Article Figures & Data
Figures
- Figure 1.
Experimental setup and protocol. A, Participants performed submaximal isometric plantar flexions or knee extensions. Electromyographic signals were recorded using grids of surface electrodes and decomposed into motor unit discharge times. We used an online feedback motor unit (MU) control paradigm to test the hypothesis that the central nervous system mainly adopts a rigid control of the motor units during a single-joint task. B, Participants first performed a series of 10 contractions with torque feedback (reference contractions, B). In the second part of the experiment, motor unit firing rates were estimated in real time and converted into visual feedback displayed to the participant (online control contractions, C). The visual feedback consisted of a two-dimensional space in which a cursor moved according to the firing rates of two motor units. Two targets (Target 1 and Target 2) were alternately displayed to the participant, who had to differentially modulate the firing rates of the two motor units to alternately move the cursor toward one of the two targets. B, The firing rates of the motor unit 1 from the gastrocnemius lateralis (GL) and motor unit 7 from the gastrocnemius medialis (GM) of one participant are displayed across the 10 trials. Note that the two-dimensional space is displayed in panel B but was not provided as feedback to the participants during the reference contractions. This example shows a clear dissociation of the activity of two motor units from different muscles, i.e., GL and GM, during the online control task. Such a dissociation was not observed for GM–GM and vastus lateralis–vastus medialis motor units.
- Figure 2.
Motor unit behavior during the reference contractions. A, Biplots in which the coordinates of each motor unit represent its weights within the two latent factors, as identified using non-negative matrix factorization (data depicted for three participants: P3, P5, and P14). Angles between vectors were calculated to estimate the level of common inputs for each pair of motor unit. It is worth noting that the firing activity of GL–GM motor units was either well correlated (P5) or uncorrelated (P3), while the firing activity of VL–VM motor units systematically covaried (P14). B, Results are displayed for all participants, where each gray dot represents a pair of motor units and each colored dot represents a pair that was used as feedback in the second part of the experiment. The box denotes the 25th and 75th percentiles of the values distribution. The line is the median. The thick horizontal black lines denote a significant statistical difference between muscles (p < 0.05).
- Figure 3.
Volitional control of motor unit firing rates. A, Typical examples of cursor positions are depicted for the 10 reference contractions (only torque feedback) and the online contractions (5 for each target). The success rate is displayed for each target. It is calculated as the percentage of time spent within the target (while maintaining the torque constant) over the total duration of the trial. B, Relation between the angle between motor units estimated from the reference contractions and the success rate in Target 2 (true flexible control) for all the displayed motor units. Each data point is a pair of motor units, and the color code is the same as in C. C, Success rate for Targets 1 (T1) and 2 (T2). Each data point is a pair of motor units, the box denotes the 25th and 75th percentiles of the distribution of scores, and the line is the median. The thick horizontal black lines denote a significant statistical difference between muscles (p < 0.05).
- Figure 4.
Changes in firing rate at the motor unit population level. A, We calculated the correlation coefficients between the smoothed firing rates of the motor units displayed to the participant and the smoothed firing rates of the nondisplayed units. In this example, the correlation coefficient between the firing rates of the two displayed motor units was 0.01 (left panel), showing a perfect dissociation of the activity of the two units. We extended the analysis to the correlation between the firing rate of the displayed motor units from the GL (central panel) or GM (right panel) and all the other motor units from GM. Each pair of motor units has a different color. B, Correlation coefficients between the motor units (displayed and nondisplayed to the participants) during the reference contractions and the online control contractions. Each data point is a pair of motor units, and the connecting line denotes the change in correlation between conditions. The box denotes the 25th and 75th percentiles of the distribution of coefficients, and the black line is the median. The thick horizontal black lines represent a significant statistical difference between conditions. C, The changes in correlations observed between reference contractions and online control trials were separated between all pairs of motor units displayed to the participant (top panel). The changes in correlations between the displayed units and those nondisplayed units are shown in the bottom panel.
- Figure 5.
Control strategies during the online task. A, To investigate the strategies associated with the dissociation of GL and GM motor unit firing rates, we identified the time stamps where the cursor was in the target favoring GL (TGL) or GM (TGM) motor unit recruitment. TGas refers to a cursor position between the two targets. Ankle moments and EMG amplitudes of the tibialis anterior and vastus lateralis were averaged over these periods and compared. B, Average ankle force moments during the reference (Ref) contractions and the three targets. Each data point is a participant and the connect lines denote the variations between conditions. The box represents the 25th and 75th percentiles of the distribution of moments and the black line is the median. C, Each data point is a participant and the connecting lines denote the variations between conditions. The box represents the 25th and 75th percentiles of the distribution of EMG amplitudes and the black line is the median. The thick horizontal black lines represent a significant statistical difference between conditions (p < 0.05).
