Florence Bouhali^1,2, Ting Qi^1,3, Maria Luisa Mandelli¹, Cheng Wang^1,4, Rian Bogley¹, Jocelyn Caballero¹, Rayburn Tang¹, Ellie Carpenter¹, Maria Cecilia Ferrer Ladao¹, Christa Pereira Watson¹, Kevin Weiner⁵, Maria Luisa Gorno Tempini¹, Fumiko Hoeft⁶; ¹University of California San Francisco, ²Aix-Marseille University, France, ³Beijing University of Posts and Telecommunications, China, ⁴Fujian Medical University, China, ⁵University of California Berkeley, ⁶University of Connecticut

Developmental dyslexia is a specific learning disorder characterized by challenges in fluent reading, despite conventional instruction and average to superior intelligence. Interestingly, many studies also report domain-general challenges in visual processing in individuals with vs. without dyslexia, with less fine-grained exemplar discriminations of visual stimuli (e.g., faces). Moreover, less neural adaptation to repeated stimuli, including images of words, faces and objects, has been reported in dyslexia. Several theories therefore propose that these characteristics present before literacy instruction may be causally implicated in the emergence of dyslexia. Yet, less adaptation to various visual categories has also been observed in illiterate adults, together with differences in the functional organization of category-selective regions in ventral occipito-temporal cortex (vOTC), thereby questioning the causality between reading proficiency and general visual processing. Here, we test the impact of typical and atypical reading development on visual processing in 39 children with dyslexia and 36 without, aged 7 to 17, using a standard repetition detection (one-back) localizer task in the fMRI. We find altered stimulus-level representations in children with dyslexia or young readers with low raw reading scores, with significantly lower hit rates across all categories (words, numbers, limbs, faces, objects and places), regardless of ADHD and other confounding variables, and neural differences in repetition effects. Visual processing differences were also found at the categorical level: dyslexic/young readers showed lower categorical distinctiveness (i.e., lower differences in multivoxel activation pattern similarity computed within a category than across categories) in the left and right vOTC for bodies, faces and words – suggesting noisier neural representations. Our results therefore confirm domain-general visual challenges in dyslexia, linking behavioral and neural effects. Together with results in illiterate adults, our work further suggests that these visual challenges in dyslexia may be secondary to reading difficulties, with implications for existing theories of dyslexia.