This article introduces a special issue comprising research on efforts to personalize learning in different academic subjects. We first consider the emergence of personalized learning (PL) and the myriad of definitions that describe its essential features. Thereafter, we introduce the articles in the special issue by examining their alignment to extant theories of learning, the instructional design features that personalize the learning experience based on a learner characteristic, and the relationships between PL design and outcomes achieved in an educational context. Based on observations of contemporary PL research, we identify key issues to be addressed by the field and recommendations for future researchers to undertake to advance a PL theory. Chief among issues with PL are the role of technology, the agency of the learner, and the absence of a consistent theoretical grounding to motivate PL design choices. Future directions that would advance PL include the adoption of a theory of change in PL design, a design-based research approach to refine PL initiatives, more intensive and iterative research in authentic classroom contexts, and a greater focus on student input into and ownership of the PL experience.

Students experience mathematics in their day-to-day lives as they pursue their individual interests in areas like sports or video games. The present study explores how connecting to students’ individual interests can be used to personalize learning using an Intelligent Tutoring System (ITS) for algebra. We examine the idea that the effects of personalization may be moderated by students’ depth of quantitative engagement with their out-of-school interests. We also examine whether math problems designed to draw upon students’ knowledge of their individual interests at a deep level (i.e., actual quantitative experiences) or surface level (i.e., superficial changes to problem topic) have differential effects. Results suggest that connecting math instruction to students’ out-of-school interests can be beneficial for learning in an ITS and reduces gaming the system. However, benefits may only be realized when students’ degree of quantitative engagement with their out-of-school interests matches the depth at which the personalized problems are written. Students whose quantitative engagement with their interests is minimal may benefit most when problems draw upon superficial aspects of their interest areas. Students who report significant quantitative engagement with their interests may benefit most when individual interests are deeply incorporated into the quantitative structure of math problems. We also find that problems with deeper personalization may spur positive affective states and ward off negative ones for all students. Findings suggest depth is a critical feature of personalized learning with implications for theory and AI instructional design.

We develop a theory of grounded and embodied mathematical cognition (GEMC) that draws on action-cognition transduction for advancing understanding of how the body can support mathematical reasoning. GEMC proposes that participants’ actions serve as inputs capable of driving the cognition-action system toward associated cognitive states. This occurs through a process of transduction that promotes valuable mathematical insights by eliciting dynamic depictive gestures that enact spatio-temporal properties of mathematical entities. Our focus here is on pre-college geometry proof production. GEMC suggests that action alone can foster insight but is insufficient for valid proof production if action is not coordinated with language systems for propositionalizing general properties of objects and space. GEMC guides the design of a video game-based learning environment intended to promote students’ mathematical insights and informal proofs by eliciting dynamic gestures through in-game directed actions. GEMC generates several hypotheses that contribute to theories of embodied cognition and to the design of science, technology, engineering, and mathematics (STEM) education interventions. Pilot study results with a prototype video game tentatively support theory-based predictions regarding the role of dynamic gestures for fostering insight and proof-with-insight, and for the role of action coupled with language to promote proof-with-insight. But the pilot yields mixed results for deriving in-game interventions intended to elicit dynamic gesture production. Although our central purpose is an explication of GEMC theory and the role of action-cognition transduction, the theory-based video game design reveals the potential of GEMC to improve STEM education, and highlights the complex challenges of connecting embodiment research to education practices and learning environment design.

Generative artificial intelligence has become prevalent in discussions of educational technology, particularly in the context of mathematics education. These AI models can engage in human‐like conversation and generate answers to complex questions in real‐time, with education reports accentuating their potential to make teachers' work more efficient and improve student learning. This paper provides a review of the current literature on generative AI in mathematics education, focusing on four areas: generative AI for mathematics problem‐solving, generative AI for mathematics tutoring and feedback, generative AI to adapt mathematical tasks, and generative AI to assist mathematics teachers in planning. The paper discusses ethical and logistical issues that arise with the application of generative AI in mathematics education, and closes with some observations, recommendations, and future directions.

Instruction can be made relevant to students when it draws upon and utilizes their interests, experiences, and “funds of knowledge” in productive ways to support classroom learning. This approach has been referred to as “context personalization.” In this paper, we discuss the cognitive basis of personalization interventions, including theories related to grounding via concrete experience, activation of funds of knowledge, and cognitive load. We discuss a series of interventions in which different approaches to personalization, each of which involves a different operationalization of how instruction can be made “relevant,” are compared in studies of mathematics learning, often involving intelligent tutoring systems. We show how results vary for these interventions based on the degree to which personalization is designed to match with and draw upon learners' unique quantitative experiences, based on prior cognitive characteristics of learners and based on the degree of ownership and control students experience in the learning environment.