A nonlinear quantitative method for assessing locomotor performance

We have developed a nonlinear method based on information theory and bootstrap techniques to quantify the global characteristics of stepping that are associated with limb coordination. Since many relationships in physiology, which are inherently nonlinear may not become evident using linear methods the use of such nonlinear methods is important. Correlation, regression, ANOVA, and the like are all techniques that look for linear relations between or among variables. We have developed a technique that can be used to investigate whether the neural-control system of the lumbosacral spinal cord can retain or learn a neuromotor-control strategy. Our results suggest that some learning-related processes can occur within the spinal circuitry. This method has general applicability for finding nonlinear associations among many variables. In general, this approach allows the investigator to quantitatively assess the level of simultaneous dependency among many variables without making any assumptions about their functional relationships or about their underlying distributions.