Walking performance declines with age (Sudarsky, 1990; Shimada et al., 2010). The gait pattern associated with aging is mainly characterized by decrease in gait speed, step length, and swing phase and increase in step width, double support time, and gait variability (Sudarsky, 1990). The decrease in walking ability that accompanies aging increases the risk of falls, decreases daily living activities and quality of life, and has a strong association with near-term mortality (Shimada et al., 2010; Studenski et al., 2011). Some elderly people change their gait patterns to compensate for reduced physical performance and reduce falls (Maki, 1997). These changes in gait patterns are particularly evident in the characteristics of muscle activity while walking. Elderly adults show a greater degree of activation of the tibialis anterior and soleus muscles than young adults during the loading response and mid-stance phases (Mickelborough et al., 2004). In addition, old adults have a higher degree of activation of the vastus lateralis and medial hamstring muscles during the loading response and mid-stance phases than do young adults, suggesting increased coactivation of the ankle and knee joint muscles (Schmitz et al., 2009).

Previous studies have investigated how age-related decline in gait capacity influences brain function (Harada et al., 2009) as gait is influenced by higher-order cognitive and cortical control mechanisms (Mihara et al., 2008). Central motor networks, including the primary motor (M1), premotor (PM), and prefrontal (PFC) cortices, are activated during walking (Mihara et al., 2008). Gait characteristics such as speed and variability are closely related to cortical activity that is regulated by multiple supraspinal control mechanisms (Wilson et al., 2019). In elderly people, slow gait speed is related to decline in the frontoparietal control network, which is associated with executive function, and gait stability is linked to the function of the dorsal attention network (Hsu et al., 2019). The gait control mechanism of the cranial nervous system is also affected by aging, and use of indirect motor pathways in the PM and supplementary motor areas and the PFC increases to maintain gait control despite age-related functional declines (Herold et al., 2018). Although studies have not measured real-time functional brain activity during actual walking tasks, they have delineated the neural correlates of gait via conventional neuroimaging tools, such as functional magnetic resonance imaging (fMRI) and positron emission tomography (PET). Combined functional near-infrared spectroscopy (fNIRS) and electroencephalography (EEG) is a recently developed imaging tool for measuring activities during walking or daily living activities (Gramigna et al., 2017; Jacobsen et al., 2021). Gait studies using fNIRS and EEG can help elucidate the characteristics of brain activity through real-time signal measurement (Khan et al., 2021).

Gait is a product of complex interactions between the central nervous and musculoskeletal systems (De Groote and Falisse, 1946). Simulations that predict movement patterns based on mathematical models of the neuro-musculoskeletal system can reveal the principles of gait movement by explaining causal relationships (Falisse et al., 2020). There is an increasing number of studies aiming to understand the cortical gait control mechanism based on experimental data. A study reported corticomuscular connectivity using EEG and electromyography (EMG) during treadmill walking as a novel technique based on reliable source localization and effective connectivity analysis. Those authors demonstrated that the motor cortex drives the movement of leg muscles through corticomuscular connectivity even during stereotypic locomotion (Artoni et al., 2017). These corticomuscular connections confirm cooperation between the cortex and muscles during walking, but additional research is needed to demonstrate how cortical activity relates to muscle activity. The relationships among aging, lower limb muscle activity, and cortical gait control mechanisms are also unclear. Therefore, investigation of age-related changes in cortical activities during gait using fNIRS will help elucidate the central mechanism of aging-induced decline of gait function in the elderly.

The decline in mobility among elderly people accelerates with age (Shumway-Cook et al., 2007). For example, people who are aged 75 years or older walk much more slowly with shorter steps and increased step width than people aged 65–74 years (Thaler-Kall et al., 2015). Therefore, geriatric studies often investigate the characteristics of aging by dividing the elderly population into two distinct groups; the young-old (65–74 years) and the old-old (75 years old and older; Magaziner, 1989). Walking ability and dual-task performance are lower in old-old people than in young-old people (Gimmon et al., 2018).

This study investigates the relationship between lower limb muscle activity and cortical activation in elderly people during walking and examines differences in that relationship between young-old and old-old subjects. This study also examines the effects of fast walking speed and cognitive dual-task walking on lower limb muscle activity, cortical activation, and the relationship between them in those two age groups.

In this study, we investigated the relationships between lower limb muscle activity and cortical activation during treadmill walking in young-old and old-old participants. In the old-old group, whose gait function was more deteriorated than that in the young-old group, mGCM muscle activity and cortical activation showed larger increases than in the young-old group during treadmill walking. In addition, muscle activity correlated strongly with cortical activation in the young-old group, whereas a significant but weaker correlation was observed in the old-old group. This corticomuscular connectivity was influenced by gait speed and addition of a cognitive task, with speed loading increasing the connection and cognitive challenge weakening it.

A previous study demonstrated significant differences in gait function between young-old and old-old participants divided at the age of 75 years (Ferraro, 1980). The results of this study also show significant differences in gait and balance between those two groups. The old-old group was slower and had poorer physical performance than the young-old group, which is consistent with acceleration of decreasing mobility with age (Forrest et al., 2006). Furthermore, the age-related decreases in dual-task cost and gait performance in the old-old group were greater than those of the young-old group under high-attention load conditions (i.e., walking under dual-task conditions; Pothier et al., 2015). Therefore, age-related changes in corticomuscular connectivity between elderly people in different age groups should be investigated further.

During comfortable walking, the old-old group had greater activity of the GCM in the initial contact and initial swing phases than the young-old group. However, ES muscle activity during the terminal swing phase decreased significantly in the old-old group. Elderly people compensate for their reduced physical performance by changing their gait patterns to prevent falls (Maki, 1997). Those changes in gait patterns are particularly evident in the characteristics of muscle activity (Mickelborough et al., 2004). Activation of the TA and soleus muscles in elderly adults is greater than that in young adults during the loading response and mid-stance phases (Schmitz et al., 2009). GCM activity also increased in the old-old group, and coactivation of the ankle joint muscle significantly increased in the initial contact and swing phases. In a previous study, such coactivation in elderly people was significantly higher than that among young and middle-aged subjects while walking (Lee et al., 2017). Muscle coactivation correlated significantly with gait speed in elderly participants, suggesting that the slow gait speed in the old-old group is associated with increased coactivation of the ankle joint muscles.

During comfortable walking, the overall oxyHb concentration within the measured cortical areas was higher in the old-old group than in the young-old group. These results are consistent with the cognitive resource theory of aging that brain activation increases during comfortable walking with age (Beurskens and Bock, 2013). Similarly, older subjects exhibit greater neural activation (hyperactivation) than younger individuals when they perform exercise-related tasks (Ward, 2006). These findings of hyperactivation of the cerebral cortex among old-old people during walking could reflect their more pronounced gait degradation compared with young-old people. Additionally, elderly with cognitive decline due to dementia demonstrated higher oxyHb concentration during walking than did a healthy group (Teo et al., 2021). It was hypothesized that increased cortical activity during walking in people with cognitive decline may represent some form of neural compensation, as suggested by the compensation-related utilization of neural circuits hypothesis (CRUNCH) (Reuter-Lorenz and Cappell, 2008). Since the K-MMSE score is lower in the old-old than in the young-old, hyperactivation in the old-old can be considered as compensation for cognitive decline.

Hyperactivation of the cerebral cortex in the old-old may be associated with depressive symptoms. The GDS-SF score of the old-old was significantly higher than that of the young-old. Functional neuroimaging studies of depressed patients report abnormal activity in the left prefrontal and prefrontal cortex compared to healthy subjects (Okada et al., 2003). A previous study demonstrated increased cortical activity in a depressed group during a working memory task (Matsuo et al., 2007; Fitzgerald et al., 2008). This indicates that hyperactivation of cortical activity due to depressive symptoms is evident in high-demand cognitive environments or task demands (Colich et al., 2017). As the average GDS-SF score of the old-old was 4.85 ± 4.21 (range: 0 ~ 13), which is lower than the cut-off score of 9 (Laudisio et al., 2018), the depressive symptoms in the old-old might be related to hyperactivation.

In this study, lower limb muscle activity correlated with cortical activation among all our elderly participants during comfortable treadmill walking. In recent EEG and EMG studies, coherence and directional analyzes have investigated whether the motor cortex contributes to muscle activity (Winslow et al., 2016; Jensen et al., 2019). In those reports, coherence between the cortical motor region and the TA and BF was enhanced in the swing phase, and the TA had higher connectivity than the BF. In this study, the oxyHb concentration in the cerebral cortex was measured via fNIRS, and its correlation with muscle activity was analyzed. Our results support those previous reports showing a close connection between cortical and muscle activity, as measured by EEG–EMG coherence.

The correlation between lower limb muscle activity and cortical activation during comfortable walking was stronger in the swing phase than in the stance phase and was likely related to the CI of the joint muscles. In the swing phase, elderly people can experience an increase in muscle coactivation to gain ankle and knee joint stability, which increases cortical activity. That increased coactivation reinforces the rigidity of the leg joints in elderly people to help them maintain balance and joint stability and thus prevent falls (Thompson et al., 2018). In other studies, the stance phase of walking has been shown to be necessary to maintain an upright posture under dynamic and challenging conditions (Takakusaki, 2017). Our results are consistent with that conclusion because a specific cortical region is involved in muscle activation. In a previous EMG study of reciprocal inhibition during walking, the swing phase required greater control of cortical activation than the other gait phases (Lavoie et al., 1997). In a transcranial magnetic stimulation study using lower extremity muscle motor-evoked potentials, the corticospinal tract was closely linked to muscles that control ankle flexion/extension during the swing phase of the gait cycle (Takahashi et al., 2014).

The correlation between lower limb muscle activity and cortical activation during gait likely weakened in the old-old group, and their gait function decreased compared with that of the young-old group. These findings suggest that old-old people experience greater cortical activation than younger people while walking, but the correlation between cortical activation and muscle activity decreases. The gait control mechanism of the cranial nervous system is also affected by aging, and the use of indirect motor pathways from the PMC, SMA, and PFC increases to maintain gait control despite aging-related functional decline (Herold et al., 2018). In other words, decreased walking ability due to aging requires involvement of secondary motor areas, the PFC, and brain areas other than the primary motor cortex during walking; as a result, the association between cortical activity and muscle activity is reduced.

When walking speed increased, cortical activity decreased. Several studies have reported increased activity in the PFC when walking speed increases, but those studies compared a slow walking speed (1.5 km/h) with a normal walking speed (3.0 km/h; Kim et al., 2017). Thus, comparisons of cortical activity at increased speeds compared to slower speeds can be controversial. In this study, a decrease in cortical activity was observed when walking at a speed of 130% compared to self-selected walking speed. A previous study reported decrease in PFC and M1 activity when elderly participants walked at fast speed compared to slow speed (Herold et al., 2018). In contrast, activity of the subcortical and spinal structures that govern automatic movement increase during fast speed walking (Nordin et al., 2019). Based on these findings and results of this study, cortical activities in the motor and frontal cortical regions are reduced and lower limb muscle activities are increased during fast walking compared to comfortable walking, but the correlation between these two is strengthened, which supports the different roles of cortical activity in speed walking.

This study has some limitations. First, tasks were conducted on a treadmill, which could limit the generalizability of our results. Several studies have reported differences in gait performance and cortical activity between a treadmill and overground walking (Roeder et al., 2018). In our study, we used a treadmill to position the fNIRS device close to the subjects during walking. In future studies, cortical activity should be measured during overground walking in various environments. Second, the correlation between muscle and cortical activities does not imply a direct connection. In an EEG–EMG corticomuscular coherence study, the cortex modulated lower muscle movement in the beta band, including 20 Hz. However, the sampling rate of fNIRS used in the present study was up to only 10 Hz; consequently, the direct connection in the beta band could not be confirmed. Therefore, the correlation between the oxyHb concentration in the cortex and %MVC in the muscles during walking in elderly people does not indicate a direct connection between the brain and muscles. Third, no control group of healthy young adults was used. In this study, the characteristics of cortical and muscle activities were compared between the young-old and the old-old without comparison with health young adults. To help understand overall aging, comparison with healthy young adults is necessary, and there is a limit to interpreting the differences between two aged groups. Fourth, in fNIRS, changes in cerebral blood flow were measured by setting the maximum source detector separation (SDS) to 30 mm without using short SDS channels. Therefore, the recorded fNIRS signals reflect extra- and intracerebral changes; correction using short SDS signals could have reduced the influence of changes in skin blood flow. Therefore, in future studies, skin blood flow should be omitted using short SDS signal measurement equipment (Scholkmann et al., 2013).

Results of this study indicated significant correlations between lower limb muscle activity and cortical activation during walking, which differed by age group of the elderly. These correlations were also modulated by additional load of gait speed or cognitive dual task. These results suggest that gait function is compensated by regulating corticomotor correlation as well as brain activity during walking in the elderly. We believe that this study deepens understanding of the central-to-peripheral gait control mechanism in elderly people and provides data useful for developing specific gait training targets to prevent falls and to develop a gait intervention program using a robotic device or brain–computer interface for the elderly.

The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.

The studies involving human participants were reviewed and approved by Institutional Review Board of Samsung Medical Center (2020–09-172). The patients/participants provided their written informed consent to participate in this study.

JK, JL, and Y-HK designed the experiment. JK and GL collected the data. JK performed data and statistical analysis with assistance on approach and interpretation from JL, GL, WC, M-HK, W-KY, G-HR, and Y-HK and wrote the manuscript. Y-HK critically evaluated the manuscript. All authors contributed to the article and approved the submitted version.

This study was supported by a National Research Foundation of Korea (NRF) grant funded by the Korean government (NRF-2020R1A2C3010304); the MSIT (Ministry of Science and ICT), Korea, under the ICT Creative Consilience program (IITP-2021-2020-0-01821) supervised by the IITP (Institute for Information & Communications Technology Planning & Evaluation); and a Korean Medical Device Development Fund grant funded by the Korean government (MSIT, Ministry of Trade, Industry and Energy, Ministry of Health & Welfare, Ministry of Food and Drug Safety; Project Number: KMDF-RS-2022-00140478).

JK was employed by Ybrain Inc.

The remaining authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.

The Supplementary material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fnagi.2022.1059563/full#supplementary-material