Batuhan Erkat¹, Dinesh Venugopal¹, Roksana Sadeghi², Tran Chau³, Will Gee³, Brittnee Livingston⁴, Gislin Dagnelie², Arathy Kartha¹; ¹SUNY College of Optometry, Department of Biological and Vision Sciences, New York, NY, United States, ²The Johns Hopkins Wilmer Eye Institute, Baltimore, MD, United States, ³BaltiVirtual, Baltimore, MD, United States, ⁴Central Association for the Blind and Visually Impaired, Utica, NY, United States

People rely on vision to guide them in crossing streets safely, yet it remains unclear how severe visual impairment alters visual behaviors, including head-scanning strategies, that support this visual guidance. To address this gap, we examined horizontal head-scanning behavior in a virtual street-crossing task with systematically varied environmental demands conducted with individuals with ultra-low vision (ULV). Ten normally sighted adults (NV) completed the task under both normal and simulated ULV (sULV). Fourteen adults with clinical ULV completed the same task across four complexity levels: A (empty road), B (parked cars with pedestrians), C (moving cars that yield to pedestrians), and D (yielding moving cars and pedestrians). Measures included latency to initiate walking, total absolute angular displacement along the yaw axis (total horizontal scan), scanning speed, and a pre- versus in-motion scanning index, indicating the proportion of scanning occurring before versus during walking. Before crossing, sULVs showed greater total scan and longer latencies than NVs (respectively, p = 0.004; p < 0.001), with similar scanning speed (p = 0.33). During crossing, total scan and scanning speed did not differ significantly between the NV and sULV conditions (p = 0.78; p = 0.35), nor between the NV and ULV groups (p = 0.50; p = 0.98). However, we observed that higher environmental complexity modulates total scanning. NV participants showed similar scanning index values across simple and complex conditions (−78%, p = 0.24), whereas both sULV and ULV showed decreased horizontal scanning during street-crossing in complex scenarios (−1.46-fold, p = 0.03; −0.61-fold, p = 0.002). Our results suggest that ULV participants exhibit more head-scanning before stepping off the curb and less during walking in visually demanding environments. By characterizing these behavioral differences, we identify specific behaviors that can be targeted through mobility training for individuals with ULV to support safer street-crossing.

Acknowledgements: R00EY033031 (AK), and R01EY028452 (GD), Research to Prevent Blindness (GD)