Generation and transformation of second-order nonlinearity in catfish retina.

A large part of the response from catfish retinal neurons evoked by a white-noise modulated light stimulus is reconstructed by the linear and the second-order nonlinear components, which shows that the first- and second-order kernels represent the major response characteristics. In catfish retina, amacrine cells are classified as type-C and type-N cells. Type-C cells produce a stable and stereotyped second-order kernel that can be reproduced by squaring an underdamped first-order kernel. This is a linear filter followed by a static nonlinearity and is modeled by a cascade of the Wiener structure. A second-order kernel from the other class of amacrine cells, type-N cells, is reproduced by a simple linear filtering of type-C cell response. This is a static nonlinearity sandwiched between two linear filters and is modelled by a cascade of the Korenberg structure. These findings may greatly simplify future attempts to reconstruct retinal circuitry and may give some insight into the process of complex signal processing in the inner part of the vertebrate retina.