Individual Differences in Adult Reading Are Associated with Left Temporo-parietal to Dorsal Striatal Functional Connectivity

Abstract

Reading skills vary widely in both children and adults, with a number of factors contributing to this variability. The most prominent factor may be related to efficiency of storage, representation, or retrieval of speech sounds. This phonological hypothesis is supported by findings of reduced activation in poor readers in left hemisphere ventro-lateral prefrontal and temporo-parietal phonological processing regions. Less well explained by phonological theories are reported hyperactivation in prefrontal, striatal, and insular regions. This study investigated functional connectivity of a core phonological processing region, the temporo-parietal junction (TPJ), in relation to reading skill in an adult community sample. We hypothesized that connectivity between TPJ and regions implicated in meta-analyses of reading disorder would correlate with individual differences in reading. Forty-four adults aged 30-54, ranging in reading ability, underwent resting fMRI scans. Data-driven connectivity clustering was used to identify TPJ subregions for seed-based connectivity analyses. Correlations were assessed between TPJ connectivity and timed-pseudoword reading (decoding) ability. We found a significant correlation wherein greater left supramarginal gyrus to anterior caudate connectivity was associated with weaker decoding. This suggests that hyperactivation of the dorsal striatum, reported in poor readers during reading tasks, may reflect compensatory or inefficient overintegration into attention networks.

Keywords: caudate; functional connectivity; reading; temporo-parietal junction.

Figure 1.

TPJ clusters and associated networks for 3-cluster solution. Central panels show TPJ clusters for the left and right hemispheres overlaid on inflated and flattened brains (posterior pointed toward the center). Detailed information about seed clusters is presented in Table 2. Each cluster was used as a seed to identify an associated network across the group. These are shown around the periphery, on cortical surfaces, color-coded to match central clusters. Significant negative associations are shown in dark blue across all panels. The green (angular gyrus) cluster was associated with posterior cingulate, medial prefrontal, and inferior parietal DMN regions. The violet (superior temporal) cluster showed connectivity with cingulate and insular regions of the salience network. The orange (anterior/IPL) cluster was associated with intraparietal sulcus and dorsal prefrontal regions typically linked with the DAN or fronto-parietal network. These are similar to networks identified in Mars et al. (2012). Network overlays are thresholded at P < 0.05 FWE-corrected. LH, left hemisphere; RH, right hemisphere; DMN, default-mode network; DAN, dorsal attention network; IPL, inferior parietal lobule; TPJ, temporo-parietal junction.

Figure 2.

TPJ clusters and associated networks for 8-cluster solution, left hemisphere. Central panels show TPJ clusters for the left hemisphere overlaid on an inflated hemisphere (posterior pointed toward the right). Detailed information about seed clusters is presented in Table 3. Each cluster was used as a seed to identify the associated network across the group. These are shown around the periphery, on cortical surfaces, color-coded to match central clusters. Significant negative associations are shown in dark blue across all panels. These networks resemble the DMN (green), DAN (orange), salience (violet), similar to the 3-cluster solution. Visual (light purple) and STS (blue) networks emerged, as well as networks with connectivity patterns that do not correspond to canonical networks (red, yellow, cyan). Network overlays are thresholded at P < 0.05 FWE-corrected. DMN, default-mode network; DAN, dorsal attention network; STS, superior temporal sulcus; TPJ, temporo-parietal junction.

Figure 3.

Correlations between 3-cluster seeds and phonemic decoding scores. (a) Negative correlation between left SMG/IPL seed and bilateral caudate head ([12, 26, 6], n = 245, Z = 4.78), and right dorsal caudate ([22, 4, 28], n = 142, Z = 4.26); positive association for this cluster with PDE scores in pericalcarine cortex ([4, −72, 4], n = 152, Z = 3.76) in lower panel. Panels are color-coded to match seed regions from Figure 1. (b) Scatter plot of parameter estimates from the peak anterior caudate voxel against phonemic decoding scores, with least squares regression line. IPL, inferior parietal lobule; SMG, supramarginal gyrus; PDE, phonemic decoding efficiency.

Figure 4.

Correlations between left hemisphere 8-cluster seeds and phonemic decoding scores. Significant negative association between IPL to bilateral caudate head ([12, 26, 6], Z = 5.19 and [−18, 28, 0], Z = 4.4, n = 506) and right dorsal caudate ([22, 4, 28], n = 215, Z = 4.42) connectivity and phonemic decoding scores. Panel is color-coded to match TPJ cluster seed regions from Figure 2. TPJ, temporo-parietal junction; PDE, phonemic decoding efficiency; IPL, inferior parietal lobule.

Figure 5.

Correlations between right hemisphere 8-cluster seeds and phonemic decoding scores. Significant positive association between AG to left occipito-temporal ([12, 28, 6], n = 133, Z = 4.5) connectivity and phonemic decoding scores. Panel is color-coded to match TPJ cluster seed regions from Figure 2. AG, angular gyrus; PDE, phonemic decoding efficiency; TPJ, temporo-parietal junction.