Abstract:
Especially important in developing improved cochlear implants is to develop a deeper understanding of the processing of sound in the central auditory nervous system, for both acoustic and electrical stimulation of the auditory system. This thesis contributes to this objective through cochlear implant psychoacoustic research and modelling of auditory system sound processing. The primary hypothesis of the thesis was that the same underlying mechanisms are responsible for sound perception in both electric and acoustic hearing. Thus, if appropriate models are created for normal acoustic hearing, they should be able to predict psychoacoustic data from electric hearing when the model input is changed from acoustic to electrical stimulation. A second hypothesis was that electrode interaction could be measured by gap detection and that predictions of current spread in the cochlea could be obtained from gap detection data. Measured gap detection thresholds in three cochlear implant users were a function of the physical separation of electrode pairs used for the two stimuli that bound the gap, resulting in a U-shaped "tuning curve" for this across-channel condition. Models of gap detection in acoustic and electric hearing were created to explain these U-shaped curves. A technique was developed to obtain estimates of cochlear current spread from gap detection data. Predictions of electrode discrimination were obtained from the current spread estimates, and these were compared to data measured in cochlear implant users. The model for acoustic hearing could predict the U -shaped curves found in acoustic hearing, and when the input spike train statistics were adapted appropriately, the same model could also predict gap detection data for electric hearing. Predictions of current spread exhibited current peaks close to the electrodes and had length constants between 0.5 mm and 3 mm, similar to measured data quoted in literature. Predictions of electrode discrimination correlated well with measured data in one subject, but not in two others. The primary conclusion from the modelling results is that if the mechanisms of central auditory nervous system signal processing of acoustic stimulation are understood, these same mechanisms may be applied to understand the signal processing in auditory electrical stimulation and to predict psychoacoustic data for electrical stimulation. A second conclusion is that spatial mechanisms, as opposed to temporal mechanisms, may determine gap detection thresholds in the across-channel condition. This is important in cochlear electrical stimulation, where spike trains are strongly phase-locked to the stimulus and temporal mechanisms cannot predict gap detection thresholds. A third conclusion is that gap detection can be used to measure channel interaction and to predict current distributions in the cochlea, although there is still uncertainty about the accuracy of these predictions. However, the gap detection data and predictions for current distributions indicate that electrodes are not discriminable when they are closer than 1.5 mm. The implication of these last two conclusions taken together is that research should focus on obtaining better spatial resolution in cochlear implants.
