Awarded in 2022
Home Department: Neurosciences
Faculty Advisors: Stephen Baccus (Neurobiology) and Nick Haber (Education and Computer Science)

Research Title: Rapid Perceptual Learning in Rewarded Tasks – The Efficient Learning Hypothesis

Research Description: Historically, reinforcement learning (RL)—a machine learning approach for task-based problem solving, and visual neuroscience experiments have relied heavily on artificial stimuli to study decision-making. Because current models of RL fail to capture the rapid learning of animals in natural contexts, it remains unclear how biological and RL agents can learn in naturalistic environments. Javier will bridge neurophysiology, naturalistic virtual reality, and deep reinforcement learning to investigate decision-making and cortical plasticity of rodents in natural environments. He will test the hypothesis that efficient visual coding in the primary visual cortex combined with optimal learning strategies support rapid task-based visual decision-making. Success of this work will aid in investigating behavior and neural plasticity by combining biological experimentation with computational modeling methods and interactive deep learning techniques.