, the instruction indicated that these exercises were not mandatory, and participants completed them only if they wanted to. Participants could choose to solve zero to seventeen tasks. We used this number as a behavioral indicator of task persistence. We expected that using a behavioral measure increases ecological validity compared to self-report measures.

Prior Growth Mindset. The mindset was measured using four items from the original Theory of Intelligence Questionnaire (Dweck, 2000). The scale measures the extent to which intelligence is perceived as a malleable factor. Respondents indicated how much they agreed with statements such as “You have a certain amount of intelligence, and you can’t really do much to change it” (α = .91). Items are rated on a six-point Likert scale from “strongly disagree” (1) to “strongly agree” (6).

Dispositional Mindfulness. The Cognitive and Affective Mindfulness Scale (Feldman et al., 2007) was used to determine dispositional mindfulness. This 12-item scale measures everyday mindfulness and focuses on the degree to which examinees experience their thoughts and feelings. Respondents indicated how much statements such as “I am able to accept the thoughts and feelings I have” relate to them (α = .73). Items are rated on a four-point Likert scale from “rarely/not at all” (1) to “almost always” (4).

Achievement Goals. The Achievement Goal Questionnaire (Elliot et al., 2011) was used to assess goal setting regarding the IQ test and the practice items. Two items corresponding to each of the six achievement goal types in the 3 × 2 model: task-approach (e.g., “To know the right answers to the questions”), task-avoidance (e.g., ”To avoid getting a lot of questions wrong”), self-approach (e.g., “To do better on this test than I typically do in this type of situation”), self-avoidance (e.g., “To avoid doing worse on this test than I normally do on these types of tests”), other-approach (e.g., “To outperform other students”) and other-avoidance (e.g., “To avoid doing worse than other students”) goals. The scale ranged from “not at all like me” (1) to “completely like me” (7). The six achievement goals (using task-based, self-based, and other-based standards of competence, in terms of approach and avoidance) were assessed for the IQ test and the practice tasks, resulting in a total of twelve achievement goal variables.

Task-related boredom and stress. Task-related boredom questions were developed based on the Hungarian translation of the Achievement Emotions Questionnaire (Pekrun et al., 2013). We included three items: “How boring was the IQ test for you?”; “How exciting was the IQ test for you?”; “How bored were you during the assessment?”. To check task-related anxiety and stress, we developed six new items specifically related to students’ experiences in this study. Participants responded to the following items: “How nervous were you during the assessment?”; “How stressful did you find the IQ test?”; “How calm were you when completing the IQ test?”, “Did stress affect your performance on the IQ test?”, “To what extent has stress worsened your performance on the IQ test?”, “How challenging was the IQ test for you?”. However, the last item loaded to a separate factor, hence we did not include it in the scale. Thus, anxiety was measured using five items. All nine items were rated on a six-point Likert scale from “not at all” (1) to “very much” (6). Scales were extracted using exploratory factor analysis (see later), and factor scores were used for the analysis.

Statistical analyses were performed with R 4.1.0 (R Core Team, 2021). Within R, the tidyverse package (Wickham et al., 2019) was used for data transformation and visualization. For assessing the effect on the number of completed optional tasks, zero-inflated negative binomial regression was performed using the pscl package (Zeileis et al., 2008). The zero-inflated regression fits two models: the zero-inflated part is a binomial logistic model that attempts to predict if a participant solved at least one extra task. The other is the count model, which attempts to predict the number of tasks solved by the participant. The effect of induction on performance was investigated using multiple linear regression. Assumption checks were performed using the performance package (Lüdecke et al., 2021). The psych package was used for factor analysis (Revelle, 2021). To get the task boredom and anxiety scores, we used exploratory factor analysis on the items of the post-study questionnaire using direct oblimin rotation. We used the factor scores for the task boredom and anxiety factors in the subsequent linear regressions (see details in the supplementary material S1). Bayes Factor (BF₁₀) was calculated using the formula: exp(ΔBIC₀₁/2) (Wagenmakers, 2007).

Table 1 shows the variable means and standard deviations for each group. We found no significant difference between the groups in the sample characteristics or any of the traits measured before the manipulation (see supplementary material S2 for statistical tests).

As Figure 3 shows, the group difference in agreement with the question: “Your intelligence is an attribute that you cannot change over time” suggests that we successfully induced the respective mindset in the participants (fixed mindset group mean (SD) = 3.61 (1.25), growth mindset group mean = 2.76 (1.14); t(380) = -8.36, p < .001, Cohen’s d = 0.71, 95% CI [0.50, 0.91]). On the other hand, the lack of group difference in the answer to the question: “How much could you get into a calm and focused state of mind?” suggests that we were not able to perceivably induce a mindful state in the participants (control group mean (SD) = 3.41 (1.39), mindfulness group mean = 3.34 (1.32); t(380) = -0.48, p = .629; Cohen’s d = 0.06, 95% CI [-0.14, 0.26]). Both statistical tests were controlled for the pre-manipulation mindset or mindfulness, respectively.

We tested the main hypotheses that the mindset (growth vs. fixed) and/or mindfulness induction (vs. control) — increases the number of optional exercises chosen for practice, using Poisson regression (as pre-registered). Although we found a significant main effect of the mindset induction in the expected direction — the growth mindset manipulation increased task persistence — the assumptions for Poisson regression were not met. The tests showed both overdispersion (dispersion = 5.71, $χ$² = 2162.18, p < .001) and zero-inflation (observed zeros = 131, predicted zeros = 11, ratio = 0.08). Perumean-Chaney and colleagues (2013) report that over-dispersion and zero-inflation can increase type-I error, and they suggest using zero-inflated negative binomial regression in these cases. Therefore, a more adequate zero-inflated negative binomial regression was conducted. Neither the count model, nor the zero-inflated model yielded significant effects for mindset induction (p = .260 and .923), mindfulness induction (p = .976 and .915), or the interaction of the two (p = .642 and .962; see Table 2). The low Bayes Factor (BF = 4.88 × 10^-8) for the model suggested that it is very unlikely that any of the inductions affected the outcome.

Apart from the experimental inductions, we also used participants’ dispositional growth mindset and mindfulness as moderators on the effect between the inductions and task persistence (Table 3). For the count models, we found that both pre-manipulation mindset beliefs and dispositional mindfulness moderated the effect of the respective experimental induction on the number of completed optional tasks. Moreover, dispositional mindfulness — but not pre-induction mindset — also significantly moderated the effect of mindfulness induction on completing at least one task in the zero-inflated model.

The effect of mindset induction on the number of completed optional tasks increased by a factor of 1.31 (or +31%; 95% CI [1.02, 1.67]) for a unit increase in the pre-induction growth mindset. This means that participants who had a higher pre-induction growth mindset were more likely to solve extra tasks if their belief was reinforced by the induction.

Similarly, the effect of mindfulness induction on the number of completed optional tasks increased by a factor of 1.46 (or +46%; 95% CI [1.15, 1.86]) for each unit increase in dispositional mindfulness. Moreover, according to the zero-inflated model, the effect of mindfulness induction increased the chance of solving at least one optional task 2.45 fold (or +145%; 95% CI [1.10, 5.48]) for each unit increase in dispositional mindfulness. These results suggest that those with higher dispositional mindfulness displayed more task persistence to solve extra tasks after a brief mindfulness meditation session as compared to people with lower dispositional mindfulness (see Figure 4).

We also tested if the growth mindset or mindful state was associated with better performance after negative feedback on the “real” IQ test (i.e., where we instructed participants to solve the IQ items from the final version of the IQ test). As the assumption for homoscedasticity was violated, we used heteroscedasticity-consistent standard errors to test the significance of the predictors. In line with non-significant effects on task persistence, neither the predictors (mindset p = .533, mindful state p = .911, interaction p = .800, see Table 4), nor the linear model overall, F(3, 379) = 0.45, p = .716 were significant on IQ-test performance. In contrast to results on task persistence, dispositional mindfulness did not have a moderating effect on performance (see supplementary material S4).

Achievement goal setting was proposed as a potential mediator mechanism for the inductions to affect task persistence. As the inductions did not increase task persistence, uncovering mechanisms became a less important aim for the study. Nevertheless, we analyzed the effect of the mindset and mindfulness inductions on achievement goals and found no significant effect. Moreover, we also investigated the effect of inductions on anxiety and boredom, but again, we found no evidence that these secondary outcomes were affected by the inductions. Analysis of the aforementioned outcomes can be found in Supplementary material S5. We also investigated further moderator effects that are outside of the scope of this paper, see Supplementary material S3.

Currently, little is known about the effectiveness of brief inductions that might facilitate task persistence after receiving potentially demotivating negative feedback. In a parallel-groups experiment, we tested whether brief mindset and mindfulness inductions can mitigate the adverse effects of negative feedback on students’ task persistence. We aimed to induce fixed or growth mindset beliefs, and gave participants negative feedback on a bogus intelligence assessment task. Then they took part in either a guided mindfulness induction or a control condition. Afterward, we offered participants optional practice tasks and used the number of completed tasks as a behavioral measure of task persistence. In contrast to our hypotheses, we have found that task persistence was not affected by the mindset or the mindfulness inductions. Nonetheless, we found that the inductions were effective for participants with a higher pre-induction growth mindset or dispositional mindfulness. We also investigated if mindset and mindfulness inductions affected secondary outcomes, such as achievement goals, anxiety, task boredom, and task performance. However, we found no such effects.

Prior studies suggest that a growth mindset might be associated with learning and mastery-oriented attributions, motivations, and behaviors that tend to facilitate task persistence when facing setbacks (e.g., Burnette et al., 2013; Dweck et al., 1995; Nussbaum & Dweck, 2008). However, only a handful of studies have investigated the effect of growth vs. fixed mindset induction on mastery behaviors in an experimental setting (Hong et al., 1999; Mueller & Dweck, 1998; Niiya et al., 2010; Nussbaum & Dweck, 2008). These studies have found that a growth mindset induction can facilitate mastery behaviors in the face of setbacks. Thus, we expected that a growth mindset induction would have a positive effect on people’s effort to persist longer on optional tasks. We assumed that the growth mindset induction would increase the perceived usefulness of persistence and effort (Hong et al., 1999; Miele et al., 2011; Miele & Molden, 2010; A. J. Mrazek et al., 2018) and participants would attribute their failure experience to the lack of effort invested in the IQ test (Blackwell et al., 2007; Dweck & Yeager, 2019). Based on this, we expected that the growth mindset induction would boost task persistence after negative feedback. Even though the induction seemed to be effective, it did not produce the predicted effect.

There might be some explanations for not being able to detect the effect of the growth mindset induction on task persistence for all participants. It might be possible that inducing a growth mindset may need more time to manifest in actual behavior. For instance, Hong and colleagues (1999) assumed that a short-term psychological mechanism in which a growth mindset was induced could promote effort after receiving negative feedback. They found that mindset induction activated the “meaning framework” of the person in a way that facilitated the attribution of performance to effort rather than ability. Yet, it is possible that the “meaning framework” might not always be activated instantly and may rather take effect in the longer term. While Mueller and Dweck (1998) could induce a growth or a fixed mindset by praising effort and persistence (“You must have worked hard at these problems’‘) or ability (’‘You must be smart at these problems’’) of the participants, Li and Bates (2019) could not replicate this effect. Thus, it appears that brief growth mindset inductions might not always lead to the expected outcomes. Studies that have successfully changed the mindset meaning system (Rege et al., 2020) and/or affected behavior usually applied more thorough and elaborate mindset manipulations; possibly affecting deeper thoughts and feelings. Noteworthy examples are online wise interventions (e.g., Paunesku et al., 2015; Yeager et al., 2019), or multi-week in-person interventions (e.g., Blackwell et al., 2007). The reason for finding the moderating effect of dispositional growth mindset might be exactly because the short induction could reinforce and activate the growth mindset beliefs of those individuals who already had a congruent belief.

Another explanation might be that the effect of growth mindset induction depends on the context or the characteristics of the participants (see Yeager & Dweck, 2020). It may be possible that growth mindset inductions can be more efficient in a different context or in a different population (see Walton & Yeager, 2020). Just like in our case, the inductions only boosted task persistence among those with a pre-existing growth mindset belief. Similarly, growth mindset induction has been shown to be more effective among, lower-achieving high school students (Bettinger et al., 2018; Paunesku et al., 2015; Yeager, Romero, et al., 2016), racial minorities, and first-generation college students (Broda et al., 2018; Orosz et al., 2017, 2020; Yeager, Walton, et al., 2016). The present sample was recruited from the most selective, prestigious higher education institution in Hungary with a low proportion of the aforementioned academically at-risk groups.

Lastly, prior studies showed that growth mindset messages could lead to greater performance benefits if the educational context is supportive of growth mindset messages (Yeager et al., 2019). It is possible that the educational context did not provide the most optimal conditions for the growth mindset meaning system for all students. However, the pre-existing growth mindset measure might have reflected students’ context-specific mindset-related experiences, and those who experienced a growth mindset context — those scoring higher on the growth mindset scale — could benefit more from the growth mindset induction. In sum, our results suggest that contexts that facilitate a growth mindset can promote the beneficial effects of growth mindset messages and interventions.

As already mentioned, we have found that the growth mindset induction was effective among participants with congruent dispositions, similar to results with other psychological constructs (e.g., mindfulness; Tang & Braver, 2020). However, our hypothesis referred to the moderation effect in the other direction as well (i.e., students with a pre-existing fixed mindset belief could benefit more from the induction). We proposed this hypothesis because some studies found that growth mindset interventions — which are much longer than brief inductions and meant to facilitate a growth mindset in the longer term — improved grades the most (Blackwell et al., 2007; Yeager et al., 2014), and decreased self-handicapping (Török et al., 2022) among participants with more of a fixed mindset belief. We did not find evidence for this presentiment, such that the growth mindset induction was more useful for those who already had a pre-existing growth mindset belief.

It should be noted that we only found a significant moderation effect for solving more extra tasks (count model), and not for solving at least one extra task (zero-inflated model). This means that solving at least one task was not predicted by the interaction of mindset induction and pre-induction growth mindset. As we argued earlier, mindset beliefs are supposed to affect goal setting, which should increase the motivation for practicing in general. We currently do not see a well-supported theoretical explanation for this finding.

While a growth mindset is known to be beneficial in the learning context, a recent study has shown that facing setbacks even with a growth mindset, can lead people to experience fixed-mindset thoughts and emotions (Orosz et al., 2020). One strategy that could help people deal with ruminating thoughts and negative emotions is mindfulness (Schonert-Reichl & Lawlor, 2010; Schonert-Reichl & Roeser, 2016; Zenner et al., 2014). There have been only a few attempts to probe the potential beneficial role of mindfulness elements in facilitating the implementation of growth mindset beliefs. For example, Orosz and colleagues (2020) found that an online growth mindset intervention, by integrating certain mindfulness elements, increased the grade point average of both high school and university students months later. Although this work reports promising behavioral outcomes, it did not provide extensive details about how mindfulness-related psychological mechanisms are responsible for academic benefits. However, building on their results, it appears that a combined growth mindset with mindfulness treatment may result in stronger behavioral intentions. Thus, in this experiment, we expected that a brief mindfulness exercise could help people to handle negative thoughts and feelings from an optimal distance with acceptance and without self-judgment. As a result, we expected people to invest more effort in self-improvement after negative feedback, however, our study did not support this hypothesis.

Some explanations for why mindfulness induction failed to boost task persistence may come from specific effects of mindfulness that could interfere with motivation. It is possible that the mindfulness induction made participants more disengaged from the intelligence test situation, made them see the situation from a broader perspective, or made them more accepting of their cognitive abilities. All of these would decrease the subjective significance of the IQ task, potentially causing lower task persistence. It is also possible that certain participants were not open to the mindfulness meditation exercise (for example, they might think that it was too esoteric) and these negative attitudes could have a negative impact on the effectiveness of the mindfulness induction.

While mindfulness-based interventions are effective in terms of emotion regulation and coping with negative feedback and stress (Zenner et al., 2014), such interventions are typically longer than a simple nine-minute-long experimental induction. Thus, a single meditation session may have a minimal effect on novice meditators, especially in a potentially stressful situation such as right after receiving self-threatening negative feedback. It is conceivable that a short induction is only effective for people with a trait that matches the induction, that is, people with high dispositional mindfulness. This is in fact what we found in the present study. Previous literature supports the idea that the efficacy of mindfulness induction is moderated by personality traits (Winning & Boag, 2015) and by dispositional mindfulness specifically (Tang & Braver, 2020). For instance, a mindfulness induction was found to be only effective for stress recovery in couples’ conflicts for participants with high dispositional mindfulness (Laurent et al., 2015).

Furthermore, Eberth and Sedlmeier (2012) found that complex mindfulness programs such as the Mindfulness-Based Stress Reduction (MBSR) program are more effective than meditation practice alone. This might be explained by the fact that such complex programs include psychoeducational components. Thus, it is debated whether mindfulness meditation practice alone can be an effective strategy for stress management for everyone. The present mindfulness induction not only lacked these psychoeducational elements but it was also separated from the other parts of the experiment. This means that participants just listened to the mindfulness meditation without getting any contextual information about why they needed to do this exercise.

It would be possible to provide more context (e.g., the reasons why they received a mindfulness exercise) and intertwine the message of mindfulness with other relevant components of the experiment or intervention. This could include psychoeducative elements (e.g., being explicit about how mindfulness can help in coping with setbacks) that could contribute to the success of such manipulation, especially for those who are unfamiliar with mindfulness. Similar to the importance of contexts that are conducive to establishing a growth mindset meaning system, it appears that embedding mindfulness practices in a broader mindfulness-related meaning system might be useful. For instance, in Orosz and colleagues’ (2020) intervention study, the mindfulness elements (e.g., acceptance of negative experiences, decentering, self-distancing, letting them go) appeared as an integral part of the growth mindset material and instead of involving any sort of meditative practice, they were carefully and precisely contextualized in terms of how to use mindful strategies in academic setback situations.

Despite these potential reasons for finding null results, the basic mindfulness induction applied in the present study brings important theoretical relevance to the results. While it might be ecologically less valid, we believe that the basic mindfulness induction in the present study is methodologically more rigorous. The study controlled for confounding effects of related psychoeducational components and coping mechanisms, and purely tested the effects of state mindfulness as opposed to the effects of providing coping mechanisms to participants.

Previous studies have also found that mindfulness induction is not effective in facilitating stress recovery after failure in perfectionist students (Azam et al., 2015). Along the same line, some intervention studies show that a key mechanism for the beneficial effects of mindfulness programs on well-being might be the cultivation of self-compassion (Gu et al., 2015; Roeser et al., 2013). The mindfulness meditation material used in the present study did not refer to self-compassion. In sum, it appears that our mindfulness induction was only effective in fostering task persistence following negative feedback for participants with high dispositional mindfulness. Even without providing an optimal context to embed the mindfulness messages, these participants might be more susceptible to any mindfulness-related stimuli. We assume that for participants with lower levels of dispositional mindfulness, a longer mindfulness intervention, possibly supplemented by elements of self-compassion might be needed. It might be possible that exposing those with high dispositional mindfulness to a mindfulness induction may bear an increased effect that might resemble a longer-lasting or more complex intervention.

According to our results, dispositional mindfulness can help in using mindful strategies when one faces difficulties. However, it is also possible that not only traits, but students’ mindsets about mindfulness can be also beneficial. For instance, in Orosz et al’s (2020) intervention studies, college students’ beliefs about the changeability of their mindfulness skills mediated between the mindfulness treatment and the academic outcomes. This study, along with other ones (Kong & Jolly, 2019; A. J. Mrazek et al., 2018), suggest that not only practicing mindfulness can be beneficial, but inducing students’ beliefs about the malleability of their mindfulness can also make a difference.

It should be noted that while there was a positive effect of the mindfulness induction on task persistence among participants with high dispositional mindfulness, we did not find that the induction was successful. That being said, we applied a single question asking participants about how much they could get into a calm and focused state, which might not be an optimal assessment. In contrast, Hafenbrack and colleagues (2014) asked three questions regarding the degree to which participants were aware of their breathing, the sensation of breathing, and their body in general or posed a question about how much they were “absorbed in the present moment”.

As secondary hypotheses, we also proposed that growth mindset and mindfulness inductions would beneficially influence goal-setting (Cury et al., 2006; McCarthy, 2011). As people change their goals from fulfilling expectations to self-improvement, they would experience decreased anxiety (Schleider & Weisz, 2018; Zenner et al., 2014) and less boredom (Karumbaiah et al., 2017), ultimately resulting in better cognitive performance (Blackwell et al., 2007; M. D. Mrazek et al., 2013). However, we found no such effects. The explanation for the absence of these effects might also be related to the fact that these studies built on the effects of mindset and mindfulness in more elaborate interventions (e.g., Orosz et al., 2020) and for more specific groups of people (Broda et al., 2018; Orosz et al., 2017, 2020; Yeager, Walton, et al., 2016). Being exposed to a stressful self-threatening situation, and being mindful of the emerging negative experiences, may not affect those who are not familiar with practicing it. Furthermore, providing a persuasive article about the malleable nature of intelligence for those people who have opposing thoughts, may not be helpful (but see Blackwell et al., 2007; Török et al., 2022 for contradicting results; Yeager et al., 2014).

To sum up, some previous studies showed that students with a growth mindset may still be judgmental towards themselves, ruminate about failures, and are threatened by ego-relevant failure (Niiya et al., 2010; Orosz et al., 2020). Thus, we expected that mindfulness meditation (as a complementary strategy) could help them to observe the flow of the resulting negative thoughts and feelings from an optimal distance with acceptance and without self-judgment. Therefore, as the result of mindfulness induction, we expected higher task persistence after negative feedback, a more positive attitude toward the task, and decreased anxiety. We did not find these effects in our study. As we discussed above, the mindset manipulation might not have been effective enough for those who did not have a pre-existing growth mindset belief to make a difference in the behavioral task persistence measure. It might be also true for the mindfulness manipulation in which the induction check did not provide any support for the success of our manipulation. Instead, participants characterized by high dispositional mindfulness attempted to complete more optional tasks in the mindfulness condition compared to their low mindfulness counterparts, while this pattern was not found in the control condition. Overall, our results suggest that applying more elaborate growth mindset interventions and mindfulness meditations and contextualizing why the exercises are necessary might be helpful in future studies.

One of the strengths of the study was that it used an experimental design, reducing potential biases that might limit causal conclusions. The methodology and data analysis plan were pre-registered, ensuring the confirmatory nature of the study and reducing the researcher’s degrees of freedom (Wicherts et al., 2016). We also used a relatively large offline sample that may reduce the noise in the data due to the controlled environment. We also carefully considered if participants were aware of the false-negative feedback manipulation, and excluded those who reported that they did not believe that the feedback was based on actual performances. An additional strength of the study is that we used a behavioral measure of task persistence, increasing the ecological validity of the findings.

On the other hand, the study also has several limitations. We found that the group that received the mindfulness induction did not report higher levels of relaxation and focus than the control group. This might mean that the induction could not produce the desired effect. As discussed previously, such brief mindfulness induction alone may not be enough to facilitate a mindful state. Alternatively, we may have been unable to measure the change using the induction check question. The main reasons for using a single item were (1) not to attract too much attention to the manipulation, (2) not to break the experimental flow, and (3) to avoid reducing the effects of mindfulness meditation by asking participants to describe their current mental states.

Moreover, in our pre-registration, we used certain assumptions to calculate the required sample size, and some of these assumptions (i.e., the distribution of the outcome variable) were not met in the final dataset. This means that the study might not have had the adequate statistical power to find a significant difference between the groups. In other words, the study is only able to detect a larger difference than was originally planned. The initial preregistration aimed for approximately a hundred participants per cell to achieve 95% power to detect the effect of one optional task in any treatment group. Using a zero-inflated censored negative binomial distribution — that we got in the study — we could achieve 80% power with a difference of three optional tasks. However, given the low Bayes Factor value for the model, it is very unlikely that more data would lead to a different conclusion.

The validity of the behavioral task may also have some limitations. Our behavioral measure of task persistence was the number of optional tasks that participants completed as practice, before the “real” IQ test. A recently developed and thoroughly validated instrument — the PERC task — uses a very similar approach to measuring mastery behaviors (i.e., persistence, effort, resilience, and challenge-seeking). Therefore, although our assessment of task persistence may be imperfect, it is very similar to an extensively validated instrument. A potential criticism of this assessment is that it could have been more sensitive, such as in studies that use the time spent on effortful activities as a proxy of effort (e.g., Galla et al., 2014; Porter et al., 2020). We used a 45-second time limit for all IQ tasks, thus, we could not use time as a metric of task persistence. To allow students to practice the IQ tasks as much as they would like to, the time limit could be dropped during the optional task block. By this approach, we could prospectively get a more sensitive proxy of the actual willingness students would invest in learning (which might be especially important for more difficult items).

One might suppose that the participants found the task too boring to make an effort. The experiment might have been relatively long and demanding, with 36 obligatory and 17 optional IQ tasks, in addition to the inductions and questionnaires. In the pretests of the research protocol, we found that fewer IQ tasks made the cover story of the study less believable. Admittedly, we tried to find a balance where the research protocol was long enough to make the cover story credible while keeping it brief enough to maintain participant engagement. We investigated this option by looking at the items on the task-related boredom scale. We found that the averages of the two boredom items were below (Ms = 2.74, 2.59), while the excitement item was above (M = 3.50) the scale midpoint (3.0). Thus, it does not seem plausible that the task was too boring for participants. Based on these item scores we believe that the research protocol could maintain participant interest in general. Although there was a weak negative correlation between task boredom scores and the number of completed tasks, r(383) = -.14, p = .003, meaning that participants, who found the IQ tasks more boring, also completed fewer tasks. However, this variability was not associated with mindset or mindfulness inductions.

Another shortcoming is that we did not use a true control condition for the mindset induction, thus we were not able to investigate if exposure to negative feedback resulted in lower levels of additional task completion. Future mindset studies may consider adding a condition that involved no manipulation of mindset beliefs.

Lastly, it is also possible that the feedback we gave (“Your IQ is below average”) was rather person-oriented which could have undermined the effect of the mindset manipulation. For this reason, in future studies, it might be better to use more task-oriented feedback such as “Your IQ test performance was below average”.

In a pre-registered experiment, we found that a brief growth mindset induction and mindfulness meditation might not be sufficient to mitigate the short-term demotivating effects of negative academic feedback. However, the inductions seemed to effectively boost task persistence for participants with higher levels of pre-induction growth mindset or dispositional mindfulness. This suggests that in order to help a wide range of students to overcome stressful academic situations, such as feedback on an unsuccessful exam, we might need to aim for more pervasive changes instead of one-time inductions. Possible directions seem to be shifting students’ mindsets with wise social interventions (e.g., Yeager et al., 2019) and altering dispositional mindfulness by applying longer interventions (e.g., Quaglia et al., 2016).

Substantial contributions to conception and design: TN, KS, LT, BB, GO

Acquisition of data: TN, KS, LT

Analysis and interpretation of data: TN, KS, LT, ZKT

Drafting the article or revising it critically for important intellectual content: TN, KS, LT, BB, ZKT, GO

Final approval of the version to be published: TN, KS, LT, BB, ZKT, GO

We appreciate the contribution of research assistants in the data collection process: Adrienn Csiba, Eszter Csernák, Zsófia Gergó, Markus Lien, Mónika Lőrincz, Márton Mezőbándi-Nemes, Fanni Mészáros, Emese Misák, Júlia Molnár, Marietta Molnár, Dóra Kántor, Orsolya Szőke, Virág Teodóra Fodor.

TN received financial support from the Hungarian National Research, Development and Innovation Office (Grant No.: FK124225, PD131954), GO from the Young Researcher STARS, DSG1, DSG2, and CPJ grants from Conseil Régional Hauts de France. BB was supported by the Merit Scholarship Program for Foreign Students (PBEEE) awarded by the Ministère de l’Éducation et de l’Enseignement Supérieur (MEES) during the finalization of the paper.

The authors declare no competing interests.

All stimuli, presentation materials, participant data, and analysis scripts can be found on this paper’s project page on OSF: https://osf.io/gtbyc/