“An open problem remains on how to best deliver sensory feedback to the person using a prosthesis. It's difficult to use artificial feedback, something that's not from your own nervous system, to coordinate quick movements because your brain, as capable and fast as it is, has to constantly work to interpret this new artificial feedback.
“If you're trying to give the person feedback on their residual limb, which is the most natural place to get feedback, you’ll get a lot of crosstalk. Some people have tried putting vibration motors or some type of feedback actuator on different parts of the body, but that's a harder problem for your brain to solve. Even if it can be solved, the processing delay makes the feedback unnatural and not very useful.
“People have not yet found much success with vibration feedback, because the resolution is not good enough to make a difference. In the case of a prosthetic hand, for example, vibration feedback is no better than just observing the prosthesis with their own eyes. That's where some of my work was motivated—why bother to have all these attempts at sensory feedback for prostheses? We took a Bayesian approach of thinking about the uncertainty. If uncertainty of a type of sensory feedback is really high, then people won't rely on it. We need to figure out how to give them information that will reduce their uncertainty."