A hierarchical foundation for models of sensorimotor control.

摘要: Successful performance of a sensorimotor task arises from the interaction descending commands brain with intrinsic properties lower levels system, including dynamic mechanical muscle, natural coordinates somatosensory receptors, interneuronal circuitry spinal cord, and computational noise in these elements. Engineering models biological motor control often oversimplify or even ignore because they appear to complicate an already difficult problem. We modeled three highly simplified systems that reflect essential attributes tasks: acquiring target face random torque-pulse perturbations, optimizing fusimotor gain for same minimizing postural error versus energy consumption during low- high-frequency perturbations. The emergent maintained stability feedback delays, resolved redundancy over-complete systems, helped estimate loads respond suggest general hierarchical approach modeling which better reflects real problem faced by brain, as first step toward identifying actual neurocomputational steps their anatomical partitioning brain.