bmm

bmm implements Bayesian hierarchical mixture models for visual working memory tasks, enabling group and condition comparisons with lower trial numbers through an efficient linear model syntax.

Key Features:

• Hierarchical Bayesian estimation: Provides robust and reliable parameter estimates with lower trial numbers compared to maximum likelihood estimation for single-subject data.
• Linear model syntax: Integrates hierarchical Bayesian estimation with an efficient linear model syntax that enables adaptation of mixture models to practically any experimental design.
• Flexible parameter variation: Allows model parameters to vary over arbitrary groups or experimental conditions.
• Informed priors: Supports specification of informed priors and hierarchical structures that improve subject-level parameter estimation and solve estimation problems.

Scientific Applications:

• Visual working memory research: Estimates mixture models for continuous report recall tasks, a highly popular measurement model in visual working memory research.
• Group comparisons: Enables flexible comparison of model parameters across different participant groups.
• Condition comparisons: Allows comparison of model parameters across experimental conditions within a study.

Methodology:

Bayesian hierarchical mixture models are specified and fit using an R package implementation with hierarchical Bayesian estimation procedures and linear model syntax for parameter variation over groups or conditions.