Exploring Theory of Cognition for General Theory of Software Engineering

In recent years, there has been significant interest in general theories of software engineering. In this article, we explore the utility of a theory of cognition, ACT-R, as a component of such a general theory. The ACT-R theory was instantiated to predict the effort of programming language comprehension for two cases: (i) a C program, and (ii) the corresponding Assembly program. An experiment was then conducted to generate empirical data on the two comprehension tasks. The theoretical predictions were compared to the empirical results. The theoretical model predicted that the effort of understanding the considered program in C is 37% of the effort of understanding a comparable program written in Assembly. The experiment generated 33% as the corresponding percentage number. The concordance between theoretical model and experimental data was surprisingly high, encouraging further investigations into the utility of cognitive theories in software engineering.