EMC2: An R Package for cognitive models of choice
Andrew Heathcote (Amsterdam Mathematical Psychology Laboratory)
Psychological Science Senior Common Room (2D17, Priory Road Complex)
Hosted by the School of Psychological Science
Abstract: I introduce EMC2, an R package for Bayesian hierarchical analysis of cognitive models of choice. EMC2 bridges the gap between standard regression analyses and cognitive modelling through linear-model specifications for each type of cognitive-model parameter. The flexible implementation of the linear modelling language allows users to map model parameters directly to complicated designs and hypotheses. EMC2 implements recent developments in Bayesian parameter estimation and hypothesis testing, including powerful and efficient sampling and marginal likelihood estimation algorithms, so it is computationally feasible to estimate many
different cognitive models, and perform inference among them. Using two leading evidence-accumulation models, we illustrate how EMC2 provides a workflow that makes it easy to specify diverse parameterisations and informative priors, and to evaluate, refine, compare, and interpret models. A preprint about the base package can be found at https://doi.org/10.31234/osf.io/2e4dq, in this talk I address more advanced features implemented in the development version of the package and discuss plans for future expansions.