Cerebral cortical mechanisms of copying geometrical shapes: a multidimensional scaling analysis of fMRI patterns of activation.

We used multidimensional scaling (MDS) to characterize the integrative neural mechanisms during viewing and subsequently copying nine geometrical shapes. Human subjects initially looked at a central fixation point ("rest" period), then looked at a geometrical shape ("visual" period) which they copied without visual feedback ("copying" period). BOLD signal was recorded from voxels in 28 cortical areas (14 from each hemisphere) using a 4 Tesla magnet. For each voxel, signal ratios of "Visual versus Rest" (VR), and "Copy versus Visual" (CV) were calculated and used to construct two sets of Euclidean distance dissimilarity matrices for the nine shapes, with separate matrices defined for each region of interest (ROI) across subjects. The relations of perceptual and motor aspects of the shapes to MDS dimensions and specific ROIs were assessed using stepwise multiple regressions. The optimal individually scaled (INDSCAL) solutions were 2-dimensional. For the VR condition, MDS dimensions were significantly associated with the presence of crossing in a shape (Dimension 1), and with perimeter, height, cycles, peak segment speed, and horizontal symmetry (Dimension 2). ROIs most prominently associated with these dimensions essentially comprised the medial frontal lobe bilaterally, the inferior frontal gyrus bilaterally, and the left intraparietal sulcus (Dimension 1), and visual areas, including the calcarine sulcus and cuneus bilaterally (Dimension 2). These results document the expected involvement of visual areas and support the hypothesis advanced on the basis of previous findings (Lewis et al. 2003a) that a motor rehearsal of the upcoming shape copying is occurring during this visual presentation period. For the CV condition, practically one motor feature (number of segments drawn) dominated both dimensions, with a secondary engagement of horizontal symmetry in Dimension 1. The right postcentral gyrus, right intraparietal sulcus, right superior parietal lobule and right inferior parietal lobule contributed mostly to Dimension 1; the superior frontal gyrus bilaterally, right middle frontal gyrus, left postcentral gyrus, left inferior parietal lobule contributed mostly to Dimension 2; and the left superior parietal lobule and left intraparietal sulcus contributed to both dimensions approximately equally. CV BOLD activation of ROIs contributing to Dimension 1 (or to both dimensions) was significantly associated with the number of shape segments drawn. Since the direction of movement differs in successively drawn shape segments, the number of segments (minus one) equals the number of changes in the direction of movement. We conclude that this fundamental spatial motor aspect of drawing geometrical shapes is the critical variable, independent of the particular shape drawn, that dominates cortical activation during copying.