Text Comprehension after Brain Injury: An fMRI Study of Inference Processes in Patients with Anterior Temporal or Fronto-Medial Lesions

Text comprehension requires the continuous integration of utterances with both general world knowledge and the prior discourse context. These inference or coherence building processes have been shown to consistently activate the left fronto-medial cortex and anterior temporal regions bilaterally (Mazoyer et al., 1993; Ferstl & von Cramon, 2001, 2002). However, in behavioral tests, there was evidence for an inferencing deficit after left frontal lesions only, but not after anterior temporal lesions (Ferstl, Guthke & von Cramon, 2002). The goal of the present study was to characterize the network necessary for higher level language processing by describing the changes caused by lesions to one of the key components. We used event-related whole head fMRI at 3 Tesla to directly study cerebral blood flow during an inference task in brain damaged patients. 80 sentence pairs were auditorily presented over head phones. The participants were instructed to judge using a YES/NO response whether there was a pragmatic, content-based connection between the sentences. The brain-injured participants were selected from a pool of former patients of the outpatient clinic for cognitive neurology at the University of Leipzig. For the fMRI analysis participants with more than 25% errors were excluded. Data of 16 patients were included in the analyses: 10 patients with anterior temporal lesions (n=6 left temporal, n=4 right temporal), and 6 patients with a likely involvement of the FMC. These latter patients were closed-head injury patients whose anatomic MRI scan showed bifrontal contusions or evidence for microbleeds along the superior frontal gyrus. The fMRI data were first analyzed for the whole group. Comparing all language trials to the control task (listening to pseudo-word sentences), we found very similar activation patterns as in the previous studies with healthy control participants. As expected, the FMC group had bilateral involvement of the anterior temporal lobes, whereas for the temporal groups this region was active in the contra-lesional hemisphere only. Furthermore, the patient group showed large regions of activation in the left-sided prefrontal cortex, in particular in the pars triangularis of the inferior frontal gyrus and near the junction of the inferior frontal sulcus and the precentral sulcus. Activation in the fronto-medial wall depended on the patient's performance level, confirming its role for inference processes.