1. Introduction and problem

Between ages 3 and 6 —kindergarten, preschool, CENDI, nursery school— a package of three artefacts has been installed that claims to stand for executive function development. The first is a cognitive training app: a screen that repeats attention, memory or inhibition items and declares that the child "already trains executive functions". The second is an adaptive game that scores attention: a system that adjusts difficulty according to latencies, errors or "time on screen" and presents them as if the score were inhibitory control. The third is a model that classifies impulsivity: an algorithm that labels behaviour, predicts working memory deficit or generates "executive risk" profiles and delivers them as if the output were cognitive flexibility. All three are visible and cheap in coordination time. They allow the centre to display that it "already does executive functions with artificial intelligence". The leap —from completing a digital item or receiving an algorithmic label to claiming executive functions— is not authorised either by cognitive training evidence or by what classrooms observe when there is real inhibition, working memory sustained in a shared situation and rule change mediated by an adult.

This article's thesis is restrictive. A cognitive training app, an adaptive game that scores attention or a model that classifies impulsivity does not develop executive functions in early childhood education. At this stage executive functions are inhibitory control, working memory and cognitive flexibility built through shared play, conversation, mediated waiting, movement and adult scaffolding on real everyday conflicts; not digital drill or algorithmic attention scoring. Trujillo-Trujillo, Cadavid Ruiz and Sanabria-Mazo (2025), in a meta-analysis of thirty-five studies with children aged 3 to 6, estimate small effects of cognitive training: d = 0.13 on verbal working memory, d = 0.10 on inhibitory control and d = 0.02 on nonverbal working memory. That does not authorise translating "there is a training app" as "there are executive functions". It authorises asking what was measured: often performance on constrained laboratory or screen tasks, not inhibition of an impulse in a turn-taking conflict, not retention of an instruction while the group negotiates a rule, not strategy change when the adult modifies the game. Ackerman and Friedman-Krauss (2021), in their synthesis report on preschool executive function, emphasise that inhibition, working memory and cognitive flexibility are cognitive control constructs with their own trajectories and specific measures. An algorithm that classifies impulsivity does not walk with the child. It administers a label on data the professional practice did not interpret in the classroom.

The problem is aggravated by a category confusion that product sheets do not mention and that this article rigorously separates from socio-emotional self-regulation. Cuartas, Hanno, Lesaux and Jones (2022), with a latent variable approach in 1,032 preschool children, show that executive functions and self-regulation share variance but are not the same construct. Finders, Duncan, Korucu, Bryant, Purpura and Schmitt (2021) review the literature and document the same conceptual overlap that the market exploits: when an app promises "self-regulation" or "emotion management" and the centre declares it as executive functions, a category error has occurred. This work does not recycle a socio-emotional competencies article. The question is cognitive control: what counts as executive function when an early childhood centre "does AI and executive functions". Confusing the product —app, score, classifier— with the process is the error this work names. Eng, Tsegai-Moore and Fisher (2024) demonstrate feasibility of gamification with machine learning to assess executive functions in 3- to 5-year-olds; Kara and Çağıltay (2024) design digital games oriented to executive functions. Marked pedagogical inference: the "AI for early childhood executive functions" market inherits that distribution and turns it into a development promise. It converts algorithmic assessment and item drill into the whole of cognitive control.

The contributions are three: reconstruct the state of the art separating inhibitory control, working memory and cognitive flexibility from digital training; examine three families of empirical cases; and offer four tests to decide when a kindergarten can claim executive functions, and not only an app, an adaptive score or an impulsivity profile generated by a model.

2. State of the art: from executive functions as cognitive control to the displayed artefact

It is useful to separate four strata that the "AI for early childhood executive functions" market usually mixes. The first is the construct of executive functions aged 3 to 6 as inhibitory control, working memory and cognitive flexibility (Ackerman and Friedman-Krauss, 2021; Cuartas et al., 2022; Finders et al., 2021). The second is the relational and bodily practice that cultivates them —shared play, teacher scaffolding, quality of adult–child interaction, mediated waiting— (Halliwell et al., 2021; Etokabeka, 2025; Etokabeka, Van Heerden and Du Preez, 2022; Veraksa et al., 2024; Sankalaite et al., 2021). The third is evidence on digital cognitive training, adaptive games, gamification with machine learning and predictive models (Trujillo-Trujillo et al., 2025; Birtwistle et al., 2025; Eng et al., 2024; Gao et al., 2026; Kara and Çağıltay, 2024; Panesi and Ferlino, 2023; Fitzpatrick et al., 2025). The fourth is the framework of rights, systems and developmentally appropriate practice, treating the 3–6-year-old as a subject of process interactions, not a drill interface user (OECD, 2021, 2023; NAEYC, 2022; UNESCO, 2021; Miao and Holmes, 2023; European Commission, 2022; Chen, 2024; Su and Yang, 2022; Ljungcrantz, 2026).

In the construct stratum, Ackerman and Friedman-Krauss (2021) synthesise assessment, relations with academic achievement and intervention in preschool, distinguishing inhibition, working memory and flexibility as measurable but non-reducible components. Cuartas et al. (2022) model executive functions and self-regulation as separable though correlated factors. Finders et al. (2021) warn that the literature mixes terms and constructs. Status: finding that executive functions are cognitive control, not synonymous with socio-emotional self-regulation. Inference: an app promising "emotion regulation" or "social self-control" does not cover, by existing, inhibition, working memory and flexibility.

In the relational stratum, Sankalaite et al. (2021) systematically review the relation between teacher–child interaction and executive functions: interaction quality —scaffolding, sensitivity, structuring— associates with executive development. Veraksa et al. (2024), with 947 children followed over one year, show that high-quality adult–child interaction predicts better executive function development. Halliwell et al. (2021) implement the Brain Builder Program: ten tabletop games with an adult, with the explicit instruction "wait for it", and document executive function improvements versus control. Etokabeka (2025), with eight South African teachers, analyses scaffolding in structured play to enhance executive functions. Etokabeka, Van Heerden and Du Preez (2022) link that scaffolding with executive skills in South African preschool. Status: finding that executive functions are verified in shared encounters, not in a latency log. Inference: an adaptive game scoring attention without an adult does not reproduce that practice.

In the digital stratum, Trujillo-Trujillo et al. (2025) set a small meta-analytic ceiling for cognitive training in preschool. Birtwistle et al. (2025), with N = 57, report g = 0.23 in a randomised controlled trial of cognitive training. Eng et al. (2024) combine gamification and machine learning to assess executive functions with feasibility in ages 3–5. Gao et al. (2026), with N = 141, reach AUC = 0.854 predicting working memory deficit in children with ADHD symptoms —prediction, not intervention. Kara and Çağıltay (2024) conduct design research on digital games for executive functions. Panesi and Ferlino (2023) pilot digital and analogical training. Fitzpatrick et al. (2025), with N = 315 in a longitudinal design, estimate B = −9.30 in inhibitory control for high screen-time trajectories. Status: field finding on training ceiling, algorithmic assessment feasibility and screen risk; not finding that an impulsivity classifier develops cognitive flexibility. Inference: training, scoring and classifying is what the three artefacts sell as "there are already executive functions with AI".

In the systems stratum, OECD (2021) anchors ECEC quality in meaningful process interactions. OECD (2023) situates digitisation in meaningful uses, not an executive item feed. NAEYC (2022) requires developmentally appropriate practice. UNESCO (2021) requires human oversight. Miao and Holmes (2023) set age thresholds and pedagogical validation for generative AI. Chen (2024), Su and Yang (2022) and Ljungcrantz (2026) map AI in ECE without equating it to kindergarten executive functions. The European Commission (2022) and U.S. Department of Education (2023) agree on not replacing professional judgement. Inference: a four-year-old is not the operator of a training app nor the passive subject of an adaptive score. They are the protagonist of cognitive control that an adult guarantees in play, conversation and real conflict.

3. Review method

A critical narrative review was conducted, not a primary meta-analysis. The purpose was not to estimate a homogeneous effect size of digital training, but to articulate a pedagogical category argument with verified sources. Inclusion criteria: (a) 2021–2026, with explicitly marked transfer when the sample is not 3–6; (b) executive functions, inhibitory control, working memory, cognitive flexibility, cognitive training, digital games, algorithmic assessment, teacher–child interaction or AI in early childhood education; (c) relevance for kindergarten, preschool, CENDI or ages 3–6; (d) peer-reviewed journal, DOI or NAEYC, UNESCO, OECD, European Commission or department of education report; (e) verifiable DOI or publisher page. Excluded as central object were axes already used in this series —socio-emotional competency as substitute for executive functions, generative tutoring, gaps, privacy, UDL, STEM, numeracy, formative assessment, family–school, continuous training— though some appear as limits.

The search was executed on 27 August 2026 on DOI pages, Springer, Elsevier, Wiley, SAGE, Frontiers, MDPI, OECD iLibrary, UNESDOC, ERIC, NAEYC, JAIR and publisher sites. Each source was verified against at least one of those pages. Empirical finding, conceptual or normative framework, and marked pedagogical inference were distinguished. The executive functions / socio-emotional self-regulation distinction was prioritised following Cuartas et al. (2022) and Finders et al. (2021).

4. Case 1. Digital cognitive training, adaptive games and classification models do not develop executive functions

Trujillo-Trujillo, Cadavid Ruiz and Sanabria-Mazo (2025) publish in Child Neuropsychology the synthesis that best names the first artefact when presented as executive function development. They meta-analyse thirty-five studies with children aged 3 to 6 exposed to cognitive training. Effects are small: d = 0.13 on verbal working memory, d = 0.10 on inhibitory control, d = 0.02 on nonverbal working memory. Evidence status. Empirical synthesis finding: cognitive training in preschool associates with modest gains on constrained measures. It is not finding that installing a training app develops executive functions in classroom life, nor that inhibition, working memory and flexibility are covered. The datum the market does not cite is magnitude: small effects on constructs measured in laboratory or screen tasks. Marked pedagogical inference: this is the gesture a kindergarten copies when it "does executive functions with an app". The product is installed, item performance is measured and cognitive control is declared fulfilled. What exists is an artefact. Executive functions, in Ackerman and Friedman-Krauss (2021) and Halliwell et al. (2021), require inhibiting an impulse in a shared situation, retaining an instruction in play with an adult and changing rules when context demands. A drill app does not observe that on the playground or in the play corner.

Birtwistle, Chernikova, Wünsch and Niklas (2025) saturate the portrait with the trial that best illustrates the second gesture's ceiling. They randomly assign 57 preschool children to cognitive training or control and report g = 0.23 on post-intervention executive functions. Status: empirical RCT finding with modest N. Not finding that an adaptive game scoring attention in real time develops inhibitory control outside the trained task. Marked inference: the "adaptation" the second artefact sells as executive personalisation is, in this corpus, item adjustment with small gain. A system modifying difficulty by latencies inherits that logic: it optimises the score, not turn negotiation or mediated waiting that Halliwell et al. (2021) name with "wait for it".

Eng, Tsegai-Moore and Fisher (2024) and Gao, Chen and Wang et al. (2026) represent the evaluative-algorithmic family. Eng et al. document feasibility of gamification with machine learning to assess executive functions in ages 3–5. Gao et al., with 141 children, reach AUC = 0.854 predicting working memory deficit in ADHD symptoms. Status: feasibility and prediction finding. Not finding that classifying impulsivity or predicting working memory equals building cognitive flexibility. Inference: this is the third gesture —model classifying impulsivity— turned into a development promise. Kara and Çağıltay (2024) design digital games for executive functions through design research; Panesi and Ferlino (2023) compare digital and analogical training in a pilot. Fitzpatrick et al. (2025) warn otherwise: high screen-time trajectories associate with worse inhibitory control (B = −9.30) in 315 longitudinally followed children. Status: screen-associated risk finding, not proof that more trained screen resolves the deficit. Inference: a CENDI delivering an "AI-generated executive profile" has done classification. Executive functions are verified in whether the group waited, negotiated, remembered the instruction and changed rules with an adult. They are not verified in the model's AUC.

5. Case 2. What the kindergarten actually does when there are executive functions: shared play, scaffolding and mediated waiting

Halliwell, MacKinnon, Plomp, Gibb, Coelho, Van Rootselaar and Gonzalez (2021) publish in Frontiers in Psychology the study that best names early executive functions as shared practice, not drill interface. They implement the Brain Builder Program: ten tabletop games with an adult, with explicit emphasis on waiting before acting —"wait for it"—, turns and impulse regulation in playful situation. They document executive function improvements versus control in a randomised controlled trial. Evidence status. Empirical finding on adult-mediated play intervention. Not finding on AI nor that an adaptive game without adult mediation reproduces that practice. Marked pedagogical inference: this is the object an early childhood centre can call inhibitory control and working memory in real situation. Waiting is not a screen timer: it is an adult holding the pause while the child contains the impulse to grab, throw or respond. A game scoring attention does not place the child in turn negotiation with peers. A model classifying impulsivity does not mediate the toy conflict.

Etokabeka (2025) saturates the portrait from teacher practice. With eight South African teachers, they analyse how scaffolding in structured play enhances executive functions in preschool: naming the instruction, chunking the task, modelling waiting, immediate feedback on impulse regulation. Etokabeka, Van Heerden and Du Preez (2022) link that scaffolding with executive skills measured in South African context. Status: qualitative/quantitative process finding in small, situated sample. Not generalisable to a Latin American kindergarten without marked transfer. Inference: a kindergarten cannot treat "we already use scaffolding on the tablet" as equivalent to scaffolding in structured play with a present adult. Veraksa, Gavrilova, Yakushina, Karimova and Solopova (2024), with 947 children followed over one year, show that teacher–child interaction quality predicts executive function development. Sankalaite et al. (2021), in systematic review, confirm that sensitivity, structuring and scaffolding in interaction associate with executive development. Status: longitudinal and synthesis finding on interaction, not screen. Inference: executive functions jointly include inhibition in real conflict, instruction retention in shared play and flexibility when the adult changes the rule. An app scoring only latencies does not describe that bundle. It describes a cut-down.

Movement and conversation complete the case. OECD (2021) anchors quality in meaningful process interactions —exactly the terrain where an adult names waiting, repeats the instruction and helps change strategy. NAEYC (2022) requires developmentally appropriate practice: cognitive control aged 3–6 is not reducible to completing isolated items. Inference: when there are executive functions in early childhood, there is regulated bodily movement, conversation about what to do first and next, and everyday conflict —turn, material, transition— mediated by an adult who does not replace their judgement with an adaptive game's score.

6. Case 3. Executive functions are not socio-emotional self-regulation: a category distinction

Cuartas, Hanno, Lesaux and Jones (2022) publish in PLOS ONE the study that best prevents recycling executive functions as socio-emotional self-regulation. With 1,032 preschool children and latent variable models, they dissociate executive functions and self-regulation: they share variance but constitute distinguishable factors. Evidence status. Construct dissociation empirical finding. Not AI finding. Marked pedagogical inference: when an app or model promises "self-regulation", "emotion management" or "social self-control" and the centre declares it as executive functions, the error this article forbids has occurred. Executive functions are cognitive inhibition, working memory and flexibility —control of mental processes to reach a goal—; they are not, without more, emotional calm, empathy or global socio-emotional competency.

Finders, Duncan, Korucu, Bryant, Purpura and Schmitt (2021) systematically review executive functions and self-regulation in early childhood and document the terminological overlap feeding confusing product sheets. Ackerman and Friedman-Krauss (2021) fix the assessment map: inhibition, working memory and cognitive flexibility with their own instruments and trajectories. Status: review and assessment framework finding. Inference: an impulsivity classifier trained on behavioural signals may label observable behaviour; it does not alone demonstrate the child developed verbal working memory or cognitive flexibility in strict sense. Trujillo-Trujillo et al. (2025) show even explicit cognitive training produces small effects on those constructs. Reinforced inference: confusing SEL with EF is not a terminological detail. It is a category error authorising false pedagogical declarations.

This article does not convert Halliwell et al. (2021) into a socio-emotional wellbeing study: the retained object is waiting and turns as inhibitory control in play. It does not convert Veraksa et al. (2024) into leadership or school climate study: the retained object is teacher–child interaction quality. Chen (2024), Su and Yang (2022) and Ljungcrantz (2026) map AI in ECE; Miao and Holmes (2023) set age limits for generative AI. Inference: the only AI use not contradicting executive functions aged 3–6 remains on the side of the adult who walks, waits, names instructions and reorganises play —as support to record observations or vary materials—, subject to pedagogical validation. It does not enter as autonomous trainer nor classifier replacing professional gaze.

7. Inferential framework: four tests to claim executive functions, not an artefact

The following framework is this article's pedagogical inference, anchored in cases and verified instruments. It is not a new international standard. It distinguishes four tests. If a kindergarten, preschool, CENDI or nursery school does not pass them, it cannot declare that a cognitive training app, an adaptive game scoring attention or a model classifying impulsivity develops executive functions.

7.1. Test of inhibitory control in real conflict, not isolated item. Trujillo-Trujillo et al. (2025) document small training effects on inhibition measured in task. Halliwell et al. (2021) situate inhibition in play with adult and explicit waiting. Fitzpatrick et al. (2025) associate high screen with worse longitudinal inhibition. Inference: executive function evidence is verified in whether the child inhibited an impulse in turn, transition or material negotiation. If the "evidence" that the centre develops executive functions is the log of item hits or adaptive attention score, the centre has made inventory, not inhibitory control.

7.2. Test of working memory sustained in shared situation, not algorithmic profile. Ackerman and Friedman-Krauss (2021) define working memory as central component. Gao et al. (2026) predict deficit with AUC = 0.854 —assessment, not construction. Veraksa et al. (2024) link quality interaction with executive development. Inference: the 3–6-year-old is not the bearer of a working memory profile generated by a model. They are the subject retaining the game instruction while the adult modifies a rule and peers negotiate. If AI enters, it enters as teacher observation support, not classifier replacing the shared sequence.

7.3. Test of mediated cognitive flexibility, not drill repeating the same routine. Birtwistle et al. (2025) show modest gains in structured training. Etokabeka (2025) shows flexibility when the teacher restructures play and models strategy change. Cuartas et al. (2022) prevent confusing flexibility with global self-regulation. Inference: cognitive flexibility is changing rule, strategy or attention when shared context demands. An adaptive game only speeding or slowing items does not demonstrate that change in real situation.

7.4. Test of EF/SEL distinction and professional judgement, not product catalogue. Finders et al. (2021) and Cuartas et al. (2022) require separating executive functions from socio-emotional self-regulation. UNESCO (2021) requires human oversight. Miao and Holmes (2023) require pedagogical validation. The European Commission (2022) and U.S. Department of Education (2023) require not replacing the teacher. NAEYC (2022) and OECD (2021, 2023) anchor quality in process interactions. Inference: an early childhood centre cannot treat the child as cognitive training operator nor bearer of algorithmic impulsivity labels. Executive functions are not fulfilled by scoring attention better on screen. They are fulfilled by waiting, remembering, changing rules and negotiating everyday conflicts with adults holding scaffolding.

The framework admits digital when subordinated to validated teacher observation and not replacing shared play (Eng et al., 2024; Panesi and Ferlino, 2023, as assessment pilots or complement, not autonomous development). It rejects declaring executive functions through cognitive training, adaptive score or impulsivity classifier (Trujillo-Trujillo et al., 2025; Birtwistle et al., 2025; Gao et al., 2026; Fitzpatrick et al., 2025).

8. Discussion

Three tensions organise the discussion. The first is between displaying an artefact and observing cognitive control. It is finding that cognitive training produces small effects (Trujillo-Trujillo et al., 2025; Birtwistle et al., 2025); that gamified assessment with machine learning is feasible (Eng et al., 2024); and that algorithmic working memory prediction reaches high AUC (Gao et al., 2026). It is framework that quality is played in process interactions (OECD, 2021; NAEYC, 2022). It is not finding that an app, score or classifier produces the practice Halliwell et al. (2021), Etokabeka (2025) and Veraksa et al. (2024) observe. The three artefacts measure what engineering can deliver and declare what only shared play would authorise.

The second is between screen and body. Fitzpatrick et al. (2025) associate high screen trajectories with worse inhibitory control. Kara and Çağıltay (2024) design digital games for executive functions. Inference: insisting the kindergarten "already develops executive functions" because there is tablet training is inverted pedagogy when longitudinal evidence penalises passive screen exposure. The child becomes item operator; the adult, score supervisor.

The third is between executive functions and socio-emotional self-regulation. Cuartas et al. (2022) and Finders et al. (2021) show they are not the same. Inference: the market selling SEL as EF aggravates the category error. The only AI use coherent with ages 3–6 remains on the side of the adult walking with the group, subject to pedagogical validation. An autonomous trainer or impulsivity classifier because the system needs a product is not that.

9. Limits

This review is narrative. It does not apply PRISMA nor estimate combined primary effects. Trujillo-Trujillo et al. (2025) group heterogeneous designs. Birtwistle et al. (2025) are N = 57. Eng et al. (2024) report feasibility, not development efficacy. Gao et al. (2026) predict deficit in ADHD symptoms, not general kindergarten population. Halliwell et al. (2021) are tabletop games in controlled context. Etokabeka (2025) are eight South African teachers. Veraksa et al. (2024) are 947 children in Russian context. Fitzpatrick et al. (2025) measure global screen time, not a specific app. Cuartas et al. (2022) are not an AI study. Chen (2024), Su and Yang (2022) and Ljungcrantz (2026) map AI in ECE, not kindergarten executive functions. NAEYC, UNESCO and OECD are framework. Panesi and Ferlino (2023) are small pilot. No Latin American trials were located comparing scaffolding in real play versus autonomous cognitive training in CENDI. Section 7 inferences are pedagogical category hypotheses, not implementation evidence.

10. Conclusions

A cognitive training app, an adaptive game scoring attention or a model classifying impulsivity does not develop executive functions in an early childhood centre. Verified evidence does not authorise that declaration. Thirty-five studies estimate small cognitive training effects (d = 0.13 verbal working memory; d = 0.10 inhibitory control; Trujillo-Trujillo et al., 2025). An RCT with 57 children reports g = 0.23 (Birtwistle et al., 2025). Gamification with machine learning assesses with feasibility (Eng et al., 2024); prediction reaches AUC = 0.854 (Gao et al., 2026); high screen trajectories associate with worse inhibition (B = −9.30; Fitzpatrick et al., 2025). Conversely, when there are executive functions in early childhood, there is shared play, mediated waiting and scaffolding: ten games with adult and "wait for it" (Halliwell et al., 2021); teacher scaffolding in structured play (Etokabeka, 2025; Etokabeka, Van Heerden and Du Preez, 2022); longitudinal interaction quality (Veraksa et al., 2024; Sankalaite et al., 2021). Executive functions are distinguished from socio-emotional self-regulation (Cuartas et al., 2022; Finders et al., 2021; Ackerman and Friedman-Krauss, 2021). Current guidelines require process interactions, human oversight, pedagogical validation and not replacing professional judgement (OECD, 2021, 2023; NAEYC, 2022; UNESCO, 2021; Miao and Holmes, 2023; European Commission, 2022; U.S. Department of Education, 2023).

Where sources do not measure a kindergarten, this article does not claim it. Where they measure training, score or classification, it does not translate them into developed executive functions. Accompanying three- to six-year-olds in cognitive control is exercising inhibition in real conflicts, sustaining working memory in shared play, flexibilising rules with adult mediation, moving the body with intention and conversing about what to do first and next. The rest is digital drill and algorithmic attention scoring. They are not executive functions, and must not be presented as what they are not.

Laboratorio Editorial de NEXTECH.IA / Ingeniero Mitre.

References

  1. Ackerman, D. J., and Friedman-Krauss, A. H. (2021). Preschool executive function: Assessment, relations with academic achievement, and intervention. ETS Research Report Series. https://doi.org/10.1002/ets2.12148
  2. Birtwistle, E., Chernikova, O., Wünsch, N., and Niklas, F. (2025). Can cognitive training improve executive functions in preschool children? A randomized controlled trial. Sage Open, 15(1). https://doi.org/10.1177/21582440241311060
  3. Chen, J. J. (2024). A scoping study on AI affordances in early childhood education: Mapping the global landscape, identifying research gaps, and charting future research directions. Journal of Artificial Intelligence Research, 81, 701–740. https://doi.org/10.1613/jair.1.16882
  4. Cuartas, J., Hanno, E., Lesaux, N. K., and Jones, S. M. (2022). Disentangling executive function and self-regulation in early childhood: A latent variable approach. PLOS ONE, 17(11), e0277013. https://doi.org/10.1371/journal.pone.0277013
  5. Eng, T., Tsegai-Moore, A., and Fisher, A. V. (2024). Gamification and machine learning for executive function assessment in preschool children: A feasibility study. Brain Sciences, 14(5), 451. https://doi.org/10.3390/brainsci14050451
  6. Etokabeka, P. P. (2025). Teachers' scaffolding of structured play to enhance executive function skills in South African preschoolers. Early Childhood Education Journal, 53, 2973–2982. https://doi.org/10.1007/s10643-024-01827-1
  7. Etokabeka, P. P., Van Heerden, A., and Du Preez, C. (2022). Executive function skills in South African preschool children: The role of teacher scaffolding during structured play. South African Journal of Childhood Education, 12(1), a1141. https://doi.org/10.4102/sajce.v12i1.1141
  8. European Commission. (2022). Ethical guidelines on the use of artificial intelligence (AI) and data in teaching and learning for educators. Publications Office of the European Union. https://doi.org/10.2766/153756
  9. Finders, J. K., Duncan, R. J., Korucu, I., Bryant, L. M., Purpura, D. J., and Schmitt, S. A. (2021). Executive function and self-regulation in early childhood: A review of the literature. Frontiers in Psychology, 12, 721282. https://doi.org/10.3389/fpsyg.2021.721282
  10. Fitzpatrick, C., Florit, E., Lemieux, A., Garon-Carrier, G., and Mason, G. (2025). Screen time trajectories and inhibitory control in early childhood: A longitudinal study. Academic Pediatrics, 25(2), 102603. https://doi.org/10.1016/j.acap.2024.102603
  11. Gao, Y., Chen, X., Wang, L., et al. (2026). Machine learning prediction of working memory deficits in preschool children with ADHD symptoms. BMC Pediatrics, 26, 112. https://doi.org/10.1186/s12887-025-06465-z
  12. Halliwell, E., MacKinnon, D. P., Plomp, H., Gibb, R., Coelho, C., Van Rootselaar, N., and Gonzalez, A. (2021). The Brain Builder Program: A randomized controlled trial of a play-based intervention to improve executive function in preschoolers. Frontiers in Psychology, 12, 720225. https://doi.org/10.3389/fpsyg.2021.720225
  13. Kara, N., and Çağıltay, K. (2024). Design research on digital games for executive function development in early childhood. International Journal of Human-Computer Interaction, 40(23), 8259–8273. https://doi.org/10.1080/10447318.2023.2279403
  14. Ljungcrantz, L. (2026). The interaction of AI and early childhood education. A state-of-the-art review 2020–2024. Early Childhood Education Journal, 54, 3565–3581. https://doi.org/10.1007/s10643-025-02079-3
  15. Miao, F., and Holmes, W. (2023). Guidance for generative AI in education and research. UNESCO. https://doi.org/10.54675/EWZM9535
  16. NAEYC. (2022). Developmentally appropriate practice in early childhood programs serving children from birth through age 8 (4th ed.). NAEYC. https://www.naeyc.org/resources/pubs/books/dap-fourth-edition
  17. OECD. (2021). Starting Strong VI: Supporting meaningful interactions in early childhood education and care. OECD Publishing. https://doi.org/10.1787/f47a06ae-en
  18. OECD. (2023). Empowering young children in the digital age (Starting Strong). OECD Publishing. https://doi.org/10.1787/50967622-en
  19. Panesi, S., and Ferlino, L. (2023). Digital and analogical executive function training in preschool: A pilot study. International Journal of Information and Education Technology, 13(4), 604–613. https://doi.org/10.18178/ijiet.2023.13.4.1844
  20. Sankalaite, A., Huizinga, M., Dewandeleer, A. C., Xu, C., de Vries, M., Hens, K., and Baeyens, D. (2021). Teacher-child interaction and executive function: A systematic review. Frontiers in Psychology, 12, 718262. https://doi.org/10.3389/fpsyg.2021.718262
  21. Su, J., and Yang, W. (2022). Artificial intelligence in early childhood education: A scoping review. Computers and Education: Artificial Intelligence, 3, 100049. https://doi.org/10.1016/j.caeai.2022.100049
  22. Trujillo-Trujillo, F., Cadavid Ruiz, C., and Sanabria-Mazo, J. P. (2025). Effects of cognitive training on executive functions in preschool children: A meta-analysis. Child Neuropsychology. https://doi.org/10.1080/09297049.2025.2595075
  23. UNESCO. (2021). Recommendation on the ethics of artificial intelligence. UNESCO. https://unesdoc.unesco.org/ark:/48223/pf0000381137
  24. U.S. Department of Education, Office of Educational Technology. (2023). Artificial intelligence and the future of teaching and learning: Insights and recommendations. U.S. Department of Education. https://www.ed.gov/sites/ed/files/documents/ai-report/ai-report.pdf
  25. Veraksa, N., Gavrilova, M., Yakushina, I., Karimova, R., and Solopova, O. (2024). Teacher-child interaction quality and executive function development in preschool: A longitudinal study. Frontiers in Education, 9, 1421037. https://doi.org/10.3389/feduc.2024.1421037