Figure 4. Generalized Cascade Model of the Auditory Transduction Chain.

The model is composed of a sequence containing two linear temporal filters, l(τ) and q(t), and two static nonlinear transformations, namely a quadratic nonlinearity and an output nonlinearity g˜(·), which may differ from the nonlinearity g(·) of the click-stimulus model (see Protocol S1). First, the stimulus A(t) is convolved with the filter l(τ) (linear integration). Second, the result is squared (nonlinear transformation). Third, the result of the previous step is convolved with the filter q(τ), yielding the effective stimulus intensity J(t) (linear integration). Fourth, a final transformation g˜ of J(t) (nonlinear transformation) determines the response, which in this generalized model is the time-dependent firing rate r(t). The model thus corresponds to an LNLN cascade. This abstract structure directly follows the sequential configuration of the biophysical processing steps shown in Figure 1.