Integrating top-down signals in face and object processing

点击翻译按钮获取中文摘要

When we talk about "seeing the world", we implicitly refer to the people and things associated with the places where we expect to find them. We may take these abilities for granted, but engineers have yet to build a synthetic visual system that performs as accurately, with the level of speed and efficiency as our visual system. My research is devoted to understanding how the brain codes faces and objects. One unresolved question is how the activity of brain cells and their interactions lead to our visual aptitudes. One secret to our visual system's success may be our use of two self-generated signals to enhance face and object processing: (i) our eye movements that actively scan the environment, and (ii) memory from previous experiences. We will record the activity of groups of brain cells as faces and objects are detected or discriminated. We will measure the changes in the rate of action potentials and how the changes are distributed across cells, as well as whether action potentials 'march to the beat' of local neural rhythms, or act independently of them. Furthermore, (i) we will determine how this code is influenced by the dynamics of active search, which places new visual information on the retina ~3-5 times per second. "On your mark, get set"...reacting to "Go!" is much faster when you're the one calling the shot. Similarly, could the brain use the eye movement it generated to help prepare the visual system for incoming signals from the retina? (ii) Another way to prepare or shape brain activity is through memory. We find objects in clutter better when it's our familiar clutter than when we stumble upon someone else's, and we have an easier time recognizing a familiar friend than a new acquaintance. Experience can have a dramatic impact on our perceptual abilities. We will study the influence of eye movements and of past experience on neural activity in cells that respond to faces or objects. Understanding how faces and objects are coded in the brain, and if it may be shaped for better performance, is a fundamental question that could change our views of brain processing in general, and inform artificial systems that seek to emulate our remarkable ability to "see the world".