Mathematics learning gains depend on pedagogical reasoning rather than

An editorial collection of twenty-one research contributions across school, university, and teacher education environments indicates that digital technology expands mathematical learning only when paired with sound pedagogical design, teacher agency, and context-sensitive implementation. Rather than treating technology access as an automatic driver of educational achievement, the compiled evidence underscores that the value of digital tools depends on how learners encounter, represent, and investigate mathematical concepts. Across dynamic geometry, simulations, digital video, blended courses, intelligent tutoring, generative artificial intelligence, and computational evaluation, the findings collectively demonstrate that meaningful integration requires rigorous attention to mathematical reasoning and the empirical evidence guiding instructional decisions.

Investigations into dynamic representations highlight measurable learning benefits alongside unresolved questions regarding the specific drivers of improvement. In Uganda, Kirabo and colleagues documented higher post-test performance among in-service primary teachers engaging in GeoGebra-supported problem-based learning compared to comparison participants, with classroom observations pointing to the combined influence of visualization, collaboration, and applied tasks. Similarly, Zenteno Ruiz and colleagues observed favorable achievement outcomes following a PhET-supported intervention on mathematical functions among Peruvian university students. While both investigations demonstrate the promise of linking interactive tools to structured mathematical tasks, researchers noted that the separate contributions of software features versus underlying pedagogy remain to be established with greater precision.

Visual approaches also provided avenues for addressing abstract logical and spatial relationships, though design limitations warrant measured interpretation. Orhani and colleagues documented improved student performance and positive attitudes following instruction on De Morgan's Laws using visual methods, yet their single-group research design could not isolate the intervention from alternative explanations for the observed change. In synchronous online geometry instruction, Yeung and colleagues connected GeoGebra applets, diverse examples, and interactive dialogue to gains in achievement and cognitive engagement among nineteen students, with learning gains maintained during a two-week delayed assessment. The authors emphasized that the exploratory nature and small sample size of the setting require caution before generalizing outcomes to routine provision.

Broadening the scope of representation, several studies explored multimodal and embodied avenues for mathematical understanding. Villa-Ochoa and Farsani examined prospective teachers' mathematical modelling videos, illustrating how audiovisual production enables students to organize mathematical meaning and connect abstract ideas to lived experience. Taking an ecological and embodied perspective, Videla and colleagues analyzed material construction, artificial-intelligence-mediated bodily classification, and immersive virtual reality. Their analysis located geometric understanding within coordinated actions spanning learners, instructional tasks, and physical or virtual environments, prompting educators to structure opportunities where students actively manipulate and construct mathematical relationships rather than observe them passively.

The collection further demonstrates that sophisticated digital resources cannot compensate for gaps in teacher agency or pedagogical preparation. In rural classrooms, Malale and Mbhiza observed technological knowledge deficits and severe infrastructure constraints during teachers' GeoGebra-supported Euclidean geometry lessons, where difficulties in generating accurate representations illustrated the necessity of linking technical training with mathematical precision and pedagogical judgement. These operational realities align with the perspectives of ninety Chilean mathematics teacher educators analyzed by Dockendorff and colleagues, who emphasized that developing digital competence demands substantial time and must remain grounded in mathematical reasoning and purposeful tool selection.

Alternative teacher preparation models highlighted the contrast between passive procedural observation and active instructional reasoning. Mensah demonstrated that a tightly controlled GeoGebra model lesson, while successful at demonstrating procedural steps, restricted prospective teachers from participating in the reasoning behind instructional choices, revealing an institutional barrier to cultivating autonomous professional judgement. In contrast, Huang and colleagues utilized visual learning analytics to assess GeoGebra artifacts and micro-teaching videos from a year-long professional development program, formulating recommendations for open educational practices. In Zimbabwe, Jamba and Chauraya reported that collaborative STEAM activities integrating digital and artificial intelligence tools supported prospective teachers' conceptual reconstruction, explanatory flexibility, and critical digital literacy through sustained lesson design and revision.

Addressing the material realities of educational access, Mulovhedzi identified overlapping infrastructure, pedagogical, and administrative leadership constraints within rural South African schools. While educators developed adaptive strategies, including offline preparation and device-sharing arrangements, the findings indicate that local resilience must be matched by dependable infrastructure investments and sustained instructional support. In urban secondary schools, Yanisko and colleagues evaluated an engineering-embedded algebra elective where students appreciated digital resources and practical application but reported difficulties and resistance regarding interdisciplinary demands, demonstrating that access to challenging mathematical curricula requires explicit instructional scaffolding, transparent expectations, and responsive teaching.

At the post-secondary level, curriculum coherence emerged as a decisive factor in sustaining technology-supported learning. Kämpf and colleagues evaluated a spiral-curricular blended learning course that coordinated mathematical methods with physics instruction, finding positive course evaluations and evidence of conceptual knowledge consolidation. The authors concluded that effective educational delivery relies on coordinated instructional sequencing, usable interactive materials, and ongoing learner support. They also noted that open educational resources offer viable pathways for sustainable practice only when educators are granted sufficient institutional time and capacity to adapt materials to their specific classroom settings.

Evaluations of artificial intelligence underscored the necessity of separating technological enthusiasm from verifiable educational efficacy. In the related STEM area of computer-robotics programming, Omeh and Ayanwale recorded favorable engagement, motivation, and achievement outcomes from AI-supported problem-based learning; however, because experimental conditions were situated across different universities, unmeasured institutional differences remain an alternative explanation. Examining generative AI, Mahlaba tracked prospective teachers evaluating ChatGPT-generated solutions to mathematical word problems, observing that while some participants identified and corrected reasoning errors, others accepted flawed logic or relied on superficial assessments, highlighting the critical role of teacher metacognition when reviewing machine-generated outputs.

Technical scrutiny of assessment automation exposed significant operational risks in evaluating student work. Ableitinger and Dorner tested the GPT-4 Turbo model on coding student solutions, identifying persistent handwriting-recognition failures, symbol misinterpretations, and inconsistent error classifications. These empirical findings establish that automated classification models cannot replace defensible human evaluation, requiring teachers and researchers to retain ultimate responsibility for interpreting student errors and allocating scores. Methodological measurement concerns were echoed by Susanibar Ramirez and colleagues, who adapted a survey measuring teachers' reported AI usage in Peru; despite showing internal consistency, confirmatory factor analyses revealed substantial model misfit, proving that reliability alone is insufficient to justify high-stakes score interpretation.

Large-scale empirical investigations directly challenged the assumption that greater technology exposure inherently elevates student achievement. Analyzing Japanese PISA data, Zhang and colleagues observed that mathematics self-efficacy was a substantially stronger predictor of student performance than information and communication technology variables, with associations varying across distinct usage types. In a longitudinal evaluation involving 940 students across 55 German classrooms, Schaaf and colleagues detected no statistically significant association between intelligent tutoring system usage frequency and student learning gains at either the individual or classroom level. Complementing these findings, Bian and Fang applied double and debiased machine learning to textbook evaluation, noting that while flexible estimation mitigates restrictive functional forms, causal inferences remain strictly contingent on valid identification assumptions.

Synthesizing the broader findings across the research collection, the authors conclude that digital tools advance learning only through the structured interaction of mathematical tasks, teacher judgement, learner participation, and institutional conditions. While dynamic representations, multimodal media, and artificial intelligence present novel opportunities to investigate mathematical ideas, their educational efficacy remains strictly contingent on sound instructional design and equitable access. Critically, the editorial synthesizes three clear boundaries identified across the gathered evidence: favorable student or teacher perceptions do not demonstrate actual learning gains, higher technology usage frequency does not reflect instructional quality, and internal test consistency alone does not guarantee valid score interpretations. Addressing these barriers therefore requires shifting focus toward the specific conditions under which digital technologies foster equitable mathematical reasoning, supported by rigorous evaluations of accessibility and sustained learning.

To translate these insights into practice, the editorial outlines structured priorities for classroom educators, teacher preparation programs, and school leadership. Teachers and teacher educators are urged to position mathematical reasoning as the central organizing principle of technology-enhanced instruction rather than treating software operation as an isolated skill. Effective professional development requires sustained opportunities where educators collaboratively co-design tasks, test dynamic representations, critically evaluate artificial intelligence outputs, and refine lesson plans using direct classroom evidence. For school administrators and educational policymakers, the collection stresses that investments in hardware and connectivity must be coupled with ongoing technical maintenance, dedicated professional learning time, and explicit curriculum alignment. Furthermore, policy planning must provide resources adapted to environments with limited connectivity, incorporating local teacher and student priorities while judging implementation by genuine learning opportunities rather than mere device uptake.

Methodological and technological governance forms the final dimension of the call to action, establishing strict standards for researchers and artificial intelligence developers. The authors urge educational researchers to adopt rigorous comparative and longitudinal research designs that account for institutional clustering, measure long-term retention and conceptual transfer, and transparently document local implementation constraints. Research instruments also require independent validation to justify their practical interpretations and high-stakes uses. Concurrently, developers and deploying institutions must document specific software versions and prompts, systematically evaluate computational errors within realistic educational settings, safeguard student data privacy, and ensure humans retain authority over consequential pedagogical choices. Ultimately, the authors advocate for sustained cross-sector partnerships between educators, researchers, policymakers, and technology developers to align digital innovation with verifiable mathematical comprehension and equitable educational participation.

Subject of Research: Technology integration and pedagogical conditions in mathematics education

Article Title: Editorial: Bridging barriers: technology integration in mathematics education

Article References: Ayanwale, M. A., Sunzuma, G., & Chirinda, B. (2026). Editorial: Bridging barriers: technology integration in mathematics education. Frontiers in Education, 11, Article 2038958. https://doi.org/10.3389/feduc.2026.2038958

Image Credits: AI Generated

Keywords: mathematics education, technology integration, dynamic geometry, GeoGebra, PhET simulations, intelligent tutoring systems, generative artificial intelligence, teacher professional development, mathematical reasoning, digital equity, computer-robotics programming, PISA mathematics achievement, handwriting recognition errors, pedagogical decision-making, double debiased machine learning, embodied geometric understanding, STEAM education, visual learning analytics