Personalized Spatial Remapping of Text Improves Word Reading in the Presence of Simulated Central Field Loss

Shifting letters to peripheral retinal locations during reading (remapping) may improve performance for people with central field loss (central scotomas). Shapes of scotomas vary from person to person, and we hypothesized that different remapping strategies would be better for different scotoma shapes. We tested normally-sighted participants with simulated central scotomas that were either: A) circular, B) horizontally-elongated and C) vertically-elongated. Remappings were constructed using results of a letter-based perimetry task that measured individual letter recognition performance across the visual field for each scotoma condition. We tested five remapping strategies, and describe letter placement from left to right. Letters that would have been covered by the scotoma were: 1) placed on a curved path above or below the scotoma along a diagonal trajectory (‘diagonals’), 2 & 3) shifted horizontally or vertically to bypass the scotoma (‘horizontal gap’ and ‘vertical gap’), 4) shifted vertically individually to the location in that column with best recognition accuracy (‘max accuracy’), 5) shifted together to the best continuous row of the display that was uninterrupted by the scotoma (‘max row’). Words ranging in length from 3 to 10 characters were presented for 200 msec along the remappings. Participants (n = 30; 10 per scotoma condition) typed the word they saw, and word recognition accuracy was calculated for each remapping. Performance differed reliably between remapping strategies for the three scotoma shapes (all ps < 0.001; all η2 > 0.53). The max accuracy remapping resulted in the worst performance across all scotoma types, likely because letters jumped vertically along the trajectory multiple times due to the letter locations being selected without adjacency constraints. The other more spatially continuous remapping strategies elicited better word recognition. Critically, the best remapping varied between scotoma conditions, indicating that personalizing remapping strategy based on scotoma shape optimized remapping performance.