Relationship Between Model-Based Decision-Making and the Comprehension Performance of Source Code With Confusing Patterns

Background: Confusing source code requires deliberate comprehension. Recent psychology studies have characterized individual decision-making differences as model-free (fast and automatic) and model-based (slow and deliberative) decision-making. A framework has been proposed to estimate an individual differences in the degree of model-based control. Aims: This study investigates the correlation of the degree of model-based control and the comprehension performance of confusing source code and compares the correlation of comprehension performance between the degree of model-based control and developer skills reported in previous studies. Method: We conducted an observational study using source code with and without confusing code patterns. We measured the degree of model-based control for each participant. Results: Multiple regression analysis on the results of 91 software engineers showed that the degree of model-based control has a positive correlation with the percentage of correct answers for questions about source code with confusing code patterns with statistical significance. Conclusion: In source code reviews, refactoring, and enhancement development, the appropriate developer assignment criteria using developer attributes differ between source code with and without confusing code patterns.