In chapter 8, we saw that cooperation can evolve if agents play repeated social dilemma games. We call this direct reciprocity. However, in many situations we interact with strangers just for one time, and we see high levels of cooperation. For example, the exchange on a market between a buyer and a seller, taking a taxi, or letting a car join the traffic.

Can we use models to understand the conditions in which we can expect high levels of cooperation if we don’t have repeated interaction? This indirect reciprocity depends on recognizing attributes of the agents. Can we trust the agent? How will my reputation be affected if I defect?