Neurophysiological and subjective measures of listening effort

Speech perception in suboptimal environments requires extra investment of neurocognitive resources. This is coined listening effort and is a commonly reported problem. Yet, the neurophysiological processes underlying listening effort are not clearly understood. The goal of this study was to investigate the relationship between subjective and objective measures of listening effort when listening to speech in a noisy background. Various approaches to asses listening effort have been devised and there is an ongoing debate on whether objective, physiological measures are associated with the subjective percept of effortful listening.

Using a modified version of the adaptive categorical listening effort scale (ACALES; Krüger et al., 2017) we asked 20 normal-hearing adults to rate their subjective listening effort when listening to the German Oldenburger sentence test (OLSA) in five fixed SNRs (-5, -2.5, 0, +2.5, +5 dB SPL) and for two types of noise (OlNoise, International Female Fluctuating Masker (IFFM)) presented at an intensity of 65 dB SPL. While participants were performing the task, an electroencephalogram (EEG) was recorded using unobtrusive behind the ear EEG electrodes (cEEGrids, Debener et al., 2015). EEG data analysis focused on alpha band activity (8-12Hz) which has been suggested to reflect an inhibitory function of processing task-irrelevant noise signals (Obleser et al., 2012, Strauß et al., 2014) and can be considered as an index of listening effort. Also, in a subset of participants (N=15) speech intelligibility at above mentioned SNRs was measured.

For all SNRs and both noise types speech intelligibility ranged between 94 and 100%, while ACALES ratings indicate that with increasing SNR the perceived listening effort. The EEG data reveal that alpha power spectral density values are higher during poor SNRs as compared to advantageous SNRs. Further analyses show positive correlations between subjective ratings and EEG alpha power. The results confirm that listening to speech in noise induces listening effort even in positive SNRs as well as a linear relationship between subjective listening effort scores and EEG alpha power spectral density with more effort related to more alpha power. These findings support the proposal of alpha as an indicative parameter for detecting changes in listening effort objectively. The speech intelligibility data confirm that measures of listening effort are more sensitive than SI measures for positive SNRs. Finally, the study shows that unobtrusive behind the ear electrodes can be used to measure changes in listening effort revealing the potential to use this technology outside the lab.