Katrín Fjóla Aspelund¹ (), Ömer Dağlar Tanrikulu¹; ¹University of New Hampshire

This study examines whether the retinal size of objects influences ensemble size judgments in realistic contexts. Prior research has demonstrated that average size estimates are based on perceived size representations that are rescaled based on viewing distance. However, these studies often relied on explicit ensemble judgments on simplified stimuli that lacked naturalistic visual properties. To address this limitation, we use an immersive virtual reality (VR) environment with familiar objects and full-depth cues to investigate whether participants rely on retinal size or perceived size when averaging objects at distinct depth planes. Participants completed a two-alternative forced-choice (2AFC) task in a realistic supermarket VR environment, comparing the average sizes of two groups of virtual flour sacks placed on checkout counters at distances of 2.5m and 5m from the observer. We used a method of constant stimuli to calculate participants' Point of Subjective Equality (PSE) values. The closer group had a fixed mean size, while the farther group varied in mean size. Results showed a bias in participants' judgments, with the farther group needing to be slightly larger to appear equal to the closer group. PSEs were larger than the fixed mean of the closer group, indicating partial size-distance rescaling. Despite the availability of rich depth cues and realistic object properties, ensemble size judgments were still influenced by retinal size, indicating a measurable bias toward retinal size in average size judgments. Unlike prior studies assuming a linear relationship between retinal and perceived size, our 2AFC method avoided this assumption. The task was also implicit and realistic, asking participants which group contained more flour, simulating practical decision-making. Our results highlight the importance of studying ensemble processing in realistic environments to understand how it functions in daily life.