The association between cognitive flexibility and creativity across childhood and adolescence: a scoping review

Abstract

Cognitive flexibility (CF), the ability to adaptively shift between perspectives, ideas, or strategies, is increasingly recognised as an important executive function underlying creativity. Creativity broadly refers to the capacity to generate ideas or solutions that are both unique and suitable to a given context. A positive association between CF and creativity is well-documented in adults. However, evidence from childhood and adolescence, which are developmental periods characterised by heightened neuroplasticity and the maturation of executive control, remains fragmented. This is largely due to substantial methodological heterogeneity across studies. This scoping review synthesises empirical studies published in the past decade that examine the association between CF and creativity in individuals aged 0–18 years. Specifically, it examines: (1) how CF and creativity have been conceptualised and measured in this age group; (2) whether there is an association between CF and creativity; and (3) whether this association varies across preschoolers, school-going children, and adolescents. Fourteen studies met the inclusion criteria for the scoping review. Definitions and measurement approaches varied widely, spanning performance-based tasks and self‑report instruments. Nine of the studies found a significant positive association between CF and creativity, particularly when CF was assessed as a complex, multifaceted construct and when creativity was assessed by traditional tools. In contrast, five studies reported null or inconsistent findings; these studies often employed narrow task-switching paradigms or creativity tasks incorporating other cognitive demands (e.g., reasoning). Importantly, no consistent age‑related differences were identified in the CF–creativity association. The results combined suggest that creativity may rely more on higher-level, strategic forms of CF that support the integration and exploration of ideas, rather than on simple rule-switching ability. By consolidating existing evidence and highlighting key conceptual and methodological gaps, this review offers a clear direction for future research to take when investigating the association between CF and creativity across developmental stages.

Supplementary Information

The online version contains supplementary material available at 10.1186/s40359-026-04886-8.

Introduction

Creativity in everyday life has been shown to be instrumental to academic achievement [1], fostering a sense of purpose [2], and enhancing the overall quality of life [3]. Scholarly work on creativity consistently identifies flexibility as a core cognitive component of creative thought [4–6]. Theoretical accounts such as the Dual Pathway to Creativity Model explicitly position cognitive flexibility (hereafter known as CF) as a central mechanism supporting creative thinking [7], as creative cognition requires the ability to move beyond habitual responses to make novel connections between seemingly unrelated concepts, which are processes that depend on flexible shifts in attention and perspective [8, 9]. In addition, both CF and creativity involve generating, manipulating, and switching between multiple ideas or techniques in response to changing task requirements [10, 11]. Further, empirical studies in adults [12] consistently show that higher CF is associated with more creative thinking and the use of creative problem-solving strategies. In sum, both theoretical and empirical work suggest that CF and creativity are closely related.

Despite scholarly evidence suggesting the close association between CF and creativity, the developmental trajectory of the CF-creativity relationship remains poorly understood. Although lifespan frameworks for CF [13] and creativity [14] have gained traction, they have largely evolved independently with limited integration. As a result, research on the CF-creativity link remains disproportionately adult-focused [14–16], while studies in children and adolescents are relatively scarce and methodologically inconsistent. This is particularly concerning given that childhood and adolescence are marked by dynamic neural and cognitive development [17, 18], offering potentially critical windows to understand how CF supports emerging creativity.

To move the field forward, there is a pressing need to consolidate what is currently known about CF and creativity across development. A comprehensive review of the literature can clarify how these constructs have been conceptualized and measured in younger populations, reveal patterns or inconsistencies in their association, and identify critical gaps in knowledge. Understanding where the field currently stands is a necessary step towards building an integrated developmental framework of CF and creativity.

Cognitive flexibility

Conceptualisation, origins and the shifting paradigm

CF refers to the capacity to adapt one’s thoughts, strategies or behaviour in response to changing task or contextual demands [9, 19]. It is related to but distinct from executive function (hereafter known as EF), which is a family of top-down executive control processes that support goal-directed behaviours through the regulation of attention, thoughts, and actions [17, 20]. Within the seminal EF framework by Diamond (2013) [17], CF is conceptualised as one of the EF subcomponents, alongside inhibition and working memory. While EF represents a broader set of control processes that coordinate behaviour in accordance with internal goals, CF specifically refers to the ability to shift between task rules, mental representations, or strategies. Distinguishing CF from the broader EF construct is important, as creativity research increasingly highlights the role of executive control [21, 22], while recognising the distinct role of CF in enabling individuals to generate, shift between, and integrate multiple ideas during creative problem solving [23].

Lesion studies demonstrate that damage to specific prefrontal regions impairs flexible adaptation, highlighting the PFC’s central role in supporting CF [24, 25]. This evidence establishes CF as a key component of EFs, which rely on prefrontal mechanisms to coordinate working memory, inhibitory control, and flexible rule implementation [26]. By supporting the ability to shift between competing task rules, representations, or strategies, CF enables the dynamic reconfiguration of cognitive resources [27], distinguishing it from other EF components. Importantly, these flexible processes are thought to underpin creative cognition, as they allow individuals to generate, manipulate, and integrate multiple ideas or solutions in response to changing task demands [23, 28]. Thus, examining CF provides important insights into how specific components of executive functioning contribute to creative problem solving across development.

Additionally, CF has also been studied through the task-switching (or set-shifting) lens [27, 29], with CF termed as task-set shifting or task switching. This line of research has been closely associated with the development of the task-switching paradigm in experimental psychology. A comprehensive review of this paradigm and its role in understanding flexible control is provided by Stephen Monsell (2003) [29], who synthesised decades of work examining how individuals shift between competing task sets. In Monsell’s (2003) [29] review, early experimental studies investigating mental set and attentional shifting (e.g., Jersild, 1927 [30]; Biederman, 1972 [31]) laid the foundation for later paradigms that operationalised CF through tasks requiring participants to alternate between stimulus–response rules. In these paradigms, flexibility is typically quantified through switch costs, referring to the performance decrement observed when participants switch tasks relative to repeating the same task. Building on this tradition, contemporary research commonly employs set-shifting paradigms such as the Wisconsin Card Sorting Test (WCST) and Trail Making Test (TMT) to operationalise CF within broader EF frameworks [32].

Neurocognitive accounts further link flexible cognition to prefrontal control systems that regulate behaviour through the interaction between bottom-up and top-down processing. Bottom-up processes are largely stimulus-driven and support well-learned or automatic responses, whereas top-down control involves internally maintained task goals that direct behaviour in accordance with current demands [33]. Miller and Cohen (2001) [34] proposed that the prefrontal cortex supports cognitive control by maintaining representations of task rules and goals that guide processing in other brain systems. These top-down signals allow individuals to select task-relevant responses when the task goal competes with more habitual or salient stimulus-driven responses. Classic paradigms such as the Stroop task and the WCST illustrate this principle: successful performance requires the suppression of a dominant or previously reinforced rule as well as the flexible adoption of a new one when task contingencies change. Within the framework proposed by Miller and Cohen, CF reflects the capacity to update and shift between task rules through prefrontal top-down control mechanisms that override more habitual bottom-up prepotent responses and support adaptive behaviours in changing environments.

Taken together, CF has been understood in several complementary ways: it has been described as a shift in mindset [35], a multidimensional competence involving cognitive, affective, and knowledge-based components that support adaptive performance [36], and a dispositional trait, reflecting one’s awareness and willingness to adapt [37–40]. Specifically, it is found that CF improves rapidly across early and middle childhood [13, 41]. In children, researchers have distinguished between different kinds of flexibility, with a recent account distinguishing between reactive flexibility, modifying behaviour in response to external changes, and spontaneous flexibility, the ability to independently generate varied ideas [42]. In adolescents, distinct developmental trajectories of facets of CF – cognitive shifting and contingency-based switching – have been proposed [43]. In Parr et al.’s [43] work, cognitive shifting refers to the capacity to redirect attention, adjust strategies, or change behavioural responses when task goals, rules, or mental sets change, aligning with the broader definition of CF provided above. In contrast, contingency-based switching involves modifying one’s behaviour in response to changing reward patterns or probabilities. These distinctions are particularly relevant for studies examining the CF–creativity link in pre-adulthood phases. Developmental accounts therefore suggest that the mechanisms supporting flexible cognition may shift across childhood and adolescence, with younger children relying more on reactive or bottom-up forms of flexibility, and older individuals increasingly engaging top-down executive control processes supported by prefrontal maturation [17, 27, 34].

Measurement

The performance-based approach to CF often indexes the behavioural and executive aspects of CF, that is, how efficiently individuals shift between task rules, stimulus dimensions, or response sets under changing conditions. CF tasks representative of this approach include the WCST [44] and its child-friendly versions such as the Dimensional Change Card Sort (DCCS) [45, 46], in which individuals infer and flexibly shift sorting rules based on feedback. Similar set-shifting demands can be found in colour–shape switching tasks [47] and their variants (e.g., the Figure Matching Task by Ellefson et al. [48]). Other task paradigms such as the TMT [49] capture set-shifting speed and sequencing, while reversal learning tasks assess flexibility in adaptive decision-making by requiring participants to reverse previously learned stimulus–reward contingencies when feedback patterns change [44]. Fluency-based measures are also commonly used, such as verbal fluency tasks [50], where participants generate as many category-based (e.g., fruits, animals) or phonemically constrained (e.g., F, A, S) words as possible within given time limits [51]. The fluency-based measures typically index rapid strategic retrieval and flexible search processes. Complementing these, the Five-Point Test [52] provides a nonverbal index of CF, requiring participants to connect dots within structured arrays to produce as many original designs as possible in three minutes.

Self-report measures instead assess perceived or dispositional flexibility, that is, how individuals evaluate their ability to adapt their thinking and behaviour in everyday contexts. The Cognitive Flexibility Scale (CFS) [38] captures the ability to recognise alternatives, adjust behaviour, and perceived control. The Cognitive Flexibility Inventory (CFI) [53] extends this approach to clinical and developmental settings, assessing the ability to recognise multiple interpretations and to perceive difficult situations as controllable. Its two subscales: (1) Cognitive Alternatives (the ability to generate alternative interpretations in response to changing situational demands) and (2) Cognitive Control (the individual’s appraisal of difficult situations as being within their control) have been linked to indices of mental resilience and reduced depressive symptoms [53].

In preschoolers and very young children, CF measurement primarily relies on motor or eye-movement responses. Perseverative reaching/looking tasks rooted in the A-not-B paradigm test children’s ability to shift to novel responses on switch trials [54, 55]. Early CF has also been examined using sequential-touch categorisation tasks, which assess whether children flexibly group objects [15, 56, 57]. In addition, tasks such as the Eyes Task [58] incorporate both inhibition and shifting: children view faces whose gaze direction signals either a congruent response (making a same-side button press whenever the eyes look downward) or an incongruent response (making an opposite-side button press when the eyes look diagonally across). Successful performance requires learning and flexibly applying two competing rules, making the task a sensitive measure of early set-shifting and inhibitory control. More recently, instrumented toys with embedded sensors have enabled fine-grained capture of children’s sequential touching and squeezing behaviours which are thought to reflect key CF developmental processes [59].

Across different age groups, correlations between behavioural and self-report measures tend to be weak and inconsistent. Dang et al. [60] argue that these methods engage different response processes: behavioural tasks typically involve unfamiliar, artificial stimuli and yield objective indices (accuracy, reaction time), whereas self-reports draw on subjective judgments about everyday behaviour, which are more susceptible to bias. Similar discrepancies between ability-based and self-perceived measures are well established in the CF literature [61–63].

In summary, CF is assessed through a range of behavioural and self-report tools, each capturing different aspects of CF. However, weak correlations between methods such as task-based and self-report instruments reflect underlying conceptual and measurement discrepancies, which may be exacerbated in developmental stages due to ability-based disparities between young children and older populations. These challenges highlight the need for more integrated, developmentally sensitive frameworks that align task demands with age-related capacities.

Creativity

Conceptualisation and key frameworks

Creativity has been conceptualised through multiple frameworks [64, 65]. Componential models emphasise the interaction of person, process, environment, and product, whereas cognitive process models focus on the mental operations that generate creative ideas [66, 67]. Common to these accounts is the recurrent theme of flexible thinking. Classic associative theories propose that creativity relies on forming remote or unconventional associations [68], supported by a flexible semantic memory system capable of reorganising concepts in novel ways [10]. Other perspectives highlight deliberate, effortful cognitive processes involving attention and memory to support flexible decision-making [69].

One influential framework linking executive control processes and creativity is the Dual Pathway to Creativity Model (DPCM), developed iteratively across several studies [7, 70]. The model proposes two complementary pathways to creative performance: flexibility and persistence. The flexibility pathway involves broad exploration, generation of diverse associations, and shifting between ideas, whereas the persistence pathway entails sustained, effortful search within a narrower conceptual space. Critically, DPCM specifies that state factors (e.g., mood, motivational activation) and trait-level individual differences (e.g., openness, approach-related traits) modulate engagement of these pathways, influencing whether flexible or persistent processing predominates [71].

From a developmental perspective, these pathways map onto the bottom-up versus top-down processes. In younger children, due to developing prefrontal areas and executive control abilities [72], creative responses may rely more heavily on bottom-up, stimulus-driven mechanisms, such as automatic attentional shifts and associative processes [73]. Older children and adolescents on the other hand, are found to increasingly recruit top-down executive processes, enabling goal-directed, effortful engagement of flexibility or persistence pathways in accordance with task demands [74, 75]. State (mood and motivation) and trait (individual dispositions) factors further modulate this balance: activating moods or higher dispositional motivation enhance top-down engagement, supporting deliberate strategy selection and idea generation, whereas deactivating moods or lower trait motivation encourage bottom-up, stimulus-driven processing [71].

Within the DPCM, CF aligns particularly with the flexibility pathway, supporting the dynamic generation, shifting, and integration of multiple ideas, a capacity that increasingly benefits from top-down control as EFs mature and are modulated by state and trait factors. This developmental lens also clarifies observed differences in creativity measures: divergent thinking (DT), which is related to the generation of multiple original ideas, primarily reflects exploratory flexibility, whereas convergent thinking (CT), which focuses on identifying one optimal solution from multiple alternatives, rely more on persistence and controlled, goal-directed processes [37, 38]. Considering CF within the DPCM thus illustrates how EFs contribute to creative cognition across childhood and adolescence, highlighting a shift from bottom-up, stimulus-driven flexibility in younger children to top-down, effortful, and context-sensitive flexibility in older developmental stages, reflecting reliance on linguistic versus visuospatial systems of creativity.

In addition, previous research showed that in early and middle childhood, creativity is often embedded in play-based and imaginative contexts, while in adolescence it becomes increasingly intentional, goal-directed, and domain-specific [76]. These changes highlight the importance of clarifying how creativity is defined and assessed at different developmental stages.

Measurement

Creativity has been assessed through ability-based tasks, self-report instruments, and, in early childhood, play- or motor-based measures. Performance-based tasks remain the most widely used ones for studying creative cognition. The Alternative Uses Task (AUT) [77], also known as the Unusual Uses Test [77, 78], requiring participants to generate unusual uses for everyday objects, measures verbal divergent thinking, with responses typically scored for fluency, flexibility, and originality [77, 79]. The compound Remote Associates Test (RAT) [79] assesses convergent creativity, requiring participants to generate a single meaningful association linking seemingly unrelated words, a process aligned with analytic problem-solving [80]. For figural creativity, the visual subtests of the Torrance Tests of Creative Thinking (TTCT), with its child-friendly version the Thinking Creatively in Action and Movement task [80, 81], remain the dominant measure, which captures visual-perceptual creativity and originality through completion of partial shapes [82]. Variants or similar measurements as that of the TTCT such as the PIC-N [83] and Complete the Drawing task [84] have also been used as creativity measurements. The AUT, RAT, and TTCT are considered classic tools in the creativity literature that have consistently demonstrated strong reliability and validity [85, 86] and remain the gold-standard for creativity measurement. In young children, the Pattern Meanings task assesses divergent thinking by asking children to name as many things as possible that a line drawing can resemble [87]. Real-world problem-solving tasks such as the hook-innovation task [58] require children to manipulate physical materials to achieve a concrete goal: bending a pipe cleaner to retrieve a sticker from a tube, thus assessing creativity in a practical, goal-directed context. Other creativity measures for young children include the Unusual Box test [87], which assesses divergent thinking by asking children to come up with novel ways to interact with a complex, unusual item.

Self-report measures offer a complementary perspective to task-based measures by capturing domain-specific creative dispositions and activities. For instance, the Kaufman Domains of Creativity Scale (K-DOCS) [88, 89] and the Creative Achievement Questionnaire (CAQ) [90] assess real-world creative accomplishments in adults across domains. Age-graded tools such as the Inventory of Creative Activities for Young Adolescents (ICAYA) [91] and the Creative Self-Efficacy Scale for Children and Adolescents (CASES) [92] extend self-report assessment to children and adolescents aged 3–16. Similar to CF measurements, past research showed poor correlations between self-assessed and task-based creativity measures [93–95], suggesting the need for conducting comparable measurements across developmental stages to draw more accurate conclusions on the developmental trajectory of creativity.

The present study

A considerable amount of research demonstrates a robust positive association between CF and creativity [96] in adults, supported by both behavioural [12, 97] and neural evidence [98]. These studies suggest that CF supports creative cognition through mechanisms such as perspective shifting, associative processing, and flexible semantic retrieval [99, 100]. While these adult findings provide a useful foundation, they cannot be directly applied to younger populations, whose cognitive, linguistic, motor, and expressive capacities differ markedly from adults.

In comparison, research on the CF-creativity relationship in childhood remains limited and yields inconsistent results. Some studies report null findings. Hoffmann and Russ [101] examined 61 children (kindergarten through fourth grade; ~5–10 years) using the Wisconsin Card Sorting Task-64 (WCST-64) to assess CF, and both a divergent thinking task (i.e., the Alternate Uses Task) and a storytelling task to assess creativity. No significant association was observed between WCST performance and either creativity measure, suggesting that traditional CF tasks may not capture the control processes supporting creativity in younger children. De Chantal and Markovits [102] similarly observed no link between the DCCS task performance and performance on creative prompts in children aged between 32 and 64 months. In their study, preschool children completed a DCCS task to assess set-shifting abilities and a generation task to capture the capacity to produce alternative ideas (which was considered a measure of early creativity). Their results indicated that performance on the DCCS did not predict responses on the creative prompts, whereas the generation task significantly influenced reasoning outcomes. These findings suggest that, at this early developmental stage, standard CF measures such as the DCCS may not yet be closely linked with creative idea generation. Conversely, Wang et al. [46] showed that preschoolers (N = 26, 51–82 months) who performed a DCCS task and a nonverbal creative thinking task (the Unusual Box Test; UBT) recruited overlapping lateral prefrontal brain regions during both tasks. While behavioral performance on the DCCS and UBT were not significantly correlated, neuroimaging data indicated that children who strongly engaged ventrolateral prefrontal regions during the post-switch phase of the DCCS also recruited these regions during the UBT. These findings suggest that, even in early childhood, CF-related neural mechanisms may partially support creative thinking, despite behavioral measures not showing a direct link. These mixed findings likely reflect heterogeneity in task design (e.g., DCCS, WCST, UBT, generation tasks), creativity measures (divergent thinking, idea generation, storytelling), age ranges (4–5 vs. 5–10 years), and analytic approaches, as well as developmental differences in how cognitive flexibility supports creative thinking across early and middle childhood.

Adolescence, in contrast to childhood, presents a critical period where creativity becomes more intentional, goal-directed, and domain-specific [103, 104]. Evidence further supports that creativity maturation is particularly pronounced during this period, as several studies show notable gains in originality and divergent thinking across mid-adolescence [105, 106]. In addition, adolescence appears to be a particularly formative period for the CF–creativity link, with individuals showing heightened responsiveness to targeted training. For instance, Stevenson et al. [107] examined creative ideation training in adolescents (13–16 years, N = 71) and adults (22–31 years, N = 61), using eight 20-minute sessions over two weeks in three conditions: (1) alternative uses generation (creative ideation), (2) object characteristic generation (general ideation), and (3) rule-switching. Participants’ performance on the AUT was assessed for fluency, flexibility, and originality. While all participants improved in fluency and originality across conditions, adolescents showed greater gains in originality compared with adults, highlighting developmental differences in learning sensitivity and neuroplasticity. These findings suggest that creative ideation and cognitive flexibility can be enhanced in adolescence, supporting the view that this stage represents a window of increased flexibility, exploration, and capacity for intervention-driven improvement in creative thinking. Further, research on CF development highlights adolescence as a period of specialised flexibility where distinct components of CF follow different maturational trajectories. Cognitive shifting abilities improve steadily from childhood to adulthood, reflecting gradual refinement of neural processes such as brain-state flexibility, neural variability, and excitatory/inhibitory balance. In contrast, contingency-based switching exhibits an adolescent-specific peak, driven by dopamine-mediated increases in exploration, reward sensitivity, and flexible adaptation to changing environmental contingencies. This adolescent mode of flexibility is proposed to scaffold adult-like cognitive shifting and supports the emergence of mature, goal-directed cognitive flexibility, emphasising the developmental importance of this period for fostering flexible thought [43]. Together, these findings suggest that adolescence represents a formative period in which creativity and CF co-develop through complementary processes. Steady improvements in cognitive shifting, combined with an adolescent-specific peak in contingency-based switching, reflect underlying neurodevelopmental changes including dopamine-driven exploration, neural variability, and excitatory/inhibitory balance. These developmental trajectories, alongside heightened responsiveness to CF and creative ideation training, suggest that adolescents are particularly sensitive to interventions and environmental demands that foster flexible, goal-directed creative thinking. Despite strong theoretical convergence in how CF and creativity are understood, where both involve the ability to generate, shift between, and integrate multiple ideas or perspectives [7, 9], the empirical literature linking these constructs during development remains fragmented. Much of what is known arises from adult studies, with considerably less work examining how CF and creativity relate across childhood and adolescence, even though these periods represent critical windows for the emergence of flexible thought [13] and creative behaviour [108]. Furthermore, both constructs are measured using diverse tools that vary substantially across age groups, from motor-based paradigms in very young children to performance tasks and self-reports in older children and adolescents. These methodological differences introduce uncertainty regarding how CF and creativity are conceptualized, how comparable measures truly are across development, and whether observed associations are robust or anecdotal.

With these considerations, a scoping review is particularly appropriate in this domain for three reasons. First, the developmental literature is very limited, requiring an exploratory approach that can incorporate a wide range of study designs and measurement tools. Second, flexibility and creativity are assessed using age-dependent paradigms (see discussion above), making it essential to chart this measurement landscape before synthesizing associations. Third, preliminary evidence suggests that the CF–creativity link may shift across age [102], but no review has yet examined this systematically across developmental stages. Guided by these needs, the present scoping review addresses three exploratory research questions for individuals before 18 years old in the recent 10 years:

1. How are CF and creativity conceptualized across studies on individuals aged between 0 and 18 years old?

2. How are CF and creativity measured across studies on individuals aged between 0 and 18 years?

3. Is there a CF-creativity association in developmental studies and how does this association differ in strength across preschoolers, school-going children, and adolescents?

Method

Search and selection of studies

The Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines [109] informed our literature search and reporting of results. In our effort to compile relevant studies systematically, we used the following combination of keywords: “(Divergent thinking OR Creati* OR Originality OR Convergent thinking) AND (Cognitive flexibility OR Executive funct* OR Attention* OR Shifting OR Switching OR Mental flexibility OR Behavioural flexibility OR Neural flexibility OR Brain flexibility)”. Keywords were chosen for their relevance to creativity and CF, drawing on prior research investigating these constructs in children and adolescents [9, 110]. We then searched the following databases: PubMed, Web of Science, ERIC, and APA PsycINFO. The search spanned articles published within the recent 10 years from January 1, 2016, to October 12, 2025.

Study selection adhered to these inclusion criteria: the study must (i) be a peer-reviewed journal article that is published in English between January 1, 2016, and October 12, 2025, (ii) be empirical in nature, (iii) involve participants aged 0 to 18 years from neurotypical populations without clinical diagnoses, (iv) include at least one measure of creativity and (v) one measure of CF, (vi) and be published in peer-reviewed journals.

To maximise search sensitivity and reduce the risk of missing relevant studies due to inconsistent age indexing across bibliographic databases, we applied age restrictions as one of the inclusion criteria rather than as a search filter [111]. Following previous work on the relationship between executive functions and creativity [112], this review focuses on typically developing children and adolescents, while neurodivergent or clinical populations were excluded. This scope of investigation would allow for more coherent integration of evidence and thus more in-depth interpretation of findings. This decision is also informed by the fact that including neuroatypical groups could introduce additional heterogeneity due to disorder-specific cognitive profiles. Such heterogeneity in executive functioning, as reflected in behavioural performance, would complicate the synthesis of evidence.

Titles and abstracts were first screened by two researchers, (anonymised for review) and (anonymised for review) independently. The Cohen’s Kappa for inter-rater reliability between reviewers was 0.89, indicating strong agreement. Full texts were screened and all authors were involved in decisions on inclusion/exclusion of the relevant articles. During the full-text screening stage, two additional inclusion criteria were applied: (vii) CF had to be measured as an independent construct rather than as a subcomponent of EFs; and (viii) the study had to include an analysis of the relationship between CF and creativity. In addition, following the screening process, (anonymised for review) conducted a snowball search to ensure that no relevant studies were missed. The search resulted in an initial total of 915 studies: 58 from ERIC, 12 from PubMed, 427 from APA PsycINFO, and 418 from Web of Science. After removing duplications (n = 152), we identified 35 papers for full draft review. From these 35 papers, 21 were excluded for the following reasons: seven studies measured and analysed overall EFs without isolating CF; seven did not include any measure of CF; six did not examine the relationship between CF and creativity; and one was not a peer-reviewed publication. Details of the screening procedure are provided in Fig. 1. The screening process yielded a total of 14 studies which fully met the inclusion criteria outlined above.

Fig. 1.

PRISMA flow diagram of the literature screening process

Quality control

To assess the methodological quality of the cross-sectional studies, we applied the Joanna Briggs Institute (JBI) Critical Appraisal Checklist [113], which provides a clear scoring system and comprehensive evaluation criteria. The checklist consists of eight items assessing sample selection, participant description, exposure measurement, assessment of participants’ conditions, identification of confounders, strategies to control confounding, outcome evaluation, and statistical analysis. Each item is rated as yes, no, unclear, or not applicable, with the justification that the answer “yes” will account for 1 point. A total score above 7 would indicate a low risk of bias. In our sample, all 14 studies included in the review scored at least 5, indicating at least a moderate risk of bias. As an exploratory step, we further examined whether study quality moderated the observed associations; the individual JBI scores for each study are provided in the Appendix (Table A1).

Data coding

Data extraction was conducted by one author (anonymised for review), who collected the required information from the included studies. Another author (anonymised for review) independently reviewed this information to verify its accuracy and ensure completeness. The information extracted included basic details such as authors, country of research, age range, and sample size. We also collected detailed data on the definitions, measurements, task types, and data formats related to the selected assessments of creativity and CF, as well as whether the study identified an association between creativity and CF. Test statistics and effect sizes, as well as the outcomes of the respective statistical analyses are reported as follows. Detailed literature filtering steps can be found in the OSF file of this project: https://osf.io/b9hxy/.

Results

A summary of the results can be found in Table 1. A detailed data report can be found in the OSF file of this project: https://osf.io/b9hxy/.

Table 1.

Study results summary

Study	Age Range of Participants	Age group category	Sample Size	CF Conceptualization	CF Measurement	Creativity Conceptualization	Creativity Measurement	CF-Creativity Link
Vaisarova & Carlson (2021) [114]	4 and 6 years old	Preschool & Childhood	N = 103	Shifting	Flexible Item Selection Task (FIST)	Production of ideas or outcomes	Alternate Uses Task (AUT)	Yes
Veraksa et al. (2022) [84]	6–7 years old	Childhood	N = 206	NA	Dimensional Change Card Sort (DCCS)	Internal cognitive disposition or thinking style	Complete the Drawing task	Yes
Bai et al. (2021) [115]	5.45–6.53 years old	Preschool & Childhood	N = 102	Shifting	1) Dimensional Change Card Sort (DCCS); 2) Animal Shifting	Internal cognitive disposition or thinking style	Alternate Uses Task (AUT)	No
Wang, Sakata, & Moriguchi (2021) [116]	50m20d- 81m24d	Preschool & Childhood	N = 26	Shifting	Dimensional Change Card Sort (DCCS)	Production of ideas or outcomes	Unusual Box Test (UBT)	Yes
Arán Filippetti & Krumm (2020) [42]	8–13 years old	Childhood	N = 177	Emergent cognitive construct encompassing shifting	1) Wisconsin Card Sorting Test (WCST); 2) The Semantic Verbal Fluency test (SVF) 3) The Phonological Verbal Fluency test (PVF); 4) Five-Point Test (FPT)	Internal cognitive disposition or thinking style	1) Torrance Tests 2) Creative Intelligence	Yes
Segundo-Marcos et al. (2025) [83]	9–12 years old	Childhood	N = 60	Shifting	Number-Letter Task	Dynamic cognitive or problem-solving process	PIC-N (Torrance like Creativity Imagination Test)	No
Beck et al. (2016) [58]	6–8 years old	Childhood	N = 43	Shifting	Eyes task	Production of ideas or outcomes	Hook-innovation task	No
Taranu et al. (2019) [117]	6, 8 and 10 years old	Childhood	N = 63	Shifting	Verbal Fluency Task	Internal cognitive disposition or thinking style	Pattern meanings task	Yes
Kupczyszyn et al. (2024) [118]	9–12 years old	Childhood	N = 304	Emergent cognitive construct encompassing shifting	1) Wisconsin Card Sorting Test (WCST); 2) The Semantic Verbal Fluency test (SVF); 3) The Phonological Verbal Fluency test (PVF)	Dynamic cognitive or problem-solving process	Creative Personality Scale	Yes
Krumm et al. (2018) [119]	8–13 years old	Childhood	N = 209	NA	1) Wisconsin Card Sorting Test (WCST); 2) The Semantic Verbal Fluency test (SVF); 3) The Phonological Verbal Fluency test (PVF)	Internal cognitive disposition or thinking style	1) Torrance Tests 2) Creative intelligence	Yes
Sampedro & Pena (2019) [120]	9–12 years old	Childhood	N = 224	NA	Five-Point Test (FPT)	Production of ideas or outcomes	Torrance Tests	Yes
Caliskan & Altun (2025) [121]	10th grade High school students	Adolescence	N = 341	Multidimensional Adaptive ability	Cognitive Flexibility Scale for high school level students	Internal cognitive disposition or thinking style	the Oil Spill problem from PISA	No
Wang & Chang (2022) [116]	11–17 years old	Childhood & Adolescence	N = 765	Multidimensional Adaptive ability	The Cognitive Flexibility Inventory (CFI)	NA	Creativity Scale	Yes
Zhao et al. (2023) (Study1) [122]	13–15 years old	Adolescence	N = 73	Shifting	Switching Task	Production of ideas or outcomes	1) The Remote Association Test (RAT); 2) Torrance Tests	No

We also conducted a preliminary pilot meta-analysis, the results of which are reported in the Appendix. Given the heterogeneity in CF measurement approaches across the 14 studies, this exploratory analysis disaggregated effect sizes by measurement modality: cognitive tasks, questionnaire-based measures, and verbal fluency tasks. Full forest plots and pooled estimates for each modality are presented in Appendix Figures A1 to A3.

Definition and conceptualisation of CF

Among the 14 studies included in the analysis, 11 provided explicit conceptual definitions of CF, while 3 did not. For the 11 studies that explained the concept of CF, their conceptualisations can be categorized into three themes: (1) CF as shifting, (2) CF as a systemic property arising from the coordination of multiple cognitive processes, and (3) CF as an ecologically-relevant multidimensional ability suited to real-world contexts.

The first conceptualisation of CF as shifting was found in a total of 7 studies (n_total = 7; n_preschoolers = 3; n_children = 3; n _adolescents = 1). This conceptualisation is largely grounded in the broader EF literature, with CF being a key component of EFs alongside inhibition and working memory. In these studies, CF is typically described as “switching between modes of thinking” [118: 2], and as the ability to “to switch between multiple tasks, operations, or mental sets, allowing efficient focus on changes in environmental demand and adapting behaviors to new situation” [88: 6] .

The second theme, situated within an integrative perspective, yielded a total of 2 studies (n_total = 2; n_children = 2). CF in these studies is framed as “a property of the cognitive system rather than a specific cognitive ability” [43: 3]. Arán Filippetti and Krumm [42] further distinguish CF into two subtypes: reactive and spontaneous flexibility. CF was also described as a more advanced component of EF, “recognized as an emergent property of the cognitive system” [120: 2].

The third theme involves a more complex conceptualisation of CF, i.e., CF as a form of multidimensional adaptability (n_total = 2; n_adolescents = 2). Specifically, CF was described as a multidimensional process involving the generation of multiple ideas, consideration of alternatives, and flexible adaptation to changing task demands [121], typically associated with problem-solving, trial-and-error learning, and fluid intelligence, and as a combination of cognitive control, flexible behavior, and social adaptability [116] suggesting its relevance to real-world contexts.

For the 3 studies that did not provide explicit definitions, CF was mentioned as one of the EF components [84, 119, 120].

Definition and conceptualisation of creativity

All but one of the reviewed studies offered explicit definitions for creativity, which can be broadly categorised into three themes: (1) creativity as being able to come up with original and useful ideas or outcomes, (2) creativity as a dynamic cognitive or problem-solving process, and (3) creativity as an internal cognitive disposition or thinking style.

The first conceptualisation of creativity was found in a total of 5 studies (n_total = 5; n_preschoolers = 2; n_children = 2; n_adolescents = 1). In these studies, creativity is understood as producing ideas that are novel and effective [118: 1] and as “the capacity to produce novel and useful outcomes” [47: 1].

A second group of studies approach creativity as a dynamic cognitive or problem-solving process (n_total = 2; n_children = 2), which involves stages such as problem identification, idea generation, testing, and refinement. Creativity in this genre is defined as a cognitive process generating effective, new, and original ideas by Segundo-Marcos et al. [83] and as a process that enables individuals to detect problems, formulate and test hypotheses, and communicate results by Kupczyszyn et al. [118].

Finally, a third genre of studies conceptualise creativity as an internal cognitive disposition or thinking style, rather than as a product or process (n_total = 6; n_preschoolers = 2; n_children = 3; n_adolescents = 1). This includes definitions highlighting divergent thinking, imagination, flexibility, and the ability to combine unrelated concepts. For example, creativity is viewed as the ability to generate novel ideas, switch perspectives, and combine unrelated concepts in different ways by Arán Filippetti and Krumm [42] and as a thinking process involving the production of original ideas through exploring multiple possibilities by Bai et al. [115].

Measurements of CF

For the measurements of CF, among the 14 studies included in the analysis, 12 studies assessed CF through task-based performance (n_preschoolers = 4; n_children = 7; n_adolescents = 1), while 2 used self-report measures (n_adolescents = 2). The types of data collected from the task-based performance measures fell within five main categories, which included rating or performance scores (n_total = 5; n_preschoolers = 2; n_children = 3), the number of recalled items (n_total = 2; n_children = 2), switch-cost related data (e.g., switch cost, switch ratio and switch frequency, n_total = 4; n_preschoolers = 2;n_children = 2), response time (n_total = 1; n_adolescents = 1), and accuracy (n_total = 1; n_preschoolers = 1).

The 12 studies that employ task-based performance can be further grouped into two categories: (1) those with a focus on cognitive shifting and (2) those with a focus on spontaneous and strategic flexibility, an advanced form of CF that enables individuals to adaptively adopt and utilize strategies across different contexts. In studies that focused on cognitive shifting (n_total = 7; n_preschoolers = 4; n_children = 2; n_adolescents = 1), the tasks used include the WCST [42, 119], the DCCS [46, 84, 115], the Flexible Item Selection Task [114], the Switching Task [122], the Eyes Task (a measure of task switching), the Animal Shifting Task [115], and the Number-Letter Task [83]. The types of data extracted and analyzed from these tests or tasks include rating scores (n_total = 2; n_preschoolers = 2), switch cost (n_total = 2; n_children = 2), switch ratio (n_total = 1; n_preschoolers = 1), switching frequency (n_total = 1; n_preschoolers = 1), response time (n_total = 1; n_adolescents = 1), and accuracy (n_total = 1; n_preschoolers = 1), with some studies incorporating multiple types of data.

For studies that focus on spontaneous and strategic flexibility (n_total = 5), all of them involve children as participants. The tests used for assessing spontaneous and strategic flexibility include the Verbal Fluency Task [117], the Semantic and Phonological Verbal Fluency tests [42, 118, 119], and the Five-Point Test [42, 120]. The types of data extracted and analyzed from these tests include aggregated performance scores (n_total = 3; n_children = 3) and the number of recalled items (n_total = 2; n_children = 2). In addition, the 2 studies that measure CF using self-report scales both target adolescents. The Cognitive Flexibility Scale for high school students was employed by Caliskan and Altun [121], while the CFI was used by Wang and Chang [116].

Measurements of creativity

For measurements of creativity, 12 studies employ task-based performance measures (n_preschoolers = 4; n_children = 6; n_adolescents = 2), and 2 studies use teacher-rating or self-rating scales (n_total = 2; n_children = 1; n_adolescents = 1).The types of data collected from task-based performance include rating scores (n_total = 8; n_preschoolers = 2; n_children = 5; n_adolescents = 1), the number of responses, valid, or unique ideas (n_total = 3; n_preschoolers = 2; n_children = 1), accuracy (n_total = 1; n_adolescents = 1), and whether children finish the task on time (n_total = 1; n_children = 1).

Among the 12 studies that use task-based performance, Torrance Tests are the most frequently used assessment (n_total = 5; n_children = 4; n_adolescents = 1). This was followed by the AUT (n_total = 2; n_preschoolers = 2) and the Creativity Intelligence measure (n_total = 2; n_children = 2). The Complete the Drawing task [84], the UBT [46], the hook-innovation task [58], the Pattern Meanings task [117], and the RAT [122] were each used in one study. In addition, the Oil Spill problem from PISA was employed to assess high school students’ creative problem-solving skills [121].

Two self-report creativity measures were used in the reviewed studies. The Creative Personality Scale was used in one study on children and was rated by teachers [118], whereas the Creativity Scale was used in one study with adolescents which they self-rated [116].

The association between creativity and CF across age groups

Of the 14 studies, 9 reported that CF is fully or partially associated with creativity, while 5 did not find evidence for such an association. Age-group-specific results are presented in the following sections.

Preschoolers

In the 4 studies on preschoolers, 3 identified a creativity-CF association while 1 did not. In studies where an association was observed, different measurement approaches were used, and the strength of the association varied accordingly: Vaisarova & Carlson [114] found that the Flexible Item Selection Task was significantly negatively correlated with both the log fluency (r = − .23, p < .05) and log originality (r = − .24, p < .05) of the AUT, leading them to surmise that CF negatively predicts creativity in preschoolers. Veraksa et al. [84] used the DCCS and the Complete the Drawing Task to assess CF and creativity respectively. In contrast to other studies, this study investigates how creativity influences CF, and after controlling for age and gender, DCCS performance was found to be significantly influenced by imagination flexibility in the drawing task. Specifically, preschoolers whose drawings showed fewer repetitions demonstrated enhanced cognitive switching abilities (η²p = .023, p < .05), but with very small effect size. Wang et al. [46] used the DCCS to measure CF and the UBT (Fluency and Originality) to assess creativity. They found no direct association between CF and creativity at the behavioural level. Nevertheless, functional near-infrared spectroscopy data indicated positive associations between activation in the left ventrolateral prefrontal cortex when performing the UBT and on the post-switch trials of the DCCS (r = .61, p < .01) and the right ventrolateral prefrontal cortex (r = .43, p < .05) with large effect size. However, the study conducted by Bai and colleagues [115] using the DCCS and Animal Shifting task to assess CF and the AUT (Fluency and Originality) to assess creativity found no significant association between CF and both fluency and originality in the AUT.

School-going children

Of the 7 studies on school-going children, 5 identify a positive association between CF and creativity, while 2 do not.

Among the 5 studies reporting a CF-creativity link, Krumm et al. [119] assessed CF using the WCST, verbal and phonological fluency tasks, and the Five-Point Test, and measured creativity with the Torrance Tests and the Creativity Intelligence measure. All CF measures were positively correlated with creativity measures, with verbal fluency and the Five-Point Test showing medium to large effects (rs = 0.33–0.52) and the WCST showing small effects (rs = 0.16–0.20). Hierarchical regression indicated that performance on the Five-Point Test predicted scores on both the Torrance (β = 0.17, p < .05) and the Creativity Intelligence (β = 0.17, p < .001) tests, while the verbal fluency measures predicted only Creativity Intelligence performance (β = 0.25, p < .001). Two Structural Equation Modeling (SEM) analyses indicate that CF, as indexed by verbal fluency tasks and the Five-Point Test predicted creativity (β = 0.68-0.77). Arán Filippetti and Krumm [42] employed assessment methods similar to those used in Krumm et al. [119]. However, they view WCST as a measure of reactive CF and verbal and phonological fluency tasks alongside the Five-Point Test as measures of spontaneous flexibility. Correlation results found significant positive correlations with all the CF and creativity measures (r = .17 − .54), with hierarchical regression showing that verbal fluency significantly predicts creativity (β = 0.22, p = .01 for the Torrance Tests; β = 0.33, p < .001 for Creativity Intelligence), as did the Five-Point Test (β = 0.22, p = .01 for the Torrance Tests; β = 0.31, p < .001 for Creativity Intelligence). In addition, SEM analysis indicated that only spontaneous flexibility (β = 0.77) contributes to creativity. The study by Kupczyszyn et al. [118] used similar tasks, assessing CF with the WCST and verbal and phonological fluency tasks (indexing spontaneous flexibility), alongside the Creative Personality Scale for creativity (with the five subdimensions and a total score). All three CF measures significantly correlated with all the subdimensions of the Creative Personality Scale, with small to medium effect sizes (rs = 0.16–0.31). However, only verbal fluency positively predicted the total Creative Personality Scale score (β = 0.20, p = .001). In addition, their SEM analysis showed that spontaneous flexibility, represented by verbal fluency tasks significantly predicts creative personality (β = 0.13). Kupczyszyn et al.’s [118] results suggest that spontaneous flexibility may be a stronger predictor of creativity than reactive flexibility.

Sampedro and Peña [120] assessed CF using the Five-Point Test and measured creativity with the verbal and figural components of the Torrance Test. They found that performance on the Five-Point Test correlates with verbal creativity (r = .39, p < .001), figural creativity (r = .27, p < .001), and overall creativity (r = .42, p < .001). In their mediation analysis examining bilingualism, CF, and creativity, CF was also found to directly predict verbal (β = 0.34, p < .001) and figural creativity (β = 0.19, p < .001). Finally, Taranu et al. [117] assessed CF using a verbal fluency task and measured creativity with the Pattern Meanings task. They found a significant positive correlation between CF and creativity, with a medium effect size (r = .48 before controlling for age; r = .45 after controlling for age).

As for the two studies that did not find an association between CF and creativity in children, Segundo-Marcos et al. [83] assessed CF using the Number–Letter Task and measured creativity with the Creative Imagination Test (a measure conceptually similar to the Torrance Test). Beck et al. [58] used the Eye Task (a task-switching–type measure) to assess CF and the Hook-Innovation Task to evaluate children’s ability to solve a physical reasoning problem.

Adolescents

Only three studies in our review were found to have examined the link between CF and creativity during adolescence, with one finding supporting evidence for the link and the other two not. The study by Wang and colleagues [116] used the self-report CFI Scale and a creativity scale and reported significant correlations between both CF dimensions, i.e., cognitive alternatives (r = .68, p < .01) and cognitive control (r = .35, p < .01), and creativity, with medium to large effect sizes. SEM analysis also found that cognitive alternatives (β = 0.57, p = .001) and cognitive control (β = 0.13, p = .001) positively predicted creativity. In the other studies that did not report a significant positive association, Caliskan and Altun [121] used the self-report CFI Scale to assess CF, and implemented the Oil Spill problem from the PISA assessment as the creativity measure. Zhao et al. [122] measured CF using a switching task and assessed creativity using the RAT and the TTCT.

Discussion

This scoping review aimed to examine the association between CF and creativity, focusing on how both constructs are conceptualised and measured, the robustness of the CF–creativity link, and whether this link varies across developmental stages. Overall, the evidence across ages is mixed, and the pattern does not follow a linear age-related trajectory. Studies on preschoolers, school-going children, and adolescents each show a combination of significant and null findings, suggesting that age alone does not reliably moderate the CF–creativity relationship. Instead, variability appears more strongly driven by differences in theoretical assumptions, conceptual definitions, and measurement approaches of the two constructs.

Conceptualisations and measurements of CF and creativity

Our findings show that developmental studies conceptualise CF in three main ways: (1) as shifting or set-switching, or rule-switching, assumed in studies involving preschoolers, school-going children, and adolescents, (2) as an integrated capacity involving multiple cognitive control components alongside shifting, assumed mostly in studies focusing on middle childhood and (3) as a multidimensional form of adaptability, supported by neural maturation, enabling metacognitive, behavioural, and social flexibility, assumed mainly in studies involving adolescents. These patterns align with established developmental trajectories showing that EFs tend to be more unitary in early childhood, within which CF is often operationalised as a simple set-shifting skill, before EF later fractionates into multiple specialised components capable of supporting a wider range of control operations [123]. This point will be elaborated further when discussing the link between CF and creativity below.

In contrast to CF, creativity is conceptualised across developmental studies through three main approaches, but without a consistent age-related pattern. In the first approach, creativity refers to coming up with unique and effective ideas [124]. This perspective is assumed in studies involving preschoolers, school-going children, and adolescents alike, suggesting that researchers across ages value novelty and utility as central components in the study of creativity. The second approach frames creativity as a dynamic cognitive or problem-solving process, involving stages such as problem identification, idea generation, testing, and refinement [83, 118]. This conceptualisation is mostly observed in child studies but is not exclusive to a specific age group. The third approach treats creativity as an internal cognitive disposition or thinking style, highlighting divergent thinking, imagination, and the ability to combine unrelated concepts [42, 115]. This perspective also appears across studies sampling preschoolers, school-going children, and adolescents. As a result, different facets of creativity, rather than age-specific capacities of it, are emphasised in the reviewed studies. The differences in conceptualisations appear more related to research focus or theoretical framing than to age, reflecting the multifaceted and context-sensitive nature of creativity in developmental research.

Task selection for construct measurement aligns closely with suitability to age group, particularly for CF. Across studies, CF measurement is dominated by task-based assessments in preschool and school-aged children, with self-report measures appearing only in adolescent samples. This uneven distribution likely reflects developmental constraints: young children often have limited metacognitive insights, who could be inadequately capable in terms of linguistic comprehension and response consistency required for valid self-report [125]. As a result, it appears that researchers default to performance-based tasks to index CF in early and middle childhood. Zlotnik and Toglia [126] suggest that adolescents’ self-rated EF abilities closely correlate with their actual task performance, which is not the case for younger children [127]. Nonetheless, using different measurements across age groups raises significant concerns about construct equivalence. First of all, self-report assessments or questionnaires are meant to assess EF skills in context, which are highly ecologically valid, while task-based measures often assess EF skills in distraction-minimised contexts, which are not typical of real-world environments [128, 129]. Further, it is noteworthy that the EFs underlying performance in self-rated and task-based measures show different trajectories of development in adolescence [130]. This means that developmental comparisons using different instruments across ages risk confounding true age-related change with measurement artefacts. It can therefore be argued that CF, as one component of the broader EF system, is tapped in different forms when measured through task-based and self-report assessments. Apparent increases in CF from childhood to adolescence [131, 132] may therefore reflect the transition from tasks to self-reports rather than genuine cognitive maturation.

As for creativity assessment, both task-based performance and self-report measurement approaches are employed consistently across developmental stages, whereas the specific forms and features of tasks vary substantially. Studies on preschoolers tend to rely on drawing, pretend-play, or imagination tasks that tap early symbolic representation and basic divergent thinking [133, 134]. Studies on school-going children more often use structured divergent-thinking measures such as the TTCT [135, 136], while adolescent studies incorporate convergent or complex problem-solving assessments (e.g., the RAT or PISA Oil Spill), which place heavier demands on linguistic ability, reasoning, and executive control [137, 138]. These task-level differences mean that the same aspect of creativity is not assessed across age groups, with each measurement approach privileging different cognitive processes. Consequently, apparent developmental differences in CF–creativity associations may reflect shifts in task demands rather than genuine age-related changes in the underlying constructs.

In addition, creativity tasks such as the TTCT and the AUT often yield multiple indices, including fluency, originality, and flexibility, which reflect distinct aspects of the creative ideation process rather than a single unified construct [124]. Prior work suggests that executive control processes may play a greater role during later stages of ideation, when individuals move beyond generating numerous responses (fluency) to producing more novel ideas (originality), as later phases of creative problem solving often require more deliberate and controlled processing than early associative idea generation [70, 139]. These patterns of our results suggest that CF may support specific stages of creative ideation, particularly those involved in generating more novel or uncommon ideas rather than simply producing a larger number of responses. However, some studies included in this review reported creativity outcomes as composite scores or emphasised overall divergent-thinking performance rather than analysing specific dimensions such as originality and fluency separately. As a result, it was not always possible to determine whether observed CF–creativity associations were driven primarily by a distinct factor within creativity.

The CF-creativity link across developmental stages

In this section, we discuss the CF-creativity link in terms of the following three conceptualisations of CF: (1) as a shifting ability, (2) as an integrative aspect of cognition involving multiple control components, and (3) as an ecologically-relevant multidimensional ability suited to real-world contexts. We will also discuss the CF-creativity association with respect to the interactions between concept definition, task measurement and age group for greater clarity on the CF-creativity developmental framework.

CF as a shifting ability

Across the preschool and early school-age years, CF is most commonly conceptualised as a shifting ability, and correspondingly, studies used simple rule-switching tasks such as the DCCS, Animal Shifting, and Flexible Item Selection Task among preschoolers and fluency-based tasks such as verbal fluency, Number–Letter, or Eyes tasks among older children. Creativity in these studies is typically assessed using tasks emphasising idea generation, including the AUT, drawing tasks, or Unusual Box Test in preschoolers, and Torrance-like imagination tasks in middle childhood. Across these age groups, findings showed a mixed pattern: some studies report significant positive [46] and negative CF-creativity links [114], while others find no CF-creativity associations [58, 83, 115]. By early adolescence, studies using similar shifting-based tasks (e.g., switching tasks paired with the RAT and Torrance Tests) predominantly find no association [122]. Together, these findings suggest that when CF is operationalised primarily as a shifting ability, its association with creativity is inconsistent and may depend on how creative idea generation is measured across developmental stages.

Our findings also show that the cognitive mechanisms supporting creativity may differ developmentally. In preschool and early childhood stages, divergent thinking responses appear to rely largely on associative semantic fluency and spontaneous idea generation, with minimal demands on controlled retrieval or executive monitoring, which could explain the negative association found between CF and creativity in Vaisarova and Carlson [114]. By adolescence, divergent thinking increasingly requires controlled episodic and semantic retrieval, working-memory maintenance and updating, relational integration, and evaluative monitoring via frontoparietal networks [140–144]. This shift from relying on bottom-up processes in childhood to top-down ones in adolescence is also reflected in the complexity of responses generated. In particular, originality shows the greatest gains from childhood to adolescence, with adults and adolescents generating higher quality, more novel responses than children [145, 146]. Further, in preschool and school-going child samples, CF was primarily assessed using generative tasks (e.g., the Five-Point Test, verbal fluency), which share cognitive demands with divergent-thinking measures of creativity (e.g., TTCT), as both rely on broad retrieval abilities [147]. This could have likely inflated the CF-creativity associations as a result of processing similarity, disguising the deeper developmental mechanisms. In contrast, adolescent studies rely mostly on self-report measures of CF, which capture metacognitive beliefs about flexibility rather than executive performance [62]. These self-report indices correspond poorly with task-based creativity measures as explicated above, which may explain the weaker or null associations observed in older samples. These reasons could account for why CF–creativity associations were more consistently observed in younger children in our review, where both constructs were measured using shifting-based and basic divergent thinking tasks, compared with adolescents engaging in complex or metacognitive creativity measures.

CF as an integrative aspect of cognition involving multiple control components

Studies adopting this conceptualisation typically use multicomponent CF measures such as the WCST, semantic and phonological verbal fluency, or the Five-Point Test [42, 117–119]. Creativity, in turn, in these studies is framed as either a thinking style or cognitive disposition, assessed through Torrance Tests, creative intelligence measures, pattern-meaning tasks, or creative personality scales. The studies generally involve older children aged 8–13, in which the CF–creativity link is consistently found. These relatively salient effects are likely due to the strong conceptual alignment between constructs: CF and creativity are both operationalised as broad, higher-level cognitive constructs rather than narrow, task-specific outcomes.

The observation of a robust CF-creativity link when conceptualised as a systemic property of the cognitive system in middle childhood also coincides with emerging literature showing that this stage marks a period of increasing EF integration, during which children become more capable of coordinating various EF processes within complex cognitive activities. For instance, Brydges et al. [123] conducted a longitudinal study examining children’s performance in inhibition, working memory, and shifting at approximately 8 and 10 years of age. Their findings showed that at age 8, a unitary EF model best explained performance across tasks. By age 10, performance was better explained by a fractionated two-factor model, comprising working memory, and a combined inhibition/shifting factor. This study aligns with the present review that the developmental trajectory of EF begins with a more global control system in early to middle childhood, which then differentiates into more specialised subcomponents in late childhood, closely resembling the EF structure observed in adults, which then fit better with the requirements of higher-level creativity measures.

CF as an ecologically-relevant multidimensional ability suited to real-world contexts

In adolescent samples, CF is most often conceptualised as a multidimensional adaptive ability reflecting metacognitive flexibility and the capacity to adjust thinking in everyday contexts. This approach relies primarily on self-report instruments such as the CFI or Cognitive Flexibility Scale. Creativity is similarly assessed as an applied, real-world ability, using self-report creativity scales or complex problem-solving tasks such as the PISA Oil Spill scenario. Studies in this group show mixed findings: Wang and Chang [116] report a positive association between multidimensional CF and creativity, whereas Caliskan and Altun [121] find no link when creativity was measured through structured analytic problem-solving (e.g., the Oil Spill problem from PISA). These results suggest that ecologically-relevant CF measures such as self-rated assessments tend to correlate with creativity only when creativity tasks also index metacognitive regulation and real-life problem-solving. Overall, these adolescent findings highlight the importance of conceptual alignment: multidimensional, self-perceived CF predicts creativity chiefly when creativity tasks similarly capture flexible thinking in ecologically-grounded contexts.

Limitations and future directions

Several limitations of the present scoping review limit us from establishing a developmental framework of the CF-creativity association. First, substantial conceptual and methodological heterogeneity across studies constrained our ability to draw firm developmental conclusions. CF was defined and operationalised through markedly different approaches: from narrow task-based shifting indices to broad multicomponent EF composites, while creativity was assessed through assessments such as AUT performance, self-rated creativity assessments, pattern completion tasks, or personality-based creativity dispositions. To complicate matters, divergent-thinking (DT) tasks often yield multiple indices, such as fluency, originality, and flexibility, which are frequently analysed as composite scores when examining associations with CF, thereby obscuring potential differences between specific dimensions of creativity.

Such inconsistencies in construct conceptualisation and measurement makes cross-study comparisons challenging and may partly account for variability in reported associations.

Second, most studies relied on laboratory-based behavioural measures that offer limited ecological validity. The criterion validity of neuropsychological tasks for predicting real-world functioning is typically poor to moderate [148, 149], and the association between neuropsychological EF tasks and self-rated assessments is similarly weak [130]. These issues complicate interpretation of CF–creativity associations, especially in studies using single-task indicators of either construct.

Third, evidence is notably sparse for adolescents, with few studies sampling individuals aged 13–18, limiting confidence in claims about developmental transitions in the CF–creativity relation from childhood into adulthood. In addition, the present review is limited in its capacity to determine whether the observed relationships between cognitive flexibility and creativity generalize to neurodivergent populations, in which executive functioning and creative processes may follow distinct developmental trajectories and functional patterns. Future research should systematically examine these associations within neurodivergent groups to clarify whether similar, attenuated, or divergent mechanisms underlie the interplay between cognitive flexibility and creativity across different neurodevelopmental profiles.

Future research should distinguish between creativity dimensions such as fluency, originality, and flexibility when examining CF–creativity relations, as CF may differentially contribute to specific stages of the ideation process. In addition, future research should adopt more theoretically coherent and ecologically-grounded approaches to studying CF–creativity relations across development. First, developmental work should differentiate associative from controlled processes within divergent-thinking tasks as previous evidence clearly shows that children and adolescents/ adults use associative processes and cognitive control processes in divergent thinking respectively [114] through methodological innovations such as process-dissociation paradigms [150] or by applying large language models [151]. Longitudinal and cohort-sequential designs would clarify how increasing executive-system integration across middle childhood and adolescence supports the change from bottom-up associative responding to more top-down controlled task performance. Second, the field would benefit from using aligned CF measures within broader EF batteries and creativity assessments that reduce construct-level heterogeneity. Third, given that self-rated CF and creativity measures often demonstrate stronger correspondence with real-world functioning, there is a growing need to develop tasks that bridge laboratory performance with everyday cognitive–creative demands. Current evidence suggests that neuropsychological tasks show only modest criterion validity for functional outcomes [148, 149] and correlate weakly with self-report assessments [130]. Designing tasks that capture both controlled cognitive operations and their real world manifestations would substantially strengthen the field’s capacity to identify which aspects of CF, and related executive control processes support creative performance outside the laboratory. Such efforts would also guide practitioners in selecting or designing interventions with higher ecological and translational value, ensuring that improvements observed in controlled settings more faithfully generalise to real-world creative behaviour [152].

Conclusion

In sum, this scoping review demonstrates that both CF and creativity are conceptualised and measured heterogeneously across developmental studies, with CF definitions and operationalisations showing clear age-related patterns. CF in preschoolers and younger children is predominantly conceptualised as shifting or set-switching, whereas studies examining CF in late childhood and adolescence apply more integrated and multidimensional conceptualisations. This shift mirrors the broader developmental trajectory of EFs, which become increasingly differentiated into distinct components, including CF. Creativity itself shows less age-specific conceptual variation, though task demands differ substantially across stages, from drawing and imaginative tasks in preschoolers to structured divergent- and convergent-creativity assessments in older children and adolescents. The CF–creativity link is inconsistent, with stronger associations in younger samples potentially inflated by shared task demands, and weaker associations in adolescents partly reflecting differences between self-report and task-based instruments. These findings underscore the need for multimodal, developmentally appropriate, and conceptually aligned assessments to accurately chart CF–creativity relations across preschoolers, school-going children and adolescents. In addition, the findings further emphasise that the CF–creativity association cannot be understood independent of a lifespan framework. Although the studies reviewed span preschool to adolescence, they do not collectively trace developmental continuity. Instead, each age group is characterised by different conceptualisations, task types, and theoretical assumptions, limiting our ability to draw conclusions about how CF–creativity relations unfold across developmental periods. Bringing these findings together therefore highlights a major gap: current evidence does not map onto a coherent developmental trajectory. Rather, it presents fragmented snapshots that likely reflect methodological choices more than true developmental change. Future work grounded in lifespan developmental science through tracing how CF and its relationship to the broader EF system evolve and differentially support creativity, will be essential for building a unified developmental account of the CF–creativity link.

Abbreviations

CF

Cognitive Flexibility

EF

Executive Function

FIST

The Flexible Item Selection Task

DCCS

The Dimensional Change Card Sort Task

DPCM

The Dual Pathway to Creativity Model

WCST

The Wisconsin Card Sorting Test

SVF

The Semantic Verbal Fluency test

PVF

The Phonological Verbal Fluency test

FPT

The Five-Point Test

CFS

The Cognitive Flexibility Scale

CFI

The Cognitive Flexibility Inventory

AUT

The Alternate Uses Task

K-DOCS

The Kaufman Domains of Creativity Scale

CAQ

The Creative Achievement Questionnaire

ICAYA

The Inventory of Creative Activities for Young Adolescents

CASES

The Creative Self-Efficacy Scale for Children and Adolescents

WCST-64

Wisconsin Card Sorting Task-64

UBT

The Unusual Box Test

RAT

The Remote Association Test

Authors’ contributions

Kastoori Kalaivanan drafted the introduction, literature review, and discussion; prepared the original manuscript draft; and revised the manuscript based on feedback. Rui Wang oversaw the overall project administration, resources, and supervision, and validation of the study; coordinated manuscript writing, revision, and submission; collected and analyzed data; and contributed to the writing of the abstract and results as well as review and editing. Jiani Ren collected and analyzed data, drafted the results section, and supported the literature review. Mengquan Liu collected and analyzed data and drafted the methods section. Xueni Zhang contributed to conceptualization and validation of the study, guided the research questions and overall manuscript structure, and assisted with review and editing. Teo Chew Lee contributed to conceptualization, supervision, validation, theoretical grounding, and review and editing. S.H. Annabel Chen contributed to conceptualization, supervision, validation, and review and editing, particularly in relation to research methodology, procedures, and interpretation of findings.All authors read and approved the final manuscript.

Funding

This work was supported by grants to the NTU-Cambridge Centre for Lifelong Learning and Individualised Cognition (CLIC), a project by the National Research Foundation, Prime Minister’s Office, Singapore, under its Campus for Research Excellence and Technological Enterprise (NRF-CREATE SoL) Programme with the funding administered by the Cambridge Centre for Advanced Research and Education in Singapore Ltd. (CARES) and housed at the Centre for Research and Development in Learning (CRADLE@NTU).

Data availability

The datasets analysed during the current study consist of published articles retrieved from publicly accessible bibliographic databases. No new datasets were generated. All search strategies and screening procedures are documented in the manuscript and supplementary materials are uploaded to Open Science Framework. The bibliographic data were obtained from PubMed, Web of Science, ERIC, and APA PsycINFO, which are accessible to users through institutional or individual subscriptions. Extracted data used for analysis in this review are available from the corresponding author upon reasonable request.

Declarations

Ethics approval and consent to participate

Not applicable.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.