Expanding a behavioral repertoire by tuning spinal primitives

Abstract

Work in the frog spinal cord, and more recently in mammals, suggests that the degrees of freedom problem in motor planning may be simplified by building motor actions from combinations of motor primitives. How motor primitives arise from spinal circuits, their properties and their plasticity are important issues for this framework. Most recently, we have used various decomposition techniques (independent components analysis, matching pursuit cosine packet analysis, best basis algorithm and wavelet methods) in both simple and complex behaviors to examine the muscles that are controlled as groups or units, and to investigate the pattern and time scale of action of primitives in behaviors. Our data are beginning to provide a description of the way in which spinal elements are controlled for incorporation into more complex behaviors. We outline this description and our methods in this paper.