On the role of beliefs in observational flavor conditioning