Table 1.
Experiment-1: School-aged sample (n = 55).
Demographic and behavioral performance and their correlations with the discrepancy index (DiscInd), decoding (DECODE) and reading comprehension skills (COMP).
Table 2.
Experiment-2: Demographic and behavioral performance at Time 1 and Time 2 and their correlations with the discrepancy index (DiscInd) at Time 2.
Fig 1.
Scatter-plot of decoding (DECODE; WRMT-WA SS) and reading comprehension scores (COMP; WRMT-PC SS).
Participants from Experiment-1 are displayed in red. DECODE and COMP measures showed a significant positive correlation (r = 0.67, p < 0.001). The discrepancy index (DiscInd; i.e., COMP “minus (-)” DECODE) was greater than 10 SS (Standard Score) for 27.3% of the participants (triangles) i.e., they were better at comprehending than decoding. Discrepancy was in the opposite direction and negative for 21.8% of the participants, i.e., less than -10 SS and they were better at decoding than comprehending (squares). Participants that did not belong to either group are those without a large discrepancy between DECODE and COMP (50.9% of the participants, circles). Participants from Experiment-2 at Time 2 are displayed in green. DiscInd was greater than 10 SS for 20.9% (triangles), less than -10 SS for 41.9% (squares), and non-discrepant (between -10 and 10 SS) for 37.2% of the participants (circles).
Fig 2.
The neural correlates of decoding, reading comprehension and DiscInd.
A. Experiment-1. Association between GMV and behavioral measures of decoding (DECODE) and reading comprehension scores (COMP). A-1. Association between GMV with COMP (blue) and DECODE (pink). A-2. Association between GMV with COMP partialling out DECODE (cyan). There were no regions that correlated between GMV and DECODE when COMP was partialled out. B. Experiment-1 and Experiment-2. Association between GMV and DiscInd, i.e., a score subtracting DECODE from COMP. B-1. The cluster that shows association between GMV and DiscInd in school-age children from Experiment-1 whole brain analysis is shown in red. In Experiment-2, the cluster that shows association between GMV of beginning readers and DiscInd 3 years later within the cluster obtained from Experiement-1 using ROI small volume correction is shown in green. B-2. Mean values extracted from the clusters in B-1 are plotted as scatterplots to show that the positive relationships between DiscInd scores and left dorsolateral prefrontal cortex (DLPFC) GMV are not driven by children with dyslexia in Experiment-1 (red full circle) or from the children with a family history of dyslexia in Experiment-2 (green full circle).
Table 3.
Experiment-1: Brain regions where gray matter was associated with the discrepancy index (DiscInd).
Fig 3.
Reverse inference results in Experiment-3.
A. Resting state functional connectivity (rsFC) from the left DLPFC cluster’s center of gravity (COG) as seed (MNI x = -40, y = 38, z = 16) using the 1000 Functional Connectome data using threshold of r > 0.30 B. The voxel-by-voxel values from the resulting rsFC map and reverse inference map for each term were correlated. The resulting correlation values are plotted against each key term with threshold of r > 0.1.
Fig 4.
Description of the Reading System Framework and contributions of the current study to this Framework.
The Reading System Framework by Perfetti is represented in blue lines. It emphasizes the lexical component that mediates decoding, reading comprehension and the integration between language knowledge with reading processes. Current findings are represented in red lines. They indicate that the left DLPFC network, and possibly by inference, cognitive control and working memory, influence reading comprehension (single-lined red arrow), also strongly modify the strength of the relation between the two (double-lined red arrow).