In recent years, the mechanism called "indirect reciprocity" has been drawing wide attention as a plausible candidate explaining the evolution of cooperation among primates, especially among humans. Several experiments on humans support that the "discriminator", who cooperates only with cooperators, plays an important role in it. However, in Public-Goods-game-type social interactions, one can find that adopting such a strategy yields no benefit to the players. By using "adaptive dynamics", I show that the discriminating strategy cannot emerge spontaneously from the asocial population. This result implies that we have to consider "which strategy wins" and "which strategy evolves" separately. In the latter half of my talk, I pay attention to punitive aggression in nature. When punishment exists, individuals tend to participate in cooperative behavior, otherwise they will suffer serious damage. The punitive aggression itself, however, also costs some to the actor, so we face another question; "Who pays that cost ?" This is the well-known "second-order free rider problem". I show that this can be solved when reputation exists in the population.