Understanding Software Merge through Empirical Studies and EEG Data

This presentation explores software merging from empirical, technical, and neurocognitive perspectives, investigating how developers integrate software artifacts, manage conflicts, and experience cognitive effort during integration tasks. It summarizes industrial studies involving 297 UML model compositions, more than 2,500 classes, and 56 weeks of observation in the oil and gas industry. Findings reveal that detecting and resolving inconsistencies requires substantially more effort than applying semi-automated composition techniques. The presentation also examines how change characteristics, conflict complexity, and developers' technical reputation influence integration effort. Building on these findings, the presentation introduces an experimental investigation in which 42 participants performed source code merging tasks while researchers monitored their brain activity using electroencephalography (EEG). The analysis of EEG microstates reveals distinct neural patterns associated with task performance, highlighting the potential of neurophysiological measurements for understanding developers' cognitive processes. The presentation also introduces the openly available BrainMergeEEG dataset and discusses opportunities for developing cognitively adaptive software engineering environments. Overall, it connects empirical software engineering with neuroscience to advance our understanding of software integration and support more effective, human-centered development tools.