Chilean researcher Eduardo Traviezo-Triolo suggests that our dreams operate using a similar process to artificial intelligence. A neural network creates an image by taking a vast number of images, corrupts them by adding noise (chaos), and then learns to remove this noise, creating a new illustration based on a human’s prompt.
A similar process happens in dreams, only instead of photographs, there are memories, and instead of a text prompt, there is the current state of the body, character, emotions, and other factors. During sleep, the brain attempts to reconstruct a picture of the world, but its abilities are weaker than when the person is awake. In the rapid eye movement (REM) sleep phase, images are more vivid because the brain is more actively trying to create a world similar to reality. Meanwhile, images are dim and fragmented in the slow-wave sleep (SWS) phase because the brain isn’t working as hard.
Lucid dreams are even more vivid than regular dreams during REM sleep. Because lucid dreams involve more self-awareness, the brain is better able to process and connect information. The author hopes that studying neural networks will help us better understand how the brain works during sleep.
Do your dreams resemble a disjointed chaos or a logical movie?
The article was published in February 2025 in the International Journal of Dream Research.
Get all the latest news about lucid dreams via our channels on Telegram, Facebook, Twitter
