Dynamic gaze metrics in virtual reality: an AOI-based methodological framework applied to sexualized virtual characters

Abstract

Eye-tracking research has traditionally relied on standard metrics, such as fixation-related and time-based measures, to characterize visual attention. While informative, these measures primarily describe where attention is directed and for how long. They are less suited to capturing the continuous spatial and dynamic organization of gaze behavior, which is particularly relevant in VR environments that allow head movements and involve dynamic stimuli. The present study proposes a methodological framework that extends conventional fixation-based analyses by integrating complementary AOI-centered dynamic and entropy-based measures. The framework was applied to sex-related gaze patterns toward neutral and sexualized virtual characters as an empirical case study. Specifically, fixation count and dwell time were used to characterize attentional allocation; Gaze Radial Angular Deviation (GRAD) mean quantified the average angular distance from the AOI center; the SD and CV of GRAD characterized the absolute and relative spatial variability, respectively; correlation dimension D2 captured the temporal complexity of the GRAD signal; and spatial gaze entropy (SGE) and gaze transition entropy (GTE) described the broader distribution and transition structure of the scan path across the AOIs. Bringing these measures together within a single analytical framework provided a more comprehensive characterization of gaze behavior than conventional fixation-based summaries alone. The proposed approach extends the characterization of gaze beyond conventional fixation-based analyses and provides a broader account of visual behavior in immersive environments.