Description

Coordinating one’s behavior with the behavior of other individuals is a fundamental feature of everyday social interaction. A defining feature of such behavior is that it is dynamic, that is, it evolves over time. This is true whether one considers the linguistic, gestural and non-verbal coordination that occurs between two or more individuals engaged in a conversation or the physical movement coordination that occurs when two or more people clear a dinner table or load a dishwasher together. Such behavior is also emergent and robust to sudden changes in task context or unexpected environmental or social perturbations. Accordingly, robust social action and multi-agent coordination is synergistic, with co-acting individuals adapting to each other and the environment around them in a mutual and reciprocal manner. Research investigating the behavioral dynamics of joint-action and multi-agent coordination has steadily increased over the last several decades. Spurred by several factors, including (i) the increased accessibility of technologies for recording and extracting the time-evolution of multi-agent behavior (e.g., motion tracking, eye-tracking, EEG), (ii) the development of new nonlinear techniques for analyzing behavioral and linguistic time-series data, and (iii) a growing appreciation that social cognition, perception, and action are interdependent, embodied and embedded processes, this research has not only been directed towards measuring and identifying the stable patterns of coordinated social and multi-agent activity that emerge over time, but also how these stable pattern are activated, dissolved, transformed, and exchanged over time. Not surprisingly, researchers have begun to investigate the implications of this behavioral dynamics perspective for understanding social cognitive processes as well as clinical disorders with social deficits such as autism and schizophrenia. Attempts at modeling the dynamics of social action and multi-agent behavior using various nonlinear and complex systems methods has also increased over the last several years, with many researchers demonstrating how simple low-dimensional dynamical or computational models can be employed to capture and explain the dynamics of ongoing joint-action and multi-agent behavior. A characteristic feature of these dynamical models is that they reveal how stable social action and multi-agent coordination arises naturally from the interaction of the physical, biomechanical, neural, informational, and social cognitive properties of a joint-action task context and goal, and cannot be ascribed to any one singular processes, agent, or level of analysis. The implications of these modeling endeavors for the design of robust and adaptive human-machine systems and robotic agents has not gone unnoticed, with a growing body of work now devoted to such joint-action, (bio)-inspired human-robotic interaction initiatives.