Bayesian outcome-based strategy classification.

Area of Science:

• Cognitive Psychology
• Decision Science
• Computational Modeling

Background:

• Existing models like Weighted Additive (WADD) and Take-The-Best (TTB) are foundational in multi-attribute decision research.
• Hilbig and Moshagen (2014) proposed a Minimum Description Length (MDL) approach for model inference.
• There is a need for complementary methods to enhance understanding of human decision processes.

Purpose of the Study:

• To present a novel Bayesian approach for inferring decision-making models in multi-attribute forced choice tasks.
• To offer theoretical and methodological advantages over existing inference methods.
• To demonstrate the utility of the Bayesian approach using existing data and models.

Main Methods:

• Development of a simple graphical model implemented in JAGS (Just Another Gibbs Sampler).
• Application of a fully Bayesian inference framework.
• Utilizing prior predictive analysis and posterior inferences for model comparison and parameter estimation.

Main Results:

• The Bayesian approach provides a robust framework for inferring decision strategies.
• Demonstrated the application of the method to data from Hilbig and Moshagen (2014).
• Identified which decision models people use and their parameter settings.

Conclusions:

• The proposed Bayesian method offers a valuable alternative for analyzing human decision-making.
• This approach enhances the understanding of cognitive processes in complex choices.
• The framework facilitates detailed insights into model selection and parameter usage in decision tasks.