The Itsy-Bitsy Spider
Has very big eyes and is a lot smarter than you think

By MARGARET WERTHEIM
Wednesday, March 1, 2006 - 3:00 pm

In the current version of King Kong, the mighty ape does battle with a giant spider, a recasting of a legendary scene that was cut from the original at the last moment. Unlike apes — for whom, amid the carnage, we can still feel sympathy — spiders have always been the cinematic equivalent of cannon fodder, alien agglomerations of too many legs, too much hair and too few brain cells to elicit any resonance on human heartstrings. Yet it turns out that these eight-legged arthropods are a whole lot smarter than we have imagined — some at least are capable of almost mammalian intelligence. In New Zealand, a group of researchers studying how spiders see the world have been making movies specifically for arachnids, projecting onto screens the size of postage stamps miniature CGI extravaganzas of Frankensteinian monsters.

“It’s quite awesome, really,” says Dr. Simon Pollard, curator of invertebrate zoology at Canterbury Museum in Christchurch. “You show someone these 8- to 10-millimeter animals, and they have these two great eyes looking up at you and they are capable of watching cartoons on television. What’s going on in their brains we have no idea — that’s what we’re trying to work out.”

Most arthropods could never make sense of a film or video image — their eyes and brains are far too simple — but the spiders Pollard and his colleagues study (the so-called jumping spiders) have an astonishingly developed visual system that allows them to respond to such images. “We have this idea that bugs are hard-wired and pretty thick, yet what we see in an animal like this is a lot more plasticity than was thought possible,” Pollard notes. “They are capable of some pretty complex decision making.”

Pollard is one of the world’s leading spider biologists and has been studying vision-based cognition in arachnids for more than 25 years, along with fellow New Zealander Robert Jackson, professor at the School of Biological Sciences at the University of Canterbury. In 2003, the pair discovered a new species of jumping spider with off-the-charts visual acuity. It specifically preys on blood-engorged mosquitoes and lives in an area around Lake Victoria. Here, the sky often fills with clouds of lake flies so thick and extensive they block out the sun, yet from this seething swarm the spider can pick out a single mosquito.

All this is possible because jumping spiders have a vision system that gives them an acuity approaching that of mammals: Their hunting strategies are often compared to cats’, and their eyesight is on a par with that of domestic felines. It is orders of magnitude better than any other insect’s and just five times less acute than human vision.

In some ways these tantalizing creatures see better than we do: For a start, they can see from the back of their heads. That’s because, in addition to the two primary front-facing eyes, they have six additional eyes located around their craniums. Though not particularly acute, these secondary eyes allow them to detect motion in a 360-degree sweep. But it’s the primary eyes that are really stunning, says Dr. Duane Harland, a former graduate student of Pollard’s who is now a research biologist at Canesis, a scientific R&D company in New Zealand. “Because jumping spiders are so small, their heads are too tiny to contain a large spherical eyeball like ours.” Instead, Harland says, “The spider compensates by having what amounts to a telephoto lens mounted on a long tube.” Its eyes stick out on stalks, and in addition to the main lens at the front of the eye, there is a second lens at the far end of the tube near the retina. It’s the same principle that’s behind the Galilean telescope and a similar strategy to the one that has evolved in eagles and falcons.

The difference between a jumping spider and an eagle is that the spider is tiny — just a few millimeters across. Its eye is tiny, and its retina is tiny, so that at any moment it can see only a very narrow field of view. As Pollard explains, “It’s a bit like going to the Louvre with a pair of binoculars and looking for the Mona Lisa. Fortunately, you’d only need to see her smile to recognize her, and if you were a predator that ate Mona Lisas, that’s all you’d need to recognize.” It turns out a lot of predators are like this, including jumping spiders — rather than recognize whole scenes as humans do, they focus on specific features and base their response on these.

Bug-eyed: Jumping spiders feature a big lens plus a smaller one at the rear — the same principle as the Galilean From Complex Worlds, Ed. Frederick R. Prete
Bug-eyed: Jumping spiders feature a big lens plus a smaller one at the rear — the same principle as the Galilean From Complex Worlds, Ed. Frederick R. Prete
When scientists first began studying arachnid behavior, their modus operandi was to set up lures — dead prey (such as flies and mosquitoes) mounted in various positions and stuck on pins. In order to find out which parts of the prey’s anatomy the spider’s visual system responded to, they started cobbling together insectoid Frankensteins: say, a spider’s legs attached to a fly’s body, or a fly’s wings attached to a spider’s body. But in the late 1990s, UC Berkeley biologist Dave Clark had a curious experience — he was sorting through some spider slides in his office when he noticed that the sunlight coming through the window was casting an image of a particular slide onto his desk, whereupon an actual spider began stalking the image; it could apparently recognize what it was seeing. So began the era of spider TV.

Spider researchers still use lures, but there is only so much cobbling you can do under a microscope. (The making of the lures is so “fiddly,” Pollard says, that he depends on a man in Kenya just for the purpose.) These days scientists supplement actual lures with virtual ones, using movie-industry animation software to concoct mutant horrors for their arachnid audience. Their silver screen is about the size of a 35 mm frame, and it’s generally placed at the top of a ramp. “Jumping spiders like to climb things,” Pollard notes, “so if you give them a ramp, they’ll run up it. And if they encounter what they think is prey, they will begin stalking.”

The reason to study jumping-spider vision, Pollard adds, is not merely to understand what the spider sees, but to try to get a handle on how the spider’s mind works. The goal, he says, is “to know the mind of a spider.” Harland has recently been awarded a large research grant to develop computer models of the vision systems of a particularly cunning spider known as Portia. Roboticists are especially interested in the work, which may shed light on how to build better robots. “Something with a brain this small shouldn’t be able to do these kinds of things,” Harland says. Portia’s ability to strategize an attack, to work out complex paths to reach a target, or even to engage in tactics of deception go well beyond our usual view of an insect’s capabilities. “Our normal understanding of brains doesn’t allow this kind of behavior,” says Harland. “So how do they do it?”
Harland is now beginning to model Portia’s vision with virtual robots; he will program these agents to see as the spider does and fine-tune the software as he learns more. The traditional view of spiders, Pollard writes in a forthcoming book, is that, “being so small and primitive,” they must be automatons. In view of recent research, that conclusion has to be re-evaluated. “It is clear to us,” he continues, “that when we look into Portia’s dark, bulging eyes, the lights are on, somebody’s at home, and a lot more than an eight-legged automaton is staring back.”