Speaker：郭佳卉，德克萨斯大学达拉斯分校 行为与脑科学学院 助理教授

Time：10:00-11:30, July 1, 2024

Venue：Room 1113， Wangkezhen Building

Host：鲍平磊 研究员

Abstract

An accurate perception of the surrounding environment is critical for us to function effectively in the complex real world. For example, we need to recognize our families and friends, navigate ourselves in new places, and make social interactions with others. Studies on human high-level perception help us better interact with the world and provide insights into the functional principles of the human brain. In this talk, I will introduce my research on human high-level perception using naturalistic stimuli combined with computational methods, neuropsychological populations, and deep neural networks. First, I will demonstrate that functionally specialized brain areas for human high-level perception can be estimated with high fidelity from movie watching using hyperalignment, and can be generalized across different movie contents, languages, and other parameters. Second, I will discuss how we can estimate functional topographies in congenital blind and deaf patients. Finally, I will briefly discuss studies on using deep neural networks to understand human cognitive and neural representations under naturalistic situations.