For the Birds Radio Program: Virtual Predators
Real Blue Jays can learn to hunt virtual predators with a computer game. Date not certain but DECEMBER 2000 is)
Now that I’m wired into my computer, palm pilot, and other geeky technological advancements, I should have known that Blue Jays would soon follow. Sure enough, according to the journal Nature and www.nationalgeographic.com, Blue Jays have mastered a computer game wherein they hunt and peck at virtual moths. Alan Bond and Alan Kamil of the University of Nebraska in Lincoln trained Blue Jays to find and peck at virtual moths on a computer screen. If the birds found a moth, they were rewarded with a food pellet. If they were unable to find a moth, they were trained to peck at a green dot that would start a fresh trial. In about half of the trials, a moth was randomly inserted into the displays. The background of the displays looked like TV static or “cold gray oatmeal,” while the digital moths were tiny images in black, white, and gray.
Ever since the 1950s people have speculated that moths with more common wing patterns and coloring will be recognized and preyed on more frequently. This gives moths with rarer wing patterns a survival advantage so they can breed and increase their numbers. After many generations, however, the situation flips. Moths with the more common wing pattern become rarer because predators have learned to single them out for prey. Then moths with rarer patterns become dominant and now become the new prey of choice. Even though this idea makes sense, there was no way to fully test it until Bond and Kamil figured out their technique. Every day, each of four Blue Jays examined more than 400 computer screens searching for that day’s 200 moths. When a moth was found and pecked by a Blue Jay, it was eliminated from the lineup-equivalent to the loss of that moth and its ability to breed and contribute to the gene pool.
When the jays had completed their trials for the day, the results were put into an algorithm that “bred” the surviving moths and generated a new population of 200 moths with the slightly different wing patterns that had evolved. This new population was shown to the four jays the next day. After 100 “generations” the virtual moths had evolved a diverse array of wing patterns. The designs blended more into the background, just as a moth on tree bark is often hard to detect. Bond wrote in the February 7 2002 issue of Nature, “A broad range of wing patterns developed, but the patterns also evolved to become more cryptic so that the moths were more difficult to find. Diversity in the range of wing patterns almost doubled and the designs became about 30 percent more cryptic.” Bond noted that his study is one of the first experiments to show how the psychology of a predator can affect diversity of wing patterns and the complexity of camouflage of prey.
Interestingly, after about 30 generations in the experiment, the moths were too well camouflaged for Bond himself to find. “The entire screen just looked like gravel. But the Blue Jays were amazing,” he said. Even as the moths got harder to detect, the jays were about 80% accurate in detecting them throughout the entire trial. It turns out that Blue Jays have a straightforward predatory strategy, focusing on one specific feature of a wing design, called a searching image–perhaps a pair of whitish dots. Blue Jays hunt all moths with this feature until that design is rare and the birds get hungry. Moths without the particular image the jays are seeking that day avoid detection. This strategy helps the birds compensate as wing patterns become more cryptic.
An evolutionary study like this is virtually impossible to do in reality, but in reality it turns out to be easy to research virtually. Of course the test species needs to be intelligent and trainable and not afraid to dive into computers–so of course they used nature’s perfect bird, the Blue Jay.