We put forth a general theory of boundedly rational behavior and learning for symmetric normal-form games with unique symmetric Nash equilibria. A class of evidence-based behavioral rules is specified, which includes best-responding to a prior and Nash play. A player begins with initial...