Uncovering executive function profiles within interindividual variability: A data driven clustering exploration of design fluency in school-aged children

Uncovering executive function profiles within interindividual variability: A data driven clustering exploration of design fluency in school-aged children | bioRxiv /* */ /* */ <!-- <!-- /*! * yepnope1.5.4 * (c) WTFPL, GPLv2 */ (function(a,b,c){function d(a){return"[object Function]"==o.call(a)}function e(a){return"string"==typeof a}function f(){}function g(a){return!a||"loaded"==a||"complete"==a||"uninitialized"==a}function h(){var a=p.shift();q=1,a?a.t?m(function(){("c"==a.t?B.injectCss:B.injectJs)(a.s,0,a.a,a.x,a.e,1)},0):(a(),h()):q=0}function i(a,c,d,e,f,i,j){function k(b){if(!o&&g(l.readyState)&&(u.r=o=1,!q&&h(),l.onload=l.onreadystatechange=null,b)){"img"!=a&&m(function(){t.removeChild(l)},50);for(var d in y[c])y[c].hasOwnProperty(d)&&y[c][d].onload()}}var j=j||B.errorTimeout,l=b.createElement(a),o=0,r=0,u={t:d,s:c,e:f,a:i,x:j};1===y[c]&&(r=1,y[c]=[]),"object"==a?l.data=c:(l.src=c,l.type=a),l.width=l.height="0",l.onerror=l.onload=l.onreadystatechange=function(){k.call(this,r)},p.splice(e,0,u),"img"!=a&&(r||2===y[c]?(t.insertBefore(l,s?null:n),m(k,j)):y[c].push(l))}function j(a,b,c,d,f){return q=0,b=b||"j",e(a)?i("c"==b?v:u,a,b,this.i++,c,d,f):(p.splice(this.i++,0,a),1==p.length&&h()),this}function k(){var a=B;return a.loader={load:j,i:0},a}var l=b.documentElement,m=a.setTimeout,n=b.getElementsByTagName("script")[0],o={}.toString,p=[],q=0,r="MozAppearance"in l.style,s=r&&!!b.createRange().compareNode,t=s?l:n.parentNode,l=a.opera&&"[object Opera]"==o.call(a.opera),l=!!b.attachEvent&&!l,u=r?"object":l?"script":"img",v=l?"script":u,w=Array.isArray||function(a){return"[object Array]"==o.call(a)},x=[],y={},z={timeout:function(a,b){return b.length&&(a.timeout=b[0]),a}},A,B;B=function(a){function b(a){var a=a.split("!"),b=x.length,c=a.pop(),d=a.length,c={url:c,origUrl:c,prefixes:a},e,f,g;for(f=0;f<d;f++)g=a[f].split("="),(e=z[g.shift()])&&(c=e(c,g));for(f=0;f<b;f++)c=x[f](c);return c}function g(a,e,f,g,h){var i=b(a),j=i.autoCallback;i.url.split(".").pop().split("?").shift(),i.bypass||(e&&(e=d(e)?e:e[a]||e[g]||e[a.split("/").pop().split("?")[0]]),i.instead?i.instead(a,e,f,g,h):(y[i.url]?i.noexec=!0:y[i.url]=1,f.load(i.url,i.forceCSS||!i.forceJS&&"css"==i.url.split(".").pop().split("?").shift()?"c":c,i.noexec,i.attrs,i.timeout),(d(e)||d(j))&&f.load(function(){k(),e&&e(i.origUrl,h,g),j&&j(i.origUrl,h,g),y[i.url]=2})))}function h(a,b){function c(a,c){if(a){if(e(a))c||(j=function(){var a=[].slice.call(arguments);k.apply(this,a),l()}),g(a,j,b,0,h);else if(Object(a)===a)for(n in m=function(){var b=0,c;for(c in a)a.hasOwnProperty(c)&&b++;return b}(),a)a.hasOwnProperty(n)&&(!c&&!--m&&(d(j)?j=function(){var a=[].slice.call(arguments);k.apply(this,a),l()}:j[n]=function(a){return function(){var b=[].slice.call(arguments);a&&a.apply(this,b),l()}}(k[n])),g(a[n],j,b,n,h))}else!c&&l()}var h=!!a.test,i=a.load||a.both,j=a.callback||f,k=j,l=a.complete||f,m,n;c(h?a.yep:a.nope,!!i),i&&c(i)}var i,j,l=this.yepnope.loader;if(e(a))g(a,0,l,0);else if(w(a))for(i=0;i (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0];var j=d.createElement(s);var dl=l!='dataLayer'?'&l='+l:'';j.src='//www.googletagmanager.com/gtm.js?id='+i+dl;j.type='text/javascript';j.async=true;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-M677548'); Skip to main content Home About Submit ALERTS / RSS Search for this keyword Advanced Search New Results Uncovering executive function profiles within interindividual variability: A data driven clustering exploration of design fluency in school-aged children View ORCID Profile Myriam Sahraoui , View ORCID Profile Karim Jerbi , View ORCID Profile Vanessa Hadid , View ORCID Profile Bruno Gauthier doi: https://doi.org/10.1101/2025.08.19.670430 Myriam Sahraoui 1 Département de Psychologie, Université de Montréal, McGill University , Montreal, Canada 2 Cognitive and Computational Neuroscience Lab (CoCo lab), McGill University , Montreal, Canada 3 Laboratoire d’études en neuropsychologie de l’enfance et de l’adolescence (LÉNEA), McGill University , Montreal, Canada Find this author on Google Scholar Find this author on PubMed Search for this author on this site ORCID record for Myriam Sahraoui For correspondence: myriam.sahraoui{at}umontreal.ca Karim Jerbi 1 Département de Psychologie, Université de Montréal, McGill University , Montreal, Canada 2 Cognitive and Computational Neuroscience Lab (CoCo lab), McGill University , Montreal, Canada Find this author on Google Scholar Find this author on PubMed Search for this author on this site ORCID record for Karim Jerbi Vanessa Hadid 1 Département de Psychologie, Université de Montréal, McGill University , Montreal, Canada 2 Cognitive and Computational Neuroscience Lab (CoCo lab), McGill University , Montreal, Canada 4 McGill University Health Centre, McGill University , Montreal, Canada Find this author on Google Scholar Find this author on PubMed Search for this author on this site ORCID record for Vanessa Hadid Bruno Gauthier 1 Département de Psychologie, Université de Montréal, McGill University , Montreal, Canada 3 Laboratoire d’études en neuropsychologie de l’enfance et de l’adolescence (LÉNEA), McGill University , Montreal, Canada Find this author on Google Scholar Find this author on PubMed Search for this author on this site ORCID record for Bruno Gauthier Abstract Full Text Info/History Metrics Preview PDF Abstract This study applied unsupervised machine learning to performances on a design fluency task, to identify distinct executive function (EF) profiles among 113 neurotypical Canadian children (62% female, 74% White, aged 7-13). By tracking design, repetition, and strategy production every minute for five minutes, the results revealed two profiles: Profile A generated fewer designs but exhibited a more stable production pattern across time. Profile B, initially more productive, showed a steeper increase in repetition errors and a decline in design production. These findings demonstrate how EF interindividual variability in neurotypical children can be captured through naturally emerging performance patterns, highlighting the value of temporal analysis in differentiating executive functioning profiles. Main Executive functions (EFs) consist of a set of cognitive processes essential for self-regulation, purposeful behavior, and successful adaptation to new or complex tasks. These cognitive abilities, including inhibition, cognitive flexibility, working memory, and planning ( Miyake & Friedman, 2012 ), are especially important during childhood as they support critical developmental outcomes spanning academic achievement, social skills, emotional regulation, and long-term cognitive and adaptive functioning ( Judd et al., 2024 ; Kamphorst et al., 2021 ; Nesayan et al., 2019 ). Although substantial interindividual variability in executive functioning has been observed even among typically developing children ( Kochhann et al., 2018 ; Mareva & Holmes 2024 ; Muñoz-Parreño et al., 2023 ), relatively few studies have investigated whether distinct profiles of executive functioning occur naturally within this population. Identifying such cognitive profiles could clarify how executive processes interact during tasks, potentially informing individualized educational and developmental strategies ( Astle et al., 2019 ); Judd et al., 2024 ). Traditional approaches for evaluating EFs typically rely on extensive neuropsychological test batteries designed to measure discrete cognitive components in isolation. While these methods are effective in assessing specific executive components and predicting cognitive outcomes ( Bablekou et al., 2023 ; Chen et al., 2022 ; Scope et al., 2010 ), they often fall short in capturing the complexity and interplay of EFs as they occur online in realistic settings ( Baez et al., 2020 ; Gauthier et al., 2016 ; Litkowski et al., 2020 ; Sandilos et al., 2019 ). This limitation stems in part from the fact that any cognitive task tends to recruit multiple executive functions — and often engages broader cognitive processes as well — making it difficult to isolate specific constructs, a challenge commonly referred to as test impurity (see Suchy et al., 2017 ). Additionally, global (composite) scores are often used in these studies in an attempt to capture a broad range of information across multiple cognitive domains. However, these scores often present a somewhat static view of cognition. In fact, as demonstrated by Gauthier et al. (2016) , process measures - i.e. measurements taken across time - provide valuable insights into the dynamic production changes that occur during task execution, as executive functions are specifically associated with these dynamic changes ( Tessier et al., 2022 ). Furthermore, traditional approaches to EFs assessment usually depend on predefined normative criteria, which may obscure subtle yet meaningful qualitative differences among neurotypical children ( Astle et al., 2019 ). Consequently, previous literature investigating executive profiles in childhood has yielded heterogeneous results, identifying anywhere from two to six distinct cognitive profiles ( Astle et al., 2019 ; Bablekou et al., 2023 ; Muñoz-Parreño et al., 2023 ; Younger et al., 2024 ). For instance, Muñoz et al. (2023) found two EF profiles in neurotypical children aged ten to thirteen years old, distinguishing between generally higher and lower performers across tasks. Similarly, Bablekou et al. (2023) identified two profiles in preschool and early primary school children. One was marked by lower task performance but overestimated metacognitive judgments, and the other by higher performance but underestimated judgments. In contrast, Younger et al. (2024) identified six EF profiles among children aged 7 to 15, including low, mixed, and high EF performance, as well as three profiles distinguished by either strong flexibility, working memory, or attention. Astle et al. (2019) also reported six profiles in a broader age range (5 to 18 years), but this time including a neurodiverse sample. These inconsistencies in methodology and results highlight the need for a more flexible approach that can better capture the dynamic and heterogeneous nature of executive functioning in children. It should also overcome key methodological challenges, such as global composite scores, time-intensive administration, and reliance on a priori hypotheses ( Miller & Barr, 2017 ; Péron, 2024 ). Visuospatial design fluency tasks have emerged as an effective assessment tool for examining executive function in children ( Fournier et al., 2020 ; Gauthier et al., 2016 ; Stievano & Scalisi, 2016 ). During these tasks, children are asked to generate as many novel patterns as possible within parameters that vary depending on the test, within a limited time period. Performance in these tasks engages multiple executive functions simultaneously: cognitive flexibility (creating new patterns), inhibition (avoiding repeated patterns), planning (strategically applying modifications), and working memory (keeping track of previously generated patterns in the absence of visual reminders; Stievano & Scalisi, 2016 ). According to Goebel et al. (2009) , Lezak (1995) , and Stievano and Scalisi (2016) , a higher percentage of strategies in design fluency tasks leads to a higher production of total designs and a lower production of repeated designs. For example, Bolduc (2023) recently provided evidence suggesting that the number of strategies, as well as the number of total and correct designs, were significantly associated with verbal fluency and cognitive flexibility, while the percentage of repetitions was inversely related to visual working memory. This relation was also found in Stievano and Scalisi (2016) , as well as an inverse relation with visual working memory, problem-solving, and visual-motor integration. Furthermore, the results of Bolduc (2023) suggest that the total number of correct unique designs produced in a tablet-based design fluency task may serve as an integrated measure of executive functioning, given its convergence with established tasks such as the Stroop, Corsi blocks, and the Tower of London. Although research has primarily focused on using design fluency tasks to differentiate clinical groups, the potential of this task extends further when combined with advanced analytical methods, such as unsupervised clustering. For instance, Gauthier et al. (2016) employed Self-Organizing Maps (SOM; Kohonen, 1982 ), an unsupervised clustering technique, to uncover distinct performance profiles in children with and without ADHD. While the total number of designs produced did not differentiate the groups, their time-based production patterns revealed meaningful distinctions. This highlights how combining design fluency tasks with machine learning can not only help in differentiating clinical and non-clinical populations, but also be used to explore interindividual variability within neurotypical children, to reveal distinct profiles that reflect different modes of executive functioning, and to provide deeper insights into individual cognitive strategies. However, whether similar unsupervised methods can reveal naturally occurring executive function profiles among neurotypical children remains unexplored. More broadly, unsupervised clustering algorithms, such as K-means, offer a powerful alternative for identifying cognitive subgroups without relying on predefined classifications ( Shah et al., 2024 ; Stamatis et al., 2021 ). These methods have been used in neuropsychological research to uncover latent cognitive patterns that traditional methodologies might overlook and to classify cognitive subtypes in clinical populations. While clustering techniques have proven effective in differentiating clinical and neurotypical groups (e.g. Gauthier et al., 2016 ), a parallel stream of research has also explored the use of ML-based clustering approaches across multiple EF tasks to classify executive profiles in neurotypical children. For instance, Usai et al. (2018) applied a data-driven clustering approach to six EF tasks in five-year-olds, identifying four distinct EF profiles (i.e. optimal, typical, low working memory, and deficient), while Martarelli et al. (2018) used a similar approach with four EF tasks and additional measures of social functioning in children aged 6–8 years, also revealing four cognitive profiles. Although previous work illustrates the potential of ML for characterizing EF interindividual variability, to our knowledge, no study has explored whether a single EF task, such as a design fluency task, can be used with unsupervised clustering to identify cognitive profiles in neurotypical children. To address this, the current study applies K-means clustering to performance data from an app-based version of a design fluency task to uncover naturally occurring profiles. We hypothesized that distinct EF profiles would emerge, possibly reflecting variability in cognitive control strategies and executive resource allocation rather than a simple dichotomy of high versus low performers. Specifically, we expected that while fluency would differentiate profiles, the underlying executive mechanisms—such as working memory, cognitive flexibility, and strategy adaptation—would be key in distinguishing them. Given the role of proportional measures in fluency-based tasks, we anticipated that differences in how children regulate strategy use and suppress repetition errors over time would drive performance variations. Finally, although EFs generally improve with age, we expected that chronological age alone would not fully explain the observed clusters, as individual differences in cognitive control mechanisms would likely play a more critical role in shaping these profiles. Methods Participants The data of 113 children (70 girls) aged 7 to 13 years old (M=10.28, SD=1.34) were included in this study. Of the participants, 74% were White, 5% were Black, 4% were Hispanic, 3% were Arabic, 2% were First Nations, 2% were Asian, and 10% identified as other. The participants data came from a previous study in the Laboratoire d’études en neuropsychologie de l’enfance et de l’adolescence (LÉ NEA), (see: Laniel & Gauthier, 2022 ). Three cognitive assessment tests were included: the current design fluency task, the WISC-4 Vocabulary Test, and the WISC-4 Matrix Test (See Measures section below for details). All neurotypical participants who completed the design fluency task were included in the present study. To be included in the study, the participants had to be schooled in French since the beginning of their education and speak French at home. The exclusion criteria were: (1) a grade under 60% in French on their school report card, (2) repeating a school year, (3) being followed by a speech therapist or a remedial teacher for written language, (4) a history of brain trauma or a neurodevelopmental disorder diagnosis. Participants who scored more than two standard deviations below the mean for their age in reading ability or intellectual quotient (IQ) were also excluded. Additionally, participants with inattention or hyperactivity and impulsivity scores more than two standard deviations above the mean were excluded ( Laniel & Gauthier, 2022 ). Procedure The data used in this study was approved by the comité d’éthique et de la recherche en arts et sciences de l’Université de Montréal (n°: CERAS-2015-16-080-P). Participants were recruited in different primary schools and day camps of six different regions in the province of Quebec, Canada. Parents were given a solicitation form, and if they were interested in participating in the study, they were contacted to verify if they were eligible and make an appointment. Participation was on a voluntary basis. All children completed a battery of cognitive tests (see Laniel and Gauthier (2022) for more details). During testing, the parents were in the waiting room and completed multiple questionnaires. The administration of the design fluency task was always performed first in the assessment. For the administration of the subtests vocabulary and matrix of the WISC-IV, the standard administration procedure was followed. Measures AppFLO – Design fluency The Design fluency task used in this study is a beta version of the tablet-based AppFLO test, developed in the LÉNEA. In this task, five dots are displayed on the screen, arranged in a standardized configuration resembling the face of a die. The size and placement of the dots remain constant across all devices used for assessment. Using their index finger, participants are instructed to draw a design by connecting at least two dots with straight lines. Once a design is completed, they tap an arrow on the right side of the screen to proceed to the next design. The AppFLO application automatically records the following three variables ( Bolduc, 2023 ) that were used in the present study’s analyses: Total number of designs : All designs produced by the child. This is considered a global measure of executive functioning. Number of repetitions : All designs that were repeated. This measure is associated with visual working memory. Number of strategies : The number of strategic modifications applied across designs, including additions subtractions (adding or removing a line relative to the previous design), rotations (turning a previous design), and reflections (mirroring a previous design). This measure reflects planning and cognitive flexibility. All these metrics were recorded at one-minute intervals throughout the task. A difference of AppFLO compared to other design fluency tasks, is that only the design currently being produced is visible to the participant, which increases the demands on visual working memory ( Bolduc, 2023 ). Figure 1 provides illustrative examples of four designs produced by a participant. Download figure Open in new tab Figure 1 Examples of correct designs completed by a participant on AppFLO To account for individual differences in total task productivity, two percentage-based metrics were computed to normalize strategic modifications, repetitions, and unique production relative to total design production. These metrics provide a more precise assessment of executive control, ensuring that variations in overall output do not confound interpretations of strategy use or repetition tendencies. (i) Percentage of repetitions : (Total repetitions/Total designs) × 100 (ii) Percentage of strategies : (Total strategies/Total designs) × 100 Proportional measures such as these have been widely used in Design fluency research to assess executive function performance ( Goebel et al., 2009 ; Stievano & Scalisi, 2016 ). By normalizing repetitions and strategic modifications relative to total designs, percentage-based metrics allow for a fairer comparison across participants and time points. Wechsler Intelligence Scale for Children - Fourth Edition (WISC-IV) The WISC-IV Wechsler (2003) is a standardized test used in children aged 6 to 17 years to assess intellectual quotient (IQ). Two subtests of the scale were used to create an IQ estimate for the analysis: (1) Vocabulary consisting of asking the participant to explain the meaning of different words, (2) Matrix Reasoning consisting of showing participants a matrix with a pattern and asking them to choose an image to complete a missing part of the pattern. Both subtests have strong validity and fidelity ( Wechsler, 2003 ) and were chosen based on their high correlation with their indices (verbal comprehension and non-verbal reasoning), as well as with global IQ. Main Analyses To determine if the dataset was prone to clustering, a Hopkin test was performed as it offers a quick way of establishing the distribution of the dataset pre-clustering. This test compares the data distribution with a randomly generated dataset to evaluate the likelihood of clusters forming. An H-value close to 0 indicates uniform data, whereas a value closer to 1 suggests a tendency toward clustering ( Wright, 2022 ). For clustering analysis, we used K-means clustering, which aims to maximize intergroup distance while minimizing intragroup variance. In the parameters, a k-means++ initialization was used to minimize randomness by selecting initial centroids based on a probability distribution. The silhouette coefficient was used to assess the quality of the output clusters, and to define the optimal number of clusters for the data. The Hopkins test, K-means, and silhouette coefficient were implemented with Python using the Scikit-learn library ( Pedregosa et al., 2011 ). Permutation tests To assess differences between the possible clusters, and since our data was not normally distributed, permutation tests were used to compare executive function metrics across clusters. Permutation testing was chosen as it provides a non-parametric alternative to traditional hypothesis testing, reducing assumptions about data distribution. For each permutation test, the observed difference between the clusters were calculated, followed by 10,000 resampled iterations where cluster labels were shuffled. A two-sided p-value was computed by determining the proportion of permutations that produced a difference greater than or equal to the observed difference. Permutation tests were conducted for the following metrics, both globally (collapsed across time points) and at each individual time point (T1–T5): (1) Total number of designs (overall productivity in the task). (2) Total number of repetitions (raw count of repeated designs). (3) Total number of strategies (raw count of strategic modifications: additions, subtractions, rotations, reflections). (4) Percentage of repetitions (proportion of repeated designs, indexed to total designs, reflecting inhibitory control). (5) Percentage of strategies (proportion of modifications, indexed to total designs, reflecting cognitive flexibility). The temporal fluctuations of performance were explored by applying permutation tests at each time point (T1–T5) to assess when, during the task, differences emerged. Additionally, permutation tests were used to compare age distributions between clusters, ensuring that observed differences in executive function profiles were not driven by age-related effects. Similarly, Mann-Whitney U tests were used to compare sex distributions, verbal, and non-verbal IQ estimate scaled scores between clusters. These tests were used to determine whether the proportion of repetitions and strategies differed between clusters, as well as to examine whether executive function profiles varied based on age, sex, or cognitive abilities. Correlation analyses To examine the relation between executive function performance and age, Spearman rank-order correlations were conducted separately for each cluster. This non-parametric approach was selected for its robustness against non-normal distributions. Correlations were computed using both raw measures (total number of designs, repetitions, and strategies) and derived metrics (percentage of repetitions and percentage of strategies) collapsed across time points. Results Pre-clustering Analysis A Hopkins statistic test was performed to evaluate the clustering tendency of the data set, averaging 1,000 iterations to account for variability. The Hopkins value obtained (H = 0.69) confirmed the presence of distinct but overlapping clusters (H = 0 indicates random data, while values closer to 1 indicate a tendency to exhibit clusters). Clustering and Cluster Validation The optimal number of clusters was determined using the K-means algorithm, with the silhouette coefficient used to assess the separation between clusters (tested across 2 to 10 clusters). The highest silhouette coefficient (s = 0.37) indicated a moderate level of cohesion and separation, supporting a two-cluster solution. Given this result, a two-group clustering solution was retained for further analysis. To aid visualization, Principal Component Analysis (PCA) was applied independently to the same dataset used for K-means clustering. The clusters were then projected onto the first two principal components, as shown in Figure 2 Download figure Open in new tab Figure 2 Note . Cluster separation of the K-means, displayed on two components PCA. Cluster 1 is colored in blue, Cluster 2 is colored in pink, and the centroids of each cluster is marked by an X. The plot shows a distinct separation between the clusters on 70% of the variance, illustrating the profiles identified. Cluster Differences Permutation tests were conducted to compare the two clusters across key executive function measures, including total designs, repetitions, and strategies, both globally and at individual time points. Table 1 presents the descriptive statistics for each metric, along with test statistics (tperm) and significance values. Cluster 2 exhibited significantly higher values across time for total designs, repetitions, and strategies. View this table: View inline View popup Table 1 Total Number of Designs On average, participants in Cluster 2 consistently produced more designs than those in Cluster 1 throughout the task (tperm = −6.23, p < 0.001), with differences emerging at T1 and persisting across all time points. At T1, Cluster 2 generated 15 ± 0.6 designs, significantly more than the 6.8 ± 0.4 produced by Cluster 1 (tperm = −8.20, p < 0.001). Cluster 2 exhibited a gradual decline in the number of designs produced over time, decreasing from 15 ± 0.6 at T1 to 10.2 ± 0.5 at T5. In contrast, Cluster 1 showed a more stable pattern, producing between 5.1 and 6.9 designs across time. The difference between clusters remained significant at all time points (tperm range: −8.20 to −5.03, all p < 0.001) ( Figure 3 ), indicating a consistently higher design production in Cluster 2 despite a progressive reduction across the task. Download figure Open in new tab Figure 3 Note. Comparison of the two clusters on the total number of designs, repetitions, and strategies across time. Cluster 2 consistently showed significantly higher values than Cluster 1 in total designs, repetitions, and strategies. Total Number of Repetitions Participants in Cluster 2 produced, on average, three times as many repetitions as those in Cluster 1 (3.35 vs. 1.11). This difference was statistically significant overall (tperm = −2.23, p < 0.001), with differences already apparent at T1. At T1, Cluster 2 produced 1.3 ± 0.3 repetitions, significantly more than Cluster 1 (0.5 ± 0.2, tperm = −0.63, p = 0.0074). Repetition counts remained relatively stable across time for Cluster 1, ranging from a minimum of 0.5 ± 0.2 at T1 to a maximum of 1.6 ± 0.3 at T5. In contrast, Cluster 2 showed an increase in repetitions across time, with values rising from a minimum of 1.3 ± 0.3 at T1 to a maximum 4.4 ± 0.5 at T5. The disparity between clusters became more pronounced from T2 onward, where Cluster 2 produced consistently higher repetition counts (tperm range: −2.13 to −3.11, all p < 0.001) ( Figure 3 ). Total Number of Strategies The total number of strategies employed was significantly higher in Cluster 2 across the task (tperm= −2.27, p < 0.001), with the largest differences emerging early on. At T1, Cluster 2 applied 3.6 ± 0.4 strategies, more than twice the number used by Cluster 1 (1.3 ± 0.2, tperm = −3.26, p < 0.001). Strategy use remained stable for Cluster 1 across time, ranging from 1.0 to 1.74 strategies per time point. In contrast, Cluster 2 exhibited a decline, dropping from 3.6 ± 0.4 at T1 to 2.4 ± 0.3 at T5. While Cluster 2 continued to use more strategies overall, the difference between clusters gradually diminished over time (tperm values ranging from −2.76 at T2 to −1.45 at T5, all p < 0.001), indicating that the initial advantage in strategic modifications was reduced by the end of the task. Age, Sex, and IQ Between-Group Comparisons Permutation t-tests revealed no significant differences between clusters in age (Cluster 1 : M = 121 months ± 16; Cluster 2 : M = 127 months ± 15), sex (Cluster 1 : 78.1% female, Cluster 2: 62.5% female) or non-verbal IQ (Cluster 1: M = 10.34, SD = 3.29; Cluster 2: M = 10.88, SD = 2.61; estimated by the WISC-IV Matrices scaled score). However, a small but statistically significant difference was found in verbal IQ (estimated by the WISC-IV Vocabulary scaled score), with Cluster 2 scoring slightly higher (tperm = −1.13, p = 0.047). Percentage-Based Metrics To account for differences in total design production, percentage-based metrics were analyzed, offering a normalized comparison of repetition and strategy use between clusters ( Figure 4 ). While absolute counts of repetitions and strategies provided important insights into group-level differences, percentage scores allowed us to assess EF relative to overall task performance. Download figure Open in new tab Figure 4 Percentage-based performance metrics per cluster across time Note . Percentage-based metrics for repetitions and strategies across time points (T1–T5) for Cluster 1 and Cluster 2. Panel (a) shows the percentage of repetitions, with Cluster 2 exhibiting a significantly higher rate than Cluster 1, particularly from T2 onward. Panel (b) displays the percentage of strategies, with Cluster 2 initially using more strategies but showing a diminishing advantage over time. Statistical significance is indicated for each comparison, with tperm values reported for each time point. Percentage of Repetitions Cluster 2 exhibited a significantly higher proportion of repetitions across the task (tperm = −9.10, p < 0.001), with the most pronounced differences occurring from T2 onward. At T2, Cluster 2 showed a repetition rate of 22.5% ± 1.4, compared to 14.5% ± 1.2 in Cluster 1 (tperm = −7.98, p = 0.0305). This gap widened by T5, where Cluster 2 exhibited 41.9% ± 2.1 repetitions significantly higher than Cluster 1 (29.5% ± 1.8, tperm = −12.40, p = 0.0139) ( Figure 4a ). Percentage of Strategies Cluster 2 employed overall a significantly higher percentage of strategies (tperm = −9.01, p < 0.001), but this advantage diminished over time. At T1, Cluster 2 used strategies in 28.3% ± 1.5 of their designs, compared to 19.4% ± 1.3 in Cluster 1 (tperm = −16,63, p < 0.001). However, by T4 and T5, the percentage of strategies no longer differed significantly between clusters ( Figure 4b ). This shift highlights a divergence in performance patterns: while Cluster 2 initially employed strategies at a higher rate, this advantage did not persist throughout the task. Age-Related Correlations Spearman correlations were conducted separately for each cluster to examine relations between age and EF metrics, collapsed across time points. In Cluster 1, a significant positive correlation was found between age and the percentage of strategies (r = 0.28, p = 0.015), suggesting that older children were more likely to apply strategic modifications relative to their total designs. In Cluster 2, no significant correlations were found between age and EF metrics when collapsed across time points. Despite these age-related trends within Cluster 1, age did not significantly differ between Cluster 1 and Cluster 2 ( p > 0.05), suggesting that the observed differences in EF metrics were unlikely to be driven solely by age. Discussion Understanding EF interindividual variability in typically developing children remains a challenge, as traditional assessment methods often impose predefined classifications that may overlook meaningful individual differences. This study addressed this gap by applying a data-driven clustering approach to a single robust EF task, the design fluency task, to reveal naturally occurring EF profiles. Unlike previous studies that relied on composite EF scores or tested multiple executive functions separately, our approach used a single robust EF task and an unsupervised clustering method to uncover naturally occurring cognitive profiles. This allowed us to examine how fluency, inhibition, working memory, and strategy usage interact over time without imposing predefined categories, providing a more process-oriented view of executive control. As hypothesized, distinct EF profiles emerged, revealing differences in cognitive control strategies and executive resource allocation. This finding suggests that EF variability in children is not simply a matter of high vs. low difference, but may reflect different EFs dynamics, emphasizing the need for process-based assessments over static outcome measures. The results revealed two distinct EF profiles, differing not only in fluency but also in how executive control evolved over time. While fluency levels varied between clusters, profiles were best characterized by how the performance of children evolved over time. This finding underscores that executive efficiency is not solely a function of overall output but depends on how fluency is dynamically regulated across the task. To determine whether naturally occurring clusters were present, we applied the Hopkins statistic, which confirmed a tendency toward clustering. K-means clustering was then implemented, with the silhouette coefficient supporting a two-cluster solution. Although the clusters appeared relatively well-separated on 70% of the variance, the observed overlap suggests that EF variability is better conceptualized as a continuum rather than within discrete categories. We retained two clusters to optimize within-group cohesion and between-group differentiation while maintaining interpretability. Prior research using unsupervised clustering has identified varying numbers of EF profiles. However, methodological differences may account for these inconsistencies, as previous studies often included broader age ranges, neurodivergent samples, or multiple EF tasks ( Astle et al., 2019 ). By applying a parsimonious clustering approach to a single, robust EF task, our study was able to isolate meaningful variations in executive control strategies without introducing confounds from task or participant heterogeneity. Profile A (Cluster 1) showed a controlled, accuracy-focused approach, with lower initial fluency (fewer total designs) and a gradual increase in repetition errors (percentage of repetitions) over time. While cognitive flexibility (percentage of strategies) remained lower than Profile B’s, it increased progressively compared to their own early performance (T1). Profile B (Cluster 2) exhibited a high-output approach early in the task, generating more designs but with a steeper increase in repetition errors (percentage of repetitions) and a progressive decline in cognitive flexibility and planning (percentage of strategies). These findings confirm that EF should be understood as a time-sensitive, adaptive process rather than a static ability measured at a single point in time ( Gauthier et al., 2016 ; Stievano & Scalisi, 2016 ). Profile Differences in Executive Control Tradeoffs A central finding of this study is that the two profiles illustrate distinct approaches to the speed-accuracy tradeoff, revealing different trajectories of executive functioning during the task. The speed-accuracy tradeoff refers to the well-established principle that more accurate responses typically require more time, necessitating a balance between speed and accuracy that can vary across both individuals and situations ( Heitz, 2014 ). Profile B consistently generated a higher number of designs than Profile A, accompanied by a steeper increase in relative repetitions over time, suggesting a preference for speed. In contrast, Profile A produced significantly fewer designs but showed a more controlled increase in repetition errors, suggesting a focus towards their repetition monitoring over global production. These divergent patterns suggest that Profile B prioritized fluency and speed, potentially at the expense of monitoring their responses. This interpretation is supported by a sharper increase in relative repetitions in Profile B, as well as a gradual decline in the percentage of strategies used over time. Meanwhile, Profile A produced significantly fewer designs but showed a more progressive increase in the percentage of strategies and repetitions, despite fluctuations at each time point. This trajectory may reflect a more developmentally driven approach to executive control, in line with models emphasizing the evolving interplay between fluency, flexibility, planning abilities and inhibition ( Dumont et al., 2022 ; Goodgold-Edwards et al., 1997 ; Rommelse et al., 2016 ). The inhibitory resource depletion hypothesis offers an additional explanation for this pattern, suggesting that sustained cognitive effort gradually depletes attentional and inhibitory resources, increasing reliance on automatic responses and, consequently, repetition errors ( Dumont et al., 2022 ; Poreh, 2022 ). This was particularly evident in Profile B’s rapid rise in repetition errors, which became significantly higher than those of Profile A by T2. In contrast, the behavior of Profile A participants suggested greater caution, with only a modest rise in repetition errors and a steadier increase in strategic productions. Rather than relying on high speed or fluency, their tendency to maintain more stable output and strategy use relative to total designs was a key factor in sustaining their performance over time. Interestingly, the percentage of strategies used in Profile A showed a significant moderate positive correlation with age, while no such relation was observed in Profile B. This suggests that executive functions in Profile A follow a more typical developmental trajectory, with older children in this cluster demonstrating greater strategic behaviors, as reflected in their increasing percentage of strategies over time. This finding aligns with models of EF maturation, which propose that as children develop, they refine their ability to strategically allocate cognitive resources, balancing fluency, inhibition, and flexibility more effectively (Best & Miller, 2010). In contrast, Profile B’s lack of correlation with age suggests that their executive functioning may be less aligned with typical age-related cognitive maturation. Instead, their performance could reflect greater variability in self-regulation, task engagement, or cognitive control challenges specific to high-output performances (i.e. a higher production early during the task likely increases the cognitive load and monitoring demand). Although some studies have reported associations between age and EF task performances (Best & Miller, 2010; Klenberg et al., 2001 ), findings remain mixed regarding cognitive profiles and design fluency. For example, Regards et al. (1982) found a positive association between age and design fluency, and Gabrig et al. (2018) reported significant differences between EF profiles and age. However, other studies (e.g. Astle et al., 2019 ; Hurks, 2013 ; Stievano & Scalisi, 2016 ) have found only weak or no correlation between age and performance, or age and cluster separation. In the current study, chronological age alone did not explain profile differences, suggesting that individual differences in design fluency tasks may play a more significant role than age itself in shaping performance trajectories. Similarly, the relations between IQ and profiles was limited, with a small but significant effect observed only for Verbal IQ. These findings align with prior research that has shown mixed results. Some studies report a link between IQ and design fluency (e.g. Goebel et al., 2009 ; Regard et al., 1982 ), while others find no significant effects ( Hurks, 2013 ; Tucha et al., 2012 ). The literature is more consistent concerning sex, with most studies reporting negligible or no effects of sex, on EF and design fluency tasks ( Albert et al., 2009 ; Bolduc & Gauthier, 2023; Klenberg et al., 2001 ). Our results also supports this lack of difference. Taken together, these results indicate that inter-individual variability plays a more central role in the differentiation of EF profiles. Executive Efficiency Under Working Memory Constraints A key factor contributing to the difference between the profiles may be the increased working memory demands imposed by the task format. Unlike traditional fluency tasks, where participants can visually track their previous designs, the tablet-based version displayed one design at a time, requiring children to rely entirely on working memory to monitor their responses and avoid repetitions. This feature introduces a higher cognitive load, especially for participants producing a greater number of designs. Because Profile B consistently generated more designs than Profile A, they had more past responses to monitor, increasing the load on working memory resources. Their steeper rise in relative repetitions over time suggests that they may have struggled to manage the growing cognitive load, leading to more repetition errors. This aligns with research showing that when working memory becomes overloaded, impulse regulation weakens, leading to performance declines on sustained cognitive tasks ( Cabbage et al., 2017 ). Conversely, Profile A’s ability to sustain production over may be explained by a combination of relatively lower cognitive load and distinct strategic adaptation. Their more gradual increase in relative repetitions and more stable strategy usage indicate that they may have regulated cognitive load differently, which could have helped prevent a sharp decline in unique designs over time. Interestingly, Profile A’s Percentage of strategies fluctuated at each time point, rather than following a strictly linear trajectory. This pattern suggests that Profile A was actively adjusting their cognitive strategies, alternating between periods of higher cognitive flexibility (more strategies) and more conservative execution (lower strategy use, higher inhibition). In contrast, Profile B exhibited a more consistent trajectory, with fewer fluctuations in strategy use. This suggests that both profiles may each balance cognitive flexibility and inhibition in different ways, which could reflect varying demands on working memory and strategic regulation. Furthermore, each profile may be better suited in different situations, with Profile B potentially performing better on tasks requiring higher fluency, and Profile A potentially performing better on tasks requiring greater accuracy and performance monitoring. Limitations and Future Directions While this study provides important insights into EF profiles, several limitations must be acknowledged. First, the AppFlo version used in this study was a beta version and may have differed from other versions of the application and standard paper-based design tasks. This may have provided slightly different results in some metric counts, potentially impacting the results of this study. However, while not validated, multiple informal verifications were conducted, leading us to believe that no major changes would have occurred in the results. Nonetheless, future research should try to employ the now prevalidated version of AppFLO, to improve the generalizability of the results. Second, although the sample size was sufficient for clustering, it remains relatively modest (n=113) and predominantly constituted of French-speaking White children from Quebec. Future studies should aim to replicate these results by including a larger and more diverse sample. Third, while performance variability across the task may have reflected cognitive fatigue or reduced self-regulation over time, such factors were not directly measured. This makes it harder to differentiate preference-based profiles and deficit-based profiles. Future studies should explore how fatigue susceptibility contributes to EF variability, particularly in clinical populations such as ADHD, where speed-accuracy tradeoff mechanisms may be more pronounced ( Rommelse et al., 2016 ). Future studies should also explore whether these EF profiles could have some transdiagnostic components, as well as the impact academic performance, problem-solving tasks, or daily executive function challenges to enhance clinical and educational applicability. Finally, while the use of K-means was parsimonious and efficient in this data-driven study, further research aiming at exploring or validating profiles may benefit from the adoption other machine learning techniques, including neural networks, to allow for a deeper understanding of gradual and intermediate changes in measures between these profiles. Conclusion This study successfully identified two distinct EF profiles in neurotypical children using an unsupervised clustering approach, revealing opposing speed-accuracy tradeoffs. Profile A demonstrated a more a prioritization of being careful over fluency, while Profile B exhibited higher fluency but reduced accuracy over time. Crucially, process-based measures provided new insights into how children regulate executive function dynamically throughout a design fluency task, highlighting the importance of assessing EF as an evolving process rather than a static ability. Beyond these differences, Profile A’s cognitive flexibility increased with age, whereas Profile B’s executive function strategies remained independent of age-related maturation. While this suggests that EF regulation may follow different developmental trajectories, further research is needed to determine whether these profiles remain stable over time or shift with context, experience or cognitive training. Future studies should seek to replicate these findings in larger and more diverse samples to assess the stability and generalizability of these EF profiles. If confirmed, these profiles could redefine perspectives on cognitive variability in neurotypical populations and provide novel insights into the mechanisms underlying executive control. Identifying these distinct EF strategies could inform targeted cognitive interventions—for example, supporting Profile B in developing a stronger monitoring control, while optimizing Profile A’s flexibility and fluency for more efficient executive function regulation. Additionally, understanding how these EF profiles influence learning and problem-solving could guide personalized educational strategies and early identification of executive function difficulties in children. Acknowledgment This project was partially funded by a grant from the Fonds de recherche du Québec – Société et culture (FRQSC), through the Audace program. Additionally, we gratefully acknowledge the Fondation TALAN ( www.fondationtalan.org ), which also partially funded this research. Finally, we would like to thank Sina Esmaeili and Fateme Aliyari for their invaluable support and careful proofreading of this paper. This study was not preregistered. Funder Information Declared Fonds de Recherche du Québec - Société et culture , Audace Fondation TALAN, Montreal, CA Footnotes myriam.sahraoui{at}umontreal.ca , karim.jerbi{at}umontreal.ca , vanessa.hadid{at}umontreal.ca , bruno.gauthier{at}umontreal.ca References ↵ Albert , D. , Opwis , K. , & Regard , M . ( 2009 ). Effect of drawing hand and age on figural fluency: A graphomotor study with the five-point test in children . Child Neuropsychology , 16 ( 1 ), 32 – 41 . doi: 10.1080/09297040903049061 OpenUrl CrossRef ↵ Astle , D. E. , Bathelt , J. , Team , T. C. , & Holmes , J . ( 2019 ). Remapping the cognitive and neural profiles of children who struggle at school . Developmental Science , 22 ( 1 ), e12747 . doi: 10.1111/desc.12747 OpenUrl CrossRef PubMed ↵ Bablekou , Z. , Chrysochoou , E. et Kazi , S. ( 2023 ). Executive functions, listening comprehension, and metacognitive processes in childhood: Developmental profiles . Psychology: The Journal of the Hellenic Psychological Society , 28 ( 2 ), 48 – 68 . doi: 10.12681/psy_hps.36222 OpenUrl CrossRef ↵ Baez , A. C. , Dajani , D. R. , Voorhies , W. , Parladé , M. V. , Alessandri , M. , Britton , J. C. , Llabre , M. M. et Uddin , L. Q. ( 2020 ). Parsing heterogeneity of executive function in typically and atypically developing children: A conceptual replication and exploration of social function . Springer . doi: 10.1007/s10803-019-04290-9 OpenUrl CrossRef Best , J. R. , Miller , P. H. , & Jones , L. L . ( 2009 ). Executive Functions after Age 5: Changes and Correlates . Dev Rev , 29 ( 3 ), 180 – 200 . doi: 10.1016/j.dr.2009.05.002 OpenUrl CrossRef PubMed Web of Science ↵ Bolduc , F. ( 2023 ). Prévalidation d’un test de fluidité graphique sous forme d’application mobile pour les enfants de 6 à 12 ans ( Unpublished doctoral dissertation ). Université de Montréal . ↵ Cabbage , K. , Brinkley , S. , Gray , S. , Alt , M. , Cowan , N. , Green , S. , Kuo , T. , & Hogan , T. P . ( 2017 ). Assessing Working Memory in Children: The Comprehensive Assessment Battery for Children - Working Memory (CABC-WM) . Journal of visualized experiments : JoVE , ( 124 ), 55121 . doi: 10.3791/55121 OpenUrl CrossRef ↵ Chen , J. , Jiang , H. , Justice , L. M. , Lin , T. J. , & Purtell , K. M . ( 2022 ). One size doesn’t fit all: Profiles of isolated children in early childhood . Journal of Applied Developmental Psychology 80 , 101397 . doi: 10.1016/j.appdev.2022.101397 OpenUrl CrossRef ↵ Dumont , É. , Castellanos-Ryan , N. , Parent , S. , Jacques , S. , Séguin , J. R. , & Zelazo , P. D . ( 2022 ). Transactional longitudinal relations between accuracy and reaction time on a measure of cognitive flexibility at 5, 6, and 7 years of age . Developmental science , 25 ( 5 ), e13254 . doi: 10.1111/desc.13254 OpenUrl CrossRef PubMed ↵ Fournier , A. , Gauthier , B. , Guay , M. C. et Parent , V. ( 2020 , 01 Mar). Design fluency in children with ADHD and comorbid disorders . Brain Sciences , 10 ( 3 ), article n° 172. doi: 10.3390/brainsci10030172 OpenUrl CrossRef ↵ Gabrig , I. A. , Oliveira , R. M. , & Charchat-Fichman , H . ( 2018 ). Subgroups in cognitive development: Distribution of brazilian students aged 7-14 years . Psychology and Neuroscience , 11 ( 2 ), 168 – 179 . doi: 10.1037/pne0000112 OpenUrl CrossRef ↵ Gauthier , B. , Parent , V. , & Lageix , P . ( 2016 ). Exploring the dynamics of design fluency in children with and without adhd using artificial neural networks . Child Neuropsy-chology , 22 ( 2 ), 238 – 246 . doi: 10.1080/09297049.2014 OpenUrl CrossRef ↵ Goebel , S. , Fischer , R. , Ferstl , R. , & Mehdorn , H. M . ( 2009 ). Normative data and psychometric properties for qualitative and quantitative scoring criteria of the five-point test . The Clinical Neuropsychologist , 23 ( 4 ), 675 – 690 . doi: 10.1080/13854040802389185 OpenUrl CrossRef PubMed ↵ Goodgold-Edwards , S. A. , Beshere , N. , Murphy , K. , Macneil , D. , & Daoust , B . ( 1997 ). Cognitive strategies during a reciprocal tapping task . Physical & Occupational Therapy in Pediatrics , 17 ( 3 ), 1 – 19 . doi: 10.1080/J006v17n03_01 OpenUrl CrossRef ↵ Heitz , R. P . ( 2014 ). The speed-accuracy tradeoff: history, physiology, methodology, and behavior . Frontiers in neuroscience , 8 , 150 . doi: 10.3389/fnins.2014.00150 OpenUrl CrossRef PubMed ↵ Hurks , P. P . ( 2013 ). Administering design fluency tests in school-aged children: Analyses of design productivity over time, clustering, and switching . The Clinical Neuropsychologist , 27 ( 7 ), 1131 – 1149 . doi: 10.1080/13854046.2013.821170 OpenUrl CrossRef PubMed ↵ Judd , N. , Aristodemou , M. , Klingberg , T. , & Kievit , R . ( 2024 ). Interindividual differences in cognitive variability are ubiquitous and distinct from mean performance in a battery of eleven tasks . J Cogn , 7 ( 1 ), 45 . doi: 10.5334/joc.371 OpenUrl CrossRef PubMed ↵ Kamphorst , E. , Cantell , M. , Van Der Veer , G. , Minnaert , A. , & Houwen , S. ( 2021 ). Emerging school readiness profiles: Motor skills matter for cognitive and non-cognitive first grade school outcomes . Frontiers in Psychology , 12 . doi: 10.3389/fpsyg.2021.759480 OpenUrl CrossRef ↵ Klenberg , L. , Korkman , M. , & Lahti-Nuuttila , P . ( 2001 ). Differential development of attention and executive functions in 3-to 12-year-old finnish children . Developmental neuropsychology , 20 ( 1 ), 407 – 428 . doi: 10.1207/s15326942dn2001_6 OpenUrl CrossRef PubMed Web of Science ↵ Kochhann , R. , Gonçalves , H. A. , Pureza , J. d. R. , Viapiana , V. F. , Fonseca , F. d. P. , Salles , J. F. , & Fonseca , R. P. ( 2018 ). Variability in neurocognitive performance: Age, gender, and school-related differences in children and from ages 6 to 12 . Applied Neuropsychology: Child , 7 ( 3 ), 277 – 285 . doi: 10.1080/21622965.2017.1312403 OpenUrl CrossRef PubMed ↵ Kohonen , T . ( 1982 ). Self-organized formation of topologically correct feature maps . Biological cybernetics , 43 ( 1 ), 59 – 69 . doi: 10.1007/BF00337288 OpenUrl CrossRef Web of Science ↵ Laniel , P. , & Gauthier , B . ( 2022 ). Validation clinique et normes préliminaires des sous-tests de lecture et de dictée du test d’évaluation du langage écrit québécois . Canadian Journal of Speech Language Pathology & Audiology , 46 ( 2 ). OpenUrl ↵ Lezak , M . ( 1995 ). Neuropsychological assessment . Oxford university press . ↵ Litkowski , E. C. , Finders , J. K. , Borriello , G. A. , Purpura , D. J. , & Schmitt , S. A . ( 2020 ). Patterns of heterogeneity in kindergarten children’s executive function: Profile associations with third grade achievement . Learning and Individual Differences , 80 . doi: 10.1016/j.lindif.2020.101846 OpenUrl CrossRef ↵ Mareva , S . et Holmes , J. ( 2024 ). Mapping neurodevelopmental diversity in executive function . Cortex: A Journal Devoted to the Study of the Nervous System and Behavior , 172 , 204 – 221 . doi: 10.1016/j.cortex.2023.11.021 OpenUrl CrossRef PubMed ↵ Martarelli CS , Feurer E , Dapp LC , Roebers CM . Profiles of executive functions and social skills in the transition to school: A person-centred approach . Inf Child Dev . 2018 ; 27 : e2114 . doi: 10.1002/icd.2114 OpenUrl CrossRef ↵ Miller , J. B. , & Barr , W. B . ( 2017 ). The Technology Crisis in Neuropsychology . Archives of clinical neuropsychology : the official journal of the National Academy of Neuropsychologists , 32 ( 5 ), 541 – 554 . doi: 10.1093/arclin/acx050 OpenUrl CrossRef PubMed ↵ Miyake , A. , & Friedman , N. P . ( 2012 ). The Nature and Organization of Individual Differences in Executive Functions: Four General Conclusions . Current directions in psychological science , 21 ( 1 ), 8 – 14 . doi: 10.1177/0963721411429458 OpenUrl CrossRef PubMed ↵ Muñoz-Parreño , J. A. , Belando-Pedreño , N. & Valero-Valenzuela , A. ( 2023 ). Emotional intelligence, spelling performance and intelligence quotient differences based on the executive function profile of schoolchildren . European Journal of Neuroscience , 58 ( 8 ), 3879 – 3891 . doi: 10.1111/ejn.16152 OpenUrl CrossRef PubMed ↵ Nesayan , A. , Amani , M. & Asadi Gandomani , R . ( 2019 ). Cognitive Profile of Children and its Relationship With Academic Performance . Basic Clin Neurosci , 10 ( 2 ), 165 – 174 . doi: 10.32598/bcn.9.10.230 OpenUrl CrossRef PubMed ↵ Pedregosa , F. , Varoquaux , G. , Gramfort , A. , Michel , V. , Thirion , B. , Grisel , O. , Duchesnay , E . ( 2011 ). Scikit-learn: Machine learning in Python . Journal of Machine Learning Research , 12 , 2825 – 2830 . OpenUrl ↵ Poreh , A . ( 2022 ). Development of normative data for the Five Point Test using the cognitive reliability and time to failure rate theory . Applied Neuropsychology: Adult , 29 ( 5 ), 1141 – 1151 . doi: 10.1080/23279095.2020.1855181 OpenUrl CrossRef PubMed ↵ Péron , J. A . ( 2024 ). Challenges and prospects in advancing clinical neuropsychology . Cortex , 179 , 261 – 270 . doi: 10.1016/j.cortex.2024.08.001 OpenUrl CrossRef PubMed ↵ Regard , M. , Strauss , E. et Knapp , P. ( 1982 ). Children’s Production on Verbal and Non-Verbal Fluency Tasks . Perceptual and Motor Skills , 55 ( 3 ), 839 – 844 . doi: 10.2466/pms.1982.55.3.839 OpenUrl CrossRef PubMed ↵ Rommelse , N. N. J. , van der Meer , J. M. J. , Hartman , C. A. , & Buitelaar , J. K. ( 2016 ). Cognitive Profiling Useful for Unraveling Cross-Disorder Mechanisms: Support for a Step-Function Endophenotype Model . Clinical Psychological Science , 4 ( 6 ), 957 – 970 . doi: 10.1177/2167702616638826 OpenUrl CrossRef ↵ Sandilos , L. E. , Whittaker , J. V. , Vitiello , V. E. et Kinzie , M. B. ( 2019 ). Preschoolers’ school readiness profiles and the teacher-child relationship: A latent transition approach . Journal of Applied Developmental Psychology , 62 , 185 – 198 . doi: 10.1016/j.appdev.2019.02.010 OpenUrl CrossRef ↵ Scope , A. , Empson , J. et McHale , S. ( 2010 ). Executive function in children with high and low attentional skills: Correspondences between behavioural and cognitive profiles . British Journal of Developmental Psychology , 28 ( 2 ), 293 – 305 . doi: 10.1348/026151009X410371 OpenUrl CrossRef PubMed ↵ Shah , M. , Shandilya , A. , Patel , K. , Mehta , M. , Sanghavi , J. , & Pandya , A . ( 2024 ). Neuropsychological detection and prediction using machine learning algorithms: a comprehensive review . Intelligent Medicine , 4 ( 3 ), 177 – 187 . doi: 10.1016/j.imed.2023.04.003 OpenUrl CrossRef ↵ Stamatis , C. A. , Batistuzzo , M. C. , Tanamatis , T. , Miguel , E. C. , Hoexter , M. Q. et Timpano , K. R. ( 2021 ). Using supervised machine learning on neuropsychological data to distinguish OCD patients with and without sensory phenomena from healthy controls . British Journal of Clinical Psychology , 60 ( 1 ), 77 – 98 . doi: 10.1111/bjc.12272 OpenUrl CrossRef PubMed ↵ Stievano , P. et Scalisi , T. G. ( 2016 ). Unique designs, errors and strategies in the Five-Point Test: The contribution of age, phonemic fluency and visuospatial abilities in Italian children aged 6-11 years . Child Neuropsychology , 22 ( 2 ), 197 – 219 . doi: 10.1080/09297049.2014.988607 OpenUrl CrossRef ↵ Suchy , Y. , Ziemnik , R. E. et Niermeyer , M. A. ( 2017 ). Chapter 22 - Assessment of Executive Functions in Clinical Settings . Dans E. Goldberg ( dir .), Executive Functions in Health and Disease (p. 551 – 569 ). Academic Press . doi: 10.1016/B978-0-12-803676-1.00022-2 OpenUrl CrossRef ↵ Tessier , M. , Desmarais , A. , Leclerc , J. B. , Lavoie , M. E. , O’Connor K , P. et Gauthier , B. ( 2022 , 01 Apr ). Capturing Subtle Neurocognitive Differences in Children with and without Tourette Syndrome through a Fine-Grained Analysis of Design Fluency Profiles . Journal of Clinical Medicine , 11 ( 7 ), article n° 1946. doi: 10.3390/jcm11071946 OpenUrl CrossRef ↵ Tucha , L. , Aschenbrenner , S. , Koerts , J. et Lange , K. W. ( 2012 ). The Five-Point Test: Reliability, validity and normative data for children and adults . PLoS ONE , 7 ( 9 ). doi: 10.1371/journal.pone.0046080 OpenUrl CrossRef PubMed ↵ Usai , M. C. , Viterbori , P. et Traverso , L. ( 2018 ). Preschool executive function profiles: Implications for math achievement in Grades 1 and 3 . Journal of Research in Childhood Education , 32 ( 4 ), 404 – 418 . doi: 10.1080/02568543.2018.1496957 OpenUrl CrossRef ↵ Wechsler , D . ( 2003 ). Wechsler intelligence scale for children (4th ed.) . Pearson . ↵ Wright , K . ( 2022 ). Will the real hopkins statistic please stand up? The R Journal , 14 , 282 – 292 . doi: 10.32614/RJ-2022-055 OpenUrl CrossRef ↵ Younger , J. W. , Schaerlaeken , S. , Anguera , J. A. et Gazzaley , A. ( 2024 , 2024/09/01/). The whole is greater than the sum of its parts: Using cognitive profiles to predict academic achievement . Trends in Neuroscience and Education , 36 , 100237 . doi: 10.1016/j.tine.2024.100237 OpenUrl CrossRef View the discussion thread. Back to top Previous Next Posted August 24, 2025. Download PDF Email Thank you for your interest in spreading the word about bioRxiv. NOTE: Your email address is requested solely to identify you as the sender of this article. Your Email * Your Name * Send To * Enter multiple addresses on separate lines or separate them with commas. You are going to email the following Uncovering executive function profiles within interindividual variability: A data driven clustering exploration of design fluency in school-aged children Message Subject (Your Name) has forwarded a page to you from bioRxiv Message Body (Your Name) thought you would like to see this page from the bioRxiv website. Your Personal Message CAPTCHA This question is for testing whether or not you are a human visitor and to prevent automated spam submissions. Share Uncovering executive function profiles within interindividual variability: A data driven clustering exploration of design fluency in school-aged children Myriam Sahraoui , Karim Jerbi , Vanessa Hadid , Bruno Gauthier bioRxiv 2025.08.19.670430; doi: https://doi.org/10.1101/2025.08.19.670430 Share This Article: Copy Citation Tools Uncovering executive function profiles within interindividual variability: A data driven clustering exploration of design fluency in school-aged children Myriam Sahraoui , Karim Jerbi , Vanessa Hadid , Bruno Gauthier bioRxiv 2025.08.19.670430; doi: https://doi.org/10.1101/2025.08.19.670430 Citation Manager Formats BibTeX Bookends EasyBib EndNote (tagged) EndNote 8 (xml) Medlars Mendeley Papers RefWorks Tagged Ref Manager RIS Zotero Tweet Widget Facebook Like Google Plus One Subject Area Neuroscience Subject Areas All Articles Animal Behavior and Cognition (7621) Biochemistry (17644) Bioengineering (13867) Bioinformatics (41865) Biophysics (21413) Cancer Biology (18548) Cell Biology (25442) Clinical Trials (138) Developmental Biology (13360) Ecology (19866) Epidemiology (2067) Evolutionary Biology (24289) Genetics (15587) Genomics (22470) Immunology (17705) Microbiology (40304) Molecular Biology (17142) Neuroscience (88454) Paleontology (666) Pathology (2826) Pharmacology and Toxicology (4815) Physiology (7634) Plant Biology (15110) Scientific Communication and Education (2042) Synthetic Biology (4285) Systems Biology (9812) Zoology (2268)