Maurice Smith, Gordon McKay Professor of Bioengineering, SEAS
Maurice studies how your brain forms motor skills through practice at a basic science level in terms of algorithms – like an engineer or roboticist might think of it. Despite all our recent progress, one of the areas that is behind is physical applications. Maurice remembers watching Star Wars as a child and the dream for robots to be ubiquitous. The problem is that we don’t understand yet how to make robots’ motor skills robust yet – eg uneven ground.
Maurice proceeded to discuss several examples:
Maurice described the symptoms of this condition (which can be caused by a variety of things including alcohol misuse) – poor walking skills, broad-based gait, tremor and past-pointing.
In one study, they put prism glasses on people that change their perception and made them play darts. In healthy people, they learn in around 12 shots to start throwing the dart towards the middle of the board, and on removal of the prisms, there is an aftereffect then a correction. In people with cerebellar disease, there is no learning, and the dart shots remain off to the side:
Maurice studies this sort of cerebellar learning in his lab.
He asks individuals to move their hands from one point to another holding a handle with a motor attached. They can then add various forces to the handle using the motor – trying to prevent a straight path – and measure change in the forces made by the person’s hand. Healthy people learn where the forces are perturbing them, and learn to move their hands in a straight path again. You can display this in a learning curve.
Two processes for motor learning
They have found that there is a fast process (low retention, fast adaption) and slow process (high retention, slow adaption) for motor adaption. They have found that over time, the slow process contributes more and more to the motor movement.
To learn more about these processes, they carried out a test to see how stable to learning was.
Maurice is also doing work in motor control in Alzheimer’s (AD) – which he didn’t have time to go into today, but showed that human data points to an extrinsic learning deficit in AD.