Towards human-centric software complexity metrics: a neuroscience-based approach

Current code complexity metrics used in the software industry do not consider the human factor from an individual perspective (i.e., the programmer, or tester). In software development, the developers play a critical and fundamental role to accomplish each development activity, and the limitations induced by them are inevitable. Therefore, the ability to capture perceived code complexity accurately is very important for the software quality and its adequate maintenance. This Ph.D. thesis focuses on exploring the possible approaches to incorporating a human-centric dimension in software complexity assessment and management. To reach this purpose, we use the information captured from wearable and non-intrusive devices (e.g., ECG and Eye-tracking with Pupillography) to identify the relevant cognitive states (such as mental effort, stress and mental fatigue) of individual developers during the software development process, and use such data to develop new complexity metrics that consider the genuine human perception of the challenges in comprehending code. Expected new human-centric code complexity metrics must consider the developers’ perspective and provide an accurate assessment of software complexity to improve the software’s quality and reduce bugs, time consumed, and unnecessary costs.