Dora Matzke (University of Amsterdam)

Evidence-accumulation models ‒models that assume that rapid decisions are made by accumulating a threshold amount of evidence‒ have a long and successful history of accounting for response time and accuracy data from a wide range of domains, including lexical decisions, memory retrieval, and perceptual decision making. This workshop will provide an introduction to Bayesian estimation and model comparison in the context of evidence-accumulation models, such as the Diffusion Decision Model and the Linear Ballistic Accumulator Model. The Bayesian approach has important advantages, such as providing coherent ways to address model complexity, ensuring viable parameter estimation in sparse data environments, and enabling the assessment of group- and individual-level differences through hierarchical modeling. Using a series of short lectures and practical exercises, the workshop will cover individual and hierarchical parameter estimation using Differential Evolution Markov Chain Monte Carlo sampling as implemented in the Dynamic Models of Choice (DMC) package, posterior inference with plausible values, and model comparison using the Bayes factor.