Cortical Over-representation of Speech in Older Listeners Correlates with a Reduction in both Behavioral Inhibition and Speech Intelligibility
Date:
Abstract
The ability to segregate speech from noise deteriorates with aging. Aging is also associated with an exaggerated representation of the speech envelope in auditory cortex. However, the relation between this cortical over-representation and intelligibility of speech in noise in older listeners remains an open question. Here we develop a robust and novel approach based on information theory to investigate the cortical response phase-locked to speech envelopes of both attended and unattended speakers, using mutual information between the neural responses and the speech stimuli. The responses were recorded by magnetoencephalography (MEG). The results show an over-representation of mutual information of both unattended and attended speakers’; speech in older listeners above and beyond the cortical over-representation previously reported. In particular the relative enhancement of mutual information for attended over unattended speech decreases with decreasing SNR, in older listeners for the late response component (~250 ms latency). In contrast, younger listeners show an increasing trend, as might be expected with increasing attention. In older listeners, the mutual information in this same late response to attended speech negatively correlates with behavioral inhibition, as measured by performance in a visual Flanker task, and also negatively correlates with speech intelligibility in noise, as measured by QuickSIN score. Neural source space analysis shows that this late response in older listeners is lateralized to right hemisphere auditory cortical regions. The results suggest that the over-representation of speech in the late response in the aging auditory cortex may be due to the same neural mechanism responsible for decreasing behavioral inhibition, and deterioration of speech intelligibility, in older listeners. From a broader view, this research demonstrates a powerful tool for analyzing phase-locked cortical responses and their relationship to relevant behavioral measures. Download Poster