Modelling optimal learning pathways: A Markov Decision Process approach to the pedagogy–heutagogy continuum

Abstract

This paper presents a novel application of an AI model based on a Markov Decision Process (MDP) and leverages Dynamic Programming and value iteration to model the student learning journey across pedagogical, andragogical, and heutagogical learning paradigms. In contrast to a traditional static educational model, the proposed approach can adapt to changing cohort engagement and progress to provide more personalised and effective learning. Thus, the model can provide deeper insights into how students navigate self-directed learning, stay motivated, and achieve job readiness by considering different learning states, choices, and their associated outcomes. This work offers a theoretical contribution by formalising the pedagogy-heutagogy continuum and a practical framework through integration of analytics systems to optimise the learning process. It establishes a conceptual shift where personalisation moves from a design choice to a mathematically optimised strategy, bridging educational theory with computational decision science. While the current model uses illustrative data, it establishes a scalable foundation for future empirical integration using learning analytics.