How Disciplinary Norms Influence Mathematicians' Views of Programming in Undergraduate Mathematics

TL;DR

This study explores how mathematicians' perceptions of programming influence its role in undergraduate teaching, revealing four epistemic archetypes and the conditions for broader integration.

Contribution

It identifies four epistemic archetypes among mathematicians and analyzes how these norms shape programming's role in research and teaching.

Findings

Light-touch programming is common in research but less visible in teaching.

Legitimacy of programming is tied to substantive integration in dedicated courses.

More extensive computational work fits within specialised numerical courses.

Abstract

Programming is deeply embedded in contemporary mathematical practice, yet its epistemic status in university mathematics teaching remains contested. Little is known about how mathematicians themselves understand the legitimacy of programming in their professional work, and how these views shape their teaching. We address this gap through semi-structured interviews with 15 mathematicians at a Northern European university with over two decades of systematic integration of programming across STEM subjects. Drawing on Cultural-Historical Activity Theory and Communities of Practice, we examine how mathematicians articulate the role of programming across research and teaching. We identify four epistemic archetypes - classical pure, classical applied, computational applied, and computational pure - each expressing coherent norms governing legitimate use of programming. Across archetypes, light-touch programming (e.g., numerical exploration) was widely used in research but largely invisible in teaching, where legitimacy was tied to more "substantive" integration. We argue that this gap reflects epistemic continuity across practices combined with teaching's focus on established analytic outcomes, which reduces the epistemic visibility of computational work. Given how mathematicians articulate legitimate uses of programming, our findings suggest that integration is most widely accepted when substantive programming is taught in dedicated courses, while light epistemic use - such as numerical exploration - is used to support learning in traditional theoretical courses. More extensive computational work is generally viewed as fitting naturally within specialised numerical or computational mathematics courses.