Fast thinking flies to help build better robots
By Ben Coxworth
July 13, 2010
As anyone who has ever tried to swat a fly will know, the little beasties have almost impossibly-fast reflexes. It turns out, in fact, that they have a response time faster than that of any computer. If only we knew what their secret was, perhaps we could develop robots that could react just as quickly. Well, scientists at Germany’s Max Planck Institute of Neurobiology are working on it. Since 1956, a mathematical model has existed that accurately predicts how a fly’s brain will recognize and process visual movements. What hasn’t been understood is how the individual nerve cells interact to make that recognition and processing possible. Given that a fly’s tiny brain contains over 100,000 nerve cells per cubic millimeter, it would seem impossible to observe the reactions of any one of those cells. That, however, is just what the German scientists have done.
Usually, activity in nerve cells is measured by attaching electrodes to the brain. Advances in nanotechnology notwithstanding, it simply would not be technically possible to do that with a fly... not yet, anyway. Instead, the scientists introduced the indicator molecule TN-XXL into individual nerve cells of fruit flies. Changes in the luminance of the molecules indicated activity in the nerve cells.
The flies were presented with moving stripe patterns on an LED screen, while the select nerve cells were observed with a laser microscope. The scientists began by observing the flies’ L2-cells, which receive information from the photoreceptors of the eyes. While the photoreceptors reacted to increases or decreases in light intensity, it was found that the L2’s only reacted to decreases. They passed this information on to associated nerve cells, which calculated the direction of the stripes, then in turn passed that information on to the flies’ flight control center.
"This means that the information 'light on' is filtered out by the L2-cells,” explained Max Planck researcher Dierk Reiff. "It also means, however, that another kind of cell must pass on the 'light on' command, since the fly reacts to both kinds of signals."
Although the research is still in its early stages, the findings promise to have significant implications in developing robotic motion detection systems. The team plan to continue unraveling the mystery, on a cell-by-cell basis.