UC Irvine

The hierarchical Bayesian approach to cognitive modeling often provides a quality of inference that cannot be matched with other analytical methods. In addition, the general approach is quite flexible, and can be utilized to great effect in many analytical settings. I illustrate these qualities in two applications of hierarchical Bayesian cognitive models. In the first application, I revisit a transfer-of-training study. First, I discuss a hierarchical cognitive model that describes the transfer-of-training data. I then illustrate how that hierarchical cognitive model can be further extended in order to create a cognitive latent variable model. Critically, this cognitive latent variable model directly models the latent effects of training and transfer on the cognitive parameters that drive participant behavior. I then provide an in depth analysis to illustrate how this cognitive latent variable model provides a quality of inference that far surpasses more standard analytical approaches. In the second application, I perform a cognitive meta-analysis on the spatial congruency bias literature. To do this, I extend the hierarchical Bayesian cognitive model into an integrative data analysis, creating a Model-based Integrative Data AnalysiS (MIDAS). Using this model, I create a model that is capable of simultaneously estimating cognitive effects at the individual, within-experiment, and between-experiment levels, which is what allows us to estimate cognitive effects in a meta-analytical setting.