Skip to content ↓

Experiment sheds new light on movement-learning process

MIT researchers have made some surprising discoveries about the way the human brain stores motor memories, the memories associated with learning skilled movements.

In an experiment developed by the scientists, subjects learned to move their arms toward a series of targets while an externally applied force field pushed their hands, disturbing their movements. Surprisingly, however, the scientists found that a subject's learning was erased if he or she was then required, within the next eight hours, to move toward the same set of targets while the force field pushed his or her hand in a different direction.

The researchers' results suggest that memories of learned movements initially exist in a vulnerable state. In this state, the memories can be destroyed when a different task must be learned soon after the initial task. If sufficient time is allowed, however, these vulnerable memories gradually become consolidated. After a motor memory is consolidated, it will not be erased when a new task is learned.

The work has implications for the way people should teach skilled movements. It could also provide insight into diseases of the nervous system that affect the ability to learn new movements.

The work was presented by Thomas Brashers-Krug, a graduate student in the Department of Brain and Cognitive Sciences, at a meeting of the Society for Neuroscience earlier this month. Mr. Brashers-Krug conducted the research with postdoctoral associate Dr. Reza Shadmehr and Professor Emilio Bizzi, head of the Department of Brain and Cognitive Sciences.

To study motor memories, the MIT scientists measured the performance of 22 right-handed subjects in a motor-learning task. The subjects, 16 men and 14 women, all between 18 and 35 years old, were tested either two or three days in a row.

The researchers asked their subjects to hold a handle attached to one end of a robot arm. The subjects moved the handle to guide it to a series of targets displayed on a computer monitor. Each time the subject reached one target, that target vanished and a new target appeared 10 centimeters away. The subject then had to guide the handle to this new location. In this way, the subject moved the handle to approximately 40 targets per minute.

The robot arm was receiving instructions from a computer to push the handle in a clockwise direction at the same time that the subject tried to guide the handle from target to target. In this way, the robot arm made it nearly impossible for the subjects to move straight to any of the targets. The task thus challenged the subjects to predict exactly how the robot would push their hands.

After about five or 10 minutes, most subjects learned to predict the robot's actions and so could move the robot arm fairly easily. When the subjects achieved this level of performance, they once again were able to move the handle straight to the targets. When a subject could move the handle smoothly, in spite of the robot's pushing and pulling, the researchers knew that the subject had learned the task well.

"Once subjects learned this task, we set out to answer two questions," Mr. Brashers-Krug said. "First, we wanted to know whether a subject who went home and came back the next day would remember what he or she learned. In other words, did a subject form a motor memory of this task? We found that, yes, they did.

"The second and more interesting question was whether learning a second task would affect a subject's ability to remember the first task. Our results suggest that the answer to that question is also yes."

Subjects in one group learned the robot arm task and immediately afterward were given a second task to learn. In the second task, they had to move the handle to the same series of targets, but the robot arm produced a very different set of forces-rather than pushing the handle in a clockwise direction, it now pushed the handle in a counterclockwise direction.

The following day, the researchers discovered that this group of subjects had completely forgotten how to move straight to the targets when exposed to the clockwise force field. Training in the second task had erased what the subjects had learned about the first task.

However, when a different group of subjects was exposed to the counterclockwise force field 24 hours after learning to move in the clockwise force field, they did not forget the first task. In this group of subjects, training in the second task did not erase the learning of the first task.

Since this second group did not learn the second set of forces until 24 hours after the first set, these results suggest that time itself could be a critical factor in the consolidation of motor memories.

These experiments showed an unexpected interference in learning two different motor tasks. This interference appears to disappear within 24 hours. The studies also provide a simple model for scientists to study the neurophysiological understanding of the acquisition of motor tasks.

Modern techniques available at MIT will allow the researchers to simultaneously record the activity of many neurons in an animal's brain as that animal learns these motor tasks. In this way, they will be able to discover the pattern of brain activity that underlies this interference.

This work was sponsored by grants from the National Institutes of Health and the Office of Naval Research.

A version of this article appeared in the November 30, 1994 issue of MIT Tech Talk (Volume 39, Number 13).

Related Topics

More MIT News