Abstract:
We propose a theory that explains why rational agents start to believe in a causal relationship between unrelated events. Agents send and collect messages through a communication network. If they are convinced of a relationship between two events, they send messages confirming their belief with higher probability than messages contradicting it. The network aggregates this communication bias over individuals. Therefore, agents may find a strong relationship between unrelated events even if the communication bias is very small. We apply this model to an informal economy where the fear of punishment by supernatural forces prevents agents from cheating others.