Perceptual Characteristics of Computerized Program Visualizations can Differentially Affect Detection and Localization Performance

Program visualizations (PVs) are sometimes less effective in teaching computer algorithms than desired. One reason may be that PV designers have largely ignored the users' perceptual capabilities. We examined perceptual characteristics of bar displays, similar to those used in teaching sorting algorithms. Within each experimental condition we varied the number of bars displayed, number of flashing cues, number of bars changing height, and whether bars were labeled. Across experimental conditions we examined placement of bars within the visual field and whether bar profiles were fixed or varied from trial to trial. Thirty-six university students participated. Bars placed in peripheral locations can harm performance because of human retinal eccentricity effects. Moreover, many perceptual characteristics that help localization of critical changes (e.g., labels and cues) do not help detection, whereas some that help detection (e.g., 2 simultaneous changes) hurt localization. Our results suggest that considera...