Interpersonal Synchrony: Analyzing and Modeling Social Interaction Dynamics

The central focus in this dissertation is on modeling interpersonal interaction dynamics from underlying mechanisms and pathways. In this work, the aim was to let computational modeling and empirical science mutually benefit from each other. More specifically, the mechanisms based on generic and more specific relationships and pathways described in empirical literature were used as the foundation for a programmatic series of computational models. One of the major advantages of computational modeling is that it can test whether proposed correlations between such mechanisms and phenomena indeed can be confirmed computationally by in silico experiments (experiments through computer simulations). This allows researchers to verify whether the presumed relationships indeed lead to the emergence of certain observed phenomena. In recent years, interpersonal synchrony has attracted a great deal of attention from researchers addressing interpersonal interaction dynamics. In this dissertation, I explored how multimodal interpersonal synchrony can be simulated by interacting agents. One the one hand, this concerns the mechanisms that underlie the emergence of multimodal interpersonal synchrony among two agents taking into account internal cognitive and affective mental states and mechanisms from neuroscience. A specific focus was on modeling how adaptive agents can be able to subjectively detect interpersonal synchrony with other agents and how they subsequently can adapt the interaction behavior based on this subjective interpersonal synchrony detection. This adaptive interaction behavior has been differentiated into short-term and long-term adaptivity. In addition to detected synchrony itself, detected disruptions or transitions in synchrony can also have an effect on the adaptation of interaction behaviour. Within social interactions, relationship-specific and relationship-independent adaptation can be learnt. Patterns of transitions in synchrony and transferences of learnt interaction behavior over relationships have been investigated as well in this dissertation. The increased interest in interpersonal synchrony has also led to the development and application of a large number of statistical methods to detect synchrony in pairs of time series or in larger sets of time series. To achieve further progress in this field, I have developed a form of organization and structuring of all these methods and software to support the use of these methods by applied researchers. Moreover, it has been demonstrated how computational agent models can compare and integrate different concepts and methods in the study of interpersonal synchrony. I integrated and compared multiple time lags and multiple methods for synchrony detection and transition detection in my adaptive agent models. Furthermore, I considered how subjectively detected synchrony can differ for the two agents involved in an interaction and how the detected synchrony of both agents can differ from objectively detected synchrony by an external observer. The underlying idea behind this was that the type of adaptive behavior studied here, for each agent will be driven by its own subjective view and not by objectively detected synchrony from an external observer viewpoint.