Impact of Aging on Theta-Phase Gamma-Amplitude Coupling During Learning: A Multivariate Analysis

Aging is associated with cognitive decline and memory impairment, but the underlying neural mechanisms remain unclear. Phase-amplitude coupling (PAC) between mid-frontal theta (5 Hz) and occipital gamma (>30 Hz) oscillations is a proposed marker for parallel storage of multiple items in working memory. However, research has mainly focused on young individuals and epilepsy patients, with only a few studies on aging populations. Moreover, these studies have relied on univariate PAC methods, which can be flawed by potential spurious or biased PAC estimates due to non-stationarity of EEG signals. Additionally, these methods typically assess PAC at the level of individual electrodes, potentially overlooking the broader functional significance of theta-gamma coupling in coordinating neural activity across distant brain regions.

To address these gaps, we employed multivariate PAC (mPAC) through generalized eigendecomposition (GED) analysis, which avoids the pitfalls of non-sinusoidal oscillations. 113 young and 117 older healthy participants engaged in a sequence learning paradigm (6423 sequence repetitions, 55’944 stimuli), in which they learned a fixed sequence of visual stimuli over repeated observations, allowing us to track the mPAC during the incremental process of learning.

Behavioral results revealed that younger participants learned significantly faster than older participants. Neurophysiological data showed that mPAC increased over the course of learning in both age groups and could identify fast and slow learners. However, older participants exhibited lower mPAC compared to younger counterparts, which suggest compromised parallel storage of items in working memory in older age. Finally, stratification analysis revealed that mPAC effects persist across performance groups with similar mid-frontal theta levels, suggesting that theta alone does not account for these effects. These findings shed light on the age-related differences in memory formation processes and may guide interventions to enhance memory performance in older adults and slow learners.