In recent years, computational models have given insight into processes underlying cultural transmission, but efforts have mostly focused on small-scale cognitive problems. In this project, we will be extending a video game playing agent that learns to play games in human-like ways in order to model knowledge transmission in complex game worlds. Just as a person might play a new video game for an hour and then quickly explain the most important mechanics to a friend, our modified agent will be able to learn to play games during a quick gameplay session and then decide which information/strategies to pass to naive copies of itself to help them learn the game. We hope our framework will lead to a new direction for a more quantitative study of knowledge transmission in complex environments.

I have been in the UROP program for two years, and I think that it has been the key to my success at MIT. I have loved applying my knowledge from machine learning and cognitive science classes in my research and look forward to a longer‐term research project and hopefully publishing a paper!