COMPTEACH · Computational mechanisms of teacher-pupil knowledge transfer

Computational mechanisms of teacher-pupil knowledge transfer

Reinforcement learning is how humans and other animals learn through trial and error: we act, observe outcomes, and adjust. This kind of learning shapes everyday life, from mastering motor skills and planning efficiently to preparing for exams or refining medical decisions based on past results. Computational models of reinforcement learning have made it possible to describe learning in measurable components, improving our understanding of how individuals adapt their behaviour. In real life, however, learning rarely happens alone. Much of what we learn is taught to us, yet we still lack a clear mechanistic understanding of what makes teaching effective and how strategies are transmitted from teacher to learner. At a time when societies are rethinking learning after the pandemic and computational modelling is progressing rapidly, the opportunity is unique to advance a scientific understanding of teaching. COMPTEACH addresses this gap by developing a computational account of human teaching, combining methods from computational cognitive science and social psychology with tools from reinforcement learning and language analysis. The project tests three core ideas: (i) successful teaching transfers behavioural strategies (not just information) from teacher to learner; (ii) this transfer depends on what teachers say and how they structure their explanations; and (iii) teachers’ own experience and strategic knowledge shape how effectively they transmit learning strategies. By modelling teachers and learners together and linking strategy transfer to features of written explanations, COMPTEACH lays the groundwork for more realistic models of knowledge transmission. While the project is primarily fundamental research, it supports a pathway to impact by enabling future applied studies in education and health. For example, a next step is already planned through a collaboration with the Bligny Hospital Center (Briis-sous-Forges, France) to study teaching and learning mechanisms in people living with anxiety and depression, conditions known to affect social learning. As mental health difficulties become increasingly common among adolescents and students, understanding how guidance and instruction shape learning strategies can help build evidence for improving pedagogical and supportive interactions in real-world settings.

Data: CORDIS, © European Union