Daniel Butts
(University of Maryland)
Wednesday 11th January 2012
16:00
B10 Seminar Room, Basement,
Alexandra House, 17 Queen Square, London, WC1N 3AR
Linking nonlinear models to mechanism in pre-cortical vision
How visual information is represented through the pre-cortical processing stages of retina and lateral geniculate nucleus (LGN) is an essential component of understanding downstream processing of vision. While the familiar center-surround receptive field provides a coarse description of what aspects of visual information are being conveyed, such descriptions gloss over the complexity of retinal processing, which involves multiple linear and nonlinear stages resulting in many different types of output neurons (retinal ganglion cells). Here, we apply biologically motivated nonlinear modeling framework to recorded spike trains in the retina and LGN, as well as intracellular recordings in identified RGC subtypes, to characterize nonlinear aspects of pre-cortical processing and infer potential underlying mechanisms. We find a more complex (and cell-type specific) description of pre-cortical processing that potentially accounts for well-known nonlinear response properties such as contrast adaptation and generation of temporal precision. It suggests how particular biophysical mechanisms contribute essential components of the spike trains going into cortex, and more generally presents modeling methods to characterize nonlinear computation in sensory systems.