|Institution:||University of Washington|
|Keywords:||computational model; object form; object recognition; primate; vision; Neurosciences; Behavioral neuroscience|
|Full text PDF:||http://hdl.handle.net/1773/40838|
Primates use vision to understand and interact with the world around them. Several interconnected visual areas are responsible for processing information about visual objects, ultimately enabling object and scene recognition and perception. Visual area V4 is an intermediate stage in the pathway of object processing, and V4 neurons are sensitive to multiple aspects of objects such as their form, texture, and color. The first goal of this dissertation was to investigate how responses of neurons in visual area V4 reflect information about object boundaries and interiors. Unlike prominent computational models of object recognition which rely on boundaries, the majority of neurons in area V4 displayed a modulation of responses by the presence or absence of an object interior. I developed two computational model modifications that successfully incorporated both of the response characteristics that I observed experimentally: sensitivity to boundaries and sensitivity to the presence of an object interior. The second goal of this dissertation was to examine whether V4 response selectivity for shape and color of visual objects depended on the task which the animal was performing. I compared responses during a task where the animal was discriminating object shape to a task where the animal was passively viewing the same objects. I found that the majority of neurons displayed not only a change in the magnitude of responses between the two tasks, but that response changes were additionally dependent on object shape and color. Together, these studies contribute to our understanding of how neurons encode visual information, and how this encoding is changed during behavior.Advisors/Committee Members: Pasupathy, Anitha (advisor).