Altruism and cooperation, which impregnate human behavior, are usually explained by either kin or group selection mechanisms or by direct and indirect reciprocity. Direct reciprocity can be explained by the existence of some memory device capable of storing previous actions or by limited dispersal. However, the evolutionary origin and stability of indirect reciprocity can be explained only when the actions are observed and classified by the members of society with the help of a social norm. Knowing the actual score (reputation) of a potential recipient (and of the donor) and the norm followed by the potential donor, she can decide whether its recipient is worth of donation or not. If freeriders are excluded effectively from the interaction by this norm, indirect reciprocity can be maintained. My project presents two problems connected with the evolution of cooperative social norms and the effectiveness of altruistic punishment based on strong reciprocity. Indirect reciprocity and systems of norms in general are based on information about the social status of the potential partner in the population. Most models assume that individuals are well informed about the reputation of their partners due to direct perception or reliable information transfer among individuals. Although different algorithms are applied for the information transfer in the theoretical papers, they have a common nature, viz.,they assume that communication is honest, and, hence, the information is reliable. Similarly, there is no incentive to cheat in the experimental settings. However, this is not sufficiently the case in the real world: in addition to information transfer, communication also enables manipulation, mainly by the social denigration of others. I am interested in the question whether reliable information transfer can co-evolve with a cooperative system of norms in a population structure characteristic of ancient humans. I use the general framework of the indirect reciprocity game, and I assume that individuals follow second-order norms and that cooperative or defective actions are classified as ´good´ or ´bad´ according to their norm. In this new model, the behavior of the potential receiver is estimated by the information from the observers of the previous action of the receiver. Is it possible that cultural selection maintains honest communication? If honest communication is possible, what are the main factors responsible for it? Are there polymorphisms in gossip rules in the evolutionary equilibrium? In the second part of the project I study how social and environmental circumstances impinge on the success of egalitarian and strong reciprocity strategies. It is generally accepted that humans follow ´strong reciprocity´ in social dilemmas, i.e., that they obey cooperative norms and (altruistically) punish its violators. The role of strong reciprocity in maintaining cooperation among unrelated human individuals is questioned by some recent experiments, which highlight that there is an alternative explanation for the punishment of norm violators. Social norms can be effective only if participants analyze situations correctly. Moreover, social norms maintain cooperation among unrelated individuals only if the information transfer among individuals is widespread and reliable. Thus we have the following hypotheses: If ´social noise´ is high in the population, it is much easier to evaluate individuals according to their income rather than their behavior, i.e., the egalitarian strategy is more effective than cooperative norms. To study this hypothesis, I work out a model that follows up on Fehr and Gachter´s (2002) experimental set-up studying these questions: Are there circumstances in which strong reciprocity is the winner of the selection, while in other cases the egalitarian motive is the better strategy? Does social noise alone determine the outcome of the selection, or are the costs and benefits of the games important as well?