Paper:

Ehrlich, D. B.^*, Stone, J. T.^*, Brandfonbrener, D., Atanasov, A., & Murray, J. D. (2021). PsychRNN: An Accessible and Flexible Python Package for Training Recurrent Neural Network Models on Cognitive Tasks. ENeuro, 8(1). [DOI]

Presentation:

Prefer listening to a 15 minute talk to see if PsychRNN is for you? Check out our talk at Neuromatch 3.0.

Full documentation is available at psychrnn.readthedocs.io.

This package is intended to help cognitive scientists easily translate task designs from human or primate behavioral experiments into a form capable of being used as training data for a recurrent neural network.

We have isolated the front-end task design, in which users can intuitively describe the conditional logic of their task from the backend where gradient descent based optimization occurs. This is intended to facilitate researchers who might otherwise not have an easy implementation available to design and test hypothesis regarding the behavior of recurrent neural networks in different task environements.

Release announcments are posted on the psychrnn mailing list and on GitHub

Code is written and upkept by: Daniel B. Ehrlich, Jasmine T. Stone, David Brandfonbrener, and Alex Atanasov.

Contact: psychrnn@gmail.com

Start with Hello World to get a quick sense of what PsychRNN does. Then go through the Simple Example to get a feel for how to customize PsychRNN. The rest of Getting Started will help guide you through using available features, defining your own task, and even defining your own model.

• python = 2.7 or python >= 3.4

• numpy

• tensorflow >= 1.13.1

• For notebook demos, jupyter

• For notebook demos, ipython

• For plotting features, matplotlib

PsychRNN was developed to work with both Python 2.7 and 3.4+ using TensorFlow 1.13.1+. It is currently being tested on Python 2.7 and 3.4-3.8 with TensorFlow 1.13.1-2.2.

Note: TensorFlow 2.2 does not support Python < 3.5. Only TensorFlow 1.13.1-1.14 are compatible with Python 3.4. Python 3.8 is only supported by TensorFlow 2.2.

Normally, you can install with:

pip install psychrnn

Alternatively, you can download and extract the source files from the GitHub release. Within the downloaded PsychRNN folder, run:

    python setup.py install

[THIS OPTION IS NOT RECOMMENDED FOR MOST USERS] To get the most recent (not necessarily stable) version from the github repo, clone the repository and install:

    git clone https://github.com/murraylab/PsychRNN.git
    cd PsychRNN
    python setup.py install

Please report bugs to https://github.com/murraylab/psychrnn/issues. This includes any problems with the documentation. Fixes (in the form of pull requests) for bugs are greatly appreciated.

Feature requests are welcome but may or may not be accepted due to limited resources. If you implement the feature yourself we are open to accepting it in PsychRNN. If you implement a new feature in PsychRNN, please do the following before submitting a pull request on GitHub:

• Make sure your code is clean and well commented
• If appropriate, update the official documentation in the docs/ directory
• Write unit tests and optionally integration tests for your new feature in the tests/ folder.
• Ensure all existing tests pass (pytest returns without error)

For all other questions or comments, contact psychrnn@gmail.com.

All code is available under the MIT license. See LICENSE for more information.