Operant conditioning of cortical waves through a brain-machine interface

At the surface of the cerebral cortex, activity dynamics measured at a mesoscopic (about 0.1 to 1 mm in mouse cortex) scale are characterized by both spontaneous and behavior-related dynamical waves of synchronized neuronal activity. These waves are thought to participate in information propagation and processing, but remain poorly understood.

To assess if such mesoscopic coordinated neuronal dynamics can be controlled in a goal-directed manner, we implemented a task in which mice are trained to generate waves with specific trajectories in order to obtain rewards. We tracked propagating waves at the surface of the somatosensory-motor cortex of head-fixed mice in real time by means of wide field calcium imaging, and conditioned the delivery of rewards to the detection of wave trajectories responding to specific spatiotemporal criteria.

We found that the majority of the trained mice significantly increased their performance, mainly by increasing the frequency of rewardable waves. As the mice learned to achieve this task, we observed changes in the spatiotemporal patterns of the cortical waves. By revealing that, upon learning, neuronal activity can be shaped at the mesoscopic scale to generate specific waves patterns, our work opens up new perspectives for brain-machine interfacing.