Slide Slam F10
Different listening strategies for natural speech in elderly: band-specific aging effects
Pieter De Clercq1, Jonas Vanthornhout1, Maaike Vandermosten1, Tom Francart1; 1KU Leuven
One recent advance in the field of neurobiology of language is to work with natural, connected speech. When listening to natural speech, the brain tracks the envelope of the stimulus, which contains important cues for speech understanding . Research in healthy aging found, paradoxically, increased neural tracking of the envelope in elderly while speech understanding was decreased by aging [2,3]. These studies investigated the brain’s response in a 1-8 Hz frequency range, containing both the theta and delta band. Disentangling these frequency bands might provide additional information, as they are believed to represent distinct processes. The timescale of the theta band corresponds to syllables and was found to uniquely contribute to acoustically processing the speech signal. The delta band on the other hand, encodes words and phrases and contributes more to actual speech understanding . In the present study, we disentangled these frequency bands and separately investigated aging effects. Method: We did a re-analysis of previously published data demonstrating enhanced cortical tracking of natural speech in elderly (with a 1-8 Hz frequency band) . 11 Young (17-37 yo), 32 middle-aged (42-60 yo) and 12 healthy older adults (62-82 yo) listened to a 15-minutes long story while electroencephalogram (EEG) was recorded. We calculated the mutual information (MI) between the speech envelope and the EEG data in the delta (0.5-4 Hz) and theta (4-8 Hz) bands. The MI can be considered as a measure of information sharing between two signals, with higher values reflecting a cortical response that is more tuned to the speech signal. This measure was recently applied in this field, demonstrating enhanced cortical responses for elderly with higher statistical power compared to linear methods . Results: For the 1-8 Hz broadband signal, we replicated previous findings: a non-parametric cluster-based permutation test revealed a significant enhanced MI for older individuals in early processing stages (5-80 ms, p = 0.029). However, for the separate delta and theta bands a linear mixed effects model revealed a strong significant interaction effect between age and band (F = 8.15, p = 0.006). The MI was negatively correlated with age in the delta band (i.e., higher MI for younger individuals, Spearman’s R = -0.26) and correlated positively with age in the theta band (i.e., higher MI for older individuals, Spearman’s R = 0.31). Conclusion: The enhanced cortical response to natural speech stimuli is band-specific rather than generic. We therefore hypothesize that aging is accompanied by a shift in speech listening strategy, with elderly relying more on acoustic cues of the stimulus. These results can be integrated with recent work demonstrating that enhanced cortical tracking originates from an early acoustic response in the auditory cortex  and work reporting decreased cortical responses to language features at the semantic level in elderly .