Resilience and vulnerability of neural speech tracking in children with cochlear implants

Abstract

AbstractInfants are born with biological biases that favour language acquisition. One is the auditory system’s ability to track the envelope of continuous speech, a pivotal feature for spoken language comprehension in adulthood. However, the extent to which neural speech tracking relies on postnatal auditory experience remains unknown. In this case-control study, we tested children with or without access to functional hearing in the first year of life after they received cochlear implants (CIs) for hearing restoration. We measured neural speech tracking in CI users with a congenital bilateral profound deafness (CD) or who acquired it later in development (AD; minimum auditory experience after birth 12 months), as well as in two groups of hearing controls listening to original (HC) or vocoded-speech (HC-v). Remarkably, neural speech tracking in children with CIs was unaffected by the absence of perinatal auditory experience. Regardless of deafness onset, CI users and HC exhibited a similar neural tracking magnitude at short timescales ∼50– 130 ms (P1TRF) of brain activity. However, this neural tracking phase (P1TRF) was delayed in CI users, and its timing depended on the age of hearing restoration. Conversely, at longer timescales ∼130–260 ms (N2TRF) of brain activity, speech tracking was substantially dampened in participants with CIs, thereby accounting for their comprehension deficits. Speech tracking in HC listening to vocoded-speech and in a phantom head-model with CIs suggested that neural processing differences between HC and CI children could not merely be explained by the degraded acoustic stimulation, nor by electrical artifacts of the implants. These findings highlight (i) the resilience of sensory components of neural speech tracking to the lack of hearing in the first year of life, (ii) the crucial role played when hearing restoration takes place in mitigating the impact of atypical auditory experience, (iii) the vulnerability of higher hierarchical levels of speech processing in CI users.