Bayesian Cognitive Model Builder

To specify priors for hierarchical Bayesian cognitive models, select priors that respect cognitive constraints and apply regularization across participants. This includes choosing between centered and non-centered parameterizations for your Stan or PyMC model structure.

The best way to run MCMC diagnostics for cognitive modeling is to use a robust diagnostic checklist that evaluates convergence diagnostics, trace and rank plots, and posterior predictive checks to identify MCMC pathologies in your hierarchical models.

You perform posterior predictive checks for hierarchical Bayesian models by evaluating model adequacy through domain-relevant workflow guidance. This involves generating explicit diagnostic checklists and comparing posterior predictions against observed cognitive data.

Yes, this hierarchical Bayesian modeling guidance applies to common cognitive-model families including DDM, SDT, RL, MPT, and IRT. It provides domain-validated priors specification and model-structure decisions across these frameworks using Stan and PyMC.

Use centered or non-centered parameterization in hierarchical cognitive models based on model-structure decisions and the need for regularization across participants. This guidance helps determine the appropriate approach to avoid MCMC pathologies and ensure convergence.

To compare hierarchical Bayesian cognitive models using PSIS-LOO or WAIC, apply model comparison and reporting workflows that include information criteria, Bayes factors, and sensitivity analyses. This ensures reproducible evaluation of your cognitive model adequacy.