Poster B29, Thursday, August 16, 3:05 – 4:50 pm, Room 2000AB

Dynamic Conceptome: Semantic hubs and spokes form functional modules in the whole-brain graphs with intra-/inter-modular connectivity

Seyedehrezvan Farahibozorg1,2, Olaf Hauk1;1MRC Cognition and Brain Sciences Unit, University of Cambridge, 2Wellcome Centre For Integrative Neuroimaging, Nuffield Department of Clinical Neurosciences, University of Oxford

The fast temporal dynamics of connectivity between nodes of the semantic networks while a concept unfolds in the brain are largely unknown. To tackle this, in this study we proposed an unprecedented translation of the predictions of the hub(s)-and-spokes framework (Lambon-Ralph et al. 2016, Patterson et al. 2007) into quantifiable concepts of network community and modularity from graph theory. Importantly, without imposing any restrictions on the number or locations of the potential hub and spoke modules and connectivity among them, we aimed at identifying these characteristics of the semantic networks using a data-driven approach. For this purpose, we recruited 19 healthy native English speakers (age 18-40) who performed a semantic target detection task in a visual word recognition paradigm while Electro-/Magnetoencephalography (EEG/MEG) data (70/306 channels) were recorded simultaneously (Elekta Neuromag). We compared word categories with strong visual (e.g. sun), auditory (e.g. whistle) and hand-action (e.g. wrench) attributes and hypothesised that while hub modules should be involved in all the three pairwise comparisons, spoke modules for each sensory-motor modality should only be modulated in the two pairwise comparisons that involved their corresponding category. Preprocessing included Maxfilter, band-pass filtering (0.1-45Hz), ICA artefact rejection and epoching. Forward modelling and source estimation were performed on combined EEG/MEG data and based on individual MRIs using boundary element models and L2 minimum norm estimation. Whole-brain networks were reconstructed by parcellating the cortex based on EEG/MEG-adapted Desikan-Killiany Atlas (Farahibozorg et al. 2018) and computing a parcel-parcel (74 x 74) spectral coherence matrix. We reconstructed networks in three time windows (50-250ms, 150-350ms and 250-450ms) and three frequency bands of Alpha (8-12Hz), Beta (13-30Hz) and Gamma (30-45Hz). Thereafter, we identified functional modules for each network using Louvain approach (Blondel et al. 2008) and found consensus modules across subjects, conditions, times and frequencies using a method similar to Lancichinetti & Fortunato (2012). Finally, we conducted statistical comparisons between module-module connectivity matrices of different word categories and applied cluster-based permutations in order to correct for multiple comparisons using a method similar to Zalesky et al. (2010). Key results comprised: (i) Eight consensus modules were identified across time windows, frequencies, conditions and subjects. Interestingly, while the left anterior temporal lobe (ATL), posterior middle temporal (pMTG) and angular gyrus (AG) were clustered together and formed a single temporo-parietal module, the right ATL, pMTG and inferior parietal cortices were identified as stand-alone modules; (ii) Right ATL, right pMTG and right parietal modules were identified as potential semantic hub modules; (iii) Bilateral occipital and left parietal modules were identified as visual spokes, a bilateral central module as a hand spoke and a left temporo-parietal module as well as bilateral frontal cortices as auditory spokes; (iv) We found modulations of three types of intra-/inter-modular connections including hub-hub, spoke-spoke and hub-spoke connectivity. These results provide the first data-driven characterisation of global integrator semantic hubs, spokes and connectivity among them. Additionally, results can further specify the role of the right-hemispheric semantic hubs and unravel temporo-spectral profiles of the involvement of ATL and parietal cortex as integrator semantic hubs.

Topic Area: Meaning: Lexical Semantics

Back