Listening effort, referring to the cognitive resources engaged when understanding speech in challenging conditions, can be assessed through behavioral and physiological measures. While the behavioral measures tend to be subjective, the use of physiological measures, such as electroencephalography (EEG), has shown mixed results. The multidimensionality of listening effort can thus be measured in different ways. Objective measures, such as EEG, offer the possibility to better understand the neural mechanisms underlying speech processing. In this study, we used EEG to examine how alpha oscillations are modulated during effortful listening and whether these changes originate from different neural sources. Alpha oscillations have been described in relation to listening in complex environments in contradictory ways. Some studies report increases in alpha power during effortful listening, while others describe a decrease in alpha power. One explanation for these inconsistencies seems to rely on the presence of multiple alpha generators in the brain. In this study, we investigated alpha oscillations across different difficulty levels in both speech-in-noise (SIN) and speech-in-speech (SIS) conditions.

Speech intelligibility, subjective listening effort and electrophysiological data were recorded during speech-in-speech and speech-in-noise performance of 30 participants. Spectral and time-frequency analyses were used to assess cortical responses to different levels of auditory difficulty. Because interpreting alpha oscillations in the time-frequency domain alone can be ambiguous, we extended the analysis by applying independent component analysis (ICA) to the EEG data. This approach allowed us to determine whether observed alpha dynamics reflected distinct underlying neural components. In addition, exploratory source localization was performed to identify the cortical origins of these components.

Results showed that, in the time-frequency domain, alpha power increased in the left temporo-parietal region during both SIN and SIS conditions, consistent with greater engagement in difficult listening. However, ICA revealed that these changes were associated with different independent components. Some components exhibited alpha synchronization, while others showed desynchronization. This pattern indicates that alpha oscillations during effortful listening are not generated by a single source but by multiple alpha generators with distinct functional roles. Alpha-band EEG activity may reflect greater neural inhibition, which could indicate a change in listening processes during the task. This modulation seems linked to task difficulty and may also relate to listening effort. By refining these results by extracting independent components related to alpha activity described in the literature, we confirmed that alpha power observed in the time-frequency domain is the result of different independent components, and that both enhancement and suppression of alpha can be observed during listening in SIN and SIS.