The neural basis of automatic and controlled semantic cognition: Spatiotemporal dynamics revealed by magnetoencephalography and brain stimulation

This thesis investigated the neural basis of different aspects of semantic cognition using magnetoencephalography (MEG) and chronometric transcranial magnetic stimulation (cTMS). The primary aim was to examine the evidence for a potential functional dissociation between ATL and pMTG; (i) in terms of how they support semantic retrieval, and (ii) whether they support different types of semantic knowledge. Chapter 2 investigated the predictions of the Controlled Semantic Cognition framework (CSC - Lambon Ralph et al., 2017). According to this view, ATL and pMTG within the temporal lobe have distinctive roles in semantic cognition: the ATL is argued to correspond to a semantic store and in this way shows processing advantages for patterns of retrieval that are highly coherent with the structure of long-term conceptual knowledge, whereas pMTG is implicated in more controlled aspects of retrieval that might promote the accessibility of weaker associations that are not dominant within the ATL representations. The brain’s oscillatory response to strongly- and weakly-related word pairs was characterised in left anterior temporal lobe (ATL) and left posterior temporal gyrus (pMTG). ATL showed a larger response to strongly-related words, while pMTG responded more strongly to weakly-related words, suggesting these regions contribute to automatic/coherent and controlled semantic retrieval respectively. The chapter also investigated the causal role of ATL and pMTG in automatic and controlled retrieval using cTMS. Results showed disruption for weak associations in pMTG at an early time point, and slightly later disruption for strong associations in ATL. Chapter 3 further contrasted the CSC framework with the dual hub view, which alternatively proposes that (i) ATL and pMTG (plus angular gyrus; AG) represent different types of semantic information – with taxonomic links in ATL (e.g., DOG and RAT) and thematic relationships in pMTG/AG (e.g., DOG and BONE). According to the CSC framework, the functional division between these brain regions is better characterised by automatic/coherent retrieval in ATL/AG and more controlled retrieval in pMTG. Results indicated that ATL, pMTG and AG all responded to both taxonomic and thematic relationships; thus, I did not observe empirical support for the dual hub view. There was greater engagement of ATL for strong associations and pMTG for weaker associations, consistent with the CSC view. These findings together suggest that pMTG may play a role in maintaining a semantic context and detecting situations in which controlled retrieval processes may need to be engaged. In contrast, ATL may support patterns of coherent semantic retrieval between highly-related concepts. Thus, the findings of this thesis help to elucidate the specific roles of these regions in the semantic network.