Investigating Sensitivity to Auditory Cognition in Listening Effort Assessments: A Simultaneous EEG and Pupillometry Study

Abstract

Background: It is still not fully explained what kind of cognitive sources the methods used in the assessment of listening effort are more sensitive to and how these measurement results are related to each other. The aim of the study is to ascertain which neural resources crucial for listening effort are most sensitive to objective measurement methods using differently degraded speech stimuli.

Methods: A total of 49 individuals between the ages of 19 and 34 with normal hearing participated in the study. In the first stage, simultaneous pupillometry, electroencephalogram (EEG), and single-task paradigm reaction time (RT) measurements were made during the challenging listening and repetition task with noise-vocoded speech. Two speech reception thresholds (SRT) (50% and 80%) for two vocoding conditions (16 and 6 channels) were collected, resulting in 4 conditions. In the second stage, the Rey Auditory Verbal Learning Test (RAVLT) and the test of attention in listening (TAIL) were applied. Stepwise linear regression analyses were conducted to examine the predictors of listening effort measurements.

Results: A significant difference was found between 6 and 16 channel stimuli in both pupil dilation change and EEG alpha band power change. In the hardest listening condition, whereas RAVLT scores are significant predictors of pupil dilation change, TAIL scores are significant predictors of EEG alpha power. As the stimulus difficulty decreased, the factors that predicted both EEG and pupillometry results decreased. In the single-task paradigm, a significant regression model could not be obtained at all four difficulty levels.

Conclusion: As a result of the study, it was found that the pupil dilation change was more sensitive to auditory memory skills and the EEG alpha power change was more sensitive to auditory attention skills. To our knowledge, this study is the first to investigate the sensitivity of different listening effort measurement methods to auditory cognitive skills.

Keywords: auditory attention; auditory cognition; listening effort; pupillometry; working memory.

FIGURE 1

Speech stimulus used in the study.

FIGURE 2

Single‐task paradigm RT results under different difficult listening conditions. Although statistically significant difference was found as a result of the repeated measures of ANOVA test (p = 0.03), there wasn't any significant difference found in intra‐group comparisons (p > 0.05). RT, reaction time.

FIGURE 3

Pupil dilation change results under different difficult listening conditions. The repeated measures of ANOVA result was statistically significant (p < 0.001, F = 61.48, η _p ² = 0.86). When we look at intra‐group comparisons with Bonferroni correction, the highest pupil dilation change was seen in the 506 condition with 6 channels and 50% SRT, and this difference was statistically significant (p < 0.001). In intra‐group comparisons, the only match that was not statistically significant was found between the 806 and 8016 conditions with 80% SRT (p > 0.05). SRT, speech reception thresholds.

FIGURE 4

The percentage change in EEG alpha band amplitude results under different difficult listening conditions. The repeated measures of result were found to be statistically significant (p < 0.001, F = 112.33, η _p ² = 0.78). In intra‐group comparisons with Bonferroni correction, the lowest alpha band amplitude change was seen in the 506 condition with 6 channels and 80% SRT, and this difference was found to be statistically significant (p < 0.001). In intra‐group comparisons, the only match that was not statistically significant was found between the 5016 and 8016 conditions with 16 channels, 50% SRT, and 80% SRT (p > 0.05). EEG, electroencephalogram; SRT, speech reception thresholds.

FIGURE 5

Scatter graph of pupil dilation change in 506 condition and its predictors in the stepwise linear regression model.

FIGURE 6

Scatter graph of EEG alpha power change in 506 condition and its predictors in the stepwise linear regression model. EEG, electroencephalogram.