Organizing principles of pulvino-cortical functional coupling in humans

Abstract

The pulvinar influences communication between cortical areas. We use fMRI to characterize the functional organization of the human pulvinar and its coupling with cortex. The ventral pulvinar is sensitive to spatial position and moment-to-moment transitions in visual statistics, but also differentiates visual categories such as faces and scenes. The dorsal pulvinar is modulated by spatial attention and is sensitive to the temporal structure of visual input. Cortical areas are functionally coupled with discrete pulvinar regions. The spatial organization of this coupling reflects the functional specializations and anatomical distances between cortical areas. The ventral pulvinar is functionally coupled with occipital-temporal cortices. The dorsal pulvinar is functionally coupled with frontal, parietal, and cingulate cortices, including the attention, default mode, and human-specific tool networks. These differences mirror the principles governing cortical organization of dorsal and ventral cortical visual streams. These results provide a functional framework for how the pulvinar facilitates and regulates cortical processing.

Fig. 1

Functional distinctions within the pulvinar. a The ventral half of the pulvinar showed greater responses to stimulation of contralateral visual space (left, p < 0.05, FDR-corrected, n = 5). Anatomical extent of the left hemisphere pulvinar in posterior-most slice outlined in dark red. Graphs show the group average (horizontal black line), standard deviation (vertical black lines), and 95% confidence interval (shaded area) for laterality and attention indices. Grey circles illustrate individual subjects. The ventral pulvinar (vPul) showed greater laterality of responses (middle, t(4) = 5.96, p = 0.004, n = 5), while the dorsal pulvinar (dPul) showed greater attentional modulation (right, t(4) = 3.16, p = 0.034, n = 5). b Repeated presentations of intact movie stimuli evoked consistent responses in both the ventral and dorsal pulvinar. Only the dorsal pulvinar showed greater consistency to repeated presentations of intact vs. temporally-scrambled movies (r > 0.15, n = 11). c A posterior medial portion of the ventral pulvinar responded preferentially to face vs. scene stimuli in the voxel-wise contrast (left, p < 0.05, FDR corrected, n = 16) and in the ROI analysis (right, z = 3.10, p = 0.002). Graph conventions same as a

Fig. 2

Organization of pulvino-cortical functional coupling. a 2D plot of multidimensional scaling on the group average (n = 13) pattern similarity of pulvino-cortical functional coupling between 39 cortical areas. Only ipsilateral correlations were considered. Data clustering yielded 2 groups. One cluster (blue) mainly contained occipital and temporal cortical areas associated with the ventral visual stream. The other cluster (red) mainly contained frontal and parietal cortical areas associated with the dorsal visual stream. Line thickness and color coding reflects similarity strength between areas. Surface figures illustrate all cortical areas tested; fill color matched to assigned cluster. b The difference between the two cluster’s pulvino-cortical connectivity maps segmented the pulvinar into dorsal and ventral sections. Colors correspond to the blue and red clusters shown in (a), threshold at an p < 0.05 FDR-corrected. c Same analysis as in (b), but from calculating pulvino-cortical connectivity maps using the Human Connectome Project’s 180 cortical areas as the cortical correlation profile

Fig. 3

Consistency of pulvino-cortical coupling between hemispheres and across individuals. a Multidimensional scaling for left and right hemisphere pulvino-cortical functional connectivity. Procrustes analysis was performed to align MDS of the left hemisphere to the right hemisphere. Lines illustrate the distances between areas matched between hemispheres. Dots were color-coded based on clustering performed on each hemisphere separately. The only difference in clustering between hemispheres was area pSTS (i.e., the only blue and red dots linked by a line). b MDS for each subject’s pulvinar connectivity (averaged across hemisphere). Procrustes analysis was performed to align MDS of each subject to the group average. Small black dots illustrate the locations of individual subject areas. Lines illustrate distances of between individual subjects and the group average for matched areas. Dots were color-coded based on clustering performed on the group average

Fig. 4

Relationship between cortical distance and pulvino-cortical coupling. (left) Half matrices showing pairwise cortical distances and overlap in pulvino-cortical functional connectivity for all 39 cortical areas. Area labels are colored based on clustering in Fig. 2. Thin vertical black lines separate each block of four areas. (right) Scatter plot of cortical distance vs. overlap of pulvino-cortical connectivity for area pairs that were both part of the ventral pulvinar cluster (blue), dorsal pulvinar cluster (red), or for area pairs between clusters (grey). The correlation between cortical distance and pulvino-cortical connectivity overlap was significant for areas within the occipital-temporal cluster (r = −0.63, p < 0.0001), but not within the frontal-parietal (r = −0.10, p > 0.10). For all area pairs, Dice’s coefficient, 2 |A ∩ B| / (|A| + |B|), was calculated

Fig. 5

Functional coupling with dorsal pulvinar. a Group average (n = 13; p < 0.01 from one-sample t test across subjects) correlations within the dorsal pulvinar for three cortical seed areas (IPS2 (dark green), FEF (light green), and IPL (blue)). Surface figures illustrate the cortical correlation patterns from seed in IPS2 in a single subject. Circle colors correspond to area labels for functional coupling with dorsal pulvinar. b Three-dimensional plot of cortical connectivity map peaks in the dorsal pulvinar. Peak coordinates from the left hemisphere were reflected across the midline and averaged with the right hemisphere. Spheres depict the 3D spatial location of each area’s peak connectivity within the dorsal pulvinar. 2D projections of each data point are plotted on the walls and floor of the graph. Pulvino-cortical connectivity within the dorsal pulvinar was clustered into three groups. Within the largest cluster (green) containing the IPS1-4, FEF, and IFS, connectivity reflected cortical distance with IPS1/2 located most posteriorly, followed by IPS3/4 and then FEF and IFS located anterior. The peak connectivity for SPL, which is located medial to the IPS maps cortically, was located medial to the IPS1-4 peak correlations in the pulvinar. A second cluster (red) contained tool-selective regions in anterior parietal and lateral temporal cortex as well as IPS5. The third cluster (blue) contained medial, lateral, and inferior parietal areas as well as the precuneus. Surface figures illustrate the outlines of all cortical areas tested and color filled areas correspond to those included in the clustering of the dorsal pulvinar

Fig. 6

Functional coupling with ventral pulvinar. a Group average correlation maps (n = 13; p < 0.01 from one-sample t-test across subjects) are shown for four cortical areas defined based on their functional specialization: occipital face area (orange), fusiform face area (cyan), anterior temporal face area (blue), and parahippocampal place area (magenta).  Correlations were strongest within the ventral and posterior-most portions of the pulvinar for each area. Surface figure illustrates the localization of face- and scene-selective regions in an individual subject. Circle color code matches area labels on right. Each cortical region showed greater activity for face vs. scene stimuli (p < .01, FDR-corrected). b Three-dimensional plot of the peak occipital and temporal pulvino-cortical connectivity within the ventral pulvinar. Spheres depict the 3D spatial location of each area’s peak connectivity within the ventral pulvinar. 2D projections of each data point are plotted on the walls and floor of the graph. Clustering revealed 2 groups that differentiated lateral (red) and medial (blue) portions of the ventral pulvinar. c Same plot as in (b), but for a subset of ventral areas. Clustering of connectivity within the ventral pulvinar revealed 4 groups