By integrating functional activity, structural morphology, and cognitive skill performance, we were able to identify brain regions where functional activation during orthographic processing was related to cortical thickness. The correlation between activation and naming speed (shown in ) allowed us to determine that greater difference in activation between false font strings and words reflects a more “mature” pattern of activation. This “mature” pattern of activation was associated with thinner cortex in the right fronto-parietal region (), where cortex thins during childhood and adolescence (Sowell et al. 2003
; Gogtay et al. 2004
). These data are consistent with the notion that more “mature” cortical thickness patterns (i.e., thinner) are associated with more “mature” activation patterns. This relationship was preserved in the fronto-parietal attention network even after controlling for age or IQ.
One assumption of the present study is that the orthographic processing task used is facilitated by naming skills. The faster children are at naming letters, the easier it should be to decide if a tall letter is in the stimulus. Naming speed is a learned skill but it also correlates with age. Coupling of age and skill level confounds activation due to genetic influences that unfold with age and activation associated with higher skill level. We attempted to dissociate the 2 by using multiple regression to examine effects of skill level while statistically controlling for age effects. Results showed that activation in the fronto-parietal attention network was still associated with naming speed once the effects of age were covaried out. That is, chronological age does not fully explain the relationship between thinning fronto-parietal cortex and faster-naming speed. As both age and experience with print contribute to naming speed, it is tempting to speculate that the relationship between the fronto-parietal attention network and naming speed reflects experience effects. However, the present study design does not allow for such inference because “experience” was neither quantified nor measured. We are not the only group to have parsed out age effects from activation–skill level relationships. Our findings are consistent with Schlaggar and colleagues’ reports of functional activation that varies independently of age and is performance related (Schlaggar et al. 2002
; Brown et al. 2005
In a separate analysis, we controlled for IQ while examining the relationship between activation and skill level. It is possible that individuals with higher intelligence have faster-naming speed and that activation–naming speed relationship found in fronto-parietal regions is driven by effects of intelligence. Activation of the fronto-parietal attention network was associated with naming speed independent of IQ. This finding is important because it suggests that general cognitive ability does not fully explain activation in the fronto-parietal attention network during orthographic processing. Faster-naming individuals activate this fronto-parietal attention network more regardless of effects of general cognitive ability.
The fronto-parietal attention network is involved in both overt and covert orienting of attention to spatial location (Fan et al. 2005
; Raz and Buhle 2006
). As reviewed by Kastner and Ungerleider (2000)
, top-down bias from the attention system to visual processing includes enhancement of neural responses to an attended stimulus and increasing stimulus salience. Functional connectivity of the superior parietal lobule and frontal eye fields put them in a position to serve as sources of top-down biasing signals to visual processing streams (Kastner and Ungerleider 2000
). The strongest determinant of neural responsiveness of the lateral intraparietal area turns out to be the salience of the stimulus (Colby and Goldberg 1999
). The most parsimonious explanation of enhanced activation of this fronto-parietal attention network among individuals with more advanced skills, after controlling for age or general cognitive ability, is that false font strings appear more salient than real words to those with more advanced naming skill. Naming speed can be enhanced by experience, so sensitivity of this parietal–frontal attention network to false font strings may be a specific consequence of experience. In other words, false font strings and real words may be equally novel to slow-naming individuals (i.e., less experience with print, after accounting for their age or IQ) resulting in less difference in activation between the 2 conditions. But among fast-naming individuals (i.e., more experience with print, after controlling for age or IQ), word stimuli were likely quite familiar and only the false font strings were novel and salient, resulting in greater activation of fronto-parietal attention networks to the false font strings relative to real words.
The functional task used here was initially designed to elicit implicit reading (Turkeltaub et al. 2003
). We did not observe statistically significant relationships between cortical thickness and implicit reading activation in classical brain language regions. Cortical thickness in these regions have been shown to increase with age into young adulthood, but variability is much higher and effects much smaller than the thinning that occurs in frontal and parietal cortices during this age range (Sowell et al. 2003
). Cortical thickening was not observed in these regions in the present sample, perhaps due to limited cross-sectional sample size or a relatively restricted age range. Activation related to the implicit reading aspects of the task may be less robust than activation related to attentional aspects of the task, as suggested by the observation that activation in posterior perisylvian language regions did not survive correction for multiple comparisons. That reading requires attentional resources is not a novel finding, as Turkeltaub et al. (2003)
also reported activation in fronto-parietal attention areas for words and false font strings relative to rest using the same task. Our study highlights that the difference in activation pattern of children from that of adults reported in other studies may not reflect differences in the targeted cognitive system under study, but may in fact reflect supportive cognitive systems required to complete the task. “Mature” activation patterns for a given task may involve supportive systems as much as the target system, consistent with other developmental fMRI reports (reviewed in Durston and Casey 2006
The current results are consistent with the notion that the protracted course of functional activation development in the human brain is associated with skill level, at least in fronto-parietal attention networks, and activation corresponds to more “mature” morphology in these regions. From this, we posit that structural brain development may also be related to skill level and not merely to genetic influences that unfold with age. There may be relationships between mature patterns of activation and morphology in other regions of the brain that we lacked sufficient power to detect. The current study is a correlational study and cannot address if structural maturation allows for learning to take place, or if experience leads to morphological maturation. Nevertheless, integrating structural MRI with functional MRI and neurocognitive performance holds tremendous promise in elucidating structure–performance relationships of high-level cognitive functions specific to humans.