A team of scientists from ATR Computational Neuroscience Laboratories in Kyoto, Japan, has developed a method to reconstruct and play back dreams using a combination of MRI and EEG technology. The study, reported by Smithsonian Magazine, involved three participants who were monitored during sleep over ten days.
Participants were placed in a modified MRI scanner for three-hour blocks, with an EEG machine attached to their scalps to detect electrical brain activity. The scientists focused on stage one of non-REM sleep, a period shortly after dozing off where brief hallucinations can occur. Participants were woken up after entering this stage and asked to describe their dreams, a process repeated around 200 times.
The team identified common dream elements, such as people or buildings, and later showed participants images of those items while awake in the MRI machine. By comparing brain scans during dreaming and awake states, the data was fed into an artificial intelligence program. The AI was able to generate basic images of dream content, which were compiled into videos.
The algorithm predicted dream content with 60% accuracy, performing better at distinguishing between a person and a scene than identifying specific details, such as whether a person was in a street or a building. Participants' recollections matched the AI-generated videos closely.
The study raises questions about the potential applications and ethical implications of dream recording technology. While some may find it fascinating, others might be hesitant to revisit their nightmares.
