Any (causal) filter that does not follow the form (11.3) is called a nonlinear filter. Recall from Section 11.2, that the operation of the human auditory system is almost a linear filter, but exhibits characteristics that make it into a nonlinear filter. Linear filters are preferred because of their close connection to spectral analysis, or frequency components, of the signal. Even if the human auditory system contains some nonlinear behavior, analysis based on linear filters is nevertheless valuable.