You will learn how to implement and tune a model of perceptual decision making. The ‘high-level’ objective will be to understand how to set up neural simulations, but also to get an appreciation of their limitations, as well as their dependence on the parameters/assumptions.
In week 1, you will get to know the model with which we will working in this project.
Tasks:
- Read through background materials 1 - 2.
- Work on exercises 1 and 2.
- create your fork of the repository by accepting the github classroom assignment
- work through the notebook, add code and plots in code cells, add text answer in markdown cells.
- commit regularly, your last commit before the deadline counts as submission.
This week you will analyze the model. Link to github classroom assignment.
Tasks:
- Recap week 1, read through the solutions and make sure you understand them. Disuss with your colleages or open an issue in case you have questions or spot a mistake in the solutions.
- Work on exercises 3 and 4. Discuss with your colleages or open an issue if you have questions or if you are stuck.
This week you will learn about Bayesian parameter inference via simulation-based inference, using the example of the drift-diffusion model. Link to github classroom assignment.
Tasks:
- Read through the instructions in the notebook carefully and install the required packages in your conda environment.
- Work on exercises 1-3 (optional 4) and discuss with your colleages or open an issue if you have questions.
- Chapter 16 of Neuronal Dynamics et al.: https://neuronaldynamics.epfl.ch/online/Ch16.html
- Wong, K.-F. & Wang, X.-J. A Recurrent Network Mechanism of Time Integration in Perceptual Decisions. J. Neurosci. 26, 1314–1328 (2006)
- Goncalves, Lueckmann, Deistler et al. 2019. Training deep neural density estimators to identify mechanistic models of neural dynamics
We summarised all the logistics you need to take care of in logistics.md.