A class of stochastic visual textures is used to analyze the components of the visual evoked potential (VEP). This procedure exploits the differential sensitivity of populations of visual neurons to aspects of contrast and pattern. A simple transformation of VEP responses elicited by these stimuli separates components that reflect complex aspects of visual processing from those that reflect elementary aspects.
Simultaneous recordings of the VEP and cellular activity in the cat lateral geniculate nucleus are obtained. Responses to traditional VEP stimuli contain a mixture of intracortically-generated and precortically-generated components. A theoretical and experimental analysis demonstrates that the present approach cleanly separates intracortical generators of the VEP from precortical generators.