Discourse Processing in Autism: Effects of Linguistic Connectives

Background: Previously we reported a study examining the intersection of Theory of Mind (ToM) processing with narrative discourse comprehension in autism (Mason et al, Neuropsychologia, 2008). The results of that study suggested that although the ToM network was activated in individuals with autism, it was inefficient. Whereas the control participants selectively activated a ToM network only when appropriate, the autism group processed all inferences similarly; they used the ToM network even when no additional ToM processing was required. The cortical system in autism appears to meet the cognitive challenge created by discourse processing by indiscriminately engaging RH language and ToM areas. This previous study did not have any explicit linguistic cues forcing the reader to causally link the presented information. We were interested in the further study of discourse processing in high-functioning individuals with autism.

Objectives: To investigate whether the brains of individuals with autism always responded in a similar manner during discourse processing or whether the processing was sensitive to two different types of textual information: 1) linguistic cues that explicitly invited integration of information; and, 2) whether the inference was moderately or distantly connected to the textual information.
Methods: Participants complete a reading task presented via computer while lying in a 3T Siemens Allegra magnetic resonance imaging scanner. Stimuli consist of 40 reading passages, 10 in each of 4 conditions in a 2 X 2 design: moderately- vs. distantly-related inference and the presence or absence of one of four connectives (thus, hence, consequently, and therefore). Participants are high-functioning adults with autism and age and IQ-matched controls with Full Scale and Verbal IQs ³ 80. Standard fMRI analysis measures are being performed including group comparisons of regions and significant levels of activation in the experimental conditions. Functional connectivity differences by group and condition will also be examined.

Results: The behavioral, functional imaging, and functional connectivity data from approximately 25 individuals with HFA and 25 age and Verbal IQ-matched controls will be presented.

Conclusions: The predictions we are testing are that the autism group will activate ToM network and language processing networks during the moderately-related discourse processing condition, indicating some integration of information. However, these networks may be recruited less during the distantly-related condition as the autism group treats the information as unrelated text. Explicit cuing may change this processing pattern, forcing integration in the distantly-related passages that was not previously observed. Processing of explicit cues may also be reflected in an increase in frontal-posterior functional connectivity as compared to the condition without linguistic cues.