Auditory and visual connectivity gradients in frontoparietal cortex

Abstract

A frontoparietal network of brain regions is often implicated in both auditory and visual information processing. Although it is possible that the same set of multimodal regions subserves both modalities, there is increasing evidence that there is a differentiation of sensory function within frontoparietal cortex. Magnetic resonance imaging (MRI) in humans was used to investigate whether different frontoparietal regions showed intrinsic biases in connectivity with visual or auditory modalities. Structural connectivity was assessed with diffusion tractography and functional connectivity was tested using functional MRI. A dorsal-ventral gradient of function was observed, where connectivity with visual cortex dominates dorsal frontal and parietal connections, while connectivity with auditory cortex dominates ventral frontal and parietal regions. A gradient was also observed along the posterior-anterior axis, although in opposite directions in prefrontal and parietal cortices. The results suggest that the location of neural activity within frontoparietal cortex may be influenced by these intrinsic biases toward visual and auditory processing. Thus, the location of activity in frontoparietal cortex may be influenced as much by stimulus modality as the cognitive demands of a task. It was concluded that stimulus modality was spatially encoded throughout frontal and parietal cortices, and was speculated that such an arrangement allows for top-down modulation of modality-specific information to occur within higher-order cortex. This could provide a potentially faster and more efficient pathway by which top-down selection between sensory modalities could occur, by constraining modulations to within frontal and parietal regions, rather than long-range connections to sensory cortices. Hum Brain Mapp 38:255-270, 2017. © 2016 Wiley Periodicals, Inc.

Keywords: auditory; connectivity; frontoparietal cortex; functional; functional magnetic resonance imaging; gradients; resting-state; structural; tractograpy; visual.

Figure 1

Schematic of pipeline for relative auditory and visual structural and functional connectivity analysis.

Figure 2

Visual–auditory connectivity differences in frontoparietal cortex. (A) Structural connectivity was assessed by counting the number of probabilistic tractography streamlines (“streams”) that reached the auditory and visual targets cortices. Relative structural connectivity was assessed by comparing the visual and auditory maps (V/A; S_VA). This analysis revealed marked connectivity differences, where dorsal regions were biased toward visual connections, while ventral regions showed an auditory bias. (B) Functional connectivity was assessed using multivariate methods (see Fig. 1). A similar dorsal to ventral gradient was observed in the ratio image (F_VA), with dorsal regions favoring visual, and ventral regions favoring auditory targets. (C) The relative Euclidian (straight line) distance to the targets was regressed out of the structural data to test for a potential confound. This analysis still revealed distinct foci of visual and auditory bias in dorsal and ventral regions respectively. (D) The structural results were regressed out of the functional results to test whether the structural pattern could explain the observed functional biases. Residual functional connectivity differences were observed in both prefrontal and parietal cortices.

Figure 3

Connectivity gradients in prefrontal cortex. Functional and structural connectivity analyses both revealed a graded transition across the prefrontal cortices. In general, dorsal regions favored visual and ventral regions favored auditory targets. In this figure, connectivity values along the medial‐lateral axis were projected onto the 2D grid displayed; therefore each grid element represents the average of a vector of voxels.

Figure 4

Connectivity gradients in parietal cortex. Structural gradients followed a dorsal–ventral transition similar to the functional results (see Fig. 3) with a locus of visual bias in SPL. Functional gradients were more complex, with evidence for a posterior–anterior (as well as dorsal–ventral) gradient in the right hemisphere. In contrast, the left parietal lobe showed a posterior–anterior but not dorsal–ventral gradient. In this figure, connectivity values along the medial–lateral axis were projected onto the 2D grid displayed; therefore each grid element represents the average of a vector of voxels.

Figure 5

Gradient regression results. To quantify the imaging results, a linear regression was used to test whether the position of each voxel along a horizontal (anterior to posterior) and vertical (ventral to dorsal) axis explained the structural and functional connectivity patterns (top row). Functional and structural gradients were strongly detected in the prefrontal cortex (red colors). These gradients represented a visual–auditory transition in the dorsal–ventral and posterior‐anterior directions, respectively. In the parietal lobe (blue colors), similar structural gradients were observed in the dorsal–ventral direction. However in the horizontal axis the gradient followed an opposite pattern to the prefrontal cortex. We also repeated the analysis while covarying out the relative distance to targets (middle row), and covarying out the structural results (S_VA) from the functional data (lower row).