October 2017: Neural Functional Connectivity Predicts Reading Performance

Overview

The set of brain regions (or the brain “network”) thought to be critically involved in reading is fairly well characterized at this point. Reading studies find that brain activity for reading is, like language, stronger on the left side. There are two pathways (one high and one low) from visual regions (at the back of the brain) forward to the speech and language regions behind the ears and above the left temple. This popular dual-route model of reading is supported by evidence that early readers use a phonology-based strategy that favors the higher pathway from vision to speech, whereas skilled readers favor the lower pathway along the bottom of the cortex (e.g., Church, Coalson, Lugar, Petersen, & Schlaggar, 2008). However, despite the widespread agreement on these primary pathways, there is growing interest in additional brain regions that may play important roles in both learning to read and skilled reading.

Alcauter et al. (2017) are particularly interested in the striatum’s role in reading and learning. The striatum is a collection of cellular structures in the center of the brain that are critical for motor skills, actions, and reward processing and thus may be important for learning to read. To assess the role of the striatum in reading, Alcauter and colleagues analyzed a relatively new type of functional MRI brain data—resting state fMRI. These scans are easy to collect because the participants lie still with their eyes closed, or look at a symbol on a screen, and do not press any buttons or make any decisions. However, these scans can be difficult to collect because the job of “think about whatever you want but hold very still and stay awake” is challenging for child participants, as they do not have anything to distract them during the scan. These data are also sensitive to motion—false relationships between brain regions emerge with higher participant motion during the MRI scan. However, the data that emerge from this type of MRI scan reveal what parts of the brain fluctuate in harmony with each other while at rest, or the brain’s “functional connectivity.” This paper is novel in its attempt to relate multiple tests of reading ability to the resting-state scan data. The authors also used the brain data itself to identify a reading-related brain network for the participants, instead of applying the established network from the literature. This approach is called probabilistic independent components analysis (ICA), and it identifies statistically independent subcomponents of the MRI signal, like identifying individual speakers at a cocktail party from an audio recording. This approach is thus open to including regions not typically included in the classic reading network.

Alcauter and colleagues analyzed data from child participants in a larger study of brain networks and academic skills in Queretaro, Mexico. This study had 60 children (25 boys, 35 girls), 6.7 to 9.8 years old (average 8.46 years old), with resting state fMRI scans and good-quality brain pictures. No multilingual participants were included, so all were native Spanish speakers. All participants completed the Evaluacion Neuropsicológica Infantil (ENI), which tests reading abilities in Spanish speakers, as well as executive and cognitive functions. Accuracy, comprehension, and speed were all assessed by the ENI, and a combined reading score was converted to a standard score based on a Spanish-speaking Latin American reference sample.

Key Findings

ICA analysis identified 32 brain networks, and one network contained 75% of literature-derived reading regions.

Alcauter and colleagues used the ICA approach to identify sets, or networks, of regions of the brain that were highly correlated with each other at rest. Among the resulting networks of brain regions was one that aligned well with previous studies of reading. The regions found in this network were primarily in the left hemisphere and included most of the regions found across studies of reading. They also found some regions thought to be important for executive functions, or control over goal-oriented behaviors. These regions were part of the same reading network and included frontal brain regions, the anterior insula, the striatum, and parietal cortex on both sides. These results suggest that task fMRI studies (studies where children or adults actively read or rhyme during the MRI scan) may need to consider additional regions that support reading or reading development. However, because the authors did not directly compare their results to the established reading network, the authors cannot be certain that the network that emerged is important for reading per se, as it was generated from data where children lie quietly at rest.

Reading speed was positively correlated with functional connectivity in this network.

The authors found that the strength of functional connectivity in many brain regions, including the striatum, within this identified network related to reading speed, consistent with this network being important to reading. This finding is a novel result, as many imaging studies do not link brain activity to tests of actual reading behaviors, and many reading behavior studies lack brain data. These results were found in the left hemisphere and included classic reading regions such as the frontal region above the left temple important for reading and speech, but also regions not widely implicated in reading, including striatum regions. The involvement of the striatum in reading speed adds to a growing number of studies linking the striatum to reading and reading difficulty. Relationships between the striatum and frontal regions, in particular, may drive important motor control, sensory, and learning processes of the brain during the reading process.

One limitation of this paper is that the ICA analysis had small amounts of data per person to use for statistical analysis, perhaps resulting in poor network identifications. Further, the authors did not see significant relationships with reading comprehension or reading accuracy, calling into question the specificity to reading of the reading speed measure (it could be capturing a more general speed of processing measure, which would be less interesting).

Interpretations of Study Findings and Implications for Future Research

The purpose of Alcauter and colleagues’ study was to identify additional brain regions important for reading via their association with reading-related regions at rest. By identifying subcortical regions, like the striatum, and frontal cortex regions as important in a broader network of the reading brain at rest, the authors lend support for targeting these regions in studies of reading intervention and in broadening our understanding of the brain regions important for early reading development.

Future research should replicate this study with more data per person and test for relationships with reading scores beyond just reading speed. Further, the authors do not find results in the cerebellum, which is part of motor processing with the striatum, and they suggest that the cerebellum may be more involved at older ages. This could be testable in the large public resting-state databases available through the National Institutes of Health (e.g., the ABCD project) and other repositories. The Texas Center for Learning Disabilities’ resting-state fMRI data for fourth-grade children could potentially be used to replicate this investigation of striatal relationships to other regions in struggling readers.  

References

Alcauter, S., García-Mondragón, L., Gracia-Tabuenca, Z., Moreno, M. B., Ortiz, J. J., & Barrios, F. A. (2017). Resting state functional connectivity of the anterior striatum and prefrontal cortex predicts reading performance in school-age children. Brain and Language, 174, 94–102.

Church, J. A., Coalson, R. S., Lugar, H. M., Petersen, S. E., & Schlaggar, B. L. (2008). A developmental fMRI study of reading and repetition reveals changes in phonological and visual mechanisms over age. Cerebral Cortex, 18(9), 2054–2065.