Search Abstracts | Symposia | Slide Sessions | Poster Sessions
Characterizing the contributions of anterior temporal lobe areas to semantic language processing
Poster Session A, Friday, September 12, 11:00 am - 12:30 pm, Field House
Leonardo Fernandino1, Jia-Qing Tong1; 1Medical College of Wisconsin
While the anterior temporal lobe (ATL) has been known to contribute to semantic language processing, the precise nature of these contributions remains controversial. Does the ATL constitute/contain a single high-level semantic hub where all concepts are represented, or do different regions of the ATL represent different kinds of concepts? Taking advantage of recent advances in fMRI acquisition and analyses techniques, we investigated these questions using representational similarity analysis (RSA) and multiparametric univariate analyses across 3 independent data sets with high spatial resolution. The results provide novel insights into the functional neuroanatomy of language processing and highlight the differential contributions of distinct ATL regions to semantic cognition. Methods: We analyzed fMRI data from 3 independent experiments. In all experiments, participants performed semantic judgments on individual lexical concepts, which were visually presented as written English nouns. All data sets were acquired with 2 x 2 x 2 mm voxels (gradient-echo EPI). Study 3 used multiband multi-echo EPI acquisition designed to minimize signal drop-out in the anterior temporal lobes. Data were preprocessed with fMRIprep and analyzed with AFNI. Study 1 (n = 39) consisted of 320 unique nouns and Study 2 (n = 40) consisted of 300 unique nouns, each presented 6 times over 3 scanning sessions (Fernandino et al., 2022; Tong et al., 2022). Study 3 (n = 8, data collection still ongoing) consisted of 600 unique nouns, each presented 3 times over 3 scanning sessions. Regions encoding lexical semantic information were identified via multivoxel RSA searchlight (5-mm radius) using a model of word semantics based on experiential feature ratings, controlling for orthographic and phonological similarity. Additionally, a whole-brain general linear model (GLM) analysis including word concreteness, number of phonemes, word frequency, phonological Levenshtein distance (PLD), and response time as predictors was conducted to identify regions where neural activity varied parametrically according to each predictor. Results: At the group level, Studies 1 and 2 converged in revealing four distinct RSA searchlight clusters in the ATL, namely, in the anterior superior temporal sulcus (STSva/STSda), anterior STG (STGa), temporal pole (TGd/TGv), and perirhinal cortex (PeEc). The GLM analysis revealed similar clusters, such that the STSva/STSda, STGa, and TGd/TGv were more strongly activated by abstract than by concrete concepts, while the PeEc was more strongly activated by concrete concepts. Word frequency correlated positively with activation in STSva/STSda and temporal pole, and negatively with activity in PeEc. While activation in the STS cluster was also correlated with the number of phonemes, this was not the case for the other ATL clusters. PLD predicted activation along most of the STG as well as in other areas associated with ortho-phonological processing, but not in area STGa or any of the other ATL clusters. In Study 3, this pattern of results was observed in each individual participant, ruling out potential artifacts of spatial blurring due to cross-participant normalization. Conclusion: The results indicate that different portions of the ATL represent distinct aspects of word meaning, differentially contributing to the processing of concrete and abstract concepts.
Topic Areas: Meaning: Lexical Semantics,