“What Does the Brain Control?”

Gerald E. Loeb, M.D., Professor of Biomedical Engineering, University of Southern California

Attempts to use signals from the brain to control prosthetic devices usually start with a guess about what parameters of movement or muscle activity the brain might be encoding, followed by tuning a linear network to convert the available neural activity to those parameters as inputs to the prosthetic controller.  Research in monkeys indicates that modest correlations can be found between at least some of the neural activity and any reasonable set of parameters, and that subjects can learn to generate arbitrary neural activity patterns to control an unrelated movement parameter.  So why is the performance of these interfaces so poor on even simple tasks such as pointing to objects without mechanical interactions?  We hypothesize that the problem is the unrealistic design of the whole scheme.  Rather than controlling motor behavior explicitly, the brain normally acts indirectly by programming powerful intermediary systems that mix descending command signals with sensory feedback.  One such system is the spinal cord, whose circuitry is relatively well-described and which is capable of generating autonomously the complex and adaptive patterns of muscle coordination required for tasks such as locomotion.  Models of upper extremity biomechanics and realistic neural regulators are complex (with hundreds of control points) but they have unexpected emergent properties that appear to simplify and accelerate learning to produce complex movements.  That would provide a basis for their evolution and retention.  

Reference:  Raphael, G., Tsianos, G.A. and Loeb, G.E. Spinal-like regulator facilitates control of a two degree-of-freedom wrist.  J. Neuroscience 30:9431-9444, 2010.