Figure 1.
Experimental setup and sample force and EMG data.
A) Top view of the experimental set up. We isolated the subjects’ wrist middle, ring and little fingers to allow only abduction of the index finger. The elbow was flexed at 90° and the shoulder was in 20° forward flexion and 30° abduction. B) Representative example of force and EMG signals at 5% and 30% MVC. The top row demonstrates the force task at 5 and 30% MVC, whereas the bottom row represents the interference EMG recorded at 5 and 30% MVC.
Figure 2.
Quantification of low-frequency oscillations in force and muscle activity.
A) The force task (left column) and its corresponding interference EMG (right column). B) The force signal (10–20 s; left column) and the corresponding rectified EMG signal used for analysis (right column). C) Both the force (left column) and the rectified EMG (right column) were low-pass filtered at 2 Hz. This low-pass filtering demonstrates the important frequencies in force and EMG bursting. D) The power spectrum density of the low-pass filtered force (left column) and low-pass rectified EMG (right column). Most power occurred below 0.5 Hz.
Figure 3.
Force variability and oscillations in force.
A) The SD of force for 5 and 30% MVC. As expected, the variability of force was greater with higher force. B) Normalized power spectrum density of force from 0–2 Hz. In this figure we present the average normalized power spectrum density because it was similar for the two force levels. The greatest power (∼75%) occurred below 0.5 Hz. C) The association between low-frequency oscillations of force and SD of force was strong (R2 = 0.82). This indicates that ∼80% of force variability is due to the oscillations in force below 0.5 Hz.
Figure 4.
Normalized power spectrum density of low-pass rectified EMG power from 0–2 Hz.
In this figure we present the average normalized power spectrum density because it was similar for the two force levels. The greatest power (∼40%) occurred below 0.5 Hz.
Figure 5.
Coherence between the force and EMG burst oscillations.
A) Representative example of a force and an EMG burst oscillations signal for 20 s. It is obvious that both signals exhibit common low-frequency oscillations. The dotted line represents low-pass filtered force and EMG burst at 2 Hz, whereas the solid line represents low-pass filtered force and EMG at 0.5 Hz. B) The overall coherence between the force and EMG burst oscillations when low-pass filtering the signals at 0.5 Hz and 2 Hz. The coherence for the two filtering procedures was similar, which indicates that low-pass filtering the force and EMG signals at 0.5 Hz captures well the synchrony between the two signals.
Figure 6.
Changes (from 5 to 30% MVC) in force and EMG burst oscillations below 0.5 Hz.
The association between changes in force oscillations below 0.5 Hz and changes in EMG burst oscillations below 0.5 Hz was moderate (R2 = 0.51). This indicates that ∼50% of the force oscillations below 0.5 Hz are related to the EMG burst oscillations below 0.5 Hz.
Figure 7.
Changes (from 5 to 30% MVC) in EMG burst oscillations below 0.5 Hz and interference EMG.
Based on the multiple regression model, the change in power from 35–60 Hz in the interference EMG strongly predicted the EMG burst oscillations below 0.5 Hz (R2 = 0.95). This finding indicates that modulation of interference EMG from 35–60 Hz is strongly associated with EMG bursts below 0.5 Hz.
Figure 8.
The associations between changes in force and muscle activity.
We explain these associations starting at the bottom of the diagram. The change in variability of force (SD of force) from 5 to 30% MVC was strongly related to the change in force oscillations below 0.5 Hz (R2 = 0.82). The change in force oscillations below 0.5 Hz was related to the change in EMG burst oscillations from 0–0.5 Hz (R2 = 0.51). Interestingly, the change in EMG burst oscillations from 0–0.5 Hz was related to the change in power from 35–60 Hz in the interference EMG.