Modelling the auditory system: preprocessing and associative memories using spiking neurons

摘要

We implemented a model of the auditory system incorporating the cochlea, the cochlear nucleus, the inferior colliculus and the primary auditory cortex. For the simulation of the cochlea gammaton filters are used. The other areas are simulated by neural networks consisting of leaky integrate and fire neurons. The peripheral processing in the cochlear nucleus and the inferior colliculus is modelled by special neural networks for pitch processing and the primary auditory cortex is implemented as an associative memory. Binaural interactions are not considered in the model. The results bear resemblance to biological findings. For the inferior colliculus and the primary auditory cortex we get a tonotopic/periodotopic map, which can be found in the human brain. By using this map we can detect and separate vowels in the primary auditory cortex.