Periodic and aperiodic contributions to theta‐beta ratios across adulthood

Abstract

The ratio of fronto-central theta (4-7 Hz) to beta oscillations (13-30 Hz), known as the theta-beta ratio, is negatively correlated with attentional control, reinforcement learning, executive function, and age. Although theta-beta ratios have been found to decrease with age in adolescents and young adults, theta has been found to increase with age in older adults.Moreover, age-related decreases in individual alpha peak frequency and flattening of the 1/f aperiodic component may artifactually inflate the association between theta-beta ratio and age. These factors lead to an incomplete understanding of how theta-beta ratio varies across the lifespan and the extent to which variation is due to a conflation of aperiodic and periodic activity. We conducted a partially preregistered analysis examining the cross-sectional associations between age and resting canonical fronto-central theta-beta ratio, individual alpha peak frequency, and aperiodic component (n = 268; age 36-84, M = 55.8, SD = 11.0).Age was negatively associated with theta-beta ratios, individual peak alpha frequencies, and the aperiodic exponent. The correlation between theta-beta ratios and age remained after controlling for individual peak alpha frequencies, but was non-significant when controlling for the aperiodic exponent. Aperiodic exponent fully mediated the relationship between thetabeta ratio and age, although beta remained significantly associated with age after controlling for theta, individual peak alpha, and aperiodic exponent. Results replicate previous observations and show age-related decreases in theta-beta ratios are not due to age-related decreases in individual peak alpha frequencies but primarily explained by flattening of the aperiodic component with age.

Resting EEG power spectra across middle to late life: Associations with age, cognition, APOE-ε4 carriage and cardiometabolic burden

“…However, we did not detect a statistically significant association between age and the aperiodic exponent, with the Bayes factor indicating moderate support for the null hypothesis ( r = -0.12, p = .287, BF 10 = 0.24, Figure 2A). Please note that these are results of pairwise correlations which do not include other variables (such as cognitive performance) as covariates, following the approach in previous work (e.g., Voytek et al, 2015; Finley et al, 2022).…”

“…Using this approach, they have reported decreases in power across the delta (<4Hz) and alpha (7-13Hz) bands with increased age, as well as a slowing of alpha peak frequency (e.g., Celesia, 1986; Klimesch, 1997; Babiloni et al, 2006b; Vlahou et al, 2015; Sghirripa et al, 2021; reviewed in Scally et al, 2018). Age-related increases in theta band power across mid- to late-life have also been reported, however these changes have been attributed to brain activity typically associated with the alpha-band occurring at slower frequencies in older adults (reviewed in Klimesch, 1999; Finley et al, 2022).…”

“…the rate of decrease in power with increasing frequency, here termed the aperiodic slope) also appears to be closely linked to band power ratio measures used in resting EEG research (Donoghue et al, 2020b). For example, Finley et al (2022) reported that the ratio of power in the theta compared to the beta band was lower in older adults, but this association was largely due to changes in the shape of the aperiodic component. This example underlines the importance of separately modeling aperiodic and periodic components of Fourier power spectra (for further discussion see Donoghue et al, 2020a).…”

“…Almost all of the studies listed above compared younger and older groups of markedly different age ranges (e.g., 20-30 year-olds compared to 60-70 year-olds in Voytek et al, 2015). Studies treating age as a continuous variable have recruited very small samples (n=15 in Voytek et al, 2015, n=19 in Waschke et al, 2017), making it difficult to assess the functional form of age-aperiodic slope associations (but see Finley et al, 2022 for an approximately linear association in a large sample of 36-84 year-olds). Importantly, flatter aperiodic slopes have also been reported in individuals with worse cognitive performance (Voytek et al, 2015; Ouyang et al, 2020), and cognitive performance in many tasks steadily declines with increasing age.…”

Intellectually able adults with autism spectrum disorder show typical resting-state EEG activity

“…It is well-established that some EEG features have an age-dependent effect. Both peak alpha frequency 91 , theta/beta ratio 92 and 1/f exponents have been found to be lower in old compared to young adults 78 , 93 – 95 . DFA exponents have also been found to increase from childhood into early adulthood before it stabilizes 96 .…”