All model code and instructions can be found here: https://github.com/anayebi/mental-sim.

This repo contains pretrained PyTorch models that are optimized to predict the future state of their environment. As these models are stimulus-computable, they can be ``dropped in'' on any new video for further neural and behavioral comparisons of your design.

This repository is based on our paper:

Aran Nayebi, Rishi Rajalingham, Mehrdad Jazayeri, Guangyu Robert Yang

"Neural foundations of mental simulation: future prediction of latent representations on dynamic scenes"

37th Conference on Neural Information Processing Systems (NeurIPS 2023). Selected for spotlight.

Here's a video recording that explains our work a bit.