BCI Toolbox¶
Bayesian Causal Inference for multisensory research
BCI Toolbox is a Python package and graphical workflow for fitting, simulating, visualizing, and exporting Bayesian causal inference models for behavioral data. It is designed for researchers who want a reproducible model pipeline without writing custom analysis code for every experiment.
What You Can Do¶
Fit behavioral datasets with discrete or continuous responses.
Compare model-averaging, model-selection, and probability-matching decision strategies.
Build models with any number of modalities and one or more stimulus dimensions.
Fit individual participants or grouped datasets and export reproducible results.
Run parameter recovery, diagnostics, posterior-predictive checks, and model comparison.
Use the graphical interface for import, fitting, plotting, and export.
Use the v0.3.0 Python API for reproducible scripts and advanced workflows.
Recommended Reading Path¶
To fit your own behavioral data from Python, start with the Quickstart — five worked examples. For the graphical workflow, open Installation and then Graphical User Interface. If your experiment contains two task dimensions, such as numerosity and time, continue with 2D GUI Tutorial.
Citation¶
If BCI Toolbox supports your work, please cite:
Zhu, H., Beierholm, U., & Shams, L. (2024). BCI Toolbox: An open-source python package for the Bayesian causal inference model. PLOS Computational Biology, 20(7), e1011791. https://doi.org/10.1371/journal.pcbi.1011791
For the 2D BCI module, please also cite:
Zhu, H., Zhang, Y., Beierholm, U., & Shams, L. (2026). Crossmodal interaction of flashes and beeps across time and number follows Bayesian causal inference. Psychonomic Bulletin & Review, 33, 58. https://doi.org/10.3758/s13423-026-02857-z
Contributors¶
Haocheng Zhu, Dr. Ulrik R. Beierholm, and Dr. Ladan Shams.
Questions and feedback are welcome at evanszhu2001@gmail.com.