The study responds to a long-standing paradox in early childhood education. Although the visual arts are widely celebrated for fostering creativity and self-expression, they are frequently enacted in product-focused ways, with little attention paid to the thinking developed through the process. When art is positioned as decorative or as an end point, its capacity to support inquiry, dialogue, and sustained thought is obscured. Prior research has also shown that teachers, particularly those new to inquiry approaches, can struggle to know when and how to integrate the arts meaningfully, a difficulty often compounded by limited confidence rooted in their own experiences of art education. The new framework is designed to give educators a conceptual language for recognising artistic processes as genuine epistemic activity rather than filler activities or display material.

The evidence base was built in two phases. Phase One involved a nationwide qualitative questionnaire sent to all 2,139 early childhood settings registered on the national database of education settings, comprising seven open-ended questions about pedagogical influences, practices, benefits, and challenges of inquiry-based approaches; sixty-three settings responded. From those expressing interest, six settings were purposively selected for Phase Two to provide variation in service type, geography, size, age groupings, cultural demographics, and philosophy. Thirty-eight teachers and leaders participated, alongside children aged six months to six years, who took part through naturally occurring inquiries. Researchers spent roughly a week at each site, gathering group interviews, field notes, photographs, video recordings, observations, and pedagogical documentation, including children’s visual representations.

Methodologically, the study adopted a qualitative, interpretivist design guided by narrative inquiry and informed by sociocultural and bioecological theories of learning. Kaupapa Māori theory shaped the research praxis, emphasising whakawhanaungatanga, the ongoing enactment of relational accountability, and ako, a reciprocal ethic in which teaching and learning are shared, dialogic processes. Analysis was iterative and dialogic: researchers reviewed site data individually, then brought interpretations to regular online wānanga for collective discussion and refinement, re-storying episodes of practice and comparing interpretations across sources. Emerging accounts were checked with participating teachers, who were also invited to write their own narratives of the inquiries, strengthening their position as research partners. The framework itself was developed after the formal conclusion of the wider project through a further cross-site analysis, with successive versions refined by the team and checked against the data.

Theoretically, the framework weaves together Dewey’s experiential learning, Vygotsky’s sociocultural account of mediated thought, Māori onto-epistemologies, and the Reggio Emilia tradition associated with Loris Malaguzzi. Within this synthesis, the visual arts function as languages through which children externalise, test, and refine ideas. Vea Vecchi’s description of art-making as poetic thinking is central: sensory experience is integral to knowledge construction, and drawing, painting, sculpting, and constructing act as cultural tools that mediate cognition. Because visual representations are tangible artefacts, they render thinking visible and available for shared interpretation, allowing ideas to be revisited, contested, and extended over time. In co-constructivist classrooms, these artefacts become shared cognitive resources; as researcher Margaret Brooks has argued, drawing allows knowledge to exist in a shared state before being assimilated into new perspectives.

Materials themselves are treated as active participants in inquiry rather than neutral instruments. Building on work by Sylvia Kind, the study describes children’s encounters with clay, wood, charcoal, or acorns as vibrant social-ecological assemblages in which humans and non-humans are in constant relation. For infants and toddlers, whose investigations are primarily sensorimotor, this matters enormously: their inquiry may centre not on symbolic representation but on discovering what materials can do, their affordances, resistances, and transformations. Neuroscience supports this picture, showing that embodied encounters with materials activate attentional and feedback systems that sustain curiosity. The bicultural context adds further depth, with Māori concepts such as āta, deliberate care and attentiveness, and whanaungatanga, relationships of connection and reciprocal responsibility, framing material exploration as an ethical, place-responsive practice aligned with the national curriculum, Te Whāriki.

The resulting framework is deliberately not linear. It identifies recurring, interconnected processes through which children and teachers move as ideas, materials, and relationships evolve, and it is articulated in two age-responsive versions. For infants and toddlers, the processes are ignite, explore, revisit, and re-ignite, emphasising embodied engagement and meaning made through repetition. For young children, two further processes, plan and consolidate, are added, reflecting growing capacities for anticipation, representation, and collaboration. Crucially, the authors stress that this is not a simplified version of inquiry for the youngest learners but a challenge to assumptions about their capabilities, foregrounding their capacity to think with materials, peers, and adults in sustained cycles of sense-making.

Vignettes from the participating centres illustrate each process vividly. Ignition is captured in an infant-toddler inquiry where, following children’s fascination with painted rocks found in the neighbourhood, teachers introduced mosaic as a shared art form; toddlers gathered fragments, arranged them, applied grout, and the finished tile was placed near the entrance for families, making visible the Te Whāriki strand of contribution. Exploration appears in a toddler inquiry centred on acorns collected beneath an oak tree, where children compared sizes, sorted baby and mummy acorns, and later drew them, with marks, gestures, sounds, and narratives intertwining as movement became twirly whirly or rolling away to see the world. Revisiting is exemplified by children who, after discovering a mystery garden, theorised about a caretaker they named Nature Man, returning repeatedly to drawings and conversations until a shared narrative, papier-mâché figures, and eventually a book and theme song emerged, each return transforming earlier thinking rather than merely repeating it.

Planning and consolidation are shown to be equally material. In one kindergarten, children drew their ideas of what a taniwha, the guardian being of a local pūrākau, might look like; when construction moved to the carpentry area, the square and rectangular wood available forced them to adapt curved designs, demonstrating how planning unfolds through interaction between imagination and material conditions. Consolidation was embodied in a textile artwork inspired by tīvaevae, Cook Islands appliqué quilts, made at the conclusion of an inquiry into community and belonging. Early responses to the question of what community is included not sure and watering the plants; after months of shared making, children offered answers such as everyone together and looking after people, changes the authors read as evidence of collectively consolidated meaning, publicly displayed to families and the wider community.

The discussion repositions the visual arts as a generative space in which thinking is slowed down, made visible, and opened to reinterpretation. Materials produced moments of cognitive tension that one scholar terms cognitive knots, launching points for further investigation, while the pedagogical conditions that sustained inquiry included protected time, collegial dialogue, attentive listening, documentation that kept ideas in circulation, and deliberate re-offering of materials. The authors acknowledge the study’s grounding in a small number of New Zealand settings and offer the framework as a flexible conceptual resource requiring critical adaptation to local cultural and curriculum contexts rather than a universal model. Even so, the message is pointed: positioning the arts at the centre of inquiry means rethinking how knowledge itself is constructed, shared, and sustained, treating young children as capable participants whose ideas can be made visible, revisited, and taken seriously within a learning community.

Subject of Research: How visual arts and material exploration support inquiry-based learning and children's thinking in early childhood education.

Article Title: Thinking Through Making: Supporting Children’s Inquiry Through Visual Arts in Early Childhood Education

Article References: Thinking Through Making: Supporting Children’s Inquiry Through Visual Arts in Early Childhood Education. (n.d.). https://doi.org/10.1007/s10643-026-02341-2

Image Credits: AI Generated

DOI: 10.1007/s10643-026-02341-2

Keywords: early childhood education, visual arts, inquiry-based learning, Thinking Through Visual Inquiry, Te Whāriki, Kaupapa Māori, working theories, multimodal representation, pedagogical documentation, Reggio Emilia, materials, teacher pedagogy

The design philosophy behind MatBrain rests on a simple observation about how scientific work actually happens. When a materials scientist attacks a problem, they alternate between two very different mental modes. One mode is analytical and knowledge-heavy: interpreting a diffraction pattern, judging whether a proposed crystal structure is physically plausible, or reasoning about how a catalyst might bind nitrogen. The other mode is executive and procedural: deciding which computational tool to call next, in what order, and with what parameters. The researchers argue that forcing a single model to master both modes spreads its capacity thin, especially when that model must remain small. MatBrain therefore splits the workload across two specialized modules.

The first module, named Mat-R1, is a 30-billion-parameter model that serves as the analytical brain. It has been trained to perform expert-level domain reasoning about crystallography, materials properties and synthesis chemistry, drawing on a curated instruction-tuning corpus called Mat-252K-SFT. The second module, Mat-T1, is a 14-billion-parameter model acting as the executive hand of the system. Its job is orchestration: planning sequences of tool-based actions, invoking external computational resources, and managing the flow of information between steps. Together the two models form a collaborative agent in which Mat-R1 decides what should be concluded and Mat-T1 decides what should be done.

One of the more striking technical contributions of the work is a diagnostic method the team used to verify that the two modules really have become functionally specialized. The researchers performed an entropy analysis of each model’s output distributions and found distinct statistical signatures. The executive model, when planning tool calls, produces output distributions with a character that differs measurably from the analytical model when it reasons about materials science problems. Entropy, in this context, acts as a kind of fingerprint of the cognitive mode a model is operating in. The finding provides quantitative evidence that the dual-model architecture carves the research workflow into genuinely different computational roles rather than duplicating the same capabilities twice.

Training such an agent required a carefully staged pipeline. The team began with supervised fine-tuning on the Mat-252K-SFT dataset, which instills the domain knowledge and instruction-following behavior the models need. They then moved to reinforcement learning using a separate dataset of 20,000 examples, Mat-20K-RL, which sharpens the models’ ability to plan multi-step tool use and to reward correct reasoning chains rather than superficially plausible text. Throughout, the researchers took precautions against data leakage and maintained controlled benchmark splits with audit results, so that reported performance reflects genuine generalization rather than memorization of test items.

The payoff is efficiency. Current general-purpose large language models typically require hundreds of billions of parameters, yet published evaluations show they still struggle with the domain-specific reasoning and tool coordination that materials science demands. MatBrain, with 44 billion parameters distributed across its two modules, is competitive with frontier large language models on crystal materials tasks while remaining small enough to run on local infrastructure. For laboratories that cannot ship sensitive data to cloud providers or pay for massive inference clusters, that distinction matters enormously. It also reduces the energy footprint of each research cycle, an increasingly important consideration as AI-driven science scales up.

Versatility is another headline claim. MatBrain handles the full breadth of computational materials research: generating candidate crystal structures from compositional or functional specifications, predicting the properties of proposed materials, and planning realistic synthesis routes. In benchmark comparisons presented in the study, the system performed strongly across these tasks, suggesting that the dual-model split does not fragment competence but rather concentrates it where it is needed at each stage of the research lifecycle.

The most concrete demonstration comes from catalyst design. Applied to the search for bio-inspired nitrogen fixation catalysts, MatBrain generated 30,000 candidate crystal structures and, through automated screening, narrowed the field to 38 promising materials within 48 hours. The authors report that the system significantly reduced the human-active time required for materials design and computational screening, meaning that the scarce resource of expert attention was spent only where it added the most value. Nitrogen fixation is a problem of global consequence, since industrial ammonia production consumes vast amounts of energy, and catalysts inspired by biological systems could transform that picture. An AI agent capable of navigating the candidate space autonomously moves that goal closer.

Transparency was clearly a priority for the team, led by researchers at the Shenzhen Institutes of Advanced Technology of the Chinese Academy of Sciences. The Mat-252K-SFT and Mat-20K-RL datasets have been released publicly through HuggingFace, and the full source code of the MatBrain system is archived on Zenodo, allowing other groups to reproduce, scrutinize and extend the work. The study also includes extensive supplementary material with additional entropy analyses and their interpretation, alongside source data for the published figures, giving readers the tools to check the claims independently.

The broader significance of MatBrain may lie less in any single material it discovered than in the architectural lesson it teaches. As agentic AI systems spread through chemistry and materials science, most approaches have scaled up monolithic models and hoped that raw size would subsume specialized competence. MatBrain suggests an alternative path: divide the cognitive labor deliberately, train each component on the distribution of tasks it will actually face, and use statistical diagnostics such as entropy profiles to confirm that specialization has taken hold. If that recipe generalizes to other scientific domains, the future of autonomous research may belong not to a few colossal models behind corporate firewalls, but to federations of modest, specialized agents running in laboratories around the world, each one an expert collaborator rather than a generalist imitation.

The study arrives at a moment when autonomous laboratories and large-scale computational screening have begun to reshape how new materials are found. Recent efforts such as the autonomous synthesis laboratory demonstrated by Szymanski and colleagues, and the deep-learning materials discovery campaign of Merchant and co-workers, have shown that machine-driven exploration can surface candidate compounds at a pace no human team could match. What these approaches have often lacked, however, is a flexible reasoning layer capable of deciding what to compute, what to synthesize and when a result is trustworthy. MatBrain positions itself precisely in that gap, using language-model-based agency to coordinate established computational resources rather than to replace them.

The tool-orchestration challenge that Mat-T1 addresses is well documented in the broader literature. Benchmarks such as ToolQA and ToolSandbox were created specifically because general-purpose language models frequently fail at stateful, multi-step tool use, even when their raw question-answering ability is strong. In chemistry, systems like ChemCrow demonstrated that augmenting a language model with expert tools can markedly improve practical problem-solving, but those demonstrations typically relied on large commercial models accessed through application programming interfaces. The MatBrain results suggest that a 14-billion-parameter executive model, trained with reinforcement learning on domain-specific tool trajectories, can shoulder comparable orchestration duties without external dependencies.

On the analytical side, Mat-R1 benefits from a decade of investment in open materials databases and machine-learned potentials. Resources such as the Materials Project and the Open Quantum Materials Database, together with universal interatomic potentials like CHGNet, provide the computational substrate against which any proposed crystal structure can be evaluated. An agent like MatBrain does not need to internalize the physics of interatomic bonding in its weights; it needs to know how to interrogate the tools that do. This division of labor between learned reasoning and established simulation infrastructure is arguably what allows a 30-billion-parameter model to reach expert-level conclusions on domain tasks.

The evaluation context also deserves note. The authors benchmarked against demanding tests of scientific reasoning, including graduate-level question sets of the kind exemplified by GPQA, while maintaining leakage-controlled splits to guard against contamination of training data. Such precautions address a persistent criticism of language-model evaluations in science, where test items can leak into web-scale pretraining corpora and inflate apparent competence. The release of the benchmark splits and audit results alongside the training datasets makes this scrutiny possible for independent groups.

Finally, the emphasis on human-active time reframes what automation in science should optimize. Rather than measuring only wall-clock speed or raw throughput, the study highlights how much expert attention each design cycle consumes. In fields where trained crystallographers and computational chemists are scarce, an agent that compresses months of candidate generation and screening into two days of largely autonomous operation, while confining human involvement to validation and judgment, offers a practical template for laboratories seeking to expand discovery capacity without proportional growth in personnel.

Subject of Research: A collaborative agent with two lightweight synergistic models for autonomous crystal materials research

Article Title: A collaborative agent with two lightweight synergistic models for autonomous crystal materials research

Article References: Shi, T., Li, Y., Li, Z., Liu, Q., Zhou, J., Xu, W., Li, Y., Dai, D., He, R., Zhou, W., Wang, J., & Yu, X.-F. (2026). A collaborative agent with two lightweight synergistic models for autonomous crystal materials research. Nature Machine Intelligence. https://doi.org/10.1038/s42256-026-01298-6

Image Credits: AI Generated

DOI: 10.1038/s42256-026-01298-6

Keywords: collaborative, agent, lightweight, synergistic, models, autonomous, crystal, materials, research, scientific research