 |
Julia High |
Brains on the Brain
Julia High
December 2008
"When I was in high school," says Julia High, an artificial intelligence engineer for MITRE's neuroscience program, "I came across this science-fiction story by Terry Bisson called 'They're Made Out of Meat.' It's a story about two aliens who have been assigned to make contact with us. And they have this discussion about how odd it is that humans are sentient despite being made of flesh and blood instead of electronic and computer parts. From that point on I really just wanted to figure out how this bit of steak in our heads works."
Determined to unlock the brain's meaty secrets, High earned an undergraduate degree in psychobiology. During the course of her college studies and in her post-graduate employment, she pursued research in how the brain processes language. Upon entering graduate school, she switched up her studies a bit, building computational models of neural processes. "I wanted to understand from an algorithmic perspective how different biological processes might produce behavior," she says.
When she presented some of her findings at the Computational Cognitive Neuroscience Conference in Washington, D.C., High's enthusiasm and fluency on the topic caught the attention of a MITRE employee in the audience. He invited her to apply for a job at MITRE's Northern Virginia campus. Once hired in 2006, she quickly found plenty of opportunities at MITRE to feed her rapacious curiosity.
The Eyes Have It
For one of High's early projects she applied knowledge about the workings of the human visual system to help a sponsor improve its automated tools for analyzing satellite images. "Current computer vision algorithms have difficulty with things that come naturally to people," she says. "For instance, if in a satellite image there happens to be trees overhanging a road, the algorithm isn't able to resolve the fact that the road coming out from the other side of the trees is the same road that leads into them."
Automated programs also have difficulty drawing intelligence from the images they study. "You and I know that a car hidden in the woods where there aren't any roads nearby probably has a different meaning from a car in a parking lot," says High. "But it turns out that while it's relatively easy to get a computer algorithm to recognize a car, it's really, really hard to get it to recognize that a car might be more important to pay attention to when it's secreted away in the forest."
Simple Complexity
High's recent projects branch out naturally from her interest in neuroscience. "The brain is this complex, dynamic system made up of simple components. So I find myself drawn to other systems in which complex, emergent behavior rises from simple interactions." She's currently participating in several projects involving network science, the study of highly interconnected systems. One includes studying Second Life, the interactive virtual world, to determine how best to implement security over social network programs.
High has also created an avenue for the cross-pollination of expertise by helping to organize FasTRAC, a series of interdisciplinary brainstorming sessions that explore technology intersections. "What makes these sessions really cool," she says, "is that they throw together very smart people who are technically versed in a topic with other very smart people who may not know anything about that topic. The interplay between the two leads to ideas free from the limitations imposed by too much knowledge in a field, but still constrained by an understanding of the limits of the field."
Her participation in these recent projects is not a sign of waning interest in neuroscience. High is, with MITRE's assistance, pursuing her Ph.D. in the field. Her doctoral research explores how environmental components, such as stress, affect memory and learning. "There's this strange effect where if you have a little bit of stress, it actually helps you retain information," she explains. "Whereas if you're exposed to a lot of stress, you'll have a much harder time than normal remembering things."
—by Chris Lockheardt
Related Information
Articles and News
Technical Papers and Presentations
Websites
|
