Effects of Hearing Impairment and Hearing Aid Amplification on Listening Effort: A Systematic Review
OBJECTIVES Although previous research indicates that cognitive skills influence benefit from different types of hearing aid algorithms, comparatively little is known about the role of, and potential interaction with, hearing loss. This holds true especially for noise reduction (NR) processing. The purpose of the present study was thus to explore whether degree of hearing loss and cognitive function modulate benefit from different binaural NR settings based on measures of speech intelligibility, listening effort, and overall preference. DESIGN Forty elderly listeners with symmetrical sensorineural hearing losses in the mild to severe range participated. They were stratified into four age-matched groups (with n = 10 per group) based on their pure-tone average hearing losses and their performance on a visual measure of working memory (WM) capacity. The algorithm under consideration was a binaural coherence-based NR scheme that suppressed reverberant signal components as well as diffuse background noise at mid to high frequencies. The strength of the applied processing was varied from inactive to strong, and testing was carried out across a range of fixed signal-to-noise ratios (SNRs). Potential benefit was assessed using a dual-task paradigm combining speech recognition with a visual reaction time (VRT) task indexing listening effort. Pairwise preference judgments were also collected. All measurements were made using headphone simulations of a frontal speech target in a busy cafeteria. Test-retest data were gathered for all outcome measures. RESULTS Analysis of the test-retest data showed all data sets to be reliable. Analysis of the speech scores showed that, for all groups, speech recognition was unaffected by moderate NR processing, whereas strong NR processing reduced intelligibility by about 5%. Analysis of the VRT scores revealed a similar data pattern. That is, while moderate NR did not affect VRT performance, strong NR impaired the performance of all groups slightly. Analysis of the preference scores collapsed across SNR showed that all groups preferred some over no NR processing. Furthermore, the two groups with smaller WM capacity preferred strong over moderate NR processing; for the two groups with larger WM capacity, preference did not differ significantly between the moderate and strong settings. CONCLUSIONS The present study demonstrates that, for the algorithm and the measures of speech recognition and listening effort used here, the effects of different NR settings interact with neither degree of hearing loss nor WM capacity. However, preferred NR strength was found to be associated with smaller WM capacity, suggesting that hearing aid users with poorer cognitive function may prefer greater noise attenuation even at the expense of poorer speech intelligibility. Further research is required to enable a more detailed (SNR-dependent) analysis of this effect and to test its wider applicability.