Search Abstracts | Symposia | Slide Sessions | Poster Sessions
Differential fMRI correlates of discourse production, sentence repetition, and receptive minimal semantic composition support the dual-stream model but not entirely
Poster Session E, Friday, October 2, 11:00 am - 1:00 pm, Wangari Maathai
Joana rosselló1, Pablo Del Olmo1,2, Pilar Salgado-Pineda2, Paola Fuentes-Claramonte2, Aranza Palacios-Guerrero2, Peter Mckenna2, Edith Pomarol-Clotet2; 1University of Barcelona, 2FIDMAG. Germanes Hospitalàries Research Foundation. CIBERSAM, ISCIII, Barcelona
An ongoing fMRI large-scale study on the neural correlates of language processing in schizophrenia distinguishing patients with and without disorganized speech allows us to present the results obtained in the control group (N=31; Spanish-speaking participants) as evidence in favor of the dual-stream model of language in the brain. However, regarding the fine-grained predictions of the most recent version of the dual-stream model (Matchin & Hickok, 2020), only the production task of our study further supports the model, the semantic and repetition tasks rather qualify it. The production task consisted of overt discourse consisting of answering open questions of general interest (e.g., What will the world be like in a hundred years?) contrasted with the repetition of simple sentences (e.g., The brown table has four legs) in alternation (30 second each) for a total of three open questions and three sentence-repetitions. The comprehension task, inspired in Graessner et al. 2021, consisted of assessing whether an adjective+noun composition (noun+adjective, in Spanish) was semantically plausible or not (e.g., unfair law, strong siesta). This condition was contrasted with a non-semantic task in which participants judged whether two written pseudowords were identical or minimally different. The whole-brain fMRI statistical activation maps obtained by our team —and compared with a Chat GPT coarse assessment of 4 of our images— were consistent with the acknowledged gross divide between speech production (greater activation in the left inferior frontal cortex, premotor cortex, SMA, insula and cerebellum than in repetition), receptive semantic processing (greater activation in left temporo-frontal regions than in the pseudoword contrast) and repetition (relevant contrast showing increased bilateral activation in superior temporal, temporo-parietal and sensorimotor cortices). This division of labor is coherent with the dual-stream model in general terms. The main finding speaking in favor of the most recent version of the dual-stream model is related to the activation of the pars triangularis of the left frontal inferior gyrus in the discourse production task (MNI coordinates -56, 22, 2; z = 6.93; 766 voxels; p<0.001). In the model, pTri —rather than the pars opercularis (Friederici 2017)— is considered the principal syntactic frontal region whose function is to transform hierarchical structures formed in the pMTG into sequences of morphemes during speech production. We interpret the pTri differential activation in production vs. repetition as suggesting that the hierarchy-sequence transformation does not necessarily operate during sentence repetition, since meaning construction (hierarchy) is at least partially reduced in the task. What deserves more judgment in relation to the model is that repetition was more bilateral than discourse production, since it would seem excessive to assume that left-lateralization in production was driven by pTri activation alone. Even more challenging for the model is the left lateralization of the frontal and temporal regions during the minimal compositionality judgment task, a comprehension task. In particular, the significant activation of the pTri (MNI coordinates -50, 32, -2; z = 4.73; 482 voxels; p<0.001) is unexpected, which is made more remarkable by the low involvement of the angular gyrus.
Topic Areas: Language Production, Syntax and Combinatorial Semantics