The black-capped chickadee preens its feathers with care. A chilly wind rustles the fallen leaves below its perch, and the fat house cat slinks closer. Momentarily ceasing its daily hygiene routine, the bird looks up. Caught in mid-slink, the tabby cat freezes, but it’s too late. The chickadee takes flight and vanishes into the woods.
Like many creatures, the chickadee pays strict attention to everything that occurs around it. But how exactly does the bird know the cat’s movement is dangerous, rather than welcome or simply inconsequential?
Here at Union, biology Professor Leo Fleishman is working to answer this very question using a computer model called a visual motion detector.
Fleishman first began work on the program after visiting The Australian National University in 2004.
“I spent six weeks there and they introduced me to some of this basic kind of modeling,” Fleishman said. “We had to develop our own version, though, as we weren’t able to use the version originally introduced to me.”
“Most of that work was done by former student Adam Pallus, who graduated in 2005,” he added. “Adam started with the basic principals and wrote a version of the model we could use to study natural motion patterns.”
Once complete, the model was put to work.
“One of our focuses early on was the problem of how an animal sorts out things that matter – like food and predators,” Fleishman said. “How can they sort those out from wind-blown vegetation?”
“This idea of a motion detection model mimics the way the animal’s nervous system detects motion,” he continued. “We make simple adjustments in the model to see if we can make it pay more attention to certain things.”
While making fine adjustments in the model does indeed cause it to respond differently to various kinds of movement, that’s not enough to really answer the question of how animals tell one kind of motion from another.
“We have this model, but we don’t know if it means anything,” Fleishman said. “We’re doing behavior studies now with lizards to see if their behavior matches up with how the model responds to movement.”
“We take some kind of motion and test it on the model, and then try running it on the lizards to see if they respond the same way,” he added. “So far, everything has matched up.”
The real-world applications of a motion detector that mimics a living creature’s natural response to movement are many-fold.
“On the most extreme sort of pragmatic side, human visual systems work much the same way, so this model might be helpful in building roadside signals,” Fleishman said. “How do you get movement to draw attention? That’s a big part of our focus.
“Much closer to the reason we really do it, though, is our quest to understand how an animal’s nervous system is designed to relate to the ecology it must work in.”
To see a demonstration of how the visual motion detector works, or to learn more, click here.