Cognitive factors contribute to speech perception in cochlear-implant users and age-matched normal-hearing listeners under vocoded conditions

Abstract

This study examined the contribution of perceptual and cognitive factors to speech-perception abilities in cochlear-implant (CI) users. Thirty CI users were tested on word intelligibility in sentences with and without semantic context, presented in quiet and in noise. Performance was compared with measures of spectral-ripple detection and discrimination, thought to reflect peripheral processing, as well as with cognitive measures of working memory and non-verbal intelligence. Thirty age-matched and thirty younger normal-hearing (NH) adults also participated, listening via tone-excited vocoders, adjusted to produce mean performance for speech in noise comparable to that of the CI group. Results suggest that CI users may rely more heavily on semantic context than younger or older NH listeners, and that non-auditory working memory explains significant variance in the CI and age-matched NH groups. Between-subject variability in spectral-ripple detection thresholds was similar across groups, despite the spectral resolution for all NH listeners being limited by the same vocoder, whereas speech perception scores were more variable between CI users than between NH listeners. The results highlight the potential importance of central factors in explaining individual differences in CI users and question the extent to which standard measures of spectral resolution in CIs reflect purely peripheral processing.

Serial Recall Predicts Vocoded Sentence Recognition Across Spectral Resolutions

“…The observed correlation between the WASI Perceptual Reasoning Index and overall serial span ability is similar to the previously observed correlations between short-term memory and general fluid intelligence latent constructs (Gignac & Weiss, 2015;Unsworth & Engle, 2007a). There was a nonsignificant trend relating the WASI Perceptual Reasoning Index to sentence recognition, which had approximately the same effect size as previously reported relationships between other tests of fluid intelligence and vocoded speech (O'Neill et al, 2019). Previous research has found weak correlations between speech recognition and various cognitive subscales (Dryden et al, 2017), which may reflect the general association between fluid intelligence and speech recognition.…”

“…This relationship is most evident in listeners with hearing loss, who often struggle with speech recognition even if they use hearing aids or cochlear implants. Listeners with hearing loss often differ from listeners with typical hearing in their relationships between speech recognition and various measures of cognition, such as fluid intelligence, memory, and attention ( Kaandorp et al, 2017 ; Kronenberger et al, 2014 ; Moberly, Harris, et al, 2017 ; Moberly et al, 2016 ; Moberly, Houston, et al, 2017 ; O'Neill et al, 2019 ; G. N. Smith et al, 2019 ). Currently, it is unclear how to interpret these differences across groups of listeners.…”

Amplified cortical tracking of word-level features of continuous competing speech in older adults

“…Thus, although responses to these two surprisal measures cannot be directly compared, the stronger tracking of surprisal by older adults in our study is consistent with increased reliance on predictive processes in this population. This is in agreement with behavioral results demonstrating greater reliance on semantic context in populations with compromised representations of speech, such as those with hearing loss (Benichov et al, 2012;Lash et al, 2013) and cochlear implants (Amichetti et al, 2018;Dingemanse and Goedegebure, 2019;O'Neill et al, 2019). Importantly, the seemingly conflicting pattern of results between these studies could in fact reflect two distinct contributors to speech perception difficulties in older adults, namely decreases in the fidelity of lower level representations, and cognitive decline.…”

Effects of Age on Cortical Tracking of Word-Level Features of Continuous Competing Speech

“…Specifically, a key compensatory mechanism for the poorer fidelity of peripheral representation in the aging population may be increased reliance on the successfully identified segments of speech to aid inference about speech segments with lower SNR. Indeed, greater dependence on contextual cues for speech comprehension has been demonstrated in populations with compromised representations of speech, including those with hearing loss ( Benichov et al, 2012 ; Lash et al, 2013 ) and cochlear implants ( Amichetti et al, 2018 ; Dingemanse and Goedegebure, 2019 ; O’Neill et al, 2019 ). In the context of our feature-tracking model, a possible manifestation of heightened reliance on predictive processes would be increased tracking of surprisal.…”