Sensory feedback remains an elusive goal for powered prosthesis users. Clinicians and researchers have explored feedback strategies since the early nineteen sixties with varying degrees of success, but none are used in clinical devices. Some studies show that additional sensory feedback enhances performance but many others show no clear improvement. We don't understand whether amputees continuously rely on feedback during prosthesis movements, or if they use feedback intermittently for error corrections. To guide sensory feedback strategies for prostheses, we need a better understanding of how amputees use feedback during movement. The influence of feedback during movement can be predicted with Bayesian models of sensorimotor adaptation. These models have successfully described how able-bodied subjects use feedback during movement planning, movement execution, and how they use feedback to adapt in response to errors. This framework is especially appropriate for prosthesis control because it considers the effects of noise and uncertainty in both feedforward and feedback information. Prosthesis users presumably experience high levels of uncertainty in both feedforward and feedback information. The noise of electromyographic (EMG) signals used for prosthesis control is much greater than the noise of force and position signals used by able-bodied subjects to control objects. Thus, feedforward uncertainty may be increased in amputees. Feedback uncertainty is presumably also higher in amputees because prosthesis users are limited to visual feedback and incidental feedback transmitted through the socket. These factors affect adaptation and influence how feedback is used during movement. In our preliminary work, amputees adapted to visual perturbations in a manner consistent with Bayesian predictions. These preliminary results provide a strong motivation for a full investigation of adaptation behavior during powered prosthesis control. This proposal will investigate adaptation during more realistic conditions. We will study how amputees adapt to inertial perturbations, which are the most commonly encountered in everyday activities. Characterization of adaptation behavior will help in designing sensory feedback strategies that can most effectively reduce uncertainty for amputees using powered prostheses. Enabling effective sensory feedback strategies will improve performance with powered upper limb prostheses, recovering greater functional ability for amputees. The proposed adaptation framework can model how prosthesis users rely on feedback and how certain types of feedback can improve performance. This framework has the potential to improve poorly understood aspects of sensory feedback, such as the success of a given strategy for some tasks but not for others. The application of this framework to prosthesis control also provides insight on the adaptation of people affected by other movement or neurological disabilities.
People that use powered prosthetic arms must currently operate their devices with only visual feedback; which requires a burdensome level of concentration and high uncertainty. The objective of this project is to develop a framework that predicts which sensory feedback information will be most useful in reducing uncertainty. This framework will be helpful in understanding the interaction between the person and prosthesis.
