Computing involving physiological data such as heart rate and electrodermal activity has been used as a way to enrich K-12 students’ computing learning experiences. This study explored a novel way to engage elementary students of color as “developers” through a series of physiological computing lessons during a summer learning program. Learners used real-time physiological data (i.e. muscle energy) as computer inputs to build computer programs. The study used an explanatory mixed method by combining multi-dimensional eye-tracking metrics data with qualitative investigation to reveal the visual attention of students with different programming proficiency levels. The regression analysis revealed three statistically significant predictors of students’ programming performances. Subsequent qualitative case studies provided additional insights into students’ problem-solving processes. Interpretations and caveats are presented and discussed.