Inducing Global and Focal View in UML Class Diagram Comprehension

Human visual perception alternates between a global mode that captures overall scene structure and a focal mode that entails detailed inspection. Building on research in perceptual psychology and radiological expertise, we extend this global–focal framework to UML class diagram comprehension. We introduce a controlled methodology that combines task framing, gaze-contingent viewing (restricting peripheral vision), and information reduction (omitting textual details) to induce global versus focal viewing during UML class diagram comprehension tasks. In a within-subjects eye tracking study with 29 software engineering students, fixation durations are significantly shorter in the global condition than in the focal condition (202 vs. 234 ms, 𝑝 = .014), while saccade amplitudes are significantly longer (7.02◦ vs. 2.77◦, 𝑝 < .001). Together, these findings provide evidence that our theory-driven study design successfully induces distinct viewing modes under controlled conditions. The presented paradigm thus provides a foundation for studying how perceptual mechanisms and expertise-related differences shape UML class diagram comprehension.