ATTENTION TO SPEECH MODULATES DISTORTION PRODUCT OTOACOUSTIC EMISSIONS EVOKED BY SPEECH-DERIVED STIMULI IN HUMANS

Humans show a remarkable ability to extract speech from acoustically complex environments. Although the separation of concurrent sound sources is largely attributed to processing in the auditory cortex, extensive descending feedback pathways connect cortical and brainstem centers to the cochlea. The cochlea houses the mechanosensitive hair cells whose mechanical activities equip the cochlea with an active, frequency-selective amplification mechanism. A measurable byproduct of this nonlinear process are distortion-product otoacoustic emissions (DPOAEs), which can be recorded non-invasively from the ear canal. In the present study, we used speech-like DPOAEs derived from the harmonic structure of voiced speech to investigate whether cochlear responses are influenced by selective attention to one of two competing talkers and by intermodal attentional demands [1]. We observed that speech-like DPOAEs associated with spectrally resolved harmonics were significantly reduced when the corresponding voice was attended compared to when it was ignored. In contrast, no attentional modulation was detected for unresolved harmonics of the target voice when the competing voice contained unresolved components in the same frequency range. This pattern suggests that attentional effects at the cochlear level are selective for spectrally resolved elements of speech. Together, these results support the notion that the cochlea’s active process contributes to selective attention in noisy listening situations. Furthermore, the speech-like DPOAE approach introduced here provides a novel method for probing the role of peripheral auditory mechanisms in auditory scene analysis under naturalistic conditions.
[1] J. Steinebach & T. Reichenbach (2025). bioRxiv. doi:10.1101/2025.08.15.670505