The modeling of opinion dynamics in social systems has attracted a good deal of attention in the last decade. Even though based on intuition and observation, the mechanisms behind many of these models need solid empirical grounding. In this work, we investigate the relation among subjective variables (such as the personality), the dynamics of the affinity network dynamics, the communication patterns emerging throughout the social interactions and the opinions dynamics in a series of experiments with five small groups of ten people each. In order to ignite the discussion, the polemic topic of animal experimentation was proposed. The groups essentially polarized in two factions with a set of stubborn individuals (those not changing their opinions in time) playing the role of anchors. Our results suggest that the different layers present in the group dynamics (i.e., individual level, group dynamics and meso-communication) are deeply intermingled, specifically the stubbornness effect appears to be related to the dynamical features of the network topologies, and only in an undirected way to the personality of the participants.

The modeling of small group dynamics represents a hard challenge despite the effort of disciplines such as sociophysics and social psychology. The interaction between the complex topology of human social structures/communities and the cognitive processes characterizing humans at the microscopic level, are the focus of the classical social cognition paradigm, and it has been deeply researched in the last century. In the present study we used a web based Chat room as an experimental environment for the study of social interactions within a small group of people. The target of the present study is to explore the relations between the affinity among individual and their communication dynamics. We designed three different experimental tasks (social problem), with a crescent degree of social complexity, in order to test the impact of different social constraints on the evolution of the affinity network, as well as on the dynamics of communication. Our aim is to define the "cognitive recipes" used by the subjects to solve the required social problems. Our results show that the complexity of the social problem affects the relation between affinity and communication networks, influencing at the same time both affinity and opinion. We use the sociophysics and social cognitive models in order to interpret the results, showing the limits of the most diffused sociophysics models when aiming at forecasting the dynamics of a small group.