Socio-technical systems are characterized by high structural and behavioral complexities, which impede understanding and modeling of such systems. In particular, reciprocal relations between diverse local system processes that determine global system dynamics are not well understood. In this article we focus on the problem of establishing relations between cognitive processes that determine the social dynamics of an actor in a socio-technical system. To address this problem, a formal, integrated, agent-based modeling approach is proposed. The approach provides rich analysis possibilities of qualitative and quantitative properties, specified at local (agent) and global (system) levels. An application of the approach in the context of a subway evacuation case is described in the article. Copyright © 2011 Taylor & Francis Group, LLC.