Presentation

Search Abstracts | Symposia | Slide Sessions | Poster Sessions | Poster Slams

Network models of semantic cognition theories show differential mediation of language production by white matter tracts

Poster A58 in Poster Session A, Thursday, October 6, 10:15 am - 12:00 pm EDT, Millennium Hall

Shreya Parchure¹, John Medaglia^1,2, Densie Harvey¹, Harrison Stoll², Apoorva Kelkar², Jared Zimmerman¹, Dani Bassett¹, Roy Hamilton¹; ¹University of Pennsylvania, ²Drexel University

Semantic cognition is critical to language production and comprehension. Patients with semantic impairment due to neural disease have language disabilities that profoundly disrupt daily living. However, the anatomical basis of semantic cognition is as yet poorly understood, with proposed theories implicating left-lateralized regions in the peri-sylvian fissure to a broader network comprising nearly the entire brain. To evaluate the various models of semantic cognition as predictors of language production performance, we used a network neuroscience approach along with behavioral tests. N=31 native English-speaking healthy adults (Age= 25.6 +/- 6 yrs, 13 male) participated in the study. All subjects underwent Diffusion Spectrum Imaging (DSI) scans, with adjacency matrices computed from pairwise region to region streamlines among 234 brain regions reconstructed from tractography within each subject. Each participant completed 2 language tasks: Verb generation and Sentence completion. Networks corresponding to each language model were constructed for each subject: 1. Left inferior fronto-occipital fasciculus (IFOF), 2. Inferior longitudinal fasciculus (ILF), 3. Uncinate fasciculus (UF), 4. Peri-sylvian language network, and 5. Connections from language network to rest of brain. The mean of streamline edge weights, a measure of strength of anatomical network connections, was obtained for each modeled network. A stepwise linear mixed effects modeling approach using Ln(response times) as the behavioral outcome measure was used. For Verb generation, the IFOF (p<0.001) and ILF (p=0.015) significantly predicted responses. UF, Peri-sylvian language network, and edges from language network to rest of brain were not significant (p>0.05). For Sentence completion, models of UF (p=0.0044) and Peri-sylvian language network (p<0.001) were significant. ILF, IFOF, and edges from language network to rest of brain were not significant (p>0.05). This research represents a novel use of network neuroscience to evaluate theories of semantic cognition, identifying relationships between semantically demanding language tasks and specific tracts with connections across the brain. The results corroborate prior fMRI studies on the role of IFOF in semantic processing and retrieval, and the relevance of IFOF and ILF but not UF in Verb generation. Further, a double dissociation between networks predicting Verb generation (IFOF and ILF) and Sentence completion (UF and Peri-sylvian language network) was seen. These findings suggest a neurobiological basis of differential recruitment of language network regions according to the specific semantic processing task.

Topic Areas: Computational Approaches, Control, Selection, and Executive Processes