Periodic Reporting for period 1 - NEURACT (Untangling population representations of objects. A closed loop approach to link neural activity to mouse behavior.)
Período documentado: 2023-03-01 hasta 2025-08-31
The NEURACT project focuses on understanding how the brain processes and recognizes objects by studying the activity of large groups of neurons in mice. Recognizing objects is a vital part of how humans and animals interact with their environment, but the underlying brain mechanisms are still not fully understood. NEURACT aims to fill this gap by developing a new approach that connects brain activity directly to mouse behavior using cutting-edge technology.
To do this, the project uses an innovative "closed-loop" system where mice are trained to perform visual tasks while their brain activity is recorded and analyzed in real time. This setup allows researchers to see how changes in what the mice see affect how their brains respond and how they behave. By building digital models—or “digital twin”—of the mouse brain, the project explores how different visual features are processed and how these brain responses guide behavior.
The insights gained from NEURACT could improve our understanding of visual perception, inspire new approaches in neuroscience and artificial intelligence, and lead to more efficient tools for brain research. Importantly, the project promotes interdisciplinary collaboration by combining neuroscience, engineering, computer science, and behavior analysis. It also embraces open science through publicly shared tools like EthoPy behavioral training system, supporting broader scientific progress.
To do this, the project uses an innovative "closed-loop" system where mice are trained to perform visual tasks while their brain activity is recorded and analyzed in real time. This setup allows researchers to see how changes in what the mice see affect how their brains respond and how they behave. By building digital models—or “digital twin”—of the mouse brain, the project explores how different visual features are processed and how these brain responses guide behavior.
The insights gained from NEURACT could improve our understanding of visual perception, inspire new approaches in neuroscience and artificial intelligence, and lead to more efficient tools for brain research. Importantly, the project promotes interdisciplinary collaboration by combining neuroscience, engineering, computer science, and behavior analysis. It also embraces open science through publicly shared tools like EthoPy behavioral training system, supporting broader scientific progress.
During the first part of the project, several key scientific and technical milestones were reached:
• Behavioral Training System: The team developed and improved EthoPy, a flexible and open-source software tool for training mice and collecting behavioral data. It allows for precise, automated experiments and has been adopted by other labs.
• Visual Discrimination Tasks: Mice were successfully trained to distinguish between different objects on a screen using changes in orientation, size, and background features. The findings showed how both simple parameters and more complex object features affect recognition.
• Virtual Reality for Mice: A unique 2D virtual reality setup was created where mice interact with a digital world while their movements are tracked. This system includes visual and olfactory cues, and can adjust stimuli in real time based on mouse behavior.
• Imaging Brain Activity: A large-field two-photon mesoscope was used to record brain activity from thousands of neurons. These recordings cover areas from basic visual processing regions to higher-order visual centers.
• Digital Twin Model: A deep learning model was trained using data from over 1,000 neurons to predict how the brain reacts to visual stimuli. The predictions from this model matched well with actual mouse behavior, demonstrating a powerful link between brain activity and perception.
These achievements have created a strong foundation for future experiments and provided valuable new tools and insights for the neuroscience community.
• Behavioral Training System: The team developed and improved EthoPy, a flexible and open-source software tool for training mice and collecting behavioral data. It allows for precise, automated experiments and has been adopted by other labs.
• Visual Discrimination Tasks: Mice were successfully trained to distinguish between different objects on a screen using changes in orientation, size, and background features. The findings showed how both simple parameters and more complex object features affect recognition.
• Virtual Reality for Mice: A unique 2D virtual reality setup was created where mice interact with a digital world while their movements are tracked. This system includes visual and olfactory cues, and can adjust stimuli in real time based on mouse behavior.
• Imaging Brain Activity: A large-field two-photon mesoscope was used to record brain activity from thousands of neurons. These recordings cover areas from basic visual processing regions to higher-order visual centers.
• Digital Twin Model: A deep learning model was trained using data from over 1,000 neurons to predict how the brain reacts to visual stimuli. The predictions from this model matched well with actual mouse behavior, demonstrating a powerful link between brain activity and perception.
These achievements have created a strong foundation for future experiments and provided valuable new tools and insights for the neuroscience community.
NEURACT has delivered innovative tools and discoveries that go well beyond current methods in neuroscience:
• The digital twin of the visual cortex is a groundbreaking model that simulates brain activity in response to visual inputs. This allows researchers to predict how the brain reacts and how these reactions shape behavior, offering a new way to study brain function.
• EthoPy, the open-source behavioral control platform, integrates real-time control with database management, allowing for high-throughput, reproducible behavioral experiments. It supports collaboration and method sharing across labs, accelerating innovation.
To ensure continued progress and wider use of these results, the following are key needs:
• Additional research and testing in new behavioral tasks.
• Greater access to funding and infrastructure for imaging and modeling.
• Support for technology transfer and commercialization of EthoPy.
• Establishment of standards and best practices for using digital twin models in neuroscience.
By bridging advanced neural recordings, behavior, and machine learning, NEURACT offers powerful tools for decoding how brains recognize objects—paving the way for new research, technologies, and applications.
• The digital twin of the visual cortex is a groundbreaking model that simulates brain activity in response to visual inputs. This allows researchers to predict how the brain reacts and how these reactions shape behavior, offering a new way to study brain function.
• EthoPy, the open-source behavioral control platform, integrates real-time control with database management, allowing for high-throughput, reproducible behavioral experiments. It supports collaboration and method sharing across labs, accelerating innovation.
To ensure continued progress and wider use of these results, the following are key needs:
• Additional research and testing in new behavioral tasks.
• Greater access to funding and infrastructure for imaging and modeling.
• Support for technology transfer and commercialization of EthoPy.
• Establishment of standards and best practices for using digital twin models in neuroscience.
By bridging advanced neural recordings, behavior, and machine learning, NEURACT offers powerful tools for decoding how brains recognize objects—paving the way for new research, technologies, and applications.