How is sensory feedback of force and knowledge of prior movement experiences used in feedforward control of movement?

It remains unclear what information is used by the central nervous system (CNS) to improve performance of a movement task during training. We used a force pulse generated by a manipulandum and force applied during random catch trials to investigate how information about force is used to change feedforward movement control. Catch trials consisted of locking the manipulandum to measure isometric force. Participants experienced sets of two force pulses and were required to oppose the second pulse to minimise positional error. We investigated how prior experiences influenced feedforward commands by controlling the distribution of forces experienced over the course of the experiment. Electromyographic recordings of arm muscles were used to determine whether cocontraction was being used to minimise positional error. Forces measured on catch trials indicate that participants were unable initially to accurately oppose the force pulse and instead increased shoulder and elbow joint stiffness to minimise displacement. Force estimates were initially biased toward sensory information from the first force pulse but with practice became increasingly biased toward the mean of the distribution. Positional error decreased with practice. This suggests that with continued practice participants use sensory information about force to more accurately estimate the required force. We also showed that participants use information about their prior experience in their estimate and use this prior experience more when knowledge of the prior experience is greater and when the distribution is narrower. These results suggest that the CNS uses Bayesian models when estimating force requirements of a movement task