Auditory information coding by modeled cochlear nucleus neurons
Journal of Computational Neuroscience
| Hi-index | 0.00 |
The human auditory system excels in the detection of signals in background noise. We evaluate the principles of robust processing with a detailed inner ear model and a model of octopus neurons in the cochlear nucleus. These neurons reject steady-state excitation and fire on signal onsets with extremely high temporal precision. Spike-triggered reverse-correlation analysis revealed that octopus neurons fire preferentially if many coincident spikes follow a short interval of relative low excitation. The frequency spectrum of the reverse-correlation revealed that octopus neurons perform a band-pass analysis of the incoming signal, with the pass-band ranging from about 110 to 650 Hz. The low-frequency slope was approximately 6 dB/oct, which indicates that octopus neurons process the first derivative of the input signal. This mechanism not only removes steady-state activity, which accentuates onsets, but also enhances amplitude modulation in the frequency region predominant in speech.