Robust Predictive Mechanisms in Aging: Insights from Behavioral and Neuroimaging Research
Poster Session E - Monday, March 31, 2025, 2:30 – 4:30 pm EDT, Back Bay Ballroom/Republic Ballroom
Sandra Martin1 (martin@cbs.mpg.de), Merle Schuckart2,3, Jonas Obleser2,3, Gesa Hartwigsen1,4; 1Max Planck Institute for Human Cognitive and Brain Sciences, Leipzig, Germany, 2Department of Psychology, University of Lübeck, Germany, 3Center of Brain, Behavior and Metabolism, University of Lübeck, Germany, 4Wilhelm Wundt Institute for Psychology, Leipzig University, Germany
Language prediction is crucial for understanding and facilitating everyday communication. For instance, when navigating through heavy traffic while following navigation instructions, it acts as a catalyst that enhances processing efficiency. However, there is ongoing debate how much language prediction taxes executive resources. Cognitive aging provides an intriguing context to explore this interaction due to age-related sensory and executive decline. We present data from two large studies: a behavioral study (n=175) and a neuroimaging study (n=90), using a dual-task paradigm that combines text reading with an n-back task of varying cognitive demands. Language predictability is measured by word surprisal, derived from GPT-2. Behavioral study results, using linear and non-linear mixed-effects regression, show that limiting executive resources reduces the benefits of language predictability on reading times across the adult lifespan (20-84 years). Interestingly, even under high cognitive demand, older adults exhibit the strongest language predictability effects, with faster reading times for low surprisal words. These findings have significant implications. They highlight the resource-dependent nature of predictive mechanisms in language comprehension, suggesting that linguistic predictions rely on executive resources. The age-dependent effects underscore the susceptibility to increased processing costs with age, while also demonstrating robust predictive mechanisms in older age. This skill may serve as a compensatory mechanism for executive decline with age. Initial neuroimaging results indicate that predictability processing occurs in a left-lateralized fronto-temporal network. Analyses targeting changes associated with advancing age will further emphasize the age-dependent modulation of this network, offering insights into the neural basis of language prediction in aging.
Topic Area: LANGUAGE: Lexicon