ENCODING OF ACOUSTIC AND ELECTRICAL COCHLEAR IMPLANT STIMULATION IN THE AUDITORY CORTEX ARE DISTINCT

Background: Cochlear implants are neuroprosthetics that restore hearing and speech perception to folks with profound hearing loss. A long-standing question is how auditory pathways encode cochlear implants and how these overlap with encoding similar acoustic stimuli under normal hearing conditions. Here, we determined the encoding of tones and cochlear implant channels across the auditory cortex by using micro-electrocorticography (µECoG), a subtype of intracranial electroencephalography (iEEG).
Methods: Tone evoked responses were measured from N=7 normal hearing (NH) rats and cochlear-implant evoked responses from N=7 bilaterally deafened rats unilaterally implanted with an 8-channel intracochlear array. Responses were measured from 60 cortical surface electrodes (3.25 x 3.25 mm grid) placed over the auditory cortex and analyzed using event related potentials (ERPs, 2-100 Hz) or high gamma (HG, 70-140 Hz).
Results: Both tones (NH) and intracochlear electrodes (CI) evoked clear single-trial and trial-averaged transients that were topographically organized. Next, we found single-trial evoked responses were more variable for CI vs NH rats. Then, we trained classifiers on NH and CI evoked responses and found that they could correctly predict NH and CI stimuli. Lastly, we trained a classifier on tone-evoked responses, tested them on intracochlear-evoked responses, and found minimal information transfer between responses evoked by both modalities.
Conclusions: The lack of transferable representations suggests cochlear implant users may experience qualitatively different precepts that cannot be entirely predicted by acoustic hearing. Our results also suggest refinement of cortical responses may improve inter-neuronal variance at the population level and improve auditory perception.