Neuromechanistic models for auditory perception

In this thesis, I developed two neuromechanistic models, pitch comparison model and pitch detection model, to explain how the brain extracts sound features and how stimulus history can influence auditory perception.;The pitch comparison model is presented in the first part. We developed a recurrent, firing-rate network model that accounts for context-dependent perception of pitch change direction. We focus especially on an ambiguous comparison; listeners experience opposite percepts (either ascending or descending) for the ambiguous tone pair depending on the spectral location of preceding context tones. Our model detects local frequency change direction of successively played stimuli using asymmetric inhibition. A novel adaptation mechanism, slow-facilitating inhibition, successfully accounts for the context-dependent perception demonstrated in behavioral experiments.;The pitch detection model is presented in the second part. Stimuli with distinct spectra can elicit the same pitch. We developed a multi-stage neuronal model for pitch detection using biophysical slope detectors. The slope detectors can detect coincidence among convergent auditory nerve fibers with high temporal precision and fire regularly with the same periodicity as that of the sound. Therefore, pitch can be decoded from their inter-spike interval histograms. Our model successfully accounts for a variety of pitch phenomena. The regular firing pattern of our slope-detectors is invariant to stimulus type, thus can serve as a neural basis for pitch equivalence. Different from other temporal models, our model does not require a delay line, thus more physiologically realizable.