Investigating the negative link between perfectionism and emotional divergent thinking

Abstract

Previous research has shown that perfectionism was negatively associated with the generation of original ideas in Divergent Thinking (DT) tasks, while striving for excellence was positively associated with it. However, the explanatory variables for these effects remain unclear. This study investigated the mediating roles of doubts about actions, concerns over mistakes, openness to experience, empathy, and emotions during DT tasks. Additionally, it examined an emotional DT task (i.e., naming frustrating things and things that affect one’s self-esteem). From a sample of n = 282 university students, we replicated the negative association between perfectionism and DT abilities, though the effect size was smaller than in prior studies. Perfectionism correlated with lower empathy and greater primary negative emotions (e.g., fear) but similar openness to experience compared to excellencism. Mediation analyses revealed that doubts and concerns were unrelated to DT abilities. Openness to experience and empathy were positively correlated with DT abilities. Primary negative emotions during the tasks were negatively associated with the originality of answers. In contrast, positive emotions and secondary negative emotions (e.g., embarrassment) predicted more original ideas. These findings emphasize the importance of promoting excellencism over perfectionism to foster original ideas. This study has implications for overcoming barriers and supporting the creative process of individuals high on perfectionism. It also has implications for creativity researchers investigating the role of empathy and emotions in DT abilities.

Supplementary Information

The online version contains supplementary material available at 10.1007/s44202-025-00330-x.

Introduction

Surprisingly few studies have investigated the link between perfectionism and the generation of creative ideas. On the one hand, perfectionism is widely legitimized in modern workplaces and many individuals are encouraged to strive for perfection [1, 2]. On the other hand, creative abilities are targeted by most enterprises, as well as international and national task forces concerned with the set of skills needed for todays and tomorrow’s world [3]. It remains unclear if perfectionism is detrimental or beneficial to creative thinking.

Perfectionism can have a dual impact on creative thinking, potentially bringing both benefits and challenges. Setting very high standards for oneself can have motivating effects, but it can also paralyze. Perfectionism has been associated with strategic goal pursuit and high self-efficacy—i.e., the belief that one is able to perform and meet a goal [1, 4]. It drives high-quality work through persistent efforts. However, standards of perfection are often accompanied by a fear of failure, which stifles the creative process of individuals, making them hesitant to take risks or experiment with new ideas and activities. Aiming for perfect outcomes is a weight which can result in procrastination, stress, anxiety, and burnout [4, 5], and overall hinders creative thinking and productivity [6]. Despite that meta-analyses have associated perfectionism with achievements in school [7] and in sports [8, 9], studies have also repeatedly shown that perfectionism was a risk factor for psychological health and interpersonal relationships [4–6, 10]. Why, how, and when perfectionism has positive or negative impacts on individuals and their creative abilities still requires more research to elucidate [11].

Setting excessively high standards of perfection is not the only way to be successful. The Model of Excellencism and Perfectionism (MEP) defines the concept of excellencism as a tendency to “aim and strive toward high yet attainable standards in an effortful, engaged, and determined yet flexible manner” ([11], p. 200). Striving towards excellence has been associated with better academic achievement than striving towards perfection, as well as protective effects against harmful academic outcomes [6, 10]. People who strive towards excellence attain excellent results and are satisfied by it (i.e., “when it's good enough, it's good enough”). In contrast, people who strive for perfection are not satisfied with excellence. They feel like it is not good enough, and do not consider the job done [11]. Their effort continues until perfection is met and this quest is often never ending. Logically, perfection goes beyond excellence. This additional effort can be (a) beneficial, (b) unneeded, or (c) harmful, depending on the outcome and the circumstances.

Perfectionism, excellence, and creative ideas

An important part of the creative process involves generating original ideas and exploring new possibilities [12–14]. Researchers have hypothesized that the rigid thinking associated with perfectionistic strivings can hamper this stage [15, 16]. In contrast, striving towards excellence suggests a more flexible attitude towards life. In a recent study [17], the authors observed that perfectionism hindered divergent thinking and associative abilities, as well as openness to experience—compared to excellencism—but was neutral with respect to general self-efficacy, creative self-efficacy, and creative personal identity. These results suggest that perfectionism is counter-productive towards divergent and associative modes of thinking. The goal of the current study was to explore reasons why perfectionism might be detrimental to the generation of original ideas in divergent thinking tasks, and why excellencism might be beneficial.

Generating creative ideas—i.e., ideas which are both original and appropriate [18]—is one aspect of the creative process, along with problem framing, gathering knowledge, evaluating ideas, and implementing solutions. The idea generation phase is largely encompassed by Divergent Thinking (DT) abilities. DT is an umbrella term which regroups many cognitive processes involved in the generation of variations and alternatives [19], including the retrieval of relevant information from long-term memory, the inhibition of dominant responses, the capacity to shift perspective, and the maintenance of goal-oriented search strategies [20, 21]. DT is an important mode of thinking used to generate variations, but it does not guarantee a “hit”. Finding original ideas is also highly dependent upon the problem and the solution-space, the strategy used to search that space, and the subjective value of ideas. DT abilities can be assessed with open-ended questions such as “what would happen as consequences if glasses and corrective lenses had never been invented” or through visual tasks requesting to design an image using available shapes. The number of responses given is often used as a proxy for one’s fluency of thoughts, whereas the extent to which a response is surprising, unconventional, remote, or clever can be used as a proxy for the originality and creative quality of answers. The set of responses is often evaluated subjectively by experts or qualified individuals, or more recently by Large Language Models [22]. Importantly, there exists a vast variety of ways to measure aspects of DT. In fact, different DT tasks only weakly correlate to one another [23]. From verbal to visual, musical, and scientific, DT in one domain barely predicts an individual's propensity to generate original ideas in another domain. This underscores the importance of using multiple tasks to either assess a 'general' component of DT or to employ specific tasks while keeping their scores distinct.

The current study used two classic DT items, namely the generation of original uses for a brick and a newspaper, and two emotionally oriented items (e.g., name all the things which can be frustrating [24]). The two types of tasks are expected to behave differently in relation to perfectionism and excellencism. A re-analysis of an open dataset which used those DT tasks [24] revealed that perfectionism was more detrimental to the emotionally oriented items compared to the classic DT items. The results of this analysis are reported in Table 1. We statistically tested the difference by introducing a new parameter in the model which represented the difference between the two unstandardized coefficients. Perfectionism predicted significantly less original answers in the emotional task compared to the classic task (non-standardized ∆_parameter= − 0.083, SE = 0.03, p = 0.006), whereas the fluency score was equivalent across both tasks (non-standardized ∆_parameter = − 0.33, SE = 0.21, p = 0.11). For excellencism, the difference between the task was non-significantly different from zero for the scores of originality and fluency (respectively non-standardized ∆_parameter = 0.063, SE = 0.04, p = 0.16, and ∆_parameter = 0.57, SE = 0.29, p = 0.05). This suggests that participants high on perfectionism generated an equivalent number of ideas in the emotional DT task as they did in the classic DT task, but their ideas were less original in the emotional task. However, in both the emotional and classic DT task, there was a significant negative correlation between perfectionism and originality as well as fluency. These creativity metrics were positively correlated with excellencism.

Table 1.

Reanalysis of results from [24], showing standardized regression coefficients from a multivariate multiple regression with perfectionism and excellencism predicting divergent thinking scores

Variables	DT classic		DT emotional
	Originality	Fluency	Originality	Fluency
Excellencism	0.30**	0.21**	0.42**	0.25**
Perfectionism	− 0.24**	− 0.20**	− 0.48**	− 0.23**
R2	7.2%	4%	18.6%**	5.5%

DT  divergent thinking

n = 184. **p < 0.01

Why perfectionism might be detrimental to divergent thinking?

Why perfectionism may hinder the process of generating creative ideas? And why emotional ideas in particular? Genuine creative problem-solving situations is inherently a risky process. When the outcome is uncertain, proposing original solutions may come with insecurities and anxieties related to fear of failing, fear of being different, fear of criticism or ridicule, fear of rejection, and fear of consequences [25, 26]. These emotions and the evaluative concerns which often accompany perfectionistic standards [27, 28] may hinder the capacity to let go, improvise, play, and fail, which are an integral part of a divergent mode of thinking.

The concept of emotional DT overlaps with the emerging literature on emotion regulation flexibility which corresponds to “the ability to choose how we address life’s inevitable ups and downs, its disappointments, triumphs, and challenges” ([29], p. 29). Being able to display emotional flexibility involves reinterpreting the situation and the associated emotions. This flexibility is a form of creative ability. Perfectionism has been associated with greater ruminations [30], rigid emotional responses to goal progress, as well as patterns of social disconnection, and dichotomous, black and white, thinking [31]. Those tendencies may be the consequence of a reduced capacity to explore different emotions to a particular situation (i.e., reduced emotional DT).

According to the dual-pathway theory of creativity [32], generating creative ideas can be achieved through a cognitive flexibility pathway or a cognitive persistence pathway. The flexibility pathway is best represented by the use of broad and inclusive cognitive categories, a propensity to switch between different approaches, a defocused attention, the consideration of different perspectives, and the instantiation of associative thinking through a spreading of activation within memory and semantic networks. In contrast, the persistence pathway represents the possibility to reach creative ideas through hard work, a systematic and effortful exploration of possibilities, and an in-depth exploration of only a few categories or perspectives. The persistence pathway implies a greater degree of sustained attention, task-directed effort, systematic search, and analytical processing. Both pathways are conducive of creative ideas in their respective way.

Different states and traits can influence creative thinking through their impact on one or both pathways [32, 33]. The trade-off between flexibility and stability may suggest that the two pathways are negatively related. However, this is not necessarily the case. People can switch from more flexible to more systematic processing modes while solving a problem. For instance, a person can exploit a promising category of solutions up to a point where the return on investment is too low, in which case the individual switches to another category of solution. All else being equal, relying on both pathways could be more optimal than relying on only one of the two. Aiming for perfection might promote the generation of creative ideas through the persistence pathway, while potentially negatively affecting the flexibility pathway, in which case the relation between perfectionism and creativity becomes difficult to predict and will depend on other factors [32]. Excellencism might contribute to creative thinking through a greater focus and persistence compared to individuals with lower standards of performance. Likewise, excellencism might give an edge over perfectionism on creative tasks through the flexibility pathway.

Five explanatory mechanisms

This study explores five potential explanatory mechanisms of the associations between perfectionism, excellencism, and DT abilities.

Doubts about actions and concerns over mistakes Doubts about actions refers to skepticism that one can do things “right”. Concerns over mistakes refers to a tendency to react negatively to mistakes, to interpret mistakes as synonymous with failure, and to hold the belief that failure will result in losing others' respect [28]. Both are recurrent experiences associated with perfectionism. These avoidance goals can negatively impact the exploration of a space of solutions and risk-taking tendencies, and hence might have negative effects on cognitive flexibility. The current study is the first to our knowledge to assess the relationships between doubts about actions, concerns over mistakes, perfectionism, and DT abilities.

Openness to experience Open individuals are curious, have broad interests, they enjoy intellectual challenges, and are sensible to their emotions and to art [34, 35]. Perfectionism has been found to be negatively associated with openness to experience [17], while openness to experience is a robust positive predictor of various creative outcomes, including DT [34, 36–38]. A strong desire to do things perfectly from the start and to avoid failing may narrow one’s perspectives and propensity to try new activities. Perfectionism can diminish cognitive flexibility through lower openness to new activities and experiences. Thus, we hypothesized that a lower openness to experience would account for the negative association between perfectionism and DT.

Empathy A third potential explanatory variable is empathy. The association between empathy and DT is receiving increasing empirical attention. The cognitive mechanisms responsible for empathy have been theorized to overlap with those underlying DT [39]. Empathy entails the flexible exploration and simulation of someone else feelings. Researchers have suggested and observed that empathy was partly an imaginative and constructive process, relying on many of the same imagination abilities involved in creative thought [40, 41]. The capacity to take someone else’s perspective can help producing new ideas and more useful ideas to others [42, 43]. Engaging in perspective taking can be seen as a form of DT ability as it requires inhibiting one's own perspective and generating alternatives emotions to a situation. Empathy has been shown to be directly associated with creativity [42–44]. A recent study by Kripal and Reiter-Pamon [44] found that empathy was moderately associated with problem construction and the originality of solutions for ambiguous relationship problems.

Perfectionism has been associated with more self-reported empathy—not less [45]. Nonetheless, this literature has not distinguished between perfectionism and excellencism. Therefore, positive results may be in part attributed to a positive association between excellencism and empathy (this pattern has been found to be the case for academic grades and creative indicators [6, 17]). More research is needed to disentangle the relationship between excellencism, perfectionism, empathy, and DT.

Emotions during the DT tasks The emotions that individuals feel while completing a task can impact their internal search process. Research repeatedly showed that positive emotions lead to broader visual search patterns, more creative thoughts, more inclusive social interactions, and greater flexibility in goals and mindsets [46–49]. People’s moods interact with the task. Positive moods can signal that a situation is safe and with no consequences, which promotes the exploration of possibilities in an unconstrained and flexible way [32]. Negative moods can signal that a situation is problematic and that specific actions must be undertaken to address it, which promotes a systematic, constrained and analytical approach. In the current study, we investigated positive (e.g., joy), negative (e.g., angry), and self-conscious emotions (e.g., embarrassment; [50]), also called secondary negative emotions [49].

This study

This study investigated the differences between perfectionism and excellencism on classic and emotional DT tasks. By identifying explanatory mechanisms, we aim to contribute to an understanding of what in perfectionism and in excellencism impact their creative potential. We hypothesized that:

H1: Individuals high on perfectionism would have more doubts and more concerns over mistakes compared to individuals high only on excellencism; in turn, doubts and concerns would be negatively related to DT abilities. The effect is expected to be equivalent for both DT tasks, classic and emotional.

H2: Individuals high on perfectionism would be less open to experience than individuals high only on excellencism; in turn openness to experience would be positively related to DT abilities. Openness to experience should be more highly associated with emotional DT task considering that emotional awareness is an important component of that construct.

H3: Individuals high on perfectionism would be less empathic than individuals high only on excellencism; in turn, empathy would be positively related to DT abilities. The effect of empathy should be greater towards the emotional DT task, assuming that empathy is a mechanism of emotional exploration.

H4: Individuals high on perfectionism would experience fewer positive emotions, more primary negative emotions, and more secondary negative emotions during the DT tasks than individuals high only on excellencism; in turn, positive emotions would be positively related to DT abilities, whereas primary and secondary negative emotions could be positively or negatively related to DT abilities. The direction could depend on whether participants are cognitively stimulated by the emotion. We expect the effect of emotions to be greater towards the emotional DT task.

Method

Participants

A sample of 282 undergraduate students (74.8% female) from a French/English bilingual university in the province of Ontario was retained for analysis. The participants were recruited through an integrated system of participation in research designed for students where they receive course credits for a class to compensate their participation. From the initial 342 participants, 35 were removed due to missing data on more than 70% of the items and 18 participants were removed due to duplicate participation. The duplicate participants completed the study twice. This often happen when participants do not receive their credit instantaneously. They sometimes try again thinking there is an issue with their first completion. We retained only their first participation as valid. Finally, three participants were removed due to completing the study in less than 20 min. This cutoff was derived from pilot participants, whereby 20 min to complete the study represented a completion time 50% shorter than the average. This cutoff differs from the pre-registration which initially set the minimum to 30 min. The 30 min cutoff was deemed to strict after examining the pilot participants. 40 participants completed the study within 20 to 30 min, and were thus included in the study. Univariate outliers’ inspection revealed four extreme cases (whose z scores were beyond ± 3.5; [51]) which were removed from further analyses. Multivariate inspections (Mahalanobis distance, using a critical distance from a χ² (11), p < 0.001) revealed no additional extreme cases. The final sample was n = 282.

The age ranged from 17 to 49 years. Mean age was 19.4 (SD of 3.0). Participants identified as white (49.0%), Asian (20.3%), African/African American (11.5%), Arab (9.8%), Hispanic (1.7%), First Nations, Métis, or Inuit (0.3%) and others (7.3%). Participants reported that their first language was English (52.3%), French (27.7%) and others (20.0%). Their program of study included health sciences (26.0%), psychology (20.7%), management (15.1%), social sciences (9.5%), natural sciences (7.4%), arts (5.3%), computer science (1.4%) and others (14.7%). Participants reported coming from different social classes: lower (2.8%), lower average (9.6%), average (52.1%), upper average (33.0%) and upper class (2.5%).

Procedure and measures

The study was pre-registered (see OSF link). The pre-registration was uploaded during data-collection (13/11/21) but before data collection ended (8/12/21). The pre-registration included some analyses which we did not include in the current study. More specifically, one associative thinking task and some exploratory questions on the strategy used by participants during the DT tasks were completed but not included in this study. The study was completed online. Participants answered socio-demographic questions first, then completed questionnaires interlaced with four cognitive tasks measuring divergent thinking. The present study was approved by the University of Ottawa's Research Ethics Board and performed in accordance with the University of Ottawa's Research Ethics Board guidelines. All participants provided informed consent prior to their participation in this study.

The measures are described below. The independent variables are the scores of Excellencism and Perfectionism. The mediating variables are: Doubts about actions, Concern over mistakes, Openness to experience, Empathy, and the Emotions felt during the DT tasks. Finally, the dependent variables are the originality of answers in classic DT and the originality of answers in emotional DT. The originality scores correspond to the creativity of answers according to independent judges. Fluency scores (i.e., the number of generated answers to the DT tasks) were coded but not maintained in the models for parsimony and because those scores are less revelatory of creative abilities [52].

Measurements

Perfectionism and excellencism The long version of the Scale of Perfectionism and Excellencism (SCOPE; 22 items, [6]) was used to measure the constructs of perfectionism and excellencism. Perfectionism is defined as “the tendency to aim and strive toward idealized, flawless, and excessively high standards in a relentless manner’ [11] (p. 200). Excellencism is defined as a “tendency to aim and strive toward very high yet attainable standards in an effortful, engaged, and determined yet flexible manner” [11] (p. 200). Items were preceded by the stem: “My goal in life is…”. A representative item of perfectionism is “…to be exceptionally productive all the time”. A representative item of excellencism is "…to be very productive". Participants rated their answers on a scale ranging from 1 (Not at all) to 7 (Totally). The scale has been validated across five studies (total N = 2157; [6]). The validation studies suggested excellent composite reliability (Omega, ω) for Excellencism ω = 0.956 95%CI [0.948, 0.963] and Perfectionism ω = 0.973, [0.968, 0.977], as well as factorial validity for the two-factor solution and construct validity within a nomological network composed of Need for Achievement, Fear of Failure, Self-Compassion, Satisfaction and Frustration of the Need for Competence, and other measurements of perfectionism.

Doubts about actions and concern over mistakes The constructs of Doubts About Actions and Concerns Over Mistakes were assessed with the subscales of the Multidimensional Perfectionism Scale (MPS; [28]). Both subscales Doubts About Actions (4 items) and Concerns Over Mistakes (9 items) showed good internal consistency with coefficients of 0.72 and 0.90 respectively in a validation study [53]. Moderate to high correlations with measures of dysfunctional beliefs, anxiety, and depression supported the validity of the scale total score [28]. An example item for Doubts about actions is: “Even when I do something very carefully, I often feel that it is not quite right.” An example item for Concern over mistakes is “I should be upset if I make a mistake”. Participants rated their answers on a scale ranging from 1 (Strongly disagree) to 5 (Strongly agree).

Openness to experience Openness to experience was assessed with the Big Five Aspects Scale (20 items; [54]). Openness to experience captures intellectual curiosity, interest in arts, emotional awareness, wild imagination, as well as non-conformity. The subscale of Openness to experience had good internal consistency in a validation study (α = 0.78; [54]). The Openness to experience factor correlated well with the same factor in other established measures of personality such as the NEO-PI-R [54]. An example item is "I see beauty in things that others might not notice." Participants responded on a 5-point scale ranging from 1 (Highly disagree) to 5 (Highly agree).

Empathy Empathy was assessed with the Interpersonal Reactivity Index questionnaire (IRI, 28 items; [55]). The 28-item IRI is a self-report measure consisting of four 7-item subscales: Perspective Taking (tendency to adopt the psychological point of view of others); Fantasy (tendency to transpose oneself imaginatively into the feelings and actions of fictitious characters in books, movies, and plays); Empathic Concern (feelings of sympathy and concern for others); and Personal Distress ("self-oriented" feelings of personal anxiety and unease in tense interpersonal settings). The IRI correlates moderately with other empathy measures, such as the Empathy Quotient and social skills assessments (0.11 < r < 0.63), indicating that it effectively captures an empathy construct [55]. The IRI is based on a multidimensional model of empathy which considers each subscale to be a unique dimension of empathy. However, an evaluation of the structure of the empathy construct led Cliffordson to observe that a general factor virtually identical to the Empathic Concern subscale explained a great deal of the variance in the Perspective Taking and Fantasy subscales [56]. The personal distress subscale behaved differently than the others, showing weak correlations with the other three. Cliffordson concluded that the three former subscales could be averaged into a total score and interpreted as a measure of a general empathic disposition encompassing both, a cognitive and an emotional component. Based on these findings, we computed a general score by averaging the Perspective Taking, Fantasy, and Empathic concerns subscales. An example item of the empathic concern subscale is: “When I see someone being abused, I feel a little protective towards them.” An example item of the perspective taking subscale is: “Before criticizing somebody, I try to imagine how I would feel if I were in their place.” An example item of the fantasy subscale is: “I really get involved with the feelings of the characters in a novel.” Participants rated their answers on a 4-point scale ranging from 1 (Doesn't describe me) to 4 (Describes me totally).

Emotions felt during the divergent thinking tasks A list of twenty-three triads of emotions were presented to the participants, preceded by the stem “During the creative tasks, I felt…”. Inspired by the initial Differential Emotions Questionnaire [57], Fredrickson and colleagues [58] added eight discrete positive emotions to the eleven emotions which appeared in the original scale, namely amusement, awe, contentment, gratitude, hope, love, pride, and sexual desire. We further added three items inspired by the literature on perfectionism: “I felt Analytical, judgmental, critical”, “I felt embarrassed, inadequate, ashamed”, and “I felt Anxious, panicky, nervous”. Participants rated their answers on a scale from 1 (Not at all) to 5 (Extremely). The list of items used is reported in Appendix A.

The items can be divided in two subscales, called Positive and Negative emotions [57] with reported reliability coefficients of α = 0.79 and α = 0.69 respectively. Galanakis and colleagues [49] further examined the psychometric qualities of the scale in a sample of 11,422 Greek adults aging from 18 to 83 years old. Exploratory Factor Analysis revealed a three-factor solution, called Positive emotions, Primary Negative emotions, and Secondary Negative emotions. In addition, they observed problematic loadings for four items, namely (1) Sympathy, concern, compassion, (2) Surprised, amazed, astonished, (3) Disgust, distaste, revulsion, and (4) Awe, wonder, amazement. They combined the Primary and Secondary Negative emotions subscales and found positive correlations with measurements of stress, anxiety and depression, as well as negative correlations with life satisfaction, psychological resilience, optimism, inspiration, hope and subjective happiness. The Positive emotions subscale was negatively correlated to stress, anxiety and depression and positively correlated to life satisfaction, psychological resilience, optimism, inspiration, hope and subjective happiness [49].

Classic divergent thinking Participants were instructed to generate as many original answers as possible to the following two traditional divergent thinking items (1) Tell us all the different ways you could use a brick, and (2) Tell us all the different ways you could use a newspaper [13]. The tasks were presented as “creative games”. Participants were given three minutes per items and explicitly asked to be creative. The originality of answers was evaluated by two trained graduate students based on the snapshot scoring methods [59]. This method consists of providing a single holistic rating of originality to the entire set of responses of each participant, and then to average the score across the judges. The method evidenced good reliability with two judges [59]. The originality of answers was anchored on a scale from 1 (not creative at all) to 4 (very creative) and assessed according to the guidelines found in [59], whereby answers that are uncommon, remote, or clever are given higher scores. The Intra-Class Correlation for the originality score in the current study was 0.85.

Despite their widespread use, classic DT tasks face criticisms for lacking construct validity, failing to capture all aspects of the creative process, neglecting domain specificity and expertise, and exhibiting questionable predictive, ecological, and discriminant validities [60]. For example, there is little relation between various subtests of DT [23]. Each subtest may evaluate one aspect of creative thinking, but neither a single subtest nor DT as a whole take into account the complete creative process. However, DT scores have been found to be predictive of creative achievements [61–63]. Overall, DT test scores can be interpreted as indicators of creative ideation potential, but not as a measure of the whole creativity phenomenon [52].

Emotional divergent thinking Participants were instructed to generate as many original answers as possible to the following two items (1) Name all the things you can think of that can be frustrating, and (2) Name all the things you can think of that can positively or negatively affect your self-esteem [24]. These emotional items were exploratory; their specificity to an emotional domain has not been demonstrated, but these items were added to replicate the results where perfectionism was more negatively associated with these emotional items compared to the classic ones. The tasks were presented as “creative games”. Participants were given three minutes per items and explicitly asked to be creative. An originality score was computed the same way as the classic DT. The Intra-Class Correlation for the originality score in the current study was 0.85.

Results

Descriptive statistics

The dataset is available online (see OSF link). Following data screening, we computed averages. Descriptive statistics are given in Table 2. We noted that skewness was below one for all variables. Table S1 (supplementary file) shows the correlations between the various measurements (see OSF link).

Table 2.

Descriptive statistics for the main variable of this study

Measure	Min	Max	Mean	SD	Skew	α
Excellencism*	2.91	7.00	5.67	0.89	− 0.306	0.93
Perfectionism*	1.00	7.00	3.92	1.58	0.012	0.97
Concerns over Mistakes**	1.00	5.00	3.31	1.00	− 0.413	0.80
Doubts about Actions**	1.00	5.00	3.00	0.94	− 0.126	0.90
Openness to experience**	2.50	4.65	3.61	0.41	− 0.025	0.73
Empathy**	2.00	3.64	2.85	0.34	− 0.010	0.79
DT Originality Classic***	0.88	3.25	1.75	0.47	0.486	0.85
DT Fluency Classic	2.00	24.50	10.14	4.10	0.793	0.87
DT Originality Emotional***	1.00	3.13	1.89	0.45	0.476	0.85
DT Fluency Emotional	3.00	31.00	13.48	5.51	0.611	0.71

*On a scale from 1 to 7; **: on a scale from 1 to 5; ***: on a scale from 1 to 4

Analytic approach

Data were analyzed using Mplus 8 [64]. Given their conceptual similarity, we combined the explanatory variables of Doubts about actions and Concerns over mistakes in a single mediation model. We combined the explanatory variables of Openness to experience and Empathy in one mediation model. These two models were fully saturated (all paths were estimated), therefore no fit indices were calculated. We assessed the significance of the indirect (mediation) effect with a bias-corrected bootstrapping procedure involving 5000 bootstrapped samples and the Maximum Likelihood (ML) estimator.

The emotions felt during the tasks were examined using structural equation models. The use of latent factors was particularly appropriate for this construct considering that it involves many differentiated emotions which are expected to be explained by two or three higher-order factors. We tested from one to three-factor solutions with Confirmatory Factor Analysis (CFA) model where all cross-loadings and correlated residuals were fixed to zero. We assessed the fit of the measurement models first, before adding the regression paths. This procedure allowed to investigate potential problems with the latent constructs without the influence of their associations with the outcomes [65]. We used conventional cut-off of goodness-of-fit indices (i.e., χ² closer to zero, BIC closer to zero, RMSEA < 0.10, CFI > 0.90, TLI > 0.90, SRMR < 0.05, primary loadings > 0.40 [66, 67]) to assess the validity of the latent constructs. These criteria have been found to be more appropriate with small sample sizes (n < 500) compared to more stringent criteria (e.g., CIF > 0.95, RMSEA ≤ 0.06, and SRMR ≤ 0.08; [67]). Although indices of goodness-of-fit are widely used, these indices should always be complemented with the comparison of alternative models [65]. We contrasted from one to three-factor solutions.

Perfectionism, excellencism and divergent thinking

The results of a multivariate multiple regression are reported in Table 3. As seen, Perfectionism was associated negatively with the emotional originality score (β = − 0.19 95%CI [− 0.31, − 0.05]) more so than the classic originality score (β = − 0.11 [− 0.25, 0.03]). Excellencism was positively associated with the emotional originality score (β = 0.18 [0.04, 0.31]) more so than the classic originality score (β = 0.14 [0.004, 0.26]). The effect sizes were smaller compared to the original study (compare Table 3 with Table 1). The fluency scores on the other hand were nearly identical to what was found in the original study.

Table 3.

Standardized regression coefficient from a multivariate multiple regression with perfectionism and excellencism predicting divergent thinking

Variables	DT classic		DT emotional
	Originality	Fluency	Originality	Fluency
Excellencism	0.14*	0.21**	0.18**	0.15*
Perfectionism	− 0.11	− 0.22**	− 0.19**	− 0.23**
R2	1.7%	4.9%	3.5%	4.3%

n = 282. *: p < 0.05, **: p < 0.01

We tested the difference between the originality scores of perfectionism on the emotional and classic tasks by reparameterizing the model to introduce a parameter representing the difference between the two coefficients [68]. The difference was not significant (non-standardized ∆_parameter = − 0.021, SE = 0.02, p = 0.32). The difference for excellencism was also not significant (non-standardized ∆_parameter = 0.02, SE = 0.03, p = 0.64). This suggests that participants high on perfectionism generated ideas that were as original in the emotional task and the classic task. However, in both the emotional and classic DT task, there was a significant negative correlation between perfectionism and originality.

The directions of the effects were consistent with the hypotheses, despite that one association did not reach statistical significance. More specifically, perfectionism was not significantly associated with the originality score of classic DT (β = − 0.11, p = 0.14).

Explanatory mechanisms

Doubts about actions and concerns over mistakes We tested whether Doubts about actions and Concerns over mistakes explained the difference between perfectionism and excellencism in their capacity to generate original answers. The results of a mediation analysis are presented in Fig. 1. The results of the indirect effects (mediation effects) are reported in Table S2 (supplementary file).

Fig. 1.

Doubts about actions and concerns over mistakes mediating the relationship between excellencism and perfectionism towards the originality scores of DT. *: p < 0.05, **: p < 0.01. n = 282

As can be seen in Fig. 1, none of the mediators were significantly associated to the originality DT scores. Figure 1 shows that perfectionism was significantly and positively associated to Doubts about actions (β = 0.41, p < 0.001), as well as Concerns over mistakes (β = 0.20, p = 0.003). Excellencism was not significantly associated to neither Doubts about actions, nor Concerns over mistakes. Hence, Hypothesis 1 was not supported by the data.

Openness to experience and empathy The results of mediation analyses with Openness to experience and Empathy as explanatory variables are presented in Fig. 2. The results of the indirect effects (mediation effects) are reported in Table S3.

Fig. 2.

Double-mediation: openness to experience and empathy mediating the relationship between excellencism and perfectionism towards the originality scores of DT. *: p < 0.05, **: p < 0.01. n = 282

As can be seen in Fig. 2, perfectionism was not significantly associated with openness to experience (β = − 0.09, p = 0.17). In turn, openness to experience was positively associated with original ideas in both DT tasks. However, perfectionism was negatively associated with empathy (β = − 0.18, p = 0.006), whereas empathy was positively associated with original ideas in both DT tasks. Empathy significantly mediated the association between perfectionism and emotional DT.

Excellencism was positively and moderately associated with both Openness to experience (β = 0.34, p < 0.01) and Empathy (β = 0.30, p < 0.01). Openness to experience and Empathy both significantly mediated the relationship between excellencism and originality in both DT tasks. The mediations were complete (the total effects were no longer significant after controlling for the mediators).

It should be noted that whenever excellencism significantly predicts an outcome and perfectionism has a non-significant relationship to it (as seen with openness to experience), this suggests that perfectionism fails to contribute to the prediction of excellencism (see [69]) and that perfectionism predicts that outcome to the same extent, i.e., neither more nor less than excellencism. We provided a graphical representation of the association between excellencism and perfectionism towards Openness to experience and Empathy in Figures S1 and S2 (supplementary files), using the Shinny app designed by Gaudreau and colleagues [69].

We conclude that perfectionism and excellencism were not significantly different on openness to experience, therefore Hypothesis 2 was not supported. However, openness to experience explained the better DT abilities associated with excellencism compared to inviduals low on excellencism and low on perfectionism. Hypothesis 3 regarding empathy was partially supported. More specifically, perfectionism was associated with less empathy compared to excellencism; in turn, empathy was positively associated to DT abilities. However, the effect of empathy on DT was not greater for the emotional task.

Emotions felt during the task We investigate the role of 22 emotions1 felt during the DT tasks. We specified a latent Structural Equation Model to regroup the emotions. We tested from one to three factors. The two factors model had a latent factor labelled ‘Positive’ emotions (comprising the emotions 1 to 11 from Appendix A) and a second latent factor labelled ‘Negative’ emotions (comprising the emotions 12 to 22 from Appendix A). The three factors CFA model separated the negative emotions into “Primary Negative” and ‘Secondary Negative” emotions. The former consisted of primary emotions such as fear, angriness, and sadness (i.e., emotions 12 to 17 from Appendix A), the later consisted of self-conscious emotions such as shame, embarrassment, and self-judgments (i.e., emotions 18 to 22 from Appendix A). The triad “Repentant, guilty, blameworthy” was specified to fit on the Secondary Negative latent construct. However, modification indices indicated that restricting this item to the Primary Negative latent factor was better suited considering the data. Table 4 reports the fit indices of these three measurements. The three-factor model with post hoc modification (“Repentant, guilty, blameworthy” loading on the Primary Negative construct) had a better fit; hence we retained that model. The fit was not fully satisfying according to conventional criteria. However, the fit was decent considering the number of indicators (22) and the strict model imposed by a CFA (i.e., cross loadings and residual correlations fixed to zero). All primary loadings were higher than λ > 0.43.

Table 4.

Fit of the latent measurement models

	LL	χ2	df	BIC	SRMR	RMSEA	CFI	TLI
Measurement models
One factor	− 8343.62	1701.56**	209	17,059.61	0.206	0.159 [0.152, 0.166]	0.48	0.42
Two-factors	− 7736.79	660.00**	208	15,851.58	0.091	0.088 [0.080, 0.095]	0.84	0.82
Three-factors	− 7712.47	616.44**	206	15,814.24	0.092	0.084 [0.076, 0.092]	0.85	0.84
†Three-factors	− 7669.49	536.32**	206	15,728.26	0.086	0.075 [0.068, 0.083]	0.89	0.87
Regression model
†Three-factors with mediators and outcomes	− 8818.21	661.10**	282	18,172.40	0.079	0.069 [0.062, 0.076]	0.88	0.86

n = 282. **: p < 0.01 †item 15 loading on “primary negative emotion” construct instead of “secondary negative emotion”

The results are presented in Fig. 3. Perfectionism was significantly and positively associated with all three latent factors, namely Positive emotion (β = 0.16, p = 0.02), Primary Negative emotion (β = 0.29, p < 0.001), and Secondary Negative emotion (β = 0.22, p = 0.001). Excellencism was not significantly associated with Positive emotion (β = 0.04, p = 0.53), but it was significantly and negatively associated with Primary Negative emotion (β = − 0.25, p < 0.001) and negatively associated with Secondary Negative emotion (β = − 0.21, p = 0.002).

Fig. 3.

Structural equation model with excellencism and perfectionism predicting three latent emotions factors, in turn predicting the originality DT scores. *: p < 0.05, **: p < 0.01. n = 282

Positive emotion was significantly associated with classic DT originality (β = 0.24, p < 0.001) but not significantly associated with emotional DT originality (β = 0.08, p = 0.25). Primary Negative emotion was significantly and negatively associated with classic DT originality (β = − 0.43, p < 0.001) and emotional DT originality (β = −0.47, p < 0.001). Lastly, Secondary Negative emotion was positively associated with both tasks, namely classic DT originality (β = 0.29, p < 0.001) and emotional DT originality (β = 0.35, p < 0.001).

The mediation effects are reported in Table S4 (supplementary file). Overall, perfectionism, predicted more Positive emotions, more Primary Negative emotions, and more Secondary Negative emotions. Therefore, the effects counterbalanced themselves. Specifically, Positive and Secondary Negative emotions predicted better DT abilities, whereas Primary Negative emotions predicted worse DT abilities.

Excellencism predicted fewer Primary Negative emotions, which explained their higher originality on both DT tasks. However, half of this effect was counterbalanced by lower Secondary Negative emotions which predicted more original ideas in both tasks.

General discussion

This study investigated how the pursuit of perfection could hinder the generation of original ideas during divergent thinking tasks, and how the pursuit of excellence could contribute to this capacity. We compared an emotional and a classic DT task. We expected that the emotional rigidity associated with perfectionism would hinder their performance during the emotional task more so than the classic task. We partially replicated previous results [24] whereby perfectionism was negatively associated with the originality of ideas. However, participants high on perfectionism performed similarly in the emotional task compared to the classic task, contrary to what was found in [24].

We investigated five potential explanatory mechanisms to understand what in perfectionism and in excellencism may impact the creative potential of individuals. We tested the mediating role of Doubts about actions, Concerns over mistakes, Openness to experience, Empathy, and the Emotions experienced during the DT tasks. These mediators were assumed to influence the generation of original ideas by either activating (or inhibiting) the flexible pathway to creativity—where diverse and loose categories are explored in surface—or by activating (or inhibiting) the persistence pathway to creativity, which involves the exploitation of a few categories in more depth.

Of these explanatory variables, only doubts about actions and concerns over mistakes did not significantly explain any variance in the generation of original ideas.

Doubts about actions and concerns over mistakes

Our results showed that perfectionism was associated with more doubts about actions and concerns over mistakes than excellencism. This is consistent with the literature on perfectionism. Perfectionism has been repeatedly associated with constructs such as fear of failure, higher levels of stress, and performance anxiety [4, 5, 25, 70]. We hypothesized that doubts and concerns would hinder individuals’ creative thinking by limiting their explorative tendencies and accentuating the presence of irrelevant thoughts during the creative task. However, doubts and concerns did not predict less original ideas on DT tasks.

Despite their high standards, individuals who aim for excellence seem more inclined to consider failure as a normal part of the learning process [6]. It is plausible that creative and ambitious individuals also have high concerns and doubts, and therefore these constructs are not good indicators of exploratory behaviors. In that case, doubts and concerns would not hinder the cognitive flexibility pathway, but instead be a collateral of taking risks and exploring uncharted paths. For example, a study using interviews with recognized artists revealed a common theme across artistic disciplines. Most artists described going through a series of “crises” coupled by “constant self-doubts and a desire to start everything afresh” [71] (p. 5). The relationship between doubts, concerns, and creative thinking might be positive in some cases and negative in others. Perfectionism or emotion regulation could be moderators of that link. Future studies should continue to investigate the link between doubts, concerns, and DT as it seems to be an inherent part of the creative process.

Openness to experience and empathy

Perfectionism was associated with an equivalent level of openness to experience than excellencism. Both, excellencism and perfectionism were associated with more openness to experience than individuals with lower standards. This contradicts the results of [17], where a negative and significant association between perfectionism and openness to experience was observed. However, that previous study also observed no difference between perfectionism and excellencism on creative identity and creative self-efficacy. Our hypothesis that perfectionism would be associated with less original ideas because of lower openness to experience, thus less flexible cognition, was not supported by the data.

Perfectionism is not incompatible with a creative identity and creative self-efficacy. Standards of perfection are nourished in many disciplines including artistic domains. Nevertheless, the avoidance orientation and rigidity of thinking which often accompany perfectionistic strivings would suggest a negative association with openness to experience. It is plausible that individuals high on perfectionism overestimate their openness to experience to protect their self-image. This may occur because creativity is associated with productivity, fame, and extraordinary results. Perfectionists might have a bias towards these qualities and a desire to project an image of a perfect individual. To examine this question, objective measurements coupled with external evaluators to assess openness to experience could be used in the future.

Empathy was another potential explanatory mechanism with its effects assumed to go through the flexible pathway to creativity. The construct of empathy entails a tendency to flexibly adopt the psychological point of view of others as well as feelings of sympathy and concern for others. Due to this imaginative and constructive process, empathy is thought to rely on many of the same abilities involved in divergent thinking [40].

Perfectionism was associated with less empathy than excellencism, which partly explained the poorer performances on DT tasks. Empathy positively predicted more original ideas in both DT tasks over and above openness to experience. However, empathy did not predict the emotional DT task more than the classic one. Gallo [39] discussed the fact that creative individuals “possesses unusual perceptual and personal openness, and a marked capacity for empathic identification with the other” (p. 316). The trait Openness to experience contains an important emotional component, described by Deyoung [34] as an “awareness of both one’s own and others’ emotions” (p. 36). Moreover, openness to experience is often enhanced by being in contact with a diversity of individuals, cultures, and life stories, empathizing with them, and accepting their perspectives. The fact that both, openness to experience and empathy, were uniquely conducive of DT abilities in the current study suggest distinct mechanisms. The association between empathy and DT deserves more attention by creativity researchers. It is a promising line of research which can contribute to delineating the mechanisms behind the generation of creative ideas and explain in part why excellencism is positively associated with DT abilities and why perfectionism is negatively associated to these abilities.

Emotions during DT

The emotions that individuals experience can impact their internal search process. Three latent factors accounted for the emotions participants felt during the DT tasks. The factors were Positive emotions, Primary Negative emotions (e.g., angry, sad or scared), and Secondary Negative emotions (e.g., embarrassed, ashamed, or analytical). These factors are consistent with the validation study of Galanakis et al. [49], however our solution indicates potential areas for improvement.

The results suggested that perfectionism was associated with more Positive emotions, more Primary Negative emotions, and more Secondary Negative emotions compared to excellencism. Whereas excellencism predicted significantly less Primary and Secondary Negative emotions with small-to-moderate effects. In turn, Positive emotions were associated with more original ideas in the classic DT task with a small-to-moderate effect; Primary Negative emotions were associated with less original ideas in both tasks with large effects, whereas Secondary Negative emotions were associated with more original ideas in both tasks with moderate effects.

The effects largely counter-balanced themselves. More specifically, perfectionism was associated with more original ideas through greater Positive and greater Secondary Negative emotions, but also less original ideas through greater Primary Negative emotions. Excellencism was associated with more original ideas through fewerPrimary Negative emotions, but also less original ideas through fewer Secondary Negative emotions.

The Affect Infusion Model [72] and the dual pathway model of creativity [32, 33] both suggest that individuals in positive states rely more on heuristics, intuitive judgments and maintain a broader scope of attention, hence influencing creative ideas through the flexibility pathway, while those in negative moods engage in more systematic analytical thinking. Negative moods would have a positive effect on creative ideas through the cognitive persistence pathway. This could explain the positive effect of Secondary Negative emotions. It is important to acknowledge that despite a positive relationship with DT abilities, Secondary Negative emotions—which involve feelings of embarrassment, self-critics, and shame—may suggest an unhealthy relationship with the task. It may encourage disproportionate concerns over mistakes, doubts, performance anxiety, negative self-talk, and harsh self-criticism, which together may affect well-being. Hence, despite a positive effect on DT abilities, Secondary Negative emotions may become problematic if they are chronic or obsessive. Ambitious creative projects may accentuate this distinction between perfectionism and excellencism. More research is needed to understand the effect of Secondary Negative emotions on DT abilities. Lastly, it remains surprising that excellencism was not associated with more Positive emotions than individuals with lower standards. Their good performance on the tasks and their desire for intellectual challenges was expected to translate into more positive emotions. However, the task might be just as pleasurable regardless of one’s performance, and therefore individuals low on excellencism and perfectionism may have just as much positive emotions during the task than those with excellence standards.

It has been shown that certain individuals use negative emotional experiences as an instrument to be motivated towards their goals. For example, individuals high on neuroticism perform better on cognitively demanding tasks when they feel worried during the task [73, 74]. This phenomenon has been called the trait-consistent hypothesis. This hypothesis suggests that negative emotions benefit individuals with certain personality dispositions. For example, the experiences of worry or anxiety would be “trait-consistent experiences” for individuals high on neuroticism or perfectionism, and would facilitate their performance on creativity tasks [73]. One possibility is that emotions of embarrassment, shame, and self-judgments were triggered by the task and benefited individuals high on perfectionism. Cognitively demanding tasks require larger effort to optimize performance, which accentuates the trait-consistent effect of emotions. This hypothesis has not been tested with perfectionism and constitutes a promising line of future research.

An additional future direction is to extend the present study from creative idea generation to creative idea evaluation, which is another major component underlying individual’s creative abilities. Perfectionists not only establish exceptionally high and often inflexible standards for themselves, they also focus intensely on the process of evaluation [50]. They view life as a continuous sequence of assessments, where their performance is constantly measured, with their name always expected to be at the top. The evaluation of ideas and the metacognitive feelings associated with the perceived ease or difficulty of generating solutions are used to inform the decision to continue generating ideas or to stop [75]. It is an integral part of the creative process of generating, evaluating, and selecting ideas, and it is likely to contribute to understanding the effect that high standards have on creative thinking.

Implications

This study aimed to better understand the influence of perfectionism and excellencism on divergent thinking. In many competitive settings, as in some school environments, workplaces, sports, and artistic milieux, it is not uncommon for individuals to set very high standards of performance. These standards and goals can be energizing but they can also cause distress, anxiety, doubts, and many negative emotions. A key objective of creativity research is to find ways to help individuals unlock their creative potential.

Understanding the experiences of people who aim for perfection and excellence can help support them in schools, sports, and workplaces. This study suggested that perfectionism should preferably not be encouraged for outcomes associated with the generation of original ideas to simple problems. This study contributed to an understanding of the factors that are associated with the generation of original ideas, such as openness to experience, empathy, and positive as well as secondary negative emotions. The relationships people have with their goals and with themselves is of utter importance to their creative potential. By anticipating the emotions associated with excellencism and perfectionism during DT tasks, and understanding which emotions might be detrimental or beneficial, it is possible to better support the creative process of individuals.

Limitations

This study used self-reported questionnaires, which introduced social desirability biases. Furthermore, the use of a cross-sectional design to test indirect effects has limitations such as causality issues, omitted variable bias, and static snapshot. Moreover, the measurement of emotional DT was limited in scope. It would be necessary to examine a greater variety of DT tasks in the future to extend our results. Lastly, for reasons of parsimony, we focused the analyses on the originality of answers to DT tasks. Answers to DT tasks can be coded for the fluency (number of answers) and flexibility (number of different categories), which can reveal interesting patterns of associations which we did not investigate in this study. Future studies should inspect those indicators of DT abilities.

Conclusion

To conclude, different mechanisms must be considered to fully explain the associations between perfectionism, excellencism, and DT abilities. We identified the role of empathy, openness to experience, and the emotions during the task. It is not yet fully clear how those factors interact and whether different reasons for pursuing excellence and perfection could moderate these effects.

Although this study partially replicated the effects observed in [24], the present effects were smaller in size. Overall, this study contributed to the literature by examining associations which have been infrequently studied by creativity researchers. Among the key findings, we note the absence of associations between doubts about actions, concerns over mistakes, and the originality scores of DT; a positive and robust association between empathy, openness to experience, and DT, a large negative association between Primary Negative emotions and DT, as well as a moderate positive association between Secondary Negative emotions and DT abilities.

Future studies should continue to examine the role of empathy and emotions among individuals aiming for perfection and excellence as these types of strivings are frequently reinforced in competitive environments. Moreover, future research could investigate the trait-consistent hypothesis which suggests that secondary negative emotions can facilitate the performance of those pursuing perfection, due to a match with their personality dispositions. Lastly, investigating performance anxiety among individuals high on perfectionism and excellencism may contribute to comprehending its potential effects on creative cognitive performances.

Electronic supplementary material

Below is the link to the electronic supplementary material.

Appendix A: Emotions during the task

Positive emotions

1. Amused, fun-loving, silly

2. Awe, wonder, amazement

3. Content, serene, peaceful

4. Glad, happy, joyful

5. Grateful, appreciative, thankful

6. Hopeful, optimistic, encouraged

7. Interested, alert, curious

8. Love, closeness, trust

9. Proud, confident, self-assured

10. Surprised, amazed, astonished

11. Sympathy, concern, compassion

Negative emotions

• 12.Angry, irritated, annoyed
• 13.Contemptuous, scornful, disdainful
• 14.Disgust, distaste, revulsion
• 15.Repentant, guilty, blameworthy [15]
• 16.Sad, downhearted, unhappy
• 17.Scared, fearful, afraid

Worrisome emotions

• 18.Embarrassed, self-conscious, blushing [7]
• 19.Ashamed, humiliated, disgraced [3]
• 20.Anxious, panicky, nervous [22]
• 21.Embarrassed, inadequate, ashamed [23]
• 22.Analytical, judgmental, critical [24]

Excluded from the design

• *23.Sexual, desiring, flirtatious

*Emotion 23 (I feel “sexual, desiring, flirtatious”) was removed due to its irrelevance and not fitting any of the factor.

Author contributions

J-C. G-P., M-F. B., and D. C. conceptualized the study. M-F. B. collected the participants. J-C G-P. cleaned the dataset and executed the statistical analyses. J-C. G-P. and M-F. B. wrote the main manuscript text. J-C. G-P. prepared the Figures and Tables. D. C. supervised the project. D. C. validated the data analyses. D.C. reviewed and edited the main text.

Funding

Fonds de recherche du Québec—Nature et technologies, Natural Sciences and Engineering Research Council of Canada, 2016-03906.

Data availability

The dataset and supplementary materials are available at this OSF link: https://osf.io/94pb3/.

Declarations

Competing interests

The authors declare no competing interests.