Repetitive transcranial magnetic stimulation (rTMS) is a promising neuromodulatory therapy for post-stroke rehabilitation, offering the potential to modulate brain network dynamics. However, its clinical effectiveness remains debated due to the high inter-individual variability in patient responses, largely attributed to heterogeneous patterns of neural reorganization after stroke. In this study, we proposed an EEG ensemble learning framework (EEG-ELF) to predict motor recovery outcomes following rTMS therapy in stroke patients. The framework was evaluated on EEG recordings acquired during a motor imagery task from 20 individuals with subcortical stroke, collected both before and after a four-week rTMS intervention, along with the upper-limb Fugl-Meyer Assessment (FMA) scores. EEG-ELF's prediction performance was assessed through both regression and classification tasks using leave-one-subject-out cross-validation. The results demonstrated a strong correlation between predicted and actual recovery rates (Pearson's R = 0.92, p $< $ 0.001), and the framework achieved 95% accuracy in classifying patients into good and poor prognosis groups. These findings suggest that EEG-ELF can accurately predict individual motor outcomes from pre-treatment EEG features, highlighting its potential to guide personalized rTMS interventions through data-driven stratification of stroke patients.

Ageing induces structural and functional changes in the neuromuscular systems that impair voluntary force production, compromising daily function and wellbeing. We examined whether older adults preserve the capacity for motor unit adaptations to a short-term strength training intervention previously shown to enhance neural drive to muscle in young adults. Twenty‑three older adults were assigned to a training group (INT, n = 13, 71 ± 4 years of age) or a control group (CON, n = 10, 69 ± 2 years of age) and completed pre- and postintervention assessments of ankle dorsiflexor maximal voluntary force (MVF). Motor unit behaviour was analysed from high‑density surface EMG recorded from tibialis anterior during submaximal trapezoidal contractions. The INT group performed a 4 week supervised isometric strength training programme, whereas the CON group maintained habitual activity. High‑density surface EMG signals were decomposed into individual motor units, tracked longitudinally across sessions. Training increased MVF by 17.6% and enhanced motor unit discharge rate at recruitment (+8.2%, P = 0.031) and constant force (+11.3%, P < 0.001), without changes in recruitment or derecruitment thresholds. Estimates of persistent inward currents (delta frequency) increased (+1.0 pulses per second) and were positively correlated with changes in discharge rate, which, in turn, were correlated with gains in MVF (rrm = 0.54-0.57, P < 0.05). This pattern suggests that enhanced intrinsic excitability and synaptic input to motor neurons contributed to improvements in strength. These results demonstrate that, despite age-related motor unit remodelling, the ageing nervous system remains responsive to targeted strength training, preserving the capacity for meaningful neural adaptations. KEY POINTS: We assessed whether a short-term intensive strength training intervention, previously shown to increase spinal motor output to the muscle significantly in young adults, would also be effective in older adults. High-density surface EMG was used to identify and longitudinally track the same motor units before and after a 4 week isometric strength training intervention. We found significant strength gains in older adults, with the increase in muscle force output being positively associated with higher motor unit discharge rate and persistent inward currents, indicating that neural drive enhancement was a key contributor to the observed improvements in force. Despite age-related motor neuron remodelling, the older nervous system remains highly responsive to strength training, exhibiting qualitatively similar but attenuated motor unit adaptations compared with young adults.

Spinal motor neurons serve as the link between the nervous system and muscles. As the final common pathway of the neuromuscular system, they receive inputs from both higher-level controllers and afferent pathways. It is often assumed that spinal motor neurons are primarily driven by continuous common inputs (cCI) within different frequency bands. Within this framework, the motor neuron pool behaves as a linear amplifier of the cCI. This implies that the frequency content of descending and spinal oscillatory signals is preserved and faithfully transmitted to the muscles; thus, the spectral content at the output of the motor neuron pool corresponds to that of the cCI. However, this framework overlooks the possibility that motor neurons could also be driven by impulsive common inputs (iCI), which can induce synchronization among them and disrupt the linear transmission of other synaptic inputs at the pool level. To test this hypothesis, computational simulations and experimental data from two different human muscles were used to characterize different aspects related to motor neuron spiking synchronization at the pool level. Our findings suggest that, indeed, iCI can account for relevant features observed in experimental data such as the presence of synchronization events at the pool level. We also observed that such impulsive inputs can affect the linearity in the transmission of cCI by the motor neuron pool. This study represents pioneering indirect evidence of the existence of iCI as inputs to motor neurons.Key points The current understanding of the motor control of voluntary movements assumes a continuous control, driven by oscillatory common signals. Some aspects of motor unit pool behaviour (particularly in terms of spiking synchronization and spectral content) typically observed in experimental recordings cannot be reproduced in simulations that only use continuous common inputs (cCI) to motor neurons. This study provides evidence indicating that spinal motor neurons receive a portion of their synaptic input in the form of impulsive common inputs (iCI) that synchronize their activity. The study also shows how such iCI can affect the linear transmission of other cCI by the motor neuron pool. These findings constitute a fundamental paradigm shift in the understanding of motor control and impact the development of interfaces that extract information from the activity of spinal motor neurons.

This paper presents Wireless Wearable Myolink (W2 Myolink), a high-density, battery-powered system for wireless acquisition of surface electromyography(EMG) signals, and demonstrates its practicality across diverse experimentalparadigms. The device features a compact form factor (110x70x35 mm), low weight(160 g), and continuous operation for up to 6 h. It supports low-noise (1.34 µVRMS),high-resolution (24 bit) recordings from 128 EMG channels with real-time wireless datatransmission over Wi-Fi. The capabilities of W2 Myolink have been validated througha series of experiments. Conventional EMG studies included the recording and analysisof muscle activity during various gestures and hand movements, as well as concurrentEMG–EEG measurements for the investigation of cortico-muscular coherence. Thesystem’s suitability for clinical and pathological scenarios was examined using surrogateEMG signals representative of healthy subjects and of conditions such as myopathy,neuropathy, and Amyotrophic Lateral Sclerosis (ALS). More complex experimental usecases involved human–robot interaction with ISYBOT robotic arms, intraoperativeEMG monitoring during different types of surgery with and without exoskeletonassistance for the surgeon, and EMG acquisition during psychedelic experiencesinduced by psilocybin. Collectively, these results demonstrate that W2 Myolink canreliably acquire and wirelessly transmit high-quality, high-density EMG data in a broadrange of research and application contexts.

Neuromuscular compartments provide the functional substrate for finger‑specific motor control, yet their in vivo three‑dimensional (3D) architecture remains unresolved because of the absence of clear fascial landmarks for imaging. As a representative example, in the flexor digitorum superficialis (FDS), the primary finger flexor, this lack of compartment‑resolved anatomy and fiber trajectories limits our understanding of the neuromechanical control of hand movements and prevents realistic finger‑specific computational models. Here, we present a multimodal imaging framework that combines ultrasound and magnetic resonance imaging (MRI) to resolve this gap by providing both volumetric compartment segmentation and in vivo architectural reconstruction. During isolated finger flexion, tissue movement directions estimated from dynamic ultrasound delineate compartment contours, which are mapped into a 3D MRI reference to generate volumetric compartment masks. Using these masks as constraints, tractography reconstructs 3D fiber trajectories and yields compartment-specific architectural measurements. When applied to ten healthy individuals, the framework yielded compartment topology and fiber orientations consistent with classic cadaveric anatomy. Moreover, the estimated compartments accurately predicted the experimental distribution of surface electromyography amplitude on the skin surface. These results demonstrate that in vivo compartment mapping can simultaneously resolve the missing in vivo anatomy and architecture of the FDS, providing 3D fiber trajectories and quantitative parameters for finger-specific computational models while delivering anatomically grounded definitions of the functional units that underlie the neuromechanical control of the hand. By supplying compartment-level ground truth for both biomechanics and electromyography, this framework establishes a foundation for subject-specific neuromuscular digital twins and next-generation neuromuscular interfaces.

The conventional framework of motor unit (MU) control assumes that MUs in a MU pool are constrained by a fixed recruitment order and a common input. This rigid control framework has been challenged by recent findings suggesting that MU activity could be flexibly modulated, potentially mediated by descending cortical inputs. In this study, rather than evaluating flexibility from the perspective of recruitment thresholds, we investigated control flexibility by assessing whether human participants can voluntarily modulate MU firing rates beyond rigid control constraints. Specifically, we examined whether participants could voluntarily modulate the firing rates of a pair of MUs from the tibialis anterior muscle during real-time feedback. Two tasks involving target reach with different visual feedback derived from the MUs' firing rates were conducted. In both tasks, there was no evidence that participants were able to change MU firing rates in a way that would violate rigid control robustly. Our findings demonstrate limited flexibility in MU control in human tibialis anterior muscle within single-session training, even when real-time MU activity feedback was provided. The results suggest that MU flexibility is not inherently present in the human lower limb.NEW & NOTEWORTHY How flexibly can the central nervous system control individual motor units? We tested whether humans can achieve independent control of pairs of motor units from the tibialis anterior muscle in two experiments with real-time visual feedback of motor unit firing rates. The results of both experiments provided no evidence of independent control.

High-intensity interval training (HIIT) and continuous endurance exercise (END) induce distinct neuromuscular adaptations, with END particularly enhancing fatigue resistance during sustained submaximal contractions. However, the motor unit (MU) mechanisms underlying these effects remain unclear. This study investigated MU firing adaptations associated with changes in time to task failure following END and HIIT. Sixteen healthy men were randomly assigned to END or HIIT (n = 8/group) and completed six sessions over 14 days. HIIT involved 8-12 × 60-s intervals at 100% peak power output, separated by a 75-s recovery; END involved 90-120 min of continuous cycling at ∼65% peak oxygen uptake (V̇o2peak). Before and after training, participants performed a nonfatiguing isometric contraction at 50% maximal voluntary contraction (MVC), followed by a sustained contraction at 30% MVC until failure, while high-density surface EMG signals were recorded from the vasti muscles. Signals were decomposed, and MUs were tracked across sessions. MU firing rates displayed a biphasic response to fatigue: an initial decline (first phase) followed by a later increase (second phase). Postintervention, only the END group increased time-to-task failure and delayed the onset of the second phase (P = 0.021), which correlated with time to failure (r = 0.70). The END group also showed less attenuation in firing rate at failure (50% vs. 30% MVC difference: END = 0.56 Hz; HIIT = 2.8 Hz; P = 0.011), which was also associated with total endurance time (r = 0.72). These findings suggest that END-induced fatigue resistance is associated with specific MU firing adaptations that enhance the central nervous system's ability to optimise MU recruitment and firing dynamics during fatigue development.NEW & NOTEWORTHY Two weeks of endurance training (END), but not high-intensity interval training (HIIT), prolonged time to task failure during sustained submaximal contractions. This improvement in performance was linked to distinct motor unit adaptations, delayed discharge rate increase, and reduced firing rate attenuation that explained ∼50% of performance gains. Short-term END thus elicits unique neuromuscular changes that enhance fatigue resistance at low-to-moderate intensities.

Soft robotic suits have the potential to rehabilitate, assist, and augment the human body. The low weight, cost, and minimal form-factor of these devices make them ideal for daily use by both healthy and impaired individuals. However, challenges associated with data-driven, user-specific, and comfort-first design of human-robot interfaces using soft materials limit their widespread translation and adoption. In this work, we present the quantitative evaluation of ergonomics and comfort of the Elevate suit - a cable driven soft robotic suit that assists shoulder elevation. Using a motion-capture system and force sensors, we measured the suit's ergonomics during assisted shoulder elevation up to 70 degrees. Two 4-hour sessions were conducted with one subject, involving transmitting cable tensions of up to 200N with no discomfort reported. We estimated that the pressure applied to the shoulder during assisted movements was within the range seen in a human grasp (approximately 69.1-85.1kPa), and estimated volumetric compression of <3

Restoring sensory function post amputation remains a major challenge. Peripheral nerve stimulation and targeted reinnervation may partially restore somatotopic feedback, but their need for surgery hinders widespread adoption. Here, we investigate the feasibility of transcutaneous spinal cord stimulation (tSCS) as a non-invasive approach for sensory restoration in upper-limb amputees. In a study involving seventeen able-bodied participants and five individuals with upper-limb amputation, we show that tSCS can evoke a range of sensations, including touch, tapping, vibration, and movement, perceived as originating from the missing limb. Notably, these perceptions were primarily isolated to the missing limb and absent in the residual limb in 98% of trials. Participants with amputations found tSCS tolerable, with some reporting increased comfort during stimulation. tSCS evoked sensations in the fingertips of 93% of able-bodied participants, though these were mainly paraesthetic. We further characterised how stimulation parameters, including electrode placement, carrier frequency, and burst frequency, modulated the quality and type of perceived sensations. Additionally, we show that tSCS maintained force proprioception necessary for effective prosthesis control. These findings support the potential of tSCS as a non-invasive sensory feedback approach for upper-limb prosthesis users.

Individuals suffering from progressive neuromuscular diseases gradually lose all muscle control and therefore are forced to repeatedly adapt to new control interface technologies to maintain some level of independence. Consequently, ideal interface technology should adapt to the progression of paralysis. We propose an adaptive tongue-brain hybrid interface framework for the three-dimensional control of a robotic arm. The interface was tested with able-bodied individuals and individuals with amyotrophic lateral sclerosis. The experiments demonstrated the importance of flexible frameworks for cooperation between control modalities as this allows a critical optimization of the control performance relative to the disease stage. The hybrid framework allowed for a 4-32% stepwise decrease in performance while some tongue-functionality would exists, rather than a 200% decrease when moving directly from a tongue to a brain control interface. This hybrid framework is the first step towards a new concept of assistive robotic control with a higher focus on adapting to the functionality of disabled individuals.

Abstract A motoneuron pool is often regarded as a rigid controller because the largely shared synaptic input across motoneurons leads to strongly correlated activity. However, brief deviations from this correlated behavior have been observed even in some constrained tasks, raising the question of whether these results reflect limitations of the rigid view of motoneuron pool control. Here we show that they do not. We developed a biophysical model of a motoneuron pool receiving shared excitatory and inhibitory synaptic inputs that also included the motoneuron-specific effects of neuromodulation; model parameters were tuned based on large-scale motoneuron recordings in humans. Simulations showed that the intrinsic differences in how motoneurons respond to neuromodulation are both necessary and sufficient to transiently decorrelate pairs of motoneurons receiving a shared synaptic input. Crucially, such transient decorrelation is only observed when motoneurons have different sensitivity to neuromodulation, consistent with experimental observations during volitional control in humans. Our model also explains how participants can improve their ability to transiently decorrelate the activity of motoneurons innervating the same muscle by leveraging refined behavioral strategies that exploit the differential response of motoneurons to neuromodulation, rather than through physiological changes. These results identify that heterogeneous sensitivity to neuromodulation enables brief flexibility in the otherwise rigid control of motoneurons enforced by a shared synaptic input, and show how practice allows participants to exploit latent flexibility within otherwise rigid constraints.

This study evaluated the sensitivity and robustness of a motor unit (MU)–based EEG filtering approach for estimating corticomuscular coherence. The effects of three factors were examined: the number of MUs (No. MUs) used to construct the MU-based EEG filter, the EEG extension factor (FEEG) applied during preprocessing, and muscle contraction level. By applying the MU-based EEG filter to the EEG signals, we obtain an EEG component (EEGcomp) that is coupled to MU activity. Coherence between the EEGcomp and the cumulative spike train was computed and quantified using three metrics: the maximum value of coherence (COHmax), the number of frequency bins exceeding the significance threshold (COHNbin), and the Root Mean Square (RMS) of coherence (COHRMS). Analyses were performed separately in the alpha, beta, and low gamma frequency bands. Across all bands, coherence metrics increased with No. MUs, with the strongest differences observed between small and large No. MUs, and saturation occurring at approximately 10–12 MUs. FEEG had a limited influence on the alpha and beta bands, whereas in the gamma band its effect was stronger, with saturation occurring at relatively high FEEG values (FEEG ≥ 20). Contraction level significantly affected coherence and interacted with No. MUs in the alpha and gamma bands, indicating contraction-dependent modulation of MU-related coherence estimates.

Postural stability during quiet human stance relies heavily on neural feedback control. However, delays in this feedback can markedly reshape closed-loop dynamics and potentially compromise balance stability. Despite this, from a theoretical perspective, the mechanisms by which such delays affect balance stability remain poorly understood. Most existing studies rely on numerical simulations of neuromechanical models of human standing, while rigorous mathematical analysis of the system's dynamic properties is still lacking. To address this gap, this study proposed a frequency-domain analytical framework based on a widely adopted neuromechanical model of human postural control. By transforming the postural control system into the complex frequency domain, we derived analytical solutions and systematically investigated how the system's characteristic roots in the Laplace domain evolved with increasing neural feedback delay. This analysis revealed the mechanisms by which delays induce instability. Furthermore, we derived critical conditions for system destabilization and provided an exact analytical expression for the delay threshold leading to instability. Based on these results, a stability criterion was proposed, providing a theoretical basis for assessing the robustness of postural control. The proposed framework applies to the study of postural stability in populations such as older adults and patients with neurodegenerative diseases. Over all, this research provides a solid theoretical foundation, both qualitative and quantitative, for understanding instability in human postural control induced by varying neural feedback delays.

OBJECTIVE:This study aimed to investigate cortico-spinal interactions during movement execution 'Go' and movement inhibition 'NoGo', with a focus on frequency-specific coupling between cortical activities and spinal motor neuron discharges. METHODS:High-density surface EMG (HD-sEMG) was recorded from the right fore arm muscles in 16 healthy adults, and the cumulative spike train (CST) of the active motor units was estimated using the spatial spike detection (SSD) method. Electroen cephalography (EEG) and CST signals were analyzed in the time-frequency domain, andtheir couplings were assessed with the coherence and cross-correlation functions over three task-related time windows (steady phase, preparation, execution/inhibition). RESULTS:We observed a strong δ-band coupling between EEG and CST during movement execution, a phenomenon absent in the 'NoGo' condition. This effect was most pronounced during the movement execution window and localized at the contralateral senso rimotor cortex. By contrast, high-frequency coherence exhibited lateralization but showed no significant differences among the three task-related time windows in 'Go' and 'NoGo' conditions. CONCLUSION:These findings revealed that δ-band cortico-spinal couplings were specifically as sociated with rapid movement execution, whereas high frequency activities primarily reflected force maintenance. SIGNIFICANCE:By combining EEG and SSD-based CST estimation, this study highlights the critical role of δ-band cortico-spinal coupling in rapid motor control and provide a potential neurophysiological indicator for assessing cortico-spinal tract integrity during functional motor tasks.

The flat, stiff sole of energy-storage-and-return prosthetic feet hinders adaptation to irregular terrains. This is among the causes both of high falling risk and of the consequential arising compensatory mechanisms in prosthetic users. To overcome that, we introduce the SoftFoot Pro, an anthropomorphic and adaptive prosthetic foot featuring a flexible and inextensible sole that passively adapts to obstacles, widening the ground contact area. Experimental comparison to a traditional carbon fibre foot in two unilateral transtibial prosthetic users highlights that the adaptive design reduces stance torque and power consumption at the contralateral knee and at both hips, both on level and uneven grounds. The more even load distribution between the two limbs reduces compensatory strategies and gait asymmetries, resulting in biomechanics closer to that of unimpaired individuals. These findings hold promise for enhancing quality of life for individuals with limb loss, potentially improving stability and reducing fall risk. Irregular terrains challenge users of classic prosthetic feet. Here, the authors introduce an adaptive prosthetic foot that equalizes ground contact pressure and reduces compensatory strategies and gait asymmetries compared with a traditional carbon fibre foot.

Background: We present a miniature (30 × 34 mm) 64-channel data acquisition headstage optimized for high-density surface electromyography. Methods: The headstage is made up of a multi-channel ASIC analogue front-end utilizing only MOS transistors, fabricated in 350 nm CMOS technology (IC die dimensions 6.9 × 1.8 mm), combined with an off-the-shelf multi-channel current-input ADC (DDC264, Texas Instruments). The ASIC analogue front-end employs MOS-based capacitors for both processing and AC-coupling. Results: The combination of these two sub-circuits enables the simultaneous recording of 64 channels at a typical sampling rate of 4 KHz with a maximum analogue bandwidth of 0.5–1500 Hz and a resolution of 20-bits. Typical input-referred-noise, determined by the analogue front-end, is 3.5 μVRMS for a surface EMG bandwidth of interest of 20–500 Hz. This two-chip solution results in a power consumption of 5 mW per channel. Analogue performance variability of the custom ASIC was characterized across a dataset of 960-channels (15 dies) from two fabrication runs. Conclusions: This work practically demonstrates the viability of using both a MOS-only analogue front-end and commercially available off-shelf high-performance back-end hardware already developed for medical imaging applications to record high-density surface biosignals. The aforementioned techniques can be employed to reduce the size and cost for systems or wearable devices; facilitating the translation of high-density bio-acquisition setups from the research environment to more affordable commercial products.

Although tremor is the most common movement disorder, the role of the peripheral musculoskeletal system in shaping tremor is not fully understood. Elucidating how tremorogenic muscle activity propagates through the peripheral musculoskeletal system on its way to becoming tremor at the hand is important for understanding the effects of (and improving) tremor-suppression strategies.We present the first model of tremor propagation throughout the upper limb from tremorogenic muscle activity to hand tremor. Using this linear, time-invariant multi-input multi-output model, we simulated the hand tremor caused by all 50 upper-limb muscles (excluding intrinsic hand muscles) individually and collectively, across seven postures representative of daily activities and clinical assessments. To ensure robustness, Monte Carlo simulations were repeated at many different input and model parameter values.Modeling revealed that to the extent that the musculoskeletal system of the upper limb can be approximated as linear and time-invariant during postural tremor, single-frequency tremorogenic drive to any number of muscles causes the hand to trace an elliptical path once steady state is reached. Each muscle is associated with a tremor ellipse, characterized in terms of direction and magnitude. Muscles acting on the same degrees of freedom (whether synergists or antagonists) tended to produce tremor in similar directions. Tremor direction varied significantly with posture, but taken together, the individual tremor ellipses formed a relatively flat ellipsoid whose dominant plane remained nearly perpendicular to the long axis of the forearm and hand. According to our simulations across a variety of postures, the peripheral musculoskeletal system generally amplified tremorogenic input to distal muscles (particularly wrist muscles) more than other muscles.These results highlight the critical role of peripheral biomechanics in shaping hand tremor. Tremor ellipses offer a useful framework for understanding the first-order effects of the musculoskeletal system on tremor and for estimating the potential of individual muscles to contribute to hand tremor.

INTRODUCTION:Noninvasive neural interfaces promise scalable access to neural information without the risks of implanted sensors, but their fundamental limitation is the transformation imposed by the volume conductor between neural sources and sensors. Biological tissues spatially and temporally filter, mix, and disperse neural activity, such that recorded signals (e.g. EEG, ENG, surface EMG) primarily reflect convolution with tissue-dependent impulse responses rather than the underlying neural information. AREA COVERED:We frame this challenge using a generic convolutive model in which neural sources are observed through volume-conductor filters and noise. Two complementary strategies are discussed: direct compensation, which seeks to separate and recover subsets of neural sources through deconvolution methods, and indirect compensation, which learns representations that are invariant to volume-conductor variability from large, diverse datasets. EXPERT OPINION:We argue that progress in noninvasive interfacing will depend on explicit recognition and compensation of the volume conductor effect, either directly or indirectly. Together, these strategies point toward noninvasive neural interfaces that can scale beyond subject-specific calibration by isolating neural information from tissue-dependent distortions.

Human-machine interfaces (HMIs) have been widely integrated with motor rehabilitation and augmentation systems. Forecasting movement transitions during human-robot interaction is crucial to ensure system safety, intuitiveness, and reactivity, particularly in anticipating human motor intentions under sudden perturbations or emergency scenarios. In this study, we investigated pre-movement neural signatures preceding sudden movement transitions during ongoing bimanual tasks. Informed by these findings, we propose a physiology-informed EEG Transformer (PI-EEGformer) for EEG-based motor intention recognition. An EEG dataset collected from a bimanual movement task, where one hand was required to switch motor states in response to unexpected cues, was used to evaluate the performance of the PI-EEGformer in comparison with seven state-of-the-art models. Results showed that, prior to the movement transition, EEG power spectrum decreased, and movement-related cortical potentials (MRCPs) could be accurately extracted from the contralateral motor cortex. PI-EEGformer reached an average accuracy of 0.912 in inter-subject tests and 0.829 in cross-subject tests in detecting movement transitions using EEG from 500 ms to 100 ms prior to the actual movement. This performance was superior to all the state-of-the-art models tested. These results demonstrate that EEG neural signatures can predict sudden movement transitions during ongoing bimanual tasks. The PI-EEGformer, designed with these physiological signatures, can enable accurate prediction of sudden movement transitions. This study will help improve the response of HMI systems to sudden disturbances, contributing to a more realistic HMI system.

Abstract Brief high-frequency bursts of action potentials shape neural coding. In spinal motoneurons, they appear as doublets: closely spaced spike pairs, observed for a century, that amplify muscle force nonlinearly through the catch-like property. However, the cellular and anatomical origins of doublets remain unresolved. Combining intracellular recordings, a conductance-based model, and human motor unit and force data, we show that initial and repetitive doublets arise from the same mechanism. Both spikes are initiated at the axon initial segment, but the second is driven by a persistent sodium (NaP) current at the first node of Ranvier, which returns toward the initial segment and sums with somatic NaP current, the calcium-mediated afterdepolarization, and the passive membrane response. In our model, blocking any of these active currents abolishes the doublet, and monoaminergic facilitation of NaP determines whether doublets occur. Human motor units discharged repetitive doublets whose interval lengthened over time, a time course set by inactivation of the somatic NaP current, and produced nonlinear increases in force. Because monoaminergic drive both gates the doublet and sets how its interval evolves, neuromodulation shapes not only the gain but also the timing of motoneuron output.