Final Report Summary - OPT (Optimality Principles in Human Motor Control)
The goal of my EU funded research is to understand how the brain generates an appropriate muscle activation pattern given a particular movement task. This is a difficult question because the large number of joints and muscles provide the musculoskeletal system with very many degrees of freedom. Therefore, a typical task −like reaching for a target− can be performed in infinite ways, both with respect to kinematics (one can arrive at the target in infinite ways) and muscle activations. Interestingly, when measuring kinematics and muscle activation during a particular movement task, there is very little variation both within and across humans. Thus, on the basis of what generates the brain appropriate muscle activation patterns?
One essential part of my research was to develop a unique detailed Optimal Control musculoskeletal model of the human arm that would allow investigating what movements would look like under various assumptions such as minimizing energy and muscle force. This work was complicated at took well over 14 months to complete, however proven to be the key ingredient for my research. It was essential to reveal a new role for special mechano-receptors found in human tendons (Golgi-tendon organs) in posture/movement control. In addition it provided insight in the tight coupling found in the spinal cord between its sensory afferents and those of muscle spindles. In another study, the optimal control model was used to predict movements while minimizing costs at: the muscle input level (e.g. minimizing muscle activation), mechanical level (e.g. minimizing the muscle forces) and the kinematic level (e.g. minimizing jerk). These predictions were done for experimental conditions involving robot-arm induced force-fields and special visual perturbations such to separately manipulate the cost at the three levels mentioned before. Surprisingly, it was found that movements made by human subjects are only consistent with a costs at the kinematic level: mechanical costs and control costs do not play a role in movement path selection by the brain. An interesting follow-up question arises directly from the latter study. If it is true that the brain selects movements using kinematic costs only, muscle activation patterns must be generated in a separate step (which only then could involve mechanical/control costs). So: Given a particular movement path, how does the brain come up with the muscle activation patterns that we can observe experimentally? I have devised an experimental setup aimed to dissociate between mechanical and control costs. The results show that the brain favors muscle activation patterns that are energetically advantageous, and does so by favoring muscles that can efficiently contribute to the positive mechanical work done. Yet, the experimentally observed muscle activations (EMG) revealed that the brain adds substantial (energetically costly) co-contraction and remains an issue for further research.
During the last year of my EU funded research I have successfully applied for new funding; one grant for a PhD student (co-applicant) and a personal EU grant. Based on the latter success, I was awarded an Assistant Professorship at the host institution with a three year tenure track.
One essential part of my research was to develop a unique detailed Optimal Control musculoskeletal model of the human arm that would allow investigating what movements would look like under various assumptions such as minimizing energy and muscle force. This work was complicated at took well over 14 months to complete, however proven to be the key ingredient for my research. It was essential to reveal a new role for special mechano-receptors found in human tendons (Golgi-tendon organs) in posture/movement control. In addition it provided insight in the tight coupling found in the spinal cord between its sensory afferents and those of muscle spindles. In another study, the optimal control model was used to predict movements while minimizing costs at: the muscle input level (e.g. minimizing muscle activation), mechanical level (e.g. minimizing the muscle forces) and the kinematic level (e.g. minimizing jerk). These predictions were done for experimental conditions involving robot-arm induced force-fields and special visual perturbations such to separately manipulate the cost at the three levels mentioned before. Surprisingly, it was found that movements made by human subjects are only consistent with a costs at the kinematic level: mechanical costs and control costs do not play a role in movement path selection by the brain. An interesting follow-up question arises directly from the latter study. If it is true that the brain selects movements using kinematic costs only, muscle activation patterns must be generated in a separate step (which only then could involve mechanical/control costs). So: Given a particular movement path, how does the brain come up with the muscle activation patterns that we can observe experimentally? I have devised an experimental setup aimed to dissociate between mechanical and control costs. The results show that the brain favors muscle activation patterns that are energetically advantageous, and does so by favoring muscles that can efficiently contribute to the positive mechanical work done. Yet, the experimentally observed muscle activations (EMG) revealed that the brain adds substantial (energetically costly) co-contraction and remains an issue for further research.
During the last year of my EU funded research I have successfully applied for new funding; one grant for a PhD student (co-applicant) and a personal EU grant. Based on the latter success, I was awarded an Assistant Professorship at the host institution with a three year tenure track.