Sidney Lehky
Sidney Lehky | |
|---|---|
| Born | Chicago, Illinois, United States |
| Residence | United States |
| Nationality | USA |
| Alma mater | University of Chicago |
| Known for | Biology, Neuroscience, General Interest |
| Scientific career | |
| Fields | Neuroscience, Computational neuroscience, Vision science |
| Institutions | Salk Institute for Biological Studies; Purdue University |
| Doctoral advisor | Hugh Wilson |
Sidney R. Lehky is a computational neuroscientist whose research addresses vision and neural coding. He is affiliated with the Computational Neurobiology Laboratory at the Salk Institute for Biological Studies in La Jolla, California, and has also held an appointment in the Department of Psychological Sciences at Purdue University. He is best known for introducing, with Terrence Sejnowski, the concept of the neural "projective field."
Biography
Lehky was born in Chicago, Illinois. He earned a Bachelor of Arts in biological sciences from the University of Chicago and a Ph.D. in biophysics and theoretical biology from the same institution, where his doctoral work in visual psychophysics was supervised by Hugh Wilson. He subsequently held postdoctoral positions in neural modeling with Terrence Sejnowski and in monkey neurophysiology with researchers including Robert Desimone, John Maunsell, and Keiji Tanaka. He has since been based at the Salk Institute's Computational Neurobiology Laboratory.
Work
Lehky's research combines computational modeling with analysis of neurophysiological data to study how the visual system represents information. In a 1988 paper in Nature with Terrence Sejnowski, "Network model of shape-from-shading: neural function arises from both receptive and projective fields," he described a three-layer neural network trained by backpropagation to extract surface curvature from shading. The receptive fields that emerged in the network's hidden units resembled the edge- and bar-selective receptive fields observed in the visual cortex, even though the units were computing shape from shading. This led to the argument that a neuron's function cannot be inferred from its receptive field alone, and to the idea of the "projective field" — the pattern of a neuron's outward connections to the units it drives.
His later work has examined population coding in the primate visual system, including studies of selectivity and sparseness in the responses of cortical neurons, models of binocular depth perception and stereoacuity, and the influence of eye position on inferotemporal cortex responses. He has proposed the notion of "pseudosparse" neural coding, in which apparently sparse responses to particular stimulus sets reflect narrow tuning across a high-dimensional stimulus space rather than genuine sparseness. He also authored the Scholarpedia article on the projective field.