Eye tracking in expertise assessment case studies

This research investigates the cognitive mechanisms behind a person's analysis of source code (Python programming language) images and error detection, focusing specifically on the impact of professional visual error search skills in Python code on eye movement control. Method: The study employed eye-tracking technology using the "Neurobureau" system for psychophysiological research. Programmers, with 1-13 years of experience, were tasked to 1) explain and 2) find errors in 10 Python code stimuli, featuring syntax highlighting. The stimuli were normalized for length and complexity. The tasks were not time-bound. Results: The study discovered that with increasing professional skill, programmers develop efficient eye movement strategies, characterized by fragmenting the code into analytically significant units. More experienced programmers displayed fewer fixations, shorter scanning paths, and larger saccade amplitudes. Notably, there was an increase in the speed of saccades, especially in large searching movements, correlating with professional skill in code explanation tasks. Error detection in visual searches was found to be primarily influenced by recognizing text details, relying on the semantics and grammar of the programming language. This differs from the processing of natural scenes or texts in natural languages. Professional experience was observed to reduce the effort required in such activities. Conclusion: The study highlights how professional experience shapes eye movement strategies in source code analysis, differentiating it from natural language reading. These findings can inform the development of neuromorphic algorithms for code generation and correction; for an automated assessment of professional skill. No conflicts of interest were identified in this research.