Decoding selective attention in cochlear implant users with ipsilateral residual hearing

Cochlear implant (CI) users with ipsilateral residual hearing receive electric and acoustic stimulation (EAS). The combination of both stimulation modalities results in significant benefits in terms of speech understanding when listening to EAS in comparison to electric stimulation (ES) alone or to acoustic stimulation (AS) alone. This benefit is typically measured through behavioral speech understanding tests. However, these assessments do not provide insights on the neural mechanisms underlying electric acoustic integration nor can they be reliably administered to very young children or individuals unable to provide behavioral responses. Recently, it has been suggested that neural tracking to speech from electroencephalography (EEG) can predict speech understanding in normal hearing listeners and in CI users. The main goal of this work is to investigate behavioral speech understanding alongside with cortical auditory evoked potentials (CAEPs) and selective attention decoding in EAS subjects listening to ES, AS and to combined EAS.

Seven EAS subjects participated in the study. First, behavioral speech understanding performance was measured with ES, AS and EAS listening modes. The speech material consisted of target HSM sentences with a competing talker of opposite gender. Integration efficiency for speech understanding was then calculated. Second, CAEPS were recorded by presenting broadband noise with ES, AS and EAS listening modes. Third, selective attention was measured through EEG by monaural presentation of two concurrent speech streams with EAS, ES and AS. During selective attention task, subjects were asked to attend to one of the two concurrent speech streams.

Behavioral performance showed significant improvement of speech score with EAS compared to ES and AS listening conditions. A "supra-additive" electric acoustic integration during the behavioral speech understanding task was observed in 6 out of 7 participants. Moreover, a significant correlation between integration efficiency and amount of residual hearing was observed. In the CAEPs, no significant effect of listening condition on N1 latency or N1P2 amplitude was observed. Selective attention decoding revealed a significant difference between the attended and unattended correlation coefficients only with the EAS listening condition, indicating a possible integration of electric and acoustic simulation to facilitate segregation of the speech streams.

In summary, the main findings of this work are: 1) listening with EAS results in significant improvement in behavioral speech understanding compared to AS and ES; 2) Integration efficiency in speech understanding significantly correlated with amount of residual hearing in the current group of participants; 3) A significant difference between attended and unattended speakers was possible only with EAS listening condition showing a possible integration of both stimulation modalities to facilitate selective attention decoding, however, further research in a larger cohort is necessary.