Humboldt-Universität zu Berlin - Lebenswissen­schaftliche Fakultät - Institut für Psychologie

Proposal for a Collaborative Research Center

IDEAL

Interaction Dynamics in EverydAy Life: From individual decisions to adaptive systems

Speakers:

Prof. Dr. Ursula Hess

Prof. Dr. Jens Krause

Interactions between individuals are highly dynamic, both from a temporal and spatial perspective: they ebb and flow, grow and whither and go in and out of synchronization. These interactions are often highly complex. These exchanges occur over time and are dynamically dependent on each other.  Importantly, these dynamics can be observed at different levels of organization. For example, at a more micro level of individual interactions, people who like each other will synchronize their expressive behavior in contrast to those who do not (Hess & Fischer, 2013). At a more macro level, interactive principles of synchronization with those nearby can also explain the behavior of larger groups. Interestingly, such organizing principles can be found for human and non-human groups. Thus, the way in which a crowd of football fans leaving a stadium responds to an obstacle on the way bears close similarity to how a school of fish reacts to an obstacle along the way (Couzin & Krause 2003). Consequences of dynamic interactions range from the building and breaking of friendships, to the finding of new jobs and the migration to other countries, as well as the build-up of aggression in groups and the transmission of disease but also of ideas among socially connected individuals. As such, a better understanding of dynamic interaction processes, how they develop over time, to what degree they depend on the homogeneity versus heterogeneity of the group and to what degree they scale up and down between group sizes is highly relevant for many domains in society.

The study of social interactions has a long tradition in a diverse range of fields including psychology, sociology, social sciences, biology. More recently, computer scientists and physicists have begun to study interactions between virtual agents or robots and humans. These different research domains, have tradiontally developed largely independent resulting in dispersed knowledge across disciplines.

In recent years, statistical physics has developed tools for modeling large-scale interactions and networks. In parallel, the development of sensor systems and their miniaturization by electronic engineers and computer scientists allows large-scale automatic encounter detection and the tracking of free ranging animals as well as the quantifation of physical and/or environmental conditions under which encounters take place (Dressler et al. 2016). These approaches have resulted in rich data that can be applied to the study of social interactions. Yet, while we tend to understand the structure of networks, much less is known about the interaction dynamics that create this structure (Martin, 2009; Schröder, Hoey, & Rogers, 2016). Conversely, when we have detailed information on the interaction dynamics, we tend to lack information on whether these interaction dynamics can be scaled-up. To illustrate, psychologists tend to well understand synchronization processes in single individuals who react to a standardized (i.e., non-reactive) target (Hess & Fischer, 2013). Yet, whether the same interaction dynamics are obtained for dyads and larger groups is unclear. Similarly, when co-regulation in couples is observed over time, analyses typically focus on small-scale interactions (Hoppmann & Gerstorf, 2016), raising the question of whether these interactions depend on the characteristics of specific individuals or whether more general dyad-level mechanisms and patterns can be discerned. By contrast, researchers who focus on large groups and networks often can discern such patterns (e.g., Couzin & Krause 2003), but in turn are faced with the question of whether individual differences become more and more important as the observed groups get smaller and smaller (Herbert-Read et al., 2013).

In traditionally disconnected scientific fields such as theoretical biology and psychology, we are now in a position to measure and compare interactions between social animals or humans at a quantitative level where every single interaction between individuals in large populations and networks over extended periods can be detected and quantified. Importantly, there is still a large gap between research that focuses on the level of the individual and research aimed at understanding the dynamics that underlie the behavior of larger groups.

To further this knowledge is the goal of the CRC IDEAL. The research program is guided by the main aim to adapt the perspectives and methods used for the study of large scale interactions and the study of individuals to understand the collective behavior in humans and other organisms. By combining the two approaches, it becomes possible to, on the one hand, better detail the processes that underlie similar behaviors in humans and other organisms, while on the other hand to abstract from small-scale views to large-scale organization. 

IDEAL aims to address this challenge by combining methods and insights from different disciplines to allow us to accommodate this scaling problem for human and non-human interactions. Specific focus will be on the question of whether and how we can generalize from the individual to the group and back from the group to the individual so as to gain a better understanding of the processes that underlie the dynamics of interaction between and among human and non-human entities. Our multidisciplinary approach is essential to this endeavor because the combination of the different knowledge systems and methods developed within the disciplines involved will allow us to transcend the traditional limits of inquiry within disciplines. In particular, IDEAL is guided by a philosophy that brings together researchers who are experts in dynamic interactions and who study these at different levels of social organization. All research groups of the CRC have designed projects for the first four years which include cross-links to researchers who consider related processes at a different level of analysis. The goal of these cross-link projects is to share experience and expertise and to devise new research designs and apply new research methods that will allow this generalization. These new approaches will then be employed in follow-up research for the second four-year period and finally rounded out with transfer projects in the last four-year period.

IDEAL will address three core questions:

1. Generalizability: How does social behavior scale with group size? When and how can properties of groups be predicted from knowing the characteristics of its constituent individuals?
2. Individual differences: Does it matter who interacts with whom, that is, do individual differences matter? Or is it more relevant how individuals interact? In other words, is it the nodes of a network (the organism) or the links of a network (the interaction mode) that drive the relevant effects or processes?
3. Temporal structure: What types of changes in interactions networks are observed over time? What causes and drives these changes? When we observe interactions over time, what is the optimal temporal resolution to study the phenomenon under consideration?