Coordination Dynamics meets Active Inference and Artificial Intelligence (CD + AI2): A multi-pronged approach to understanding the dynamics of brain and the emergence of conscious agency

How do humans discover their ability to act on the world? By tethering a baby’s foot to a mobile (Fig. 1a) and measuring the motion of both in 3D, we explore how babies begin to make sense of their coordinative relationship with the world and realize their ability to make things happen (N= 16; mean age = 100.33 days). Machine and deep learning classification architectures (e.g., CapsNet) indicate that functionally connecting infants to a mobile via a tether influences the baby movement most where it matters, namely at the point of infant∼world connection (Table 1). Using dynamics as a guide, we have developed tools to identify the moment an infant switches from spontaneous to intentional action (Fig. 1b). Preliminary coordination dynamics analysis and active inference generative modeling indicate that moments of stillness hold important epistemic value for young infants discovering their ability to change the world around them (Fig. 1c). Finally, a model of slow~fast brain coordination dynamics based on a 3D extension of the Jirsa-Kelso Excitator successfully simulated the evolution of tethered foot activity as infants transition from spontaneous to ordered action. By tuning a small number of parameters, this model captures patterns of emergent goal-directed action (Fig. 1d). Meshing concepts, methods and tools of Active Inference, Artificial Intelligence and Coordination Dynamics at multiple levels of description, the CD + AI2 program of research aims to identify key control parameters that shift the infant system from spontaneous to intentional behavior. The potent combination of mathematical modeling and quantitative analysis along with empirical study allow us to express the emergence of agency in quantifiable, lawful terms.