Seol mar théacs é seo: The possibility of Bayesian learning in repeated games