The posterior marginal is the distribution of the parameter(s) of interest given the data. e.g. P(Xi∣E=e); P(Listening = true∣Status = excited)