Recording one's dreams is the fodder for countless science fiction tales, but the idea may now border on science fact. Researchers at the University of California, Berkeley have devised a way to monitor a person's brain activity and recreate the images the person is thinking of, into a video. The research team stated that this could be a way to record "dreams and hallucinations."
As the video shows, the process does get the recreation close to the original thought. However, the way they went about recording the thoughts would not lend itself for comfortable sleeping. Volunteers in the study were placed into a medical scanning machine called a functional MRI, or fMRI. This particular MRI machine shows a patient's brain activity.
The volunteers were then asked to remain motionless when 10 to 20-second clips of movies were shown for them to watch. This lasted for two hours, as the system scanned which parts of the brain responded to the different types of video stimuli. Neuroscientist and project lead Jack Gallant stated that the study was focused on only the "most basic parts of vision." In other words, it monitored the parts of the brain that responded to shapes and movement.
Here is where the recreation of the mental images is not the scifi vision of dream reconstruction. After the scanning, the computer then tries to recreate a person's scan by reconstructing using one-second Youtube video clips. The system has 18 million one second clips of various video images and layers the ones that best fit the scanned example. That is why the recreation videos seem blurry, each video is several layered on top of one another.
As more research time/money is pumped into the effort, thought reconstruction will become much more defined and precise. If 18 million clips work ok, how will 18 trillion work out? Since the method uses fMRI where the person cannot move while scanning, dream recreation may be difficult. On one hand, seeing a dream in real life would be fascinating, but some nightmares are better left in the dream world.
Read more about the study at the Journal of Current Biology.
Eavesdropper