Exploring How the Neuromuscular System Recruits Muscles and Motor Units

This is purely an exploration of how the neuromuscular system of one individual solves the control problem of loaded, isometric, shoulder abduction at 90 degrees. I used surface electromyography (sEMG) to investigate how the nervous system orchestrates the recruitment of motor units and muscles to generate the torque needed to control joint position. This is not a scientific study anymore than playing a scale is a symphony or mixing paint is art or counting to 12 makes one a personal trainer. This is an expression of my current understanding of the technology, anatomy, physiology, mathematics, and literature related to the neuromuscular system and surface electromyography. 

The setup

Equipment: Delsys Trigno wireless sEMG and 5lb. dumbbells.
Subject: Female colleague who is an avid snowboarder and exercise specialist.

sEMG probes were placed on the middle deltoids and upper traps of a colleague. I asked her to hold 5lb. weights in each hand at 90 degrees of frontal plane shoulder abduction (elbows extended), with her "elbow pits” facing forward (gleno-humeral joint position relative to the longitudinal axis of the humerus), for as long as she could. I collected tri-axial accelerometer data in addition to sEMG data for her middle deltoids and upper trapezius muscles. This first graph is of the accelerometer data for the deltoids: blue is for the right and green is the for the left. Each trace represents the position of her shoulders starting at 90 degrees of abduction. When she moves out of position (either abduction, up, or adduction, down), the trace shifts.

Accelerometer Data. Figure 1

Accelerometer Data. Figure 1

The Control Problem (not an issue, more like a math problem)

In figure 1, Notice that her positional control of the left and right shoulders is not the same. You can see that she holds the position on the left, overshooting the position a little, until a dip in the position between 50 and 75 seconds. Then, after some prompting, she is able to regain the original position and struggles to hold it from 76 seconds until failure at 99 seconds. After 99 seconds, she can no longer hold the position and her shoulder adducts. On the right side, she loses position at 29 seconds and, after some prompting, returns to the original position at 75 seconds. At 80 seconds, she fails to hold the position and her shoulder begins to adduct, even though she is trying her best to maintain the position. To summarize, the left side failed at 99 seconds and the right side failed at 80 seconds.

Fatigue and the Recruitment of Motor Units (MUs)

In this case, it is fair to say that failure is the point at which the Moto Units (an alpha motor neuron and all the cells that it innervates) that the nervous system has recruited have fatigue beyond their ability to contribute the force needed to hold the position.  A plot of the mean frequency of the sEMG signal for the right and Left deltoids shows a declining mean frequency.

Mean Frequency of Deltoids. Figure 2

Mean Frequency of Deltoids. Figure 2

The downward slope of the lines in Figure 2 indicate that both deltoids are immediately fatiguing and continue to fatigue as time goes on. In other words, the muscle fibers initially recruited were unable to maintain enough force a hold the position, so the nervous system began immediately recruiting more motor units (MUs). As these additional, larger MUs are added, the mean frequency decreases. Each time the currently recruited fibers were not able to keep up with the demand, more MUs were recruited (decreasing the mean frequency even more). This process continues until all of the muscle fibers of the available recruited MUs are too fatigued to maintain the force needed to hold the position.

The upper traps, on the other hand do not exhibit the same rate of fatigue. Check out the graph of the mean frequency for the upper traps.

Mean Frequency of Trapezius. Figure 3

Mean Frequency of Trapezius. Figure 3

It is challenging to detect much of a decrease in the MF of the upper trap sEMG until 45 seconds, then it goes back up, and at 80 seconds begins to decrease again. So the MUs recruited early in the exercise are, for the most part, able to manage the load without recruiting many additional MUs.

Plotting the best fit curve of the mean frequency gives us a better appreciation for the rates of decline of the mean frequency, i.e. the rates of fatigue. The rate of fatigue is different for each muscle in this isometric exercise.

Best Fit Curve for Mean Frequency of Deltoids. Figure 4

Best Fit Curve for Mean Frequency of Deltoids. Figure 4

The left deltoid fatigues at a faster rate than the right until at 60 seconds there’s a decrease in the rate of fatigue (figure 4). In addition, we can see that for the left and right traps, while the mean frequency for each side is different (the green line is higher than the blue), the rate of fatigue is close to the same (figure 5).

Best Fit Curve for Mean Frequency of Trapezius. Figure 5

Best Fit Curve for Mean Frequency of Trapezius. Figure 5

Motor Unit participation 

Which motor units did my colleague rely upon to solve this control problem? If you recall, MUs range in size. Small MUs have smaller alpha motor neurons, innervate smaller fatigue resistant fibers, are stimulated at higher rates, produce smaller action potentials, and are recruited first. They are referred to as high frequency MUs. Large MUs have larger alpha motor neurons, larger muscle fibers, are stimulated at lower rates, produce larger action potentials, and are recruited after smaller MUs. These are referred to as  low frequency MUs. The muscle fibers of low frequency MUs can be fatigue resistant or fatiguable. The lowest frequency MUs are fatiguable. 

For my colleague’s deltoids and traps, I ran a power density spectrum (PSD) of the sEMG for each muscle during the first 25 seconds of the contraction and the last 25 seconds leading up to failure. The x axis of the PSD indicates the frequency range of the MUs that are contributing to the contraction. The Y axis indicates how much that frequency is being used during the specified time interval. 

Power Spectral Density of the Deltoids - First 25 Seconds. Figure 6

Power Spectral Density of the Deltoids - First 25 Seconds. Figure 6

The PSD for first 25 seconds of the deltoid contractions (figure 6) reveals 2 things: 1) The MUs that are contributing the most energy to the contraction are at 50 and 80Hz for the left deltoid, and 50 and 75Hz for the right. 2) While the range of MUs used is very similar, the left side requires less use of the MUs in the frequency range to produce the same amount of force as the right.

Since the shoulders started to move, i.e. failed, at different times, I ran the PSD for the last 25 seconds leading to failure for each side: 55 to 80 seconds for the right deltoid, 74 to 99 seconds for the left deltoid (figure 7)

Power Spectral Density of the Deltoids - Last 25 Seconds. Figure 7

Power Spectral Density of the Deltoids - Last 25 Seconds. Figure 7

As expected with fatigue, the frequency range of the MUs used shifts toward lower frequency Mus, i.e. larger MUs. The left deltoid now uses MUs at 38Hz more often and the curve has shifted to the left indicating that more large MUs have been recruited. The right deltoid relied more on MUs at 50Hz (as indicated by the peak at 50Hz). So at the point where the right side was fatigued enough to need to recruit MUs of the 38Hz variety, for some reason, those MUs did not contribute as much as needed. The left side was able to continue to hold the position because it recruited larger MUs.

We have another opportunity for exploration. My colleague's right shoulder failed to hold the original position past 80 seconds, but she continued to use maximal effort to hold a position a few degrees below the original. Let's see which frequencies of MUs are dominant in the right deltoid from 80 seconds to 99 seconds, the failure time of the left (figure 8).

Power Spectral Density of the deltoids-post failure on the right and to failure on the left. Figure 8

Power Spectral Density of the deltoids-post failure on the right and to failure on the left. Figure 8

The left deltoid is graphed for comparison. The right side is dominated by MUs at 47Hz, whereas the left has MUs of 38Hz dominating. So greater fatigue was required to recruit the lower frequency MUs on the right.

The PSD for the trapezius for the first 25 and last 25 seconds of the contraction are represented on one graph. Notice that the traces for the left trap (lime green first 25, turquoise last 25) do not have the sharp, narrow peaks that deltoids have.

Power Spectral Density of the trapezius for the first 25 seconds (Left is lime green, right is blue) and the last 25 seconds (Left is turquoise, Right is red) Figure 9

Power Spectral Density of the trapezius for the first 25 seconds (Left is lime green, right is blue) and the last 25 seconds (Left is turquoise, Right is red) Figure 9

In Figure 9, the MUs used are more spread out over the range of frequencies. The right deltoid, however, is a little different (blue first 25, red last 25). For the first 25 seconds, the trace is similar to that of the second trace for the left trapezius (turquoise). The MUs used are spread out over the range, but they are being used more (as indicated by the blue trace being higher that the turquoise). For the last 25 seconds, the right trapezius, unlike the left, there is a sharp narrowing and sizable increase in the trace of the PSD. This indicates that the MUs at at 47Hz are highly active. I bet you can come up with a hypothesis for what the right trapezius is working so hard.

Now, what does one do with this information? 

Exercise Design

Exercise design is highly experimental. I attempt to take advantage of relationships, explored in the literature, between specific forces in exercise and adaptations in the neuromusculoskeletal system (NMS). The information helps inform my decisions, therefore increasing the likelihood that the exercises that I design will be helpful to the individual with whom I am working. I do not claim that this is the only, better, or best way to improve the efficiency and force production of my colleague's right deltoid.

It must be also noted that the intra and extracellular milieux of the motor units influences the rate and type of adaptation. So, my colleague’s nutrition would need to be addressed by someone other than me:).

So here’s how I am thinking about this: let's say the goal is to have symmetrical force output and duration of that output in both shoulders.


  • The NMS adapts both acutely and chronically to force application.
  • MUs are recruited to satisfy the amount of force required to solve a control problem and MUs are also recruited strategically to optimize the duration of the contraction.(4) 
  • There is evidence for a 20 degree transfer of isometric strength. (3)
  • Isometric contractions have more "muscular energy" than concentric contractions. Concentric contractions lose 20% of the "energy efficiency” as the tissue shortens. (2)
  • An increase in cytoplasmic calcium ions (due to unaccustomed rate of activity) in the muscle cell stimulate phenotypic changes in the fiber type of the cell. (5)
  • An increase in NAD⁺ is part of a signaling pathway that stimulates mitochondria biogenesis. (6)

I. Here’s my design of a control problem (exercise) that stimulates the NMS to recruit a broad range of MUs, with emphasis on the lower frequency MUs. I'd have my client perform sub-maximal to maximal voluntary isometric contractions (MVIC) at 20 and 40 degrees of abduction, in the following way: 

  1. Ramp up the effort into shoulder abduction against an immovable object (either slightly more load than can be moved or a stationary object) from very little effort to full strength over 2 seconds. 
  2. Hold the MVIC for 5 seconds
  3. Decrease the effort over 2 seconds. 
  4. Repeats steps 1 to 3 four to six times. 
  5. Rest for one minute
  6. Repeat step 4. 

II. To further stimulate change in phenotype of muscle fibers and their metabolic characteristics: Shoulder abduction from 0 to 90 degrees with an adjustable resistance profile that decreases as the shoulder abducts. Resistance applied at hand or distal humerus to minimize shear at gleno-humeral joint. Concentric phase is 4 seconds, 1 second isometric hold, 2 second eccentric, 2 second isometric. Encourage scapular motion, if needed. I would manipulate the resistance profile within the exercise the try and keep my colleague's perceived exertion at a 3 on the Purvis effort continuum for low-intensity resistance exercise scale (1 to 5). (1) The exercise is over when:

  1. scapular position changes from one rep to the next.
  2. In spite of attempts to adjust the resistance profile, exertion is a 4 on the Purvis effort continuum for low-intensity resistance exercise scale.(1) 
  3. 60 seconds of time under tension. 

After 3 weeks of bi-weekely visits, I would run the same test and check out the results. What do you think about that?


1. D’Adamo, C. R., Mcmillin, C. R., Chen, K. W., Lucas, E. K., & Berman, B. M. (2015). Supervised Resistance Exercise for Patients with Persistent Symptoms of Lyme Disease. Medicine & Science in Sports & Exercise,47(11), 2291-2298. doi:10.1249/mss.0000000000000683

2. Criswell, E., E d D. (n.d.). Cram's Introduction to Surface Electromyography (2nd ed.). Sudbury, MA: Jones and Bartlett.

3. Knapik, J. J., Mawdsley, R. H., & Ramos, M. U. (1983). Angular Specificity and Test Mode Specificity of Isometric and Isokinetic Strength Training. Journal of Orthopaedic & Sports Physical Therapy,5(2), 58-65. doi:10.2519/jospt.1983.5.2.58

4. Luca, C. J., Kline, J. C., & Contessa, P. (2014). Transposed firing activation of motor units. Journal of Neurophysiology,112(4), 962-970. doi:10.1152/jn.00619.2013

5. Schiaffino, S., & Reggiani, C. (2011). Fiber Types in Mammalian Skeletal Muscles. Physiological Reviews,91(4), 1447-1531. doi:10.1152/physrev.00031.2010

6. White, A. T., & Schenk, S. (2012). NAD /NADH and skeletal muscle mitochondrial adaptations to exercise. AJP: Endocrinology and Metabolism,303(3). doi:10.1152/ajpendo.00054.2012