The long-term objective of this research project is to fully characterize the retinal input to the mammalian brain. This information is essential if we hope to restore vision, either through synthetic or biological means, to those blinded through retinal diseases. Through decades of research we have learned much about how the mammalian retina encodes the visual world, but substantial gaps in our knowledge remain before realistic quantitative models of retinal image coding can be developed. The research proposed in this application seeks to fill two of these gaps. First, evidence is growing which threatens the foundation of our understanding of how the eye encodes visual information. Traditionally, retinal ganglion cells have been though to transmit neural messages by independently modulating their rates of discharge of action potentials. Simultaneous recordings from pairs or more of neurons in cats and other vertebrates have shown, however, that the spike trains of retinal ganglion cells are temporally correlated. The correlated firing events have the potential to encode more information than is possible were ganglion cells independent encoders of the visual scene. One major objective of this proposal is too determine whether a coding scheme based on correlated firing of cell groups is more plausible than one based on single cell firing for the mammalian retina. Second, a realistic model of how the retina represents visual images requires detailed information about the array of ganglion cell receptive fields and the spatiotemporal integrative properties of these fields. While we recently have very good quantitative descriptions of the spatiotemporal transfer functions of ganglion cells, we lack physiological maps of the array of ganglion cell receptive fields. Instead, we have anatomical maps of the array of ganglion cell dendritic fields. The other major objective of this proposal is to determine whether these anatomical maps can correctly substitute for the physiological maps which are more difficult to measure.
Showing the most recent 10 out of 22 publications
