Trial and Error: A Bayesian Approach #69
Description
Title
Trial and Error: A Bayesian Approach
Leaders
Adam J. Tiesman, Ansley J. Kunnath
- Github: @tiesmaaj, @ansleykunnath
Collaborators
Your name will be here if you join our project!!!
Project description
- What are you doing, for whom, and why?
We’re building a pipeline to analyze how trial history shapes decision-making in psychophysical tasks, with a focus on 2-alternative forced choice (2-AFC) discrimination tasks. This pipeline will allow anyone who runs psychophysical tasks (even you and your data!) to uncover patterns in how past stimuli, outcomes, and responses influence current behavior. By focusing on trial history effects and Bayesian inference, our project aims to provide actionable insights into learning, adaptation, and decision biases in experimental data.
- What makes your project special and exciting?
This project focuses on analyzing how past trials influence current decisions using Bayesian modeling and clear visualizations of behavioral patterns. It provides a straightforward, flexible, and reproducible workflow for psychophysical data analysis, helping researchers uncover biases and understand participant adaptation. Designed for both scientific rigor and usability, this pipeline supports tackling complex questions in behavioral science. Additionally, it’s a chance to collaborate, deepen skills in Bayesian inference, and produce meaningful results in just one weekend.
Link to project repository/sources
https://github.com/tiesmaaj/trial_and_error
Concerete goals with specific tasks for Brainhack Vanderbilt 2025
We want to build data pipelines for 3 separate analyses:
- Individual Session - Create a trial history pipeline to process and analyze data from single experimental sessions.
- Comparing Conditions Within Subjects - Design a trial history pipeline to compare performance and decision-making patterns across different experimental conditions for the same participant.
- Group Level - Construct a pipeline to aggregate and analyze data across multiple participants, identifying consistent patterns and group-level trends in trial history effects.
Good first issues
- Data Preparation - write scripts to validate input data, test preprocessing scripts on example datasets
- Visualization - implement simple and complex plots such as histograms, scatterplots, and other reusable visualization templates
- Documentation and Building - write clear and concise instructions and explanations, build README file, propose future features and extensions
- Testing and Debugging - test the pipeline on multiple datasets and identify possible bugs to ensure robustness
- Bayesian Modeling - research Bayesian modeling, implement simplified version, define priors and likelihoods for specific types of tasks
Skills
All levels of coding (MATLAB and Git) are welcome! We will have tasks designed to be challenging, but not overwhelming, for all skill levels!
Onboarding documentation
Linked here is a Google drive folder that has papers and resources that lay the foundation for trial history and Bayesian modeling for behavioral data. These are great resources for understanding the basis of our research question.
What will participants learn?
First-time Brainhack participants and veteran coders will gain the following skills:
- Data Cleaning and Preprocessing - Handle raw experimental data, compute derived variables, and prepare datasets for analysis.
- Data Visualization - Create impactful plots to communicate key findings using MATLAB.
- Pipeline Development - Build modular, reusable, and scalable workflows for individual, condition-level, and group-level analyses.
- Bayesian Modeling - Implement priors, likelihoods, and posterior updates for psychophysical tasks.
- Trial History Analysis - Quantify how past trials influence decision-making.
- Collaborative Coding - Work in teams, use version control, and manage projects effectively.
- MATLAB Proficiency - Gain hands-on experience with MATLAB for analysis and modeling.
- Problem-Solving - Learn to collaborate under time constraints in a hackathon setting.
Public data to use
Example data we will use will be included on project's Github repo linked here
Number of collaborators
4+
Credit to collaborators
Project contributors are listed on the project README.
Image
Project Summary
Uncover how prior information and past decisions shape behavior! Our pipeline will analyze trial history effects in psychophysical tasks with data visualizations and Bayesian models.
Type
documentation, pipeline_development, visualization
Development status
1_basic structure
Topic
bayesian_approaches, data_visualisation, reproducible_scientific_methods, statistical_modelling
Tools
other
Programming language
Matlab
Modalities
behavioral
Git skills
1_commit_push, 2_branches_PRs
Anything else?
No response
Things to do after the project is submitted and ready to review.
- Add a comment below the main post of your issue saying:
Hi @brainhack-vandy/project-monitors my project is ready!