Cortical processing at the mesoscopic scale – MesoBrain

At the mesoscopic scale (0.5 mm typical scale in the mouse), the cerebral cortex shows a variety of spatiotemporal dynamics, including the rapid propagation of waves of neuronal depolarization that appear, vanish, reverberate and spiral within and across cortical area borders. Although they can be triggered by sensory inputs, the function of these waves of cortical activity in sensory integration and beyond remains poorly understood.

We hypothesize that these dynamics may contribute to sensorimotor integration, and that they could be the substrate for cortical information processing, independent of the detailed activity of individual neurons. To test this hypothesis, we will use the mouse as a model, and combine optogenetics, imaging and electrophysiology to simultaneously record and perturb cortical dynamics at the mesoscopic scale while the mouse, held by the head, is involved in an operant conditioning task. This closed-loop approach will allow us to interrogate in a causal way the functional role of mesoscopic cortical waves of activity.

Our first aim will be to ask whether the mesoscale spatial organization of artificial cortical inputs can have an impact on behavioral performance. Indeed, our recent data show that mice can learn a motor-control task guided by patterned photoactivations of channelrohdopsin in the primary somatosensory cortex (S1) only if this artificial sensory feedback is biomimetic and adheres to the functional map of this area. Our first goal will be to explore the contribution of the different cortical layers in the integration of mesoscopic sensory inputs. We will test the specific role of layers 2/3 and 4 in this task, using double and triple trangenic mice that will limit the expression of channelrhodopsin to these layers.

Our second aim will be to better characterize the specific ability of mice to extract information from structured stimulations at the mesoscopic scale. To do so, we will train transgenic mice in a 2 alternative forced choice task in which they will have to discriminate between photostimulations delivered in S1 that differ in their spatial organization, either at a critical scale of 500 µm, or at a scale 10 times smaller. We hypothesize that, due to the lateral propagation of activity at the cortical surface, only the mesoscopic scale features of the inputs will be efficiently discriminated by the mice.

Finally, in our third aim we will ask if beyond their potential role in sensory integration and perception, cortical waves of activity can be shaped voluntarily by the mice to solve a task. In this objective, we will record the cortical activity at the mesoscopic scale and extract in real-time the parameters of the cortical waves, including their position and orientation. We will then test whether mice can be trained to solve a task by directly controlling these parameters in order to obtain a reward.

Our research program will help to better understand the functional role of mesoscopic dynamics of cortical activity in the transfer and processing of information. Beyond this contribution, which will be of great interest to the neuroscience community, this work will also participate to a better understanding of the mechanisms at play in current brain-machine interfaces based on electroencephalography, and will open new avenues in the development of more efficient brain-machine interfaces.