Data to design

Simulating instructional strategies with agent-based modelling

Abstract

What if we could simulate a learning environment like a living, evolving ecosystem? Our research views education as a complex adaptive system and applies methods from computational science to study it. We build agent-based models that simulate how learning happens in classrooms, drawing on both quantitative and qualitative data. These models allow us to run simulations that are updated with real classroom data over time. This helps us explore questions such as: “Which teaching methods best support student learning and knowledge transfer?” Our modelling suggests that Productive Failure may be especially effective in promoting deeper learning and transfer. We conclude by discussing how these insights highlight the potential of computational modelling in education research.