A New Approximation Technique for Maxwell's Equations

Texas A&M Research Foundation, College Station TX

Investigators

Abstract

Bressloff 0209824 The long term goal of this project is to develop a dynamical theory of how neurons in primary visual cortex (V1) generate a tuned response to multiple (rather than single) features of a visual stimulus, and how these responses are spatially integrated across the cortex to generate more global information about a visual scene. A primary focus of the work is to extend current network models of orientation tuning to incorporate the fact that V1 cells are also selective for spatial frequency. This is motivated by the considerable physiological and psychophysical evidence suggesting that cortical circuits carry out a localized two-dimensional Fourier decomposition of a stimulus rather than simply performing local edge detection. Optical imaging of the surface of cortex has revealed an intricate relationship between the distribution of orientation and spatial frequency preferences across cortex. How correlations between these two feature preference maps is manifested by the local and long-range circuitry of V1, and the consequences for the large-scale dynamics of V1 is also investigated. The primary visual cortex (V1) located at the back of the brain is the first cortical area to process visual information received from the eyes. One of the classical results regarding the function of neurons (brain cells) in V1 is that they analyze very local features of a visual image, that is, they carry out image decomposition. (For example, V1 cells are sensitive to the orientation of an edge representing the boundary between a light and dark region of the image. This discovery by Hubel and Wiesel led to the Nobel prize in medicine). A very important question that follows from this is how our coherent perception of the world is reconstructed. Until recently, it was thought that the local information from cells in V1 was passed through higher order processing stages in the brain where cognition occurs. However, it is becoming clear that long-range circuitry within V1 could itself contribute to the process of reconstruction. The basic aim of the proposal is to investigate this process by developing a large-scale mathematical model of primary visual cortex that incorporates the latest anatomical data regarding its internal circuitry. Understanding how early stages in the visual brain encode images has important applications to information technology (such as the development of artificial vision systems) and biotechnology (such as the development of an artificial prosthesis for the visually impaired). In the latter case it might be possible one day to artificially stimulate primary visual cortex to induce a visual sensation, rather like a controlled visual hallucination.