We examined top-down attentional modulation of alpha-band oscillatory activity related to three different stages of the visual working memory task, in 10 year old children and young adults. the CNBT, the anticipatory stage, target detection, and post-response. Although the sensory, bottom-up information was identical in both children and adults, the pattern of top-down event-related alpha modulation was different. In adults, the proximity of the anticipatory stimuli to the forthcoming target was reflected in a gradual increase in ERD of alpha suggesting an active preparatory state for target detection. In children a lesser modulation of alpha was exhibited during anticipatory stages, but a major difference emerged after response to the target. Thus, in contrast to our primary hypothesis about the most prominent differences between children and adults to be displayed at the anticipatory task stages, the major group differences in alpha modulation were demonstrated at the post–response stage. Here alpha-band in adults synchronized to above baseline level, whereas in children it remained at a desynchronized state below the baseline. Therefore, in adults the post-target ERD was followed by an ERS, whereas in children the ERD protracted long after the response was made, suggesting a persistent state of readiness. However, one must be aware, that these post-response group-differences in the modulation of alpha are not just effects occurring on the final post-response stage of the task. The post-response final effect may be a cumulative outcome of differences in stimulus processing, baseline alpha, and other intrinsic processes, each affecting the alpha-band activation in a mode specific to a particular developmental level.
The allocation of attentional resources in visual working memory tasks has been broadly studied (Jonides, Schumacher, Smith, Koeppe, Awh, Reuter-Lorenz et al., 1998
; Smith and Jonides, 1998), but little is known about the attentional resources allocated to anticipatory strategies preceding targets in adults (Corbetta and Shulman, 2002) and almost nothing is known in children. Thus, the current study elucidates the development of processing stages in visual working memory. The modulation of anticipatory alpha in adults suggests a more refined interplay between allocation of attentional resources to task-relevant networks related to anticipatory stages, and inhibition of irrelevant processing at the post response stages. This view is consistent with the early attentional capacity models (Broadbent; 1958
; Kahneman, 1973
; Norman and Bobrow, 1975
), and more recent “biased competition” model (Desimone and Duncan, 1995
; Kastner, Pinsk, De Weerd, Desimone, Ungerleider, 1999
). In the biased competition model, the stimulus-related competing groups of neurons were suggested to be biased by top-down inhibitory control of irrelevant neuronal activation, or “selective lateral inhibition” (Konorski, 1967
). The gradual increase of ERD as by, reflecting a top-down controlled increase in preparatory activation, appears consistent with such fine-tuned inhibitory/excitatory interplay (Gevins et al., 1997
). Furthermore, it has been shown using the Granger causality technique that greater directed influence is generated from frontal eye fields to inferior parietal sulcus, than the reverse, providing anatomical support for top-down frontal-parietal attentional flow. Such increased causality was predictive of correct behavioral performance (Bressler, Tang, Sylvester, Shulman, Corbetta, 2008
). Thus, frontal-parietal networks are essential for visual top-down attentional control, and consequently, less flexible modulation of the parietal alpha-band in children may reflect immaturity of those networks.
Alternatively, the lower modulation of alpha in children may result from difficulty in internalizing task rules since the top-down voluntary control strictly follows a set of rules. This, however, is less feasible, because children performed with a speed and accuracy that was comparable to adults. Furthermore, the effect we see in children in this study may also result from a broad scalp distribution of channels with low alpha. This also seems unlikely, because in such a case the differences between adults and children would be more uniform across all stages of the task (2BT, 1BT, T, and PR). Our data points in the opposite direction showing the variation of alpha determined by variable task stages. Consistently, age-driven differences in brain activation accompanied by accuracy of performance similar to adults are one of the most consistent findings in our developmental neuroimaging studies (Ciesielski et al., 2006
During the developmental trajectory the cortical-cortical networks develop late, in parallel with formation and deepening of sulci and increasing complexity of the cortical surface (Nolte, 2008
). Capacity of visual processing and responding is progressing concurrently with development of subcortical-cortical and cortical-cortical loops (Johnson, Mareschal, Csibra, 2001
). The frontal brain areas begin to reach functional and structural maturity only during late adolescence (Happaney, Zelazo, Stuss, 2004
; Luciana and Nelson, 2003; Sowell, Thompson, Leonard, Welcome, Kan, Toga, 2004
). Since the anticipatory pre-target stages in CNBT rely on the proficiency of the frontal-parietal dorsal network, engaged in top-down attentional control (Corbetta and Shulman, 2003
), their immaturity may contribute to reduced flexibility in attentional modulation of alpha-band across the task epoch, including the response stage. Consistent with this view is the lower performance in children on tests commonly related to frontal lobes (Wisconsin Card Sorting and in Stroop Word-Color Interference Tests). Consistent with the above are prior studies in adults demonstrating a stronger coupling of alpha oscillations between parietal-occipital and frontal areas in visual working memory tasks. Higher alpha amplitude was shown in trials preceding “seen” than “not seen” trials (Buchel and Friston, 1997; Babiloni, Vecchio, Bultrini, Romani, Rossini, 2006
). Since the modulation of synchronous changes characterizing the connectivity of functional brain networks has been linked to cognitive strategies (Buzsaki, 2007), one possible explanation of child-adult differences is that children may use a different set of attentional strategies, and, possibly different developmentally available networks. Our prior studies point to participation of the earlier developing striatum a cerebellum (Ciesielski, Lesnik, Benzel, Hart, Sanders 1999
; Ciesielski et al., 2006
) closely anatomically and functionally related to the parietal cortex (Clower, Dum, Strick, 2005
; Hoshi, Tremblay, Feger, Carras, Strick 2005
; Schmahmann and Pandya, 1997
The low sensitivity of parietal alpha oscillations to cognitive top-down modulation may be yet another reason for the brain-behavior dissociation in children (Krause et al., 2007
; Petersen and Eeg-Olofson, 1971
). Although caution in interpretation is in order, one may ponder whether the post-response ERS of alpha in adults reflects a return to an idling neuronal state, or, alternatively, to an active inhibitory state within the task-unrelated networks (Jensen, Gelfand, Kounios, Lisman, 2002
; Worden et al., 2001). It has been proposed that ERS of alpha, exhibited after the task is completed, would be representative of a widely activated cortical-cortical network, which allows a quick flexible shift to local information processing after the stimulus is presented (Babiloni et al., 2005b
; Nunez, Wingeier, Silberstein, 2001
). In children such flexibility may be missing. Their ERD at the post-response stage, suggests a lower sensitivity of alpha activity to the task context and slower recovery of baseline alpha. In the present study we did not analyze frontal ERD/ERS, because little alpha-band activity was observed in the prefrontal sensors, and we were concerned about sensory-motor mu rhythm from the nearby precentral motor regions.
In summary, our results suggest significant top-down modulation of ERD/ERS in alpha-band activity during the anticipatory and post-response time-windows of the CNBT working memory task in adults, and paucity of such task-related modulation in children. Since both groups performed with similar speed and accuracy, it is tempting to consider engagement of different strategies and different neuronal networks in children. Such alternative developmental networks may be encompassing the earlier maturing striatum and cerebellum (Goldman, 1974
; Ciesielski et al., 1999
). Indeed, the modulation of synchronous changes characterizing the connectivity of functional posterior brain networks have been linked to variation in cognitive strategies (Krause et al., 2001
; Buzsaki, 2007). The parietal-occipital alpha-band activity is considered to be an universal functional property of the brain, as it correlates with stimulus-evoked activation in many sub-cortical and cortical networks (Steriade and Llinas, 1988
). Here, we provide developmental support for the emerging novel models of the inferior parietal cortex as an essential component for on-line episodic-recall in working memory (Vilberg and Rugg, 2008), and as a center (“hub”) comprising connections from all major control networks (Buckner, Andrews-Hanna, Schacter, 2008
). A search for developmentally apt top-down visual control networks engaged in modulation of parietal-occipital alpha-band activity merits further investigation.
- The prefrontal-parietal networks associated with top-down cognitive control in adults are not matured in 10-year olds, yet children perform well on visual-working memory tasks.
- We use magnetoencephalography (MEG) to monitor in children and adults event-related desynchronization (ERD) and synchronization (ERS) of parietal-occipital alpha-band oscillatory activity (8–13 Hz) during top-down anticipatory, target detection and post-response stages of the Categorical N-Back Task (CNBT).
- Since Children performed as well as the Adults on CNBT and yet displayed developmentally distinct, task-stage-determined patterns of ERD/ERS we suggest that children may be using different top-down cognitive strategies and, hence, different, developmentally apt neuronal networks.