Seminar Series

Dr Kazuhisa Shibata

In this talk, I will review psychophysical and neuroimaging studies that addressed an essential question in learning: how do we know what to learn? This question is important because the brain must learn new information processing to adapt to a current environment while protect existing information processing from being replaced with unimportant ones. The brain must have the mechanisms to resolve this so-called plasticity-stability dilemma. As an example, I will focus on visual perceptual learning (VPL), defined as the enhancement of sensitivity to a visual feature as a result of experience on the feature. First, I will describe the complementary roles of attention and reward in VPL of a visual feature depending on whether the feature is relevant to a given task. Second, I will discuss the roles of the passage of time and the amount of training in VPL of a visual feature when another feature is trained after training on the feature. The results of the studies suggest that what to learn in VPL is determined by multiple brain mechanisms, including attention, reinforcement signals evoked by reward, and stabilization based on an excitatory to inhibitory balance in the visual cortex. These findings may help us clarify how the brain resolves the plasticity-stability dilemma in other learning types, such as memory and motor learning.