Abstract
Drastic changes in visual processing occur in the first years of life, yet little is known about how visual cortex develops during this period in humans. Retinotopic mapping has been performed in children as young as five, and the organization of their visual cortex was found to be adult-like (Gomez et al., 2018). However, whether this is true in younger infants and toddlers is unclear. On one hand, visual acuity and functions improve substantially over the first 12 months (Maurer & Lewis, 2001), suggesting that visual cortex may undergo reorganization. On the other hand, a hierarchical and topographic proto-organization has been observed from birth in non-human primates (Arcaro & Livingstone, 2017). Here we investigate how the organization of human visual cortex develops by conducting retinotopic mapping using fMRI in infants. Retinotopy in infants is a challenge for two key reasons: (1) they provide low quantities of data because of motion and fussiness, and (2) it is not possible to ensure fixation during stimulus presentation. These issues make modern travelling-wave paradigms infeasible. We instead adopted block designs that were robust to eye movements: meridian mapping to identify polar-angle boundaries between areas (Schnieder et al., 1993) and full-field spatial frequency modulation to map eccentricity (Arcaro & Livingstone, 2017). These tasks engaged infants, allowing us to obtain sufficient data in 6 sessions with 5 infants between 5 and 18 months. In all participants, we observed strong and selective evoked fMRI activity in visual cortex that distinguished horizontal versus vertical meridians and high versus low spatial frequencies. In infants over a year old, for whom we have succeeded in making surface reconstructions for flat mapping, we see clear evidence of hierarchically and topographically organized visual areas. In these participants, we are able to delineate and quantify regions up to ventral V4 and dorsal V3.