Search Abstracts | Symposia | Slide Sessions | Poster Sessions
Healthy cognitive aging of the language network: A task-based EEG study of interindividual differences
Poster Session C, Saturday, September 13, 11:00 am - 12:30 pm, Field House
Stéphanie De Keulenaer1, Mathias Huybrechts1, Rose Bruffaerts1; 1UAntwerp, Belgium
Introduction In contrast to other cognitive domains, language remains largely preserved with age (Shafto & Tyler, 2014). Nonetheless, neurophysiological studies report age-related changes in the timing of neural processes during sentence processing (Barbieri et al, 2021; Pagán et al, 2025). Furthermore, aging is associated with higher interindividual variability, requiring an individual rather than a group-based approach. Notably, interindividual differences in neurophysiological responsivity during picture naming relate to individual-level cognitive performance (Bruffaerts et al., 2019). To study aging and language at the individual level, we tested the stability of individual neurophysiological responses to sentence processing within a cohort of young and older healthy adults using a functional identification approach, which has been extensively validated in young adults using fMRI (Fedorenko et al., 2010; 2024) and recently investigated using MEG (Huybrechts & Bruffaerts, preprint). Recent fMRI work using this approach has shown that the language network is preserved in aging (Billot et al., preprint). We expand these findings using EEG, as the high temporal resolution is well-suited to detect age-related neurophysiological changes. Methods We collected high-density-EEG in 21 native Dutch-speaking young controls (YHC; mean age: 23.4, education level: 16.5 years, 11 female) and 22 older controls (OHC; mean age: 67.4, education level: 15.1 years, 12 female). The functional identification task contained 80 sentences of 12 words, and 20 control sequences of 12 nonwords, matched for phonological properties and syllable length (adapted to Dutch from Fedorenko et al., 2010). Stimuli were presented sequentially for 500ms. Each sequence was followed by a probe judgement to assess attention and behavioural performance. To test whether EEG allows for a robust identification of subject-specific language-responsive sensors, we evaluated the stability of the language responses within individuals across trials. Specifically, we examined the Spearman correlation in the size of the sentence effect (percent signal change for the sentence condition vs baseline) over all sensors within each participant across odd- and even-numbered trials and compared this to the correlations between different participants (Huybrechts & Bruffaerts, preprint). Results Behavioral data shows no significant differences in response accuracy between YHC and OHC (p =0.27), although response time was significantly higher in OHC (mean YHC: 1123ms, mean OHC: 1364ms, p = 0.01). The mean correlation of the topographies of the neurophysiological responses was significantly higher within the same participant versus between pairs of participants in both YHC (mean within rho: 0.68, mean between rho: -0.01, p < 0.001) and OHC (mean within rho: 0.76, mean between rho: 0.09, p < 0.001). Conclusion The language network can be identified at the individual level in a robust and replicable way using EEG, using a similar approach as in fMRI and MEG. Although language performance remained stable with age, we observed significantly slower response times, perhaps reflecting underlying changes in neural processing efficiency. This individualized framework enables further research into age-related differences in the timing and strength of neurophysiological correlates of language. A potential application is the use of task-based EEG to discriminate between normal and abnormal cognitive aging (De Keulenaer et al., in press).
Topic Areas: Reading, Computational Approaches