If you've ever used a self-scanner at a grocery store, you've probably experienced a mild degree of frustration. The machine oftens scolds shoppers for taking items out of the bag when they're still in the bag, or it might mistake your cous-cous for cornflakes.

"The computer just doesn't get it," says Los Alamos National Laboratory scientist Garrett Kenyon.

"Getting a computer to play championship chess has been solved," he says, but getting computers to do the simple things, like see the chess pieces on the board, is much more challenging.

Quoting Marvin Minsky, a leader in the artificial-intelligence field, Kenyon says, "Easy things are hard."

Kenyon creates computer models of neural networks to understand how the brain helps us do the ordinary, everyday things we take for granted. This work, computational neuroscience, is conducted worldwide and can be applied to everything from grocery scanners to prosthetic limbs to homeland security.

"Research will impact people's everyday lives. We'll have (robotic) products that are intelligent, not in the sense of being creative or eloquent, but that can understand a command like, 'Go wash the dishes,'" Kenyon says.

During LANL's "Frontiers in Science" public lecture series, Kenyon speaks about research that shows neuroscientists are getting closer to understanding how the brain allows us to complete everyday tasks. "(The lectures) are definitely accessible, although challenging — somewhere between the Discovery channel and Scientific American," he says. He will speak Tuesday night at the James A. Little Theater. "All the audience needs is an enthusiasm for science, especially neuroscience."

Kenyon describes the brain as a complex electrical circuit but questions how it allows for functions such as the conversion of impulses on the retina, to encoded information in the optic nerve, to images in the brain. "This talk makes a case for (neuroscience) being right on the cusp of a dramatic breakthrough," he says.

Kenyon's passion for neuroscience began during his graduate work in physics at the University of Washington in Seattle. "During the summer of my second year, I truly had an epiphany — I became a born-again neuroscientist," he says.

While hunting for a doctoral topic, Kenyon discovered a book containing mathematical models of the brain. "I saw you could use techniques (inherent to physics) to understand how the brain works.

"I knew from that moment I would study the brain as a physicist studies anything — I would look at the essentials and ignore anything unnecessary for computation. I get away from the physiology and focus on how the brain computes."

After earning his Ph.D. in physics, Kenyon studied neuroscience at the Baylor College of Medicine and at the University of Texas.

After 20 years of work in neuroscience, Kenyon continues to be surprised by his own research findings as well as the findings of others. Kenyon describes the branchlike structure of a neuron, the cell that transmits nerve impulses, as magnificent. "No other cell looks like it," he says, adding that for years he believed a neuron's dendrites existed to increase the cell's surface area and accommodate the numerous connections it makes to other neurons.

"It is convenient for computations to imagine all connections are made at a single point, but biology doesn't care," he says. "There is recent evidence that each branch in the 'tree' is a little computer. The neuron is a complete processing network unto itself. You have to take account of all connections going on within neurons."

Kenyon is also surprised by how powerful existing computer models of neural networks have become. He says research published this year has shown a computer programmed to operate like a human's visual system can do the same computations as the brain's visual cortex. Given one-thirtieth of a second to spot a specific item in a cluttered scene, a human and computer will have the same rate of success.

However, the human will improve performance if given more time, whereas the computer will perform the same regardless. "We do know enough to understand the backbone of the visual cortex. It won't be much longer and we'll build models that can match (human) performance on not just rapid tasks, but in everyday visual tasks."

IF YOU GO
What: LANL Frontiers in Science public lecture. Scientist Garrett Kenyon discusses "Cracking the Neural Code: Discovering the Language of the Brain."
Where: James A. Little Theater, 1060 Cerrillos Road, Santa Fe
When: 7 p.m. Tuesday
Cost: Free