Multi-dimensional networks as a tool to model, analyze, and interpret multi-subject brain connectivity in hyperscanning settings^*

Social interactions are woven in the fabric of our lives, from birth to adulthood. Studies show that when social connections are lacking, there is a high probability of developing poor mental and physical health. Conversely, some mental disorders result in a compromised ability to form these interactions, with autism being a prime example. For these reasons, understanding the neural mechanisms underlying social interactions is crucial. Hyperscanning experiments, where brain activity is simultaneously recorded from multiple subjects during social tasks, have emerged as a powerful tool for tackling this issue, and the study of brain-to-brain connectivity has provided significant insights into social neuroscience. However, no agreed-upon methods exist for analyzing and interpreting such data, and most of the analyses still rely on techniques designed for single-subject experiments. Here, we propose a framework to model brain-to-brain connectivity through multi-dimensional networks. Under this conceptualization, both intra- and inter-brain connectivity are viewed as components of a higher-order network that can be studied by extending graph theory indices to a multidimensional case. In this article, we also discuss how these indices can be interpreted in the context of multi-subject networks. Future investigations will apply this framework to hyperscanning datasets to provide new insights into the neural circuits associated with social interactions.