EMC2: An R Package for cognitive models of choice
Niek Stevenson, Michelle C. Donzallaz, Reilly James Innes, Birte U. Forstmann, Dóra Matzke, Andrew Heathcote
Abstract
We introduce EMC2, an R package for the Bayesian hierarchical analysisof cognitive models of choice. EMC2 bridges the gap between standardregression analyses and cognitive modeling through linear-model specifica-tions for each type of cognitive-model parameter. The flexible implemen-tation of the linear modelling language allows users to map model parame-ters directly to complicated designs and hypotheses. EMC2 implementsrecent developments in Bayesian parameter estimation and hypothesistesting, including powerful and efficient sampling and marginal likelihoodestimation algorithms, so it is computationally feasible to estimate manydifferent cognitive models, and perform inference among them. Using twoleading evidence-accumulation models, we illustrate how EMC2 providesa workflow that makes it easy to specify diverse parameterisations and in-formative priors, and to evaluate, refine, compare, and interpret models.