Recording
https://www.bilibili.com/video/BV1Cy4y1w7Wx/
Speaker
Taicheng Huang
Bio
Taicheng Huang (黄泰诚) is a postdoc at Shanghai Mental Health Center, affiliated with Shanghai Jiao Tong University School of Medicine, under the supervision of Prof. Tifei Yuan (袁逖飞教授). He graduated from Beijing Normal University, and received doctoral training in cognitive neuroscience. Later, he came to Tsinghua University for formal training in cognitive science under the supervision of Prof. Jia Liu (刘嘉教授). Currently, he hopes to dive into psychiatry to integrate psychology, computational science and neuroscience to understand mental illnesses, especially addiction.
Abstract
This talk I will start with a paper entitled the hippocampus as a predictive map, which was published in Nature Neuroscience several years ago, to discuss the relations between cognitive map, predictive coding and world model. After rapidly going through the paper, one of my previous works will be introduced, which tried to build a reinforcement learning model to understand how humans accomplish stability inference in daily life, this work emphasizes the irreplaceability of cognitive map, or namely the structural knowledge for flexible behavior, in our daily activity. Later, I will summarize a general philosophical idea behind both works, then link this idea to modern artificial intelligence about the possibility of developing a human-like AI, further focus on some potential gaps that need to be filled in order to achieve the goal. Hopefully in this talk we can have a great discussion about some critical ideas, and I highly welcome audiences from different backgrounds to provide updated pioneer opinions.