Seeking a bridge between language and motor cortices: a PPI study

Abstract

The relation between the sensorimotor cortex and the language network has been widely discussed but still remains controversial. Two independent theories compete to explain how this area is involved during action-related verbs processing. The embodied view assumes that action word representations activate sensorimotor representations which are accessed when an action word is processed or when an action is observed. The abstract hypothesis states that the mental representations of words are abstract and independent of the objects' sensorimotor properties they refer to. We combined neuropsychological and fMRI-PPI connectivity data, to address action-related verbs processing in neurosurgical patients with lesions involving (N = 5) or sparing (N = 5) the primary motor cortex and healthy controls (N = 12). A lack of significant changes in the functional coupling between the left M1 cortex and functional nodes of the linguistic network during the verb generation task was found for all the groups. In addition, we found that the ability to perform an action verb naming task was not related to a damaged M1. These data showed that there was not a task-specific functional interaction active between M1 and the inferior frontal gyrus. We will discuss how these findings indicate that action words do not automatically activate the M1 cortex; we suggest rather that its enrolment could be related to other not strictly linguistic processing.

Keywords: PPI; connectivity; healthy controls; language network; sensorimotor cortex; tumor patiens.

Figure 1

(A) Whole brain analysis results for the group of patients with lesions involving M1 (M1−) performing hand clenching movements vs. rest and (B) activation maps for the group of patients with lesions sparing M1 (M1+) performing lip movements vs. rest. The two types of movement have been selected in this image to highlight the close location of the M1− to the hand representation area, as evidenced by the activation cluster, and the vicinity of the M1+ to the lip representation area, as evidenced by the activation cluster. Data were thresholded at p < 0.05 cluster corrected (Z > 2.3). (C) Overlapping of the ROIs drawn on the patients' lesions after normalization (in blue for the M1+ and in red for the M1−) and of the mask created by using the Anatomy Toolbox and the maximum probability maps (MPS) of the left and right M1 (in green) and of the left and right Pm cortex (in pink).

Figure 2

(A) Activation elicited by the verb generation task vs. rest (p < 0.05 cluster corrected (Z > 2.3) for healthy controls (upper row), for patients with lesions involving M1 (M1−, middle row) and for patients with lesions sparing M1 (M1+, lower row). (B) The image shows the activation maps generated by the PPI analysis. Brain regions showing significant increases of connectivity to the left Broca's area during verb generation task for healthy controls and for M1− are shown. For M1+ PPI analysis didn't find any area with a significant activation. (C) Overlapping of the seed regions (Broca's and M1 area) on a rendered 3D template.

Figure 3

(A) Activation maps for whole-brain GLM analysis related to verb generation task vs. rest. The contrast data between groups are presented at a threshold of p < 0.05, cluster corrected for Z > 2.3. (B) Average BOLD signal time-course extracted from the motor seed ROI (left) and from the Broca's seed ROI (right) during the verb generation task are displayed for all the groups separately (M1−, lesions involving the primary motor cortex; M1+, lesions sparing M1; HC, healthy controls).