According to a new study, future brain-computer interfaces might be improved by an AI tool that decodes what mice see.
Researchers at the Swiss university EPFL created the system, which they called CEBRA. Their goal is to reveal undiscovered connections between the brain and behaviors.
The group attempted to decipher what a mouse sees when watching a video in order to test CEBRA (pronounced "zebra").
First, the scientists gathered freely available brain data from rodents watching films. Electrode probes placed in the visual cortex of a mouse allowed researchers to measure some of the brain activity. The remainder came from optical probes of genetically altered mice, whose neurons were designed to glow green when they were active.
The base algorithm in CEBRA was trained using all of this data. As a result, the system developed a mapping between brain activity and particular video frames.
The group then used the technique on a different mouse that had watched the video. Following data analysis, CEBRA was able to precisely anticipate what the mouse had observed based solely on brain signals.
The team then used the brain activity to recreate the clip. In the video down below, you can witness the outcomes for yourself:
Unsurprisingly, the researchers aren't just curious about how rodents watch films.
Finding structure in complicated systems is CEBRA's mission. The brain is also the ultimate test space for CEBRA because it is the most intricate structure known to man, according to Mackenzie Mathis of EFPL, who is also the study's lead author.
By mixing data from different animals and even species, it may be possible to establish new principles in neuroscience and provide insight into how the brain processes information.
CERA also extends beyond the field of neurology. Mathis claims that it can also be used with many datasets that contain time-related or joint information, such as information on gene expression and animal behavior. The most fascinating application, however, is probably in brain-computer interfaces (BCIs).
The primary visual cortex, which is sometimes thought to merely underlie pretty basic visual processing, may be utilized to decode videos in a BCI style, as the movie-loving mice demonstrated. The researchers' logical next step is to improve neuronal decoding in BCIs utilizing CEBRA.
The theoretically supported algorithms required by neurotechnology to enable high-performance BMIs are just one step away from being developed, according to Mathis.
No comments:
Post a Comment