We consider various implications of information about the other player in two–player evolutionary games. A simple model of desertion shows that information about the partner's behaviour can be disadvantageous, and highlights the idea of credible threats. We then discuss the general issue of whether the partner can convince the focal player that it will behave in a specific way, i.e. whether the focal player can make credible threats or promises. We show that when desertion decisions depend on reserves, a player can manipulate its reserves so as to create a credible threat of desertion. We then extend previous work on the evolution of trust and commitment, discussing conditions under which it is advantageous to assume that a partner will behave in a certain way even though it is not in its best interest.