Separate lanes for adding and reading in the white matter highways of the human brain

Abstract

Math and reading involve distributed brain networks and have both shared (e.g. encoding of visual stimuli) and dissociated (e.g. quantity processing) cognitive components. Yet, to date, the shared vs. dissociated gray and white matter substrates of the math and reading networks are unknown. Here, we define these networks and evaluate the structural properties of their fascicles using functional MRI, diffusion MRI, and quantitative MRI. Our results reveal that there are distinct gray matter regions which are preferentially engaged in either math (adding) or reading, and that the superior longitudinal and arcuate fascicles are shared across the math and reading networks. Strikingly, within these fascicles, reading- and math-related tracts are segregated into parallel sub-bundles and show structural differences related to myelination. These findings open a new avenue of research that examines the contribution of sub-bundles within fascicles to specific behaviors.

Fig. 1

Identification of gray matter regions of the math and reading networks and their functionally-defined white matter tracts (fWMT). a FMRI experiment used to define math- and reading-related regions. Subjects viewed morphs between numbers and letters, containing either >80% letter (<20% number) or >80% number (<20% letter) information. At the beginning of each trial, a cue (Read/Add/Color) indicated which task should be performed, then four stimuli of the same morph type appeared for 1 s each, followed by an answer screen presented for 2 s. Subjects indicated their answer with a button press. Identical stimuli were presented across tasks. Trial structure is shown at the bottom. b Gray matter functional regions of interest (fROIs) of the math and reading networks. Green: Reading-related regions were defined based on higher responses in the reading task than other tasks; Blue: Math-related regions were defined based on higher responses in the adding task than other tasks; Orange: Regions that responded more strongly during reading vs. color and adding vs. color tasks. All fROIs were defined using a T ≥ 3 (voxel level) threshold in each participant’s brain. c Example fROIs and their respective fWMTs in axial slices of a representative participant. Blue: Math fROIs. Green: Reading fROIs; lighter shades of blue and green under each fROI: respective gray-white-matter-interface (GWMI) of that fROI. The fiber tracts that terminate at the GWMI of each fROI are shown in pastel colors; the colors of the tracts indicate the main diffusion direction, pink: right/left; light green: anterior/posterior; light blue: superior/inferior. Scale bar in c indicates 1 cm. IFG inferior frontal gyrus, PCS precentral sulcus, SMGr reading fROI in supramarginal gyrus, SMGm math fROI in supramarginal gyrus, STS superior temporal sulcus, ITG inferior temporal gyrus, OTS occipito-temporal sulcus, IPS intraparietal sulcus, lOTC lateral occipito-temporal cortex

Fig. 2

Functionally-defined white matter tracts (fWMT) of reading- and math-related regions. a Six fascicles (AF, SLF, pAF, VOF, ILF, and IFOF) contain >90% of all fWMT of the fROIs identified in the reading task. b The same six fascicles also contain >90% of all fWMT of the fROIs identified in the adding task. c lOTC conjunction fROI shows substantial connectivity with the AF and pAF. In a, b, c: Left: fWMT for each fROI in a representative subject’s left hemisphere. The same subject is displayed in all panels; Fascicles are color coded in accordance with the legend at the bottom. Scale bar in c indicates 2 cm. Middle: Bar graphs showing what percentage of the fWMT is associated with each of the six fascicles. The graph shows the mean across subjects ± SEM. Circles: Individual subjects’ data. Dashed horizontal line: Line is placed at 10%, which was the cutoff used for the schematics in the right columns. Right: Schematic illustration of the fascicles associated with each fROI. The thickness of the lines is derived from the bar graph, showing the relative connectivity weight of each fascicle. IFG inferior frontal gyrus, PCS precentral sulcus, SMGr reading fROI in supramarginal gyrus, SMGm math fROI in supramarginal gyrus, STS superior temporal sulcus, ITG inferior temporal gyrus, OTS occipito-temporal sulcus, IPS intraparietal sulcus, lOTC lateral occipito-temporal cortex, IFOF inferior fronto-occipital fasciculus, ILF inferior longitudinal fasciculus, SLF superior longitudinal fasciculus, AF arcuate fasciculus, pAF posterior arcuate fasciculus, VOF vertical occipital fasciculus

Fig. 3

Pairwise fWMT within and between the reading and math networks. a–d Within-network connections of the reading network. e–h Within-network connections of the math network. i–l Connections of the lOTC conjunction fROI to the math and reading networks. m–o Between-network connections. Left a, e, i, m: Pairwise white matter connections in a representative subject’s left hemisphere. Scale bar in m indicates 1 cm. Second from left b, f, j, n: Dice coefficient (DC) of pairwise connections, mean across subjects ± SEM. The DC quantifies the overlap in the fWMT of both fROIs. X-labels indicate the fROI pairing. Dashed line: Chance level DC estimated from the average connections to out of network fROIs in ventral-temporal cortex that were activated maximally during the color task. Circles: Individual subjects’ data. Asterisks: DC is significantly higher than chance (paired t-test, significance level was Bonferroni adjusted). Second from right in row 1–3 c, g, k: The relative contribution of six fascicles to the pairwise connections (legend at bottom). X-labels indicate the fROI pairing. Right d, h, l, o: Schematic illustration of the pairwise connections. Line thickness is scaled proportionally to the DC; Color indicates the fascicle with the highest relative contribution to the pairwise connection. IFG inferior frontal gyrus, PCS precentral sulcus, SMGr reading fROI in supramarginal gyrus, SMGm math fROI in supramarginal gyrus, STS superior temporal sulcus, ITG inferior temporal gyrus, OTS occipito-temporal sulcus, IPS intraparietal sulcus, lOTC lateral occipito-temporal cortex, IFOF inferior fronto-occipital fasciculus, ILF inferior longitudinal fasciculus, SLF superior longitudinal fasciculus, AF arcuate fasciculus, pAF posterior arcuate fasciculus, VOF vertical occipital fasciculus

Fig. 4

Pairwise connections of the reading and math networks are segregated and parallel in the SLF and the AF. a–c: SLF tracts connecting IFG and SMGr in the reading network and PCS and SMGm in the math network. d–f: AF tracts connecting the IFG and STS in the reading network and the PCS and ITG in the math network. g, h: AF tracts connecting the lOTC conjunction fROI with the IFG in the reading network and the PCS in the math network. a, d, g: Math (blue) and reading (green) tracts of the SLF and AF presented in the left hemisphere of a representative individual subject showing the spatial segregation of these tracts. Scale bar in g indicates 1 cm. b, e, h: Euclidean distance in millimeter (derived from x,y,z coordinates) of all tracts relative to the core (mean) tract, within-network (black) and between-networks (maroon). The distance was calculated across all tracts; the histograms show the distribution of distances across all nodes in each subject ± SEM. Upside-down triangles: Mean distance across nodes and subjects. Asterisk: Mean distances differ significantly, p < 0.05 (paired t-test). c, f, i: Performance of a linear SVM classifying math and reading tracts within the SLF and AF based on their spatial location. Data show mean classification accuracy across nodes ± SEM. IFG inferior frontal gyrus, PCS precentral sulcus, SMGr reading fROI in supramarginal gyrus, SMGm math fROI in supramarginal gyrus, STS superior temporal sulcus, ITG inferior temporal gyrus, lOTC lateral occipito-temporal cortex, AF arcuate fasciculus, SLF superior longitudinal fasciculus

Fig. 5

Tracts in the SLF and the AF that are associated with reading show shorter proton relaxation time (T₁) than those associated with math. a–c: T₁ measurements for SLF tracts connecting IFG and SMGr in the reading network (green) and PCS and SMGm in the math network (blue). d–f: T₁ measurements for AF tracts connecting the IFG and STS in the reading network (green) and the PCS and ITG in the math network (blue). g–i: T₁ measurements for AF tracts connecting the lOTC conjunction fROI with the IFG in the reading network (green) and the PCS in the math network (blue). Left a, d, g: Average T₁ for reading- and math-related tracts in the SLF and the AF. Bar graph shows mean across subjects ± SEM. Circles: Individual subjects’ data. Asterisk: T₁ for math- and reading-related tracts differs significantly, p < 0.05 (paired t-test). Middle b, e, h: Distribution of T₁ values. The histograms show the distribution of T₁ values across all nodes in each subject ± SEM. Right c, f, i: Average T₁ for reading- and math-related tracts along the SLF and the AF. Line graph shows mean across subjects ± SEM. IFG inferior frontal gyrus, PCS precentral sulcus, SMGr reading fROI in supramarginal gyrus, SMGm math fROI in supramarginal gyrus, STS superior temporal sulcus, ITG inferior temporal gyrus, lOTC lateral occipito-temporal cortex, SLF superior longitudinal fasciculus, AF arcuate fasciculus