Examining Generalization and Flexibility in Structure Learning with EEG
With the incredible amount of information available in the world, humans have to form many different behavioral strategies in order to account for the variety of situations and information we could encounter. This makes the ability to flexibly adapt behavior to different contexts a critical component of human intelligence. For example, when we use computers, we know that Macs and PCs use different operating systems. We can apply what we now from using a Mac laptop to a Mac desktop, but we know we cant apply that same knowledge to a PC. In this way, learning has to be both generalizable, so that it isnt necessary to constantly relearn information, and also flexible, to account for a variety of different situations. In my project, I will be exploring how humans learn behavioral strategies and choose which strategies to apply to novel contexts, using a combination of computational modeling and neuroimaging data from EEG. Specifically, my project will expand upon past behavioral data to test a neurally-inspired model that utilizes hierarchically structured reinforcement learning to best approximate optimal inference.
Message to Sponsor
- Major: Cognitive Science
- Sponsor: Zara Fund
- Mentor: Anne Collins