Periodic Reporting for period 1 - CORVISDEC (Cortical circuits underlying visual decision-making behaviors in mice)
Período documentado: 2018-08-01 hasta 2020-07-31
To better understand the perceptual, cognitive, and motor aspects of decision-making, this project combined a tractable mammalian model (mouse), a high-throughput neural assay (two-photon imaging), and a complex behavioral task (psychophysics) to parse the neural computations required to drive decision-making in a complex and changing environment.
Specifically, our objectives for this project were to: (1) Train mice to perform a complex visually guided behavioral task that required mice to transform uncertain sensory evidence and decision history into choice actions; (2) Train animals with an open-source behavioral paradigm and setup that we could deploy to multiple laboratories as part of a global collaboration studying visually guided behavior in the mouse; (3) Once animals were trained in the behavioral task, record activity in the mouse’s brain using two-photon imaging techniques, which afford the longitudinal and large-scale tracking of neurons over multiple brain regions and multiple days; (4) Using data from these recordings, decode the specific task variables from various neural populations to identify which cortical regions are responsible for transforming sensory evidence into motor actions.
The project's high-throughput in vivo imaging combined with a complex behavioral task hold promise for disentangling the sensory, cognitive, and motor processing required for mammals to successfully make decisions under real-world conditions. Identifying the cortical role in decision-making will direct neuroscience toward a better understanding of how we interact within complex social environments, and how this process may be disrupted in conditions such as Alzheimer’s disease, autism spectrum disorders, depression, obsessive-compulsive disorder, or drug addiction.
Surgical techniques to access sensory, parietal, and motor regions of cortex for imaging was achieved and refined during this period, particularly with respect to motor cortex. This region of the brain is notoriously difficult to access due to the anatomy of the skull and the requirements of the implant, but persistent efforts have refined our methodology to make this preparation one now available to others in the host laboratory, within our collaboration, and across the field. This preparation marks the first time that mouse motor neurons have been accessible via two-photon imaging in our lab.
Behavioral hardware and software have been developed, refined, and disseminated to multiple laboratories, whose researchers are now generating standardized behavioral data from mice across the world. In this project, we highlighted a need for more widespread collaboration across neuroscience laboratories and committed to developing an open-source behavioral 'kit’ that deployed our research technology from the host lab to other labs. This work is disseminated publicly (International Brain Laboratory, 2020, doi.org/10.1101/2020.01.17.909838).
Over the period, we successfully trained mice on a behavioral task that required the integration of sensory (visual) information with history of reward outcomes and accurate action planning. We find that mice adapt their behavior in the face of changing environmental conditions (e.g. when the reward changes or when the visual information changes), similarly to how humans would adapt under dynamic environmental realities.
Our analysis of neural data collected during the behavioral task suggests that the brain very closely links sensory information available in the task with the action plan that mice associate with that sensory information. These early (unpublished) results suggest that, in lieu of clear cortical boundaries between areas that process vision and motion, there is extensive overlap in areas that code both sensory and motor information over the time that animals prepare to act during the task. The notion of serial, separate processing stages between vision and action does not seem to be supported by the neural activity observed in the brain areas we studied, and opens the door to continued study on how precisely brain areas work in concert to guide the body’s actions and decisions.
The excitement that our work was met with in the neuroscience community holds great promise for fundamentally changing how collaborative neuroscience is undertaken and managed on a global scale. As a result of this infrastructure and the popularity of our approach, we have written another article specifically outlining the structure of the IBL collaboration as a model for similar research collaboratives in the future (Wool et al., 2020, psyarxiv.com/f4uaj).
As for the behavioral and neural data collected and analyzed for this project, we expect to undertake a final report to formalize our findings about the relationship between sensory and motor processing in the motor regions of mouse cortex. We are currently completing this analysis and expect to submit a research article for publication in early 2021. All analysis code is currently deposited on Github.