While cochlear implant users achieve surprisingly high speechrecognition scores in clean acoustic conditions, their ability to localize sounds is severely limited. One reason for this impairment is that their ability to detect interaural time differences (ITDs) is much worse than for normal hearing subjects, which is surprising, as this happens even in laboratory conditions with precise single electrode stimulations. In this talk we feed the neuronal signals derived from a detailed finite element model of the current spread in a human inner ear based on high-resolution (6 micrometer voxel size) micro-computertomography, and a multi-compartment model of auditory nerve fibers into a phenomenological model of the medial superior olive developed by Encke. In the first step, we estimated the temporal jitter of electrically evoked action potentials of nerve fibers along the cochlea. It turned out that the neurons in the medial superior olive in the mammalian auditory pathway evaluate the overlap between ipsilateral excitatory inputs and contralateral inhibitory inputs. With this knowledge we will highlight the consequence of the neuronal processing for the detection of interaural time differences and suggest why it fails in electric hearing.