Representation of Proprioceptive Information for Generation of Arm Dynamics

The goal of this thesis was to investigate how the brain represents proprioceptive information about limb state, i.e., position, velocity and acceleration to generate reaching movements. To achieve this goal, human behaviors during point-to-point reaching in multiple novel dynamic environments were examined and neural coding schemes in the brain were inferred from the observed behaviors, e.g. the pattern of generalization. The patterns of generalization across limb position in multiple dynamic conditions suggested that the internal model encodes limb position and velocity using a gain-field scheme. That is, neural elements in the brain seem to encode limb velocity and position multiplicatively. This multiplicative combining scheme seems to extend to encoding of limb velocity and acceleration as the pattern of generalization across acceleration space rejects a linearly separable encoding of acceleration and velocity. The nonlinear combining of all sensory information in the brain might originate from the properties of the proprioceptive sensors in our body, muscle spindles. Muscle spindles respond to all three sensory information in highly nonlinear way due to the nonlinear relationship between tension and muscle spindle length. Surprisingly, some tuning functions that were independently derived from the mathematical model of muscle spindle closely resemble the tuning functions derived from the gain-field scheme and thus, can explain the patterns of generalization across various state spaces. So far, at least in a force field paradigm, motor learning has been considered as an implicit procedure and the effect of explicit knowledge has been ignored. However, a proper assessment of relevant explicit knowledge indicates that there is a small but significant effect of explicit knowledge on motor performance. Interestingly, the formation of explicit knowledge is primarily driven by visual information while the formation of implicit internal model is primarily driven by proprioceptive information. Advisor: Dr. Reza Shadmehr