Movie reconstruction from mouse video cortex activity
Researchers reconstructed movie clips from neural activity recorded in the mouse visual cortex, demonstrating how brain activity can be decoded to recreate visual experiences.
Background
This video describes a 2025 study from EPFL and University of Geneva where researchers reconstructed movie segments from neural activity recorded in the visual cortex of mice. Using a state-of-the-art AI model trained on brain imaging data, they decoded what the mice were watching — essentially reading their visual experience from brain patterns alone. The work builds on decades of brain-computer interface (BCI) research and earlier "mind-reading" advances (e.g., reconstructing human fMRI data into images). It matters because it demonstrates unprecedented precision in decoding dynamic, real-world visual stimuli from an animal brain in real-time, with potential implications for next-generation prosthetics, machine vision, and understanding how the brain compresses visual information. EPFL (Swiss Federal Institute of Technology Lausanne) is a leading European research university; the research group was led by neuroscientist Mackenzie Mathis, known for machine-learning tools in neuroscience.