Embodied theories of meaning argue that the human conceptual system is implemented in distributed brain networks whose mechanisms are shared by perceptual and action processing. Such meaning representations should be accessible by different perceptual routes, including language and vision. In this context, the current research tests two hypotheses: 1. When human subjects process sentences describing physical events and images of equivalent events, the meaning encoded in these different modalities will be represented in a common network, independent of the input modality. 2. This meaning network will include parietal sensorimotor and multimodal integration areas, and more amodal frontal areas involved in sequential organization of events in time.To test this hypothesis we exposed 19 healthy young subjects to written sentences and visual images of actions in an event-related fMRI protocol. In the same subjects we recorded 61 direction diffusion images in order to reconstruct their white matter tracts.
Using SPM5, conjunction analysis of areas that were active simultaneously for sentence and image processing revealed a distributed fronto-temporo-parietal network. Frontal areas included inferior frontal gyrus BA47, BA46, BA10. Temporo-parietal activation was observed in the medial temporal gyrus BA22, BA37 extending into BA39,BA21 and in the inferior parietal lobe BA40.
For the DTI analysis, we used AFNI to compute diffusion tensors, and DTI-Query to generate and visualise deterministic tracts. We placed seed volumes within the region of principal activation in the parietal activation site and traced the white matter pathways connected with and traversing this area. In each subject, the MNI seed was projected into native space where the tractography was performed. The resulting tracts were projected back into a common space. A group map of white matter tracts revealed a trifurcating network, with a ventro-lateral component reaching into the temporal pole (by way of the inferior longitudinal fasciculus), a medial component reaching into the ventral frontal pole (by way of the inferior fronto-occipital fasciculus), and a dorsal component reaching into the premotor cortex (by way of the superior longitudinal fasciculus).
These data are consistent with the hypothesis that independent of sentence vs. image input, aspects of meaning is represented in a common network including a parietal sensorimotor simulation system interacting with higher level planning and temporal organization functions of the frontal pole and object semantics in the temporal pole.