A Computational Model of Muscle Recruitment for Wrist Movements

Ashvin Shah, Andrew H. Fagg, Andrew G. Barto

{ash | fagg | barto}@cs.umass.edu
Neuroscience and Behavior Program and Computer Science Department
University of Massachusetts, Amherst


In order to produce a desired movement, the brain must select the set of muscles that will be used to generate the appropriate joint motions. This problem is made particularly difficult by the fact that in many cases, different subsets of muscles may be used to produce the same movement. The motor system is therefore faced with a problem of redundancy that must be resolved. Hoffman and Strick (1999) recently examined the role of several different muscles in the production of two degree-of-freedom, step-tracking wrist movements involving flexion/extension and radial/ulnar deviation in primates. It was observed that agonist muscles were recruited as a function of movement direction in a roughly cosine-like fashion, with a muscle's "preferred direction" defined as the peak of the cosine fitted to the muscle's activity pattern. Furthermore, the preferred direction for some muscles differed significantly from the muscle's direction of action (or pulling direction).

We present a model of muscle recruitment that selects wrist muscle activation levels based on their ability to produce the desired movement while minimizing the total effort required by all muscles to implement the movement. Taken together, these two constraints lead to a cosine-like recruitment of the muscles. The pattern of muscle activation as a function of movement direction produced by the model closely matches that produced by experimental EMG recordings (Hoffman and Strick, 1999). The model also predicts differences in a muscle's pulling and preferred directions. These differences are due to the uneven distribution of the pulling directions of the six primary muscles involved in these movements. The pulling directions of some muscles are separated by gaps as large as 120 degrees. In these cases, movements to targets within the gap require additional activation of the two muscles whose pulling directions border the gap to compensate for the work that these muscles must do against one another. This leads to shifts of the muscle's preferred direction toward the gap. In addition, the magnitude of the shift increases as the size of the gap increases.


Presented at 2001 Spring Meeting on the Neural Control of Movement, Seville, Spain