Abstract:
Generative AI is increasingly embedded in computer science education, creating new opportunities and concerns regarding how students learn to program. This research investigates AI's impact on computer science students' cognition by examining how cognitive constructs are represented across programming-education research and how their significance varies across software engineering activities in AI and non-AI literature. A systematic mapping study, informed by established software engineering mapping guidelines and PRISMA 2020, screened 14,628 records between 2023 and 2025 from six databases and retained 191 primary studies, including 80 AI-related and 111 non-AI studies. Mixed deductive-inductive thematic analysis identified twelve cognitive and pedagogical themes. Cognitive processes, computational thinking, and comprehension were most frequently examined, while metacognition, motivation, cognitive load, misconceptions, and mental models capture complementary regulatory and interpretive dimensions. Findings indicate that cognition is task-sensitive: debugging foregrounds program comprehension, misconception diagnosis, and evaluation, whereas collaborative and pair-programming activities foreground explanation, shared problem solving, and cognitive regulation. AI-related studies predominantly address reasoning, interaction, and regulation, while foundational constructs such as comprehension, misconceptions, and mental models remain more prominent in non-AI research. Rather than viewing this distribution solely as conceptual fragmentation, this research proposes that it reflects both task specialisation and the early development of AI-mediated programming education. The mapping provides evidence that programming cognition is task-sensitive and identifies a research gap between AI-assisted code production and the verification, diagnosis, and judgement activities that remain central to durable learning. This gap motivates longitudinal research on AI amplification, cognitive debt, and emerging mental labour.
PPIG 2026 - 37th Annual Workshop
Doctoral Consortium
How Is AI Reshaping Computer Science Students' Cognition of Programming?