centaur_evaluation

Not Yet AlphaFold for the Mind: Evaluating Centaur as a Synthetic Participant

This repository accompanies the commentary paper
“Not Yet AlphaFold for the Mind: Evaluating Centaur as a Synthetic Participant”.

It contains the data and code used to evaluate the three cognitive tasks discussed in the paper:

• Reverse Learning
• Horizon Task
• Wisconsin Card Sorting Test (WCST)

---

Repository Structure

Each task has its own folder, which includes:

• The data used for the task (e.g., participant timelines) and data generated by task.
• The prompt used to run generative performance.
• The prompt used to run predictive performance.
• The notebooks for the evaluation of generative/predictive performance of all models used.
• The notebooks for recreating the final figures.

---

Contact

If you have any questions about the repository, feel free to contact:

Sabrina Namazova
📧 snamazova@uni-osnabrueck.de

Alessandra Brondetta

📧albrondetta@uni-osnabrueck.de