On the adequacy of Bayesian evaluations of categorization models: Reply to Vanpaemel and Lee (2012).

Abstract

Vanpaemel and Lee (2012) argue, and we agree, that the comparison of formal models can be facilitated by Bayesian methods. However, Bayesian methods neither precede nor supplant our proposals (Wills & Pothos, 2012), as Bayesian methods can be applied both to our proposals and to their polar opposites. Furthermore, the use of Bayesian methods to control for model complexity can be actively misleading when combined with the consideration of narrow data sets, and significant development work is required before Bayesian methods can be applied to some of the leading formal models of categorization. Even where Bayesian methods can be applied, the use of non-Bayesian methods is sometimes preferable due to their computational simplicity (Vanpaemel & Storms, 2010). We also clarify our position on arbitrarily variable parameters, and on the relationship between ordinal properties and overfitting.