Stimulating about 20 neurons sometimes triggered a learned movement in mice

Stimulating about 20 neurons sometimes triggered a learned movement in mice

The outcome depended on the selected cells and motor cortex activity before the light pulse. These findings come from Takaki Komiyama’s group in a study published in Nature on October 14, 2026.

Researchers trained mice for two weeks to press a lever after a sound. They then monitored hundreds of neurons in the upper layers of the primary motor cortex, a brain region involved in controlling movement. They selected about 20 cells that normally became active before the press.

The team used holographic optogenetics, a method for activating selected cells with light. In some trials, a three-millisecond pulse replaced the sound and activated these neurons simultaneously. These trials were randomly interspersed with sound trials. Lever presses after the light pulse earned no reward. The researchers conducted 17 experimental sessions in 11 mice.

In some trials, movements during the lever press followed a trajectory similar to the learned one. About half of the evoked movements had a similarity coefficient above 0.5. Neurons that the light did not directly target also reproduced the activity pattern associated with the learned press.

To test the role of cell selection, the researchers stimulated the same number of neurons that normally became active after movement began. The median trajectory similarity coefficient was 0.229, compared with 0.455 when they stimulated cells active before the press. These numbers describe how closely the evoked movement resembled the learned one. With cell count held constant, the outcome differed depending on when the targeted neurons normally joined the activity sequence.

Movement similarity was also associated with cortical activity before stimulation. The more closely activity during the half-second before stimulation resembled the state preceding a normal learned press, the more closely both the movement and the activity of other cells matched the learned pattern. Even when stimulating the same group of neurons, an identical pulse produced different outcomes across trials.

Stimulating a small group of cells therefore sometimes triggered an already learned movement in mice. The study does not demonstrate control of arbitrary movements or recovery of function in humans, and it provides no basis for treating the approach as an established therapy.

The link below leads to Eternal Search’s page for an organization called Experiment. The supplied material describes neither its work nor a connection to this study.

Open the related Eternal Search page

Sources
[1] nature.com
[2] biorxiv.org