Development of attentional networks: An fMRI study with children and adults

Abstract

Data on the development of the attentional systems remain scarce. We used structural and event-related functional magnetic resonance imaging to investigate differences in the neural mechanisms associated with alerting, reorienting, and executive control of attention between children (ages 8 to 12 years) and adults, while controlling for effects of performance and brain morphology. Behaviorally, children exhibited a numerically smaller alerting effect and significantly larger invalidity (reorienting) and interference (executive control of attention) effects. Neurally, children showed significantly reduced brain activation in a priori defined regions-of-interest in right-sided frontal–midbrain regions during alerting, in the right-sided temporo-parietal junction during reorienting of attention, and in the dorsolateral prefrontal cortex during executive control of attention. In addition, children activated significantly more brain regions outside the a priori defined regions-of-interest, such as the superior frontal gyrus during reorienting and the superior temporal gyrus during executive control of attention. Functional group differences overlapped with structural group differences in gray matter volume in particular within the frontopolar areas. The data suggest that there is a transition from functional yet immature systems supporting attentional functions in children to the more definitive adult networks and that the differences observed may reflect both developmental changes in cognitive strategies and morphology.

Introduction

One fundamental objective in basic and clinical neuroscience is to understand the development of the functional organization of the human brain. Such knowledge cannot only inform us about its normal anatomical and functional development but may also help us to understand the consequences of early perturbations to the developing brain or aberrant organizations thereof which are assumed to underlie developmental disorders (Casey, 2000, Schlaggar et al., 2002). Attention plays a critical role in the normal human cognitive, emotional, and social development (Johnson, 2000). Attentional deficits occur in a number of common childhood psychiatric diseases, such as Attention Deficit Hyperactivity Disorder (ADHD), autism, or childhood depression. Developmental studies suggest that some mechanisms of attention are present from early infancy; however, they also provide evidence of a differential development of particular attentional functions (Rueda et al., 2004).

Posner and Petersen (1990) developed an influential model of attentional functions, in which differential neural networks and neuromodulators are assumed to subserve different attentional functions. In this model, alerting is defined as achieving and maintaining an alert state. Orienting and reorienting are required when stimuli occur outside the current focus of attention. Finally, executive control is defined as resolving conflict among responses. While the alerting system has been associated with frontal and parietal regions of the right hemisphere modulated by norepinephrine (Witte and Marrocco, 1997), orienting and reorienting of attention are thought to be mediated by a network which is lateralized to the right temporo-parietal junction and the right inferior frontal gyrus (Corbetta et al., 2000) and which is assumed to be modulated by the cholinergic system (Thiel et al., 2005). The executive attentional control network is supposed to involve the anterior cingulate and lateral prefrontal cortex and to be modulated by dopamine (Marrocco and Davidson, 1998). Several recent neuroimaging studies have investigated these attentional networks in healthy adult subjects and generally support the view that the neural networks subserving as the sources of alerting, reorienting, and conflict effects are anatomically separable, but also stress that there is substantial functional overlap (Fan et al., 2005). In addition, there are conflicting results concerning the precise anatomical localization of these attentional systems. For example, for the alerting system, some authors point to a right-sided fronto-parieto-thalamic network (Sturm and Willmes, 2001), while others suggest left lateralized parietal and frontal regions (Coull et al., 2001). For the orienting and reorienting system, Corbetta and Shulman (2002) propose two attentional networks: The orienting network which is involved in visual selection and working memory is suggested to depend on bilateral activity in intraparietal sulcus and the human homologue of frontal eyefields. In contrast, reorienting of attention to salient events is achieved by a another network, which is strongly lateralized to the right temporo-parietal junction and inferior frontal gyrus. For the conflict system, previous work with the flanker task show primarily cingulate activity, in addition to prefrontal cortex activation (e.g., Botvinick et al., 2001, Fan et al., 2003, Ullsperger and von Cramon, 2004).

Thus, it can be summarized that the dorsolateral prefrontal cortex and the cingulate gyrus might be involved in either alerting or conflict resolution, while the inferior frontal gyrus and the parietal cortex might be associated with both alerting and reorienting of attention.

Behaviorally, developmental studies of the alertness system suggest that although children and adults encode target relevant information at equivalent rates, children make less use of the warning aspect of the cue (Smothergill and Kraut, 1989). For the orienting system, it has been found that the ability of shifting attention to exogenous cues differs little between children and adults, while the ability to disengage, the speed of shifting attention voluntarily, and the accuracy of its termination seem to improve with age (Brodeur and Enns, 1997, Trick and Enns, 1998). With regard to higher-level forms of attention such as the resolution of conflict among competing stimulus elements, i.e., executive control, the majority of developmental studies has stressed a relative lack of executive control in infants until the age of 12 years (see Ruff and Rothbart, 1996, for a review).

On the neural level, to date, nearly all developmental fMRI studies (i.e., studies that compared children with adults) have focused on aspects of executive control and found evidence for a higher susceptibility to interference in children that was paralleled by maturational differences in underlying frontostriatal circuitry (Bunge et al., 2002, Durston et al., 2002, Casey et al., 2002). Recently, Booth et al. (2003) compared the neural development of selective attention and response inhibition and observed only small developmental changes for the selective attention tasks but large developmental differences in the response inhibition task indicating differential effects for attentional processes in children aged 9 to 12 years. However, no developmental fMRI study has yet investigated all three attentional systems (alerting, orienting/reorienting, and executive control).

Accordingly, we designed a functional imaging experiment that specifically aimed at investigating developmental differences in the neural mechanisms underlying alertness, reorienting, and executive control in children aged 8 to 12 years relative to adults. This age range was chosen since it has previously been shown that the attentional functions under investigation develop till the end of puberty (Carver et al., 2001, Bedard et al., 2002). Based on previous data, we hypothesized that, behaviorally and neurally, the largest developmental differences between children and adults would occur in the executive control system but that due to the functional overlap between these networks also the alerting and reorienting systems would be affected. More specifically, we assumed that healthy adult subjects would recruit specific and separable neural networks for alerting, reorienting, and executive control of attention. In contrast, for children, we predicted less focal recruitment of the circuitries involved, in particular with regard to fronto-striatal (Giedd, 1999, Casey et al., 2002) and fronto-parietal (Sowell et al., 2002) networks. To test these hypotheses, we measured changes in neural activity in healthy children and healthy adults using event-related fMRI while subjects performed a modified version of the Attention Network Task (ANT; Fan et al., 2002) which measures alerting, orienting, and executive attention.

Since developmental studies of brain morphometry indicate long-lasting maturation processes during childhood and adolescence (Giedd, 1999, Sowell et al., 2002, Sowell et al., 2004), in particular within the fronto-parietal networks and the cerebellum, we also obtained anatomical MR images on all subjects, allowing for the assessment of possible structural differences between the two groups, using voxel-based morphometry (VBM) (Ashburner and Friston, 2000).