Modeling cortical responses to mixture stimuli reveals origins of orientation tuning variation

R Goris, E P Simoncelli and J A Movshon

Published in Computational and Systems Neuroscience (CoSyNe), (II-85), Feb 2013.

This paper has been superseded by:
Origin and function of tuning diversity in macaque visual cortex
R L Goris, E P Simoncelli and J A Movshon.
Neuron, vol.88 pp. 819--831, Nov 2015.

Neurons in primary visual cortex are typically selective for the orientation of bars and gratings. But there is considerable variation in the sharpness of orientation tuning. The selectivity of cortical cells in part originates in the linear summation of thalamic inputs, and in part in the nonlinear rectification of the membrane voltage. It is not known whether the variation in tuning sharpness is primarily determined by the linear or the nonlinear stage. To address this question, we performed a simple superposition experiment that allowed us to estimate the contribution of both stages. We studied the orientation selectivity of cortical neurons in anesthetized macaques using stimuli that varied in their orientation composition -- our stimuli were composed of multiple incoherently-drifting sinusoidal gratings whose orientations were spaced at 20 deg intervals and whose contrasts were drawn from a Gaussian distribution (standard deviation 0-55 deg). We developed a linear-nonlinear model of orientation selectivity that successfully accounted for the various changes in response gain and tuning bandwidth with stimulus bandwidth. Analysis of the model parameters revealed that both the linear and nonlinear stage significantly contribute to the variability in orientation selectivity as determined with bars and gratings. However, most of this variability originates in the linear processing stage. Thus, the variation in the sharpness of orientation tuning in primary visual cortex may largely originate in differences in the arrangement of thalamic inputs.
  • Listing of all publications