The spikes leaving the cochlear model can be fed into a neural network. Here they are fed to one neuron only to model coincidence detection. The idea is that the periodicity of a harmonic sound could be neuronally detected when single neurons spike at the frequency of the fundamental of the sound. This is indeed found in neurons in the nucleus cochlearis and the trapezoid body.
Therefore all neural outputs from the cochlear are fed into a neural model simulating the behaviour of a single neuron. It appears that coincidence detection depends on the strength of the input. With strong input periodicities up to about 300 Hz are detected by the neuron. Still above this frequency single neurons do not seem to be able to perform coincidence detection. This is pointing to a neural network performing, where single neurons showing coincidence are only doing so because they are part of a neural network.
Behaviour of a neuron with complex tone input fed from all cochlear bark bands
Perfect coincidence detection of a neuron works only up to about 300 Hz