Fig. 8

The MS-INRF model is highly nonlinear. (a) Map of MS-INRF responses to a luminance-defined sinusoidal grating as a function of its spatial and temporal frequencies; each point in the map represents a grating, and the point marked with a red dot denotes a grating that we will call G, for which the MS-INRF model produces a negligible output. (b) Map of model responses to stimuli consisting of the sum of grating G plus another grating whose spatial and temporal frequencies vary; given that the model response to G, when presented in isolation, was negligible, if the model were linear then the map on the right should be identical to the map on the left.