Arda Fidanci¹ (), Colin S. Flowers¹, Sophia Osovsky¹, Gordon E. Legge¹, Stephen Engel¹; ¹University of Minnesota

Remapping is an assistive approach for people with central vision loss (CVL) that works by relocating letters hidden by a scotoma to functioning retina outside the scotoma. How letters are relocated—for instance, by warping them around the scotoma versus moving entire lines up or down—can differentially affect letter recognition (Fidanci et al., 2024) and reading (Flowers et al., 2024). Here, we tested whether the differences in word reading due to various remappings can be predicted from letter recognition performance. We measured letter and word recognition in 10 young controls with three simulated scotoma shapes (circular, horizontally elongated, and vertically elongated). We compared five remapping strategies from prior work : (1) Maximum Row, moving letters onto the single row with highest overall letter-recognition accuracy; (2) Horizontal Gap, shifting letters horizontally to jump over the scotoma; (3) Vertical Gap, shifting letters up or down to bypass the scotoma; (4) Diagonals, positioning letters on diagonal paths around the scotoma; (5) MaxAcc, relocating each letter to the most accurate location within its column. In the letter recognition task, three letters were flashed for 200 ms at successive positions along the remapping path while maintaining central fixation. In the word recognition task, words, ranging in length from 3 to 10 letters, were presented for 200 ms along the same remapping paths. Letters and words were presented in a total of 15 conditions (strategy x scotoma shape). The letter and word recognition accuracies varied widely across conditions and were best overall in the Horizontal Gap condition (letter = 83.42%; word = 77.98%) and worst in the MaxAcc condition (62.18%; 62.65%). Across all conditions, letter and word recognition correlated highly with each other (r = 0.76, p = 0.001). The spatial configurations in remapping strategies impact letter recognition differentially, which is what likely produces differential word-reading performance.

Acknowledgements: NIH Grant EY030890