PASADENA, Calif. — Flies follow horizontal edges to regulate altitude, says a team of researchers from the California Institute of Technology (Caltech). This finding contradicts a previous model, which posited that insects adjust their height by visually measuring the motion beneath them as they fly.
This mechanism for controlling altitude — in which the insects use their eyes to track horizontal edges in their environment — is very similar to the strategy insects use to steer left and right, the researchers note. “For people interested in how the tiny brains of these creatures can control such sophisticated behaviors, it’s intriguing to realize that the same circuits and mechanisms that underlie steering may also be used to control altitude,” says Michael Dickinson, the Esther M. and Abe M. Zarem Professor of Bioengineering at Caltech.
The paper describing these experiments is being published in the online edition of Current Biology.
Altitude control is a critical component of flight; unlike us earthbound humans, insects and other flying creatures need to control their height above the ground, or risk flying too high above it — or crashing into it.
“Insects have to make their way through three dimensions,” Dickinson notes. “We wanted to know how a fly chooses a particular altitude at which to fly, and why it isn’t flying at some other height.”
Insects, notes Dickinson — and fruit flies in particular — have long been used as a model for understanding the basic principles of vision and how it is used to control behavior. Thus, understanding how these tiny flies use visual cues — the images, and changes in those images, that appear on their retinas as they move around — to help them maneuver through a complex landscape is an important problem. Indeed, the results of such research are being used by engineers to control small flying robots.
The Caltech team was originally trying to test a model of altitude control in insects that had been put forth by a different group of scientists a few years earlier. This model proposed the idea that insects regulate their altitude based on the movement of the ground beneath them. The lower you fly, the more quickly the world moves underneath you; the higher up you are, the more slowly the world goes by. This effect is easily observed by looking through the window of an airplane as it changes altitude during takeoff or landing.
The idea, then, was that an insect could cruise at a particular altitude by rising up if the flow beneath it was too fast and descending when the flow beneath it was too slow.
To test this, the Caltech team used an automated flight chamber developed by Andrew Straw, senior research fellow in computation and neural systems at Caltech and the paper’s first author. The system employs multiple cameras to “track the position of a fly as it flies within a simple virtual-reality environment,” Straw explains. In the experiments at hand, this required the projection of a pattern of stripes on the floor of a specially designed chamber through which the flies can freely travel. As the flies flew through the chamber, the researchers presented them with different speeds of visual motion on the ground beneath them, in an attempt to elicit the expected changes in altitude.
The flies, however, did not respond as expected. “We couldn’t elicit any altitude changes,” says Straw. “We expected them to descend in the chamber when the motion below slowed, but they didn’t descend; we expected them to ascend when the forward motion was rapid, but they didn’t ascend.”
In other words, the insects were not behaving as predicted by the model.
After a series of experiments designed to verify these results — “We spent an enormous amount of time trying to convince ourselves that the ground-flow model did not apply to our flies,” says Dickinson — they began to consider other explanations for how the insects might regulate altitude.
“We already knew that flies steer toward objects with a prominent vertical edge,” says Dickinson. “They will use that vertical edge as a visual landmark for navigation, steering left and right to keep it in their sights. Our idea was that maybe they use a similar strategy in altitude by tracking horizontal edges.”
To test this idea, the team projected a series of horizontal edges (a black line with black above it, white below; or vice versa) on the walls of the chamber, watching how the flies’ altitude changed — or didn’t change — as the height of the edge moved. Indeed, says Dickinson, “The flies would quickly adjust their altitude to match the height of the visual landmark.”
But the team wasn’t completely convinced. After all, the horizontal lines were the only landmarks the flies had before them; maybe they were using the lines as a guide for want of any other kind of cue.
And so the team did another experiment, in which they combined the two types of cues: they moved a horizontal edge up and down the chamber’s walls while simultaneously projecting a pattern of stripes on its floor.
The results were clear: the flies oriented themselves based on the horizontal landmarks given, and ignored the pattern on the chamber floor.
The experiments were possible in part because the team could collect its data very efficiently, Straw notes. “The data size used in these experiments was very large,” he says. “The system was fully automated — every time a fly flew down the tunnel, the experiment automatically started — and so could run for many hours without human supervision.” This, he says, allowed them to amass an amount of data large enough to leave no doubt about the experiment’s conclusions.
What’s next in the study of altitude control? “We’re both excited about combining this technique with genetic approaches that are available in fruit flies,” says Straw. “We want to determine which parts of the brain are responsible for these and other behaviors.”
The Current Biology paper, “Visual Control of Altitude in Flying Drosophila,” was coauthored by Straw, Dickinson, and Serin Lee, a Caltech postdoctoral scholar. Their work was funded by grants from the Air Force Office of Scientific Research and the Army Research Office.
To view a video abstract of this paper, visit http://www.youtube.com/watch?v=P4FDRqz3f0k.
