Isaac Liao¹ (), David E. Osher; ¹The Ohio State University

The ventral visual stream consists of a mosaic of regions specialized for processing distinct high-level visual categories like faces, bodies, objects, and scenes. It has been argued that the functional activations of brain regions rely on their specialized connectivity patterns; the function of each region is determined in large part by its unique set of connections to the rest of the brain. However, the connectivity patterns that characterize these ventral visual regions are still poorly understood. Here, we characterized the connectivity fingerprints for 18 functional regions of interest (fROI) by building linear models to predict each region based on its connectivity patterns to the whole brain, at the level of individual subjects. We then compared the connectivity patterns that predict each fROI with their task-based response. We found that, in general, the activity of a fROI tends to be predicted by distant regions that share selectivity of the same stimulus domain. The connectivity fingerprints for a fROI also tend to be most similar to regions that perform the same function. Notably, some fROIs have characteristic connectivity patterns that may distinguish them within and between domains. Together, our results demonstrate common motifs in the connectivity patterns of high-level visual regions. Future work will focus on using TMS to non-invasively disrupt the regions that predict function in the ventral visual stream.