Computer science education research has mostly focused on cognitive approaches to learning. Cognitive approaches to understanding learning do not account for all the phenomena observed in teaching and learning. A number of apparently successful educational approaches, such as peer assessment, apprentice-based learning and action learning, have aspects that are not satisfactorily explained by purely cognitive models. On the other hand, these approaches are stratagems rather than comprehensive theories, in that they do not apply in all cases. Education theories which explore learning beyond the cognitive domain, such as social construction, may provide additional insights into matters such as teaching style, curriculum design and assessment practices. This paper proposes a start towards introducing social construction into computer science education, and proposes new directions for research, curriculum development and educational practice. (Contains 4 figures.)