Output of my IHC model
I must admit that the IHC model is currently the weakest link in my electronic auditory pathway. Although I'm having a close colaboration with Ray Meddis, the father of the world famous Meddis' Hair Cell model, I'm still not sure how critical the different details of the Inner Hair Cell are for the rest of the auditory pathway to function. My current IHC model implements saturation of the output with increasing input intensity, half wave rectification, and temporal adaptation. The measured output of my model is shown in the figure above when stimulated with a 1kHz, 100mVpp sinusoid. The circuit is shown below. I used this output to feed into my spiking neuron chip.
I also supervised a student project to implement Meddis' Hair Cell Model in analog VLSI. The circuit that came out was, as expected, way to big to be really useful. A second problem with the model is the fact that there is a difference of 3 orders of magnitude between the smallest and largest conductance in the model. The analog implementation therefore needs a precision better than 0.1 percent on the largest signal, for the smallest signal to be still significant. This is simply not possible in an analog electronic model. It might be possible though to modify the model to be compatible with an analog electronic implementation, and create a real good electronic IHC model.