Modeling the formation of extrastriate primate visual field maps

Topographic maps of visual space are pervasive throughout the primate visual system. The molecular factors responsible for generating the precise topographic map of primary visual cortex, V1 are well studied. However, our understanding of the mechanisms governing postprimary cortical development remains limited. It is assumed that extrastriate maps self- organize around an established V1 map (Rosa 2002). Recent research, modeling the development of extrastriate maps as iterative propagation from the V1 map, have demonstrated that activity-based connections and wiring density limits are sufficient to generate a hierarchy of mirror-symmetric maps of visual space (Imam & Finlay 2020). Here, we extend this model by incorporating the specific cortical topology and retinotopic organization observed in primates. We measured the retinotopic organization of visual cortex in macaque monkeys using fMRI (Arcaro et al. 2017). For each subject, we segmented the grey matter and reconstructed the cortical surface from high-resolution anatomical MRIs. We then constructed a network model in which distances between nodes corresponded to the measured anatomical distances along the cortical surface. Connectivity within the model was established using an activity-based developmental program. As the program unfolded, nodes in V1 preferentially formed connections with nodes outside of V1 based on activity-dependent correlations and edge density limits within the network. Our preliminary results reveal that this activity-based modeling approach can produce a series of mirror-symmetric maps in the approximate locations corresponding to visual areas V2, V3, and V4 found in individual monkeys. Ongoing work aims to assess the impact of different distance measures (cortical vs. volumetric) and species-specific (human vs. macaque) factors related to cortical folding. Rather than requiring area-specific specifications, this work supports the view that a stereotypical layout of cortical areas unfolds within a given species through iterative self-organizing rules, requiring only a prespecification of the primary cortical map.