Suyeon Kim¹, Oakyoon Cha¹; ¹Sungshin Women's University

Even a mundane visual scene, such as a crowd on a street, presents a vast amount of visual information. Facing this challenge, the visual system extracts statistical properties to reduce redundant information from groups. Last year, we found that variability judgments were influenced by local feature contrast, with this influence increasing as task-irrelevant variability increased. Building on Pascucci et al. (2021)’s finding that attention was unevenly distributed during ensemble judgments, we speculated that participants might have applied a narrower attention window when task-irrelevant variability raised the attention load. In the present study, we investigated whether task-irrelevant feature variability influences the distribution of attention during ensemble judgments. Participants briefly viewed a 5-by-5 array of oval shapes with varying orientations, and reproduced the average orientation of all ovals. We manipulated the color variability of oval shapes (Experiment 1) and the color variability embedded in the background (Experiment 2). Embedding colorful circles in the background allowed us to increase the task-irrelevant variability further in Experiment 2. For each trial, we calculated the average orientation in three different windows (one oval in the center, ovals in a 3-by-3 window in the center, and all 25 ovals) and assessed the contribution of each window to average judgments. In Experiment 1, task-irrelevant variability did not influence how the three windows contributed to the average judgments. However, in Experiment 2, task-irrelevant variability increased the contribution of the 3-by-3 window and reduced the contribution of the 5-by-5 window, suggesting that attention was less evenly distributed and shifted toward the center. These results suggest that task-irrelevant feature variability can influence the distribution of attention, thereby making ensemble judgments more reliant on fewer numbers of objects.

Acknowledgements: This work was supported by the National Research Foundation of Korea (NRF) grant funded by the Korea government (Ministry of Science and ICT) (No. RS-2023-00211668).