Face perception is a distinctive visual skill which is thought to be unique compared to the processing of other visual categories. Accumulating evidence point to the distributed nature of the face processing system,which is composed of a number of different regions that were described in detail. Distributed systems can be investigated using networks, which are a set of elements and their pairwise connections which allow characterizing each element’s connection pattern (i.e. a “connectome”). Network analyses were proven to be successful in many domains including social, biological and traffic networks. Accordingly, over the last several years, a number of face processing studies, including our own, have

begun to adopt a network, or a connectome methodology and framework to better characterize the functional aspects of this system. Nevertheless, the structural aspects of the network as well as the structural-functional-behavioral correspondence were not fully investigated. In the current proposal we plan to implement a framework for connectome embeddings (CE), adopting methods from natural language processing (i.e., work2vec algorithm), for creating vector representations of brain networks. This will allow us to integrate structure, function and behavior within the face domain. Moreover, we expect that the proposed research would provide a powerful platform for exploring the higher-order network structure of connectome data sets, with further potential applications in modeling and comparing individual differences in human connectomes across development and in clinical and translational studies. Thus, this approach has the potential to advance the field of cognitive neuroscience research in general