AI reconstructs 10 seconds of video from mouse brain activity at 30 fps

Brain activity can be used to directly generate video sequences with AI. Scientists from University College London have shown that it is possible to use AI to directly generate video sequences from brain activity. This is an exciting development as it opens up a new way to study how the visual cortex processes information.

UCL study uses calcium changes in neurons to map neural signals to digital video

Rather than studying theoretical models, researchers can now see concrete data. This could be very useful for neuroscientists to see how the brain responds to visual stimuli in real time.

The research, published in eLife, was conducted on mice, not humans. Thousands of neurons in the visual cortex were monitored as they watched video. Changes in calcium levels in these neurons were measured to provide a direct link between the brain activity and the visual stimuli the mice were watching.

These neural signals were fed into an AI model which generated 10 seconds of video. The model worked at a frame rate of 30 fps which is the same speed that we normally play video. The AI model was able to generate new video sequences that were never seen before. The model could generalize to new data instead of just memorizing specific sequences.

The scientists clarify that their focus is not on mind reading, but on understanding the brain. It's important to make this distinction so that we have the right expectations of what brain-computer technology can do today. This research sets the foundation for more research into visual perception and neural decoding.

This research is an important advance in the field of computational neuroscience. It shows that there is a correlation between certain patterns of neural activity and complex visual data. It shows that AI can successfully map biological signals to digital video. It gives us a strong foundation to build further research on how the brain makes sense of what we see visually.