"Recognizable traits (gender, ethnicity, hair color) often provide a basis for negative discrimination, even if there is no statistical basis for differentiation. Still, people act differently towards persons with different traits, as if they had different qualities, for instance with regard to work efficiency. This leads to severe conflicts if discrimination provokes prejudice and stigmatizing that in practice mostly target asymmetrically members of one, disadvantaged category. The proposed research program enriches existing theories about the emergence and maintenance of discrimination by a new perspective: it aims to model mechanisms based on social network relations and incorporates them into the explanations. It tries to find answers under which structural configurations discrimination practices are likely to occur and under which network conditions they can be best avoided. For answering these questions, innovative methodologies will be used: agent-based simulation and experiments. These techniques have only been used to a limited extent targeting the problem of discrimination, but offer a huge potential to extend our limited insight into social mechanisms that maintain discrimination practices. With the help of these methods, the proposed research tries to delineate conditions under which an endogenous breakdown of negative discrimination is promoted by social networks. Particular attention will be devoted to cases of fierce intergroup opposition, in which network segregation is aligned with negative discrimination. Results are expected to help forming efficient policies for fighting discrimination and segregation. The practical objectives of the application are to further develop the skills of the applicant in applying these methods with a help of leading experts, and to embed the research program in a multi-disciplinary environment, where important contributions can be expected from sociologists, economists, computer scientists, and historians."