Abstract:

A great challenge in Social Neuroscience is to deal simultaneously with information from neural, behavioral and social levels. ‘Hyperscanning’ opens up a paradigm to study social interaction at multiple levels but faces the limits of experimental control inherent to investigating human interactions. We have proposed Virtual Partner Interaction (VPI) or the “human dynamic clamp” as a surrogate system to investigate human social behavior. This paradigm allows for real time interaction between a human partner and its computational mirror. Like the human, the virtual partner (VP) is described by nonlinear differential equations consisting of two terms: one for the intrinsic dynamics governing the VP, and the other specifying how the VP couples in real-time with a human subject. Parametric variations of the two terms open windows into a rich variety of social behaviors. Different time-scales of social interaction can be studied through modulation of the virtual partner’s (VP) intrinsic dynamics, including rhythmic coordination with the Haken-Kelso-Bunz model, discrete movement coordination with the Jirsa-Kelso Excitator model, and pace coordination by adding adaptive capabilities to the Excitator. We show that learning and social memory may be studied by enhancing the coupling in the Schöner-Kelso model of intentional coordination, thereby opening up opportunities for rehabilitation and therapeutic applications. Finally, we integrate the neural level with the human dynamic clamp by sampling neural activity of the human partner using suitable brain recordings (e.g. EEG, fMRI, PET, NIRS) on one side, and by plugging a realistic neurocomputational model into the virtual partner on the other side. The human dynamic clamp generalizes previous paradigms and opens up new possibilities for the development of “social computational neuroscience”.

Keywords: Dynamical Systems, Social Interaction, Computational Modelling, Human-machine Systems, Coordination