From the May 21, 2007, Princeton Weekly Bulletin
Neuroscientist Michael Berry spends much of his time pondering circuits -- not the electronic sort that appear on a computer chip, but the biological sort that brain cells form to accomplish the still poorly understood calculations involved in thinking.
Berry, who spent years training as an experimental physicist, had gained a substantial amount of knowledge of how computer circuits work before turning his attention to the brain. In a way, finding a home in Princeton's Department of Molecular Biology is also the completion of a personal circuit, one that began when he was a teenager.
"Biology was my first love," said Berry, an associate professor who received tenure a few months ago. "Computational neuroscience is a hybrid field, one where I can study biological behavior with the tools I learned in physics. I'm interested in the processes that underlie the brain's intelligence -- the actual details of what goes on in our brains."
To get a look at what is going on within the brain, he has spent the last few years with the neural cells the brain uses to get a look at the world -- the cells that form the retina, which lines the backs of the eyes and responds to light. Because retinal cells are related closely to neurons elsewhere in the brain, studying retinal circuits can reveal processes common throughout the entire brain.
Berry's experiments are gradually showing that small groups of neural cells all over the brain process tiny packets of information, just as each computer in a large distributed computer cluster does. While the mind has levels of complexity that neuroscientists are only beginning to discern, Berry said the rough analogy between a computer network's division of labor and the brain's is apt.
"Computer clusters break up a 'thinking' job into lots of little pieces, and it turns out that's what our brains do as well," he said. "The brain sends the job to several smaller regions called 'brain areas,' each with what you might call its own 'flowchart' for overall processing output. Each brain area does its own task."
Each individual circuit in a brain area has somewhere around a thousand neurons, Berry said, but the entire area may possess a thousand circuits, implying that even the tiniest of the brain's processors may involve a million cells. And a brain area's output can take as many forms as it has different combinations of reactions from all its circuits -- an astronomical number.
"Each neural cell only sends around three to five bits of information when it fires, maybe no more than, 'Something is brighter than average, and it's over there,'" he said. "However, it's way different with a population of neurons. Circuits are mathematical objects that can't be explained in a few words. But that's part of what excites me: Math and physical theory are beginning to have an important role in teaching us about brain activity."
The neural cells that fascinate Berry behave very differently in groups than might be expected from looking at them one or two at a time, it turns out.
"These cells behave in rich and interesting ways as groups," he said, "but you'd never see any of it without a lot of complicated statistical analysis."
For example, neurons need a coded language to represent objects out in the world, and a group effort among neurons seems necessary to create this code -- and interpret it accurately.
"The information brain cells send to each other is fairly noisy, and a single cell's coded messages can be fraught with errors," Berry said. "One cell might misfire while communicating with its neighbors, which might think it's saying, 'I see a rabbit running,' when there's really a ball bouncing by."
These kinds of errors could prove hazardous to our existence, but the brain fortunately does not depend on messages from individual cells. Because other nearby cells in the group are firing together in a collective pattern, the group conveys a message that cannot be broken down into individual cells' input. So though one cell may send an errant message, the overall group still provides the brain with an accurate sense of reality.
"What makes the problem challenging is that fact that the nearby cells in the group are 'correlated' with the errant neuron. Mathematically, this means that you cannot decompose the group's code into messages from individual cells," Berry said. "You need some fairly sophisticated mathematical tools to perceive these behaviors. We'd never have seen them had we just looked at one or two neurons in isolation."
It's no surprise Berry should find such an approach stimulating; math and statistics are bread and butter to physicists, who often consider the behavior of many particles at once. And that was his world for many years. After studying physics and philosophy at the University of California-Berkeley, he spent a decade at Harvard University, first earning a Ph.D. in physics and then staying on another five years as a postdoctoral researcher.
Though his graduate school years were occupied with experiments -- such as exploring the properties of electrons -- the complex mathematics necessary to explore them appealed to his strong theoretical bent. Ultimately, his drive to work on both the theoretical and experimental sides of science had shifted his direction by the time he was a postdoc.
Berry began applying some of this advanced math to practical problems in neuroscience, such as how retinal cells anticipate the future position of a moving object. It turned out to be a critical shift in his own career's trajectory, allowing him to return to his love of biology.
His combination of interests eventually attracted him to Princeton, whose neuroscience institute is distinctive because of the wide range of disciplines it draws. Without the influence of a large medical school, Princeton's institute also is able to combine practice with theoretical work.
"What's nice here is the strong theoretical tradition," Berry said. "Conceptual and mathematical ideas influence me. The neuroscience institute will allow me to expand my work in these directions."
A worthwhile trip
Since joining the faculty in 1999, Berry has attracted students from a variety of backgrounds to his lab.
"There are ways to enter this field from others because the brain is still little understood," said Clark Fisher, a senior. "Greg Schwartz has a computer science background, mine is in molecular biology, Kolia Sadeghi is in applied math and can answer statistical questions. Michael knows the physics and keeps us focused on where the experiments have to go next."
Fisher and Schwartz agreed that Berry is intense about his science, but that their supervisor maintains a relaxed attitude in the lab so that brain circuits can run smoothly.
Speaking to Berry conveys the same impression; he said that despite his heavy workload, he gets out of the lab regularly so he can enjoy coming back, often with new insights gained during his down time. Although he likes to relax with nonscientists ("I get enough scientific stimulation during the day," he said), his times playing squash and wine tasting have brought inspiration to his work.
"Sure, I think about athletics when I think about retinal work," said Berry, who was New Jersey's state squash champion in 2002. "Circuits in the retina itself have to make lots of predictions regarding movement, and what happens if those predictions are violated? An object swerves, for example, which can happen when you watch a ball get fumbled or a car shifting lanes on the freeway. We've found that a group of retinal cells fire in a unique pattern when an object suddenly reverses its motion, and that pattern can signal a violation of the retina's prediction to the brain."
"I think of the retina as a model circuit," he added. "Computations in the retina have turned out to be relevant to other senses, including those I use when I'm learning about wine."
Berry does find it curious that he stepped away from biology for years, only to find himself where he wanted to be in the first place. Despite the roundabout route that brought him to neuroscience, he said the chance to make fundamental discoveries about an organ as fascinating as the brain has made the trip worthwhile.
"What is intelligence, what is consciousness? These are profound questions with relevance to everyday life, and neuroscience is beginning to grapple with them," he said. "If we make progress in this field, it's early enough that there might be some cool practical applications around the corner, even in my lifetime."
From the May 21, 2007, Princeton Weekly Bulletin