Innovation networks today are widely considered as a promising tool for the organization of industrial R&D bringing together heterogeneous actors and their heterogeneous knowledge bases. For the analysis of the architecture and the dynamics of innovation networks it becomes necessary to explicitly focus on knowledge generation and diffusion processes in the networks - the recourse to accumulated R&D capital which characterizes models in a neoclassical fashion is not sufficient. The analysis of innovation networks has to cope with the interactions of large diversified firms, technologically specialized start-up companies and public research institutes which all might be connected in innovation networks. Agent-Based simulation models (ABMs) are designed for the analysis of interactions among heterogeneous agents and offer an alternative to standard approaches of the mainstream. ABMs allow for an explicit treatment of mutual knowledge exchange and joint knowledge development amongst the various actors participating an innovation network.

This presentation will introduce to the basic requirements of modeling learning and innovation in networks and contrast these requirements with the status-quo in the theory of industrial organization. In this discussion it is shown that technological spillovers are not involuntary knowledge flows but strongly depend on the linkages among agents and their competencies to integrate external knowledge. As an alternative to mainstream approaches conceiving knowledge as public good which freely spills over among agents, we introduce to an ABM of innovation network evolution in a Neo-Schumpeterian fashion. This model explicitly addresses the heterogeneity of the knowledge of actors as well as the creation of linkages among them which allow for a meaningful knowledge exchange and joint development of new knowledge.