The paper presents an evolutionary model illustrating the dynamics that give rise to discriminatory social norms i.e. such rules of behaviour that fulfi l two conditions: (1) they treat differently actors having the same abilities and technical options, but differing in some arbitrary sense (2) they are supported by socially enforced sanctions. In the presented model both discrimination and social norms are necessary to solve a coordination problem that arises when the situation requires different actors to perform different tasks. The properties of behavioural rules relying on discrimination and leading to various degrees of inequality are analysed. It is demonstrated that in general norms ensuring equal payoffs are easier to stabilize, but unfair norms can also be stable.

The article presents an empirical research which aimed at verifying if participation in a cooperation or competition training determined future behaviors during an anonymous, two-person trust game. We applied two variants of training: active and passive. The main hypothesis claimed that participation in a cooperation training activates the social norm of cooperation and induces more cooperative choices in trust games, whereas participation in a competition training reduces the willingness for cooperative choices. We also predicted that an active training would be more effective in activating cooperation and competition norms than a passive one. Finally, it was expected that interrelation of a cooperative decision in a more risky game would reduce the influence of a training-activated norm on the final decision. The results of the experiment confirmed all three hypotheses.