Queen's Gazette | Queen's University

Search form

News Release - Queen's University researcher begins to uncover how the brain processes tool use

Thursday, April 12, 2018

Although scientists have invented chess programs that can rival grand-masters, we have yet to design a robot that comes close to matching the dexterity of a five-year old child when it comes to manipulating the chess pieces.

One of the secrets of our success, when it comes to object manipulation and tool use, is our ability to predict the relationship between ‘motor commands’ from our brain and the resulting motion of the grasped object. This ability requires learning, and storing in memory, the mechanical properties (or dynamics) of the object and the environment it interact with.

Researchers have assumed that learning dynamics is linked to complete objects, such as a newly purchased hammer. However, new research by Randy Flanagan at Queen’s University and his colleagues at Cambridge University has shown that, instead, such learning is linked to the point on the object or tool that it being controlled.

“We can use a hammer to pound a nail or pry one loose from a board and the dynamics involved in these two tasks are very different,” says Dr. Flanagan. “Our results indicate that people can learn these different dynamics because they control the head of the hammer for one task and claw for the other. Our results suggest that, in the context of action, a tool is not represented as an object, but rather in terms of the specific tasks and control problems that the tool is being used to perform.

“Understanding memory representations underlying object manipulation tasks is potentially important because these tasks provide a powerful way to assess cognitive disorders, including learning and memory impairments that is non-verbal and does not require complex instructions.”

Flanagan and his colleagues used a virtual reality system with a robotic device to simulate forces acted on a grasped virtual object. “We found that when participants controlled different locations on the object, they could learn different dynamics linked to these locations, but could not learn the different dynamics when controlling a single location on the object.”

The task that was used can be likened to using a broom. If a person sweeps along a wall on the left, they may control the left edge of the broom (so that it rubs along the wall but not too hard) whereas if they sweep along a wall on the right they may control the right edge of the broom in a similar manner. When the broom rubs along a wall, torque (twisting force) is applied on the broom handle dependent on which side is toward the wall. Dr. Flanagan and his team discovered that a person can form two memories for these opposite torques when they exert control over either edge of the broom.

Moving forward, Dr. Flanagan and his team next want to examine how quickly and flexibly people can shift between different control points on an object during natural tasks, and whether people can simultaneously control multiple control points.

“I am also excited about using functional neuroimaging here at Queen’s to investigate the neural bases of control points,” says Dr. Flanagan. “It is well established that tool-use can change neural representations in perceptual and motor areas of the brain, so it will be fascinating to explore whether, and if so how, these representations depend not only on tool-use but which control point on a tool is being controlled.”

Working on the research with Dr. Flanagan were James B. Heald, James N. Ingram, and Daniel M. Wolpert from the Computational and Biological Learning Laboratory, Department of Engineering, University of Cambridge, Cambridge, United Kingston.

The research was published in Nature Human Behaviour.

- 30 -


Anne Craig, Media Relations Officer

613-533-2877 or anne.craig@queensu.ca

Related Experts

  • Flanagan, Randall