Paired corticospinal‐motoneuronal stimulation enhances ballistic motor learning and corticospinal plasticity in older adults

Abstract

Abstract

Late adulthood is accompanied by declines in manual motor performance and reduced neuroplasticity, which can influence the effects of motor practice and learning. Corticomotoneuronal (CM) connectivity can be targeted non‐invasively through individualized paired corticospinal‐motoneuronal stimulation (PCMS) to prime ballistic motor learning in young adults. However, the priming effects of PCMS on motor output and ballistic motor learning in older adults remain unexplored. Part one of this study investigates ballistic motor performance and learning in young (20–30 years) and older (65–75 years) adults as within‐session changes in peak acceleration of rapid index finger flexions and delayed retention 1 week later. The results demonstrate that older adults display lower maximal acceleration compared to young adults and smaller improvements with practice, indicating inferior learning and low levels of delayed retention. Part two of the study investigates the effects of PCMS on motor learning and corticospinal excitability in older adults. Corticospinal excitability was assessed throughout the experiment by recording motor evoked potentials from the first dorsal interosseous. PCMS increased subsequent ballistic learning and corticospinal excitability after practice compared to SHAM. Importantly, combined PCMS and motor practice also enhanced long‐term retention, and performance remained enhanced 7 days later. This means that PCMS effectively reinstated the otherwise absent long‐term learning in older adults. We demonstrate that PCMS primes experience‐dependent plasticity accompanying motor learning resulting in long‐term benefits on motor performance in older adults. These findings highlight the potential of PCMS to enhance the effects of motor practice and benefit functional abilities in older adults.

Key points

• Late adulthood is associated with reduced activation of spinal motoneurons during vigorous movements, resulting in slower and less precise movements.

• Older adults (aged 65–75 years) display lower ballistic motor performance compared to younger adults (aged 20–30 years); furthermore, older adults exhibit smaller improvements during practice, and lower retention.

• A single session of paired corticospinal‐motoneuronal stimulation (PCMS) increases corticospinal excitability and primes within‐session ballistic motor learning in older adults.

• A single session of PCMS improves long‐term retention following ballistic motor learning.

• We provide proof‐of‐principle that PCMS represents a potential strategy to enhance the effects of motor practice and counteract age‐related decline in motor function.

Abstract figure legend The first part of this experimental study showed that late adulthood (adults aged 65–75 years) is associated with reduced motor performance and learning of simple vigorous finger movements. The second part of the study shows that ballistic motor learning can be improved in older adults when a practice session is preceded by a session of non‐invasive paired corticospinal‐motoneuronal stimulations (PCMS). The study shows that a single session of PCMS improves long‐term retention following ballistic motor learning in older adults. Created in https://BioRender.com

Introduction

Ageing decreases the ability to effectively activate spinal motoneurons during vigorous movements due to decreased supraspinal and afferent drive, along with decreased excitability of the motoneurons (Clark & Taylor, 2012; Shaffer & Harrison, 2007). Ultimately, this leads to lower maximal discharge rates of spinal motoneurons (Klass et al., 2008) and a decline in the number of motoneurons (Drey et al., 2014), which impairs the ability to perform fast and precise ballistic movements (del Vecchio et al., 2019; Duchateau & Baudry, 2014). In addition to central nervous system changes, ageing is also associated with peripheral alterations, such as loss of muscle mass and changes in fibre type, all of which lead to further functional impairments (Deschenes, 2004; Larsson et al., 2019). Ballistic motor practice increases maximal discharge rates and motor unit recruitment through use‐dependent plasticity occurring at supraspinal and spinal loci (Giesebrecht et al., 2012; Lundbye‐Jensen et al., 2011; Muellbacher et al., 2001). While some studies have demonstrated similar relative ballistic learning effects between young and older adults (Hinder et al., 2011), other studies have shown that late adulthood is characterized by both decreased responsiveness to ballistic practice (Rogasch et al., 2009) and deficits in memory consolidation (Roig et al., 2014). Furthermore, older adults display an increased susceptibility to post‐practice interference (Roig et al., 2014) and lower (or no) interlimb transfer after motor learning (Hinder et al., 2013, 2011). Thus, finding ways to facilitate plasticity and augment motor learning in older adults is an important perspective when trying to improve and preserve functional abilities with increasing age. This is particularly relevant in rehabilitation settings.

Associative plasticity can be noninvasively induced in humans by pairing transcranial magnetic stimulation (TMS) with peripheral nerve stimulation (PNS) at specific time intervals (Harel & Carmel, 2016). These stimulation protocols are inspired by principles of cellular spike‐timing‐dependent plasticity (STDP) (Caporale & Dan, 2008). An example is paired corticospinal‐motoneuronal stimulation (PCMS) (Taylor & Martin, 2009). PCMS protocols in which TMS volleys are timed to arrive at the CM‐presynapse 2 ms before antidromic volleys evoked from PNS have been shown to enhance voluntary motor output in healthy young individuals (Taylor & Martin, 2009) and improve hand function in individuals with spinal cord injury (Bunday & Perez, 2012). PCMS is emerging as a therapeutic tool with promising applications for rehabilitation following spinal cord injuries (Christiansen & Perez, 2018; Jo et al., 2020). Existing research on the potential benefits of PCMS in neurologically intact individuals has predominantly focused on young adults (D'Amico et al., 2018, 2020; Dongés et al., 2018; Fitzpatrick et al., 2016; Taylor & Martin, 2009). However, the therapeutic potential of PCMS may be greater in late adulthood, characterized by age‐related declines in motor function and a higher prevalence of neurological disorders and injuries leading to compromised motor function (Hirtz et al., 2007). We recently demonstrated that PCMS effectively improved ballistic motor learning in a younger population (Bjørndal et al., 2024). This proof‐of‐principle study lends credence to the use of PCMS to improve the effects of motor practice to counteract age‐related declines in motor function and enhance the outcome of sensorimotor rehabilitation.

In the present study, we first compared how performance and learning of a task contingent on maximal discharge rate and recruitment gain differ between young adults (aged 20–30 years) and older adults (aged 65–75 years). We hypothesized that older adults would display inferior performance and motor learning compared to young adults. The second part of the study aimed specifically at older adults and investigated how PCMS interacted with ballistic motor training and learning. The primary aim was to determine whether PCMS, in comparison to a SHAM protocol, could augment training‐related improvements in ballistic motor performance and recover age‐related declines in retention of motor learning in older adults. Measurements of index finger acceleration were obtained before and immediately after PCMS or SHAM, after motor practice, and once again 1 week later. The secondary aim was to explore the effect of PCMS on corticospinal‐ and α‐motoneuronal excitability, assessed as motor evoked potentials (MEPs) elicited by single‐pulse TMS and F‐wave amplitudes, respectively. We hypothesized that PCMS, as opposed to the SHAM protocol, would improve motor learning, resulting in enhanced performance after ballistic motor practice, increased MEPs, and unchanged F‐waves. Moreover, we expected that the behavioural priming effect would remain robust upon reassessment 7 days later.

Methods

Ethical approval

The study was reviewed and approved by the local ethics committee (protocol H‐17019671), and all experimental procedures were conducted in accordance with the Declaration of Helsinki. The study was not registered in any database prior to recruitment. All participants received written and oral information about the study and provided written informed consent prior to enrolment.

Participants

Sixty participants were included in the study: 40 older adults aged 65–75 years and 20 young adults aged 20–30 (see Table 1 for the participant characteristics). The inclusion criteria were as follows: neurologically healthy and right‐handed according to the Edinburgh Handedness Inventory (Oldfield, 1971), without a known family history of epilepsy. Ten of the twenty young adults completed the identical protocol involving SHAM stimulation before ballistic motor practice, as part of our previous study (Bjørndal et al., 2024).

Table 1.

Participant characteristics at baseline

	Young adults (SHAM)	Older adults (SHAM)	Older adults (PCMS)
Participants (male/female)	20 (11/9)	20 (8/12)	20 (9/11)
Age, years	25.1 (2.2)	68.3 (2.6)	68.7 (3.1)
Height, cm	178.3 (11.5)	172 (8.3)	168 (7.4)
Weight, kg	78.1 (12.1)	73 (13)	72 (12)
Handedness (LQ)	0.9 (0.2)	0.9 (0.2)	0.9 (0.3)
MMSE	–	28.5 (1.3)	28.7 (1.1)
Sleep before main experiment, hours	8.0 (0.9)	7.1 (1.1)	7.6 (1.0)
Sleep before retention test, hours	7.9 (0.8)	7.2 (1.1)	7.5 (1.0)
rMT, % MSO	42 (8.6)	42.1 (9.2)	41.5 (8.0)

Values are reported as means (standard deviation), and categorical values are presented as counts. LQ, laterality quotient (from −1 (left‐handed, to +1 (right‐handed); MMSE, mini mental state examination (0–30 items); MSO, maximum stimulator output; rMT, resting motor threshold.

Experimental design

All participants completed the same protocol, with variations only in the SHAM or PCMS intervention. In the first part of the study, we compared behavioural measurements, specifically ballistic motor performance and learning, between older and young participants. The neurophysiological measurements from the young group are therefore not shown in the present study. In the second part of the study, we focused specifically on older adults to investigate how PCMS affected motor learning and neurophysiological measures. The older adults were randomly allocated to two groups: an active PCMS protocol (PCMS) or a controlled SHAM protocol (SHAM). Thus, the first part of the study compared the SHAM group of older adults with the SHAM group of young adults, while the second part analysed the effects of PCMS within the older adult group. At the beginning of the experiment, latencies of M‐waves, F‐waves, and MEPs were recorded to calculate the individual peripheral and central conduction times used to individualize the PCMS protocol (Table 2). The PCMS protocol used includes an inter‐arrival interval where a TMS volley arrives at the CM‐presynapse 2 ms before the antidromic volley from PNS (Bunday & Perez, 2012). After PCMS or SHAM, the participants practiced a ballistic motor task for three practice blocks. Measurements of motor performance (quantified as acceleration during rapid finger flexions), corticospinal excitability (quantified as peak‐to‐peak MEP amplitudes), and α‐motoneuronal excitability (quantified as F‐wave amplitudes) were obtained before and after intervention (PCMS or SHAM). MEPs were also measured every 15th minute for 45 min after motor practice to investigate the time course of corticospinal excitability after practice. Furthermore, motor performance in the ballistic motor task was reassessed after 7 days to investigate the long‐term effects of PCMS or SHAM on retention of the ballistic motor learning (Fig. 1A ).

Table 2.

Measures of latencies and calculated conduction times in the two groups: PCMS and SHAM

	PCMS	SHAM
M max latency, ms	4.1 (0.5)	4.3 (0.7)
F‐wave latency, ms	30.4 (2.7)	30.0 (3.1)
MEPactive latency, ms	23.0 (2.1)	22.9 (2.3)
Peripheral conduction time, ms	13.1 (1.2)	12.8 (1.3)
Central conduction time, ms	5.7 (1.3)	5.7 (1.0)
Inter‐stimulus interval, ms	5.4 (1.6)	5.1 (1.5)

Values are reported as means (SD). M _max, maximum M‐wave (maximal compound action potential); MEP_active, MEP latency measured during an isometric contraction (10%MVC).

Figure 1. Experimental procedure.

A, overview of the protocol. All three groups (Young adults (SHAM), Older adults (SHAM), and Older adults (PCMS) completed the same experimental procedures only varying the stimulation protocol (PCMS/SHAM). On the main experimental day, younger participants received SHAM, and older participants were randomized to SHAM or PCMS before a motor practice session (incl. three practice blocks of 50 trials, with two 2‐min break in between). Motor test trials were assessed before and after PCMS or SHAM and motor practice and again after 1 week (7‐day retention). Motor evoked potentials were also assessed before and after PCMS or SHAM and after motor practice (post‐practice 0 min), and again 15, 30 and 45 min post‐practice (PP15‐45). B, illustration of the PCMS and SHAM protocols. Interstimulus intervals were calculated individually based on the measured latencies. Both PCMS and SHAM used the same TMS intensity (150% rMT) and frequency (0.1 Hz) between the 100 pairs of TMS and PNS pulses. The TMS coil was flipped in the SHAM condition, thereby avoiding M1 stimulation. The PNS intensity was lowered to the perceptual threshold, thus avoiding activation of α‐motoneurons; see the example of single participant traces of the 100 paired stimulations from PCMS and SHAM. C, illustration of the ballistic motor task. Participants’ hands were placed in a handle that only allowed index finger flexions. Motor performance was quantified as peak acceleration, and EMG was recorded from m. FDI. Participants were presented with feedback on their performance after each practice trial, visualized as a score of peak acceleration as a percentage of their mean baseline score. Created in BioRender.com.

Ballistic motor task

The ballistic motor task was the same as described by Bjørndal et al. (2024). In brief, participants were asked to perform and improve maximal index finger flexion accelerations. The participants’ right arm was flexed at the elbow at approximately 90°, enabling the hand to grasp a custom‐built handle (Fig. 1C ). A metal splint oriented perpendicular to the handle's axis of rotation allowed only the index finger to be flexed. An accelerometer mounted on the metal splint recorded the acceleration (amplified and filtered (low pass 20 Hz) and sampled at 1 kHz on a computer with a USB6008 DAQ Board (National Instruments, Inc.). During each trial, the participants were paced by a horizontal blue trace running from left to right on a white computer screen. During a 1‐s timeframe, participants performed a rapid finger flexion, with a new trial beginning every 4th second. A green vertical midline was displayed on the screen, and participants were instructed to perform the movement near the midline to maintain rhythm. Participants received thorough oral and visual instructions about the purpose of the task, and were explicitly informed that their score would not depend on hitting the midline. Participants were allowed 5 familiarization trials before the tests of 10 trials without augmented feedback at each time point (baseline, post‐stimulations, immediate retention, and 7‐day retention). The immediate retention test was completed within 2 min following the completion of motor practice. During practice (3 × 50 trials) and the additional practice block on day 7, augmented feedback was provided following the completion of each trial, as the peak acceleration in the respective trial normalized to the mean acceleration measured at baseline. Testing motor performance both in trials with and without augmented feedback allows assessment of changes in motor performance with motor practice independently of the immediate effect of augmented feedback on performance. In this way, it allows a learning‐performance‐distinction, which is particularly relevant in the area of motor learning (for review, see Kantak & Winstein 2012). The augmented feedback was presented as a score (peak acceleration as a percentage of baseline) for 2 s before the next trial began. Additionally, neutral verbal encouragement was provided during practice at least every 10th trial for all participants. This included sentences like ‘move as fast as you can’.

Electromyography

Electromyography (EMG) was recorded from the right first dorsal interosseous muscle (m. FDI) through surface electrodes (Ag‐AgCl, 1 cm diameter) applied on the skin after preparation with medical sandpaper. The electrodes were placed in a muscle‐belly‐tendon montage, with the active electrode on the muscle belly. A zinc plate was used as the ground electrode and was placed at the base of the hand. The EMG signals were amplified (×200), filtered (band‐pass 5 Hz to 1 kHz), and sampled at 2 kHz on a computer for offline analysis (Cambridge Electronic Design 1401 with Signal software v6.05 or Spike2 software v7.10). Line noise (50 Hz) was removed using a Hum Bug noise eliminator (Digitimer).

Paired corticospinal‐motoneuronal stimulations and SHAM stimulations

Both protocols consisted of 100 paired TMS and PNS stimulations with a frequency of 0.1 Hz similar to Bunday and Perez (2012) (Fig. 1B ). For the SHAM protocol, an inactive TMS coil was used to maintain a comparable feeling of a coil on the head while avoiding the transcranial effects of the TMS pulses. In addition, an active TMS‐coil was flipped 180° and placed on top of the inactive coil, with TMS intensity settings (150% rMT) similar to the PCMS protocol to produce the TMS ‘click’ sound in both groups. PNS intensity was set to 130% M _max (maximal compound muscle action potential) intensity in the PCMS condition, and at the perceptual threshold in the SHAM condition. The SHAM protocol thereby mimicked a similar experimental environment, and to some extent controlled for the auditory and somatosensory co‐stimulation seen after TMS (Siebner et al., 2022). Importantly, the timing of stimulations was based on individualized peripheral conduction time (PCT) and central conduction times (CCT) calculated from F‐wave, M _max and MEP latencies:

$$\mathrm{PCT}=\frac{\left(\mathrm{latency}-M\mathrm{latency}\right)}{2},$$

$$\mathrm{CCT}=(\mathrm{MEP}\mathrm{latency}-\left(\mathrm{PCT}+M\mathrm{latency}\right),$$

$$\mathrm{Interstimulus}\mathrm{interval}=\left(\mathrm{PCT}-\mathrm{CCT}\right)-2\mathrm{ms}.$$

Peripheral electrical nerve stimulation

Electrical stimulation with high voltage electrical current (200 µs pulse duration, DS7A; Digitimer) was delivered to the ulnar nerve at the wrist (Bar Stimulating Electrode, Digitimer) to measure the maximal compound muscle action potential (M _max) and F‐waves. F‐wave amplitudes and persistence (the percentage of stimuli evoking a response) were obtained with the cathode positioned proximally to indirectly quantify α‐motoneuron excitability (McNeil et al., 2013). To ensure validity and reliability of F‐waves, 60 stimulations at supramaximal (130% of M _max) intensity were delivered at 1 Hz at each time point (baseline, post‐stimulations, and post‐practice) (Lin & Floeter, 2004). The F‐wave recordings were filtered offline using a 2nd order Bessel high‐pass filter (200 Hz) to allow accurate assessment of onset latency, amplitude, and persistence (Christiansen et al., 2018; Khan et al., 2012).

Transcranial magnetic stimulation

We assessed corticospinal excitability as the peak‐to‐peak amplitudes of TMS‐evoked FDI muscle responses (motor evoked potentials; MEPs). Monophasic single‐pulse TMS was applied to the contralateral M1 to the dominant hand via a figure‐of‐eight TMS coil (Magstim D70² connected to a Magstim²⁰⁰). The hotspot was defined as the stimulation site that elicited the largest and most consistent MEPs, and a neuro‐navigation system (Brainsight 2, Rogue Research, Montreal, Canada) was used to ensure stable coil positioning once the hotspot was found using a template MRI scan. The coil was placed with the centre oriented parallel to the scalp over the motor hotspot and with the handle of the coil pointing backward at an angle of ∼45° to the sagittal and horizontal axes. The resting motor threshold (rMT) was defined as the stimulus intensity needed to elicit recognizable MEPs with an amplitude above 0.05 mV in 5/10 consecutive stimulations (Rossini et al., 2015). Muscle relaxation was monitored using concurrent electromyographic (EMG) recordings. Twenty TMS stimulations (120% of rMT) were delivered at each time point (baseline, post‐stimulations, and post‐practice (PP0), PP15, PP30, and PP45) to measure MEP amplitudes.

Data analysis

Ballistic motor performance was quantified as the peak acceleration during rapid index finger flexion. In the first part of the study, age‐related differences in motor performance and learning were investigated by comparing both absolute and relative performance changes. Relative performance was expressed as peak acceleration scores normalized to the baseline value. The second part of the study focused on the effects of PCMS vs. SHAM on the effects of motor practice within the older adults group, with an expectation of similar baseline values between the two older groups (PCMS vs. SHAM). Therefore, data are presented relative to the baseline to isolate the intervention effect on ballistic motor learning.

FDI EMG was used to measure MEPs, F‐waves, and M‐waves. During recordings, we closely monitored each frame for background muscle activity 200 ms prior to stimulation. Recordings were stopped when background muscle activity appeared and participants were instructed to maintain a resting muscle state. Additional stimulations were performed to ensure the desired total of stimulations at each measurement. Despite this, electrophysiological data from two participants in the SHAM group were removed from the later analysis, because of the inability to keep the muscle in a resting state due to constant low muscle activity. For the MEP time course, we also calculated a grand mean across the time course as a summary of the overall after‐effects, allowing a general conclusion on the effect of the intervention, similar to the procedure of previous studies (Centeno et al., 2018; Ostadan et al., 2016).

Both MEPs and F‐waves were normalized to M _max at the nearest time point. MEPs were subsequently displayed as a percentage of the baseline. FDI EMG measured during the ballistic task was analysed to investigate whether the EMG during ballistic finger movements changed with motor practice. EMG data were filtered using a moving root‐mean‐square filter with a time constant of 50 ms (similar to Aagaard et al., 2002). EMG root‐mean‐square (EMG_RMS) amplitude was calculated in a time window from EMG onset to 70 ms after EMG onset. Additionally, the rate of EMG rise was calculated from EMG onset to 30 ms. EMG measurements were normalized to the individual M _max amplitude to allow between‐session comparisons (main experiment and 7‐day retention).

Statistical analysis

All statistical analyses were performed using R (R Core Team, 2022, version 4.1.3). At baseline, unpaired t tests were used to compare participant characteristics between groups. The effect of age group (older vs. young) on ballistic motor performance was investigated using a linear mixed‐effects model. Similarly, the effect of stimulations (PCMS vs. SHAM) on ballistic motor performance, corticospinal and α‐motoneuronal excitability was investigated using linear mixed effects models, which were set up individually for all dependent variables (peak acceleration, MEP, and F‐waves). Intercepts for each participant within each time block were added as a random effect. Assumptions of normality and homogeneity of variance of residuals were inspected using quantile‐quantile plots and residual plots. The MEP data followed a gamma distribution and was therefore analysed with a generalized linear mixed model (GLMM), using the glmer() function from the lme4 package (Bates et al., 2015).

For ballistic motor performance measures acquired without provision of augmented feedback (test trials) at baseline, post‐stimulation, immediate retention, and delayed retention, the following model was specified:

$$\begin{array}{ccc}& & \mathrm{Peak}\mathrm{acceleratio}{\mathrm{n}}_{\mathrm{Test}\mathrm{trials}}\sim \mathrm{Age}\mathrm{group}\times \mathrm{Time}\hfill \\ & & +\left(1|\mathrm{Subject}\right),\hfill \end{array}$$ (1a)

$$\begin{array}{ccc}& & \mathrm{Peak}\mathrm{acceleratio}{\mathrm{n}}_{\mathrm{Testtrials}}\sim \mathrm{Protocol}\times \mathrm{Time}\hfill \\ & & +\left(1|\mathrm{Subject}\right).\hfill \end{array}$$ (1b)

Model (1a) was set up for both absolute performance (peak acceleration (g)) and normalized performance (peak acceleration as a percentage of baseline) to compare Older vs. Young, with Age group as a factor of 2 levels (older adults, young adults) and Time as a factor of 4 levels (baseline, post‐stimulation, immediate retention, 7Dretention). For normalized performance, the Time factor did not include baseline. Model (1b) compared the normalized performance between PCMS and SHAM, with Protocol as a factor of 2 levels (PCMS, SHAM), and a similar Time factor as Model (1a).

Similarly, for motor performance measures acquired during ballistic motor practice (practice trials) with augmented feedback, the following model was specified:

$$\begin{array}{ccc}& & \mathrm{Peak}\mathrm{acceleratio}{\mathrm{n}}_{\mathrm{Practicetrials}}\sim \mathrm{Age}\mathrm{group}\times \mathrm{Block}\hfill \\ & & +\left(1|\mathrm{Subject}\right),\hfill \end{array}$$ (2a)

$$\begin{array}{ccc}& & \mathrm{Peak}\mathrm{acceleratio}{\mathrm{n}}_{\mathrm{Practicetrials}}\sim \mathrm{Protocol}\times \mathrm{Block}\hfill \\ & & +\left(1|\mathrm{Subject}\right).\hfill \end{array}$$ (2b)

Model (2a) was used when comparing Young vs. Older, with Age group as a factor of 2 levels (young adults, older adults), and Model (2b) was used when comparing PCMS versus SHAM, with Protocol as a factor of 2 levels (PCMS, SHAM). For both models, Time was a factor of 3 levels (start of block 1, end of block 3, and start of 7Dpractice). Finally, for MEP amplitudes, the following generalized linear mixed model was specified:

$$\begin{array}{ccc}& & \mathrm{MEP}(\mathrm{normalized}\mathrm{to}{\mathrm{M}}_{\max }\sim \mathrm{Protocol}\times \mathrm{Time}\hfill \\ & & +\left(1|\mathrm{Subject}\right),\mathrm{family}=\mathrm{gamma}\left(\mathrm{link}=\mathrm{“}\log \mathrm{”}\right),\hfill \end{array}$$ (3)

where Protocol was a factor of 2 levels (SHAM, PCMS) and Time was a factor of 6 levels (baseline, post‐stimulation, post‐practice 0, post‐practice 15 min, post‐practice 30 min, post‐practice 45 min). We also created a model with a combined post‐effect, meaning that Time was a factor of 3 levels (baseline, post‐stimulation, and post‐practice (grand mean)). In this model, we specify that the response variable follows a gamma distribution with a log link function. The log link function was used to ensure that the predictions were positive and to handle the right‐skewness of the data effectively. A linear mixed model was used for M _max and F‐waves, with Time factor of 4 levels (baseline, post‐stimulation, post‐practice 0, and post‐practice 45 min).

Significant main effects or interactions were evaluated using the R‐package lmerTest (Kuznetsova et al., 2017), which computes P values from mixed effect models using the Satterthwaite's degrees of freedom approach. If significant main effects or interactions were observed, pairwise comparisons were performed using the multcomp R‐package (Hothorn et al., 2008). Here we tested our specific hypotheses: we tested whether groups had within‐group improvements between time points. For Models (1a) and (1b), we tested differences between groups at post‐stimulation, immediate retention, and 7Dretention. For models with performance data normalized to baseline values, the between‐group comparisons reflect change scores from baseline (P values were adjusted for three comparisons). Similarly, for Model (2), we tested for differences between groups at the start and end of practice and at the beginning of day 7 practice (P values were adjusted for three comparisons). For Model (3), we tested the difference between group differences in MEP amplitudes at different time points relative to baseline (P values were adjusted for five comparisons).

These pairwise comparisons are presented as model estimates with their corresponding standard errors (SE) and confidence intervals (CI). For all pairwise comparisons, we used the Holm‐Sidak method to adjust for multiple statistical comparisons. For all statistical analyses, the significance level was set at P < 0.05. The mean (M) and standard deviation (SD) are shown when raw data are presented.

Results

Older adults demonstrate inferior ballistic performance, smaller improvements during practice, and lower retention compared to young adults

In the first part of the study, we compared ballistic motor performance and learning between Older (aged 65–75 years) and Young (aged 20–30 years) adults (Fig. 2). Analysis of absolute peak acceleration values using a linear mixed effects model (LMM) revealed significant main effects of Age group (F _(1,40) = 40.6, P < 0.001), Time (F _(3,1425) = 150.5, P < 0.001) and the interaction effect between age group and Time (F _(3,1425) = 46.9, P < 0.001), indicating that ballistic motor performance evolved differently between age groups over time. Post hoc comparisons showed that at baseline, older adults displayed significantly lower ballistic motor performance than young adults (Young: M = 2.34g (0.50) vs. Older: M = 1.51g (0.56), P < 0.001), and this difference remained significant at immediate retention (P < 0.001) and 7‐day retention (P < 0.001). Within‐group comparisons showed that the young group improved significantly from baseline to all time points (all P values <0.01), whereas the older adults displayed a significant drop from baseline to PS (P = 0.024) (Fig. 2B ). Analysing the practice trials where augmented feedback was presented after each trial showed that both the young and older adults improved performance with practice (start of practice vs. end of practice: all P values <0.01) and lasting 1 week (start of practice vs. start of 7‐day practice: all P values <0.01), with older adults performing lower than the young adults at every time point (Fig. 2C ).

Figure 2. Age‐related effects on ballistic motor performance and learning.

A, learning curves displaying the absolute difference between older (aged 65–75 years, n = 20) (orange) and young (aged 20–30 years, n = 20) (green) participants. Motor performance was quantified as the peak acceleration (g). Each data point reflects the average of ten trials. Error bars indicate standard deviation (Test trials = circular data points; Practice trials = triangular data points). Motor test trials were performed without the provision of augmented feedback both before and after stimulations and again after the three motor practice blocks (immediate retention: IR). A delayed retention motor test was performed after 1 week (7R), followed by an extra practice block (7D practice) to assess whether the performance was saturated. B, individual data: Motor test trials showing absolute performance for both age groups at baseline, after SHAM stimulation, IR, and 7R. Box plots show the median as the midline, the box bounds the 25th and 75th quartiles, and the whiskers bound the minimum and maximum values (same box plot definition in C, E, F), single data points represent individual means, and black traces are the group mean. C, individual data: Practice trials from the start of practice block 1 (start of B1) to the end of practice block 3 (end of B3) to the start of the practice block on day 7. D, relative changes in motor performance, learning curves shown as normalized performance (% of baseline) for older and young participants, and error bars indicating standard deviation. E, individual data: motor test trials showing the relative change in motor performance for both age groups after SHAM stimulations, IR and 7R. F, individual data: Practice trials, showing the relative changes in motor performance improvements from the start of practice block 1 (start of B1) to the end of practice block 3 (end of B3) and to the start of the practice block on day 7. ^*Significant within‐group difference from the first time point at P < 0.05. ^#Significant between‐group difference at P < 0.05.

To further examine learning‐related changes, we analysed relative peak acceleration (% of baseline), which also demonstrated significant main effects of Age group (F _(1,40) = 9.29, P < 0.01), Time (F _(2,1035) = 156.9, P < 0.001), and the interaction effect between Age group and Time (F _(2,1035) = 14.95, P < 0.001). Post hoc comparisons revealed no significant difference in test performance immediately after SHAM (PS) (Older: 89.5% ± 3.45, Young: 96.5% ± 3.35, P = 0.14) (Fig. 2D and  E ). Both older and young adults improved relatively with practice from post‐stimulation to immediate retention (Young: +25.1%, P < 0.001; and Older: +14.5%, P < 0.01), with older adults showing lower retention. Statistically, this was evident as significantly lower baseline‐normalized peak accelerations at immediate retention compared to young (Older: 104.0% ± 2.8 vs. Young: 121.8% ± 2.6: −17.6% ± 5.5 CI:[4.56:30.62], P = 0.01) (Fig. 2D ). This effect was robust over time: performance in the older group was also inferior compared to the young group after 1 week (Older: 100.5%±2.8 vs. Young: 118.3%±2.7: −18.68% ± 7.28 CI:[1.51: 35.8], P = 0.02 (Fig. 2E ). The 7‐day retention effect was also observed for the practice trials with augmented feedback (Fig. 2F ).

PCMS before motor practice improves ballistic motor learning in older adults

In the second part of the study, we specifically focused on assessing the effects of PCMS on ballistic motor performance and learning in older adults. No statistical differences in any of the participant characteristics were observed at baseline between the PCMS and SHAM groups (Table 1). Furthermore, no within‐group difference was observed in self‐reported hours slept the night before the main experiment compared with self‐reported hours slept before the retention test 1 week later (F _(1,70) = 0.02, P = 0.9). There were also no between‐group differences in these measures on either day (F _(1,70) = 1.6, P = 0.2). The measured latencies and calculated conduction times used to inform the individualized PCMS and SHAM protocols are presented in Table 2. Unpaired t tests revealed no between‐group differences in the measured latencies.

Individual traces of peak acceleration and EMG signals during the test trials in the ballistic motor task are shown in Fig. 3A . Analysis of absolute peak acceleration values using a linear mixed effect models revealed a significant main effect of Time (F _(3,1361) = 76.0, P < 0.001), a non‐significant main effect of Group (F _(1,39) = 1.7, P = 0.19), and a significant interaction effect of Group and Time (F _(3,1361) = 5.9, P < 0.01). Baseline performance, as measured by raw values, was comparable between the groups (PCMS: M = 1.6 g, SD = 0.47 vs. SHAM: M = 1.5g, SD = 0.58, P = 0.44). PCMS before motor practice had a positive effect on learning, as assessed in later retention tests, compared to SHAM. This difference seemed to evolve through the last practice block, with clear differences in the immediate retention test and the 7‐day retention test (see the learning curve in Fig. 3B ).

Figure 3. Effects of PCMS and SHAM on ballistic motor learning.

A, illustration of single subject traces of acceleration and EMG from separate participants from PCMS or SHAM: At baseline, immediate retention, and 7‐day retention, trials are shown in a 200 ms time window, starting 50 ms before EMG burst. B, learning curves from the PCMS (n = 20) and SHAM (n = 20) groups. Motor performance was quantified as peak acceleration as a percentage of the baseline. Each data point reflects the average of ten trials, and the error bars indicate the standard deviation. Motor test trials (circular data points) were performed without the provision of augmented feedback both before and after stimulations and again after the three motor practice blocks (Block1–Block3). During practice (triangular data points), feedback (knowledge of performance) was provided after each trial. A retention motor test was performed after 1 week (7R), followed by an extra practice block (7D practice) to assess whether the performance was saturated. C, motor test trials: between‐protocol differences in peak acceleration (unit = g) immediately after PCMS or SHAM (post‐stimulation), after motor practice (immediate retention), and 1 week later (7‐day retention). Box plots show the median as the midline, the box bounds the 25th and 75th quartiles, and the whiskers mark the minimum and maximum values. Single data points represent individual means and the black line represents group means. D, feedback trials: between‐protocol differences between the first 10 trials of practice block 1, the last 10 trials of practice block 3, and the first 10 trials of the additional practice block on day 7. Box plots show the median as the midline, the box bounds the 25th and 75th quartiles, and the whiskers mark the minimum and maximum values. Single data points represent individual means and black lines represent group means. E, similar to C but with motor test trials presented relative to baseline (peak acceleration % of baseline). F, similar to D but with feedback trials presented relative to baseline (peak acceleration % of baseline). ^*Significant within‐group difference from the first time point at P < 0.05. ^#Significant between‐protocol difference at a given time point, at P < 0.05. ^†Significant between‐protocol difference in performance change from the start of practice to the end of practice, at P < 0.05.

Learning‐related changes were analysed with a LMM setup for the relative peak acceleration (% of baseline), which showed a significant main effect of Time (F _(2,1016) = 93.6, P < 0.001), a non‐significant main effect of Group (F _(1,39) = 2.7, P = 0.10), and a significant interaction effect of Group and Time (F _(2,1016) = 5.39, P < 0.01). Post hoc comparisons showed that both groups displayed a drop in test performance (% of baseline) immediately after PCMS or SHAM (PS) (SHAM: 89.5% ± 4.35; PCMS: 95.1% ± 4.27), which was not statistically different between groups (P = 0.37) (Fig. 3E ). PCMS interacted positively with motor practice signifying a larger performance increment with practice for the PCMS group. Statistically, this was evident as significantly higher baseline‐normalized peak accelerations at immediate retention compared to SHAM (PCMS: 113.5% ± 4.2 vs. SHAM: 104.0% ± 4.3: +9.5 ± 6.06, CI:[−2.3; 21.4], P = 0.048) (Fig. 3E ). Importantly, this effect was robust over time: performance in the PCMS group was also superior to SHAM after 1 week (PCMS: 115.1% ± 1.9 vs. SHAM: 100.6% ± 2.1: +14.7 ± 6.0, CI:[5.39; 24.01], P = 0.018) (Fig. 3E ).

Participants practiced the ballistic motor task for three blocks of each 50 trials with augmented feedback on performance after the PCMS or SHAM intervention. The linear mixed model on the practice trials showed a significant main effect of Time (F _(2,1066) = 160.1, P < 0.001), a non‐significant main effect of Group (F _(1,39) = 2.49, P = 0.12), and a significant interaction effect of Group and Time (F _(2,1066) = 8.0, P < 0.001). As expected, both groups showed improved performance during practice relative to baseline (Fig. 3B , practice blocks). This was driven by the PCMS group displaying greater performance from the beginning of the additional practice block placed after the 7‐day retention test (PCMS: 126.8% ± 4.1 vs. SHAM: 112.9% ± 4.2: +13.9 ± 5.9, CI:[2.27; 25.67], P = 0.019) (Fig. 3F ). This shows that the effect of PCMS was still observed during continued practice after 1 week, indicating positive interactions across practice sessions. No significant differences were observed between the PCMS and SHAM groups in the practice blocks performed during the main experiment (all P values > 0.05), although PCMS improved performance more from the beginning of practice to the end of practice compared with SHAM (+9.1% ± 3.0, CI[3.2; 15.1], P < 0.01) (Fig. 3F ).

In addition to finger flexion acceleration, we also tested parallel changes in the EMG_RMS amplitude (see Fig. 3A for rectified EMG during trials). For EMG_RMS amplitude, linear mixed effect models showed a significant main effect of Time (F _(3,1362) = 46.9, P < 0.001), a non‐significant main effect of Group (F _(1,39) = 2.0, P = 0.16), and a significant interaction of Group and Time (F _(3,1362) = 2.8, P = 0.03). Both groups displayed a significant increase in EMG_RMS amplitude after motor practice compared to baseline (PCMS: P < 0.001, SHAM: P < 0.001); however, this was significantly greater in the PCMS group than in the SHAM group (PCMS: 137.0% ± 6.1 vs. SHAM: 120.6% ± 6.0: +16.3 ± 6.25, CI:[1.69; 31.0], P = 0.02).

Effects of PCMS and motor practice on corticospinal excitability

We assessed corticospinal excitability using peak‐to‐peak amplitudes of motor evoked potentials recorded with single‐pulse TMS (see Fig. 4A for exemplary averaged MEP traces). PCMS and motor practice had a facilitating effect on MEPs that lasted for at least an hour (Fig. 4B and  C ). Two blocks of MEPs were recorded at baseline, and no significant difference was observed between the two baseline blocks (P = 0.74). For the remaining analyses, we combined the two baseline recordings into one block. The generalized linear mixed model with all time points included showed in summary that likelihood‐ratio tests did not reveal a significant main effect of Protocol (χ²(1)  =  0.867; P  =  0.35) but did reveal a significant main effect of Time (χ²(5)  =  17.47; P < 0.01). Moreover, the likelihood‐ratio test revealed a significant interaction between Protocol and Time (χ²(5)  =  23.0; P < 0.001), suggesting that the effect of the protocol varied over time. The model with three time points (baseline, PS, and post‐grand mean) showed the following: likelihood‐ratio tests did not reveal a significant main effect of Protocol (χ²(1)  =  0.867; P  =  0.35), but did reveal a significant main effect of Time (χ²(2)  =  13.50; P < 0.01). Moreover, the likelihood‐ratio test revealed a significant interaction between Protocol and Time (χ²(2)  =  16.23.0; P < 0.001), suggesting that the effect of the protocol varied over time. Immediately after the intervention (PS), the PCMS group displayed increased MEP amplitudes as a percentage of baseline, whereas a decrease was observed in the SHAM group (PCMS: 170.9 ± 23.1% vs. SHAM: 81.4 ± 24.3%: +89.6 ± 33.5, CI:[23.8; 155.2], P = 0.043) (Fig. 4D and  E ). We observed that 14 of 20 participants in the PCMS group displayed acute MEP facilitation after PCMS, whereas only one participant displayed facilitated MEPs after SHAM stimulation (Fig. 4E ). The facilitated MEPs (perccentage of baseline) after PCMS were still present after motor practice and lasted for 45 min after motor practice. The difference in the grand average of the post‐practice MEP measurements was not statistically significant after PCMS compared with SHAM (PCMS: 188.4 ± 17.3% vs. SHAM: 150.9 ± 18.1%: +37.4 ± 25.9, CI: [−11.6; 86.5], P = 0.38) (Fig. 4D ). We observed that 17 out of 20 participants in the PCMS group displayed a continued increase in MEP (percentage of baseline) across the post‐practice period, compared to 12 after SHAM (Fig. 4E ).

Figure 4. Effects of PCMS and SHAM on corticospinal excitability.

A, illustration of exemplary single subject MEPs. The traces shown are averaged across 20 MEPs at each time point. B, individual trajectories and group‐averaged raw MEP amplitudes (measured in mV) (black) from PCMS (n = 20) and SHAM (n = 18), including the two baseline measurements (Baseline1 and Baseline2), post‐stimulation (PS), post‐motor practice (post‐practice 0, (PP0)), and the following post‐practice measurements PP15, PP30 and PP45 minutes. C, time course of group‐averaged raw MEP amplitudes (measured in mV) from PCMS (n = 20) and SHAM (n = 18) at all time points. Error bars represent standard deviation. D, box plots of MEP amplitudes (in percentage of M _max amplitude), comparing PCMS and SHAM at baseline, post‐stimulation, and post‐practice (grand mean of PP0–PP45). Box plots show the median as the midline, the box bounds the 25th and 75th quartiles, and whiskers bound the minimum and maximum values. Single data points represent individual means, and the black ‘diamond’ represents the group mean. E, individual responses (MEP % baseline) to PCMS or SHAM (the black line is the mean across participants). In D, # indicates significant between‐protocol difference (P < 0.05); in E, * indicates significant within‐protocol comparisons (P < 0.05) between time points relative to baseline.

Effect of PCMS and motor practice on maximal compound muscle action potential (M _max) and α‐motoneuronal excitability (F‐wave amplitudes)

F‐waves were recorded to indirectly quantify α‐motoneuron excitability (see Fig. 5A for individually averaged F‐wave traces). Both F‐wave and MEP amplitudes were normalized to M _max. M _max remained stable throughout the experimental session and we observed no significant interaction effect of Group and Time (F _(3,111) = 0.6, P = 0.61) or any significant main effect of Group (F _(1,38) = 0.18, P = 0.66) or Time (F _(3,111) = 2.7, P = 0.051) (Fig. 5B ). For F‐wave amplitude, the linear mixed effect model showed no significant interaction between Group and Time (F _(3,109) = 0.6, P = 0.63) or main effect of Group (F _(1,38) = 0.5, P = 0.46), but did show a significant main effect of Time (F _(3,109) = 2.73, P = 0.047) (Fig. 5C ). Furthermore, we observed no significant interaction effect between Group and Time for F‐wave persistence (F _(3,111) = 1.4, P = 0.24) or main effect of Time (F _(3,111) = 1.97, P = 0.12), but did see a significant main effect of Group (F _(1,38) = 5.0, P = 0.03) (Fig. 5 D).

Figure 5. The effect of PCMS or SHAM on α‐motoneuron excitability quantified as peak‐to‐peak F‐wave amplitudes.

A, single‐subject averaged traces of M _max and F‐waves (average of 60 frames). The y‐axis is magnified to focus on the averaged F‐wave.). B, maximal M‐wave amplitudes (mV) in both the PCMS (n = 20) and SHAM groups (n = 18). C and D, F‐wave amplitude (mV; C) and F‐wave persistence (% of F‐waves above 0.045 mV; D) at baseline, post‐stimulation (PS), post‐motor practice (PP0), and post‐practice 45 min (PP45). All box plots in B, C and D show the median as the midline, the box bounds the 25th and 75th quartiles and the whiskers bound the minimum and maximum values, single data points represent individual means, and the black ‘diamond’ represent group means.

Discussion

In two coherent experiments, the present study is the first to demonstrate that late adulthood is characterized by an impeded ability to perform ballistic movements, but also the ability to improve performance with practice. Subsequently, the study shows that a single session of PCMS adjusted to facilitate corticospinal‐motoneuronal effective connectivity primes subsequent ballistic motor learning and reinstates long‐term retention in older adults. Supporting the behavioural results, the study found acute increases in corticospinal excitability but not α‐motoneuronal excitability, which together may be the result of increased connectivity at the CM synapses. The results indicate the relevance of interventions with the potential to promote neuroplasticity and the effects of motor practice in older adults.

Reduced ability to execute and learn ballistic movements in late adulthood

First, this study demonstrated that age affects ballistic motor performance and both short‐ and long‐term retention, as indicated by both absolute and relative performance changes. This age‐related difference in initial ballistic performance was expected based on previous research and can probably be attributed to decreased supraspinal and afferent drive, along with decreased excitability of the motoneurons (Clark & Taylor, 2012; Shaffer & Harrison, 2007). In accordance with this notion, late adulthood has been demonstrated to be characterized by a right shift in TMS recruitment curves (Eisen et al., 1991; Pitcher et al., 2003; Sale & Semmler, 2005). The reduced maximal discharge rates of spinal motoneurons caused by these age‐related changes (Klass et al., 2008) impair the ability to perform fast and precise ballistic movements (del Vecchio et al., 2019; Duchateau & Baudry, 2014).

The inclusion of trials with and without augmented feedback allowed us to assess the effects of augmented feedback on both motor performance and learning (Kantak & Winstein, 2012). In addition to ballistic performance, ballistic learning was impaired in older adults. In fact, 8 of the 20 participants performed worse after practice than at baseline. Fatigue could have been a factor, but again, the amount of practice and number of trials were equal between the groups, meaning that the potential role of fatigue was similar between groups. The majority of these individuals (6/8) also displayed an initial decline in performance immediately after SHAM stimulations. Lower performance was observed during the test trials (baseline, post‐stimulation, and retention tests), which were conducted without augmented feedback on performance, measured as a percentage of their individual baseline level (knowledge of results). Importantly, no differences were observed in the magnitude of the drop in performance between the PCMS and SHAM groups; therefore, it is probably not a result of the experimental intervention. A drop in ballistic performance immediately after the SHAM protocol has also been shown in other studies using non‐invasive stimulation techniques to improve motor learning (Yamaguchi et al., 2020). The active stimulations in the PCMS group might have helped maintain participant alertness and focus, potentially mitigating the performance drop observed in the SHAM group. Additionally, warm‐up decrements could potentially explain why some older adults had difficulty transitioning from test to practice trials (Adams, 1961; Etnyre & Poindexter, 1995; Roig et al., 2014).

The drop in performance in trials with augmented feedback to trials without augmented feedback may reflect an increased age‐related dependency on visual feedback during ballistic movements. Age‐related differences related to augmented feedback have been previously reported (Swinnen et al., 1998; Wishart et al., 2002), and this age‐related difference also depends on the motor task used (Voelcker‐Rehage, 2008). Importantly, this observation does not affect the conclusion on the positive effects of the PCMS intervention on motor learning, since PCMS had beneficial effects on motor performance compared to SHAM, both in trials with and without augmented feedback. The present results are consistent with those of previous studies demonstrating diminished ballistic motor learning (Rogasch et al., 2009) and lower delayed retention (Roig et al., 2014) in older adults. Compared to the ballistic task used in the present study, repetitive thumb abduction tasks show larger relative learning effects (Cirillo et al., 2011; Rogasch et al., 2009), which could be due to changes in movement direction (vector), in addition to increased acceleration (Classen et al., 1998). In the behavioural model employed in the current study, movements were restricted to flexion/extension, meaning that directional changes did not contribute to changes in performance. Interestingly, Cirillo et al. (2010) found reduced ballistic learning in older adults, despite changes in corticospinal excitability comparable to those in young adults. It can be speculated that whereas cortical use‐dependent plasticity may be preserved, as reflected in the increased MEP amplitude, the spinal contribution is reduced. Age‐related decreases in spinal excitability (Kido et al., 2004; Scaglioni et al., 2002) and reduced descending inputs to motor neurons have been reported (Eisen et al., 1996). This provides a mechanistic rationale for targeting age‐related declines in the malleability of the spinal circuitry.

PCMS reinstates long‐term retention of ballistic motor learning in late adulthood

In line with previous reports on young adults, the present study reports facilitatory effects of PCMS on subsequent motor learning in neurologically intact older adults (Bjørndal et al., 2024; Urbin et al., 2024). PCMS has previously been shown to enhance submaximal voluntary motor output (D'Amico et al., 2018; Taylor & Martin, 2009; Urbin et al., 2017) and manipulative control of hand function acutely (Bunday & Perez, 2012). Conversely, but in line with previous findings (Bjørndal et al., 2024), this study showed no acute effects of PCMS on ballistic motor performance, which suggests that the effects of PCMS are task‐specific. Specifically, facilitatory PCMS improves dexterous and submaximal motor control, but does not affect maximal acceleration (Bjørndal et al., 2024) or force production (D'Amico et al., 2018). In contrast, the present study found that PCMS influenced the ability to improve ballistic performance with practice (i.e. learning) in older adults, which extends previous work in young adults showing positive effects of PCMS on improving ballistic movements and force control through practice (Bjørndal et al., 2024; Urbin et al., 2024). Importantly, this effect persisted for 7 days, suggesting that PCMS improves long‐term motor learning and thereby retention (Kantak & Winstein, 2012).

Effects of PCMS on corticospinal and α‐motoneuronal excitability in older adults

The superior effect of motor practice combined with PCMS compared to SHAM is probably due to the priming effects of PCMS on the neural circuitry responsible for generating ballistic movements. The present study showed that PCMS led to an acute increase in corticospinal excitability that persisted through motor practice. The facilitating effect of PCMS on corticospinal excitability is in line with previous studies in younger adults (Bjørndal et al., 2024; Bunday & Perez, 2012; Christiansen et al., 2021, 2018; D'Amico et al., 2020; Taylor & Martin, 2009). MEPs are compound potentials reflecting excitability at both cortical and spinal levels, and therefore changes in MEP amplitudes cannot inform one of the exact loci of changes within the corticospinal system. Therefore, while it cannot be excluded that the PCMS protocol could have some cortical effects, the significant improvement in motor learning observed with this protocol suggests that the primary mechanism of action is more likely related to the specific effects of the PCMS protocol targeting the spinal CM level. The interactions between cortical and spinal mechanisms are naturally complex, and the motor task naturally engages cortical areas during voluntary movements and ballistic motor learning. While we cannot exclude the possibility of cortical PCMS effects, the PCMS protocol is timed to interact at the CM level. We recently demonstrated that while PCMS protocols that were not timed to coincide at the CM synapses could lead to MEP facilitation, potentially through trans‐cortical pathways, these protocols did not affect subsequent ballistic motor learning (Bjørndal et al., 2024). In further support of a spinal origin, both low‐frequency rTMS (Taube et al., 2015) and spinal paired associative stimulation (spinal PAS) can induce change at the spinal level (Cortes et al., 2011; Leukel et al., 2012). In spinal PAS an afferent volley from a peripheral nerve stimulation converges with a descending volley elicited either by TMS over the motor cortex or the cervical segments repeatedly leading to either altered spinal (Cortes et al., 2011) or corticospinal (Leukel et al., 2012) excitability. The results support the notion that precisely timed paired stimulation can potentiate spinal synapses although for different stimulation protocols. In this study, we target intrinsic hand muscles, where eliciting H‐reflexes in resting neurological intact humans is notoriously difficult, severely reducing the possibility of changes in the network that mediates the H‐reflex (Pierrot‐Deseilligny & Burke, 2012). Secondly, the parameters of the peripheral nerve stimulation (short pulse duration and supramaximal intensity) do not favour deep afferent excitation over evoking antidromic activity in the motoneurons (Panizza et al., 1989). Third, and most importantly, the principle that the PCMS protocol operates through CM synapses is supported by previous studies demonstrating facilitatory effects on cervicomedullary evoked MEPs (CMEPs) (Bunday & Perez, 2012; Dongés et al., 2018; Fitzpatrick et al., 2016; Taylor & Martin, 2009). Post‐stimulus time histograms following cervicomedullary stimulation reveal a very narrow peak, indicative of only one synapse between the site of stimulation and recording, i.e. corticomotoneuronal synapses. Furthermore, similar to other studies (Bjørndal et al., 2024; Bunday & Perez, 2012), the present study observed no changes in F‐wave amplitudes. Notably, the two groups differed in their F‐wave persistence, as indicated by the main effect of group. The non‐significant interaction effects suggests that the observed changes in MEP were not due to changes in the intrinsic excitability of α‐motoneurons. Given all of the above, we propose that the priming effects of facilitatory PCMS adjusted to an inter‐arrival‐interval of 2 ms as in the present study are most likely mediated by changes in CM connectivity.

Modulating neuroplasticity in the ageing brain

Late adulthood brings about not only changes in the central nervous system, which is important for control of movement, but also in the adaptability of spinal and supraspinal circuits underpinning motor control. It is important to note that, whereas young adults appear to increase acceleration with ballistic training through increased drive of agonist motoneurons, older adults improve to a lesser extent and without evident changes in agonist EMG activity (Rogasch et al., 2009). As age‐related declines in motor learning appear to mainly affect tasks known to, at least in part, depend on changes in corticomotor effective connectivity (Giesebrecht et al., 2012), CM synapses are a promising target for neuromodulation in late adulthood.

Hebbian learning rules do not readily extend from youth to late adulthood. Ageing has been demonstrated to reduce or even revert after‐effects following neuromodulation targeting sensorimotor integration at the cortical level through Hebbian principles (i.e. PAS protocols) (Müller‐Dahlhaus et al., 2008; Opie et al., 2017; Tecchio et al., 2008). Previous studies have shown that PAS‐induced plasticity is lower in older adults than in younger adults (Opie et al., 2017). Opie et al. (2019); Tecchio et al., 2008 demonstrated that both facilitatory and inhibitory PAS before motor practice of a visuomotor skill decreased skill acquisition in older adults. Despite the negative effects on skill learning, this study showed that PAS can interact with subsequent practice‐dependent plasticity. In contrast to cortically targeted neuromodulatory stimulation protocols such as PAS, the present study demonstrated a relatively high number of responders to PCMS based on corticospinal excitability. This aligns with prior research focused on FDI and using PCMS with an inter‐arrival interval of 2 ms between pre‐ and post‐synapse (Bunday & Perez, 2012; Bunday et al., 2018). The targeting of CM‐synapses with PCMS may represent a potentially more effective method for enhancing neuroplasticity in older adults compared to cortically targeted neuromodulatory stimulation protocols.

Limitations

In this study, we only investigated the effects of a facilitatory protocol adjusted to ensure that presynaptic activation preceded postsynaptic activation. In young adults, we have previously demonstrated that an intended inhibitory PCMS, where postsynaptic activation precedes presynaptic activation, transiently reduces performance increments during practice (Bjørndal et al., 2024). Based on the present results, it is not possible to conclude whether a similar bidirectional Hebbian learning rule applies to older adults. However, this limitation is mainly of conceptual importance, as we are not aware of any neurological conditions that might benefit from delayed or reduced learning. We acknowledge that the difference in stimulation intensities between the real PCMS and SHAM conditions may be noticeable, particularly if the participants had prior experience with similar protocols involving TMS or PNS. However, none of the participants had prior experience with similar protocols, and the between‐group design of our study ensured that each participant was only exposed to either the PCMS or SHAM protocol, thereby minimizing this potential bias. This study included able‐bodied and well‐functioning older adults without neurological conditions. Generalization of results to other groups of individuals should therefore be performed with caution. Additionally, the study used a model for ballistic learning, in which the movement of the targeted limb is isolated. Future work is needed to explore whether the beneficial effects of PCMS in older adults extend to more unrestricted motor tasks and activities of daily life.

Conclusions

This study demonstrated an age‐related effect on ballistic motor performance and learning, with older adults showing lower baseline performance and smaller improvement after practice compared to young adults. Secondly, using a proof‐of‐principle approach, the study investigated the effects of PCMS on ballistic motor learning in older adults. The study showed that PCMS, but not a SHAM protocol, before ballistic motor practice resulted in improved learning. The results broaden the potential applicability of this technique in improving long‐term motor learning in older individuals, where age‐related changes in the central nervous system are associated with a decline in motor function. Other studies are beginning to investigate the effects of PCMS in combination with exercise and rehabilitation (Jo et al., 2023). Our results support the idea that combining PCMS with manual therapy and exercise could present a compelling approach to improving the effectiveness of rehabilitation.

Competing interests

The authors declare that they have no actual or potential conflicts of interest.

Author contributions

The experiments were performed at the Movement and Neuroscience Lab of the Department of Nutrition, Exercise, and Sports at the University of Copenhagen, Denmark. J.L.J., M.M.B., L.C., and J.R.B. conceptualized and designed the research. J.R.B. and L.J. performed the experiments. J.R.B. and L.J. analysed the data. J.R.B., L.J., M.M.B., A.K., L.C., and J.L.J. interpreted the results. J.R.B. drafted the manuscript. J.R.B., L.J., M.M.B., A.K., L.C. and J.L.J. edited and revised the manuscript. J.LJ. and A.K. supervised the work. All authors read and approved the final version submitted for publication. J.L.J. acquired the financial support for the project. All persons designated as authors qualify for authorship, and all those who qualify for authorship are listed.

Funding

We are grateful for the financial support from Nordea‐fonden (Grant nr. 02‐2019‐00045), and Danish Ministry of Culture (Grant nr. FPK.2018‐0070). Mikkel M. Beck is funded by a postdoc grant from the Research fund of the Capital Region Denmark (Region H). Anke Ninija Karabanov is supported by a Sapere Aude Statring Grant from Danmarks Frie Forskningsfond (Grant no. 0169‐00027B). Lasse Christiansen holds a postdoc grant from the Lundbeck Foundation (Grant no. R322‐2019‐2406).

Supporting information

Peer Review History

Biography

Jonas Rud Bjørndal is a PhD student in the Movement and Neuroscience Section at the Department of Nutrition, Exercise and Sports (University of Copenhagen, Denmark). His research is focused on motor learning and neurophysiology.

Data availability statement

The data that support the findings of this study are available from the corresponding author upon reasonable request.