This repository contains an implementation of the RECOLLECT architecture as described in:
A.R. van den Berg, P. R. Roelfsema & S. M. Bohte (2023). Biologically plausible gated recurrent neural networks for working memory and learning-to-learn. Biorxiv.
- Python (tested on 3.8.0)
- Numpy (tested on 1.23.1)
- Pandas (tested on 1.2.3)
- Matplotlib (tested on 3.3.4, but not required to run the code)
An example of how to use RECOLLECT has been provided for the pro-/anti-saccade task from Gottlieb and Goldberg (1999).
This software is distributed under the MIT license (see license file).