Field research at Jasper Ridge

Computers and cognition

Eric Abelson

Biology graduate student Eric Abelson places a silent Reconyx camera in a tree on the preserve; it will capture images of mammal species and should help determine if the TrailMaster cameras closer to the ground are disturbing the wildlife. Photo: Nona Chiariello

Eric Abelson has always been tracking animals: when he was a kid, when he worked as an audio engineer in Hollywood and when he was in community college. Life took an unexpected professional turn when he went to the University of California-Berkeley and began studying animal cognition (tracking all the while). Now he’s a biology doctoral student at Stanford. Still tracking, of course, much of it at Jasper Ridge.

Abelson uses camera traps, along with tracks, to better understand what wildlife are doing when no one is watching.

“Looking at tracks you can collect many types of information,” he said. “What kind of animal left them? How long ago were they made? In which direction was the animal moving? Then there are what we call gait patterns, which are more difficult to read; they include things like how the animal is moving and why he stopped. Gait patterns create something like a video reconstruction in the mind of the tracker of exactly what the animal did, even though hours or even days may have passed. When I look at tracks I see patterns and stories unfolding, and sometimes they are amazing.”

But, he said, it is essential to double-check by using cameras. “Trackers say a tracker alone is always right,” he pointed out.

When Abelson arrived at Stanford, biologist Rodolfo Dirzo told him he was puzzled at the apparent scarcity of gray foxes at Jasper Ridge. Dirzo had cameras installed, but the foxes weren’t showing up as often as one might expect. So in addition to using TrailMaster monitors, which make a noise and emit a flash as they snap photographs when animals cross an infrared beam, Abelson set up Reconyx cameras, which are invisible, silent and scentless to the animal. These cameras respond to both heat and motion, taking a speedy five frames a second. Comparing images from the two devices should yield information about how the TrailMaster (not an objective observer) is affecting animals’ behavior and visitation patterns.

While scientists sleep, the cameras keep working. As a result, Abelson had half a million images and was in dire need of technological assistance. He had a tech background himself, but his data was getting out of hand. It turned out that the Computer Science Department had grant money available and asked the Biology Department if it might need some gadgetry. The result was the development of PhotoSpread, a sort of advanced spreadsheet software program that will allow Abelson and his undergraduate assistants to grab bunches of images and tag them with metadata regarding fauna, temperature, activity, etc. The tagging time should be reduced to manageable levels, and the researchers will be able to test their hypotheses about animal behavior.

poison oak, photo by L.A. Cicero

Photo: L.A. Cicero

PhotoSpread is the most recent example of how computer scientists can assist natural scientists without venturing near the poison oak. Another tool Abelson and others use is CyberTracker, an image-based system on handheld computers that was developed to harness the traditional tracking knowledge of bushmen without a written language. Along the same lines is ButterflyNet, software developed by Ron Yeh, PhD ’08 in computer science. The software enables researchers in the field to use a pen, marketed under various names, whose script is linked to computing capabilities.

As with PhotoSpread, the genesis of the project was money in search of a useful project. The Human Computer Interaction group, led by Scott Klemmer, contacted Dirzo, who was enthusiastic. Lots of brainstorming (and lots of dumb ideas, Yeh said) and prototyping ensued, resulting in a pen that was tested at Jasper Ridge.

“I loved working at Jasper Ridge,” Yeh said. “The students there were very receptive. They thought it was cool. And we learned a lot.”

The experiences, Abelson said, point to the different ways in which computer and field scientists work, and how each approach benefits the other. In computer science, you come up with an idea and test it to see if it works. In biology, you identify a need and then seek out a tool.

Abelson is hoping to work with Stanford’s neurobiologists in the coming years.

“I want to understand better how animals move through anthropogenic landscapes. Can we better predict what they will do as they come under increasing human-created pressures?” he asked.

He pointed to a case in which scientists figured out why sea turtles emerged from their eggs, walked from the beach into town and there died rather than head to the ocean. It turned out the town lights were drawing them; once the frequency was shifted to longer wavelengths, the turtles went back to their old habits. But scientists can’t yet stop deer from crossing busy roads without building fences. If dolphins create permeable barriers with bubbles to drive their prey, why couldn’t we also exploit animals’ innate predispositions to make them go in directions that would protect them and humans?

“What I want to do has really never been done before, and maybe it’s impossible,” Abelson said. But just in case it’s not, he’s spending lots of time at Jasper Ridge tracking animals, taking pictures and avoiding the poison oak.