Inclusion Bayes factors for mixed hierarchical diffusion decision models

Area of Science:

• Cognitive Science
• Computational Neuroscience
• Psychological Modeling

Background:

• Cognitive models offer quantitative insights into latent cognitive processes, aiding theory building and empirical testing.
• Nonlinearity and parameter correlations in cognitive models present significant data analysis challenges.
• Accurate estimation requires large hierarchical datasets and robust statistical inference accounting for model uncertainty.

Purpose of the Study:

• To present Bayesian hierarchical modeling and Bayesian model averaging as solutions for analyzing complex cognitive models.
• To demonstrate the application of these Bayesian techniques to overcome estimation challenges.
• To illustrate the utility of these methods using the diffusion decision model.

Main Methods:

• Bayesian hierarchical modeling to structure and analyze large, complex datasets.
• Bayesian model averaging to account for model uncertainty and avoid biased parameter estimates.
• Application of diffusion decision model to empirical data from a collaborative selective influence study.

Main Results:

• Successfully applied Bayesian methods to address nonlinearity and parameter correlation challenges in cognitive modeling.
• Demonstrated robust estimation of the diffusion decision model parameters using the proposed techniques.
• Provided a framework for more reliable quantitative descriptions of cognitive processes.

Conclusions:

• Bayesian hierarchical modeling and model averaging effectively handle challenges in cognitive model data analysis.
• These methods enhance the reliability of parameter estimates and reduce overconfidence in model-based inferences.
• The approach facilitates cumulative theory building in cognitive science through rigorous quantitative modeling.