This dissertation focuses on the choice of priors in Bayesian model selection and their applied, theoretical and computational aspects. As George Box famously said “all models are wrong, but some are useful"; many statisticians ...