Enhancing Methodological Rigor for Computational Cognitive Science: Core Tenets and Ad Hoc Residuals

Abstract

Computational models are notoriously difficult to compare and interpret, resulting in a community segmented around modeling paradigms. In this paper, we seek to develop community standards and methodology that will make it easier to compare work across computational paradigms, discern what types of empirical predictions can be drawn from computational work, and test the validity of computational models. Using an established Bayesian model, we illustrate how our proposed methods will achieve these goals.


Back to Thursday Posters