Chemical synapses transmit information from one neuron to another throughout the brain. This is accomplished by probabilistic release of transmitter quanta that add noise to the transmitted information. If synapses in a sequential pathway add noise, then noise from one synapse might be transmitted across the next, and synaptic noise will be promulgated throughout the nervous system. Consequently, the brain should be designed to limit the amount of trans-synaptic noise. In the retina, we are beginning to understand how neural noise influences circuit design. Noise causes a measurable reduction in the detectability and discriminability of visual stimuli. To reduce the impact of noise, retinal circuitry is characterized by a pattern of divergence and convergence that sends signal through multiple neurons and synapses. Gap junctions between horizontal cells improve signal-to-nose ratio. Thresholding at the rod ? rod bipolar synapse rejects noise generated in rods. Despite the importance of noise to the function and design of the mammalian retina and brain, the amount of noise that a synapse adds to transmitted information has not been measured;few measurements have been made of the amount of noise passed from one synapse through the next. Synapses are rectifying due to the exponential relationship between membrane voltage and the calcium influx that drives transmitter release. Rectification at the bipolar ? ganglion cell synapse partitions positive and negative contrasts between On and Off pathways, but how this changes with illumination is not known. On and Off signals are recombined in the Off alpha ganglion cell where they may implement a synergistic push-pull circuit. This circuit could improve the overall coding of information, but this has not been directly tested by estimating information rate. Here we propose to investigate the how synapses contribute to retinal information processing by recording EPSCs from a defined subset of retinal ganglion cell types. We will estimate the amount of noise that bipolar synapses on the ganglion cell contribute to these currents. Our preliminary evidence suggests that once this synaptic noise is removed, the remaining noise in the EPSC is from the presynaptic circuit, possibly from quantal release from cone terminals onto bipolar cells. Preliminary evidence suggests that amacrine circuitry, and the partitioning of information into On and Off-pathway are retinal designs that reduce noise. To test these ideas we propose the following specific aims. The proposed studies, by measuring information flow through synapses, will develop explicit rules that connect low-level synaptic mechanisms to higher-level circuit architectures and thus contribute to fundamental understanding of retinal function. By examining how different types of ganglion cell integrate their synaptic input, these studies will contribute to an understanding of how parallel channels are set up by retinal circuits. The proposed studies are relevant to retinal diseases that degenerate photoreceptors but spare ganglion cells: a prosthetic device that stimulates the remaining neurons might restore sight to its original quality, but would need to match the original information rate. Thus providing the right amount and kind of information to each neuron will be critical to prosthetic design.
The proposed studies, by measuring information flow through synapses, will develop explicit rules that connect low-level synaptic mechanisms to higher-level circuit architectures and thus contribute to fundamental understanding of retinal function. The proposed studies are relevant to retinal diseases that degenerate photoreceptors but spare ganglion cells. A prosthetic device that stimulates the remaining neurons might restore sight to its original quality, but would need to match the original information rate. Thus providing the right amount and kind of information to each neuron will be critical to prosthetic design.
Showing the most recent 10 out of 16 publications
