Our primary aim was to examine similarities and differences in the RSFC–behavior relationships exhibited by typically developing children and adults. Analyses revealed both common (i.e., present in both groups) and divergent (i.e., different between groups) relationships between RSFC and reading competence, which were independent of general intelligence.
Common RSFC–reading competence relationships
Positive relationships between reading competence and RSFC within functional circuitry were associated with the PCG and IFGop seeds in both child and adult groups. The first common RSFC–behavior relationship was observed for RSFC between the PCG seed and other motor areas (i.e., the left postcentral gyrus, the supplementary motor area/posterior cingulate cortex, and the right postcentral/precentral gyrus). This finding is congruent with the motor theory of speech perception (Liberman et al., 1967
). The core claim, that speech perception involves access to motor systems, is supported by reports of activation within motor regions during speech sound recognition (Pulvermüller et al., 2006
; Meister et al., 2007
; Pulvermüller and Fadiga, 2010
). Such a close coupling between speech and motor functions also occurs during reading. Longcamp et al. (2005)
demonstrated that PCG was responsive to visually presented letters, but not to pseudo-letters (which cannot be sounded out), even in the absence of requiring motor/speech performance. Given the pivotal role of speech perception in the phonological processing underlying reading development (McBride-Chang, 1996
; Thomson and Goswami, 2008
; Goswami et al., 2011
), our findings suggest that strong functional coupling within motor regions serves as a neural framework for efficient reading, regardless of age group, presumably reflecting automatized articulation in response to written words.
The second common RSFC–behavior relationship was observed for a functional connection linking Broca's and Wernicke's areas (Kelly et al., 2010
). Specifically, reading competence positively correlated with RSFC between the IFGop seed and left superior temporal gyrus in both age groups. However, the fact that RSFC strength between Broca's and Wernicke's areas was weaker in children than in adults suggests that this connection was still undergoing development in our child participants. Nonetheless, our RSFC finding suggests that RSFC between Broca's and Wernicke's areas is important in reading competence, regardless of age group. This is in line with the Wernicke-Geschwind model of language (Geschwind, 1965
), which emphasizes “disconnection” of these two regions in individuals with a range of language impairments, including reading disorders (Catani and Mesulam, 2008
Divergent RSFC–reading competence relationships
We also observed significant age-group differences in relationships between reading competence and RSFC within functional connections associated with the FFG and IPS seeds. These group differences reflected dissociations in the direction of the RSFC–behavior relationships exhibited by children and adults.
The first divergent relationship was observed for RSFC of the FFG, a region commonly referred to as the visual word form area (Dehaene et al., 2002
). This region, located in the ventral visual pathway, is reliably activated by perception of visual forms of words, presumably resulting from extensive experience in visual word recognition during reading (Vinckier et al., 2007
; Glezer et al., 2009
). Specifically, RSFC–behavior relationships were divergent for RSFC between the FFG and both left inferior frontal gyrus opercularis (Broca's area) and left inferior parietal lobule, areas involved in phonological rehearsal (i.e., articulation) and phonological storage, respectively, during word reading (Paulesu et al., 1993
; Stoeckel et al., 2009
). While adults exhibited positive relationships between reading competence and RSFC between FFG and these phonology-related regions, children exhibited negative RSFC–behavior relationships. This suggests that stronger positive RSFC between these frontoparietal phonological regions and the FFG is beneficial for reading performance in adults (i.e., the mature brain), but not in children. Such divergent RSFC–behavior relationships may reflect the experience-dependent functional development of FFG, where the magnitude of reading task-induced activation is different between children and adults (Turkeltaub et al., 2008
A second divergent relationship was observed for RSFC between the FFG and both VMPFC and PRC/PCC. While adults who exhibited the strongest negative RSFC between these regions had the highest reading competence, children who exhibited weaker negative or even positive RSFC between FFG and both VMPFC and PRC/PCC had higher reading competence. The VMPFC and PRC/PCC constitute two core nodes of the default network, which is typically deactivated during goal-oriented performance (Raichle et al., 2001
; Buckner et al., 2008
). The RSFC–behavior relationships observed in adults suggest that, in the mature brain, reading competence is associated with increased negative correlation between FFG and default network regions. This is consistent with previous observations that stronger negative correlations between task-active networks and the default network (i.e., functional segregation) are associated with better cognitive performance in adults (Kelly et al., 2008
). Our findings suggest that functional segregation between the FFG, a region commonly associated with reading, and default network components indexes efficient reading in the mature brain. In contrast, segregation of the FFG from the default network was not associated with higher reading competence in children. Considering developmental changes in the RSFC of the default network (Fair et al., 2007
; Supekar et al., 2010
), this lack of clear functional segregation from the default network may be developmentally typical, and may not be deleterious to reading in children.
A final divergent relationship was observed for the IPS, a core region for top-down control of attention (Corbetta and Shulman, 2002
; Fox et al., 2006
). RSFC between the IPS seed and bilateral thalamus showed a positive relationship with reading competence in children, but a negative RSFC–behavior relationship in adults. That is, children exhibiting stronger positive RSFC between these regions exhibited higher reading competence, whereas adults exhibiting stronger negative RSFC between the same regions exhibited higher reading competence. The observed strong positive RSFC–behavior correlation in children suggests that children's reading relies on increased connectivity between cortical (IPS) and subcortical (thalamus) attention regions, which does not appear to be necessarily beneficial for adult reading. Although the thalamus is traditionally considered as a relay station passing sensory information to cortical regions (Jones, 1985
), recent perspectives highlight the role of the posterior thalamus, the pulvinar, in visual attention (Grieve et al., 2000
; Fan et al., 2005
). Topographically, the pulvinar largely overlaps with the thalamus cluster we observed (B.THAL) (E
). Visual attention has been hypothesized as a “spotlight” for scanning printed letters and words in a serial manner during reading, which is the inevitable and initial step for reading (Vidyasagar, 1999
). Accordingly, it can be inferred that functional coupling between cortico-subcortical attentional regions is specifically important for children, whose visual scanning during reading has not yet become effortless or automatized. Given that patterns of RSFC between the thalamus and cortical regions undergo developmental changes in childhood and adolescence (Fair et al., 2010
), the observed positive RSFC–behavior relationship (lack of segregation) may be developmentally typical and important in children.
Our results should be interpreted in light of limitations. First, the present work does not include older adolescents (>15 years old), whose intrinsic functional architecture is likely to be still undergoing dynamic developmental changes (Dosenbach et al., 2010
; Power et al., 2010
), although their reading competence approaches that of adult levels. Similarly, children younger than 7 years of age were not included, although clearly this period is highly dynamic and critical to the development of reading competence (Brem et al., 2010
). It will be of great interest for future work to provide more comprehensive examinations of potential developmental shifts in the neural correlates of reading competence across the lifespan. For young child populations, greater focus on phenotypic measures known to be early precursors of reading competence (e.g., phonological awareness) may prove to be fruitful. Finally, it is worth noting that general intelligence is highly related to reading competence in unimpaired, but not impaired readers (Hulslander et al., 2004
; Ferrer et al., 2010
). Additionally, other phenotypic variables, such as phonological awareness and working memory (Gathercole et al., 2006
), have been associated with reading competence as well. The present work included IQ as a covariate of no interest to remove the potentially confounding effects of general intelligence. Future work would benefit from inclusion of more comprehensive phenotypic information related to reading competence, to account for their potential contributions to the RSFC–reading competence relationships observed in the present work.
In conclusion, we identified both common and divergent relationships between reading competence and RSFC in children and adults. Regardless of age group, stronger coupling among motor regions, as well as between classic language/speech regions, was associated with higher reading competence. In contrast, several RSFC–behavior relationships, particularly those anchored in the left FFG (a region implicated in reading-related visual expertise), diverged between groups. Stronger coupling within FFG-based reading circuitry and stronger functional segregation between the FFG and default network components appear to be relevant for efficient and automatized reading in the mature brain (i.e., adults). The absence of these features in children who present age-appropriate reading skills highlights the importance of appreciating developmental changes within neural correlates of reading competence. Our results suggest that RSFC can be used to probe the development of reading competence, and thus may serve to facilitate identifying reading disorders in different age groups.