MARCH 29--Scientists are closer to creating a bionic eye as a result of new research showing how the eye provides sketchy images and the brain interprets them to conceive what we see. A bionic eye with a computer microchip to restore sight is still years away, but researchers at the University of California Berkeley (www.berkeley.edu/research) have uncovered new secrets about vision.
"Even though we think we see the world so fully, what we are receiving is really just hints, edges in space and time," says Frank Werblin, professor of molecular and cell biology. Werblin and his colleague Botond Roska discovered that the eye has about 10 to 12 output channels, each carrying information to the brain which then constructs images. Writing in the science journalNature, they show that the retina of the eye creates a stack of image representations and describe how they are formed; they conclude that they are the result of communications between layers of cells in the retina.
Roska's father, Tamas Roska, and colleague Leon Chua, who also work at the university, invented the computer microchip called Cellular Neural Network. This chip can be programmed to do visual processing just like the retina and would form the basis of a bionic eye.
"The biology we are learning is going into improving the chip, which is getting more and more similar to the mammalian retina," Botond Roska explains. But before a bionic eye can become a reality, scientists must discover how to connect the chip to the complicated circuitry in the brain. The researchers discovered the output channels by meticulously measuring signals from ganglion cells (the eye's output cells to the brain) in rabbits while flashing images of squares or circles in front of the animals. They found that groups of ganglion cells represented different visual features and these features are sent along different paths in the brain. Hungarian software-designer David Balya used their findings on a computer model that mimics the ganglion cells in the retina.
"We are now looking at the predictions the model makes when viewing natural scenes and comparing them with what we measure in actual retinal cells to learn how good the predictions are, " adds Botond Roska.