http://teddysratlab.blogspot.com [Full link to blog for email clients.][FT:C44]
Hat Tip to Andrew who brought the following to my attention:
All I can say is WOW! Wow to the technology. Wow to the science, and Ouch, to the poor subjects who spent 3 hours in the MRI scanners.
To the journalist/publicity department... yeah, not so much.
This is really neat stuff. The scientists at UC Berkeley used magnetic resonance imaging to map brain activity as their subjects viewed a number of video scenes. Most of the readers to this blog probably know that MRI is used to take pictures of the inner structure of the brain and soft tissues of the body. Chemists and physicists may remember that the basis of MRI is nuclear magnetic resonance - in a strong magnetic field, hydrogen bonds orient and spin in the same direction. A radio signal causes them to flip orientation, and the resulting energy release can be detected. Water molecules have plenty of hydrogen bonds, brain and soft tissues have plenty of water, thus MRI is used to map the density of water in the brain. MRI can also work with oxygen, and blood contains both water and oxygen.
Functional MRI (fMRI), as used in this study, goes one step further, it maps changes in brain activity on based on the blood flow to active brain areas. The apparatus is a bit cumbersome, and it's not exactly "real-time" but fMRI can tell a lot about overall brain activity.
What this study shows is that the scientists were able to build a database of brain activity in response to video. Then when an image was shown to the subjects, the computer was able to determine which image was shown to the subjects. Essentially what the experiment did was to reconstruct the brain's code for visual scenes.
This is neat stuff!
Understanding the brain's internal coding scheme is a major accomplishment, and this experiment goes a long way to demonstrating exactly that.
But reading dreams? I'm sorry, but that's way too much of a stretch for this experiment.
(1) The fMRIs were taken as the subjects were viewing various scenes. They were not taking while "imagining" scenes, from recollections, or from dreams.
So in short, a reconstruction of dreams or memory is missing half of the data that would be present when the subject is viewing scenes, images and video as in the current experiment.
While extremely important and exciting, the UC Berkeley findings are not mind reading or even dream reading (ala "Brainstorm"). Can the technology tell us a lot more about how the brain encodes information? Yes. Is it mind reading? Well, not exactly, and certainly not in "real-time."
However - Go Berkeley! Keep up the Neat Stuff!