Cognitive factors contributing to speech-in-noise comprehension: insights from a thousand young, normally-hearing listeners

The ability to comprehend speech under acoustically challenging conditions varies widely across individuals. This variability is typically attributed to cognitive factors that may have a role in supporting the listening effort required to comprehend speech experienced in adverse listening conditions. This notion has been formalised in a number of models that focus on the cognitive factors supporting speech comprehension in hearing impairment.

Ease of Language Understanding model (ELU, Rönnberg et al., 2013, Front. Syst. Neurosci. 7:31) posits that comprehending noisy speech specifically places demands upon processing in working memory. The Framework for Understanding Effortful Listening (FUEL, Pichora-Fuller et al., 2016, Ear Hear. 37:5S) places a decision making mechanism at the heart of the cognitive processes implicated in speech-in-noise comprehension, which weighs the demands of the task against the potential for success, executing this decision before task engagement and resource allocation. These models provide compelling bases upon which to consider the role of working memory and cognitive flexibility in adverse listening situations. However, evidence supporting these models in younger, normally-hearing, listeners remains scant (reviewed by Füllgrabe & Rosen, 2016, Front. Psychol. 7:1268). Here we use a large dataset, to probe the relationship between speech-in-noise perception ability and a battery of cognitive factors.

The Human Connectome Project (HCP, Van Essen et al., 2012, NeuroImage 62:2222), is a neuroimaging and behavioural dataset of over a thousand young (age range 22 – 37 years, Mean = 28.80, SD = 3.69), healthy participants with no self-reported history of hearing impairment. The HCP provides data on word in noise recognition (derived using the NIH words-in-noise test, hereafter “WIN”) and a battery of several dozen cognitive indicators. We used these data to examine the relevance of cognitive factors to WIN using simple linear correlations. After correcting for age, we identify a number of significant (p<.05, Bonferroni corrected for multiple comparisons) cognitive predictors of WIN, including working memory (r=.186), crystallised intelligence (r=.220) and fluid intelligence (r=.178).

While the implication of crystallised intelligence, which contains measures of vocabulary, may not be altogether illuminating, the fact that both working memory and fluid intelligence are significantly associated with speech-in-noise comprehension performance lends support to the central tenets of both the ELU and FUEL. Importantly, although the proportion of variance explained by these particular cognitive factors may be low (working memory: 3.5%, fluid intelligence: 3.2%), this establishes that they are relevant even in younger listeners, although other factors are evidently also in play.