The precision of neuronal coding in the auditory brainstem and implications for cochlear implants

Cochlear implants are extremely successful devices, they provide open speech understanding for most patients, despite their limited spectral- and temporal resolution. The largest detriments in performance are in noisy conditions, the appreciation of music, perception of tonal languages and spatial hearing. As for these tasks temporal information is thought to play a crucial role, we analyzed the precision of coding by neurons in the auditory brainstem using the framework of information theory. We used our model of peripheral sound processing, which codes sound into realistic spike trains of the auditory nerve followed by Hodgkin-Huxley type neurons in the first neural processing stage, the auditory brainstem. Stellate neurons, which integrate approximately five supra-threshold inputs from auditory nerve fibers, code the spectral energy of sound. In contrast, the so-called octopus neurons, which exhibit strong onset characteristics, achieve exquisite temporal precision by coincidence detection of their innervating auditory nerve fibers. They are able to code information with a precision in the 20 μs range, which is required for spatial hearing, where octopus neurons are believed to play a key role. For both neuron types, the major portion of information is coded with a temporal precision ranging from 0.2 to 4 ms. Our results motivate novel coding strategies for cochlear implants which code temporal information with much higher fidelity. Only temporal precision in the μs-range will allow children with bilateral implants to develop the neuronal circuitry required for spatial localization based on temporal cues.