The Omstein-Uhlenbeck neuronal model is investigated under the assumption t
hat the amplitude of the noise is signal dependent. A linear approximation
of the input-output transfer function is developed. Different types of depe
ndencies of the noise on the signal are considered. The conditions are pres
ented under which a substantial difference between transfer functions with
constant and signal-dependent noise appears. An example for which the: tran
sfer function yields contra-intuitive behavior is presented. (C) 2001 Publi
shed by Elsevier Science B.V.