Working memory (WM) is a complex cognitive function that is generally defined as the ability to temporarily store and manipulate information (Baddeley, 1992). WM is further crucial for our daily life functioning, because it is closely related to everyday functions such as attention (Oberauer, 2019), planning (Cohen and Conway, 2007, Gilhooly, 2004) and reasoning (e.g., Oberauer et al., 2008; Süß et al., 2002). Because WM capabilities decline with increasing age (Park et al., 2002, Salthouse, 1991), and thus impair autonomy and well-being in older adults (R. S. Wilson et al., 2013), it is highly important to understand the neural mechanisms of this age-related cognitive decline.

In electroencephalography (EEG), a highly influential an widely used paradigm to study the visual domain of WM (visual WM, VWM) is the lateralized color change detection task (Vogel & Machizawa, 2004), in which one cued side of a bilateral set of visual stimuli needs to be remembered. A common and well-replicated neurophysiological measure extracted in this task is the contralateral delay activity (CDA). The CDA is characterized by a sustained negative voltage deflection over posterior scalp regions contralateral to the memorized location. It has been suggested that the amplitude of the CDA represents a measure of VWM storage, because it varies as a function of VWM load and performance in younger participants (e.g., Adam et al., 2018; Luria et al., 2016; Roy & Faubert, 2022; Vogel & Machizawa, 2004). Although the CDA has also been found in older participants, there is ambiguous evidence whether and how its amplitude changes with age (Sander et al., 2011, Störmer et al., 2013) or whether its insensitive to age-related decreases in VWM capacity (Duarte et al., 2013, Jost et al., 2011, Ko et al., 2014, Schwarzkopp et al., 2016). Therefore, the role of the CDA amplitude as a potential indicator of age-associated cognitive decline, or as a marker of preserved neurophysiological processes across different age ranges, has not been fully elucidated.

Another promising candidate measure to study the neural underpinnings of age-related VWM decreases in this change detection paradigm is oscillatory activity in the alpha band, typically defined as a frequency range of 8-13 Hz (Babiloni et al., 2020). In younger adults, Sauseng et al. (2009) first reported significant lateralization of alpha power, shifting towards ipsilateral electrodes relative to the remembered stimuli. This lateralization increased for larger set sizes and correlated with task performance. Later studies on younger participants replicated the lateralization of alpha power, however, its dependency on VWM load and its relation to performance were reported inconsistently (Adam et al., 2018, Fukuda et al., 2016, Vissers et al., 2016). Because alpha power shows a similar lateralization pattern as the CDA amplitude in younger adults, it was hypothesized that both measures may represent different manifestations of the same underlying process (Mazaheri and Jensen, 2008, van Dijk et al., 2010). However, evidence from prolonged retention phases indicates that both signals show differential time signatures, and suggests that alpha power lateralization may be more directly linked to the continuous retention of the VWM stimuli (Fukuda et al., 2015).

In aging research, only few studies investigated lateralization of alpha power in the color change detection paradigm, indicating diminished lateralization in older age (Leenders et al., 2018, Sander et al., 2012, Tagliabue et al., 2019). However, none of these aging studies reported any relation between the lateralization of alpha power and the age-related decline in task performance. Thus, to better understand the mechanisms of age-related cognitive decline in VWM functions, it is crucial to investigate whether task performance and alpha lateralization merely co-vary with age, or whether reduced alpha lateralization is directly linked to the age-related decrease in performance (i.e. whether older participants with more preserved lateralization also show better performance). Otherwise, diminished alpha power lateralization alone cannot provide insights into the age-related cognitive decline, as it may be driven by unrelated, or even non-neural alterations, such as changes in conductivity.

It is important to emphasize that all of the aforementioned studies used an absolute measure of alpha power (i.e. total power) to investigate lateralization effects in VWM. However, a growing body of literature is strongly emphasizing that these conventional analyses are confounded, because they conflate periodic (i.e. neural oscillations) and aperiodic signal components of the neural power spectrum (e.g., Donoghue et al., 2020). While the aperiodic signal component has been largely neglected in previous literature, recent evidence shows that it varies dynamically and carries physiologically relevant information (e.g., Gao et al., 2017), and thus cannot be treated as background noise. Consequently, changes in the aperiodic signal may strongly bias findings that are based on total measures of frequency band power. Furthermore, standard approaches may lead to extraction of power in specific frequency bands, although no periodic component is present on top of the aperiodic signal (Donoghue et al., 2020). Specifically, in the context of aging it has been shown that the aperiodic signal flattens from adulthood to older age, and consequently induces a bias when investigating age-related changes in conventional alpha power measures (Cellier et al., 2021, Tröndle et al., 2022, Tröndle et al., 2023, Voytek et al., 2015). Furthermore, previous studies investigating WM tasks showed that the aperiodic signal dynamically changes during selective attention, and during WM encoding and retention (Donoghue et al., 2020, Gyurkovics et al., 2022, Pietrelli et al., 2022, Preston et al., 2022, Virtue-Griffiths et al., 2022, Waschke et al., 2021). In the context of VWM, it is thus unclear whether the aperiodic signal might bias previous findings on total alpha power lateralization in younger and older participants. Furthermore, the aperiodic signal may provide new insights into the underlying neurophysiological processes in VWM and their age-related changes. As the predominant physiological interpretation of the slope of the aperiodic signal (i.e. the aperiodic exponent) is the ratio of neuronal excitation and inhibition (Chini et al., 2022, Gao et al., 2017), it may show similar hemispheric differences as found in total alpha power.

To fill these knowledge gaps, we recorded EEG of 134 healthy younger and older participants during a lateralized change detection task. First, by decomposing the neural power spectrum into periodic and aperiodic signal components, our study investigated possible biases in previous literature on total alpha power lateralization in younger participants. Furthermore, we investigated potential age differences in the lateralization of both periodic alpha power and aperiodic signal components. Control analyses on total alpha power ensured the comparability to previous work. Importantly, all extracted EEG measures were related to task performance using a regression approach and a post-hoc mediation analysis, thus allowing to resolve the missing link between age-related neurophysiological changes and the observed decline in VWM performance. Finally, additional analysis of the CDA amplitude in both age groups provided new insights into the relation between the lateralization of alpha power and the CDA amplitude, and how the CDA amplitude relates to age-related decline in VWM performance.