Towards a Fine-grained Analysis of Cognitive Load During Program Comprehension

Program comprehension is inherent to all software development activities. This task may require a high mental effort (or so-called “cognitive load”), which in turn can hinder the performance of developers. In the literature, several authors have investigated the ability of biosignals to estimate developers' cognitive load during program comprehension. While the majority of these studies provide estimates at the task level, we aim for a more fine-grained level of analysis allowing to pinpoint the critical parts of code that could be associated with cognitive load. We infer these critical parts solely from eye fixation features and investigate qualitatively their relationship with those perceived as challenging by users. Being able to pinpoint critical parts in the source-code, is a first stride towards a very handy approach providing targeted support to developers to prevent them from committing errors. Furthermore, such a lightweight approach can be adapted in online settings.