A core issue in psycholinguistic research is what the online processes are by which we combine language input and our background knowledge to construct the meaning of a message. We investigate this issue in the context of reading. To build a coherent and correct mental representation of a text readers monitor incoming information for consistency with the preceding text and with their background knowledge. Prior studies have not distinguished between text-based and knowledge-based monitoring, therefore it is unclear to what extent these two aspects of text comprehension proceed independently or interactively. We addressed this issue in a contradiction paradigm with coherent and incoherent versions of texts. We combined behavioral data with neuroimaging data to investigate shared and unique brain networks involved in text-based and knowledge-based monitoring, focusing on monitoring processes that affected long-term memory representations. Consistent with prior findings, behavioral results indicate that text and background knowledge each have a unique influence on processing. However, neuroimaging data suggests a more nuanced interpretation: Text-based and knowledge-based monitoring involve shared and unique brain regions, as well as regions that are sensitive to interactions between the two sources. It appears that the (d)mPFC and hippocampus-which are important for the influence of existing knowledge on encoding processes in nonreading contexts-are particularly involved in knowledge-based monitoring. In contrast, the right IFG is primarily involved in text-based monitoring, whereas left IFG and precuneus are implicated in integration processes. Furthermore, processes during reading affect recall of information (in)consistent with prior text or background knowledge.
