Gait Speed Related Changes in Electrocortical Activity in Younger and Older Adults

Jacob Salminen; Chang Liu; Erika M Pliner; Madison Tenerowicz; Arkaprava Roy; Natalie Richer; Jungyun Hwang; Chris J Hass; David J Clark; Rachael D Seidler; Todd M Manini; Yenisel Cruz-Almeida; Daniel P Ferris

doi:10.1152/jn.00544.2024

. Author manuscript; available in PMC: 2026 Jan 14.

Published in final edited form as: J Neurophysiol. 2025 Apr 24;133(6):1761–1794. doi: 10.1152/jn.00544.2024

Gait Speed Related Changes in Electrocortical Activity in Younger and Older Adults

Jacob Salminen ¹, Chang Liu ^1,², Erika M Pliner ³, Madison Tenerowicz ¹, Arkaprava Roy ⁴, Natalie Richer ⁵, Jungyun Hwang ⁶, Chris J Hass ⁷, David J Clark ^1,⁸, Rachael D Seidler ^8,^9,¹⁰, Todd M Manini ¹¹, Yenisel Cruz-Almeida ^10,^12,^13,¹⁴, Daniel P Ferris ^1,¹⁰

^1.Department of Biomedical Engineering, University of Florida, Gainesville, FL, USA

^2.Department of Kinesiology and Nutrition, University of Illinois Chicago, Chicago, IL, USA

^3.Department of Mechanical Engineering, University of Utah, Salt Lake City, UT, USA

^4.Department of Biostatistics, University of Florida, Gainesville, FL, USA

^5.Department of Kinesiology and Applied Health, University of Winnipeg, Winnipeg, Canada

^6.Department of Neurology, University of Florida, Gainesville, FL, USA

^7.Department of Applied Physiology and Kinesiology, University of Florida, Gainesville, FL, USA

^8.Brain Rehabilitation Research Center, Malcom Randall VA Medical Center, Gainesville, FL, USA

^9.Norman Fixel Institute for Neurological Diseases, University of Florida, Gainesville, FL, USA

^10.McKnight Brain Institute, University of Florida, Gainesville, FL, USA

^11.Department of Health Outcomes and Biomedical Informatics, University of Florida, Gainesville, FL, USA

^12.Department of Community Dentistry and Behavioral Science, University of Florida, Gainesville, FL, USA

^13.Pain Research and Intervention Center of Excellence, University of Florida, Gainesville, FL, USA

^14.Department of Neuroscience, University of Florida, Gainesville, FL, USA

^✉

Correspondence: Jacob Salminen (jsalminen@ufl.edu), Work was performed at the University of Florida’s Institute of Aging

AUTHOR CONTRIUBTIONS

J.S. Performed experiments, analyzed data, interpreted results of experiments, prepared figures, drafted manuscript, edited and revised manuscript, and approved final version of manuscript.

C.L. Performed experiments, analyzed data, interpreted results of experiments, edited and revised manuscript, and approved final version of manuscript.

E.P. Performed experiments, analyzed data, interpreted results of experiments, edited and revised manuscript, and approved final version of manuscript.

M.T. Performed experiments, edited and revised manuscript, approved final version of manuscript.

A.P. analyzed data and edited and revised manuscript.

N.R. Performed experiments, interpreted results of experiments, edited and revised manuscript, approved final version of manuscript.

J.H. Performed experiments, interpreted results of experiments, edited and revised manuscript, and approved final version of manuscript.

C.J.H. Conceived and designed research, interpreted results of experiments, edited and revised manuscript, approved final version of manuscript.

D.J.C. Conceived and designed research, interpreted results of experiments, edited and revised manuscript, approved final version of manuscript.

R.D.S. Conceived and designed research, interpreted results of experiments, edited and revised manuscript, approved final version of manuscript.

T.M.M Conceived and designed research, interpreted results of experiments, edited and revised manuscript, approved final version of manuscript.

Y.C-A Conceived and designed research, interpreted results of experiments, edited and revised manuscript, approved final version of manuscript.

D.P.F Conceived and designed research, interpreted results of experiments, edited and revised manuscript, approved final version of manuscript.

PMCID: PMC12799196 NIHMSID: NIHMS2079982 PMID: 40272781

Abstract

Preferred and maximum walking speeds decline as we age and the decline has been associated with worsening health. Slowing of gait in older individuals is correlated with biomechanical and neural factors, but historically it has been difficult to measure whole-brain activity during walking. Recent advances in mobile brain imaging with high-density electroencephalography (EEG) allow for separation and localization of electrical brain activity during walking. We studied younger (N=31) and older adults (N=59) walking on a treadmill at different speeds (0.25-1.0 m/s) while we recorded electrocortical dynamics with EEG. We hypothesized that faster walking speeds would result in greater sensorimotor and posterior parietal theta band (4-7 Hz) spectral power and lower beta band (13-30 Hz) spectral power compared to slower walking speeds for older adults, consistent with previous studies on younger adults. Additionally, we used a standardized test of physical function to group older adults into high functioning [Short Physical Performance Battery, SPPB≥10] and low functioning (SPPB<10) groups for comparison. In agreement with our hypotheses, sensorimotor and posterior parietal theta power increased, and beta power decreased at faster walking speeds. We also found that left posterior parietal, mid cingulate, and cuneus exhibited differences in theta power at faster speeds between younger and older adults. The results suggest that older and younger adults activate cortical areas throughout the brain while walking at different speeds, and older adults, particularly those with lower mobility, recruit greater cognitive resources in parietal cortex compared to younger adults. These results could inform stimulation protocols targeting parietal cortex.

Keywords: Aging, Gait, Clinical, Electroencephalography, Mobile Brain Imaging

Graphical Abstract

graphic file with name nihms-2079982-f0001.jpg

NEW & NOTEWORTHY

Older and younger adults show widespread EEG beta power decreases at faster walking speeds compared to slower walking speeds. Older adults differentially alter EEG theta power while walking compared to younger adults. Prior studies with fNIRS have documented differences in prefrontal activation in older adults walking compared to younger adults, but our results show cortical changes within speed and age outside of the prefrontal cortex.

INTRODUCTION

Mobility is critical for maintaining health as humans age. The ability to walk allows us to interact with others in our community and perform activities of daily living. One important marker of mobility is walking speed. Walking speed assessments for older adults are reliable, valid, and sensitive measures of mobility and health (1-4). Preferred and maximum walking speeds decline as we age, concomitant with reduced health status (5, 6). There are multiple physiological mechanisms related to slowing gait speed in older adults, including factors related to musculoskeletal biomechanics (7-9) and sensorimotor control (10-13). Historically it has been difficult to measure ongoing brain activity during actual walking (14), leading to a lack of knowledge on how aging affects brain involvement in the control of walking.

Older adults may require more active cortical involvement in walking compared to younger adults. In neurologically intact younger adults, walking has a high degree of automaticity (11, 15, 16). This comes from a combination of neural and biomechanical factors controlling repetitive stepping. With aging, there are sensory deficits, muscle weakness, and cognitive decline that may make walking less automatic (10). Older adults show greater pre-frontal brain activity, measured with functional Near InfraRed Spectroscopy (fNIRS), compared to young adults in the same walking task (17-19). Older adults often have compromised sensory capabilities, leading to recruitment of more cortical resources to perform the same tasks as younger adults. Older adults appear to increase prefrontal and motor cortical activity more than young adults as walking speed increases based on fNIRS studies (20, 21). However, limitations with fNIRS are that it can only measure a few centimeters into the brain and it requires several seconds of steady state behavior to produce reliable results.

High-density EEG can measure electrocortical brain activity related to changes in walking speed with high temporal resolution and in both shallow and deep cortical areas. A recent comprehensive review provides a summary of recent advancements in EEG for mobile brain imaging, including what we know about EEG frequency changes related to walking (14). Source separation and localization of high-density EEG signals allows researchers to quantify brain electrical activity in brain areas that are relevant to ongoing motor tasks. Younger adults show decreases in sensorimotor alpha (8-12 Hz) and beta (13-30 Hz) spectral power at fast walking speeds (up to 2.0 m/s) compared to slow walking speeds (0.5 m/s) (22, 23). The decrease in alpha and beta in sensorimotor brain areas is likely indicative of greater localized neuronal recruitment for computation (24, 25). In contrast to alpha and beta spectral power, theta (4-7 Hz) spectral power increases (i.e. synchronization) during walking appear to relate to increased reliance on sensory system input on gait behavior (14, 26, 27). Forcing young participants to adapt to different speeds for their left and right legs on a split-belt treadmill reveals changes in alpha, beta and theta spectral power in sensorimotor, posterior parietal and cingulate cortices (26, 28, 29). However, electrocortical dynamics related to walking speed changes have not been examined in older adults. Understanding how aging affects electrical brain activity at different walking speeds may provide insight into the functional decline in mobility observed in older adults.

We used high-density EEG to measure electrocortical dynamics of older and younger adults during walking to understand changes in cortical activity with aging. We included two groups of older adults, those with higher physical function (i.e., older high functioning adults; OHFA) and those with lower physical function (i.e., older low functioning adults; OLFA). This study is part of a larger project to identify changes in gait and brain dynamics in healthy older adults (30). Our hypotheses focus on the sensorimotor cortex and posterior parietal cortex due to their roles in sensorimotor planning and gait behavior monitoring (14, 27, 31-33). We hypothesized that older adults’ sensorimotor cortex and posterior parietal cortex would display an increase in theta spectral power with increases in gait speed. Increases in gait speed cause increases in step frequency, increases in heel strike impulses, and changes in pelvis movement (7, 8, 34, 35). We hypothesized that the sensorimotor cortex and posterior parietal cortex would also show decreases in beta spectral power with increasing speed across all three groups. In addition to our two main hypotheses, we tested for changes in electrocortical spectral power with speed in other brain regions. Regarding differences between the age groups, we expected greater electrocortical power effects with increased speed for older adults with low mobility compared to younger adults and older adults with higher mobility. Younger adults have greater automaticity of walking compared to older adults (11). fNIRS studies show greater frontal brain activity with increasing walking speed in older adults compared to younger (20, 21). Therefore, older adults may require greater cortical involvement at faster walking speeds, especially lower functioning adults, compared to younger adults.

METHODS

Participants

This study was conducted as part of the larger study Mind in Motion (NCT03737760) (30). We recruited a total of 96 older adults (at least 65 years of age) and 35 younger adults (aged 20 to 40 years). Inclusion and exclusion in the study are detailed in (Clark et al., 2019; (34)). Briefly, inclusion criteria ensured that all participants are community dwelling, able to complete an unassisted 400 m walk test in 15 minutes, and English speaking. We excluded participants if they presented mild cognitive impairments graded by the Montreal Cognitive Assessment (MoCA) score <26 and other current conditions such as terminal illness, uncorrectable visual impairment, severe cardiovascular disorders, and implants that prevent the taking of a Magnetic Resonance Image (MRI). Adults were divided into two groups: older low functioning adults (OLFA) and older high functioning adults (OHFA). The group assignment was based on the Short Physical Performance Battery (SPPB) total. Subjects were graded on their performance in a sit to stand test, side-by-side stand, a semi-tandem stand, a full tandem stand, and an 8 ft walk test. Older adults with a SPPB less than 10 were considered low function (<10 SPPB, n=63) and older adults with a SPPB greater than or equal to 10 were considered high function (≥10 SPPB, n=33) (36). See [Table 2] for demographic information of our final sample of participants. After EEG pre-processing, we further excluded 28 participants due to the following reasons. 8 participants did not have an MRI scan performed. Cases vary between claustrophobia, health conditions that would jeopardize patient safety, withdrawal from study. 3 participants’ EEG data preparation produced poor impedances between the scalp and the electrode leading to noisy data. 17 participants did not perform all speed conditions, so we excluded them from analysis. Participants were unable to complete conditions due to concerns for the participants safety, unwillingness to perform the condition, or lacked the ability to walk at the specified speed [Figure 1]. For additional details on why subjects were rejected and how rejection criteria may be associated with SPPB scores see Table 1 and Figure 2, respectively. We did include subjects that had partial condition data. All participants provided written informed consent. The study and its design are in accordance with the Declaration of Helsinki and approved by the Institutional Review Board of the University of Florida (IRB 20180222700053).

Table 2.

We collected data from a cohort of younger, older high mobility, and older low functioning adults. The table shows statistics for the subjects that met the inclusion criterion. The mean, one standard deviation (SD), and range ([Min,Max]) of each group’s age and short physical performance battery (SPPB) are provided. Sex distributions are also displayed.

Group	N	Age			Sex		SPPB
Group	N	Mean	SD	[Min,Max]	Male	Female	Mean	SD	[Min,Max]
*Younger Adults*	31	24	4	[20,35]	14	17	12	0	[12,12]
*Older High Functioning Adults*	27	74	5	[64,84]	13	14	11	0.9	[9,12]
*Older Low Functioning Adults*	32	74	6	[65,90]	23	35	8	1	[6,9]

Open in a new tab

Schematic of exclusion from original study sample. First column: We considered 35 younger adults and 96 older adults for the analysis. Second column: We initially excluded participants if they were not able to complete a Magnetic Resonance Image (MRI) head scan for creating a source localization head model, if they did not walk at all 4 speeds (0.25, 0.50, 0.75, 1.0 m/s), or if issues with EEG collections prevented reliable data analysis (e.g., bridged electrodes and faulty amplifiers). Third Column: We performed further rejections after EEG processing procedures if they had less than five brain-like independent components, or if they lacked a resting state recording required for spectral principal component analysis. For additional details see Figure 2 and Table 1.

Table 1.

Rejection criteria and SPPB scores for subjects excluded from the final analysis. Acronyms: younger adults (YA), older low functioning adults (OLFA), older high functioning adults (OHFA), independent component (IC), not available (n/a), magnetic resonance image (MRI).

Subject Code	Group	SPPB	Rejection Details
H1002	YA	12	No resting condition
H1025	YA	12	Poor impedances in EEG due to gelling.
H1044	YA	n/a¹	Fewer than 5 brain-like IC after automated IC rejection.
H1046	YA	n/a¹	Poor impedances in EEG due to gelling.
H2007	OHFA	10	Fewer than 5 brain-like IC after automated IC rejection.
H2010	OHFA	12	MRI was not performed.
H2033	OHFA	11	No resting condition
H2036	OHFA	10	MRI was not performed
H2041	OHFA	12	MRI was not performed
H2072	OHFA	11	MRI was not performed
H3018	OLFA	8	MRI was not performed
H3024	OLFA	6	Did not walk at the 0.75 m/s and 1.0 m/s conditions
H3034	OLFA	9	Fewer than 5 brain-like IC after automated IC rejection.
H3042	OLFA	8	Equipment malfunction: EEG, GRF Sensors, and IMU synchs
H3046	OLFA	6	Did not walk at the 0.75 m/s condition
H3047	OLFA	7	Did not walk at the 0.75 m/s and 1.0 m/s conditions
H3053	OLFA	8	Did not walk at the 0.75 m/s and 1.0 m/s conditions
H3063	OLFA	9	Did not walk at the 0.25 m/s and 1.0 m/s conditions
H3073	OLFA	7	Did not walk at the 1.0 m/s condition
H3077	OLFA	8	Equipment malfunction: EEG, GRF Sensors, and IMU synchs
H3092	OLFA	7	Did not walk at the 1.0 m/s condition
NH3002	OLFA	7	MRI was not performed
NH3004	OLFA	9	Poor impedances in EEG
NH3006	OLFA	8	Did not walk at the 0.75 m/s and 1.0 m/s conditions
NH3007	OLFA	9	Did not walk at the 1.0 m/s condition
NH3009	OLFA	7	Did not walk at the 0.50, 0.75, and 1.0 m/s conditions
NH3010	OLFA	9	No resting condition
NH3023	OLFA	4	Did not walk for any of the conditions
NH3025	OLFA	6	Did not walk at the 0.75 m/s and 1.0 m/s conditions
NH3027	OLFA	6	MRI was not performed
NH3028	OLFA	n/a¹	Did not walk at the 0.50, 0.75, and 1.0 m/s conditions
NH3036	OLFA	9	Equipment malfunction: EEG battery died.
NH3040	OLFA	5	Fewer than 5 brain-like IC after automated IC rejection.
NH3051	OLFA	7	Did not walk at the 0.75 m/s and 1.0 m/s conditions
NH3054	OLFA	9	Fewer than 5 brain-like IC after automated IC rejection.
NH3055	OLFA	8	Fewer than 5 brain-like IC after automated IC rejection.
NH3056	OLFA	6	Did not walk at the 0.75 m/s and 1.0 m/s conditions
NH3071	OLFA	6	Did not walk at the 0.50, 0.75, and 1.0 m/s conditions
NH3074	OLFA	9	Fewer than 5 brain-like IC after automated IC rejection.
NH3082	OLFA	7	Did not walk at the 0.75 m/s and 1.0 m/s conditions
NH3104	OLFA	9	Fewer than 5 brain-like IC after automated IC rejection.
NH3114	OLFA	7	Fewer than 5 brain-like IC after automated IC rejection.
NH3123	OLFA	9	Did not walk at the 0.75 m/s and 1.0 m/s conditions

Open in a new tab

We did not collect some Short Physical Performance Battery (SPPB) for older and younger adults as they did not complete all divisions of the Mind in Motion protocol.

Comparison of Short Physical Performance Battery (SPPB) scores for older adults that were included in the study and those that were excluded due to missing condition data. The one-way ANOVA confirmed that the SPPB scores for the older adults that were unable to complete all the conditions (Excluded) were less than the scores for those that could (Included). (F(70,1)=24.33, p<0.001***).

Experimental Protocol

Before we had subjects walk on a treadmill to collect EEG data, we had participants complete an overground walking pre-test. We had participants walk 3-meters on flat ground three times. To ensure that each participant had reached their steady state walking speed on each pass we included 2-meters at the start of the walk that were not timed. A researcher instructed participants to walk at their normal pace while they used a stopwatch to time each pass. We calculated the speed of each pass, then averaged them to determine their preferred overground walking speed.

Participants walked at different speeds for an estimated 24 minutes. Participants walked on a rubberized slat belt treadmill (PPS 70 Bari-Mill, Woodway, Waukesha, WI, USA; 70 cm x 173 cm walking surface) with no objects or boundaries in their immediate vicinity (37, 38) [Figure 3]. Participants were instructed to not ride the treadmill to the back or walk closely to the front. Participants performed a sitting eyes-open resting trial for 3 minutes and walking under other conditions for an additional 24 minutes (i.e., uneven terrain treadmill walking as described in Liu, Downey, Salminen, et al. 2023). All data were included into the EEG source localization portion of our analysis (approximately 50 minutes of data), as this can improve consistency of the numerical calculation (39, 40). After we performed ICA, we focused our analysis only on the data collected when subjects walked at different speeds on the flat belt treadmill, excluding any data observed while subjects walked on the uneven terrain treadmill. The speed conditions were performed at 0.25 m/s, 0.50 m/s, 0.75 m/s, and 1.0 m/s [Figure 3]. We limited our analysis to these 4 speeds because the older adults had limitations in the walking speeds they could achieve. This is evident in subjects’ overground walking speed [Figure 5]. An overhead body harness prevented each participant from hitting the ground if they stumbled. The harness did not support body weight during walking. Two staff members were on each side of the participant to aid in the event of falls and to assist participants who felt uneasy starting and stopping the treadmill. We provided rest periods for participants between each speed trial to prevent fatiguing. At the participants’ request, we allowed additional rest time. If a participant exhibited noticeable signs of increased breathing rate or perspiration, we increased the duration of the rest period. Trials lasted 3 minutes, unless the participant could not physically continue. Some participants had shortened trials for the faster speed conditions to avoid issues with stamina (typically 1 to 2 minutes per trial, rather than 3). 5 participants had shortened trials or were unable to complete both trials due to fatigue, transient malfunctions in EEG, IMU, or ground reaction force sensor equipment, or participant unwillingness to continue.

Participants walked on a flat treadmill at 4 different speeds: 0.25 m/s, 0.50 m/s, 0.75 m/s, 1.0 m/s. Each participant was asked to walk at each speed for two trials for three minutes each. Trial order was randomized. Additionally, participants sat with eyes open for a 3-minute resting trial. Including experiment setup, collections lasted approximately 3 hours. Participants were equipped with a high-density EEG system, an inertial measurement unit attached at the posterior pelvis, and foot force sensors placed in the soles of their shoes.

Effect of speed on behavior outcomes for the younger adult (YA), older high mobility adult (OHFA), and older low mobility adult (OLFA) groups. **(A)** Overground 3-meter walking speeds by participant group. Brackets represent significant multiple comparisons after false discovery rate correction, where plus-signs indicate significance level (+p<0.05, ++p<0.01, +++p<0.001). “I”-bars show the upper and lower 95% confidence interval for the estimated marginal mean. **(B)** Effects of speed on pelvic mediolateral coefficient of variation. We found an increase in mediolateral coefficient of variation with increasing speed and an interaction between the groups and speed. Equations and coefficients for the linear fits are displayed above each group. **(C)** Effects of speed on step duration coefficient of variation. We found that step duration coefficient of variation decreased with increasing speed and had the strongest correlation of the behavioral measures observed (R²=0.58). Slope of the linear fit (m) is displayed in the top left corner and the intercepts for each group term are displayed in the top right.

EEG & MRI Data Collection Protocol

EEG.

We used a custom-made dual-layer EEG setup to record from 120 electrodes (ActiCAP snap; Brain Products GmbH, Germany) facing the scalp (traditional EEG) as well as from 120 electrodes facing away from the scalp (41). Electrodes facing away from the scalp acted as noise sensors to capture motion artifacts. A conductive graphene doped fabric (Eeontex LTT-PI-100, Marktek Inc., USA) was placed on top of the noise electrodes to simulate an artificial skin surface. We also recorded eight electromyography (EMG) channels over the neck to help capture neck muscle artifacts from bilateral sternocleidomastoid and trapezius muscles, similar to (42). Digitized locations of the electrodes on the head were mapped using a tablet based infrared scanner (Structure Sensor ST01, Occipital Inc., USA; Scanner App, XRPro LLC). Fluorescently colored stick-on dots marked fiducial landmarks (nasion, left/right preauricular) for later co-registration with the MRI derived head model. EEG data (scalp, neck, noise) were recorded at 500 Hz using four LiveAmp64 amplifiers (Brain Products GmbH, Germany; RRID:SCR_009443). Shoe insole force sensors (Loadsol 1-184 sensor; Novel Electronics Inc., USA) recorded ground reaction forces to a nearby tablet via Bluetooth (iPad Apple, USA) at 200 Hz. In addition to the left and right foot sensors, a third (extra) sensor was used to synchronize the force sensor data to the EEG data. Synchronization details are discussed in (43). Participants also wore an inertial measurement unit (IMU, Opal APDM Inc., USA) sensor over their sacrum. IMU data were recorded at 128 Hz. The IMU data were synchronized to the EEG LiveAmp by an electrical impulse at the beginning and the end of each trial.

MRI.

We collected a T1-weighted sequenced MRI for each participant. The MRI provided details of the anatomical brain structures and was pertinent to the creation of a finite element conductivity head model. The parameters for this anatomical image were: repetition time (TR) = 2000ms, echo time (TE) = 2.99ms, flip angle = 8°, voxel resolution = 0.8mm³, field of view = 256 × 256 × 167 mm2 (4:22 minutes of scan time), using a 64-channel coil array on a 3T Siemens MAGNETOM Prisma MR scanner.

Data Processing

IMU & Ground Reaction Force Measures.

We assessed subject gait behavior on each speed condition. We derived the coefficient of variation from the anteroposterior and mediolateral excursions measured using the pelvic IMU and the step durations measured using the ground reaction force sensors. For IMU data, we calculated the excursion for each gait cycle, and for ground reaction force data we calculated the time between each heel strike. Then we calculated the mean and standard deviation of excursion and step duration across all gait cycles for each speed (43). Finally, we calculated the coefficient of variation for each measure by dividing the standard deviation over the mean.

EEG Data Preprocessing.

We processed all EEG data using custom MATLAB scripts (R2023b, RRID:SCR_001622) that included functions from EEGLAB (v2022.0, RRID:SCR_007292) and FieldTrip (v20221005, RRID:SCR_004849) [Figure 4] (44, 45). Raw data from all channel types (EEG, noise, EMG) were first high pass filtered at 1 Hz (−6 dB at 0.5 Hz) to remove drift (eegfiltnew). Additionally, a 20Hz low-pass filter was applied to the EMG channels (eegfiltnew). The CleanLine plugin was applied to all data to remove 60Hz and 120Hz powerline noise. Afterwards, channels that had values that spanned greater than 3 standard deviations from the mean of that set (i.e., EEG, noise, or EMG) were removed from further analysis (42, 46, 47). We retained 95 +/− 8 channels across the population. Next, each set of channels were separately re-referenced to the common average across the set (i.e., cortical to cortical, noise to noise, EMG to EMG). Again, the channels were re-referenced to the common averaged within their respective sets. Common average referencing has shown to improve signal-to-noise ratios in EEG recordings (48)

Signal processing pipeline and brain source modeling. A) Example depiction of an EEG electrode used during subject data collections. The Scalp Electrode measures electrical potentials from the human head. The Noise Electrode, an inverted electrode that is physically coupled and electrically isolated to the Scalp Electrode, contacts an Artificial Skin Cap to measure motion artifact signals. Conductive gel provides a medium for conduction from the surfaces to the electrodes, and wires are in bundles to reduce movement artifacts. B) Flowchart for processing the recordings of the scalp, noise, and EMG signals. Signal processing algorithms (e.g., iCanClean & Cleanline) clean the input signals (46, 50, 53). An adaptive mixtures independent component analysis couples with a subject specific finite element conductivity head model to predict the location of brain sources. (Source Localization*): We used each subject’s T2 weighted MRI to produce a 6-layer finite element conductivity model of the head. Digital electrode locations, the head model, and the independent components were all used to predict the location of each component’s dipole origin. Each dipole was normalized to a MNI template. (Non-Brain component Rejection**): ICLabel percentages [38] component power spectra, and PowPowCat [39] assessed the likelihood of a particular component to be originating from a brain source. Non-brain sources were rejected from further analysis.

Next, we used the iCanClean algorithm (49, 50) to aid in removing muscle and movement artifacts from the EEG data. EEG data were cleaned using the noise sensors with a 4-second moving window and an R2 threshold of 0.65. Then, EEG data were cleaned using the EMG electrodes with an R2 of 0.4 and a 4-second moving window. This algorithm’s effective cleaning ability and its optimal parameterization have been investigated in previous work (42, 50). After motion artifact removal, the dual-layer noise sensors were removed from further analysis. Next, the clean_artifacts function in EEGLAB was applied to the EEG data to further remove bad channels and time frames. Default parameters were used except for a chan_crit1=0.5, wind_crit1=0.4, and winTol=[-inf,10] (46). The parameters were previously optimized in an investigation performed on a subset of data (46). The outcome measure of the optimization was maximizing the number of good brain components as determined by ICLabel (47, 51). 10 +/− 6 channels and a maximum of 9.0% (mean of 1.2%) of time frames were rejected. A final average re-referencing was performed. Following the previous step, the adaptive mixtures independent component analysis (AMICA) separates the EEG data into linearly independent signals, from here on referred to as sources (52).

Electrical Source Modeling.

We localized ICA sources using finite element head models constructed from subject MRIs and an electrical source density estimation based around an equivalent dipole fitting approach (ft_dipolefitting). Dipole fitting and source modeling were performed in the Fieldtrip Toolbox (v20221005). The estimation of ICA sources from EEG data is a well-established method (44, 52, 54). Each subject MRI was segmented using the headreco program from the SimNIBS toolbox (v3.2) (55). We segmented each MRI into sections of either air, cerebrospinal fluid, gray matter, white matter, skull, or scalp. Conductivity values were set at 0.33, 0.01, 1.65, 0.126, and 2.5*10⁻¹⁴, for scalp, skull, cerebrospinal fluid, white matter, and gray matter, respectively (28, 47). The prepare_mesh_hexahedral function formed a hexahedral mesh for each subject’s segmented MRI. Digitized electrode locations were aligned to the head model using fiducial landmarks and then projected to the scalp. All electrode alignments were visually inspected. The SIMBIO toolbox was then used to estimate leadfield matrices for each subject’s head model. Finally, the ft_dipolefitting algorithm estimated the location of each IC’s source within the subject-specific head model. We used the ANTs toolbox (RRID:SCR_004757) to normalize each subject’s MRI to the Montreal Neurological Institute (MNI) hi-res template. The ANTs normalization algorithm is robust to age-related changes in brain anatomy that are captured by MRI (56). The resulting transformation matrix normalized sources to the MNI template. See (46, 47) for additional details on head modeling and pipeline parameterizations.

Source Rejection.

We identified a subset of independent components which were most likely to be brain sources (47). We selected ICs using these criteria: 1) a residual variance <15%, 2) ICLabel probability of a brain source greater than 50%, 3) a negative slope of a best fit line of the 2-40 Hz range of the power spectral density (PSD), and 4) a high-frequency (>30Hz) or low-frequency (<8Hz) channel coupling of 0.3 as determined by the PowPowCat toolbox (57). All IC rejection was entirely automated. Subjects that had less than 5 ICs after rejection were removed from further analysis (10 subjects removed). In the new cohort of subjects, we retained a mean of 13 (Range 6 to 28) ICs per participant.

Source Clustering.

K-means clustering was used to group sources across subjects. We utilized the silhouette, Clinski-Harabasz, Davies-Bouldin methods to determine a K=11 for our clusterings. We clustered sources based on their normalized coordinates within the MNI head model. We used the kmeans function in MATLAB with parameters of maxiter=10,000 and replicate=1000 (58, 59). We ensured each cluster had at most one IC from each participant by selecting the IC that had the highest likelihood to be brain as determined by ICLabel. We used the Automated anatomical labeling atlas 3 (AAL3) to determine the anatomy of each cluster (60).

Spectral Power Calculations.

We performed frequency and time-frequency analysis for each cluster. First, data were separated into epochs of 5.25s (1s before and 4.25s after right heel strike). The gait cycle durations of each subject walking at their slowest speed were used to inform the epoch length. Epochs that were three standard deviations from the mean of that subject’s gait event time were removed from further analysis. We retained these amounts of epochs for each condition: 161 +/− 47 epochs for 0.25 m/s, 223 +/− 47 epochs for 0.50 m/s, 261 +/− 48 epochs for 0.75 m/s, 280 +/− 57 epochs for 1.0 m/s.

Power spectral density measures were calculated for each condition and each subject, then averaged within each cluster. The power spectral density was calculated using Welch’s method (see. pwelch in MATLAB). The power spectral density calculation was performed from 1 Hz to 200 Hz in bins of 4 Hz. Resulting measures were log transformed to the decibel scale (i.e., $10 \log_{10} (P S D)$ ). We applied spectral principal component analysis to the calculated power spectral density (see Spectral Principal Component Analysis). Spectral component analysis aided in attenuating muscle artifacts not removed in preprocessing. The Fitting Oscillations & One Over F (FOOOF) toolbox (61) was used to separate the aperiodic and periodic components of each power spectra. We applied FOOOF from 3 to 40 Hz with peak limits of [1,8], a minimum peak height of 0.05, and a maximum of 2 peaks. We subtracted out the aperiodic component from subject PSDs to compute a flattened power spectral density trace. For average frequency band power, we calculated the mean within their respective frequency ranges: Theta (3-7 Hz), Alpha (8-12 Hz), and Beta (13-30 Hz).

Spectral Principal Component Analysis.

We used a modified spectral principal component analysis to remove muscle-like artifacts from the power spectral density (62). We adapted the authors’ process used on time-frequency decompositions to be used on power spectral density. Muscle artifacts have large power and can contaminate multiple electrodes across the scalp leading to contamination at the source level after independent component analysis and subsequent frequency-power analyses. We performed a principal component analysis on the power spectral density measures calculated across all sources within each subject. Details on the entire muscle artifact removal process are listed below. Power spectral density data for sources that were previously determined to be brain in the Source Rejection were used for the final analyses.

Step-by-step process for performing spectral principal component analysis (sPCA) on power spectral density to remove muscle artifact:

We determined a grand average power spectral density by calculating the power spectral density for each walking condition and the resting state condition. Power spectral densities were calculated separately for each brain source location.
We baselined the grand average power spectral density by subtracting out the resting state power spectral density from each component.
We applied principal component analysis to the baselined grand average power spectral density in the source-frequency-power space. This procedure determined the principal components in frequency-power space that explained the most variance across all sources. We used an eigenvalue decomposition principal component analysis transformation.
From step 3, we extracted a transformation matrix that converted data from the source-frequency-power space to the principal component analysis space.
As in steps 1-2, we calculated a baselined power spectral density for each speed condition and component.
We reapplied the transformation matrix extracted in step 4 to each source-frequency-power space generated for each speed condition.
We removed the first principal component that represented muscle contamination and then transformed the remaining principal components back into the source-frequency-power space to get our final cleaned power spectral density for each speed condition.
Finally, we added the resting state power spectral density to the cleaned power spectral densities for each speed condition to retain the 1/f EEG power shape that is compatible with FOOOF.

We performed time-frequency decompositions for select brain areas. We segmented each subject’s source data using gait events marked by force data from sensors in their shoes. Gait events are ordered as right heel strike (RHS), left toe off (LTO), left heel strike (LHS), right toe off (RTO). We then time-warped each subject to a grand average time sequence calculated using the median times across all subjects. Time-warped signals were decomposed into frequency content using a Morlet wavelet transform (EEGLAB std_ersp.m; frequency range 1-200 Hz, time resolution of 160, cycles [3, 0.8], frequency factor of 4). Time-frequency data were then cleaned of muscle artifacts using spectral principal component analysis (23). Event-related spectral perturbation data within each group were baselined to the average power across 0.25 m/s, 0.5 m/s, 0.75 m/s, and 1.0 m/s. A univariate permutation based ANOVA test was performed to determine significant differences across speeds for each group (Fieldtrip in EEGLAB, ft_freqstatistics, ft_statistics_montecarlo, ft_statfun_depsamplesFunivariate; Monte Carlo permutation method, cluster-based multiple comparison correction, 2000 permutations).

Statistical Analysis

We used R (2023.12.1 build 402, RRID:SCR_001905) for gait characteristic statistical tests. We evaluated changes in the gait characteristics of coefficient of variation of the step duration (%) and coefficient of variation of pelvic mediolateral excursion (%) (N=88; Figure 4). We chose these gait kinematic parameters as they are correlated to gait stability and health outcomes in younger and older adults (63, 64). We assessed gait characteristics using a linear mixed effects model (lme4:lmer) with independent variables of speed (continuous) and group (factor). Initially, each model was fit with an interaction term between speed and group. If the interaction term was not significant, we removed the term and refit the model. Model quality was inspected using multicollinearity-check, QQ-plot, and a check for normal distribution of residuals (sjPlot:plot_model). Each model included a random intercept for each subject. Subject specific intercepts accurately control for subject-to-subject variability (65). We used a Mixed Model Analysis of Variance (MMANOVA) (a.k.a. analysis of deviance) (car:Anova, Sum of Squares Type 3) to assess significant differences for each independent variable or interaction. We assessed the effect of speeds on behavioral measures using linear mixed effects models with a random intercept for each subject. Multiple comparisons between each group were performed if a significant group or interaction term were found. Multiplicity corrections were performed for the p-values obtained by applying emmeans:emmeans function for those analyses using the false discovery rate (FDR) approach (66). We provide confidence intervals (CI 95%) of the estimated marginal means (emmeans:emmeans) to further visualize group differences. Significance level was set to alpha < 0.05 for all statistical tests.

We used MATLAB 2023b to test for differences in EEG power spectral density signatures. To understand how faster walking speeds concurred with changes in electrocortical dynamics, we quantified EEG spectral power in different brain areas while they walked. We performed stats on the power spectral density (PSD), the periodic component of the power spectral density, and the mean power in three distinct frequency bands: theta (4-8 Hz), alpha (8-13 Hz), and beta (13-30 Hz). We used the FOOOF toolbox to separate the aperiodic (e.g., the dashed lines in Figure 8.A) and periodic components (e.g., the plots displayed in Figure 8.B) of each power spectral density plot. We tested for significant changes in power spectral density and the periodic component using nonparametric permutation statistics (Fieldtrip in EEGLAB, α=0.05, 2000 iterations).

**(A)** Average scalp topography for all subjects in the cluster and the number of subjects from each group: younger adults **(YA**), older high functioning adults (**OHFA**), and older low functioning adults (**OLFA**). Toward the right of the topography are dipole locations overlaid on the Montreal Neurological Institute (MNI) template for each component in the cluster distinguished by group: younger adults (circles), older high functioning adults (triangles), older low functioning adults (diamonds). **(B)** Average power spectral densities (PSD) for younger adults (left), older high functioning adults (center), and older low functioning adults (right). Solid lines show cluster averages for each speed 0.25 (purple), 0.50 (dark blue), 0.75 (light blue), and 1.0 (green) m/s. Shaded colors corresponding to each speed represent the standard error of the power spectral densities. Vertical dashed lines mark the frequency bands of interest: theta (4-8 Hz), alpha (8-13 Hz), and beta (13-30 Hz). **(C)** Flattened average power spectral density for each walking speed. Flattening was performed by subtracting the FOOOF calculated aperiodic fit from each subject’s original power spectral density. Shaded colors corresponding to each speed represent the standard error of the power spectral densities. Shaded gray areas covering the entire vertical range of the graph are frequencies where we found significant differences across conditions using a permutation statistic. **(D)** Average power within each frequency band. From left to right: mean theta power, mean alpha power, and mean beta power. Each walking speed is colorized the same as in the power spectral density plots: 0.25 (purple), 0.50 (dark blue), 0.75 (light blue), and 1.0 (green) m/s. In each plot the 3 groups are presented in clusters representing the trends of the averaged EEG power band with speed for that group. If the linear mixed effects model found a significant speed or group effect for that EEG power band, we presented a significance level (*p<0.05, **p<0.01, ***p<0.001), a Cohen’s f² (partial) effect size (s = speed, g = group, s:g = interaction), and conditional correlation coefficient (R²). If a significant speed or interaction effect was found we displayed a linear trend equation over each group cluster. If a significant group effect was found, we display each groups estimated marginal mean and its 95% confidence interval as an “I” error bar. **Summary:** We found modulations of beta, alpha, and theta EEG power with faster walking speeds. Specifically, we found that OLFA showed differences in flattened PSD theta and beta power (C, Right) with faster walking speeds. Mean PSD outcomes showed a significant effect for speed in all EEG power bands (D; Left, Right, Center), but there was no main effect for group in any of the bands. Although no group effects were found, we observed that older adults had lower estimated marginal means in theta power (D, Left; μ_ohfa=0.43, μ_olfa=0.50) than younger adults (μ_ya=0.86), and greater estimated marginal means in beta power (D, Right; μ_ohfa=2.3, μ_olfa=2.9) than younger adults (μ_ya=1.8).

We used R (2023.12.1 build 402) to test for differences in mean flattened power spectral density. We calculated the mean across theta (4-8 Hz), alpha (8-13 Hz), and beta (13-30 Hz) frequency bands of the flattened power spectral density. Here, we follow a statistical pipeline described for the behavioral statistics above. We fit linear mixed effect models (lme4:lmer) with main effects terms for speed (continuous), group (factor), interaction, and random intercepts for each subject in the modeled brain area (65). We inspected the model fit to ensure model conformity (sjPlot:plot_model). An MMANOVA (car:Anova, Sum of Squares Type 3) confirmed significant differences in mean EEG spectral power density with respect to speed, group, and/or their interaction. If no significant interaction was found, we refit the model using only the speed and group effects. Multiple comparisons were performed given a significant group or interaction term (emmeans:emmeans), and they were corrected for multiplicity using FDR. Confidence intervals are provided in the same fashion as behavioral data analysis.

Cohen’s f² effect sizes were calculated for all linear mixed effects models (emmeans:emmeans), where a value greater than 0.02 is a small effect, a value greater than 0.15 is a medium effect, and a value greater than 0.35 is a large effect (67). We used a local (partial) Cohen’s f², which is the effect a specific term has on the fit of the model (68). First, we calculated the proportion of variance accounted for by the full model ( $R_{A B}^{2}$ , Equation 1):

R_{A B}^{2} = \frac{V_{n u l l} - V_{A B}}{V_{n u l l}}

(Equation 1)

We calculated the residual variance from a model containing the variable of interest (A) and all other variables (B) and the random intercept terms ( $V_{A B}^{⬚}$ ) and a model containing just the main intercept term and the random intercept terms ( $V_{n u l l}^{⬚}$ ). We performed this same calculation using a model containing only the variable of interest ( $R_{A}^{2}$ ). We then calculated Cohen’s $f^{2}$ local effect size of A using the $R^{2}$ terms as in (Equation 2):

f^{2} = \frac{R_{A B}^{2} - R_{A}^{2}}{1 - R_{A B}^{2}}

(Equation 2)

RESULTS

Behavioral Differences in Age Groups

The older low mobility adult group had the slowest overground 3-meter walking speed out of all groups observed. Figure 5.A shows differences in 3-meter overground walking speed by participant group. The younger adult group walked the fastest (mean 1.1, 95% CI [1.0,1.2]) and the older low function adult group walked the slowest (mean 0.73, 95% CI [0.61,0.84]) (Figure 5.A). Multiple comparisons revealed differences between younger adults and older low functioning adults (p<0.001) and between older high functioning adults and older low functioning adults (p=0.0027).

We found that faster walking speeds were associated with greater mediolateral pelvis movement variability, and greater step duration variability [Figure 7]. We found no significant interactions for step duration coefficient of variation. All three groups’ mediolateral coefficient of variation increased with increasing speed. We found a significant interaction between the group and speed terms (Cohen’s f² (partial) = 0.04, small) [Figure 5.B]. Older high functioning adults showed the greatest sensitivity in mediolateral coefficient of variation with faster walking speeds (m=10) compared to younger (m=5.5) and older low functioning (m=4.8) adults. Additionally, older high functioning adults exhibited lower initial mediolateral coefficient of variation at the slowest speed (b=7.7) compared to younger (b=11) and older low functioning adults (b=11). For step duration coefficient of variation, we found a main effect of treadmill speed (Cohen’s $f^{2}$ (partial) = 1.33, large) [Figure 5.C]. Details of each statistical analysis are available in [Table 3].

Boxes contain the anatomy labels of the brain areas and summary of differences in mean EEG power spectral density with faster walking speed, participant group, and (if appropriate) their interaction. We tested the differences in for mean power spectral density changes in theta (θ, 4-7 Hz), alpha (α, 8-12 Hz), and beta (β, 13-30 Hz). Changes in EEG power with faster walking speeds is indicated using arrows, where an increase in power with faster walking speeds is denoted by an upwards arrow and a decrease in power is denoted with a downwards arrow. Significant differences in group or interactions are denoted using a star symbol. We symbolize Cohen’s f² (partial) effect sizes for each model term using variations of each symbol’s color format. Black symbols are Cohen’s f² > 0.35 (large), gray symbols are f² > 0.15 (medium), and white symbols are f² > 0.02 (small). Brain area centroids are displayed in the central MNI MRI image as different colored dots. Brain anatomy shorthand’s are displayed within each centroid for readability. A visual depiction of a subject walking on a treadmill is provided.

Table 3.

Statistics Summary of Behavioral Data

EEG Band	Term	DF	β (95% CI)¹	Χ²-statistic	P-value	EMM (95% CI)²
Mediolateral Excursion COV (Obs. = 351 )	Speed	1	5.5 (3.3,7.8)	24	<0.001***
	Group	2		8.1	0.017*
	YA³		11 (9.3,13)			15 (13,16)
	OHFA		−3.4 (−6.1,−0.70)			14 (13,15)
	OLFA		0.10 (−2.5,2.7)			14 (13,15)
	Interaction	2		11	0.0033**
	OHFA		4.5 (1.3,7.8)
	OLFA		−0.78 (−4.0,2.4)
Step Duration COV (Obs. = 352)	Speed	1	−5.7 (−6.3,−5.1)	350	<0.001***
	Group	2		1.4	0.51
	YA³		8.7 (8.2,9.3)			5.2 (4.7,5.6)
	OHFA		0.26 (−0.39,0.90)			5.4 (4.9,5.9)
	OLFA		0.36 (−0.26,0.99)			5.5 (5.1,6.0)
Overground Walking Speed (Obs. = 82)	Group	2		13	<0.001***
	YA³		1.1 (1.0,1.2)			1.1 (1.0,1.2)
	OHFA		−0.10 (−0.26,0.05)			1.0 (0.89,1.1)
	OLFA		−0.38 (−0.54,−0.23)			0.73 (0.61,0.84)
	YA – OHFA⁴				0.38
	YA – OLFA⁴				<0.001***
	OHFA – OLFA⁴				0.0027**

Open in a new tab

Linear mixed effects model coefficient (β) and the associate confidence interval.

Estimated marginal means for each group term: younger adults (YA), older low functioning adults (OLFA), and older high functioning adults (OHFA).

For the β column this row refers to the intercept term in the linear mixed effects model, while for the estimated marginal means column this row refers to the YA group.

⁴

Multiple comparisons using estimated marginal means. P-values are corrected for Type I errors using the false discovery rate (FDR).

Identification of Brain Areas Involved in Walking

We Identified 7 clusters of dipoles that included more than half of participants within each group that were related to electrocortical activity during walking [Table 4]. Coordinates for the clusters and anatomical designations are included in Table 5. We identified brain regions using the centroid of the cluster dipole positions and the most frequent region identified for each subject’s dipole. We found clusters in the right and left sensorimotor, right and left posterior parietal, mid cingulate, left supplementary motor, and right cuneus. We also identified 4 additional clusters that did not meet our criteria of having more than 15 subjects for each group: left and right occipital, and left and right temporal [Figure 17-Figure 20; Table 12-Table 15].

Table 4.

Number of subjects per group (i.e., younger adults, older low functioning adults, and older high functioning adults) within each cluster. See Table 5 and Fig. 7. and See Table 5 and Fig. 7 for more information on the location of the clusters

Cluster Centroid	Color	Total N	Younger Adults	Older Low Mobility Adults	Older High Mobility Adults
*L Sensorimotor*	Blue	75	28	21	26
*R Sensorimotor*	Orange	62	23	17	22
*L Posterior Parietal*	Olive	71	28	20	23
*R Posterior Parietal*	Yellow	69	28	19	22
*Mid Cingulate*	Green	60	21	19	20
*R Cuneus*	Light Blue	79	27	24	28
*L Supplementary*	Pink	66	25	20	21
*R Occipital*	N/A	44	16	14	14
*L Occipital*	N/A	38	19	7	12
*L Temporal*	N/A	51	18	13	20
*R Temporal*	N/A	38	13	10	15

Open in a new tab

Table 5.

Contains the chosen cluster anatomy, the color of the anatomy depicted in the MNI images, the number of independent components in each cluster, the MNI coordinates of the centroid of the cluster, the atlas determined through the most frequent anatomy label among all dipoles in the cluster, and the atlas determined by the location of the centroid of the cluster.

Cluster Centroid	Color	Total N	MNI Coordinate	Aggregate Atlas Label	Centroid Atlas Label
*L Sensorimotor*	Blue	75	[−23, −27, 65]	L Precentral	L Postcentral
*R Sensorimotor*	Orange	62	[28, −4.3, 55]	R Precentral	R 2 Sup. Frontal
*L Posterior Parietal*	Olive	71	[−31, −57, 34]	L Mid. Cingulate	L Angular
*R Posterior Parietal*	Yellow	69	[32, −45, 50]	R Precentral	R Inf. Parietal
*Mid Cingulate*	Green	60	[−4.4, −16, 26]	L Supp. Motor Area	No Label Found
*R Cuneus*	Light Blue	79	[13, −74, 36]	R Mid. Cingulate	R Cuneus
*L Supplementary*	Pink	66	[−6.6, 24, 48]	L 2 Sup. Frontal	L Supp. Area Motor
*R Occipital*	N/A	44	[43, −45, −1.5]	R Rolandic Oper.	R Mid Temporal
*L Occipital*	N/A	38	[−18, −71, −13]	L Calcarine	L 6 Cerebellum
*L Temporal*	N/A	51	[−40.0, −19, −5.7]	L Inf. Frontal Oper.	L Insula
*R Temporal*	N/A	38	[21, 9.5, 0.30]	R Putamen	R Precentral

Open in a new tab

Table 12.

Summary statistics for right occipital.

Cluster Anatomy	EEG Band	Term	DF	β (95% CI)¹	Χ²-statistic	P-value	EMM (95% CI)²
Right Occipital (N = 176)	Theta	Speed	1	0.14 (−0.03,0.31)		0.10
		Group	2			0.092
		YA³		0.68 (0.17,1.2)			0.77 (0.25,1.29)
		OHFA		−0.50 (−1.2,0.23)			0.27 (−0.29,0.82)
		OLFA		0.34 (−0.40,1.1)			1.1 (0.55,1.7)
	Alpha	Speed	1	−0.41 (−0.72,−0.10)	6.8	0.009**
		Group	2		2.5	0.29
		YA³		3.8 (2.5,5.2)			3.6 (2.2,5.0)
		OHFA		−0.88 (−2.8,1.1)			2.7 (1.2,4.2)
		OLFA		0.74 (−1.2,2.7)			4.3 (2.9,5.8)
	Beta	Speed	1	−0.19 (−0.45,0.07)	2.1	0.15
		Group	2		1.0	0.60
		YA³		1.4 (0.87,1.9)			1.3 (0.76,1.8)
		OHFA		−0.14 (−0.91,0.63)			1.0 (0.47,1.6)
		OLFA		0.26 (−0.51,1.0)			1.2 (0.68,1.8)
		Interaction	2		6.4	0.041*
		OHFA		−0.17 (−0.56,0.21)
		OLFA		−0.49 (−0.88,−0.11)
		YA-OHFA⁴				n/a
		YA-OLFA⁴				n/a
		OHFA-OLFA⁴				n/a

Open in a new tab

Linear mixed effects model coefficient (β) and the associate confidence interval.

Estimated marginal means for each group term: younger adults (YA), older low functioning adults (OLFA), and older high functioning adults (OHFA).

For the β column this row refers to the intercept term in the linear mixed effects model, while for the estimated marginal means column this row refers to the YA group.

⁴

Multiple comparisons using estimated marginal means. P-values are corrected for Type I errors using the false discovery rate (FDR)..

Table 15.

Summary statistics for right temporal.

Cluster Anatomy	EEG Band	Term	DF	β (95% CI)¹	Χ²-statistic	P-value	EMM (95% CI)²
Right Temporal (N = 152)	Theta	Speed	1	0.28 (0.12,0.44)	12	<0.001***
		Group	2		0.22	0.89
		YA³		0.51 (0.06,0.97)			0.69 (0.23,1.1)
		OHFA		0.16 (−0.51,0.84)			0.85 (0.33,1.4)
		OLFA		0.09 (−0.52,0.69)			0.78 (0.35,1.2)
	Alpha	Speed	1	−0.12 (−0.34,0.11)	1.1	0.30
		Group	2		4.9	0.085
		YA³		0.92 (−0.06,1.9)			0.84 (−0.15,1.8)
		OHFA		1.6 (0.09,3.0)			2.4 (1.3,3.5)
		OLFA		1.1 (−0.19,2.4)			2.0 (1.0,2.9)
	Beta	Speed	1	−0.14 (−0.26,−0.02)	5.0	0.026*
		Group	2		2.1	0.35
		YA³		0.86 (0.44,1.3)			0.78 (0.35,1.2)
		OHFA		0.03 (−0.60,0.66)			0.81 (0.32,1.3)
		OLFA		−0.36 (−0.92,0.21)			0.42 (0.020,0.82)

Open in a new tab

Linear mixed effects model coefficient (β) and the associate confidence interval.

Estimated marginal means for each group term: younger adults (YA), older low functioning adults (OLFA), and older high functioning adults (OHFA).

For the β column this row refers to the intercept term in the linear mixed effects model, while for the estimated marginal means column this row refers to the YA group.

⁴

Multiple comparisons using estimated marginal means. P-values are corrected for Type I errors using the false discovery rate (FDR)..

EEG Spectral Power with Changes in Speed

Changes in spectral power with faster speeds generally concurred with the two main hypotheses. There were significant differences in spectral power with faster walking for all brain areas observed: sensorimotor [Figure 8,Figure 9], posterior parietal [Figure 10,Figure 11], mid cingulate [Figure 12], supplementary motor [Figure 13], and right cuneus [Figure 14] cortex. Statistic model information and outcomes are available for sensorimotor [Table 6], left posterior parietal [Table 7], right posterior parietal [Table 8], supplementary motor [Table 10], mid cingulate [Table 9], and right cuneus [Table 11]. Faster speeds showed increases in theta power in sensorimotor [Figure 8.D, Figure 9.D Left], posterior parietal [Figure 10.D Left], mid cingulate [Figure 12.D Left], and supplementary motor [Figure 13.D] cortex. We observed decreases in alpha power concurrent with faster walking speeds in left sensorimotor [Figure 8.D Center], right posterior parietal [Figure 11.D Center], and right cuneus [Figure 14.D Center]. Additionally, we observed decreases in beta power in the sensorimotor [Figure 8.D,Figure 9.D Right], posterior parietal [Figure 10.D, Figure 11.D Right], mid cingulate [Figure 12.D, Right], right cuneus [Figure 14.D, Right] and supplementary motor [Figure 13.D] cortex. In addition to our mean power spectral density measures, there were significant differences in power spectral density signature across speed in discrete frequencies. Generally, we observed significant differences in power spectral density around theta band and decreases around beta band [Figure 8.C Right, Figure 12.C Right & Center, and Figure 13.C Right].

Table 6.

Statistics summary for right and left sensorimotor.

Cluster Anatomy	EEG Band	Term	DF	β (95% CI)¹	Χ²-statistic	P-value	EMM (95% CI)²
Right Sensorimotor (Obs. = 248)	Theta	Speed	1	0.28 (0.15,0.41)	17	<0.001***
		Group	2		4.9	0.085
		YA³		0.69 (0.40,0.98)			0.86 (0.58,1.1)
		OHFA		−0.43 (−0.86,−0.01)			0.43 (0.10,0.76)
		OLFA		−0.36 (−0.76,0.04)			0.50 (0.21,0.79)
	Alpha	Speed	1	−0.22 (−0.45,0.02)	3.3	0.071
		Group	2		0.19	0.19
		YA³		2.5 (1.5,3.5)			2.3 (1.3,3.3)
		OHFA		−0.32 (−1.8,1.2)			2.0 (0.87,3.2)
		OLFA		−0.21 (−1.6,1.2)			2.13 (1.1,3.2)
	Beta	Speed	1	−0.32 (−0.47,−0.16)	16	<0.001***
		Group	2		4.0	0.13
		YA³		2.0 (1.2,2.7)			1.8 (1.0,2.5)
		OHFA		0.50 (−0.64,1.6)			2.3 (1.4,3.2)
		OLFA		1.1 (0.02,2.1)			2.9 (2.1,3.6)
Left Sensorimotor (Obs. = 300)	Theta	Speed	1	0.24 (0.14,0.34)	24	<0.001***
		Group	2		5.6	0.062
		YA³		0.47 (0.22,0.72)			0.62 (0.37,0.87)
		OHFA		−0.30 (−0.67,0.08)			0.32 (0.033,0.61)
		OLFA		−0.42 (−0.77,−0.06)			0.20 (−0.055,0.47)
	Alpha	Speed	1	−0.30 (−0.59,−0.01)	4.0	0.044*
		Group	2		0.40	0.82
		YA³		3.0 (1.9,4.1)			2.8 (1.8,3.9)
		OHFA		0.32 (−1.3,1.9)			3.1 (1.9,4.4)
		OLFA		−0.21 (−1.7,1.3)			2.6 (1.5,3.7)
	Beta	Speed	1	−0.39 (−0.53,−0.25)	28	<0.001
		Group	2		3.4	0.18
		YA³		2.4 (1.8,3.1)			2.2 (1.5,2.8)
		OHFA		0.78 (−0.19,1.8)			3.0 (2.2,3.7)
		OLFA		0.74 (−0.18,1.7)			2.9 (2.3,3.6)

Open in a new tab

Linear mixed effects model coefficient (β) and the associate confidence interval.

Estimated marginal means for each group term: younger adults (YA), older low functioning adults (OLFA), and older high functioning adults (OHFA).

For the β column this row refers to the intercept term in the linear mixed effects model, while for the estimated marginal means column this row refers to the YA group.

Table 7.

Statistics summary for left posterior parietal

Cluster Anatomy	EEG Band	Term	DF	β (95% CI)¹	Χ²-statistic	P-value	EMM (95% CI)²
Left Posterior Parietal (N = 284)	Theta	Speed	1	0.07 (−0.09,0.23)	0.69	0.41
		Group	2		1.4	0.50
		YA³		0.52 (0.18,0.86)			0.56 (0.23,0.89)
		OHFA		−0.26 (−0.79,0.27)			0.49 (0.096,0.88)
		OLFA		0.05 (−0.46,0.56)			0.76 (0.39,1.1)
		Interaction	2		6.2	0.045*
		OHFA		0.30 (0.04,0.55)
		OLFA		0.24 (−0.01,0.48)
		YA-OHFA⁴				0.86
		YA-OLFA⁴				0.86
		OHFA-OLFA⁴				0.86
	Alpha	Speed	1	−0.23 (−0.46,0.01)	3.6	0.059
		Group	2		1.1	0.57
		YA³		4.2 (3.3,5.1)			4.1 (3.1,5.0)
		OHFA		−0.39 (−1.8,1.0)			3.7 (2.6,4.8)
		OLFA		−0.73 (−2.1,0.62)			3.3 (2.3,4.4)
	Beta	Speed	1	−0.23 (−0.38,−0.08)	9.5	0.002**
		Group	2		0.046	0.98
		YA³		2.3 (1.8,2.8)			2.2 (1.7,2.7)
		OHFA		−0.07 (−0.84,0.69)			2.1 (1.5,2.7)
		OLFA		−0.06 (−0.79,0.67)			2.1 (1.6,2.7)

Open in a new tab

Linear mixed effects model coefficient (β) and the associate confidence interval.

Estimated marginal means for each group term: younger adults (YA), older low functioning adults (OLFA), and older high functioning adults (OHFA).

For the β column this row refers to the intercept term in the linear mixed effects model, while for the estimated marginal means column this row refers to the YA group.

⁴

Multiple comparisons using estimated marginal means. P-values are corrected for Type I errors using the false discovery rate (FDR)..

Table 8.

Statistics summary for right posterior parietal

Cluster Anatomy	EEG Band	Term	DF	β (95% CI)¹	Χ²-statistic	P-value	EMM (95% CI)²
Right Posterior Parietal (Obs. = 276)	Theta	Speed	1	0.59 (0.30,0.88)	0.93	0.3d3
		Group	2		3.4	0.18
		YA³		0.59 (0.30,0.88)			0.62 (0.34,0.91)
		OHFA		−0.42 (−0.86,0.02)			0.21 (−0.14,0.55)
		OLFA		−0.18 (−0.60,0.24)			0.45 (0.13,0.77)
	Alpha	Speed	1	−0.41 (−0.68,−0.14)	9.1	0.0026**
		Group	2		1.8	0.41
		YA³		4.1 (3.1,5.1)			3.8 (2.8,4.8)
		OHFA		−0.22 (−1.7,1.3)			3.6 (2.4,4.8)
		OLFA		−0.96 (−2.4,0.50)			2.9 (1.7,4.0)
	Beta	Speed	1	−0.24 (−0.37,−0.10)	12	<0.001***
		Group	2		1.4	0.49
		YA³		2.2 (1.6,2.8)			2.1 (1.5,2.7)
		OHFA		0.53 (−0.40,1.5)			2.6 (1.9,3.3)
		OLFA		0.38 (−0.51,1.3)			2.5 (1.8,3.2)

Open in a new tab

Linear mixed effects model coefficient (β) and the associate confidence interval.

Estimated marginal means for each group term: younger adults (YA), older low functioning adults (OLFA), and older high functioning adults (OHFA).

For the β column this row refers to the intercept term in the linear mixed effects model, while for the estimated marginal means column this row refers to the YA group.

Table 10.

Summary statistics for left supplementary motor.

Cluster Anatomy	EEG Band	Term	DF	β (95% CI)¹	Χ²-statistic	P-value	EMM (95% CI)²
Left Supplementary Motor (Obs. = 264)	Theta	Speed	1	0.38 (0.22,0.54)	21	<0.001***
		Group	2		2.6	0.27
		YA³		1.0 (0.71,1.4)			1.3 (0.96,1.6)
		OHFA		−0.33 (−0.79,0.14)			0.95 (0.60,1.3)
		OLFA		−0.33 (−0.79,0.13)			0.95 (0.61,1.3)
	Alpha	Speed	1	0.86 (0.45,1.3)	3.7	0.053
		Group	2		0.12	0.94
		YA³		0.86 (0.45,1.3)			0.96 (0.55,1.4)
		OHFA		0.10 (−0.50,0.70)			1.1 (0.60,1.5)
		OLFA		0.02 (−0.57,0.62)			0.98 (0.54,1.4)
	Beta	Speed	1	−0.15 (−0.25,−0.05)	8.6	0.003**
		Group	2		0.87	0.65
		YA³		1.3 (0.88,1.8)			1.2 (0.79,1.7)
		OHFA		0.20 (−0.47,0.87)			1.4 (0.93,1.9)
		OLFA		0.30 (−0.35,0.96)			1.5 (1.0,2.0)

Open in a new tab

Linear mixed effects model coefficient (β) and the associate confidence interval.

Estimated marginal means for each group term: younger adults (YA), older low functioning adults (OLFA), and older high functioning adults (OHFA).

For the β column this row refers to the intercept term in the linear mixed effects model, while for the estimated marginal means column this row refers to the YA group.

Table 9.

Statistics summary for mid cingulate.

Cluster Anatomy	EEG Band	Term	DF	β (95% CI)¹	Χ²-statistic	P-value	EMM (95% CI)²
Mid Cingulate (Obs. = 240)	Theta	Speed	1	0.42 (0.30,0.55)	46	<0.001***
		Group	2		7.1	0.029*
		YA³		0.68 (0.38,0.97)			0.94 (0.65,1.2)
		OHFA		−0.34 (−0.75,0.070)			0.60 (0.30,0.91)
		OLFA		−0.54 (−0.95,−0.14)			0.40 (0.10,0.69)
		YA-OHFA⁴				0.25
		YA-OLFA⁴				0.029*
		OHFA-OLFA⁴				0.60
	Alpha	Speed	1	−0.15 (−0.38,0.07)	1.9	0.17
		Group	2		0.88	0.64
		YA³		1.9 (1.2,2.5)			1.8 (0.65,1.2)
		OHFA		−0.31 (−1.2,0.55)			1.4 (0.30,0.91)
		OLFA		0.09 (−0.76,0.94)			1.8 (0.10,0.69)
	Beta	Speed	1	−0.29 (−0.40,−0.18)	29	<0.001***
		Group	2		2.8	0.24
		YA³		0.97 (0.53,1.4)			0.79 (0.35,1.2)
		OHFA		0.34 (−0.29,0.97)			1.1 (0.66,1.6)
		OLFA		0.53 (−0.10,1.1)			1.3 (0.86,1.8)

Open in a new tab

Linear mixed effects model coefficient (β) and the associate confidence interval.

Estimated marginal means for each group term: younger adults (YA), older low functioning adults (OLFA), and older high functioning adults (OHFA).

For the β column this row refers to the intercept term in the linear mixed effects model, while for the estimated marginal means column this row refers to the YA group.

⁴

Multiple comparisons using estimated marginal means. P-values are corrected for Type I errors using the false discovery rate (FDR).

Table 11.

Summary Statistics for right cuneus.

Cluster Anatomy	EEG Band	Term	DF	β (95% CI)¹	Χ²-statistic	P-value	EMM (95% CI)²
Right Cuneus (Obs. = 316)	Theta	Speed	1	0.00 (−0.16,0.15)	0.0015	0.97
		Group	2		20.0	<0.001***
		YA³		0.70 (0.33,1.1)			0.70 (0.33,1.1)
		OHFA		0.37 (−0.16,0.90)			1.1 (0.68,1.5)
		OLFA		1.1 (0.63,1.7)			1.8 (1.5,2.2)
		YA-OHFA⁴				0.35
		YA-OLFA⁴				<0.001***
		OHFA-OLFA⁴				0.015*
	Alpha	Speed	1	−0.31 (−0.58,−0.05)	5.4	0.020*
		Group	2		2.4	0.29
		YA³		5.1 (4.2,6.0)			4.9 (4.0,5.8)
		OHFA		−0.85 (−2.2,0.49)			4.1 (3.1,5.0)
		OLFA		−0.95 (−2.2,0.35)			4.0 (3.0,4.9)
	Beta	Speed	1	−0.16 (−0.28,−0.03)	5.5	0.019*
		Group	2		5.0	0.083
		YA³		2.4 (2.0,2.9)			2.3 (1.9,2.8)
		OHFA		−0.34 (−0.97,0.30)			2.0 (1.5,2.5)
		OLFA		−0.70 (−1.3,−0.8)			1.6 (1.2,2.1)

Open in a new tab

Linear mixed effects model coefficient (β) and the associate confidence interval.

Estimated marginal means for each group term: younger adults (YA), older low functioning adults (OLFA), and older high functioning adults (OHFA).

For the β column this row refers to the intercept term in the linear mixed effects model, while for the estimated marginal means column this row refers to the YA group.

⁴

Multiple comparisons using estimated marginal means. P-values are corrected for Type I errors using the false discovery rate (FDR).

Also as hypothesized, there were group differences in mean power spectral density. Interestingly, we only observed group or interaction effects in mean theta power. We found main effects of group in mid cingulate [Figure 12.D Right] and right cuneus [Figure 14.D Right] theta power. Additionally, we found an interaction effect in left posterior parietal theta [Figure 10.D left]. In mid cingulate EEG theta power was greater overall in younger adults (μ_ya=0.94, 95% CI [0.65,1.2]) than in older adults (μ_ohfa=0.60, 95% CI [0.30,0.91]; μ_olfa=0.40 95% CI [0.10,0.69]). The only multiple comparison that showed significant differences was between the older low functioning adults and the younger adults (p=0.029). In right cuneus we found an opposite trend compared to mid cingulate. Older adults exhibited overall higher EEG theta power (μ_ohfa=1.1, 95% CI [0.68,1.5]; μ_olfa=1.8 95% CI [1.5,2.2]) than younger adults (μ_ya=0.70, 95% CI [0.33,1.1]). Multiple comparisons showed no differences between older high functioning adults and younger adults (p=0.35), but they did show differences between older low functioning adults and the other two groups (p_ya-olfa<0.001, p_ohfa-olfa=0.015). In left posterior parietal cortex, we found an interaction between faster walking speed and the age groups. We found that older adults had greater slope terms (m_ohfa=0.37, m_olfa=0.31) than younger adults (m_ya=0.07). We did not find significant differences in overall EEG theta power in the multiple comparisons between groups. We did observe less overall EEG theta power in older high functioning adults (μ_ohfa=0.49 95% CI [0.096,0.88]) than older low functioning adults (μ_olfa=0.76 95% CI [0.39,1.1]).

Additional Brain Areas

We found additional brain areas that did not meet our requirement of having more than 15 subjects per group in the brain cluster. We present right occipital (Figure 17 & Table 12), left occipital (Figure 18 & Table 13), right temporal (Figure 20 & Table 15), and left temporal (Figure 19 & Table 14). We provide these results to be transparent in our findings of the study, but we caution the readers in the interpretation of the results presented in the figures given they did not meet our requirement.

Table 13.

Summary statistics for left occipital.

Cluster Anatomy	EEG Band	Term	DF	β (95% CI)¹	Χ²-statistic	P-value	EMM (95% CI)²
Left Occipital (N = 152)	Theta	Speed	1	0.40 (0.14,0.65)	9.1	0.003
		Group	2		13	0.001
		YA³		0.08 (−0.29,0.44)			0.32 (−0.019,0.66)
		OHFA		1.3 (0.59,2.0)			1.5 (0.94,2.1)
		OLFA		0.53 (−0.06,1.1)			1.2 (0.80,1.7)
		Interaction	2		12	0.003
		OHFA		−0.17 (−0.67,0.32)
		OLFA		0.62 (0.21,1.0)
		YA-OHFA⁴				n/a
		YA-OLFA⁴				n/a
		OHFA-OLFA⁴				n/a
	Alpha	Speed	1	0.11 (−0.21,0.42)	0.45	0.50
		Group	2		0.023	0.99
		YA³		2.3 (1.5,3.2)			2.4 (1.5,3.2)
		OHFA		0.02 (−1.6,1.6)			2.4 (1.0,3.8)
		OLFA		−0.09 (−1.4,1.2)			2.3 (1.2,3.4)
	Beta	Speed	1	−0.02 (−0.14,0.09)	0.15	0.70
		Group	2		0.59	0.75
		YA³		0.78 (0.42,1.1)			0.77 (0.40,1.1)
		OHFA		−0.01 (−0.69,0.68)			0.77 (0.15,1.4)
		OLFA		−0.21 (−0.79,0.36)			0.55 (0.091,1.0)

Open in a new tab

Linear mixed effects model coefficient (β) and the associate confidence interval.

Estimated marginal means for each group term: younger adults (YA), older low functioning adults (OLFA), and older high functioning adults (OHFA).

For the β column this row refers to the intercept term in the linear mixed effects model, while for the estimated marginal means column this row refers to the YA group.

⁴

Multiple comparisons using estimated marginal means. P-values are corrected for Type I errors using the false discovery rate (FDR)..

Table 14.

Summary statistics for left temporal.

Cluster Anatomy	EEG Band	Term	DF	β (95% CI)¹	Χ²-statistic	P-value	EMM (95% CI)²
Left Temporal (N = 204)	Theta	Speed	1	0.15 (0.00,0.31)	3.7	0.056
		Group	2		0.15	0.93
		YA³		0.64 (0.31,0.98)			0.74 (0.41,1.1)
		OHFA		0.08 (−0.42,0.58)			0.82 (0.43,1.2)
		OLFA		−0.01 (−0.45,0.44)			0.73 (0.42,1.0)
	Alpha	Speed	1	2.1 (1.2,3.1)	7.4	0.0065
		Group	2		7.7	0.021
		YA³		2.1 (1.2,3.1)			1.9 (0.96,2.9)
		OHFA		1.9 (0.43,3.3)			3.8 (2.7,4.9)
		OLFA		1.4 (0.14,2.7)			3.4 (2.5,4.3)
		YA-OHFA⁴				n/a
		YA-OLFA⁴				n/a
		OHFA-OLFA⁴				n/a
	Beta	Speed	1	−0.28 (−0.40,−0.16)	22	<0.001
		Group	2		0.41	0.81
		YA³		1.0 (0.64,1.5)			0.87 (0.46,1.3)
		OHFA		0.06 (−0.57,0.68)			0.93 (0.43,1.4)
		OLFA		−0.13 (−0.69,0.43)			0.74 (0.35,1.1)

Open in a new tab

Linear mixed effects model coefficient (β) and the associate confidence interval.

Estimated marginal means for each group term: younger adults (YA), older low functioning adults (OLFA), and older high functioning adults (OHFA).

For the β column this row refers to the intercept term in the linear mixed effects model, while for the estimated marginal means column this row refers to the YA group.

⁴

Multiple comparisons using estimated marginal means. P-values are corrected for Type I errors using the false discovery rate (FDR)..

DISCUSSION

Overall, our results support our hypotheses that motor areas of the cortex show increased theta spectral power and decreased beta spectral power at faster walking speeds in older adults. There were a few small differences between lower and higher functioning older adults with changes in speed, but they were neither large nor consistent. The fastest walking speed (1.0 m/s) was faster than many of our older adults’ preferred overground walking speed but less than most of our younger adults’ preferred overground walking. In some brain areas, older adult EEG theta power was differentiated from younger adult theta power. Counter to our hypothesis, there were minor inconclusive differences observed between our older low functioning adults (OLFA) and older high functioning adults (OHFA) groups. Generally, our observations indicate that older adults require more cortical resources than younger adults while walking at faster speeds but that EEG spectral power was not a strong metric to differentiate high functioning and low functioning older adults.

While we expected differences between younger adults, older low functioning adults, and older high functioning adults, we found only a few differences between groups for the EEG measures hypothesized. Based on the previous literature (4, 10, 17-19), we had thought that the older lower functioning adults would need to make higher compensations in brain activity compared to higher functioning (11, 69, 70). The basic premise was that older lower functioning adults would have to involve more brain areas and to a greater extent at faster speeds compared to the higher functioning older adults. Our EEG results did not provide strong evidence for this. The difference in preferred overground walking speed between older lower and higher functioning adults was 0.68 m/s to 0.99 m/s, respectively. There were differences in gait kinematics between the two groups where higher functioning adults had steeper changes in mediolateral excursion at faster speeds compared to older low functioning adults [Figure 5]. However, these biomechanical differences are not reflected in their EEG data. We found groupwise differences in left posterior parietal, mid cingulate, and right cuneus. The interaction observed in left posterior parietal supports our claim that older adults would have greater changes in EEG power at faster speeds. Older adults had greater sloping terms compared to younger adults in mean EEG theta power [Figure 12.D Left], with older low functioning and high functioning adults having comparable slopes. We did not expect, however, that older adults’ theta power in mid cingulate would be less than younger adults’ theta power, nor that older adults’ theta power in right cuneus would be greater than younger adults’ for all speeds. There are differences in theta power during rest between younger and older adults (71-73). , but associations between ageing and task-related theta power is unclear. We did observe a difference between older low functioning adults and younger adults in mid cingulate theta. However, the comparison may be unreliable given the overlap in confidence intervals was greater than 25% (74).

There were widespread increases in theta spectral power with faster walking speeds for all the participants [Figure 7]. Theta synchronization (i.e., increases in spectral power) during walking appears to be related to flow of information related to sensory feedback (14, 26, 27, 46, 75-77). In our study we observed strong theta synchronizations around heel strike and push-off within the gait cycle [Figure 15,Figure 16]. Greater exertion during walking increases theta power (e.g., walking uphill or at faster speeds) (22, 78), and visual perturbations have shown to increase theta power around and during the perturbation (79, 80). At faster walking speeds, there is greater negative work done with leading foot contact and greater push-off work with the trailing limb (81, 82). The increased force and speed of lower limb motion to propel and control gait could result in greater feedback from stretch and force receptors in lower limb afferents (83, 84). Additionally, older adults have reduced visual acuity, reduced somatosensation, and worsened proprioception, affecting sensory information crucial for maintaining a stable gait (11). Older adults walking at faster speeds may increase attention to afferent feedback to maintain a stable torso and body position on the treadmill. We found that all subjects have increased theta power with faster speeds in left sensorimotor [Figure 8], right sensorimotor [Figure 9], mid cingulate [Figure 12], and left supplementary motor [Figure 13]. Additionally, we observed an interaction between speed and group in left posterior parietal [Figure 10], with older adults having greater slope terms than younger adults. Together, these findings suggest that older and younger adults modulate theta power concurrently with changes in speed for sensory monitoring. The difference observed in left posterior parietal between younger and older adults may be attributed to reduced sensation and weakened muscles that occur with aging. Older adults may have reduced control over gait using afferent feedback and would compensate with increased attention to stability. Generally, theta power correlates with faster walking speeds throughout the brain for younger and older adults.

We present spectral data from younger adults (top), older high functioning adults (center), and older low functioning adults (bottom). Dark colors represent significant differences between the two conditions. **Summary:** Older adults exhibit greater changes in event-related power spectral density than younger adults in theta, alpha, and beta. Older low functioning adults show modulations in theta, alpha, and beta during stance phase, and as speed increases the power modulations within stance phase decrease. Older high functioning adults show decreases in beta and alpha power during stance phase with increasing speed. Younger adults did not show changes in power across speeds.

There were reduced beta and alpha power at faster walking speeds for all participants, but there were no observed differences between younger and older adults [Figure 7]. Beta and alpha power desynchronization (i.e. decreases in spectral power) in sensorimotor processing is typically a sign of increased neural computation (24, 25). Alpha and beta power changes are linked to states of gait adjustment (26, 28, 85), visual attention (86-88) and somatosensation (75-77, 89, 90) in the sensorimotor and posterior parietal cortices. Alpha and beta power show responses after brief delays with visual and balance perturbations in temporal, supplementary motor, and posterior parietal cortices (80, 91-94). We found decreases in beta power with faster speeds in all brain areas observed, after controlling for group effects [Figure 8-Figure 14]. Additionally, we found decreases in alpha power with faster speeds in left sensorimotor [Figure 8], right posterior parietal [Figure 11], and right cuneus [Figure 14]. These findings in older adults and younger adults support previous observations of alpha and beta power desynchronizations in younger adults walking at faster speeds. Altogether, the findings indicate that many cortical areas are involved in monitoring and adjusting gait with greater involvement at faster speeds compared to slower speeds.

The differences between older and younger adults in EEG spectral power changes with speed may be related to physical capabilities or sensory processing during walking. Increases in mediolateral and anteroposterior variance are usually a sign of decreased walking stability (63, 64, 82). Older adults often have less muscle mass and strength compared to younger adults. Reduced muscular power can make it more difficult to stabilize the trunk in and generate propulsion at faster walking speeds Figure 5.A & B] (81-83). When healthy young subjects walk on increasingly uneven terrain, they become more unstable, increase theta cingulate power, and decrease alpha and beta sensorimotor and posterior parietal power (95). Older adults can also have a reduction in sensory acuity as they age. Evidence suggests visual, vestibular, and proprioceptive acuity are all diminished with aging (96, 97). It is not clear from our data how the interplay of muscular function and sensory processing contributed separately to the changes in electrical brain activity at faster speeds. However, the older adults had more variable pelvic movements than the younger adults at all speeds, suggesting a reduced automaticity to the gait. Older adults also exhibited greater increases in left posterior parietal theta power with faster speeds compared to younger adults. Also, older adults had greater theta power across all speeds than younger adults in precuneus. The older adults may need to rely more on cognitive processing and error monitoring during gait, especially at faster speeds, than the younger adults. These factors likely contributed to the electrocortical results from our study.

Our study had limitations which affect the interpretation of the results. Older adults exhibited distinct periodic power spectral density signatures compared to younger adults. Periodic and aperiodic components of power spectral density can vary based on age (61, 72), cognitive impairment (98, 99), and psychopathology (100-102). For example, peak EEG alpha power has shown to slow with increasing age and vary across brain regions (103). We found a less pronounced alpha and beta peak in older than in younger adults in sensorimotor and posterior parietal (i.e., a plateauing effect) [Figure 8-Figure 11.C]. To a lesser degree, we found the plateauing of older adults’ power spectral density in the mid cingulate, right cuneus, and left supplementary motor. The plateauing effects may be related to a wider variation in older adult aperiodic power spectral density due to age-related and unknown changes in EEG power that are not controlled for by FOOOF. When averaging occurs at the group level this may lead to a flattening effect around the alpha and beta range [Figure 21]. This may contribute to a wider range in power spectral density outcomes and mask or multiply underlying significance. Our study design did not control for the subject’s position on the treadmill, specifically in the fore-aft direction. The increase in variability may confound EEG theta results due to subjects approaching the treadmill boundaries, especially in older adults with poor gait control. In addition, older and younger adults can have different gait characteristics when walking on a treadmill compared to overground. Compared to overground, treadmill walking can reduce the variability of step duration and preferred walking speeds, especially for older adults (34, 104, 105). Treadmill walking can change muscle activation and EMG-EEG spectral coherence (76, 77, 106). This may be due to the constraint of the treadmill surface and differences in optic flow compared to overground walking (107, 108). Changes in environment and optic flow could affect EEG beta and theta power and degrees of differences between younger and older adult EEG power.

Periodic component of the FOOOF modeled power spectral density plot in left sensorimotor. Younger adults are displayed in the left column, the older high functioning adults (OHFA) in the center, and older low functioning adults (OLFA) in the right. From top to bottom are the power spectral densities at each speed (0.25, 0.50, 0.75, and 1.0 m/s). Thick color lines represent mean periodic activity, and each gray line represents an individual subject’s periodic power spectral density plot.

Our findings suggest several areas for future study. We used the Short Physical Performance Battery to assess functional mobility in older adults. This metric has been shown to be reliable and accurate for assessing health outcomes later in life. However, the observed averaging effect on the power spectral density plots and minor differences observed between lower functioning older adults and higher functioning older adults theta and beta power changes may indicate a limitation of the Short Physical Performance Battery. A study that clusters subjects based on their power spectral density then observes clinical measures that are highly significant within populations may generate a more reliable metric for mobility impairment. We observed some older adults needing reminders to maintain a safe position on the treadmill (e.g., walking too close to the back of the treadmill belt and needing to be reminded to move forward). We averaged kinematic measures of gait over each trial. There are variabilities between gait cycles that are not captured by coefficient of variations (34, 82). Given the known variations between gait cycles in EEG power, an analysis that correlates kinematics and EEG power outcomes within each gait cycle could further discriminate brain activity related to stable and unstable walking. Additionally, using a metric that captures local dynamic stability would better quantify the stability of the participants during walking. For example, the Floquet multiplier (35) can capture the consistency cyclic activity of the pelvis across gait cycles. Using a treadmill may have influenced our findings. The constraints of the treadmill surface may lead to gait behavior unlike walking on overground at different speeds (34, 106). An experiment measuring EEG of older and younger adults during overground walking at different speeds could identify electrocortical dynamics more consistent with everyday gait challenges. Alternatively, an experiment where specific positions on the treadmill are designated for walking could also identify how constraints to gait behavior are represented in the EEG and kinematic changes observed in our study. Further, future studies could also observe variations in aperiodic and periodic power spectral density components while older and younger adults perform other motor tasks (e.g., ping pong or tossing a ball to a partner). These experiments require intricate coordination of sensorimotor systems and may recruit different brain areas and elucidate higher functions of the network observed in this paper.

In conclusion, the main findings of the study show that older adults and younger adults reduce beta spectral power in many cortical areas at faster walking speeds compared to slower walking speeds. The older adults and younger adults also increase theta spectral power in sensorimotor, supplementary motor, and cingulate cortices at faster walking speeds compared to slower walking speeds. Older adults differentially altered theta power compared to younger adults in left posterior parietal, mid cingulate, and right cuneus cortices. These findings add to our knowledge about aging effects on walking control, such that there are more areas than prefrontal cortex that show alterations with aging.

Source localization of independent component analysis on EEG signals. Top row images contain each subject’s dipole in each cluster overlaid on the MNI template. From left to right are the sagittal, coronal, and top views. Clusters are colorized for visual distinction. The bottom row of images contains the centroid of each cluster. Table 3 contains the chosen cluster anatomy, the color of the anatomy depicted in the MNI images, the number of independent components in each cluster, the MNI coordinates of the centroid of the cluster, the atlas determined through the most frequent anatomy among all dipoles in the cluster, and the atlas determined by the location of the centroid of the cluster. **Brain areas:** blue is left sensorimotor, yellow is right sensorimotor, olive is left posterior parietal, orange is right posterior parietal, green is mid cingulate, light blue is right cuneus, and pink is left supplementary motor.

ACKNOWLEDGEMENTS

We would like to thank HNL lab members for their help with data collection: Ryland Swearinger, Ryan J Downey, Quinlan Degnan, Sydney Irwin, and we thank the HNL members for their feedback and intellectual support to improve the paper and the methods for EEG analysis. We would also like to thank our study coordinators for their devotion to project goals even during Covid-19 pandemic.

GRANTS

This study was supported by the National Institute of Health (U01AG061389) for authors JSS, CL, EMP, MT, AR, NR, JH, CJH, DJC, RDS, TMM, YCA, and DPF. National Institute of Health grants F32AG072808 and T32AG062728 supported author EMP. American Heart Association Fellowship (23POST1011634, doi.org/10.58275/AHA.23POST1011634.pc.gr.161292) partially supported author CL. DPF was also supported by National Institutes of Health (R01NS104772). The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.

Footnotes

DATA AND CODE AVAILABILITY

GitHub Repository: https://github.com/JacobSal/MindInMotion_YoungerOlderAdults_BrainSpeedChanges.git

Data for younger adults are available via OpenNeuro: https://openneuro.org/datasets/ds004625/versions/1.0.2 . Older adult data is available upon request but will soon be available through OpenNeuro.

DISCLOSURES

The authors have no conflicts of interest or disclosures to state with the research conducted in this study.

DISCLAIMERS

The authors have no disclaimers to state regarding the research conducted in this study.

REFERENCES

1.Cesari M, Kritchevsky SB, Penninx BWHJ, Nicklas BJ, Simonsick EM, Newman AB, Tylavsky FA, Brach JS, Satterfield S, Bauer DC, Visser M, Rubin SM, Harris TB, Pahor M. Prognostic Value of Usual Gait Speed in Well-Functioning Older People—Results from the Health, Aging and Body Composition Study. J Am Geriatr Soc 53: 1675–1680, 2005. doi: 10.1111/j.1532-5415.2005.53501.x. [DOI] [PubMed] [Google Scholar]
2.De Rekeneire N, Visser M, Peila R, Nevitt MC, Cauley JA, Tylavsky FA, Simonsick EM, Harris TB. Is a Fall Just a Fall: Correlates of Falling in Healthy Older Persons. The Health, Aging and Body Composition Study. J Am Geriatr Soc 51: 841–846, 2003. doi: 10.1046/j.1365-2389.2003.51267.x. [DOI] [PubMed] [Google Scholar]
3.Pahor M, Guralnik JM, Ambrosius WT, Blair S, Bonds DE, Church TS, Espeland MA, Fielding RA, Gill TM, Groessl EJ, King AC, Kritchevsky SB, Manini TM, McDermott MM, Miller ME, Newman AB, Rejeski WJ, Sink KM, Williamson JD, for the LIFE study investigators. Effect of Structured Physical Activity on Prevention of Major Mobility Disability in Older Adults: The LIFE Study Randomized Clinical Trial. JAMA 311: 2387–2396, 2014. doi: 10.1001/jama.2014.5616. [DOI] [PMC free article] [PubMed] [Google Scholar]
4.Studenski S, Perera S, Patel K, Rosano C, Faulkner K, Inzitari M, Brach J, Chandler J, Cawthon P, Connor EB, Nevitt M, Visser M, Kritchevsky S, Badinelli S, Harris T, Newman AB, Cauley J, Ferrucci L, Guralnik J. Gait Speed and Survival in Older Adults. JAMA 305: 50–58, 2011. doi: 10.1001/jama.2010.1923. [DOI] [PMC free article] [PubMed] [Google Scholar]
5.Allen JL, Franz JR. The motor repertoire of older adult fallers may constrain their response to balance perturbations. J Neurophysiol 120: 2368–2378, 2018. doi: 10.1152/jn.00302.2018. [DOI] [PMC free article] [PubMed] [Google Scholar]
6.Barak Y, Wagenaar RC, Holt KG. Gait Characteristics of Elderly People With a History of Falls: A Dynamic Approach. Phys Ther 86: 1501–1510, 2006. doi: 10.2522/ptj.20050387. [DOI] [PubMed] [Google Scholar]
7.Boyer KA, Johnson RT, Banks JJ, Jewell C, Hafer JF. Systematic review and meta-analysis of gait mechanics in young and older adults. Exp Gerontol 95: 63–70, 2017. doi: 10.1016/j.exger.2017.05.005. [DOI] [PubMed] [Google Scholar]
8.Boyer KA, Hayes KL, Umberger BR, Adamczyk PG, Bean JF, Brach JS, Clark BC, Clark DJ, Ferrucci L, Finley J, Franz JR, Golightly YM, Hortobágyi T, Hunter S, Narici M, Nicklas B, Roberts T, Sawicki G, Simonsick E, Kent JA. Age-related changes in gait biomechanics and their impact on the metabolic cost of walking: Report from a National Institute on Aging workshop. Exp Gerontol 173: 112102, 2023. doi: 10.1016/j.exger.2023.112102. [DOI] [PMC free article] [PubMed] [Google Scholar]
9.Ortega JD, Farley CT. Individual limb work does not explain the greater metabolic cost of walking in elderly adults. J Appl Physiol Bethesda Md 1985 102: 2266–2273, 2007. doi: 10.1152/japplphysiol.00583.2006. [DOI] [PubMed] [Google Scholar]
10.Lord SR, Delbaere K, Sturnieks DL. Chapter 10 - Aging. In: Handbook of Clinical Neurology, edited by Day BL, Lord SR. Elsevier, p. 157–171. [DOI] [PubMed] [Google Scholar]
11.Clark DJ. Automaticity of walking: functional significance, mechanisms, measurement and rehabilitation strategies. Front Hum Neurosci 9: 246, 2015. doi: 10.3389/fnhum.2015.00246. [DOI] [PMC free article] [PubMed] [Google Scholar]
12.Monaco V, Ghionzoli A, Micera S. Age-related modifications of muscle synergies and spinal cord activity during locomotion. J Neurophysiol 104: 2092–2102, 2010. doi: 10.1152/jn.00525.2009. [DOI] [PubMed] [Google Scholar]
13.Sato SD, Choi JT. Corticospinal drive is associated with temporal walking adaptation in both healthy young and older adults. Front Aging Neurosci 14, 2022. doi: 10.3389/fnagi.2022.920475. [DOI] [PMC free article] [PubMed] [Google Scholar]
14.Richer N, Bradford JC, Ferris DP. Mobile neuroimaging: What we have learned about the neural control of human walking, with an emphasis on EEG-based research. Neurosci Biobehav Rev 162: 105718, 2024. doi: 10.1016/j.neubiorev.2024.105718. [DOI] [PMC free article] [PubMed] [Google Scholar]
15.Grillner S. Biological pattern generation: the cellular and computational logic of networks in motion. Neuron 52: 751–766, 2006. doi: 10.1016/j.neuron.2006.11.008. [DOI] [PubMed] [Google Scholar]
16.Kuo AD. Stabilization of Lateral Motion in Passive Dynamic Walking. Int J Robot Res 18: 917–930, 1999. doi: 10.1177/02783649922066655. [DOI] [Google Scholar]
17.Beurskens R, Helmich I, Rein R, Bock O. Age-related changes in prefrontal activity during walking in dual-task situations: a fNIRS study. Int J Psychophysiol Off J Int Organ Psychophysiol 92: 122–128, 2014. doi: 10.1016/j.ijpsycho.2014.03.005. [DOI] [PubMed] [Google Scholar]
18.Hawkins KA, Fox EJ, Daly JJ, Rose DK, Christou EA, McGuirk TE, Otzel DM, Butera KA, Chatterjee SA, Clark DJ. Prefrontal over-activation during walking in people with mobility deficits: Interpretation and functional implications. Hum Mov Sci 59: 46–55, 2018. doi: 10.1016/j.humov.2018.03.010. [DOI] [PMC free article] [PubMed] [Google Scholar]
19.Hwang J, Liu C, Winesett SP, Chatterjee SA, Gruber AD, Swanson CW, Manini TM, Hass CJ, Seidler RD, Ferris DP, Roy A, Clark DJ. Prefrontal cortical activity during uneven terrain walking in younger and older adults. Front Aging Neurosci 16, 2024. doi: 10.3389/fnagi.2024.1389488. [DOI] [PMC free article] [PubMed] [Google Scholar]
20.Alcock L, Vitório R, Stuart S, Rochester L, Pantall A. Faster Walking Speeds Require Greater Activity from the Primary Motor Cortex in Older Adults Compared to Younger Adults. Sensors 23: 6921, 2023. doi: 10.3390/s23156921. [DOI] [PMC free article] [PubMed] [Google Scholar]
21.de Belli V, Orcioli-Silva D, Beretta VS, Vitório R, Zampier VC, Nóbrega-Sousa P, da Conceição NR, Gobbi LTB. Prefrontal Cortical Activity During Preferred and Fast Walking in Young and Older Adults: An fNIRS Study. Neuroscience 473: 81–89, 2021. doi: 10.1016/j.neuroscience.2021.08.019. [DOI] [PubMed] [Google Scholar]
22.Nordin AD, Hairston WD, Ferris DP. Faster gait speeds reduce alpha and beta EEG spectral power from human sensorimotor cortex. IEEE Trans Biomed Eng 67: 842–853, 2020. doi: 10.1109/TBME.2019.2921766. [DOI] [PMC free article] [PubMed] [Google Scholar]
23.Seeber M, Scherer R, Wagner J, Solis-Escalante T, Müller-Putz GR. EEG beta suppression and low gamma modulation are different elements of human upright walking. Front Hum Neurosci 8, 2014. doi: 10.3389/fnhum.2014.00485. [DOI] [PMC free article] [PubMed] [Google Scholar]
24.Neuper C, Pfurtscheller G. Event-related dynamics of cortical rhythms: frequency-specific features and functional correlates. Int J Psychophysiol 43: 41–58, 2001. doi: 10.1016/S0167-8760(01)00178-7. [DOI] [PubMed] [Google Scholar]
25.Pfurtscheller G. Functional brain imaging based on ERD/ERS. Vision Res 41: 1257–1260, 2001. doi: 10.1016/S0042-6989(00)00235-2. [DOI] [PubMed] [Google Scholar]
26.Jacobsen NA, Ferris DP. Electrocortical theta activity may reflect sensory prediction errors during adaptation to a gradual gait perturbation. PeerJ 12: e17451, 2024. doi: 10.7717/peerj.17451. [DOI] [PMC free article] [PubMed] [Google Scholar]
27.Sipp AR, Gwin JT, Makeig S, Ferris DP. Loss of balance during balance beam walking elicits a multifocal theta band electrocortical response. J Neurophysiol 110: 2050–2060, 2013. doi: 10.1152/jn.00744.2012. [DOI] [PMC free article] [PubMed] [Google Scholar]
28.Jacobsen NA, Ferris DP. Electrocortical activity correlated with locomotor adaptation during split-belt treadmill walking. J Physiol 601: 3921–3944, 2023. doi: 10.1113/JP284505. [DOI] [PMC free article] [PubMed] [Google Scholar]
29.Jacobsen NA, Ferris DP. Exploring Electrocortical Signatures of Gait Adaptation: Differential Neural Dynamics in Slow and Fast Gait Adapters. . [DOI] [PMC free article] [PubMed] [Google Scholar]
30.Clark DJ, Manini TM, Ferris DP, Hass CJ, Brumback BA, Cruz-Almeida Y, Pahor M, Reuter-Lorenz PA, Seidler RD. Multimodal Imaging of Brain Activity to Investigate Walking and Mobility Decline in Older Adults (Mind in Motion Study): Hypothesis, Theory, and Methods [Online]. Front Aging Neurosci 11, 2020. https://www.frontiersin.org/articles/10.3389/fnagi.2019.00358 [5 Mar. 2024]. [DOI] [PMC free article] [PubMed] [Google Scholar]
31.Drew T, Marigold DS. Taking the next step: cortical contributions to the control of locomotion. Curr Opin Neurobiol 33: 25–33, 2015. doi: 10.1016/j.conb.2015.01.011. [DOI] [PubMed] [Google Scholar]
32.Lau TM, Gwin JT, Ferris DP. Walking reduces sensorimotor network connectivity compared to standing. J Neuroengineering Rehabil 11: 14, 2014. doi: 10.1186/1743-0003-11-14. [DOI] [PMC free article] [PubMed] [Google Scholar]
33.Mirelman A, Shema S, Maidan I, Hausdorff JM. Chapter 7 - Gait. In: Handbook of Clinical Neurology, edited by Day BL, Lord SR. Elsevier, p. 119–134. [DOI] [PubMed] [Google Scholar]
34.Dingwell JB, Cusumano JP, Cavanagh PR, Sternad D. Local Dynamic Stability Versus Kinematic Variability of Continuous Overground and Treadmill Walking. J Biomech Eng 123: 27–32, 2000. doi: 10.1115/1.1336798. [DOI] [PubMed] [Google Scholar]
35.Dingwell JB, Kang HG. Differences Between Local and Orbital Dynamic Stability During Human Walking. J Biomech Eng 129: 586–593, 2006. doi: 10.1115/1.2746383. [DOI] [PubMed] [Google Scholar]
36.Guralnik JM, Simonsick EM, Ferrucci L, Glynn RJ, Berkman LF, Blazer DG, Scherr PA, Wallace RB. A Short Physical Performance Battery Assessing Lower Extremity Function: Association With Self-Reported Disability and Prediction of Mortality and Nursing Home Admission. J Gerontol 49: M85–M94, 1994. doi: 10.1093/geronj/49.2.M85. [DOI] [PubMed] [Google Scholar]
37.Maoz SLL, Stangl M, Topalovic U, Batista D, Hiller S, Aghajan ZM, Knowlton B, Stern J, Langevin J-P, Fried I, Eliashiv D, Suthana N. Dynamic neural representations of memory and space during human ambulatory navigation. Nat Commun 14: 6643, 2023. doi: 10.1038/s41467-023-42231-4. [DOI] [PMC free article] [PubMed] [Google Scholar]
38.Stangl M, Topalovic U, Inman CS, Hiller S, Villaroman D, Aghajan ZM, Christov-Moore L, Hasulak NR, Rao VR, Halpern CH, Eliashiv D, Fried I, Suthana N. Boundary-anchored neural mechanisms of location-encoding for self and others. Nature 589: 420–425, 2021. doi: 10.1038/s41586-020-03073-y. [DOI] [PMC free article] [PubMed] [Google Scholar]
39.Onton J, Westerfield M, Townsend J, Makeig S. Imaging human EEG dynamics using independent component analysis. Neurosci Biobehav Rev 30: 808–822, 2006. doi: 10.1016/j.neubiorev.2006.06.007. [DOI] [PubMed] [Google Scholar]
40.Makeig S, Debener S, Onton J, Delorme A. Mining event-related brain dynamics. Trends Cogn Sci 8: 204–210, 2004. doi: 10.1016/j.tics.2004.03.008. [DOI] [PubMed] [Google Scholar]
41.Nordin AD, Hairston WD, Ferris DP. Dual-electrode motion artifact cancellation for mobile electroencephalography. J Neural Eng 15: 056024, 2018. doi: 10.1088/1741-2552/aad7d7. [DOI] [PubMed] [Google Scholar]
42.Studnicki A, Downey RJ, Ferris DP. Characterizing and Removing Artifacts Using Dual-Layer EEG during Table Tennis. Sensors 22: 5867, 2022. doi: 10.3390/s22155867. [DOI] [PMC free article] [PubMed] [Google Scholar]
43.Downey RJ, Richer N, Gupta R, Liu C, Pliner EM, Roy A, Hwang J, Clark DJ, Hass CJ, Manini TM, Seidler RD, Ferris DP. Uneven terrain treadmill walking in younger and older adults. PLOS ONE 17: e0278646, 2022. doi: 10.1371/journal.pone.0278646. [DOI] [PMC free article] [PubMed] [Google Scholar]
44.Delorme A, Makeig S. EEGLAB: an open source toolbox for analysis of single-trial EEG dynamics including independent component analysis. J Neurosci Methods 134: 9–21, 2004. doi: 10.1016/j.jneumeth.2003.10.009. [DOI] [PubMed] [Google Scholar]
45.Oostenveld R, Fries P, Maris E, Schoffelen J-M. FieldTrip: Open source software for advanced analysis of MEG, EEG, and invasive electrophysiological data. Comput Intell Neurosci 2011: 156869, 2011. doi: 10.1155/2011/156869. [DOI] [PMC free article] [PubMed] [Google Scholar]
46.Liu C, Downey RJ, Salminen JS, Rojas SA, Richer N, Pliner EM, Hwang J, Cruz-Almeida Y, Manini TM, Hass CJ, Seidler RD, Clark DJ, Ferris DP. Electrical Brain Activity during Human Walking with Parametric Variations in Terrain Unevenness and Walking Speed. . [DOI] [PMC free article] [PubMed] [Google Scholar]
47.Liu C, Downey RJ, Mu Y, Richer N, Hwang J, Shah VA, Sato SD, Clark DJ, Hass CJ, Manini TM, Seidler RD, Ferris DP. Comparison of EEG Source Localization Using Simplified and Anatomically Accurate Head Models in Younger and Older Adults. IEEE Trans Neural Syst Rehabil Eng Publ IEEE Eng Med Biol Soc 31: 2591–2602, 2023. doi: 10.1109/TNSRE.2023.3281356. [DOI] [PMC free article] [PubMed] [Google Scholar]
48.Ludwig KA, Miriani RM, Langhals NB, Joseph MD, Anderson DJ, Kipke DR. Using a Common Average Reference to Improve Cortical Neuron Recordings From Microelectrode Arrays. J Neurophysiol 101: 1679, 2008. doi: 10.1152/jn.90989.2008. [DOI] [PMC free article] [PubMed] [Google Scholar]
49.Downey RJ, Ferris DP. iCanClean Removes Motion, Muscle, Eye, and Line-Noise Artifacts from Phantom EEG. Sensors 23: 8214, 2023. doi: 10.3390/s23198214. [DOI] [PMC free article] [PubMed] [Google Scholar]
50.Gonsisko CB, Ferris DP, Downey RJ. iCanClean Improves Independent Component Analysis of Mobile Brain Imaging with EEG. Sensors 23: 928, 2023. doi: 10.3390/s23020928. [DOI] [PMC free article] [PubMed] [Google Scholar]
51.Pion-Tonachini L, Kreutz-Delgado K, Makeig S. ICLabel: An automated electroencephalographic independent component classifier, dataset, and website. NeuroImage 198: 181–197, 2019. doi: 10.1016/j.neuroimage.2019.05.026. [DOI] [PMC free article] [PubMed] [Google Scholar]
52.Palmer JA, Kreutz-Delgado K, Makeig S. AMICA: An Adaptive Mixture of Independent Component Analyzers with Shared Components. . [Google Scholar]
53.Leske S, Dalal SS. Reducing power line noise in EEG and MEG data via spectrum interpolation. NeuroImage 189: 763–776, 2019. doi: 10.1016/j.neuroimage.2019.01.026. [DOI] [PMC free article] [PubMed] [Google Scholar]
54.Kim H, Luo J, Chu S, Cannard C, Hoffmann S, Miyakoshi M. ICA’s bug: How ghost ICs emerge from effective rank deficiency caused by EEG electrode interpolation and incorrect re-referencing. Front Signal Process 3, 2023. doi: 10.3389/frsip.2023.1064138. [DOI] [PMC free article] [PubMed] [Google Scholar]
55.Nielsen JD, Madsen KH, Puonti O, Siebner HR, Bauer C, Madsen CG, Saturnino GB, Thielscher A. Automatic skull segmentation from MR images for realistic volume conductor models of the head: Assessment of the state-of-the-art. NeuroImage 174: 587–598, 2018. doi: 10.1016/j.neuroimage.2018.03.001. [DOI] [PubMed] [Google Scholar]
56.Avants BB, Tustison NJ, Song G, Cook PA, Klein A, Gee JC. A reproducible evaluation of ANTs similarity metric performance in brain image registration. NeuroImage 54: 2033–2044, 2011. doi: 10.1016/j.neuroimage.2010.09.025. [DOI] [PMC free article] [PubMed] [Google Scholar]
57.Thammasan N, Miyakoshi M. Cross-Frequency Power-Power Coupling Analysis: A Useful Cross-Frequency Measure to Classify ICA-Decomposed EEG. Sensors 20: 7040, 2020. doi: 10.3390/s20247040. [DOI] [PMC free article] [PubMed] [Google Scholar]
58.Multivariate Distributions. In: Multivariate Observations. John Wiley & Sons, Ltd, p. 17–58. [Google Scholar]
59.Yong CY, Sudirman R, Mahmood NH, Chew KM. Motion Classification Using Proposed Principle Component Analysis Hybrid K-Means Clustering. Engineering 5: 25–30, 2013. doi: 10.4236/eng.2013.55B006. [DOI] [Google Scholar]
60.Rolls ET, Huang C-C, Lin C-P, Feng J, Joliot M. Automated anatomical labelling atlas 3. NeuroImage 206: 116189, 2020. doi: 10.1016/j.neuroimage.2019.116189. [DOI] [PubMed] [Google Scholar]
61.Donoghue T, Haller M, Peterson EJ, Varma P, Sebastian P, Gao R, Noto T, Lara AH, Wallis JD, Knight RT, Shestyuk A, Voytek B. Parameterizing neural power spectra into periodic and aperiodic components. Nat Neurosci 23: 1655–1665, 2020. doi: 10.1038/s41593-020-00744-x. [DOI] [PMC free article] [PubMed] [Google Scholar]
62.Seeber M, Scherer R, Wagner J, Solis-Escalante T, Müller-Putz GR. High and low gamma EEG oscillations in central sensorimotor areas are conversely modulated during the human gait cycle. NeuroImage 112: 318–326, 2015. doi: 10.1016/j.neuroimage.2015.03.045. [DOI] [PubMed] [Google Scholar]
63.Hamacher D, Singh NB, Van Dieën JH, Heller MO, Taylor WR. Kinematic measures for assessing gait stability in elderly individuals: a systematic review. J R Soc Interface 8: 1682–1698, 2011. doi: 10.1098/rsif.2011.0416. [DOI] [PMC free article] [PubMed] [Google Scholar]
64.Herssens N, Verbecque E, Hallemans A, Vereeck L, Van Rompaey V, Saeys W. Do spatiotemporal parameters and gait variability differ across the lifespan of healthy adults? A systematic review. Gait Posture 64: 181–190, 2018. doi: 10.1016/j.gaitpost.2018.06.012. [DOI] [PubMed] [Google Scholar]
65.Wilkinson RD, Mazzo MR, Feeney DF. Rethinking the Statistical Analysis of Neuromechanical Data. Exerc Sport Sci Rev 51: 43–50, 2023. doi: 10.1249/JES.0000000000000308. [DOI] [PubMed] [Google Scholar]
66.Benjamini Y, Hochberg Y. Controlling the False Discovery Rate: A Practical and Powerful Approach to Multiple Testing. J R Stat Soc Ser B Methodol 57: 289–300, 1995. doi: 10.1111/j.2517-6161.1995.tb02031.x. [DOI] [Google Scholar]
67.Cohen J. Statistical Power Analysis for the Behavioral Sciences. 2nd ed. New York: Routledge, 1988. [Google Scholar]
68.Selya AS, Rose JS, Dierker LC, Hedeker D, Mermelstein RJ. A Practical Guide to Calculating Cohen’s f2, a Measure of Local Effect Size, from PROC MIXED. Front Psychol 3: 111, 2012. doi: 10.3389/fpsyg.2012.00111. [DOI] [PMC free article] [PubMed] [Google Scholar]
69.Reuter-Lorenz PA, Cappell KA. Neurocognitive aging and the compensation hypothesis. Curr Dir Psychol Sci 17: 177–182, 2008. doi: 10.1111/j.1467-8721.2008.00570.x. [DOI] [Google Scholar]
70.Sato SD, Shah VA, Fettrow T, Hall KG, Tays GD, Cenko E, Roy A, Clark DJ, Ferris DP, Hass CJ, Manini TM, Seidler RD. Resting state brain network segregation is associated with walking speed and working memory in older adults. bioRxiv: 2024.05.07.592861, 2024. [DOI] [PubMed] [Google Scholar]
71.Gyurkovics M, Clements GM, Low KA, Fabiani M, Gratton G. Stimulus-Induced Changes in 1/f-like Background Activity in EEG. J Neurosci 42: 7144–7151, 2022. doi: 10.1523/JNEUROSCI.0414-22.2022. [DOI] [PMC free article] [PubMed] [Google Scholar]
72.Voytek B, Kramer MA, Case J, Lepage KQ, Tempesta ZR, Knight RT, Gazzaley A. Age-Related Changes in 1/f Neural Electrophysiological Noise. J Neurosci 35: 13257–13265, 2015. doi: 10.1523/JNEUROSCI.2332-14.2015. [DOI] [PMC free article] [PubMed] [Google Scholar]
73.Vlahou EL, Thurm F, Kolassa I-T, Schlee W. Resting-state slow wave power, healthy aging and cognitive performance. Sci Rep 4: 5101, 2014. doi: 10.1038/srep05101. [DOI] [PMC free article] [PubMed] [Google Scholar]
74.Garofalo S, Giovagnoli S, Orsoni M, Starita F, Benassi M. Interaction effect: Are you doing the right thing? PLOS ONE 17: e0271668, 2022. doi: 10.1371/journal.pone.0271668. [DOI] [PMC free article] [PubMed] [Google Scholar]
75.Jensen O, Spaak E, Zumer JM. Human Brain Oscillations: From Physiological Mechanisms to Analysis and Cognition. In: Magnetoencephalography: From Signals to Dynamic Cortical Networks, edited by Supek S, Aine CJ. Springer International Publishing, p. 471–517. [Google Scholar]
76.Roeder L, Boonstra TW, Smith SS, Kerr GK. Dynamics of corticospinal motor control during overground and treadmill walking in humans. J Neurophysiol 120: 1017–1031, 2018. doi: 10.1152/jn.00613.2017. [DOI] [PubMed] [Google Scholar]
77.Roeder L, Boonstra TW, Kerr GK. Corticomuscular control of walking in older people and people with Parkinson’s disease. Sci Rep 10: 2980, 2020. doi: 10.1038/s41598-020-59810-w. [DOI] [PMC free article] [PubMed] [Google Scholar]
78.Bradford JC, Lukos JR, Ferris DP. Electrocortical activity distinguishes between uphill and level walking in humans. J Neurophysiol 115: 958–966, 2016. doi: 10.1152/jn.00089.2015. [DOI] [PubMed] [Google Scholar]
79.Hua A, Wang G, Bai J, Hao Z, Yang Y, Luo X, Liu J, Meng J, Wang J. Rapid reconfiguration of cortical networks after repeated exposure to visual-vestibular conflicts. Sci Rep 14: 21943, 2024. doi: 10.1038/s41598-024-73111-6. [DOI] [PMC free article] [PubMed] [Google Scholar]
80.Peterson SM, Ferris DP. Differentiation in Theta and Beta Electrocortical Activity between Visual and Physical Perturbations to Walking and Standing Balance. eNeuro 5: ENEURO.0207-18.2018, 2018. doi: 10.1523/ENEURO.0207-18.2018. [DOI] [PMC free article] [PubMed] [Google Scholar]
81.Kuo AD. The six determinants of gait and the inverted pendulum analogy: A dynamic walking perspective. Hum Mov Sci 26: 617–656, 2007. doi: 10.1016/j.humov.2007.04.003. [DOI] [PubMed] [Google Scholar]
82.Kuo AD, Donelan JM. Dynamic principles of gait and their clinical implications. Phys Ther 90: 157–174, 2010. doi: 10.2522/ptj.20090125. [DOI] [PMC free article] [PubMed] [Google Scholar]
83.Doherty TJ. Invited Review: Aging and sarcopenia. J Appl Physiol 95: 1717–1727, 2003. doi: 10.1152/japplphysiol.00347.2003. [DOI] [PubMed] [Google Scholar]
84.Kirk EA, Hope KT, Sober SJ, Sauerbrei BA. An output-null signature of inertial load in motor cortex. Nat Commun 15: 1–20, 2024. doi: 10.1038/s41467-024-51750-7. [DOI] [PMC free article] [PubMed] [Google Scholar]
85.Bulea TC, Jonghyun Kim null, Damiano DL, Stanley CJ, Hyung-Soon Park null. User-driven control increases cortical activity during treadmill walking: an EEG study. Annu Int Conf IEEE Eng Med Biol Soc IEEE Eng Med Biol Soc Annu Int Conf 2014: 2111–2114, 2014. doi: 10.1109/EMBC.2014.6944033. [DOI] [PMC free article] [PubMed] [Google Scholar]
86.Buschman TJ, Miller EK. Top-Down Versus Bottom-Up Control of Attention in the Prefrontal and Posterior Parietal Cortices. Science 315: 1860–1862, 2007. doi: 10.1126/science.1138071. [DOI] [PubMed] [Google Scholar]
87.Gola M, Magnuski M, Szumska I, Wróbel A. EEG beta band activity is related to attention and attentional deficits in the visual performance of elderly subjects. Int J Psychophysiol 89: 334–341, 2013. doi: 10.1016/j.ijpsycho.2013.05.007. [DOI] [PubMed] [Google Scholar]
88.Wróbel A. Beta activity: a carrier for visual attention. Acta Neurobiol Exp (Warsz) 60: 247–260, 2000. doi: 10.55782/ane-2000-1344. [DOI] [PubMed] [Google Scholar]
89.Anderson KL, Ding M. Attentional modulation of the somatosensory mu rhythm. Neuroscience 180: 165–180, 2011. doi: 10.1016/j.neuroscience.2011.02.004. [DOI] [PubMed] [Google Scholar]
90.Lopez S, Bini F, Del Percio C, Marinozzi F, Celletti C, Suppa A, Ferri R, Staltari E, Camerota F, Babiloni C. Electroencephalographic sensorimotor rhythms are modulated in the acute phase following focal vibration in healthy subjects. Neuroscience 352: 236–248, 2017. doi: 10.1016/j.neuroscience.2017.03.015. [DOI] [PubMed] [Google Scholar]
91.Wang G, Yang Y, Wang J, Hao Z, Luo X, Liu J. Dynamic changes of brain networks during standing balance control under visual conflict. Front Neurosci 16, 2022. doi: 10.3389/fnins.2022.1003996. [DOI] [PMC free article] [PubMed] [Google Scholar]
92.Peterson SM, Ferris DP. Group-level cortical and muscular connectivity during perturbations to walking and standing balance. NeuroImage 198: 93–103, 2019. doi: 10.1016/j.neuroimage.2019.05.038. [DOI] [PMC free article] [PubMed] [Google Scholar]
93.Barollo F, Hassan M, Petersen H, Rigoni I, Ramon C, Gargiulo P, Fratini A. Cortical Pathways During Postural Control: New Insights From Functional EEG Source Connectivity. IEEE Trans Neural Syst Rehabil Eng 30: 72–84, 2022. doi: 10.1109/TNSRE.2022.3140888. [DOI] [PubMed] [Google Scholar]
94.Oliveira AS, Schlink BR, Hairston WD, König P, Ferris DP. Restricted vision increases sensorimotor cortex involvement in human walking. J Neurophysiol 118: 1943–1951, 2017. doi: 10.1152/jn.00926.2016. [DOI] [PMC free article] [PubMed] [Google Scholar]
95.Liu C, Downey RJ, Salminen JS, Arvelo Rojas S, Richer N, Pliner EM, Hwang J, Cruz-Almeida Y, Manini TM, Hass CJ, Seidler RD, Clark DJ, Ferris DP. Electrical brain activity during human walking with parametric variations in terrain unevenness and walking speed. Imaging Neurosci 2: 1–33, 2024. doi: 10.1162/imag_a_00097. [DOI] [PMC free article] [PubMed] [Google Scholar]
96.Sturnieks DL, St George R, Lord RS. Balance disorders in the elderly. Neurophysiol Clin Neurophysiol 38: 467–478, 2008. doi: 10.1016/j.neucli.2008.09.001. [DOI] [PubMed] [Google Scholar]
97.Xing L, Bao Y, Wang B, Shi M, Wei Y, Huang X, Dai Y, Shi H, Gai X, Luo Q, Yin Y, Qin D. Falls caused by balance disorders in the elderly with multiple systems involved: Pathogenic mechanisms and treatment strategies. Front Neurol 14: 1128092, 2023. doi: 10.3389/fneur.2023.1128092. [DOI] [PMC free article] [PubMed] [Google Scholar]
98.Benwell CSY, Davila-Pérez P, Fried PJ, Jones RN, Travison TG, Santarnecchi E, Pascual-Leone A, Shafi MM. EEG spectral power abnormalities and their relationship with cognitive dysfunction in patients with Alzheimer’s disease and type 2 diabetes. Neurobiol Aging 85: 83–95, 2020. doi: 10.1016/j.neurobiolaging.2019.10.004. [DOI] [PMC free article] [PubMed] [Google Scholar]
99.Kopčanová M, Tait L, Donoghue T, Stothart G, Smith L, Flores-Sandoval AA, Davila-Perez P, Buss S, Shafi MM, Pascual-Leone A, Fried PJ, Benwell CSY. Resting-state EEG signatures of Alzheimer’s disease are driven by periodic but not aperiodic changes. Neurobiol Dis 190: 106380, 2024. doi: 10.1016/j.nbd.2023.106380. [DOI] [PubMed] [Google Scholar]
100.Karalunas SL, Ostlund BD, Alperin BR, Figuracion M, Gustafsson HC, Deming EM, Foti D, Antovich D, Dude J, Nigg J, Sullivan E. Electroencephalogram aperiodic power spectral slope can be reliably measured and predicts ADHD risk in early development. Dev Psychobiol 64: e22228, 2022. doi: 10.1002/dev.22228. [DOI] [PMC free article] [PubMed] [Google Scholar]
101.Peterson EJ, Rosen BQ, Belger A, Voytek B, Campbell AM. Aperiodic Neural Activity is a Better Predictor of Schizophrenia than Neural Oscillations. Clin EEG Neurosci 54: 434–445, 2023. doi: 10.1177/15500594231165589. [DOI] [PMC free article] [PubMed] [Google Scholar]
102.Robertson MM, Furlong S, Voytek B, Donoghue T, Boettiger CA, Sheridan MA. EEG power spectral slope differs by ADHD status and stimulant medication exposure in early childhood. J Neurophysiol 122: 2427–2437, 2019. doi: 10.1152/jn.00388.2019. [DOI] [PMC free article] [PubMed] [Google Scholar]
103.Park J, Ho RLM, Wang W, Nguyen VQ, Coombes SA. The effect of age on alpha rhythms in the human brain derived from source localized resting-state electroencephalography. NeuroImage 292: 120614, 2024. doi: 10.1016/j.neuroimage.2024.120614. [DOI] [PubMed] [Google Scholar]
104.Hollman JH, McDade EM, Petersen RC. Normative spatiotemporal gait parameters in older adults. Gait Posture 34: 111–118, 2011. doi: 10.1016/j.gaitpost.2011.03.024. [DOI] [PMC free article] [PubMed] [Google Scholar]
105.Schmitt AC, Baudendistel ST, Lipat AL, White TA, Raffegeau TE, Hass CJ. Walking indoors, outdoors, and on a treadmill: Gait differences in healthy young and older adults. Gait Posture 90: 468–474, 2021. doi: 10.1016/j.gaitpost.2021.09.197. [DOI] [PubMed] [Google Scholar]
106.Lee SJ, Hidler J. Biomechanics of overground vs. treadmill walking in healthy individuals. J Appl Physiol 104: 747–755, 2008. doi: 10.1152/japplphysiol.01380.2006. [DOI] [PubMed] [Google Scholar]
107.Salinas MM, Wilken JM, Dingwell JB. How humans use visual optic flow to regulate stepping during walking. Gait Posture 57: 15–20, 2017. doi: 10.1016/j.gaitpost.2017.05.002. [DOI] [PubMed] [Google Scholar]
108.Warren WH, Kay BA, Zosh WD, Duchon AP, Sahuc S. Optic flow is used to control human walking. Nat Neurosci 4: 213–216, 2001. doi: 10.1038/84054. [DOI] [PubMed] [Google Scholar]

[R1] 1.Cesari M, Kritchevsky SB, Penninx BWHJ, Nicklas BJ, Simonsick EM, Newman AB, Tylavsky FA, Brach JS, Satterfield S, Bauer DC, Visser M, Rubin SM, Harris TB, Pahor M. Prognostic Value of Usual Gait Speed in Well-Functioning Older People—Results from the Health, Aging and Body Composition Study. J Am Geriatr Soc 53: 1675–1680, 2005. doi: 10.1111/j.1532-5415.2005.53501.x. [DOI] [PubMed] [Google Scholar]

[R2] 2.De Rekeneire N, Visser M, Peila R, Nevitt MC, Cauley JA, Tylavsky FA, Simonsick EM, Harris TB. Is a Fall Just a Fall: Correlates of Falling in Healthy Older Persons. The Health, Aging and Body Composition Study. J Am Geriatr Soc 51: 841–846, 2003. doi: 10.1046/j.1365-2389.2003.51267.x. [DOI] [PubMed] [Google Scholar]

[R3] 3.Pahor M, Guralnik JM, Ambrosius WT, Blair S, Bonds DE, Church TS, Espeland MA, Fielding RA, Gill TM, Groessl EJ, King AC, Kritchevsky SB, Manini TM, McDermott MM, Miller ME, Newman AB, Rejeski WJ, Sink KM, Williamson JD, for the LIFE study investigators. Effect of Structured Physical Activity on Prevention of Major Mobility Disability in Older Adults: The LIFE Study Randomized Clinical Trial. JAMA 311: 2387–2396, 2014. doi: 10.1001/jama.2014.5616. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R4] 4.Studenski S, Perera S, Patel K, Rosano C, Faulkner K, Inzitari M, Brach J, Chandler J, Cawthon P, Connor EB, Nevitt M, Visser M, Kritchevsky S, Badinelli S, Harris T, Newman AB, Cauley J, Ferrucci L, Guralnik J. Gait Speed and Survival in Older Adults. JAMA 305: 50–58, 2011. doi: 10.1001/jama.2010.1923. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R5] 5.Allen JL, Franz JR. The motor repertoire of older adult fallers may constrain their response to balance perturbations. J Neurophysiol 120: 2368–2378, 2018. doi: 10.1152/jn.00302.2018. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R6] 6.Barak Y, Wagenaar RC, Holt KG. Gait Characteristics of Elderly People With a History of Falls: A Dynamic Approach. Phys Ther 86: 1501–1510, 2006. doi: 10.2522/ptj.20050387. [DOI] [PubMed] [Google Scholar]

[R7] 7.Boyer KA, Johnson RT, Banks JJ, Jewell C, Hafer JF. Systematic review and meta-analysis of gait mechanics in young and older adults. Exp Gerontol 95: 63–70, 2017. doi: 10.1016/j.exger.2017.05.005. [DOI] [PubMed] [Google Scholar]

[R8] 8.Boyer KA, Hayes KL, Umberger BR, Adamczyk PG, Bean JF, Brach JS, Clark BC, Clark DJ, Ferrucci L, Finley J, Franz JR, Golightly YM, Hortobágyi T, Hunter S, Narici M, Nicklas B, Roberts T, Sawicki G, Simonsick E, Kent JA. Age-related changes in gait biomechanics and their impact on the metabolic cost of walking: Report from a National Institute on Aging workshop. Exp Gerontol 173: 112102, 2023. doi: 10.1016/j.exger.2023.112102. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R9] 9.Ortega JD, Farley CT. Individual limb work does not explain the greater metabolic cost of walking in elderly adults. J Appl Physiol Bethesda Md 1985 102: 2266–2273, 2007. doi: 10.1152/japplphysiol.00583.2006. [DOI] [PubMed] [Google Scholar]

[R10] 10.Lord SR, Delbaere K, Sturnieks DL. Chapter 10 - Aging. In: Handbook of Clinical Neurology, edited by Day BL, Lord SR. Elsevier, p. 157–171. [DOI] [PubMed] [Google Scholar]

[R11] 11.Clark DJ. Automaticity of walking: functional significance, mechanisms, measurement and rehabilitation strategies. Front Hum Neurosci 9: 246, 2015. doi: 10.3389/fnhum.2015.00246. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R12] 12.Monaco V, Ghionzoli A, Micera S. Age-related modifications of muscle synergies and spinal cord activity during locomotion. J Neurophysiol 104: 2092–2102, 2010. doi: 10.1152/jn.00525.2009. [DOI] [PubMed] [Google Scholar]

[R13] 13.Sato SD, Choi JT. Corticospinal drive is associated with temporal walking adaptation in both healthy young and older adults. Front Aging Neurosci 14, 2022. doi: 10.3389/fnagi.2022.920475. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R14] 14.Richer N, Bradford JC, Ferris DP. Mobile neuroimaging: What we have learned about the neural control of human walking, with an emphasis on EEG-based research. Neurosci Biobehav Rev 162: 105718, 2024. doi: 10.1016/j.neubiorev.2024.105718. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R15] 15.Grillner S. Biological pattern generation: the cellular and computational logic of networks in motion. Neuron 52: 751–766, 2006. doi: 10.1016/j.neuron.2006.11.008. [DOI] [PubMed] [Google Scholar]

[R16] 16.Kuo AD. Stabilization of Lateral Motion in Passive Dynamic Walking. Int J Robot Res 18: 917–930, 1999. doi: 10.1177/02783649922066655. [DOI] [Google Scholar]

[R17] 17.Beurskens R, Helmich I, Rein R, Bock O. Age-related changes in prefrontal activity during walking in dual-task situations: a fNIRS study. Int J Psychophysiol Off J Int Organ Psychophysiol 92: 122–128, 2014. doi: 10.1016/j.ijpsycho.2014.03.005. [DOI] [PubMed] [Google Scholar]

[R18] 18.Hawkins KA, Fox EJ, Daly JJ, Rose DK, Christou EA, McGuirk TE, Otzel DM, Butera KA, Chatterjee SA, Clark DJ. Prefrontal over-activation during walking in people with mobility deficits: Interpretation and functional implications. Hum Mov Sci 59: 46–55, 2018. doi: 10.1016/j.humov.2018.03.010. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R19] 19.Hwang J, Liu C, Winesett SP, Chatterjee SA, Gruber AD, Swanson CW, Manini TM, Hass CJ, Seidler RD, Ferris DP, Roy A, Clark DJ. Prefrontal cortical activity during uneven terrain walking in younger and older adults. Front Aging Neurosci 16, 2024. doi: 10.3389/fnagi.2024.1389488. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R20] 20.Alcock L, Vitório R, Stuart S, Rochester L, Pantall A. Faster Walking Speeds Require Greater Activity from the Primary Motor Cortex in Older Adults Compared to Younger Adults. Sensors 23: 6921, 2023. doi: 10.3390/s23156921. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R21] 21.de Belli V, Orcioli-Silva D, Beretta VS, Vitório R, Zampier VC, Nóbrega-Sousa P, da Conceição NR, Gobbi LTB. Prefrontal Cortical Activity During Preferred and Fast Walking in Young and Older Adults: An fNIRS Study. Neuroscience 473: 81–89, 2021. doi: 10.1016/j.neuroscience.2021.08.019. [DOI] [PubMed] [Google Scholar]

[R22] 22.Nordin AD, Hairston WD, Ferris DP. Faster gait speeds reduce alpha and beta EEG spectral power from human sensorimotor cortex. IEEE Trans Biomed Eng 67: 842–853, 2020. doi: 10.1109/TBME.2019.2921766. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R23] 23.Seeber M, Scherer R, Wagner J, Solis-Escalante T, Müller-Putz GR. EEG beta suppression and low gamma modulation are different elements of human upright walking. Front Hum Neurosci 8, 2014. doi: 10.3389/fnhum.2014.00485. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R24] 24.Neuper C, Pfurtscheller G. Event-related dynamics of cortical rhythms: frequency-specific features and functional correlates. Int J Psychophysiol 43: 41–58, 2001. doi: 10.1016/S0167-8760(01)00178-7. [DOI] [PubMed] [Google Scholar]

[R25] 25.Pfurtscheller G. Functional brain imaging based on ERD/ERS. Vision Res 41: 1257–1260, 2001. doi: 10.1016/S0042-6989(00)00235-2. [DOI] [PubMed] [Google Scholar]

[R26] 26.Jacobsen NA, Ferris DP. Electrocortical theta activity may reflect sensory prediction errors during adaptation to a gradual gait perturbation. PeerJ 12: e17451, 2024. doi: 10.7717/peerj.17451. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R27] 27.Sipp AR, Gwin JT, Makeig S, Ferris DP. Loss of balance during balance beam walking elicits a multifocal theta band electrocortical response. J Neurophysiol 110: 2050–2060, 2013. doi: 10.1152/jn.00744.2012. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R28] 28.Jacobsen NA, Ferris DP. Electrocortical activity correlated with locomotor adaptation during split-belt treadmill walking. J Physiol 601: 3921–3944, 2023. doi: 10.1113/JP284505. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R29] 29.Jacobsen NA, Ferris DP. Exploring Electrocortical Signatures of Gait Adaptation: Differential Neural Dynamics in Slow and Fast Gait Adapters. . [DOI] [PMC free article] [PubMed] [Google Scholar]

[R30] 30.Clark DJ, Manini TM, Ferris DP, Hass CJ, Brumback BA, Cruz-Almeida Y, Pahor M, Reuter-Lorenz PA, Seidler RD. Multimodal Imaging of Brain Activity to Investigate Walking and Mobility Decline in Older Adults (Mind in Motion Study): Hypothesis, Theory, and Methods [Online]. Front Aging Neurosci 11, 2020. https://www.frontiersin.org/articles/10.3389/fnagi.2019.00358 [5 Mar. 2024]. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R31] 31.Drew T, Marigold DS. Taking the next step: cortical contributions to the control of locomotion. Curr Opin Neurobiol 33: 25–33, 2015. doi: 10.1016/j.conb.2015.01.011. [DOI] [PubMed] [Google Scholar]

[R32] 32.Lau TM, Gwin JT, Ferris DP. Walking reduces sensorimotor network connectivity compared to standing. J Neuroengineering Rehabil 11: 14, 2014. doi: 10.1186/1743-0003-11-14. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R33] 33.Mirelman A, Shema S, Maidan I, Hausdorff JM. Chapter 7 - Gait. In: Handbook of Clinical Neurology, edited by Day BL, Lord SR. Elsevier, p. 119–134. [DOI] [PubMed] [Google Scholar]

[R34] 34.Dingwell JB, Cusumano JP, Cavanagh PR, Sternad D. Local Dynamic Stability Versus Kinematic Variability of Continuous Overground and Treadmill Walking. J Biomech Eng 123: 27–32, 2000. doi: 10.1115/1.1336798. [DOI] [PubMed] [Google Scholar]

[R35] 35.Dingwell JB, Kang HG. Differences Between Local and Orbital Dynamic Stability During Human Walking. J Biomech Eng 129: 586–593, 2006. doi: 10.1115/1.2746383. [DOI] [PubMed] [Google Scholar]

[R36] 36.Guralnik JM, Simonsick EM, Ferrucci L, Glynn RJ, Berkman LF, Blazer DG, Scherr PA, Wallace RB. A Short Physical Performance Battery Assessing Lower Extremity Function: Association With Self-Reported Disability and Prediction of Mortality and Nursing Home Admission. J Gerontol 49: M85–M94, 1994. doi: 10.1093/geronj/49.2.M85. [DOI] [PubMed] [Google Scholar]

[R37] 37.Maoz SLL, Stangl M, Topalovic U, Batista D, Hiller S, Aghajan ZM, Knowlton B, Stern J, Langevin J-P, Fried I, Eliashiv D, Suthana N. Dynamic neural representations of memory and space during human ambulatory navigation. Nat Commun 14: 6643, 2023. doi: 10.1038/s41467-023-42231-4. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R38] 38.Stangl M, Topalovic U, Inman CS, Hiller S, Villaroman D, Aghajan ZM, Christov-Moore L, Hasulak NR, Rao VR, Halpern CH, Eliashiv D, Fried I, Suthana N. Boundary-anchored neural mechanisms of location-encoding for self and others. Nature 589: 420–425, 2021. doi: 10.1038/s41586-020-03073-y. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R39] 39.Onton J, Westerfield M, Townsend J, Makeig S. Imaging human EEG dynamics using independent component analysis. Neurosci Biobehav Rev 30: 808–822, 2006. doi: 10.1016/j.neubiorev.2006.06.007. [DOI] [PubMed] [Google Scholar]

[R40] 40.Makeig S, Debener S, Onton J, Delorme A. Mining event-related brain dynamics. Trends Cogn Sci 8: 204–210, 2004. doi: 10.1016/j.tics.2004.03.008. [DOI] [PubMed] [Google Scholar]

[R41] 41.Nordin AD, Hairston WD, Ferris DP. Dual-electrode motion artifact cancellation for mobile electroencephalography. J Neural Eng 15: 056024, 2018. doi: 10.1088/1741-2552/aad7d7. [DOI] [PubMed] [Google Scholar]

[R42] 42.Studnicki A, Downey RJ, Ferris DP. Characterizing and Removing Artifacts Using Dual-Layer EEG during Table Tennis. Sensors 22: 5867, 2022. doi: 10.3390/s22155867. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R43] 43.Downey RJ, Richer N, Gupta R, Liu C, Pliner EM, Roy A, Hwang J, Clark DJ, Hass CJ, Manini TM, Seidler RD, Ferris DP. Uneven terrain treadmill walking in younger and older adults. PLOS ONE 17: e0278646, 2022. doi: 10.1371/journal.pone.0278646. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R44] 44.Delorme A, Makeig S. EEGLAB: an open source toolbox for analysis of single-trial EEG dynamics including independent component analysis. J Neurosci Methods 134: 9–21, 2004. doi: 10.1016/j.jneumeth.2003.10.009. [DOI] [PubMed] [Google Scholar]

[R45] 45.Oostenveld R, Fries P, Maris E, Schoffelen J-M. FieldTrip: Open source software for advanced analysis of MEG, EEG, and invasive electrophysiological data. Comput Intell Neurosci 2011: 156869, 2011. doi: 10.1155/2011/156869. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R46] 46.Liu C, Downey RJ, Salminen JS, Rojas SA, Richer N, Pliner EM, Hwang J, Cruz-Almeida Y, Manini TM, Hass CJ, Seidler RD, Clark DJ, Ferris DP. Electrical Brain Activity during Human Walking with Parametric Variations in Terrain Unevenness and Walking Speed. . [DOI] [PMC free article] [PubMed] [Google Scholar]

[R47] 47.Liu C, Downey RJ, Mu Y, Richer N, Hwang J, Shah VA, Sato SD, Clark DJ, Hass CJ, Manini TM, Seidler RD, Ferris DP. Comparison of EEG Source Localization Using Simplified and Anatomically Accurate Head Models in Younger and Older Adults. IEEE Trans Neural Syst Rehabil Eng Publ IEEE Eng Med Biol Soc 31: 2591–2602, 2023. doi: 10.1109/TNSRE.2023.3281356. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R48] 48.Ludwig KA, Miriani RM, Langhals NB, Joseph MD, Anderson DJ, Kipke DR. Using a Common Average Reference to Improve Cortical Neuron Recordings From Microelectrode Arrays. J Neurophysiol 101: 1679, 2008. doi: 10.1152/jn.90989.2008. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R49] 49.Downey RJ, Ferris DP. iCanClean Removes Motion, Muscle, Eye, and Line-Noise Artifacts from Phantom EEG. Sensors 23: 8214, 2023. doi: 10.3390/s23198214. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R50] 50.Gonsisko CB, Ferris DP, Downey RJ. iCanClean Improves Independent Component Analysis of Mobile Brain Imaging with EEG. Sensors 23: 928, 2023. doi: 10.3390/s23020928. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R51] 51.Pion-Tonachini L, Kreutz-Delgado K, Makeig S. ICLabel: An automated electroencephalographic independent component classifier, dataset, and website. NeuroImage 198: 181–197, 2019. doi: 10.1016/j.neuroimage.2019.05.026. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R52] 52.Palmer JA, Kreutz-Delgado K, Makeig S. AMICA: An Adaptive Mixture of Independent Component Analyzers with Shared Components. . [Google Scholar]

[R53] 53.Leske S, Dalal SS. Reducing power line noise in EEG and MEG data via spectrum interpolation. NeuroImage 189: 763–776, 2019. doi: 10.1016/j.neuroimage.2019.01.026. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R54] 54.Kim H, Luo J, Chu S, Cannard C, Hoffmann S, Miyakoshi M. ICA’s bug: How ghost ICs emerge from effective rank deficiency caused by EEG electrode interpolation and incorrect re-referencing. Front Signal Process 3, 2023. doi: 10.3389/frsip.2023.1064138. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R55] 55.Nielsen JD, Madsen KH, Puonti O, Siebner HR, Bauer C, Madsen CG, Saturnino GB, Thielscher A. Automatic skull segmentation from MR images for realistic volume conductor models of the head: Assessment of the state-of-the-art. NeuroImage 174: 587–598, 2018. doi: 10.1016/j.neuroimage.2018.03.001. [DOI] [PubMed] [Google Scholar]

[R56] 56.Avants BB, Tustison NJ, Song G, Cook PA, Klein A, Gee JC. A reproducible evaluation of ANTs similarity metric performance in brain image registration. NeuroImage 54: 2033–2044, 2011. doi: 10.1016/j.neuroimage.2010.09.025. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R57] 57.Thammasan N, Miyakoshi M. Cross-Frequency Power-Power Coupling Analysis: A Useful Cross-Frequency Measure to Classify ICA-Decomposed EEG. Sensors 20: 7040, 2020. doi: 10.3390/s20247040. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R58] 58.Multivariate Distributions. In: Multivariate Observations. John Wiley & Sons, Ltd, p. 17–58. [Google Scholar]

[R59] 59.Yong CY, Sudirman R, Mahmood NH, Chew KM. Motion Classification Using Proposed Principle Component Analysis Hybrid K-Means Clustering. Engineering 5: 25–30, 2013. doi: 10.4236/eng.2013.55B006. [DOI] [Google Scholar]

[R60] 60.Rolls ET, Huang C-C, Lin C-P, Feng J, Joliot M. Automated anatomical labelling atlas 3. NeuroImage 206: 116189, 2020. doi: 10.1016/j.neuroimage.2019.116189. [DOI] [PubMed] [Google Scholar]

[R61] 61.Donoghue T, Haller M, Peterson EJ, Varma P, Sebastian P, Gao R, Noto T, Lara AH, Wallis JD, Knight RT, Shestyuk A, Voytek B. Parameterizing neural power spectra into periodic and aperiodic components. Nat Neurosci 23: 1655–1665, 2020. doi: 10.1038/s41593-020-00744-x. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R62] 62.Seeber M, Scherer R, Wagner J, Solis-Escalante T, Müller-Putz GR. High and low gamma EEG oscillations in central sensorimotor areas are conversely modulated during the human gait cycle. NeuroImage 112: 318–326, 2015. doi: 10.1016/j.neuroimage.2015.03.045. [DOI] [PubMed] [Google Scholar]

[R63] 63.Hamacher D, Singh NB, Van Dieën JH, Heller MO, Taylor WR. Kinematic measures for assessing gait stability in elderly individuals: a systematic review. J R Soc Interface 8: 1682–1698, 2011. doi: 10.1098/rsif.2011.0416. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R64] 64.Herssens N, Verbecque E, Hallemans A, Vereeck L, Van Rompaey V, Saeys W. Do spatiotemporal parameters and gait variability differ across the lifespan of healthy adults? A systematic review. Gait Posture 64: 181–190, 2018. doi: 10.1016/j.gaitpost.2018.06.012. [DOI] [PubMed] [Google Scholar]

[R65] 65.Wilkinson RD, Mazzo MR, Feeney DF. Rethinking the Statistical Analysis of Neuromechanical Data. Exerc Sport Sci Rev 51: 43–50, 2023. doi: 10.1249/JES.0000000000000308. [DOI] [PubMed] [Google Scholar]

[R66] 66.Benjamini Y, Hochberg Y. Controlling the False Discovery Rate: A Practical and Powerful Approach to Multiple Testing. J R Stat Soc Ser B Methodol 57: 289–300, 1995. doi: 10.1111/j.2517-6161.1995.tb02031.x. [DOI] [Google Scholar]

[R67] 67.Cohen J. Statistical Power Analysis for the Behavioral Sciences. 2nd ed. New York: Routledge, 1988. [Google Scholar]

[R68] 68.Selya AS, Rose JS, Dierker LC, Hedeker D, Mermelstein RJ. A Practical Guide to Calculating Cohen’s f2, a Measure of Local Effect Size, from PROC MIXED. Front Psychol 3: 111, 2012. doi: 10.3389/fpsyg.2012.00111. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R69] 69.Reuter-Lorenz PA, Cappell KA. Neurocognitive aging and the compensation hypothesis. Curr Dir Psychol Sci 17: 177–182, 2008. doi: 10.1111/j.1467-8721.2008.00570.x. [DOI] [Google Scholar]

[R70] 70.Sato SD, Shah VA, Fettrow T, Hall KG, Tays GD, Cenko E, Roy A, Clark DJ, Ferris DP, Hass CJ, Manini TM, Seidler RD. Resting state brain network segregation is associated with walking speed and working memory in older adults. bioRxiv: 2024.05.07.592861, 2024. [DOI] [PubMed] [Google Scholar]

[R71] 71.Gyurkovics M, Clements GM, Low KA, Fabiani M, Gratton G. Stimulus-Induced Changes in 1/f-like Background Activity in EEG. J Neurosci 42: 7144–7151, 2022. doi: 10.1523/JNEUROSCI.0414-22.2022. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R72] 72.Voytek B, Kramer MA, Case J, Lepage KQ, Tempesta ZR, Knight RT, Gazzaley A. Age-Related Changes in 1/f Neural Electrophysiological Noise. J Neurosci 35: 13257–13265, 2015. doi: 10.1523/JNEUROSCI.2332-14.2015. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R73] 73.Vlahou EL, Thurm F, Kolassa I-T, Schlee W. Resting-state slow wave power, healthy aging and cognitive performance. Sci Rep 4: 5101, 2014. doi: 10.1038/srep05101. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R74] 74.Garofalo S, Giovagnoli S, Orsoni M, Starita F, Benassi M. Interaction effect: Are you doing the right thing? PLOS ONE 17: e0271668, 2022. doi: 10.1371/journal.pone.0271668. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R75] 75.Jensen O, Spaak E, Zumer JM. Human Brain Oscillations: From Physiological Mechanisms to Analysis and Cognition. In: Magnetoencephalography: From Signals to Dynamic Cortical Networks, edited by Supek S, Aine CJ. Springer International Publishing, p. 471–517. [Google Scholar]

[R76] 76.Roeder L, Boonstra TW, Smith SS, Kerr GK. Dynamics of corticospinal motor control during overground and treadmill walking in humans. J Neurophysiol 120: 1017–1031, 2018. doi: 10.1152/jn.00613.2017. [DOI] [PubMed] [Google Scholar]

[R77] 77.Roeder L, Boonstra TW, Kerr GK. Corticomuscular control of walking in older people and people with Parkinson’s disease. Sci Rep 10: 2980, 2020. doi: 10.1038/s41598-020-59810-w. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R78] 78.Bradford JC, Lukos JR, Ferris DP. Electrocortical activity distinguishes between uphill and level walking in humans. J Neurophysiol 115: 958–966, 2016. doi: 10.1152/jn.00089.2015. [DOI] [PubMed] [Google Scholar]

[R79] 79.Hua A, Wang G, Bai J, Hao Z, Yang Y, Luo X, Liu J, Meng J, Wang J. Rapid reconfiguration of cortical networks after repeated exposure to visual-vestibular conflicts. Sci Rep 14: 21943, 2024. doi: 10.1038/s41598-024-73111-6. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R80] 80.Peterson SM, Ferris DP. Differentiation in Theta and Beta Electrocortical Activity between Visual and Physical Perturbations to Walking and Standing Balance. eNeuro 5: ENEURO.0207-18.2018, 2018. doi: 10.1523/ENEURO.0207-18.2018. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R81] 81.Kuo AD. The six determinants of gait and the inverted pendulum analogy: A dynamic walking perspective. Hum Mov Sci 26: 617–656, 2007. doi: 10.1016/j.humov.2007.04.003. [DOI] [PubMed] [Google Scholar]

[R82] 82.Kuo AD, Donelan JM. Dynamic principles of gait and their clinical implications. Phys Ther 90: 157–174, 2010. doi: 10.2522/ptj.20090125. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R83] 83.Doherty TJ. Invited Review: Aging and sarcopenia. J Appl Physiol 95: 1717–1727, 2003. doi: 10.1152/japplphysiol.00347.2003. [DOI] [PubMed] [Google Scholar]

[R84] 84.Kirk EA, Hope KT, Sober SJ, Sauerbrei BA. An output-null signature of inertial load in motor cortex. Nat Commun 15: 1–20, 2024. doi: 10.1038/s41467-024-51750-7. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R85] 85.Bulea TC, Jonghyun Kim null, Damiano DL, Stanley CJ, Hyung-Soon Park null. User-driven control increases cortical activity during treadmill walking: an EEG study. Annu Int Conf IEEE Eng Med Biol Soc IEEE Eng Med Biol Soc Annu Int Conf 2014: 2111–2114, 2014. doi: 10.1109/EMBC.2014.6944033. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R86] 86.Buschman TJ, Miller EK. Top-Down Versus Bottom-Up Control of Attention in the Prefrontal and Posterior Parietal Cortices. Science 315: 1860–1862, 2007. doi: 10.1126/science.1138071. [DOI] [PubMed] [Google Scholar]

[R87] 87.Gola M, Magnuski M, Szumska I, Wróbel A. EEG beta band activity is related to attention and attentional deficits in the visual performance of elderly subjects. Int J Psychophysiol 89: 334–341, 2013. doi: 10.1016/j.ijpsycho.2013.05.007. [DOI] [PubMed] [Google Scholar]

[R88] 88.Wróbel A. Beta activity: a carrier for visual attention. Acta Neurobiol Exp (Warsz) 60: 247–260, 2000. doi: 10.55782/ane-2000-1344. [DOI] [PubMed] [Google Scholar]

[R89] 89.Anderson KL, Ding M. Attentional modulation of the somatosensory mu rhythm. Neuroscience 180: 165–180, 2011. doi: 10.1016/j.neuroscience.2011.02.004. [DOI] [PubMed] [Google Scholar]

[R90] 90.Lopez S, Bini F, Del Percio C, Marinozzi F, Celletti C, Suppa A, Ferri R, Staltari E, Camerota F, Babiloni C. Electroencephalographic sensorimotor rhythms are modulated in the acute phase following focal vibration in healthy subjects. Neuroscience 352: 236–248, 2017. doi: 10.1016/j.neuroscience.2017.03.015. [DOI] [PubMed] [Google Scholar]

[R91] 91.Wang G, Yang Y, Wang J, Hao Z, Luo X, Liu J. Dynamic changes of brain networks during standing balance control under visual conflict. Front Neurosci 16, 2022. doi: 10.3389/fnins.2022.1003996. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R92] 92.Peterson SM, Ferris DP. Group-level cortical and muscular connectivity during perturbations to walking and standing balance. NeuroImage 198: 93–103, 2019. doi: 10.1016/j.neuroimage.2019.05.038. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R93] 93.Barollo F, Hassan M, Petersen H, Rigoni I, Ramon C, Gargiulo P, Fratini A. Cortical Pathways During Postural Control: New Insights From Functional EEG Source Connectivity. IEEE Trans Neural Syst Rehabil Eng 30: 72–84, 2022. doi: 10.1109/TNSRE.2022.3140888. [DOI] [PubMed] [Google Scholar]

[R94] 94.Oliveira AS, Schlink BR, Hairston WD, König P, Ferris DP. Restricted vision increases sensorimotor cortex involvement in human walking. J Neurophysiol 118: 1943–1951, 2017. doi: 10.1152/jn.00926.2016. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R95] 95.Liu C, Downey RJ, Salminen JS, Arvelo Rojas S, Richer N, Pliner EM, Hwang J, Cruz-Almeida Y, Manini TM, Hass CJ, Seidler RD, Clark DJ, Ferris DP. Electrical brain activity during human walking with parametric variations in terrain unevenness and walking speed. Imaging Neurosci 2: 1–33, 2024. doi: 10.1162/imag_a_00097. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R96] 96.Sturnieks DL, St George R, Lord RS. Balance disorders in the elderly. Neurophysiol Clin Neurophysiol 38: 467–478, 2008. doi: 10.1016/j.neucli.2008.09.001. [DOI] [PubMed] [Google Scholar]

[R97] 97.Xing L, Bao Y, Wang B, Shi M, Wei Y, Huang X, Dai Y, Shi H, Gai X, Luo Q, Yin Y, Qin D. Falls caused by balance disorders in the elderly with multiple systems involved: Pathogenic mechanisms and treatment strategies. Front Neurol 14: 1128092, 2023. doi: 10.3389/fneur.2023.1128092. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R98] 98.Benwell CSY, Davila-Pérez P, Fried PJ, Jones RN, Travison TG, Santarnecchi E, Pascual-Leone A, Shafi MM. EEG spectral power abnormalities and their relationship with cognitive dysfunction in patients with Alzheimer’s disease and type 2 diabetes. Neurobiol Aging 85: 83–95, 2020. doi: 10.1016/j.neurobiolaging.2019.10.004. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R99] 99.Kopčanová M, Tait L, Donoghue T, Stothart G, Smith L, Flores-Sandoval AA, Davila-Perez P, Buss S, Shafi MM, Pascual-Leone A, Fried PJ, Benwell CSY. Resting-state EEG signatures of Alzheimer’s disease are driven by periodic but not aperiodic changes. Neurobiol Dis 190: 106380, 2024. doi: 10.1016/j.nbd.2023.106380. [DOI] [PubMed] [Google Scholar]

[R100] 100.Karalunas SL, Ostlund BD, Alperin BR, Figuracion M, Gustafsson HC, Deming EM, Foti D, Antovich D, Dude J, Nigg J, Sullivan E. Electroencephalogram aperiodic power spectral slope can be reliably measured and predicts ADHD risk in early development. Dev Psychobiol 64: e22228, 2022. doi: 10.1002/dev.22228. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R101] 101.Peterson EJ, Rosen BQ, Belger A, Voytek B, Campbell AM. Aperiodic Neural Activity is a Better Predictor of Schizophrenia than Neural Oscillations. Clin EEG Neurosci 54: 434–445, 2023. doi: 10.1177/15500594231165589. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R102] 102.Robertson MM, Furlong S, Voytek B, Donoghue T, Boettiger CA, Sheridan MA. EEG power spectral slope differs by ADHD status and stimulant medication exposure in early childhood. J Neurophysiol 122: 2427–2437, 2019. doi: 10.1152/jn.00388.2019. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R103] 103.Park J, Ho RLM, Wang W, Nguyen VQ, Coombes SA. The effect of age on alpha rhythms in the human brain derived from source localized resting-state electroencephalography. NeuroImage 292: 120614, 2024. doi: 10.1016/j.neuroimage.2024.120614. [DOI] [PubMed] [Google Scholar]

[R104] 104.Hollman JH, McDade EM, Petersen RC. Normative spatiotemporal gait parameters in older adults. Gait Posture 34: 111–118, 2011. doi: 10.1016/j.gaitpost.2011.03.024. [DOI] [PMC free article] [PubMed] [Google Scholar]

[R105] 105.Schmitt AC, Baudendistel ST, Lipat AL, White TA, Raffegeau TE, Hass CJ. Walking indoors, outdoors, and on a treadmill: Gait differences in healthy young and older adults. Gait Posture 90: 468–474, 2021. doi: 10.1016/j.gaitpost.2021.09.197. [DOI] [PubMed] [Google Scholar]

[R106] 106.Lee SJ, Hidler J. Biomechanics of overground vs. treadmill walking in healthy individuals. J Appl Physiol 104: 747–755, 2008. doi: 10.1152/japplphysiol.01380.2006. [DOI] [PubMed] [Google Scholar]

[R107] 107.Salinas MM, Wilken JM, Dingwell JB. How humans use visual optic flow to regulate stepping during walking. Gait Posture 57: 15–20, 2017. doi: 10.1016/j.gaitpost.2017.05.002. [DOI] [PubMed] [Google Scholar]

[R108] 108.Warren WH, Kay BA, Zosh WD, Duchon AP, Sahuc S. Optic flow is used to control human walking. Nat Neurosci 4: 213–216, 2001. doi: 10.1038/84054. [DOI] [PubMed] [Google Scholar]

PERMALINK

Gait Speed Related Changes in Electrocortical Activity in Younger and Older Adults

Jacob Salminen

Chang Liu

Erika M Pliner

Madison Tenerowicz

Arkaprava Roy

Natalie Richer

Jungyun Hwang

Chris J Hass

David J Clark

Rachael D Seidler

Todd M Manini

Yenisel Cruz-Almeida

Daniel P Ferris

Abstract

Graphical Abstract

NEW & NOTEWORTHY

INTRODUCTION

METHODS

Participants

Table 2.

Figure 1.

Table 1.

Figure 2.

Experimental Protocol

Figure 3.

Figure 5.

EEG & MRI Data Collection Protocol

EEG.

MRI.

Data Processing

IMU & Ground Reaction Force Measures.

EEG Data Preprocessing.

Figure 4.

Electrical Source Modeling.

Source Rejection.

Source Clustering.

Spectral Power Calculations.

Spectral Principal Component Analysis.

Statistical Analysis

Figure 8. Left sensorimotor area’s topographies and power spectral density plots.

RESULTS

Behavioral Differences in Age Groups

Figure 7. Summary of average FOOOF corrected power spectral density modulations in the theta, alpha, and beta bands across brain areas.

Table 3.

Identification of Brain Areas Involved in Walking

Table 4.

Table 5.

Figure 17. Right occipital area’s topographies and power spectral density plots.

Figure 20. Right temporal area’s topographies and power spectral density plots.

Table 12.

Table 15.

EEG Spectral Power with Changes in Speed

Figure 9. Right sensorimotor area’s topographies and power spectral density plots.

Figure 10. Left posterior parietal area’s topographies and power spectral density plots.

Figure 11. Right posterior parietal area’s topographies and power spectral density plots.

Figure 12. Mid cingulate area’s topographies and power spectral density plots.

Figure 13. Left supplementary motor area’s topographies and power spectral density plots.

Figure 14. Right cuneus area’s topographies and power spectral density plots.

Table 6.

Table 7.

Table 8.

Table 10.

Table 9.

Table 11.

Additional Brain Areas

Figure 18. Left occipital area’s topographies and power spectral density plots.

Table 13.

Figure 19. Left temporal area’s topographies and power spectral density plots.

Table 14.

DISCUSSION

Figure 15. Within-group baselined event-related spectral perturbations in right cuneus.

Figure 16. Within-group baselined event-related spectral perturbations in the left supplementary motor.

Figure 21.

Figure 6.

ACKNOWLEDGEMENTS

GRANTS

Footnotes

REFERENCES