Good practices for Bayesian inference in statistical learning

Bayesian inference provides a unifying formalism for describing the inferences humans and other animals perform about features of the environment that change on fast timescales (moment-by-moment or trial-by-trial), as well as their learning of environmental contingencies and regularities that vary over slower time scales (i.e., statistical learning). We briefly highlight example Bayesian models of statistical learning at increasing levels of abstraction, such as the learning of parameters, causal structures, and programs. We then focus on the technical aspects of using such models and provide an overview of good practices for the key steps of fitting them to behavioral data: the validation of inference through sample-based calibration, the validation of fitting through parameter recovery, and the validation of model comparison through model recovery. While these steps undeniably require additional effort and compute, we argue that their use is essential for exploiting the full potential of Bayesian modeling.