Research‎ > ‎Papers‎ > ‎

Pruszynski - 2009 - Temporal evolution of "Automatic Gain-Scaling"


Pruszynski JA, Kurtzer I, Lillicrap TP, Scott SH. Temporal evolution of "automatic gain-scaling". J Neurophysiol. 2009 Aug;102(2):992-1003. PubMed

10 Word Summary

Fast response scales with perturbation, but steady-state does not.


The earliest neural response to a mechanical perturbation, the short-latency stretch response (R1: 20–45 ms), is known to exhibit "automatic gain-scaling" whereby its magnitude is proportional to preperturbation muscle activity. Because gain-scaling likely reflects an intrinsic property of the motoneuron pool (via the size-recruitment principle), counteracting this property poses a fundamental challenge for the nervous system, which must ultimately counter the absolute change in load regardless of the initial muscle activity (i.e., show no gain-scaling). Here we explore the temporal evolution of gain-scaling in a simple behavioral task where subjects stabilize their arm against different background loads and randomly occurring torque perturbations. We quantified gain-scaling in four elbow muscles (brachioradialis, biceps long, triceps lateral, triceps long) over the entire sequence of muscle activity following perturbation onset—the short-latency response, long-latency response (R2: 50–75 ms; R3: 75–105 ms), early voluntary corrections (120–180 ms), and steady-state activity (750–1250 ms). In agreement with previous observations, we found that the short-latency response demonstrated substantial gain-scaling with a threefold increase in background load resulting in an approximately twofold increase in muscle activity for the same perturbation. Following the short-latency response, we found a rapid decrease in gain-scaling starting in the long-latency epoch (~75-ms postperturbation) such that no significant gain-scaling was observed for the early voluntary corrections or steady-state activity. The rapid decrease in gain-scaling supports our recent suggestion that long-latency responses and voluntary control are inherently linked as part of an evolving sensorimotor control process through similar neural circuitry.


  • The experiment focused on the flexion/extension of the elbow and measured EMGs for biceps longus, brachioradialas, triceps longus, and triceps lateralus.  The subject was to try and maintain an angle of 45° with their shoulder w.r.t. to the body's ML axis and their elbow at 90° w.r.t. to the upper arm.  A background load was applied to activate either flexors or extensors. A step in torque was applied to "stretch" the activated muscles.
    • Pretty vague about exactly what kind of load they were applying.
  • Gain scaling is defined as (A_big - A_small)/(A_pre_big - A_pre_small), where A_big corresponds to the EMG amplitude from a background load of 3 Nm and A_small = 1 Nm.  The "pre" magnitudes correspond to the muscle recordings before the background load was applied.
  • Velocity sensitivity is defined as (A_med2 - A_med1)/(Ang_med2 - Ang_med1), where A_med# are the EMG magnitudes of two perturbation magnitudes for a background load of 2 Nm, and the "Ang" are the corresponding change in joint angles caused by the perturbation.
  • A factor was added to the Gain Scaling to account for "increases in muscle stiffness", which was defined by C = VelocitySensitivity*(Ang_big - Ang_small)/(A_pre_big - A_pre_small).
  • Employed a single link model to investigate muscle-property interaction with kinematics.
  • Increasing elbow perturbation = increased elbow angle excursion.
  • Increasing background load = decrease in elbow angle excursion.
  • Based on calculation, changes in background load did not significantly change muscle stiffness.
  • Background load affects Gain Scaling substantially in the first 50 ms, but in the later epochs show less of an effect.
    • The gain-scaling effect disappears at about 75 ms.
  • For this study they noticed that increase in "muscle stiffness" did not seem to affect effective stiffness.  Increased background load resulted in similar effective stiffness.
  • Decreases in Gain Scaling have implications to the Size Recruitment Principle.  It is surmised that higher level commands are responsible for the reversal.
    • The decrease in Gain Scaling is continuous. This throws somewhat of a wrench in the partitioned space idea of lower level interacting with higher level.
  • The paper thinks that voluntary and APR-like responses are highly linked.  I would agree.