How to use VBA

Some statistical theory

• Bayesian inference: introduction
• The variational Bayes approach
• VBA: structure of the generative model

Getting Started with VBA

• VBA: model inversion in 4 steps
• VBA: a fast demo (Q-learning)
• VBA: graphical output

What VBA users cannot ignore

• VBA: results and diagnostics
• VBA: controlling the inversion options
• 10 simple rules for VBA
• Identifiability and confusion analyses

Advanced VBA tricks

• Experimental design optimization
• Multisession inversions
• Mixed observations (including multinomial and gaussian variables)
• Bayesian model selection for group studies
• Bayesian model selection with large model spaces
• Volterra decompositions
• Bayesian Model Averaging (BMA)
• Setting the priors through Empirical Bayes
• Setting “hard constraints” through parameter transformations
• A few useful stand-alone VBA functions
• Extending VBA’s generative model (for power-users only)

Library of models

The VBA-toolbox already includes a large library of plug-and-play models. Below, we give a few examples of such models. We briefly expose the main theoretical and experimental issues, and point to the relevant VBA functions (demonstration scripts, evolution/observation functions, etc…). Note that this list is by no means exhaustive: users are invited to look for demonstration scripts in VBA’s \subfunctions folder.

Behavioural/cognitive models

• Reinforcement learning
• Bayesian associative learning (including, e.g., volatile systems)
• Bayesian sequence learning (BSL)
• Bayesian mentalizing (k-ToM)
• Bi-dimensional decisions (e.g., inter-temporal choices)

Neurobiological models

• Dynamic Causal Modeling
• Behavioural Dynamic Causal Modeling
• Neural Fields
• Spiking Neurons
• Calcium imaging using biophysical models

Miscellaneous models

• Statistical models (GLM, RFT, classification, mediation)
• Attractors and oscillators