Both for practical and theoretical reasons, it is important to understand the type of knowledge acquired by computer programmers, and how it is used to solve problems. A substantial amount of research has been carried out to characterise programmers knowledge and memory, a subset of which is reviewed in this paper. While several theoretical explanations have been advanced to explain the knowledge and memory of programming experts, no computational model has been developed so far to simulate the empirical data. We describe how a simplified version of CHREST, a computational model originally developed to account for perception, learning, and memory in chess, can simulate the main effects found in research into computer-program memory. While it captures the skill differences found even in cases where substantial randomisation had been done, it fails to capture within-skill effects of variations at the semantic level.
More about the programming project can be found from the CHREST home page, including an electronic (pdf) copy of this technical report.To receive an off print, please email Irene Jackson.