Drift-Diffusion Model

To fit a drift-diffusion model, you prepare two-choice RT data with cutoffs and accuracy checks, select a DDM variant, estimate parameters via MLE or Bayesian methods, and validate using posterior predictive checks.

Drift-diffusion modeling decomposes observed two-choice RT and accuracy data into latent cognitive components like drift rate, boundary separation, and non-decision time.

Choose DDM variants based on your trial counts and fitting needs: EZ-diffusion offers simple analytical estimates for limited data, while HDDM enables hierarchical Bayesian parameter estimation across subjects.

Required data quality checks for drift-diffusion modeling include applying RT cutoffs, verifying accuracy ranges, and performing parameter recovery simulations to ensure robust model fitting.

Yes, Bayesian methods are supported for drift-diffusion model fitting, offering a structured workflow for convergence checks, model comparison, and posterior predictive validation.

Inaccurate parameter recovery in drift-diffusion modeling often results from insufficient per-condition trial counts, overly restrictive RT cutoffs, or failing to perform rigorous parameter recovery simulations before fitting.