Resting state EEG mediates the association between physical activity and cognitive function in cognitively impaired elderly

Abstract

With the acceleration of the aging society, the incidence of cognitive impairment in the elderly population continues to rise. Identifying the underlying neurobiological mechanisms and demographic factors is of great significance in delaying cognitive decline. In recent years, electroencephalography (EEG), as an objective indicator reflecting brain functional states, has become an important tool for studying cognitive function. Meanwhile, the potential benefits of physical activity on cognitive health have also gained widespread attention. This study aims to explore the relationship between cognitive function and brain electrical activity in elderly individuals, analyze the direct and indirect effects of physical activity on cognitive function, and identify the potential mediating role of demographic variables on cognitive status. Cognitive function screening was conducted on 209 community-dwelling elderly individuals. The Montreal Cognitive Assessment (MoCA) scale was used to evaluate cognitive levels, while resting-state EEG was used to measure multi-region and multi-frequency neural activity. Data on physical activity levels and sociodemographic variables were also collected. Differences were tested, and Pearson correlation analysis and mediation models were employed to assess the horizontal relationships and pathways between variables. Significant differences were observed between elderly individuals with varying levels of cognitive impairment in age, marital status, educational attainment, previous occupation, dietary habits, and place of residence (p < 0.05). MoCA scores were negatively associated with theta power, particularly in the frontal (Fp1, Fp2, F4) and central (C4) regions, and were also negatively correlated with increased beta and alpha power in selected regions (p < 0.05). Physical activity were positively associated with MoCA scores and negatively correlated with theta power at Fp1 and Fp2, as well as beta1 and beta2 power at F4, C4, and O2 (p < 0.05). EEG indices jointly associated with both physical activity and cognitive function included theta power at Fp1, Fp2, F4, C4, and O2, and beta1/beta2 power at F4, C4, and O2 (p < 0.05). Mediation analyses further indicated that specific EEG markers—namely beta2 power at Fp2 and theta power at F4—may partially mediate the relationship between physical activity and cognitive function (p < 0.05). Cognitive function in elderly individuals is influenced by multiple factors, including demographic characteristics, neural activation patterns, and physical activity. Resting-state EEG markers—particularly frontal theta power (e.g., Fp1, Fp2) and alterations in beta power across frontal, central, and occipital regions—may serve as potential biomarkers for cognitive status. Physical activity may enhance cognitive performance in older adults with cognitive impairment partly by modulating these specific neural markers, such as beta2 power at Fp2 and theta power at F4.

Introduction

With the global aging population, the incidence of cognitive dysfunction in the elderly continues to rise, becoming an increasingly severe public health issue worldwide. According to the Global Burden of Disease study, there were 57.4 million dementia patients worldwide in 2019, and this number is expected to exceed 150 million by 2050^1,2. Cognitive impairment is one of the early signs of dementia, with common manifestations including mild cognitive impairment (MCI). Clinical features of MCI include decreased attention, memory decline, slowed information processing speed, and executive function deficits^3,4. If left untreated, these symptoms may progress to neurodegenerative diseases such as Alzheimer’s disease⁵. In China, with its large population and rapid aging, it is projected that by 2050, approximately 40 million elderly individuals will face varying degrees of cognitive impairment or dementia, presenting significant prevention and control challenges⁶.

Electroencephalography (EEG) is a widely used non-invasive technique that is commonly applied in cognitive neuroscience research. It reflects changes in brain function by measuring brain electrical activity. Due to its high temporal resolution, convenient data collection method, and relatively low cost, EEG has unique advantages for studying brain disorders that commonly occur in older adults, such as cognitive decline⁷. EEG effectively captures brain electrical activity related to cognitive states, emotional responses, and neurodegenerative diseases, making it an essential tool for exploring cognitive impairment, emotional fluctuations, and other brain function changes⁸. In recent years, with the continuous development and improvement of EEG technology, spectral analysis has become an indispensable part of EEG data processing⁹. By analyzing the power spectrum of EEG signals across different frequency bands, researchers can gain a deeper understanding of changes in brain function states¹⁰. Previous studies have reported significant EEG spectral characteristics in cognitive impairment, Evidence indicates that diminished alpha power is significantly associated with greater cognitive impairment^11,12. In addition, a decrease in beta power, particularly in the beta1 band, has been linked to declines in executive function, attention deficits, and memory deterioration¹³. Moreover, a reduction in beta wave power in the parietal region is often considered a physiological marker of cognitive decline, reflecting the progression of cognitive impairment¹⁴. Numerous studies have shown that engaging in regular physical activity for extended periods is a feasible and effective non-pharmacological intervention¹⁵, significantly reducing the incidence of cognitive impairment in the elderly and delaying its progression. The potential mechanisms of this effect are complex and multi-faceted, primarily involving the following aspects: First, regular aerobic physical activity can promote the expression of neurotrophic factors, such as brain-derived neurotrophic factor, enhancing neuronal plasticity and synaptic function, thereby improving cognitive flexibility and executive function¹⁶. Second, research has confirmed^17,18 that physical activity can interact with the brain through myokines secreted by muscles in a “muscle-brain axis,” triggering anti-inflammatory signaling pathways and reducing systemic inflammation, which plays a positive role in slowing neurodegeneration. At the brain network level, prolonged moderate-intensity physical activity can enhance the coupling relationship between the fronto-parietal network and the default mode network¹⁹, thus improving the efficiency of brain information processing.

Our research team has preliminarily confirmed that physical fitness in the elderly (e.g., muscle strength) can have a cross-sectional impact on cognitive function⁷, and has observed an interrelationship between physical activity, muscle strength, and cognitive ability^20,21. However, despite increasing evidence indicating that physical activity is closely related to cognitive benefits during the aging process, there remain many gaps in understanding the potential neurobiological mechanisms in elderly individuals with cognitive impairment. Past studies have started to explore the role of electroencephalography (EEG) as a potential biomarker. For example, Sanchez-Lopez et al.²², found that high levels of incidental physical activity in healthy elderly individuals were associated with more favorable EEG characteristics (such as reduced low-frequency power) and better cognitive performance. Trenado et al. further noted²³ that the combination of cognitive training and physical activity can modulate the theta and alpha wave energy in individuals with mild cognitive impairment, suggesting the practical value of EEG in tracking intervention effects. Recently, Liu et al.²⁴ discovered that specific EEG markers, such as FP1 theta and T4 alpha2, play an important role in mediating the relationship between physical function and cognition in elderly individuals with cognitive impairment.

Despite this, existing studies still have several significant limitations: On the one hand, most research focuses on organized activity or physical health, lacking exploration of habitual physical activity and neurophysiological factors in elderly individuals with cognitive impairment who are not demented and live in the community. On the other hand, although EEG parameters are correlated with cognitive or physical function indicators, systematic studies on their mediating role in the physical activity-cognition pathway are still scarce. Moreover, few studies simultaneously model EEG frequency bands across multiple brain regions (such as theta, alpha, beta waves) to identify which specific oscillatory activities play a key role in these processes. Based on these limitations, this study adopts an observational design aimed at exploring and verifying the lateral relationship between physical activity and cognitive function, as well as the underlying neurophysiological mechanisms. By combining subjective questionnaire data with multi-channel resting-state EEG, the mediating role of EEG spectral characteristics in this process will be further investigated. Therefore, the following research hypotheses are proposed: (1) The amount of physical activity is positively correlated with cognitive performance and negatively correlated with certain spectral energy values in the prefrontal and central regions; (2) EEG parameters related to these correlations play a significant mediating role in the relationship between physical activity and cognitive function. This study aims to explore specific brain electrical oscillations (such as theta, alpha, and beta waves) as key neuroelectrophysiological markers, providing a deeper understanding of the physical activity-cognition lateral relationship and offering theoretical support for the development of precise physical activity intervention strategies for elderly individuals with cognitive impairment.

Research subjects and methods

Research subjects

This study utilized G*Power software for a priori sample size estimation. For comparative analyses, the statistical test type selected was “Means: Difference between two independent means (two groups)”. With an effect size (Cohen’s d) set at 0.5, a statistical power of 0.80, and a significance level (α) of 0.05, the calculation indicated a minimum required sample size of 128 participants.For testing the hypothesized mediation model, the statistical test selected was “Linear multiple regression: Fixed model, R² increase”. The parameters were set as follows: an effect size (f²) of 0.15, a statistical power of 0.95, and a significance level (α) of 0.05. Regarding predictor variables, the “Number of tested predictors” was set to 1, representing the addition of the single mediator variable (i.e., a specific EEG Indices) in each test. The “Total number of predictors” was set to 2, corresponding to all predictors in the full model (MoCA ~ Physical Activity + EEG Indices). This calculation yielded a minimum required sample size of 89 participants.

To adequately address the research objectives and account for potential attrition or data exclusion, a convenience sampling approach was employed. Recruitment was conducted at four senior service centers in Shanghai through health promotion talks and the posting of participant recruitment advertisements. Following the principle of voluntary participation, a total of 232 older adults were recruited as study participants.This study adhered to the ethical principles outlined in the latest version of the Declaration of Helsinki and received ethical approval from the National Social Science Fund project (Approval No: 22BTY076).

Inclusion Criteria: (1) Elderly individuals aged 60 years or older; (2) Diagnosed by professional researchers with a Montreal Cognitive Assessment (MoCA) score < 26;(3) Right-handed; (4) In good physical health; (5) No severe cardiovascular or cerebrovascular diseases or major organic diseases; (6) No severe muscular diseases or contraindications to physical activity; (7) Normal vision and hearing; (8) Normal mental state, no psychiatric history, and no use of psychotropic medications; (9) Ability to communicate verbally and cooperate with the survey; (10) Willingness to sign an informed consent form.Exclusion Criteria: (1) Subjects who withdraw from the experiment; (2) Excessive EEG artifacts preventing data analysis.The subject recruitment process is shown in Fig. 1.

Fig. 1.

Recruitment Process of Research Subjects.

Testing procedure

This study employed an observational research design, and all measurements were completed at a single time point. All tests were scheduled between 13:30 and 16:30 each afternoon to control for the potential impact of diurnal biological rhythm fluctuations on cognitive function and EEG data. Each participant was required to visit the laboratory twice to complete all the measurement tasks for the study. During the first visit, the researchers provided the participants with a detailed explanation of the overall experimental process and important instructions, ensuring they fully understood the research purpose and participation requirements. This process was expected to last approximately 15 min. Afterward, participants signed an informed consent form, and the staff assisted all participants in completing a basic information form and questionnaire within 20 min. Participants were also informed of the following pre-test requirements for the second EEG session: (1) No vigorous physical activity within 24 h before the test; (2) No consumption of caffeinated or alcoholic beverages on the test day; (3) Avoid taking medications that affect the central nervous system; (4) Maintain normal sleep patterns and avoid staying up late.

During the second visit, participants underwent a one-time 5-minute resting-state EEG data collection. Prior to data collection, two professional technicians applied conductive gel to the scalp and placed the EEG cap, which took approximately 10 min. The entire EEG testing process was conducted in a quiet, dimly lit, and non-interfering environment to ensure participants were relaxed and in a stable mental state. The whole EEG test and preparation process was expected to last about 20 min. After all tests were completed, data for the included participants were organized and analyzed. The testing procedure flowchart is shown in Fig. 2.

Fig. 2.

Testing Procedure.

Testing tools

Basic information questionnaire

The basic information questionnaire includes the following items: age, gender, height, weight, marital status, previous occupation, education level, smoking habits, alcohol consumption, dietary habits, and residence.

Physical activity rating scale-3 (PARS-3)

The Physical Activity Rating Scale-3 (PARS-3) revised by Professor Liang Deqing and colleagues was used in this study. This scale evaluates physical activity participation based on three components: activity intensity, duration, and frequency. It is simple to use, reliable, and widely employed to assess activity participation levels.Calculation Method for PARS-3:Thetotal score for physical activity is calculated as: Total activity score = Intensity × Duration × Frequency.Each component has five levels: frequency and intensity are scored from 1 to 5, and duration is scored from 0 to 4. The highest possible score is 100, and the lowest is 0^25–27.

Montreal cognitive assessment (MoCA)

The MoCA scale is widely used to assess cognitive function in elderly individuals. It consists of 8 domains: visuospatial/executive function, naming, memory, attention, language fluency, abstract thinking, delayed recall, and orientation, with a total score of 30 points. The higher the score, the better the cognitive function.The MoCA total score ranges from 0 to 30: a score of 26 or higher indicates normal cognitive function²⁸, 18 to 26 points indicates mild cognitive impairment (MCI), 10 to 17 points indicates moderate cognitive impairment, and below 17 points indicates severe cognitive impairment⁷. To account for the effect of education level, participants with ≤ 12 years of education have 1 point added to their total score. If the total score exceeds 30 after the adjustment, no further addition is made²⁰. In this study, participants were divided into two groups based on their MoCA scores to explore differences in demographic variables: the mild cognitive impairment group (MoCA score 18–25) and the moderate-to-severe cognitive impairment group (MoCA score ≤ 17).

EEG signal acquisition and processing

The EEG test was conducted in a soundproof, well-ventilated laboratory with soft lighting, appropriate temperature and humidity, and no electromagnetic interference. Prior to testing, participants cleaned and dried their hair, seated themselves, and adjusted to a comfortable posture. During the experiment, participants were instructed to close their eyes, remain relaxed, quiet, and awake, and to avoid blinking, swallowing, clenching their teeth, or body movements in order to reduce movement artifacts.EEG signals were recorded using a system manufactured by Shanghai Nuocheng Electric Co., Ltd. (Model: NCERP-190012). Sixteen electrodes (Fp1, Fp2, F3, F4, C3, C4, P3, P4, O1, O2, F7, F8, T3, T4, T5, T6) were placed according to the international 10–20 system, with a ground electrode at the center of the forehead (GND) and reference electrodes placed at the bilateral mastoid (A1, A2). The impedance between the electrodes and the skin was kept below 5 kΩ to ensure signal stability and a high signal-to-noise ratio. The bandwidth of the preamplifier was set to 500 Hz to effectively filter out high-frequency noise (e.g., electromyographic noise) and ensure the quality of the EEG signal. After amplification, data were acquired at a sampling frequency of 1000 Hz, followed by a 0.1–30 Hz band-pass filter to remove low-frequency drift and high-frequency noise. Each participant had 5 min of closed-eye resting-state EEG data collected. During this resting-state period, participants remained relaxed and awake without engaging in any verbal or physical activity to minimize external interference and non-brain signal inputs. The closed-eye state helps reduce the suppression effect of visual input on alpha waves, contributing to a more stable spontaneous brain rhythm.

All EEG signals were preprocessed using the EEGLAB toolbox on the MATLAB 2023b platform. The preprocessing steps are as follows: (1) Filtering and Re-referencing: A 0.1–30 Hz band-pass filter was first applied to remove low-frequency drift (such as skin conductance changes and instrument baseline instability) and high-frequency noise (such as electromyographic activity and power line interference). The filtered signals were then re-referenced to a bilateral mastoid average reference (A1, A2) to ensure a more stable baseline and provide clearer EEG signals for subsequent analysis. (2) Manual Artifact Removal: Segments with amplitude exceeding ± 100 µV or containing non-physiological waveforms, such as spikes, jumps, or baseline drift, were visually inspected and removed. (3) Independent Component Analysis to Remove Non-brain Sources: The Infomax algorithm was used for independent component decomposition. Based on the scalp distribution characteristics, time-domain waveforms, and power spectrum features, eye movements, blinks, facial electromyographic (EMG) activity, and electrocardiographic artifacts were identified and removed. To further improve signal purity, low-amplitude, low-frequency EMG artifacts were manually inspected and removed, in order to reduce interference from facial and temporal muscle micro-activity caused by emotional tension or other states. (4) Data Quality Control: If more than 5% of channels were removed from the EEG signal or if the remaining valid data length was less than 80% of the total collected duration, the sample was excluded from further analysis.For EEG power spectrum calculation, this study used the Fast Fourier Transform (FFT) algorithm. The frequency bands were defined based on conventional standards: δ (1–4 Hz), theta (4–8 Hz), alpha1 (8–10.5 Hz), alpha2 (10.5–13 Hz), beta1 (13–20 Hz), and beta2 (20–30 Hz). Given the variability in individual alpha frequency²⁹, a preliminary check of each participant’s alpha peak was conducted before the power spectrum analysis to ensure that the alpha peak frequency for the study sample was within the conventional range of 8–12 Hz. To maintain consistency with previous studies and ensure comparability²⁴, fixed frequency band definitions were used for subsequent statistical analysis.

Statistical analysis

SPSS 29.0 software was used for statistical analysis. Continuous data were presented as mean ± standard deviation, with results rounded to two decimal places. Between-group comparisons were conducted using independent samples t-test. Categorical data were described using frequencies (n), and between-group comparisons were performed using the χ² test. Pearson correlation analysis was employed to explore the bidirectional relationships between PARS-3, cognitive function, and EEG indicators. The mediation effects of specific EEG indicators on the relationship between physical activity and cognitive function were analyzed using Model 4 in the PROCESS macro. Path analysis parameter estimates were derived using the bootstrap method with 5,000 resamples. A 95% confidence interval (CI) not including zero was considered indicative of a statistically significant mediation effect. Two-tailed tests were used for statistical inferences, with the significance level set at α = 0.05, and p < 0.05 was considered statistically significant.

Results

Demographic information of included participants

A total of 209 elderly participants were included in the study after screening with the Montreal Cognitive Assessment and Electroencephalography. A differential test was performed between the mild cognitive impairment group and the moderate-to-severe cognitive impairment group (Table 1). The results showed significant differences between the two groups in age (t = 22.177, p < 0.001)… and residence (χ² = 3.918, p < 0.05). In summary, the mild impairment group was significantly younger and had a higher educational attainment compared to the moderate-to-severe group. Significant differences were also observed in marital status, prior occupation, dietary habits, and residence. No significant differences were observed in other variables (p > 0.05).

Table 1.

Differences in basic information among elderly individuals with different cognitive function Scores.

Variables		Cognitive performance stratification		Comparative analysis
	Total (209)	Mild (n = 114)	Moderate/Severe (n = 95)	t/χ2	p
MoCA	17.50 ± 5.24	21.41 ± 2.25	12.80 ± 3.71	19.80	<0.001
Age	70.40 ± 8.08	68.11 ± 7.22	73.15 ± 8.23	22.177	<0.001
BMI	24.14 ± 3.42	24.42 ± 3.23	23.81 ± 3.63	1.630	0.203
Gender (male%)	87(41.6%)	48(42.1%)	39(41.1%)	0.024	0.878
Occupation(farmer%)	101(48.3%)	40(35.1%)	61(64.2%)	17.599	<0.001
Marital (married%)	182(87.1%)	107(93.9%)	75(78.9%)	10.243	<0.001
Education(illiterate%)	37(17.7%)	9(7.9%)	28(29.5%)	28.273	<0.001
Smoking (smoker%)	56(26.8%)	33(28.9%)	23(24.2%)	0.593	0.441
Drinking (drinker%)	58(27.8%)	33(28.9%)	17(17.9%)	1.123	0.772
Residence (rural%)	112(53.6%)	54(47.4%)	58(61.1%)	3.901	0.048
Diet (plant-based%)	54(25.8%)	22(19.3%)	32(33.7%)	9.039	0.011

Correlation between cognitive function scores and EEG indicators in elderly individuals with cognitive impairment

After controlling for confounding factors such as age, marital status, previous occupation, education level, dietary habits, and residence, the Pearson correlation analysis results (Table 2) showed significant negative correlations between MoCA scores and multiple EEG indicators in elderly individuals with cognitive impairment. Specifically, the MoCA score was negatively correlated with the following EEG indicators: Significant negative correlations were observed between MoCA scores and spectral power across multiple EEG rhythms and brain regions, with coefficients ranging from r = −0.137 to r = −0.296. Theta rhythm demonstrated the most robust and widespread associations, with the strongest correlations notably localized to the frontal areas (e.g., F4 theta: r = −0.296, p < 0.01; FP1 theta: r = −0.292, p < 0.01). Additionally, significant negative correlations were also identified for beta and alpha2 bands at various sites. These results indicate that higher cognitive performance is associated with lower spectral power, particularly in the theta frequency.

Table 2.

The correlation between MoCA scores and EEG Indices.

	delta	theta	alpha1	alpha2	beta1	beta2
FP1	−0.025	−0.292**	−0.015	0.029	−0.027	0.001
FP2	−0.017	−0.254**	−0.074	−0.087	−0.214**	−0.292**
F3	0.004	−0.123	0.005	0.064	−0.029	−0.103
F4	−0.063	−0.296**	0.010	−0.017	−0.048	−0.028
C3	−0.038	−0.129	0.051	0.022	−0.012	−0.098
C4	−0.077	−0.226**	0.001	−0.003	−0.056	−0.108
P3	−0.064	−0.250**	−0.076	−0.184**	−0.109	−0.132
P4	−0.086	−0.286**	−0.044	−0.177**	−0.104	−0.194**
O1	0.049	−0.102	−0.011	0.006	0.016	−0.074
O2	−0.075	−0.223**	−0.031	−0.003	−0.036	−0.146*
F7	0.026	−0.207**	−0.060	−0.098	−0.088	−0.158*
F8	−0.124	−0.254**	−0.049	−0.166*	−0.137*	−0.142*
T3	0.015	−0.050	0.022	0.068	−0.091	−0.110
T4	−0.063	−0.089	0.011	0.097	0.020	−0.016
T5	0.001	−0.109	−0.004	0.049	−0.028	−0.099
T6	−0.008	−0.169*	0.079	0.108	−0.020	−0.137*

**p < 0.01, *p < 0.05.

Relationship between physical activity volume and cognitive function scores in elderly individuals with cognitive impairment

After controlling for confounding factors such as age, marital status, previous occupation, education level, dietary habits, and residence, Pearson correlation analysis was used to examine the relationship between physical activity volume and cognitive function (Fig. 3). The results showed a significant correlation between cognitive function and total physical activity volume (r = 0.436, p < 0.001).

Fig. 3.

Correlation between physical activity levels and cognitive function scores in older adults with cognitive impairment.

Correlation between physical activity volume and EEG indicators in elderly individuals with cognitive impairment

After controlling for confounding factors such as age, marital status, previous occupation, education level, dietary habits, and residence, Pearson correlation analysis was used to examine the relationship between total physical activity volume and EEG indicators in elderly individuals with cognitive impairment (Table 3). The results showed significant negative correlations between total physical activity volume and the following EEG indicators: FP1 theta (r = −0.208, p < 0.01), FP2 theta (r = −0.213, p < 0.01), beta1 (r = −0.161, p < 0.05), beta2 (r = −0.140, p < 0.05), F4 theta (r = −0.210, p < 0.01), C4 theta (r = −0.247, p < 0.01), and O2 theta (r = −0.138, p < 0.05). These results suggest that higher levels of physical activity are associated with lower activation levels in these EEG indicators.

Table 3.

The correlation between physical activity scores and EEG Indices.

	delta	theta	alpha1	alpha2	beta1	beta2
FP1	−0.092	−0.208**	−0.061	−0.023	−0.061	−0.023
FP2	−0.110	−0.213**	−0.120	−0.126	−0.161*	−0.140*
F3	0.084	−0.131	−0.045	−0.001	−0.052	−0.006
F4	−0.087	−0.210**	−0.066	−0.073	−0.095	−0.032
C3	−0.057	−0.040	0.002	−0.009	0.002	−0.024
C4	−0.123	−0.247**	−0.039	−0.024	−0.082	−0.016
P3	−0.091	−0.085	−0.089	−0.096	−0.010	0.019
P4	−0.080	−0.115	−0.079	−0.128	−0.084	−0.067
O1	−0.033	−0.005	−0.073	−0.014	0.014	0.014
O2	−0.050	−0.138*	−0.125	−0.037	−0.040	−0.057
F7	−0.072	−0.089	−0.041	−0.021	−0.011	−0.018
F8	−0.130	−0.119	−0.063	−0.113	−0.082	−0.019
T3	−0.010	0.009	−0.025	−0.014	−0.078	−0.011
T4	−0.101	−0.057	−0.091	−0.052	−0.091	−0.019
T5	0.072	0.022	−0.060	0.020	0.029	0.024
T6	−0.015	−0.089	−0.025	0.033	−0.031	−0.021

**p < 0.01, *p < 0.05.

Structural relationship model of physical activity volume, cognitive function, and EEG specific indicators in elderly individuals with cognitive impairment: construction and validation

Based on the correlation analysis between EEG indicators, physical activity, and cognitive function (Tables 2 and 3), it was found that the EEG indicators jointly affected by physical activity and cognitive function include FP1 theta, FP2 theta, beta1, beta2, F4 theta, C4 theta, and O2 theta. These 7 EEG indicators may serve as specific markers for the joint effect of physical activity and cognitive function. Based on the interrelationships between physical activity, cognitive function, and specific EEG indicators in older adults with cognitive impairment, a mediation model was constructed to explore potential mediating pathways, while controlling for confounding factors including age, marital status, occupational history, education level, dietary habits, and place of residence. The results (Tables 4 and 5; Fig. 4) showed that physical activity negatively predicted FP2 beta2 (β = −0.140, p < 0.05), negatively predicted F4 theta (β = −0.210, p < 0.01), and positively predicted MoCA scores (β = 0.371, p < 0.01). FP2 beta2 negatively predicted MoCA scores (β = −0.202, p < 0.01), whereas F4 theta positively predicted MoCA scores (β = 0.173, p < 0.01).

Table 4.

Bootstrap analysis for significance testing of mediation Effects.

	Effect size	Boot SE	Bootstrap 95%CI
Total Effect	0.121	0.017	0.087, 0.156
Direct Effect	0.103	0.017	0.070, 0.137
Fp2 β2	−0.084	0.026	−0.135, −0.034
F4 θ	−0.244	0.015	−0.418, −0.071

Table 5.

Mediation analysis of EEG biomarkers in the physical activity-cognition relationship.

	Model1			Model2			Model3
	B	beta	t	B	beta	t	B	beta	t
Physical activity	−0.093	−0.140	−2.039	−0.041	−0.210	−3.087	0.103	0.371	6.057
FP2 beta2							−0.084	−0.202	−3.286
F4 theta							−0.244	−0.173	−2.782
R 2		0.020			0.044			0.272
F		4.159*			9.529**			25.508**

B indicates unstandardized coefficients. beta indicates standardized coefficients. Model 1 indicates Physical activity predicts FP2 beta2; Model 2 indicates Physical activity predicts F4 theta; Model 3 indicates Physical activity, FP2 beta2, and F4 theta jointly predict MoCA.beta indicates standardized coefficients. * indicates p < 0.05; ** indicates p < 0.01.

Fig. 4.

Path analysis of physical activity-cognition-EEG interactions.

Further analysis using the bias-corrected bootstrap method revealed that the total effect of physical activity on MoCA scores was 0.121 (95% CI 0.087, 0.156), and the direct effect was 0.103 (95% CI 0.070, 0.137). The mediation effect of FP2 beta2 was − 0.084 (95% CI −0.135, − 0.034), and that of F4 theta was − 0.244 (95% CI −0.418, − 0.071). Both the total and direct effects were statistically significant, and FP2 beta2 and F4 theta were identified as significant mediators (p < 0.05).

Discussion

This study conducted a cognitive function screening for 209 elderly individuals and analyzed their cognitive status using the Montreal Cognitive Assessment (MoCA) and electroencephalography (EEG). The results of the difference tests showed that there were significant differences in sociodemographic variables such as age, marital status, previous occupation, education level, dietary habits, and residence between different cognitive impairment groups. These findings may reflect the potential influence of socioeconomic and lifestyle factors on cognitive function in elderly individuals.Firstly, age, as an important factor in cognitive decline, may cause significant differences among elderly individuals with varying levels of cognitive impairment. With increasing age, the degradation of brain structure and function is often associated with the worsening of cognitive impairment. In the elderly population, higher age is typically accompanied by the onset of neurodegenerative diseases, further exacerbating cognitive decline^30,31. Additionally, marital status and previous occupation may also play important roles in cognitive function changes. As part of the social support system, marital status may affect an individual’s mental health and cognitive function. Studies show that individuals with spouses typically perform better in cognitive health, as the social support from a spouse can alleviate feelings of loneliness and stress, thereby supporting cognitive function maintenance³². Previous occupations may reflect an individual’s cognitive reserve. Certain high-cognitive-demand professions (e.g., education, research) may help delay cognitive decline in old age³³. Education level, as a long-term factor influencing cognitive development, also showed significant importance in this study. Higher educational levels typically correspond to stronger cognitive reserve and a lower risk of cognitive decline. This may be because individuals with higher education have accumulated more cognitive experiences and resources throughout their lives, which helps them maintain a higher cognitive level in the face of aging and degenerative diseases^34,35. Dietary habits and residence also have a close relationship with cognitive function. Studies show that healthy dietary habits, such as diets rich in antioxidants and omega-3 fatty acids, contribute to brain health and delay cognitive decline^36,37. The location of residence may reflect the impact of environmental factors on cognitive health, particularly between urban and rural elderly populations, where lifestyle, social support, and healthcare resources differ, leading to different cognitive function trends³⁸. In summary, these sociodemographic factors may influence the maintenance or decline of cognitive function through different mechanisms. Therefore, future cognitive intervention and assessment strategies should consider the combined effects of these factors and control for them in order to more effectively identify high-risk individuals and implement targeted interventions.

This study found that the MoCA scores of elderly individuals with cognitive impairment were significantly negatively correlated with several EEG indicators, particularly in the activities of theta, alpha, and beta waves. In other words, individuals with higher MoCA scores exhibited lower EEG activation levels during resting states. This finding strongly supports the neural efficiency hypothesis³⁸, suggesting that individuals with better cognitive function use brain activity more efficiently, with a more optimized allocation and utilization of neural resources^24,39. The neural efficiency hypothesis posits that individuals with better cognitive abilities typically show lower brain wave activation levels when performing cognitive tasks³⁹, particularly in low-frequency waves (such as theta) and high-frequency waves (such as beta) during resting-state EEG^40,41. Low EEG activation levels may reflect the brain’s optimized neural activity when processing information, reducing unnecessary energy consumption, thereby enhancing cognitive processing efficiency⁴¹. However, not all studies suggest that lower EEG power always correlates with better cognitive function. Some research indicates that lower brain activity may reflect inefficient or impaired brain function, particularly in the context of neurodegenerative diseases or cognitive impairments^42–44. For example, certain studies have found that low theta activity in Alzheimer’s patients may be associated with cognitive decline rather than being a sign of good cognitive function^44,45. Therefore, whether low brain wave activity always indicates the brain’s “energy-saving mode” remains an open question, potentially influenced by individual differences in physical condition, age, neural plasticity, and other factors.Nonetheless, the results of this study still support the neural efficiency hypothesis, particularly in the elderly group with better cognitive function. Specifically, lower theta activation is closely related to cognitive processing efficiency in the frontal regions, which are involved in higher cognitive functions such as decision-making, planning, and emotional regulation. During resting states, relatively low theta activity is associated with stronger executive control and less attention diversion⁴⁶. Thus, low theta activation may be a characteristic of individuals with better cognitive abilities⁴⁷. Additionally, the low activation of beta and alpha2 waves is also closely related to the resting-state EEG characteristics of individuals with better cognitive function. Beta waves are typically associated with attention focus, cognitive control, and executive functions⁴⁸, while alpha2 waves are closely linked to inhibitory control and the preparedness for information processing⁴⁹. In the P3 and P4 regions, the low activation of alpha2 waves may reflect the efficient energy utilization of the parietal regions during cognitive processing. The negative correlation between beta2 waves in the P4 region and MoCA scores might suggest that individuals with better cognitive function are able to regulate attention and alertness more effectively in specific regions, reducing interference from irrelevant neural activity, and thereby improving cognitive task performance^50,51. In conclusion, although the relationship between low EEG power and cognitive function may vary across studies, our results support the association between low EEG activation and individuals with better cognitive abilities, particularly during resting states. This characteristic may help the brain complete complex cognitive tasks with minimal energy expenditure, and by reducing unnecessary neural activation, assist cognitively strong individuals in maintaining or improving their cognitive abilities in a more efficient state.

This study also explored whether physical activity has a lateral predictive effect on the cognitive function of elderly individuals with cognitive impairment. The results showed a positive correlation between total physical activity and cognitive function, which is consistent with previous research⁵². Furthermore, this study verified that individuals with better cognitive function tend to have higher levels of physical activity, suggesting that assessing and improving physical activity behaviors in the elderly may help delay the progression of cognitive decline. Meta-analytic studies have indicated that moderate physical activity can significantly enhance cognitive abilities in elderly individuals with cognitive impairment, and the intervention effects are relatively stable^53,54. The potential mechanisms include multiple regulations of brain structure and function: physical activity not only enhances the volume of key structures like the hippocampus and cerebellum but may also promote cognitive processing efficiency by influencing brain activation levels and functional connectivity patterns⁵⁵. Neuroanatomical studies have also found that regular physical activity can effectively slow the decline in gray matter density in the elderly brain, particularly in cortical areas like the temporal, frontal, and parietal lobes, thus delaying brain atrophy⁵⁶. Moreover, this study found that higher levels of physical activity were significantly negatively correlated with the theta and beta (both beta1 and beta2) waves in multiple brain regions (such as the frontal area FP1, FP2; frontal area F4; central area C4; and occipital area O2). Specifically, individuals with higher levels of physical activity exhibited lower EEG activation levels in relevant brain regions during resting states, suggesting an energy-saving pattern of neural activity and improved brain function processing efficiency. This result supports the role of physical activity in optimizing brain function in elderly individuals with cognitive impairment⁵⁷, indicating that physical activity not only enhances the neural plasticity and functional connectivity of brain regions but may also promote efficient energy utilization in the brain during cognitive tasks⁵⁸. Additionally, this study proposed a hypothesis that EEG may have a mediating effect between physical activity and cognitive function. The results indicated that physical activity mediates the maintenance and enhancement of cognitive function by modulating specific brain region EEG indicators (specifically theta and beta wave activities). This finding not only confirms the potential benefits of physical activity for elderly individuals with cognitive impairment but also provides a new perspective on understanding the underlying neural mechanisms. Specifically, physical activity may induce an “energy-saving effect” in neural activity by influencing the theta and beta waves of brain regions, thereby delaying cognitive decline in elderly individuals.First, theta waves are closely related to the brain’s memory, attention, and emotional regulation. Previous studies have shown that enhanced theta wave activity may help complete cognitive tasks, especially in the context of age-related cognitive decline. Specifically, increased theta waves are associated with self-regulation, emotional regulation, and improved executive function^59,60, particularly in emotional regulation and attention maintenance. On the other hand, beta waves play an important role in motor control, task preparation, and execution in the brain. As people age, elderly individuals often show reduced beta wave activity, which is linked to the onset and progression of cognitive impairment⁶¹. However, existing research suggests that moderate physical activity can increase beta wave activity, and this enhanced beta wave activity, especially during the execution of complex cognitive tasks, may improve elderly individuals’ information processing speed and cognitive flexibility^61,62, thus slowing cognitive decline. Therefore, we hypothesize that by combining the changes in theta and beta waves, physical activity may slow down the rate of cognitive decline by optimizing brain wave activity⁶³. More specifically, physical activity may enhance the brain’s effective working state by improving functional connectivity between brain regions (such as the connection between the frontal lobe and hippocampus)^64,65. This mechanism is particularly important in the early stages of cognitive impairment, as it may help the brain utilize existing resources more efficiently to cope with the increasing cognitive challenges⁶⁶. In conclusion, the mediation model analysis of this study further revealed the regulatory role of physical activity in cognitive function through EEG activity, suggesting that physical activity not only directly affects cognitive function but may also indirectly promote the maintenance and enhancement of cognitive function by improving brain function efficiency. This discovery provides a new perspective for future physical activity intervention research. Based on this, the study proposed that specific EEG indicators could serve as potential mediators of the relationship between physical exercise and cognitive function. Although the effect size was small, the establishment of the model still deepens our understanding of how physical activity promotes elderly cognitive function through brain function. From a preventive perspective, even a small protective effect, when applied at the population level, may have a significant impact on delaying cognitive decline and alleviating the societal burden of dementia.

Conclusion

This study provides a new perspective on elderly individuals with cognitive impairment, exploring the potential pathways and mechanisms through which physical activity may influence brain health and cognitive maintenance:1.Demographic factors such as age, marital status, previous occupation, education level, dietary habits, and residence may affect the maintenance or decline of cognitive function through different mechanisms. 2.The study found that individuals with better cognitive abilities may exhibit lower levels of neural activation at rest, particularly in theta, alpha, and beta wave activity, which could help optimize energy expenditure. 3.A correlation between total physical activity and cognitive function in elderly individuals was identified, potentially influencing brain function efficiency and slowing cognitive decline by modulating neural activity in specific brain regions. However, as this study employed a cross-sectional design, causality cannot be definitively established. Future longitudinal studies will help verify the causal relationship between physical activity and cognitive function and further reveal the specific mechanisms in brain function maintenance.

Limitations and future prospects

(1) Due to sample size limitations, this study only grouped participants based on MoCA scores and could not comprehensively reveal differences in demographic variables across different cognitive impairment groups. Future studies should expand the sample size and analyze the demographic characteristics of groups with varying degrees of cognitive impairment. (2) The MoCA scale is primarily used for cognitive screening and does not comprehensively assess cognitive abilities. Therefore, future studies could consider using more specialized cognitive assessment tools, such as executive function tasks, to more thoroughly explore the relationship between physical activity and cognitive function. (3) This study found that individuals with better cognitive abilities may exhibit lower levels of neural activation at rest, but the relationship between lower neural activation and better cognitive function is not consistent. In some cases, the activation of EEG markers may exhibit an inverted U-shaped relationship with cognitive impairment. Future studies should expand the sample size and control for related factors to further explore this interaction. (4) As this study used an observational design, it revealed the lateral relationship between physical activity, EEG indicators, and cognitive function. Future research should design randomized controlled trials to explore the causal relationships between these variables. (5) This study did not include other aspects of physical activity. Future research may incorporate kinematic measures to explore the relationships between multiple variables, providing a basis for delaying cognitive decline. (6) Due to the design limitations, this study did not systematically compare EEG power and physical activity across different MoCA groups. Future research could consider longitudinal designs or group comparisons to reveal differences in various cognitive levels. (7) This study used traditional fixed frequency bands. Future studies may consider defining frequency bands based on individual alpha peak frequency to improve the precision of results. (8) This study analyzed EEG signals using single-electrode analysis. Future research should employ multi-electrode systems and network analysis methods to comprehensively assess the coordination between brain regions and further validate the relationship between physical activity and cognitive function.9.Although rigorous preprocessing procedures were applied to minimize EMG artifacts, the potential influence of low-amplitude EMG activity on beta-band power cannot be entirely ruled out. Therefore, interpretations of beta-band findings should be made with caution. Future studies may adopt more refined EMG–EEG separation techniques or incorporate simultaneous surface EMG recordings to further verify the neural origin of beta-band measures.

Author contributions

B. Xie: Conceptualization, Data collection, Analysis, Writing - Original Draft. C. Qiu: Data collection, Methodology, Writing - Review & Editing. C. Wei: Data collection, Analysis, Writing - Review & Editing. L. Chen: Conceptualization, Supervision, Writing - Final Draft.

Funding

This work was supported by the National Social Science Foundation of China funded project (22BTY076).

Data availability

Materials and analysis code for this study are available by emailing the corresponding author.

Declarations

Competing interests

The authors declare no competing interests.

Informed consent and institutional review board approval

For experiments involving human participants, informed consent has been obtained from all participants (all adults) in this study. For uneducated subjects, informed consent forms are signed by the individual and their guardian. Our study was approved by the ethical committee of Shanghai University of Sport (102772020RT060), All methods were carried out in accordance with relevant guidelines and regulations.