code for reproducing analyses in our paper "Contextual Flexibility Guides Communication in a Cooperative Language Game"
Our analyses use a combination of Jupyter notebooks (for deriving model predictions) and R notebooks (for visualization and statistical analysis).
There are two ways of running our python code:
-
Colab: The easiest way to reproduce our model results is by using this self-contained colab notebook. The notebook will prompt you for Google account authentication but should otherwise run in the browser the same as it would on your local machine.
-
Locally: To run the jupyter notebooks on your local machine, first download the embeddings, unpack the zip file into the
./data/directory, and then runjupyter lab.
To run the R Notebook, follow these instructions to download and open the RStudio IDE.
For more information about the data used in this project, see this OSF repository.
Kumar, A., Garg, K., Hawkins, R.D. (2021). Contextual flexibility guides communication in a cooperative language game. Proceedings of the 43rd Annual Conference of the Cognitive Science Society.