Search Abstracts | Symposia | Slide Sessions | Poster Sessions
Individual differences in the neural correlates for narrative language comprehension
Poster Session C, Thursday, October 1, 10:45 am - 12:45 pm, Wangari Maathai
Laura Giglio1, Atsuko Takashima2, Muge Ozker Sertel2,3, Florian Hintz4, Antje Meyer3, Peter Hagoort2,3; 1Institute for the Interdisciplinary Study of Language Evolution, University of Zurich, 2Donders Institute for Brain, Cognition and Behaviour, Radboud University, 3Max Planck Institute for Psycholinguistics, 4University of Marburg
The core brain network for language processing is usually investigated by focusing on group activations, that is, where participants show common activation to the same task. However, the individual patterns of brain recruitment are known to vary widely based on individual neuroanatomy and individual variability in processing more generally. In the current study, we investigated whether and how brain activity varies as a function of individual differences in processing skills. Critically, we focused on naturalistic language processing, to ask which brain areas are modulated by individual differences independently from task processing requirements set by standard fMRI language tasks. Participants (n=192) first took part in an extensive behavioural battery, which through several tasks aimed to characterize their individual differences in four factors: linguistic knowledge (LK), processing speed (PS), working memory (WM) and nonverbal reasoning (NVR). In an fMRI study, participants’ brain activity was then recorded while they listened to two short Dutch narratives for 7 minutes. The narratives were annotated for word frequency, based on SUBTLEX-NL, and word surprisal and entropy, computed using transformer model GPT-2. We first characterized which brain areas underlie naturalistic language processing. In line with ample evidence on the neural correlates of language comprehension, word frequency, surprisal and entropy modulated responses along the superior temporal gyrus, inferior frontal gyrus, as well as in clusters in middle and superior frontal areas. We then asked how brain activity elicited in response to these specific indices of word processing effort varied as a function of processing skills as captured by the behavioural scores for PS, LK, WM, and NVR. Participants with higher LK had increased activation in the medial superior frontal gyrus during word listening. Lower LK instead resulted in stronger activation in a cluster in the left superior parietal lobule in response to word processing. Lower LK was also associated with a stronger response in the right posterior middle temporal gyrus for the effect of word frequency, and a stronger response in the supplementary motor cortex in response to surprisal. Participants with lower scores on PS had higher activation in the right STG in response to word comprehension. Higher PS was instead associated with increased responses in the right middle frontal gyrus for surprisal. Finally, higher NVR scores were associated with an increased response in the right middle frontal gyrus with increasing entropy. Individual differences in WM were not associated with varying activation in naturalistic comprehension. Overall, we found that a large fronto-parieto-temporal network is modulated by incremental metrics characterizing the linguistic input participants heard. Interestingly, this network was for the most part bilateral, indicating that right homologues of core language regions are also engaged in naturalistic language processing. The brain areas sensitive to individual differences were for the most part in the right-hemisphere and not overlapping with the core responses to lexical processing, but adjacent to those clusters. The results suggest individual differences modulate the responses of broader neural systems engaged in support of language processing, such as the fronto-parietal control network and attention networks.
Topic Areas: Meaning: Discourse and Pragmatics, Meaning: Lexical Semantics