Aperiodic component of EEG power spectrum and cognitive performance are modulated by education in aging

Recent studies have shown a growing interest in the so-called “aperiodic” component of the EEG power spectrum, which describes the overall trend of the whole spectrum with a linear or exponential function. In the field of brain aging, this aperiodic component is associated both with age-related changes and performance on cognitive tasks. This study aims to elucidate the potential role of education in moderating the relationship between resting-state EEG features (including aperiodic component) and cognitive performance in aging. N = 179 healthy participants of the “Leipzig Study for Mind–Body-Emotion Interactions” (LEMON) dataset were divided into three groups based on age and education. Older adults exhibited lower exponent, offset (i.e. measures of aperiodic component), and Individual Alpha Peak Frequency (IAPF) as compared to younger adults. Moreover, visual attention and working memory were differently associated with the aperiodic component depending on education: in older adults with high education, higher exponent predicted slower processing speed and less working memory capacity, while an opposite trend was found in those with low education. While further investigation is needed, this study shows the potential modulatory role of education in the relationship between the aperiodic component of the EEG power spectrum and aging cognition.

Older adults with different educational levels did not differ on most cognitive tasks except for the working memory one [visual attention response times (t = −0.11,p = 0.91); alertness response times (t = −0.25,p = 0.79); working memory accuracy (t = 2.71, p < 0.01; Cohen's d = 0.72); delayed memory accuracy (t = −1.14, p = 0.25)], where older adults with high education performed better than older adults with lower education and more similarly to the younger adults (see also Table 1).

EEG spectral parameters and cognitive performance
In the whole sample, a higher exponent and offset significantly predicted a better performance on the visual attention task, i.e., a faster performance in terms of response time [(exponent: B = −0.44,p < 0.01; Cohen's f 2 = 25.11);(offset: B = −0.32,p < 0.01; Cohen's f 2 = 25.63)].The exponent and the offset values predicted better working memory capacity in terms of response accuracy [(exponent: B = 0.37, p = 0.04; Cohen's f 2 = 6.91); (offset: B = 0.31, p = 0.01, Cohen's f 2 = 7.96)].Significant results emerged when considering the three groups, i.e., young adults, older adults with low education, and older adults with high education, both for the exponent and the offset, in the visual attention and working memory tasks.Compared to the group of young adults, where exponent and offset did not predict any variation in cognitive performance, a significant interaction was shown in older adults depending on their educational level.In the visual attention task, low-educated older adults had a better (faster) performance at the higher aperiodic values [(exponent: B = −0.67,p = 0.04; Cohen's f 2 = 22; (offset: B = −0.56,p = 0.03, Cohen's f 2 = 21)] while those highly educated had a worse performance (slower) at the higher values of the exponent (exponent: B = 1.41, p = 0.02; Cohen's f 2 = 22), a result that was confirmed post-hoc through slope comparisons, showing a general age-related effect and also a significant difference between older adults with different education, at the third quartile of the exponent values (t = 2.45; p = 0.03), see Fig. 2.
Results showed that also in the working memory task, older adults with high education had worse performance with increasing exponent values, as compared with those with low education (B = −1.71,p = 0.03, Cohen's f 2 = 7.25).Post-hoc slope comparison showed no main education-related difference among older adults.Young adults and highly educated older adults did not differ from each other at different quartiles of exponent values (25%: t = 1.16, p = 0.47; 50%: t = 1.94, p = 0.12; 75%: t = 2.12, p = 0.08).
An additional exploratory analysis in a frontal ROI 29 showed that a significant interaction also emerged at the level of the alertness task (B = 2.01, p = 0.01, Cohen's f 2 = 9) with higher power predicting faster response times in older adults with high education.

Discussion
The present study aimed to investigate the relationship between aperiodic activity and cognitive performance, by accounting for the level of education in older individuals, compared with a control group of highly educated younger adults (i.e., high neurocognitive efficiency).
The older adults showed less cognitive efficiency compared to young adults, across all tasks, which aligns with the well-established literature about cognitive decline in healthy aging 30 .Upon considering older adults stratified based on education, the results indicated that those with higher education exhibited comparable performance to older adults with lower education, except for working memory performance.Older adults with higher education showed better working memory performance compared with older adults with lower education, which made the group of older adults with higher education more similar to the young group and suggesting a potential role of education in mitigating age-related cognitive decline, at least for some specific cognitive functions or tasks.
Consistent with previous evidence 17,31,32 , the periodic and aperiodic components of EEG differentiated between young and older adults: older adults exhibited lower values across these components, compared to younger adults, except for the parametrized power.Concerning the periodic component of the EEG signal, results showed a pattern of an age-related slowing of IAPF, reflecting findings related with previous studies 33,34 .Several interpretations have been advocated to link this periodic EEG components with aging.In addition to slowing with age, structural alterations in the brain have also been associated with the decline in power and peak frequency of alpha oscillations, particularly in older individuals 10,35 .
Additionally, the stability of power and IAPF over the life course reflects the preserved functionality of the central nervous system 36 .Regarding cognition, IAPF has previously demonstrated a positive relationship with interference resolution in working memory performance, primarily observed in the temporal lobes 29 .Our results indicate that, at least at the level of frontal brain areas (as it possible to observe in Supplementary Materials), power may play a functional role in the ability to sustain alertness and to disregard and suppress interfering information.
In relation to the aperiodic EEG components, we also found that both exponent and offset values significantly decreased with age.These results corroborate previous evidence suggesting that the aperiodic EEG component can serve as a neurophysiological marker of aging.Likewise, recent studies have revealed that aperiodic activity is influenced by various factors, including drugs 37,38 and level of arousal 39 .However, the potential mediation of education, and more specifically its influence on the relationship between aperiodic components and cognitive performance, was unexplored in the literature.
In our study, education might help in interpreting the relationship between the aperiodic component and performance on some tasks of visual attention and working memory, but not on a delayed memory task.The relation between aperiodic component and cognitive performance varied depending on education level, with a reversed pattern between exponent and cognitive performance in older adults across higher vs lower education.Older adults with lower education displayed a positive relationship between exponent and cognitive performance, while those with higher education exhibited the opposite trend.In this context, research evidence suggests that Figure 2. Relationship between exponent and cognitive performance on different tasks.On the upper side of the panel, response times in the visual attention and alertness tasks (processing speed) are reported on the leftand right-hand sides respectively.On the lower panel, accuracy in the working and delayed memory tasks are reported on the left-and right-hand sides respectively.On the x-axis, the exponent value parameterized at the occipital level and in the Alpha band (8-12 Hz) is reported.On the y-axis, the z-scores associated with the task outcome.
Vol www.nature.com/scientificreports/low exponent values (when the exponent approximates zero) may reflect an increase in asynchronous background neuronal firing, commonly called neural noise 32 .Related to the concept of neural noise, in non-linear systems like the brain, the notion of stochastic resonance proposes that information at the threshold level can be better processed within an optimal noise range than without noise 40 .If different exponent values represent varying levels of neural noise, it is possible that noise also has different effects on performance according to a specific system.In older adults with lower education levels, higher exponents-corresponding to lower noise values-may contribute to better performance.On the other hand, older adults with higher education would exhibit the opposite pattern.In this latter group, higher exponents (lower noise values) would reduce performance efficiency.These two scenarios may depend on the fact that, according to the framework of stochastic resonance, there is no ideal level of noise and its effect on performance may not follow a linear pattern: it can vary based on the specific system and compensatory dynamics.Although such a result may seem counterintuitive, a similar reversed pattern has been observed in a previous study that examined the relationship between mathematical achievement and glutamate concentrations.Glutamate has the effect of flattening the power spectrum, leading to exponent values closer to zero 37 .In a previous study 41 , it was demonstrated that the concentration of glutamate and the exponent levels could result in reversed cognitive performance outcomes depending on the participants' age.Specifically, the authors found that the concentration of glutamate (in the intraparietal sulcus) was negatively associated with mathematical achievement in younger participants, but it was positively associated with mathematical achievement in older participants.Given a possible relationship between glutamate and exponent levels 37 , these findings may be interpreted as follows: while in younger participants high levels of noise (corresponding to a lower level of glutamate and, consequently, higher exponent values) may reduce performance, in older participants high levels of noise may lead to an opposite effect, contributing to improving cognitive performance.In summary, these results, similar to ours, imply that the relationship between exponent, noise, and cognitive performance may not be straightforward, highlighting the importance of investigating possible mediators, such as education, within this complex relationship.
While tantalizing, considering education as a possible mediator in the relationship between the aperiodic component and cognitive performance may present pitfalls because education may introduce several other aspects that affect performance differently.For example, education may impact cognitive strategies, task engagement, and compensatory mechanisms, leading individuals with higher education to have a better cognitive performance with different potential explanations of the observed effect in the aperiodic EEG component.A more comprehensive definition about how this effect can be attributed to potential compensation mechanism could require further investigations.
We did observe a relationship between the exponent and processing speed, in line with a previous study 42 .Moreover, the results of the present study partially replicated what was found in some previous studies where a relationship between exponent and working memory performance was identified 15,17 .The lack of effects of aperiodic component in delayed memory task performance is of interest, as it suggests that the modulating role of aperiodic component may not be non-specific and happening for general cognitive functioning, but only for some aspects of cognitive functioning, possibly related to those case in which there is much time pressure in cognitive performance (as psychomotor or working memory tasks).
Overall, our results cannot be interpreted as exhaustive; they should emphasize the importance of considering the aperiodic component of EEG signal as a marker of neurophysiological mechanisms that relate to performance in some cognitive tasks, which can be mediated by different aspects.Our study, in particular, focused on education as one of these aspects.An important limitation is related to the fact that the LEMON database, despite having many advantages, did not have the optimal characteristics for the aims of this study.In particular, it included a cohort of participants with different ages (whereas a longitudinal dataset would have been more suited) and it included a different number of participants for each group, with a larger size for the group of younger adults as compared to the older adults.In this study, we focused on the occipital ROI.This decision was based both on previous scientific literature which led to the expectation of dominant age-related patterns in the alpha domain 8,10,[12][13][14] and to the limited spatial resolution of high-density EEG in which electrode localization data were only partially available for this dataset 43 .The additional exploratory analysis on frontal brain regions supported the potential interaction between EEG measures, education and cognitive performance that needs to be further explored through techniques with better spatial resolution 44 .
Future studies with better stratification might explore the ontogenetic trajectory of the exponent, to further investigate its role in cognitive performance across different tasks during aging.In fact, in the present study, the availability of age and education variables in a categorical form might have limited the assessment of neurobehavioral relationships and the potential use of finer analysis modeling (i.e.age was included as a factor rather than a continuous variable).
Future studies may explore the intricate connection between EEG parameters and cognition, by encompassing a broader range of variables that could modulate such a relationship, such as life experience variables, or others associated with physical health and physical activity, or to other proxies that can be traced back to the concept of "cognitive reserve", which may be crucial in understanding the complex relationship between cognitive and brain aging.Finally, it is important to stress a limitation (intrinsic to cross-sectional and quasi-experimental studies), that is the impossibility to infer cause-effect relationship.In all cases, the associations observed between EEG spectral parameters and performance should not be interpreted as evidence of causal relationships, but rather as a statistical association in which the directionality is not known and that could be mediated also by other aspects.
In summary, results from this study opens many question that may guide future research on the modulatory role of education and other cognitive reserve proxies, in the complex relationship between aperiodic EEG component and cognitive efficiency in aging.

Figure 1 .
Figure 1.Age-and Education-related differences of the EEG components.The Individual Alpha Peak Frequency, power, exponent, and offset of each group (young -high edu = younger adults with high education, old -high edu = older adults with high education; old -low edu = older adults with low education) are shown on the y-axis.

Figure 3 .
Figure 3. Age-and education-related differences in resting EEG power spectra in the occipital region, in the eyes closed condition.The plot shows a median for each group (young-high edu = younger adults with high education, old-high edu = older adults with high education; old-low edu = older adults with low education) and a 50% percentile interval, ranging from the 25 to 75 percentiles.

Table 1 .
Descriptive information about the sample and group comparisons based on participants' age and education.