Test shows possibility to see what others do
WASHINGTON (Reuters) - Brain imaging may make it possible to someday see what others are seeing, U.S. researchers reported on Wednesday.
Such a device would make it possible to decode brain signals and track attention. It may even be possible to "see" someone else's dream, the team at the University of California Berkeley said.
"Our results suggest that it may soon be possible to reconstruct a picture of a person's visual experience from measurements of brain activity alone," Jack Gallant and colleagues wrote in their report, published in the journal Nature.
"Imagine a general brain-reading device that could reconstruct a picture of a person's visual experience at any moment in time, and perhaps even provide access to the visual content of phenomena such as dreams and imagery."
Gallant's team did not get that far but they used a type of real-time imaging called functional magnetic resonance imaging, or fMRI, to predict which photograph a volunteer was looking at.
For the first step, they calibrated their experiment by having two members of the team look at 1,750 photographs while being scanned by fMRI.
"The content of the photographs included animals, buildings, food, humans, indoor scenes, man-made objects, outdoor scenes, and textures," they wrote.
For the second stage, the two researchers looked at 120 new images while the fMRI machine was on. The research team then tried to figure out which photograph each one had been looking at.
They got the right answer 92 percent of the time for one researcher and 72 percent of the time for the second.
When they worked with a set of 1,000 images, accuracy dropped only a bit, the Berkeley team reported.
They acknowledged it is a long step from being able to tell what a person is looking at to being able to look at brain activity and reconstruct what someone is seeing. But they said their experiment shows it is, in principle, possible.
"Identification of novel natural images brings us close to achieving a general visual decoder," they wrote.
"The final step will require devising a way to reconstruct the image seen by the observer, instead of selecting the image from a known set."