Slide Slam M1
Explaining semantic cognition in aging — evidence from a longitudinal large-sample study
Sarah Tune1, Jonas Obleser1; 1University of Lübeck, Germany
Healthy aging is accompanied by cognitive changes. It is generally assumed that different cognitive domains follow their own life-span trajectories: while cognitive abilities supported by fluid intelligence peak in early adulthood and decline thereafter, signatures of crystallized intelligence like vocabulary and knowledge remain relatively stable or even increase with older age (Hartshorne & Germine, 2015). Based on this simplistic distinction, one might hypothesize that semantic cognition, built on our accumulated knowledge about the world, is generally preserved in older age. However, performance in tests of semantic cognition does not only rely on the preservation of knowledge itself. Semantic cognition also requires high fidelity in the executive control processes operating around it. Notably, performance of older compared to younger adults was recently found specifically impaired in a task requiring the controlled selection of task-relevant and the suppression of competing, task-irrelevant semantic information (Hoffman, 2018). We here follow up on this effect in a longitudinal cohort of age-varying adults (N=101, 42–82 years, median at T2 = 63 years). First, we aimed at replicating the opposing effects of age on verbal intelligence and semantic control processes, and further probed their age-independent relationship. Second, we hypothesized that controlled selection abilities are largely reflective of domain-general executive control. Selection performance in semantic cognition of healthy ageing adults should thus be predictable from neural signatures of executive control recorded in an independent task approximately 1.5 years earlier. We assessed verbal intelligence using the spot-the-word test, and administered a novel German version of the 2x2 design by Hoffman (2018) which contrasts performance under different tasks (retrieval vs. selection) and levels of control demand (low vs. high). As signatures of executive control, we used the degree of 8–12 Hz alpha power lateralization and selective neural speech tracking during a difficult dual-talker listening task (Tune et al., 2021). As expected, increased age led to a decrease in accuracy that was specific to the more difficult condition (i.e. requiring a high level of control) of the controlled selection task (Task x Control demand x Age interaction; odds ratio (OR) = .67, standard error (SE) = .2, p < .5). By contrast, we observed a positive trend of verbal intelligence scores with increased age (β=.18, p=.08). When controlling for age, we found accuracy in the controlled selection task positively associated with verbal intelligence (β = .37, SE = .09, p <.001). Additionally, accuracy was predicted by the degree of alpha power lateralization (β = .23, SE = .09, p <.05), speaking to their joint involvement in executive control processes. Taken together, we provide converging evidence for the contribution of different cognitive domains to an individual’s abilities jointly referred to as semantic cognition, and for the opposing forces increased age exerts on them. The results highlight the importance of teasing apart the influences of crystallized and fluid intelligence when examining age-related performance changes in various semantic tasks.