IDENTIFYING COMPONENT DISCOURSE PROCESSING FROM THEIR fMRI TIME COURSE SIGNATURES

Identifying Component Discourse Processes Reading—from works to multiple texts (pp. 147-159). Reading is a complex task and it is accomplished by several interactive levels of skills, starting from phonological, lexical and syntactic analyses and extending to propositionalization (Perfetti & Britt, 1995) and intra-sentence integration/discourse-levels (Mason & Just 2004), which involve cognitive processing at higher/different levels such as memory and inference generation (e. among others. We suggest that there is evidence to support the characterization of the discourse-level processes in which a reader constructs an understanding of the meaning of larger units of text using the products of lexical and syntactic processing by utilizing regions across the cortex. We will refer to this set of processes as discourse comprehension Network theories of discourse comprehension (e.g., Ferstl et al., 2008; Schmalhofer & Perfetti, 2007) typically specify functional roles for different brain regions by identifying which regions activate during high level comprehension tasks such as reading passages that require inferences (Jung-The general conclusion from these studies is that there are many regions that interact to support discourse comprehension. In fact, Perfetti and Frishkoff (2008, p. 172) specify in their review of the neuroimaging literature that "Of course, there should be no surprise to learn that the components of text processing are distributed rather than localized. They depend fundamentally on processes of information encoding, memory, and retrieval, along with basic left hemisphere language processes." Occasionally, a meta-analysis across data sets (Ferstl et al., 2008) will provide insight as to which regions are generalizable beyond task-specific experimental designs. Both of these techniques are excellent ways of specifying the network(s), yet as suggested here, both can leave some things unspecified. This chapter is an attempt to revisit our initial characterization of discourse networks (Mason & Just, 2006) and suggest that there is evidence to support the characterization of cognitive processes which utilize multiple regions across discourse networks. What is presented here is a data mining technique (factor analysis of the time course of activated cortical regions) that offers a novel way of determining network makeup and function. Using the traditional approach, we (Mason & Just, 2006) put forward an integrative conceptual framework for understanding brain function specifically during discourse processing. This framework proposed that discourse comprehension involved, in addition to lexical and syntactic processing, five specialized networks: (1) a coarse semantic processing network (right middle and superior temporal); (2) a coherence monitoring network (bilateral dorsolateral prefrontal); …