An evolutionary explanation for why the prior beliefs can be assumed to coincide with the correct ones in the simple Bayesian hypothesis testings

This paper considers an evolutional process, where a simple Bayesian hypothesis testing is conducted repeatedly in parallel by many testing agents with diversified prior beliefs. Assuming that each of the agents selects a (one-shot) testing strategy to maximize the fitness value believing that the (own) prior belief is true, it is shown that only the testing agents endowed with the correct prior beliefs are survived eventually. This result provides an explanation for why the prior belief of the agent can be assumed to coincide with the correct one in the Bayesian hypothesis testing, as if the agent knows the true probability assigned by the nature, without introducing the longterm learning processes.