Auditory-cognitive contributions to speech-in-noise perception determined with structural equation modelling of a large sample

Problems understanding speech-in-noise (SIN) are commonly associated with peripheral hearing loss. But pure tone audiometry (PTA) alone fails to fully explain SIN ability. This is because SIN perception is based on complex interactions between peripheral hearing, central auditory processing (CAP) and other cognitive abilities. We assessed interaction between these factors and age using a multivariate approach that allows the modelling of directional effects on theoretical constructs: structural equation modelling. We created a model to explain SIN using latent constructs for sound segregation, auditory (working) memory, and SIN perception, as well as PTA, age, and measures of non-verbal reasoning. In a sample of 207 participants aged 18-81 years old, age was the biggest determinant of SIN ability, followed by auditory memory. PTA did not contribute to SIN directly, although it modified sound segregation ability, which covaried with auditory memory. A second model, using a CAP latent structure formed by measures of sound segregation, auditory memory, and temporal processing, revealed CAP to be the largest determinant of SIN ahead of age. Furthermore, we demonstrated the impact of PTA and non-verbal reasoning on SIN are mediated by their influence on CAP. Our results highlight the importance of central auditory processing in speech-in-noise perception.