Dr. Wolpert and his lab build computational models, informed by experiments that probe nuances of how people plan and execute movements, to unveil the movement-guiding models that brains build, summon and update over a lifetime. It’s a framework that applies to toddlers learning to walk, tennis players improving their forehands and cooks adjusting their knife work to slice a persimmon or a pepper.

“Movement usually is easy for many of us, but it’s incredibly difficult to understand how the brain plans and guides these movements ” Dr. Wolpert said. 

One line of Dr. Wolpert’s research characterizes the noise in the motor system, the part of the nervous system that controls movement. That’s the unavoidable variability in the muscular coordination behind, say, throwing darts, and in the sensory processing of, say, a dartboard’s location and of one’s arm and hand positions in space.

“I am terrible at sports,” Dr. Wolpert said with a laugh, “but I also know why,” referring to his own neural noise. 

He and his research colleagues have developed computational models that suggest brains possess neuron-based circuitry capable of predicting the effects of, and compensating for, much of this noise. Expertise (or the lack thereof) in dart throwing, driving or any other muscle-involved action is a sign of how well someone has learned to manage the noise. 

“This is how the brain deals with uncertainty,” Dr. Wolpert said. 

In recent years, he has been investigating the cognitive elements the brain deploys as it processes sensory input and plans out actions in response. 

“You have a stream of experiences from birth to death, and you build a repertoire of skills that you use in different contexts,” said Dr. Wolpert. “This sets up an interesting computational problem: When should I activate an old movement memory to help guide a current action, and when should I modify that memory or create a new one for potential use later?” 

That latter option might apply, for example, when a tennis player tries pickleball for the first time. 

As Dr. Wolpert garners clues from his computational models about how actual brains might orchestrate movement, he hopes this will inform new research as well as therapies to help so many of us affected by a disease of movement at some point in life.  

“If we can understand how to harness learning mechanisms in normal movement, we hopefully can apply that in clinical settings for those who have problems,” Dr. Wolpert said. 

In one of his collaborations, he is studying people with cerebellar disorders, which hinder the control of movements, such as reaching for objects. Even though patients with reaching issues can, for example, see their hands and witness the motor errors they are making, they still overshoot their targets. They do better, Dr. Wolpert said, with reinforcement learning, in which they are only told when their hands are getting closer or farther from an object.  

“We think reinforcement learning happens in different circuits than those in the cerebellum, so by harnessing those circuits during rehabilitation we might be able to help patients with cerebellar dysfunction regain movement functions they have lost,” Dr. Wolpert said.