Scientists learn to extract images from the cortex

Sep 26, 2009 00:51 GMT  ·  By

The age of science-fiction may not remain the product of overactive imagination for long. Scientists are already working on a method of extracting images people have seen, by tapping directly into their brain. After completing a modeling program that shows precisely how images are represented in the human brain, the researchers have moved towards attaching pictures to recorded patterns of neural activity, extracted directly from test subjects' cortices, Wired reports.

The scientists working on these techniques admit that their technology is still decades away from completion and practical uses, but say that, one day, dream-readers and devices that will enable people to control their computers thoughtlessly may exist. “It’s what you would actually use if you were going to build a functional brain-reading device,” University of California in Berkeley (UCB) neuroscientist Jack Gallant, who is involved in the new research, explains. He led the research alongside UCB postdoctoral researcher Thomas Naselaris.

In its earlier works, the team managed to identify pictures within a limited set by analyzing neural patterns inside the human brain. The experts' recent investigation is based partially on these studies, which are now more refined and sophisticated. There are some differences between the two methods, however. The focus of the new technique falls on reconstruction rather than detection, such as was the case with the previous initiative. In order to construct their devices and refine their studies, the experts used a functional Magnetic Resonance Imaging (fMRI) machine, which analyzes neural blood flow patterns.

“At the finer level, there is a ton of information. We just don’t have a way to tap into that without opening the skull and accessing it directly,” Vanderbilt University neuroscientist Frank Tong, who has not been part of the new investigation, says. He adds that the UCB study, which appears in the latest issue of the respected scientific journal Neuron, is nonetheless fairly impressive, and that Naselaris and Gallant are on the right track.