Muscle activation patterns underlying successful stopping of voluntary sway at different body lean positions
Year: 2026
Type: article
Abstract: Introduction: Human action stopping involves two distinct phases; an initial global suppression of the motor command originating from the cortex (“pause”), followed by a selective cancellation of the targeted action (“cancel”), triggered at subcortical levels through the activation of movement antagonists (Hannah et al., 2022). This process seems to be time-related, depending on whether the action is inhibited from the start or is suppressed at a later stage. Recent evidence, however, challenges the two-stage (pause-then-cancel) model supporting a single-stage global suppression framework (Chaudhuri et al., 2025). These processes have been investigated mostly in upper limb reaching (Atsma et al., 2018) and aiming tasks (Chaudhuri et al., 2025) whereas whole body actions, which involve a higher cost of control and a risk of losing balance, are less well explored.In two studies, we have investigated stopping in the context of forward voluntary sway experienced when shifting the center of gravity towards the anterior limits of the base of support. Previous analysis of body sway dynamics revealed that the ability to stop does not only depend on the stop signal delay relative to the GO command but also on the position of the center of pressure (CoP) away from the neutral stance. Here, we asked whether successful stopping is modulated by the onset of the agonist muscle (soleus-gastrocnemius) or the co-activation of the antagonist depending on the position of the stop. Based on the two-stage model we predicted that forward sway would be globally suppressed at a short stop position (confirmed in the timing of activation of the agonist muscle) or alternatively, sway could be stopped at a longer CoP position by co-activating the movement antagonists (i.e., increasing co-contraction).Methods: The study was jointly conducted at King’s College London in the Biomechanics Laboratory (ethics MRA-24/25-48273) and Aristotle University of Thessaloniki in the Motor Performance and Adaptive Physical Activity Laboratory (ethics 306/2025). Twenty-four healthy young adults (14 males, 10 females; mean ± SD age: 30.0 ± 9.8 years; height: 1.7 ± 0.8 m; body mass: 66.8 ± 13.0 kg; foot length: 0.25 ± 0.17 m) stood barefooted on a force platform facing a monitor (1.5 m) displaying a GO (green target) and END (red target) signal. Participants were fitted with surface electromyography (EMG) electrodes placed over the gastrocnemius, soleus, tibialis anterior and the deltoid (reference) muscles. EMG and CoP data were synchronously sampled (Vicon Nexus, version 2.12, Vicon Motion Systems Ltd., Oxford, UK). The CoP signal was used in real time to trigger the events using specialized software written in LabView. At the onset of the GO signal, participants performed 2 sets of 50 forward sway cycles until the CoP reached a target corresponding to 60 % of maximum leaning distance. At random cycles (28 %), the GO target disappeared signaling a STOP. Stopping was assessed in two task conditions: discrete and continuous sway. In the discrete sway task, participants were required to stop and maintain their starting position for 3 s after completing each sway cycle. In the continuous sway task, the next GO signal was triggered immediately upon the participant returned to their starting position. CoP and EMG data were processed offline in MATLAB. EMG signals were band-pass filtered (10-300 Hz), rectified and low-pass filtered (50 Hz) to get the linear envelope before normalized to the mean baseline activity (1000 ms). A STOP was considered successful if CoP velocity decreased and maintained at baseline levels (±10 % of maximum) for at least 1000 ms after the STOP command. Agonist EMG burst onset relative to the stop command (target offset) and the agonist-antagonist co-contraction index (Falconer & Winter, 1985) over a 1000 ms period after agonist burst onset were determined and compared between the two STOP positions (30 % and 45 %) and the two sway tasks (discrete vs. continuous) for the successful stops using Linear Mixed Models (LMM).Results: LMM revealed a significant main effect of position (χ²(1) = 5.04, p = 0.0247) on agonist burst onset time, indicating that gastrocnemius was recruited earlier in the long (45 % of max CoP; 0.766 ± 0.094 s) compared to the short (30 % of max CoP; 0.821± 0.093 s) stop position. On the other hand, there were no differences in agonist recruitment onset between the two task conditions. The co-activation index (CI) was not different between the two stop positions but significantly differed between the two tasks (χ²(1) = 6.46, p =0.0109), being lower in the continuous (39.1 ± 4.3) than the discrete (43.4 ± 4.3) task.Conclusions: Results confirmed that successful stopping was modulated by the timing of the agonist EMG activity that was earlier recruited in the long (45 % of max CoP) than the short (30 % of max CoP) stop position. On the other hand, the co-activation of the antagonist does not seem to be a mechanism involved in the control of successful stopping as the CI did not differ between the two stop positions. These findings align with recent evidence which challenge the two-stage (cancel-then-pause) model and suggest a sole global motor suppression as a primary mechanism of action stopping (Chaudhuri et al., 2025).Acknowledgements: The study received funding from The Royal Society (ref: IES\R2\242223 - International Exchanges 2024).
Source: Communications in Kinesiology
Institutions: Aristotle University of Thessaloniki, King's College London
Language: English
Topic: Motor Control and Adaptation
Subfield: Cognitive Neuroscience
Field: Neuroscience
Domain: Life Sciences
Open Access status: diamond
Funder: Royal Society
OpenAlex ID: W7225153161