2025년 07월 29일 (화) 오후 4시 30분

연사: 성균관대 글로벌바이오메디컬공학과 이준열 교수님

제목: Neural Mechanisms of Motion Prediction in Sensory-Motor Behavior

초록: When we interact with our environment, sensory input is often incomplete or degraded—especially in dynamic visual situations caused by movements in the environment or by our own actions. To compensate for this, the brain frequently predicts future events by drawing on prior knowledge and experience.

In this talk, I will present a series of studies examining how predictive information about future events interacts with sensory input in the brain. I will begin with a human electroencephalogram (EEG) study that investigates large-scale information transmission across the brain. Using multivariate analysis of EEG recordings, I will demonstrate how predictions about incoming visual motion are combined with actual visual input to guide sensory-motor behavior. Next, I will discuss circuit-level mechanisms in the macaque visual area MT (middle temporal area), using single-cell recordings and computational approaches to explore how predictive and sensory signals are integrated at the level of population neural activity, and how this integration influences sensory-motor behavior. Finally, I will introduce a recurrent neural network (RNN) model that simulates this integration process at the cellular level, offering insights into potential mechanisms underlying the interaction between prediction and sensory input.