Dr. Martha Umana, Ed.D., MBA, Teacher Leadership Specialist, Founder of The Bridge

“AI gives children answers.

Early childhood gives children the architecture to question them.”

Premessa editoriale

In un tempo in cui il dibattito sull’intelligenza artificiale si concentra prevalentemente su competenze, regolazioni e tecnologie, questo contributo ci invita a spostare radicalmente lo sguardo.

La Dott.ssa Martha Umana ci riporta là dove tutto comincia: nei primi anni di vita.
È lì, nelle relazioni educative più precoci, che si costruisce l’architettura cognitiva che permetterà alle future generazioni di confrontarsi con l’intelligenza artificiale in modo critico e consapevole.

Il valore di questo contributo sta nella sua capacità di rovesciare una prospettiva diffusa:
la questione non è come insegnare l’AI agli studenti, ma quali condizioni rendono possibile, fin dall’infanzia, la capacità di pensare, giudicare e non delegare.

In questo senso, l’educazione precoce emerge come una questione non solo pedagogica, ma profondamente democratica, legata ai diritti cognitivi e all’equità.

Carmine Marinucci

Abstract (EN)

This article argues that the most consequential educational response to artificial intelligence lies not in secondary school curricula or competency frameworks, but in the first five years of life. Drawing on developmental neuroscience, executive function research, and recent randomized controlled trial evidence, it proposes that early childhood mathematics, experienced within human relationships, is a primary vehicle for building the cognitive architecture that will shape how future generations engage with artificial intelligence. The article is organized around the three verbs proposed by Culture Digitali for 2026: Coltivare, Educare, and Umanizzare. It further argues that the developmental conditions that support executive function are distributed unequally along economic lines, making early childhood a site of democratic concern that the AI era is rendering newly visible. The article concludes with implications for parent education, early childhood policy, and equity.

Abstract (IT)

Questo articolo sostiene che la risposta educativa più significativa all’intelligenza artificiale non risieda nei curricoli della scuola secondaria né nei quadri di competenze, bensì nei primi cinque anni di vita. Attingendo alle neuroscienze dello sviluppo, alla ricerca sulle funzioni esecutive e alle più recenti evidenze sperimentali, l’articolo propone che la matematica nella prima infanzia, vissuta all’interno delle relazioni umane, costituisca uno dei principali strumenti per costruire l’architettura cognitiva che orienterà il rapporto delle generazioni future con l’intelligenza artificiale. Il contributo è strutturato intorno ai tre verbi proposti da Culture Digitali per il 2026: Coltivare, Educare e Umanizzare. Si sostiene inoltre che le condizioni di sviluppo che favoriscono le funzioni esecutive siano distribuite in modo diseguale lungo linee economiche, rendendo la prima infanzia un luogo di rilevanza democratica che l’era dell’IA sta rendendo sempre più visibile. L’articolo si conclude con implicazioni per l’educazione dei genitori, le politiche per la prima infanzia e l’equità.

Keywords: early childhood development, executive function, artificial intelligence, mathematics, brain architecture, cognitive rights, democratic participation, equity, parenting, digital culture

There are moments when a technology enters the world and something shifts, not only in what is possible, but in what it means to be human. I felt the edge of one of those moments as a child, holding a perforated card. I did not understand what it was. But I understood, with the particular clarity that children sometimes have before adults explain things away, that this small object was asking something new of humanity. Not a technical question. A human one. What will we need to become in order to meet what is coming?

That question has remained with me across decades and across every educational role I have inhabited: as an elementary school teacher sitting on the floor with young children and their first encounters with number, as a university professor watching adults unlearn the fear of mathematics they had carried since childhood, as someone who has lived and worked in different countries and cultures, not only studied them, learning from the inside what is constant and what is particular in how human beings learn, how families teach, and how childhood itself is understood differently depending on where you are standing. From all of these vantage points, I have watched technology arrive and reshape the terrain. And from all of them, I have arrived at the same conviction: the question is never really about the technology. It is always about the human capacity that determines how we will meet it.

We are living through another of those moments now. Artificial intelligence has arrived with genuinely unprecedented force, and the field of education has responded with an urgency I respect and with work I value. Frameworks are being built, regulations are being debated, and curricula are being redesigned. I have contributed to some of this work. And I keep returning to the same question I first asked while holding that perforated card: what will we need to become?

The answer the research compels me to give is not the answer toward which the field is currently building. The most consequential educational response to artificial intelligence is not happening in secondary schools or universities. It is happening in the first five years of life, in the homes and early childhood settings where the human capacity to think, to question, and to judge is either built or foreclosed. This article is organized around the three generative verbs that Culture Digitali has proposed as the thematic frame for 2026: Coltivare, Educare, and Umanizzare, and uses them not as metaphors but as precise descriptions of what the science tells us must happen in the earliest years if we want human beings who can live well with artificial intelligence rather than simply beside it.

Coltivare: What the Brain Requires in Order to Grow

The word cultivate derives from the Latin colere: to tend, to care for, to prepare the ground. It is exactly the right word for what early childhood development requires from the adults around a young child. The ground is not the child. The ground is the conditions.

Developmental neuroscience has established, with a consistency that should carry more policy weight than it currently does, that the brain does not develop in isolation. It develops in relationship. The Harvard Center on the Developing Child describes this process as the construction of brain architecture: a sequential, bottom-up process beginning before birth and continuing into adulthood, in which simple neural circuits provide the scaffolding for increasingly complex ones. The quality of that scaffolding depends, above all, on the quality of early relational experience (Harvard Center on the Developing Child, 2024a).

The active ingredient in healthy brain development is not stimulation in the abstract. It is a reciprocal, contingent, responsive interaction between a young child and a caring adult: the adult noticing what the child notices, interpreting the child’s signals accurately, responding promptly and appropriately, and treating the child as an active participant in a dynamic exchange. The developing brain is biologically programmed to expect this kind of interaction. When that expectation is not met consistently, the consequences are not merely developmental delays. They are neurobiological. Children who grow up without reliable, responsive adult presence show measurably different brain maturation patterns, including elevated resting theta-wave activity associated with maturational lag and long-term difficulties in executive function (Wade et al., 2024).

The plasticity of the early brain is its greatest strength and its greatest vulnerability. By age six, the brain reaches approximately 90% of its adult volume, driven primarily by the rapid expansion of the prefrontal cortex, the region most responsible for higher-order cognitive functions (Frassoni & Marzocchi, 2020). This period of intense synaptogenesis, followed by synaptic pruning, makes the child uniquely sensitive to environmental influences. The circuits that are used grow stronger. The circuits that are not used are pruned. What is not practiced in this window does not wait for a later opportunity. It becomes harder to build.

A landmark scoping review synthesizing 138 peer-reviewed studies on responsive caregiving in children from birth to age eight confirmed robust associations between the quality of early adult-child interaction and cognitive development, language acquisition, emotional regulation, and the reduction of developmental delay across all socioeconomic contexts (Lobo et al., 2026). A prospective birth cohort study following infants through the first twelve months of life found that sustained exposure to responsive caregiving reduced the risk of suspected developmental delay significantly, with the protective effect most pronounced in lower-income families (Wang et al., 2022). This finding deserves to be stated plainly: adult presence is not a privilege of the advantaged. It is a developmental necessity that can be supported across conditions when adults understand its importance.

In the absence of consistent responsive care, the body’s stress-response system activates in ways that are not temporary. Toxic stress, the prolonged activation of the stress response in the absence of a buffering adult presence, disrupts the neural architecture that executive function depends on (Harvard Center on the Developing Child, 2024b). The implications for cognitive development, and therefore for the kind of thinking that allows a child to evaluate, question, and form independent judgments about the outputs of artificial intelligence, are profound and lasting.

To cultivate, in this context, means to understand that the ground for all future human capacity is prepared in the first years of life, through the quality of the relationships a child experiences. This is not a metaphor. It is biology.

Educare: Mathematics as a Vehicle for Building Human Capacity

The word educare, in its Latin root, does not mean to fill. It means to draw out. The distinction describes two fundamentally different theories of what education is for. One theory treats the child as a vessel to be filled with information. The other treats the child as a developing being whose capacities must be drawn out through experience, relationship, and challenge. The first theory is more comfortable with AI as instructor. The second is not, and the science helps explain why.

In the preschool years, the domain that most powerfully draws out the cognitive architecture that matters for the future is early mathematics. This claim is counterintuitive. We do not typically think of counting blocks or naming shapes as preparation for living thoughtfully with artificial intelligence. But the research is now clear enough to say with confidence: early mathematics, experienced within human relationships, is one of the most powerful vehicles available for developing executive function.

What Executive Function Is and Why It Is the Capacity That Matters

Executive function refers to a set of higher-order cognitive skills that allow a person to manage their own thinking and behavior in the service of goals. Adele Diamond defines the three core dimensions as working memory, the capacity to hold information in mind and manipulate it; inhibitory control, the capacity to resist impulses, distractions, and premature conclusions; and cognitive flexibility, the capacity to shift perspectives and adapt to changing demands (Diamond, 2013). Together, these skills make it possible to set goals, monitor progress, manage frustration, evaluate information critically, and change course when a plan is not working.

These are precisely the capacities that determine whether a person can use artificial intelligence thoughtfully. A child who grows into an adult with strong executive function can hold an AI-generated answer up against what they already know, notice when something does not fit, and decide whether to accept, modify, or reject it. A child whose executive function was not supported during the developmental window in which it was most plastic may find this evaluation genuinely difficult, not because the child lacks intelligence, but because the underlying architecture was not built when it was most developmentally available.

The Relationship Between Mathematics and Executive Function

A substantial body of longitudinal research establishes that executive function in the preschool years strongly predicts later mathematics achievement. This predictive direction, from EF to math, is robust and well-established. EF measured at age four predicts mathematics performance at age six, accounting for substantial variance even after controlling for general cognitive ability (Clark et al., 2010)). A meta-analysis of cross-sectional and longitudinal studies confirms an average correlation between EF and mathematics of r = .350 during the preschool and kindergarten years (Hutchison et al., 2025).

But the relationship does not run only in one direction, and the direction that matters most for this argument runs the other way. During the preschool period, before formal schooling begins, the relationship is bidirectional: mathematical experience also builds EF. Welsh and colleagues found that emergent numeracy skills at the beginning of pre-kindergarten predicted executive function at the end of pre-kindergarten. This bidirectional relationship was specific to mathematics; it was not observed in the parallel model for literacy (Welsh et al., 2010). Schmitt and colleagues confirmed bidirectional relations between EF and mathematics across the preschool year, finding that once children entered kindergarten, the relationship became predominantly unidirectional, with EF predicting mathematics but not the reverse (Schmitt et al., 2017). The window of reciprocal influence coincides precisely with the period of rapid synaptogenesis: ages two to five.

The mechanism through which mathematical experience builds executive function is not automatic. It operates through the quality of the interaction in which mathematical activity is embedded. A recent review argues that programs targeting EF in isolation, including computerized training, fail to produce measurable improvements in mathematical outcomes or in transferred EF skills (Scerif et al., 2023). Isolated training does not work. What works is integration.

Among the strongest causal evidence for this is a preregistered randomized controlled trial published in npj Science of Learning in 2025. One hundred and three four-year-olds were assigned to an integrated EF and mathematics intervention called the ONE program, co-developed with early years educators. The intervention group improved significantly more than the control group in overall numeracy. More importantly, EF and mathematics measures showed significantly greater interconnectedness after the intervention: the network of connections between executive skills and mathematical skills became denser and more integrated as a result of experiencing mathematics in an executive function-rich, relationally mediated form. Disadvantaged children showed the greatest gains (Scerif et al., 2025).

The field has drawn a significant practical conclusion from this body of evidence: rather than treating executive function solely as a prerequisite foundation for mathematics, educators should also recognize mathematics instruction as a primary vehicle for developing executive function skills during the preschool years (Hutchison et al., 2025). These are not competing claims. They describe a reciprocal system in which each capacity strengthens the other during the precise developmental window when both are most malleable. It means that when a child counts objects with a parent, decides which pile has more, holds a rule in mind while sorting blocks by color, or notices that a pattern has broken and tries to repair it, that child is not just learning mathematics. That child is building the cognitive architecture for independent thought.

The Role of the Adult in Mathematical Experience

The critical variable in all of this is not the mathematical content. It is the adult. Frassoni and Marzocchi’s review of parenting and executive function development identifies autonomy-supportive scaffolding as the most consistent predictor of EF development in the preschool years: the adult providing the minimum necessary help to allow the child to solve the problem, following the child’s pace, asking questions rather than supplying answers, and trusting the child to take responsibility for the activity’s direction (Frassoni & Marzocchi, 2020).

Svanhild Breive’s classroom ethnography in a Norwegian kindergarten offers a vivid illustration of what this looks like in a mathematical context. Her research reconceptualizes the zone of proximal development as a symmetrical space, created when the adult positions themselves as a learner who needs the child to explain and guide. In this symmetrical space, the child develops a stronger sense of ownership over the problem-solving process, engages in deeper reasoning, and builds precisely the kind of independent epistemic agency that distinguishes a critical thinker from a passive recipient of information (Breive, 2020).

An artificial intelligence system cannot inhabit this position. It can simulate receptivity, but it cannot genuinely not know the answer. It can generate questions, but it cannot genuinely be curious about what this particular child, in this particular moment, with this particular logic, is working out. The asymmetry is structural, and it forecloses the specific quality of interaction that the research shows is most generative for early executive function development.

To educate, in the sense that matters here, is to create the conditions in which a child’s capacity to think is drawn out through a relationship with an adult who is genuinely present, genuinely curious, and genuinely willing to follow the child’s lead. Mathematics is not the only context in which this can happen. But it is an exceptionally rich one, and the research shows it is available to any adult, in any home, at any income level, during the years when it matters most.

Umanizzare: What AI Cannot Provide and Why It Matters Now

Artificial intelligence has entered education with genuine affordances and genuine risks. The affordances are real, including personalization, accessibility, immediate feedback, and adaptation to a learner’s pace. The risks are also real, and the most recent research has given them precise scientific language.

A 2026 study published in Computers & Education: Artificial Intelligence introduced the concept of epistemic narrowing to describe what happens when learners interact with AI systems optimized for convergence, that is, systems that produce singular, self-contained answers that prioritize fluency and coherence over exploration and uncertainty. Young learners who lack the epistemic norms and metacognitive strategies needed to critically assess AI content are particularly vulnerable to what the authors describe as epistemic surrogates: systems that appear authoritative while offering conclusions without justification (Vendrell & Johnston, 2026). These learners may develop that capacity less robustly because they are less often required to exercise it. The architecture for questioning was not built, and the technology does not build it.

This risk is not primarily a problem of AI design, though better design would help. It is a problem of developmental timing. The research reviewed in this article establishes that the capacity to evaluate, to persist through uncertainty, to hold competing ideas in mind and decide between them, is built in the first five years of life through specific kinds of relational experience. It is not built through instruction at age twelve, regardless of how well-intentioned that instruction is.

The field of AI in education has concentrated its attention, understandably, on students who are already in school. The UNESCO AI Competency Framework for Teachers and the AI Competency Framework for Students, both released in 2024, represent a genuine and important contribution to this work (Miao & Cukurova, 2024; Miao et al., 2024). They provide guidance on the knowledge, skills, and values that educators and learners need in an AI-saturated environment. But they are designed for K-12 teachers and school-age students. The age range from birth to eight, which developmental research identifies as the most consequential window for building the cognitive architecture on which AI literacy depends, is structurally absent from these frameworks. This is not a criticism of those who developed them. It is a gap, and one that can be named and addressed.

A systematic review synthesizing 65 empirical studies on ChatGPT and critical thinking found research concentrated almost entirely in language-learning and higher-education contexts, with virtually no studies addressing early childhood (Guo et al., 2026). The absence is itself evidence: the field has not yet recognized that the formation of the capacities that AI literacy requires begins long before a student is old enough to use AI.

To humanize, in this context, means two things. It means recognizing that the capacities that make us distinctively human as cognitive agents, namely our ability to question, to tolerate ambiguity, to form independent judgments, and to resist premature closure, are not automatically present. They are built through specific developmental conditions in the early years. And it means orienting our educational systems, our parenting cultures, and our policy frameworks toward those conditions, rather than assuming these capacities will emerge on their own or can be retrofitted through later instruction.

A Question of Democracy: Cognitive Rights and the Architecture of Opportunity

There is one dimension of this argument that the research establishes with particular clarity and that the field has been comparatively slow to name. The developmental conditions that build executive function are not equally available to all children. They are distributed, with notable consistency, along socioeconomic lines.

A comprehensive systematic review published in the Journal of Child Psychology and Psychiatry confirmed that socioeconomic status is a powerful predictor of executive function in childhood, with effects that emerge before school entry and persist across development. The key mediating mechanism is cognitive stimulation: the availability of enriching adult-child interaction that encourages curiosity, language, and active thinking. Cognitive stimulation, not income alone, was consistently found to mediate the association between socioeconomic status and executive function, meaning that the quality of what adults do with children matters more than what they own (Rakesh et al., 2025).

The mechanisms are not mysterious. Economic precarity produces parental stress, which reduces the quality and consistency of responsive caregiving. Lower levels of parental education reduce awareness of how adult interaction shapes cognitive development. Time poverty limits the daily moments of unhurried attention that build mathematical thinking and relational trust.

Children in high-income, highly educated homes receive the conditions that build executive function through the ordinary texture of daily life: dinner conversation, shared book reading, counting games during grocery shopping, and the patient adult who asks “how did you figure that out?” rather than supplying the answer. These experiences are not deliberately designed. They emerge naturally from conditions of stability, time, and knowledge.

Children in low-income homes receive those conditions far less consistently. Not because their parents love them less or are less capable. Because the conditions that make responsive, cognitively rich caregiving possible are themselves unequally distributed. A meta-analysis examining 70 empirical studies and more than 58,000 students confirmed that executive function partially mediates the relationship between socioeconomic status and academic achievement, meaning that part of the educational gap between economically advantaged and disadvantaged children is explained by an EF gap that forms before school begins (Ding et al., 2024).

The AI era transforms the stakes of this inequality. The capacity to evaluate AI-generated content critically, to resist premature conclusions, to hold competing interpretations in mind, to form independent judgments about what an authoritative-seeming system is actually claiming, is not a generic skill. It is a specific expression of executive function. A child who does not develop strong EF before age five will face artificial intelligence as an adult with a cognitive architecture that was not built for critical engagement with systems designed to produce confident, convergent answers. They will be more likely to accept, less likely to question, and more vulnerable to the epistemic narrowing the research documents. The AI era will not create this vulnerability. It will reveal and amplify a vulnerability that was already forming, in silence, before the age of five, in the homes where it was most likely to form.

This is a question of cognitive rights. The right to develop the capacity for autonomous thought, to form one’s own judgments, to participate fully in democratic life, to resist manipulation by any authority that offers confident answers in place of genuine reasoning, is being shaped, or foreclosed, before children know they have it. And it is being shaped unequally. A democracy that does not attend to the cognitive conditions of its youngest and most economically vulnerable citizens is not merely failing those children. It is also reproducing conditions of long-term civic fragility.

In this sense, early childhood is not only a developmental domain but a cultural one, because it is where habits of attention, interpretation, dialogue, and judgment first take social form. The formation of democratic citizens does not begin in school. It begins in the early years, in the quality of the relationships surrounding a child, in the moments when an adult follows a child’s thinking rather than redirecting it, in the mathematical conversations that teach a child that their reasoning matters and their questions are worth pursuing. These are the conditions that build the cognitive architecture of democratic participation. They are available, in principle, to every family. They are not, in practice, equally accessible. Closing that gap is not only a developmental imperative. It is a civic one.

Implications: Where the Work Must Go

The argument developed here has several practical implications that deserve explicit statement.

Parent education should be recognized as a form of AI readiness policy. If the quality of adult-child interaction in the first five years determines the cognitive architecture that will govern a person’s relationship with artificial intelligence for the rest of their life, then equipping parents and caregivers with a genuine understanding of responsive interaction: what it means to follow a child’s reasoning rather than correct it, to ask rather than tell, and to wait rather than solve, is among the highest-leverage investments an education system can make. This is not about prescribing parenting styles. It is about ensuring that adults understand what they are doing when they sit with a child and count objects together, and why it matters.

Pre-primary education should be repositioned within the AI-era policy conversation. The developmental window from birth to age five is not a pre-educational phase. It is the most educationally consequential phase there is, precisely because it is when the architecture for all future learning is built. Policies that treat early childhood as a period to be protected from technology while leaving its positive developmental potential unaddressed are incomplete. What is needed is not protection alone, but intentional cultivation: environments and relationships that make it possible for executive function to develop through integrated, relational, mathematically rich experience.

The equity dimension of this argument is urgent and cannot be treated as secondary. The research is unambiguous that the protective effects of responsive caregiving and rich early learning opportunities are most pronounced in lower-income families, and that the EF gap between economically advantaged and disadvantaged children forms before school entry. Addressing AI equity at age fifteen without addressing the conditions of early childhood that produce cognitive equity is treating a symptom without attending to its cause. The investment that matters most is not only in AI literacy curricula for adolescents. It is in the conditions that allow executive function to develop in the first five years of life, across all homes, not only in those where such conditions are already more available.

Finally, the evidence does not suggest that young children should be taught about AI or that mathematics should be introduced as formal academic instruction before children are ready. It suggests that the natural mathematical moments of early childhood, namely counting, comparing, patterning, estimating, and reasoning about quantities, should be recognized as opportunities for the kind of scaffolded, autonomy-supportive adult-child interaction that simultaneously builds mathematical understanding and executive function. The content matters less than the quality of the relational exchange in which it occurs.

Conclusion: The Conversation We Have Not Yet Had

I still think about that perforated card. Not because the technology was significant, but because of what I felt holding it as a child: the sense that the world was asking something new of human beings, and that I would be able to respond. That confidence did not come from anything written on the card. It came from a childhood in which adults had been present enough, attentive enough, and patient enough to help me build the architecture for wonder.

The students currently in school will face artificial intelligence with whatever cognitive resources they already have. Some will have strong foundations. Many will not. The field has responded with frameworks, regulations, and curricula. This work is necessary. But the conversation we have not yet had, and the one most likely to determine the long-term human outcome of the AI era, is the conversation about what needs to happen before a child ever enters a classroom.

The evidence is clear. By age six, the brain has reached approximately 90% of its adult volume, and much of the architecture that supports later independent thought is already in place. The executive function skills that make it possible to evaluate, question, and form independent judgments about AI-generated content are built in the preschool years, through relational experience, through mathematically rich interaction with responsive adults, through the specific conditions of being heard, followed, trusted, and challenged. These conditions are not curriculum. They are culture. They require not new technology but new consciousness: a collective recognition that the most important educational work of the AI era is happening in homes and early childhood settings, not in secondary schools and universities.

To cultivate, to educate, and to humanize in the age of artificial intelligence means, above all, to attend to the conditions of the earliest years with the seriousness they deserve. It means building curricula for parents, not just for students. It means training early childhood educators in the science of relational development alongside the science of mathematics. It means ensuring that every child, regardless of income, geography, or family circumstance, has access to the kind of adult presence that the developing brain is biologically programmed to expect.

Artificial intelligence will continue to become more capable, more fluent, and more persuasive. The question is not whether children will encounter it. They already do. The question is whether, when they do, they will have the architecture to question it. That architecture is not built by AI literacy curricula at age twelve. It is built by a responsive adult at age two, sitting beside a child who is counting, wondering, and working something out.

That is the conversation we need to have. And the time to have it is now.

References

Breive, S. (2020). Student–teacher dialectic in the co-creation of a zone of proximal development: An example from kindergarten mathematics. Early Child Development and Care, 191(14), 2249–2259. https://doi.org/10.1080/1350293X.2020.1755498

Clark, C. A. C., Pritchard, V. E., & Woodward, L. J. (2010). Preschool executive functioning abilities predict early mathematics achievement. Developmental Psychology, 46(5), 1176–1191. https://doi.org/10.1037/a0019672

Diamond, A. (2013). Executive functions. Annual Review of Psychology, 64, 135–168. https://doi.org/10.1146/annurev-psych-113011-143750

Ding, X., Li, S., Zhang, X., & Shi, J. (2024). The mediating role of executive function between socioeconomic status and academic achievement: A meta-analytic structural equation model. Learning and Individual Differences, 110, Article 102418. https://doi.org/10.1016/j.lindif.2024.102418

Frassoni, S., & Marzocchi, G. M. (2020). Il ruolo del parenting nello sviluppo delle funzioni esecutive in età prescolare [The role of parenting in the development of executive functions in preschool age]. Psicologia Clinica dello Sviluppo, 3, 375–400. https://doi.org/10.1449/98291

Guo, Y., Huang, L., Zhang, C., Li, Q., & Chen, M. (2026). ChatGPT in education: A systematic review of its impact on critical thinking skills and dispositions. Thinking Skills and Creativity, 60, Article 102106. https://doi.org/10.1016/j.tsc.2025.102106

Harvard Center on the Developing Child. (2024a). Brain architecture. https://developingchild.harvard.edu/key-concept/brain-architecture/

Harvard Center on the Developing Child. (2024b). The science of early childhood development. https://developingchild.harvard.edu/wp-content/uploads/2024/10/Science_Early_Childhood_Development.pdf

Hutchison, J. E., Tabrizian, B., Bajnauth, N., Avdellas, A., Phillips, D., & Lyons, I. M. (2025). A meta-analysis of the cross-sectional and longitudinal relations between executive functioning and math in early childhood. Journal of Numerical Cognition, 11, Article e14335. https://doi.org/10.5964/jnc.14335

Lobo, E., Mahapatra, S., Jeong, J., Babu, G. R., Srinivas, P. N., Mukherjee, D., & van Schayck, O. C. P. (2026). A scoping review of responsive caregiving in diverse populations and its association with child development. Early Human Development, 212, Article 106424. https://doi.org/10.1016/j.earlhumdev.2025.106424

Miao, F., & Cukurova, M. (2024). AI competency framework for teachers. UNESCO. https://doi.org/10.54675/ZJTE2084

Miao, F., Shiohira, K., & Lao, N. (2024). AI competency framework for students. UNESCO. https://doi.org/10.54675/JKJB9835

Rakesh, D., Lee, P. A., Gaikwad, A., & McLaughlin, K. A. (2025). Annual research review: Associations of socioeconomic status with cognitive function, language ability, and academic achievement in youth: A systematic review of mechanisms and protective factors. Journal of Child Psychology and Psychiatry, 66(4), 417–439. https://doi.org/10.1111/jcpp.14082

Scerif, G., Blakey, E., Gattas, S., Hawes, Z., Howard, S., Merkley, R., O’Connor, R., & Simms, V. (2023). Making the executive ‘function’ for the foundations of mathematics: the need for explicit theories of change for early interventions. Educational Psychology Review, 35, Article 110. https://doi.org/10.1007/s10648-023-09824-3

Scerif, G., Sučević, J., Andrews, H., Blakey, E., Gattas, S. U., Godfrey, A., Hawes, Z., Howard, S. J., Kent, L., Merkley, R., O’Connor, R., O’Reilly, F., & Simms, V. (2025). Enhancing children’s numeracy and executive functions via their explicit integration. npj Science of Learning, 10, Article 8. https://doi.org/10.1038/s41539-025-00302-9

Schmitt, S. A., Geldhof, G. J., Purpura, D. J., Duncan, R., & McClelland, M. M. (2017). Examining the relations between executive function, math, and literacy during the transition to kindergarten: A multi-analytic approach. Journal of Educational Psychology, 109(8), 1120–1140. https://doi.org/10.1037/edu0000193

Vendrell, M., & Johnston, S.-K. (2026). Scaffolding critical thinking with generative AI: Design principles for integrating large language models in higher education. Computers & Education: Artificial Intelligence, 10, Article 100572. https://doi.org/10.1016/j.caeai.2026.100572

Wade, M., Parker, V., Tang, A., Fox, N. A., Zeanah, C. H., & Nelson, C. A. (2024). Linking caregiving quality during infancy to brain activity in early childhood and later executive function. Developmental Science, 27(6), e13517. https://doi.org/10.1111/desc.13517

Wang, K., Qi, Y., Wei, Q., Shi, Y., Zhang, Y., & Shi, H. (2022). Responsive caregiving and opportunities for early learning associated with infant development: Results from a prospective birth cohort in China. Frontiers in Pediatrics, 10, Article 857107. https://doi.org/10.3389/fped.2022.857107

Welsh, J. A., Nix, R. L., Blair, C., Bierman, K. L., & Nelson, K. E. (2010). The development of cognitive skills and gains in academic school readiness for children from low-income families. Journal of Educational Psychology, 102(1), 43–53. https://doi.org/10.1037/a0016738

About the Author

Dr. Martha Umana holds a doctorate in education, a graduate specialization in teacher leadership, and an MBA. She is an independent educator, researcher, and organizational strategist, and the founder of The Bridge, a content and parent education initiative at the intersection of early childhood development, mathematics, and the AI era. She has worked as an elementary school teacher and university professor across multiple countries and cultures, and brings that breadth of experience to her research and public work. Her scholarship is particularly concerned with reframing early childhood as the most consequential educational frontier of the AI era, and with building the conditions, in homes, early childhood settings, and policy, that allow the architecture of human thinking to form before it is needed.

ORCID iD: 0009-0004-6871-1754