Analysis of spatial nonlinear responses in cortical complex cells

We investigate what is essential for the nonlinear spatial structure of the receptive fields of complex cells, specifically, whether pooling of simple-cell responses or intracellular computations of individual cells are crucial for the spatial structure. The complex cells exhibit spatial nonlinear interactions which most likely represent the organization of afferent receptive fields and their interactions. We estimate “steady” Wiener-like kernels over two-dimensions in space by carrying out network simulations of various models including an energy model and a noncascade model. The cascade models pooling simple-cell responses reproduce physiologically estimated kernels, while a noncascade model fails to reproduce the kernel. The results suggests that the afferent connections from simple cells are crucial for the spatial structure of complex cells