Direct cochlear recordings in humans show a theta rhythmic modulation of the auditory nerve by selective attention

The architecture of the efferent auditory system enables prioritization of strongly overlapping spatiotemporal cochlear activation patterns elicited by relevant and irrelevant inputs. So far, attempts at finding such attentional modulations of cochlear activity delivered indirect insights in humans or required direct recordings in animals. The extent to which spiral ganglion cells forming the human auditory nerve are sensitive to selective attention remains largely unknown. We investigated this question by testing the effects of attending to either the auditory or visual modality in human cochlear implant (CI) users (3 female, 13 male). Auditory nerve activity was directly recorded with standard CIs during a silent (anticipatory) cue-target interval. When attending the upcoming auditory input, ongoing auditory nerve activity within the theta range (5-8 Hz) was enhanced. Crucially, using the broadband signal (4-25 Hz), a classifier was even able to decode the attended modality from single-trial data. Follow-up analysis showed that the effect was not driven by a narrow frequency in particular. Using direct cochlear recordings from deaf individuals, our findings suggest that cochlear spiral ganglion cells are sensitive to top-down attentional modulations. Given the putatively broad hair-cell degeneration of these individuals, the effects are likely mediated by alternative efferent pathways as compared to previous studies using otoacoustic emissions. Successful classification of single-trial data could additionally have a significant impact on future closed-loop CI developments that incorporate real-time optimization of CI parameters based on the current mental state of the user. Significance StatementThe efferent auditory system in principle allows top-down modulation of auditory nerve activity, however evidence for this is lacking in humans. Using cochlear recordings in participants performing an audiovisual attention task, we show that ongoing auditory nerve activity in the silent cue-target period is directly modulated by selective attention. Specifically, ongoing auditory nerve activity is enhanced within the theta range when attending upcoming auditory input. Furthermore, over a broader frequency range, the attended modality can be decoded from single-trial data. Demonstrating this direct top-down influence on auditory nerve activity substantially extends previous works that focus on outer hair cell activity. Generally, our work could promote the use of standard cochlear implant electrodes to study cognitive neuroscientific questions.