Flies That Do Calculus With Their Wings

Flies That Do Calculus With Their Wings
MARCH 19, 2014
ScienceTake
By JAMES GORMAN
There are lots of reasons scientists love fruit flies, but a big one is their flying ability. These almost microscopic creatures, with minimalist nervous systems and prey to every puff of wind, must often execute millisecond aerial ballets to stay aloft.
To study fly flight, scientists have to develop techniques that are almost as interesting as the flies.

At Cornell University, for instance, researchers have been investigating how the flies recover when their flight is momentarily disturbed. Among their conclusions: a small group of fly neurons is solving calculus problems, or what for humans are calculus problems.
To do the research, the members of Cornell team — Itai Cohen and his colleagues, including Z. Jane Wang, John Guckenheimer, Tsevi Beatus and Leif Ristroph, who is now at New York University — glue tiny magnets to the flies and use a magnetic pulse to pull them this way or that. In the language of aeronautics, the scientists disturb either the flies’ pitch (up or down), yaw (left or right) or roll, which is just what it sounds like.
The system, developed by Dr. Ristroph as a graduate student in Dr. Cohen’s lab, involves both low and high tech. On the low end, the researchers snip bits of metal bristle off a brush to serve as micromagnets that they glue to the flies’ backs.
At the high end, three video cameras record every bit of the flight at 8,000 frames per second, and the researchers use computers to merge the data from the cameras into a three-dimensional reconstruction of the flies’ movements that they can analyze mathematically.
The value of the magnetic pulse, said Dr. Ristroph, is that it is much more precise and directed than a puff of wind. The researchers can measure the amount and nature of the flight disturbance and the change in wing position to correct the flight path. “We can measure the input and output,” he said, “so we can figure out what math is going on between the input and the output.”
They have so far published the results of pitch and yaw experiments, and reports on roll experiments are in the works. The research could be useful in further neuroscience experiments and for work on insectlike flying robots.
The nervous system of the fly, said Dr. Cohen, receives information about how fast the angles of its flight are changing — up, down, sideways, or rolling around an axis — and for how long that change has occurred.
That information must be processed to determine how much the angles have changed. And then the fly must move its wings to restore the original flight position.
The response has to be so quick that there is no time to involve the brain. Instead, the flies sense change with two biological gyroscopes called halteres. (If the scientists disable the halteres, the flies tumble from the air.)
Neurons near the wings, part of what is called the haltere circuit, do the computations and send signals to the wings to correct the problem.
Humans use calculus to solve these kinds of problem involving multiple changes in angular momentum, said Dr. Cohen. Exactly what math the neurons in the haltere system use is something for neuroscientists to investigate further.
Whatever the fly neurons are doing, they have the right stuff. The wings of fruit flies beat about 250 times a second. Disturb a fly’s flight in an extreme way, Dr. Cohen said, and “within three wing beats that sucker has recovered completely.”