Memory consolidation during sleep relies on the precisely timed interaction of rhythmic neural events. Here, we investigate differences in slow oscillations (SO; 0.5–1 Hz), sleep spindles (SP), and their coupling across the adult human lifespan and ask whether observed alterations relate to the ability to retain associative memories across sleep. We demonstrate that older adults do not show the fine-tuned coupling of fast SPs (12.5–16 Hz) to the SO peak present in younger adults but, instead, are characterized most by a slow SP power increase (9–12.5 Hz) at the end of the SO up-state. This slow SP power increase, typical for older adults, coincides with worse memory consolidation in young age already, whereas the tight precision of SO–fast SP coupling promotes memory consolidation across younger and older adults. Crucially, brain integrity in source regions of SO and SP generation, including the medial prefrontal cortex, thalamus, hippocampus and entorhinal cortex, reinforces this beneficial SO–SP coupling in old age. Our results reveal that cognitive functioning is not only determined by maintaining structural brain integrity across the adult lifespan, but also by the preservation of precisely timed neural interactions during sleep that enable the consolidation of declarative memories.

a b s t r a c tWe suggest that working memory (WM) performance can be conceptualized as the interplay of low-level feature binding processes and top-down control, relating to posterior and frontal brain regions and their interaction in a distributed neural network. We propose that due to age-differential trajectories of posterior and frontal brain regions top-down control processes are not fully mature until young adulthood and show marked decline with advancing age, whereas binding processes are relatively mature in children, but show senescent decline in older adults. A review of the literature spanning from middle childhood to old age shows that binding and top-down control processes undergo profound changes across the lifespan. We illustrate commonalities and dissimilarities between children, younger adults, and older adults reflecting the change in the two components' relative contribution to visual WM performance across the lifespan using results from our own lab. We conclude that an integrated account of visual WM lifespan changes combining research from behavioral neuroscience and cognitive psychology of child development as well as aging research opens avenues to advance our understanding of cognition in general.

Working memory (WM) capacity increases across childhood, peaks in young adulthood, and declines thereafter. Developmental and aging theories suggest that deficient inhibitory control processes in children and older adults may underlie the lower performance relative to younger adults. Recently, oscillatory alpha power (7-13 Hz) of the electroencephalogram (EEG) has been suggested as a neural marker of inhibition processes contributing to WM performance ). We examined 20 children (10-13 years), 12 younger adults (20-26 years), and 20 older adults (70-76 years) in a cued change-detection paradigm. Behaviorally, we observed the expected lifespan peak in younger adults. EEG alpha power was generally reduced in older adults compared to children and younger adults. In line with previous research, hemispheric differences in alpha power related to attention and WM processes during the retention interval increased with load in younger adults. In children and older adults, lateralized alpha power increased from low to medium load conditions, but decreased for high load conditions. Furthermore, older adults showed higher inter-trial phase stability shortly after stimulus onset compared to children and younger adults. Our results show that inhibitory control processes as indexed by local alpha power modulations can be observed in children and older adults but seem to break down when WM load is high. In addition, older adults are more entrained by external stimulation what may increase a need for inhibitory control during later processing. We conclude that differences in inhibitory control processes and information uptake as reflected in amplitude modulations and inter-trial phase stability of alpha rhythms interactively determine WM constraints across the lifespan.

Estimates of working memory (WM) capacity increase in children, peak in young adulthood, and decline thereafter. Despite this symmetry, the mechanisms causing capacity increments in childhood may differ from those causing decline in old age. The contralateral delay activity (CDA) of the electroencephalogram, an event-related difference wave with a posterior scalp distribution, has been suggested as a neural marker of WM capacity. Here, we examine 22 children (10-12 years), 12 younger adults (20-25 years), and 22 older adults (70-75 years) in a cued change detection paradigm. Load levels and presentation times were varied within subjects. Behaviorally, we observed the expected life-span peak in younger adults and better performance with longer presentation times. With short presentation times, task load increased CDA amplitude and decreased behavioral performance in younger adults. Both effects were less pronounced in older adults. Children showed a unique pattern: Their behavioral load effects were as strong as those of younger adults, but their CDA was unaffected by load. With long presentation times, task load modulated the CDA in children and older adults but not in younger adults. These findings suggest that age-related differences in CDA reflect changes in the top-down control over WM representations.

Working memory (WM) shows a gradual increase during childhood, followed by accelerating decline from adulthood to old age. To examine these lifespan differences more closely, we asked 34 children (10 -12 years), 40 younger adults (20 -25 years), and 39 older adults (70 -75 years) to perform a color change detection task. Load levels and encoding durations were varied for displays including targets only (Experiment 1) or targets plus distracters (Experiment 2, investigating a subsample of Experiment 1). WM performance was lower in older adults and children than in younger adults. Longer presentation times were associated with better performance in all age groups, presumably reflecting increasing effects of strategic selection mechanisms on WM performance. Children outperformed older adults when encoding times were short, and distracter effects were larger in children and older adults than in younger adults. We conclude that strategic selection in WM develops more slowly during childhood than basic binding operations, presumably reflecting the delay in maturation of frontal versus medio-temporal brain networks. In old age, both sets of mechanisms decline, reflecting senescent change in both networks. We discuss similarities to episodic memory development and address open questions for future research.Keywords: aging, binding, change detection, cognitive control, development, visual working memory A primary function of visual working memory (WM) is to maintain information of perceptual input from the environment for a short period of time so that the information can be used for goal-directed behavior (D'Esposito, 2007). A critical feature of WM is its limited capacity, which is usually estimated to include about three or four items (Luck & Vogel, 1997;G. A. Miller, 1956). Because WM capacity has been shown to be predictive for a wide range of cognitive functions (Engle, Tuholski, Laughlin, & Conway, 1999;Kane et al., 2004;Oberauer, Subeta, Wilhelm, & Wittmann, 2008), the determination of the individual WM limit has been of great scientific interest.One paradigm to measure visual WM capacity is the change detection task (Luck & Vogel, 1997). In its standard version the observer is briefly presented with a memory array followed by a one-second retention interval and then compares the representation maintained in WM to a probe array in which one item might have changed. Based on the correct and incorrect answers to displays with increasing WM load, a person-specific capacity measure (k score) can be calculated (Cowan, 2001). In the past, the change detection paradigm has been applied in a whole range of behavioral (Eng, Chen, & Jiang, 2005;Jiang, Olson, & Chun, 2000;Olson & Jiang, 2004;Vogel, Woodman, & Luck, 2001Wheeler & Treisman, 2002;Woodman & Vogel, 2005), electrophysiological (McCollough, Machizawa, & Vogel, 2007;Sauseng et al., 2009;Vogel & Machizawa, 2004;Vogel, McCollough, & Machizawa, 2005), and neuroimaging (Todd & Marois, 2004Xu, 2007;Xu & Chun, 2006;Yeh, Kuo, & Liu, 2007) studies, with slight variations. ...