Computational Learning Theories: A Mixed-Methods Framework for AI Enhanced Educational Research

Educational research stands at a critical juncture where traditional linear models prove insufficient for understanding the complex, dynamic nature of learning processes. This article proposes a computational mixed-methods approach as a necessary evolution in educational research methodology, encompassing a three-level hierarchical framework that integrates individual, social, and cultural learning processes through network-based modeling. Drawing from complexity theory, thermodynamics, and artificial intelligence, the article proposes that to guide AI applications in education effectively, learning theories must become computational to capture the nonlinear, emergent properties of learning systems. The framework consists of 14 interconnected nodes across micro (individual), meso (social), and macro (cultural) levels, offering new opportunities for mixed-methods researchers to incorporate dynamic modeling, causal analysis, and AI partnerships into their methodological toolkit. The article provides practical guidance for implementing the new approaches alongside traditional qualitative and quantitative methods, arguing that such integration is essential for advancing educational research in an AI-enhanced world.