Reference

To create and test cognitive models using ACT-R, researchers follow these steps:

1. Understanding ACT-R: ACT-R is a cognitive architecture that provides a framework for simulating human cognition. It combines a theory of human cognition with a modeling framework, focusing on defining cognitive and perceptual operations.

2. Setting up the Environment: You need a software environment that supports ACT-R. There are packages available like pyactr on GitHub, which is a Python package for creating and running ACT-R cognitive models.

3. Building the Model: Develop the model by defining goals, creating production rules (if-then statements), and specifying declarative knowledge pieces. The model is built to mirror cognitive tasks or processes.

4. Testing the Model: Once the model is developed, it's tested against empirical data. This involves running simulations and comparing output with actual human data to refine the model.

5. Using fMRI Data: In some cases, researchers use fMRI data to evaluate and constrain models by predefined Region-of-Interest (ROI) analysis. This helps in linking cognitive processes modeled in ACT-R with specific brain regions (Source).

6. Refining the Model: Based on testing, refine production rules and knowledge representations to improve the model's accuracy.

For more detailed guidance, refer to the step-by-step tutorial on using ACT-R, which explains the process in depth.