HYPHI(Φ): A PIPELINE FOR DETECTING GEOMETRIC PHASE TRANSITIONS IN HYPERSCANNING NETWORKS

Simultaneous recordings of brain activity from interacting people provide valuable insights into social neural dynamics. Yet, capturing dynamic shifts in inter-brain coupling, crucial for interpersonal attunement research, remains elusive with standard metrics. Here, we introduce a geometric framework for hyperscanning data dubbed HyPhi(Φ) that leverages discrete Ricci curvatures and their entropy to track phase transitions in brain-to-brain connectivity. We demonstrate the method’s sensitivity using simulations of coupled two-brain models and dual EEG experiments. HyPhi(Φ) provides a principled, broadly applicable way of characterizing time-varying networks of interacting dyads, and possibly uncovering the previously hidden dialectic of synchrony across brains.

Fig. 1 Overview of the analysis pipeline. (a) Dual neural signals are preprocessed and source-localized to estimate cortical activity. (b) Sliding-window inter-brain connectivity graphs are constructed. (c) Time-resolved distribution of network curvature is computed across edges. (d) Network entropy and its time derivative are derived to detect phase transitions in the evolving connectivity networks. Arrows indicate the flow of analysis, and color gradients denote curvature sign.