The COIN (COntextual INference) model [1] is a Bayesian nonparametric model of motor learning in which separate memories are acquired for different contexts. The key insight of the model is that memory creation, updating and expression are all controlled by a single computation—contextual inference. Adaptation can arise both by creating and updating memories (proper learning) and by changing how existing memories are differentially expressed (apparent learning).
The model was developed in MATLAB and has been tested on MATLAB R2020a. For Python users, please see the Python version of COIN created by Changmin Yu.
- Heald, J.B., Lengyel, M. & Wolpert, D.M. Contextual inference underlies the learning of sensorimotor repertoires. Nature 600, 489–493 (2021). [SharedIt link]
- Collins, A.G.E. & McDougle, S.D. Context is key for learning motor skills. Nature (2021).
- Download the COIN.m file.
- Install the following packages:
- "Nonparametric Bayesian Mixture Models - release 2.1" by Yee Whye Teh.
- "Lightspeed matlab toolbox" by Tom Minka.
- "Truncated Normal Generator" by Zdravko Botev.
Instructions for how to use the COIN model can be found on the COIN wiki.
Please e-mail me at jamesbheald@gmail.com if you have any questions or comments.
The COIN model is released under the terms of the GNU General Public License v3.0.