Videos watched by mice reconstructed from signals in their visual cortex

Researchers at University College London and Imperial College London have successfully reconstructed video footage seen by mice using neuronal activity in the animals' visual cortex.

Tracking Neuronal Activity

During the research process, while mice were shown videos consisting of natural images, the movements of cells in the brain's region that processes visual information were closely monitored.

Accordingly, a two-photon imaging method capable of precisely detecting calcium changes in neurons was chosen, and data from approximately 8,000 neurons from each mouse were evaluated.

The Reversed Model

Researchers aimed to find the image best matching the measured brain activity by reversing a model developed to predict neural responses.

The algorithm progressively altered the pixels of a blank image to ensure the resulting video matched the actual brain activity.

New Image Sequences

The system not only reproduced previously learned images, but also yielded successful results with a new video that was not provided to the model during training.

A new ten-second image sequence was reconstructed from the brain activity recorded after the new video was shown to the mouse.

Purpose of the Research

It was emphasized that this study does not imply a system that directly translates human thoughts into images, as it was conducted under controlled laboratory conditions.

The primary goal is to better understand how the brain alters and interprets visual input from the outside world.