Practice helps mice’s brains separate dual tasks

A mouse tries to handle two tasks at once. In the beginning, its brain’s signals collide: different neural populations compete, creating interference. After consistent training, researchers found that the animals’ secondary motor cortex, known as M2, began separating the tasks and their neural representations.

The study was led by Professor Yung Wing-ho at City University of Hong Kong and Professor Ke Ya at the Chinese University of Hong Kong. By observing M2 during dual-task performance, the team identified a shift from shared, competing resources toward specialized neurons assigned to different tasks. That reorganization reduced interference and improved multitasking performance.

The researchers then built recurrent neural network models—systems that process information through connected cycles—to test whether the same principles could work outside the brain. The models reproduced the coordination and separation mechanisms and showed that they could accelerate learning for dual tasks, giving artificial-intelligence researchers a biological strategy to examine.

For people outside the lab, the concrete consequence is not a multitasking upgrade available today. It is a possible way to design training: practice may help separate competing task signals instead of merely demanding more effort. The CityUHK team also points to potential applications in education and rehabilitation programs for neurological conditions, but those uses have not been established in people by this study.

The boundary is clear. The biological observations came from mice, and the broader implications came from recurrent neural-network models. The findings in Neuron describe how experience can reorganize neural resources; they do not yet show that the same intervention or performance change has been demonstrated in humans.