Visual crowding, the challenge in recognizing peripherally presented objects surrounded by distractors, can be mitigated through perceptual learning. This study examines the learning-induced reduction of crowding using a Landolt-C gap-detection task, investigating whether learning in one visual hemifield (left/lower) transfers to the respective opposite hemifield (right/upper), and vice versa. Participants had to indicate on each trial the relative position of the opening of the Landolt-C (gap detection: top, bottom, left, right, 4AFC) that always was presented at 10° eccentricity and that was either radially or tangentially flanked by two ring-shaped distractors. Sixty-six normally sighted subjects (aged 18-45) underwent training in either the left/right or upper/lower visual hemi-field on the Landolt-C task over four sessions. During training, a 2-down, 1-up adaptive procedure adjusted the spatial distance of flanker distractors to the Landolt-C, resulting in 70.7% correct gap detection thresholds. Pre- and post-training, participants performed the task in both hemifields at a fixed target-flanker distance of 2° eccentricity for radial and tangential flanker conditions. Results reveal significantly improved performance in the task with the fixed target-flanker distance after training in both trained and untrained hemifields. However, the learning effect is smaller in the untrained hemi-field. The findings suggest that perceptual learning of a Landolt-C optotype crowding task yields improvements that partially transfer to the respective locations in the opposite hemifield. This finding could benefit training programs for persons with central vision loss, who rely on peripheral vision for daily tasks.

Acknowledgements: This research was supported by DFG (PL641-1/1) and ZSER e.V.