Search Abstracts | Symposia | Slide Sessions | Poster Sessions
Predictive coding in the ageing brain: Evidence of inflexible predictions during speech processing in older adults
Poster Session A, Wednesday, September 30, 11:00 am - 1:00 pm, Wangari Maathai
Katerina Tampouraki1, Ediz Sohoglu1; 1University of Sussex
Everyday listening situations are characterized by background noise and competing speech streams, which increase the perceptual demands placed upon listeners. In such conditions, older adults face difficulties understanding speech, and need to exert more effort to do so (Gosselin & Gagne, 2010; Pichora-Fuller et al., 1995). These difficulties persist even when controlling for speech audibility (Fullgrabe et al., 2015). Here we explore whether a key mechanism for speech processing – predictive processing – can explain age-related variability in speech-in-noise comprehension. Behaviourally, speech is more intelligible when words can be predicted and this intelligibility benefit is comparable for older and younger listeners (Pichora-Fuller et al., 1995; Sheldon et al., 2008). This suggests predictive processing remains intact in older adults. By contrast, EEG studies demonstrate a reduced amplitude N400 component in older adults and conclude that predictive processing decreases with age (Federmeier et al., 2002). Considering older adults are more prone to false hearing (Chan et al., 2021), it is also possible that predictions remain intact in older adults but are applied inflexibly. In two online studies, we tested these three accounts of age-related differences in predictive processing. In both studies, younger listeners aged between 18 and 30 years (experiment 1: N = 49, M = 25.8, SD = 3.37; experiment 2: N = 50, M = 24.8, SD = 3.56) and older listeners aged between 60 and 90 years (experiment 1: N = 48, M = 65.4, SD = 4.34; experiment 2: N = 50, M = 69.1, SD = 4.46) heard sentences presented in speech-shaped noise (-9, -6 and 0 dB). Across experiments, sentences ended in final words that were 1) weakly predicted by the preceding words (low constraint condition), 2) moderately predicted (medium constraint), 3) strongly predicted (high constraint) and 4) strongly unpredicted (high constraint with unpredicted final words). Experiment 1 comprised all four sentence conditions, while only the latter two conditions were included in experiment 2. Critically, we controlled for consonant-in-noise discrimination abilities, to statistically adjust for bottom-up sensory differences between age groups. Final word report accuracy did not differ between age groups in the low constraint and high constraint conditions. However, when words were strongly unpredicted, older adults performed worse than younger adults. Overall, these findings point towards older adults having inflexible priors for incoming speech. These “sticky” priors may be an adaptive mechanism when encountering highly predictive speech, but in everyday, uncertain, listening situations, they could have negative effects on speech perception. In ongoing work, we are recording brain responses using EEG to establish the cortical mechanisms underlying these age-related differences in predictive processing.
Topic Areas: Speech Perception,