Brain activation and deactivation during location and color working memory tasks in 11–13-year-old children

Abstract

Using functional magnetic resonance imaging (fMRI) and n-back tasks we investigated whether, in 11–13-year-old children, spatial (location) and nonspatial (color) information is differentially processed during visual attention (0-back) and working memory (WM) (2-back) tasks and whether such cognitive task performance, compared to a resting state, results in regional deactivation. The location 0-back task, compared to the color 0-back task, activated segregated areas in the frontal, parietal and occipital cortices whereas no differentially activated voxels were obtained when location and color 2-back tasks were directly contrasted. Several midline cortical areas were less active during 0- and 2-back task performance than resting state. The task-induced deactivation increased with task difficulty as demonstrated by larger deactivation during 2-back than 0-back tasks. The results suggest that, in 11–13-year-old children, the visual attentional network is differently recruited by spatial and nonspatial information processing, but the functional organization of cortical activation in WM in this age group is not based on the type of information processed. Furthermore, 11–13-year-old children exhibited a similar pattern of cortical deactivation that has been reported in adults during cognitive task performance compared to a resting state.

Introduction

Working memory (WM) refers to temporary storage and manipulation of information in the service of goal-directed behavior (Baddeley, 1992, Baddeley, 1996). This function is essential, e.g., in learning, reasoning, problem solving and language comprehension. WM performance and the underlying neural processes and brain structures continue to develop during childhood until adolescence and young adulthood (Gogtay et al., 2004, Hale et al., 1997, Luciana and Nelson, 1998, Luna et al., 2004, Luna et al., 2001).

Visual WM has been extensively studied in adults using brain imaging methods whereas only a small number of studies have been conducted in children. Brain imaging studies in adults have shown that regions in the prefrontal (PFC) and posterior parietal (PPC) cortices are central to visual WM (Klingberg, 2006, Owen et al., 2005). Most fMRI studies in children have investigated the development of visuospatial WM (Klingberg et al., 2002, Kwon et al., 2002, Nelson et al., 2000, Olesen et al., 2007, Scherf et al., 2006, Thomas et al., 1999), and in only a few studies the development of nonspatial WM was investigated (Casey et al., 1995, Ciesielski et al., 2006, Crone et al., 2006). These studies suggest that areas of the frontoparietal network that are activated in adults are also recruited in children in the performance of visual WM tasks, although the degree of engagement of different regions may change with maturation (Berl, Vaidya, & Gaillard, 2006). Furthermore, the development of visuospatial (Klingberg et al., 2002, Kwon et al., 2002) and nonspatial (Crone et al., 2006) WM capacity has been shown to be related to increased activity in the frontoparietal cortical areas and to maturation of the frontoparietal white matter (Olesen, Nagy, Westerberg, & Klingberg, 2003). Children’s lower WM capacity has been attributed to the maturational phase of the PFC. According to pediatric neuroimaging (Giedd et al., 1999, Gogtay et al., 2004) and post-mortem studies (Chugani et al., 1987, Huttenlocher, 1979, Huttenlocher, 1990, Huttenlocher, 1997), the PFC is among the brain regions that continue to mature until early adulthood.

The PFC has a central role in WM (Fuster, 1989, Goldman-Rakic, 1987), and several models have been proposed to explain the functional organization of this cortical area in WM (reviewed e.g. by Fletcher & Henson, 2001). In adults, functional imaging studies of the “process-specific” model provide evidence that the dorsolateral PFC (DLPFC) is responsible for monitoring and manipulating and the ventrolateral PFC (VLPFC) for maintaining information in WM (D’Esposito et al., 1999, D’Esposito et al., 2000, Owen, 2000, Owen et al., 1998, Petrides, 2005, Petrides et al., 2002, Rypma, 2006). Another model, called the domain-specific model (Goldman-Rakic, 1995, Wilson et al., 1993) has also gained support from brain imaging studies demonstrating that WM processing of spatial information is handled in the DLPFC and nonspatial information in the VLPFC (Belger et al., 1998, Courtney et al., 1996, Mottaghy et al., 2002, Sala et al., 2003, Ventre-Dominey et al., 2005). The two models may not be mutually exclusive, however, as there is evidence that the organization of the PFC is based both on the type of information being processed and the cognitive process employed during the performance of WM tasks (Johnson et al., 2003, Mohr et al., 2006).

WM and attention are closely intertwined; both processes promote goal-directed behavior and show a considerable overlap of the underlying neural circuitry (Awh et al., 2006, Chun and Turk-Browne, 2007, LaBar et al., 1999, Wager et al., 2004). Attentional systems have also been divided into spatial and nonspatial domains. The results of several neuroimaging studies suggest that top-down control of spatial and nonspatial attention activates largely similar areas of the frontoparietal network but subregions of this network are specific for controlling spatial selective attention (Coull and Frith, 1998, Giesbrecht et al., 2003, Vandenberghe et al., 2001).

In the adult brain, in the absence of cognitive task performance, i.e., during a resting state, several areas of the brain are active (Gusnard and Raichle, 2001, McKiernan et al., 2003, Raichle et al., 2001). Evidence is accumulating that the spontaneous, resting-state activity that can be measured with functional brain imaging reflects fluctuations of specifically organized activity between regions that have similar functionality (for a review see Fox & Raichle 2007). Resting-state activity has also been reported in children (Kiviniemi et al., 2000), infants (Fransson et al., 2007) and monkeys (Vincent et al., 2007). The “default mode network” refers to areas in the medial prefrontal and parietal cortices, cingulate cortex and inferior parietal lobule that are active during the resting-state and are attenuated (deactivated) during attention demanding goal-directed task performance (Raichle et al., 2001). Activity observed in the default mode network has been suggested to reflect internally directed mental processes (Gusnard & Raichle, 2001) and mind-wandering (Mason et al., 2007).

The physiological mechanisms underlying deactivation are incompletely understood at present. Deactivation seen by fMRI is suggested to result from disproportionate decrease in blood flow and glucose utilization (Raichle & Mintun, 2006) and may reflect a decrease in the activity of the cells that project to the deactivated area rather than increased activity of local inhibitory interneurons in that area (Gusnard & Raichle, 2001). Deactivation has been reported almost exclusively in adults with one exception reported in infants (Dehaene-Lambertz et al., 2006) and another in adolescents (Schweinsburg, Nagel, & Tapert, 2005).

In the present study, we investigate whether attention and WM processing of spatial and nonspatial information in children aged 11–13 is segregated as suggested by domain-specific model. To our knowledge, there are no earlier brain imaging studies that have investigated the organization of spatial and nonspatial WM processing in the child brain. We used location and color 0- and 2-back tasks and a resting state and fMRI to locate the brain areas displaying visuospatial and nonspatial attention and WM related activation and deactivation.

The 2-back task requires on-line monitoring, updating, and manipulation of retained information placing great demands on the key processes within WM whereas the 0-back is an attentional task that requires detection of a predetermined stimulus but, contrary to the 2-back task, does not demand manipulation or memorizing of the stimuli presented earlier (Owen et al., 2005). In designing the experiment, we specifically attempted to get similar difficulty levels for the location and color attention and WM tasks. To minimize the effect of age and gender on the WM performance (Vuontela et al., 2003), and to keep age- (Berl et al., 2006) and gender-related (Schweinsburg et al., 2005) differences in brain activation patterns as small as possible the subject group consisted of girls within a narrow age range of 11–13 years. Based on our earlier studies (Vuontela et al., 2003) we chose to investigate this age group that mastered the tasks relatively well compared to younger children. Furthermore, the results of event-related fMRI studies of spatial WM (Klingberg et al., 2002, Kwon et al., 2002) have suggested that WM networks in children within this age group are still in the process of maturation.

In the present experiment, we employed the 0-back that is an attentional task to enable the detection of possible difference between the neural networks involved in spatial and nonspatial attention. On the basis of the adult neuroimaging data (Giesbrecht et al., 2003, Vandenberghe et al., 2001), we predicted that spatial and nonspatial attention would recruit partially segregated regions of the frontoparietal attentional network in children. In any case, we expected to see this difference in the neural network of the PPC that matures earlier than the PFC network (Gogtay et al., 2004). The 2-back task was employed to test the domain-specific model of mnemonic processing. Considering the adult neuroimaging data, visual WM processing in children is as likely to be organized according to either of the two alternative models of WM. The resting state was included in the experiment to give children a break in between the task performance and to test whether deactivation mechanisms found in adult subjects induced by cognitive task performance are functional in 11–13-year-old children.