Finishing college, saving for a down payment on a house or learning to play the cello all have something in common: they take years to accomplish. In that sense, they are uniquely human pursuits.
To figure out how our brains guide us through long-term projects in the face of endless distractions and disappointments is one of Earl Miller's goals. Dr. Miller, associate professor of brain and cognitive sciences in MIT's Center for Learning and Memory, studies the prefrontal cortex, the part of the brain most central to high-level cognitive function.
In humans, the prefrontal cortex is proportionally huge. This "most human" part of the brain is also called the brain's executive because it oversees all voluntary activity.
"The main thing we are interested in is the deepest issue in cognitive science, called executive or cognitive control. It governs how you decide what behaviors to engage in, how you make decisions, how you decide what to pay attention to, what to do with your life," Professor Miller said.
His laboratory is one of handful in the world to make strides in developing hypotheses about how the prefrontal cortex does its job.
"When I started graduate school, if you had told somebody, 'I'm going to tell you a little bit about how voluntary, goal-directed behavior works,' he or she would have said, 'You can't study that. That's a crazy topic.' But just in the past few years, we've actually developed some real hypotheses about how this all works."
Professor Miller, who studied psychology before undertaking neural science, said his team may some day be able to identify the roles played by specific neurons in learning and memory, although "the brain is arguably the most complex thing we know of. We've learned a lot in the past 50 years and certainly in the last 10 years, but I think we've just scratched the surface," he said.
BACK TO BASICS
To figure out how the overall system works, Professor Miller looks at how single neurons communicate. Single neurons are the brain's most basic units, like the ones and zeroes on which computers are based.
"We're starting from the premise that the most important feature for neurons is their ability to extract information from experience," he said. "They need to be sculpted by experience, because virtually all voluntary behavior -- everything we're saying and doing right now -- is learned."
For example, when you enter a restaurant, you know from previous experience what to expect and what actions are appropriate. While memories of dining in a specific restaurant on a specific night are probably stored in the hippocampus, the prefrontal cortex extracts the general features of previous restaurants to give you a general rule of what it means to dine in a restaurant.
Understanding how different types of information are processed and mapped out all over the brain is important but difficult, because one needs to study millions of neurons simultaneously to see the whole picture. Based on knowledge of which functions are lost by people whose prefrontal cortex is damaged by disease or accident, Professor Miller has chosen 50 or 100 neurons to study at a time.
By recording the electrical activity of these groups of neurons in rhesus monkeys, he studies the ability of the prefrontal cortex to play a role in the monkeys' ability to direct their attention and their ability to bring a visual image to mind from long-term memory.
The monkeys are trained to perform specially designed computer games, which Professor Miller has used in a variety of experiments on recall, rule-learning and spatial sense in short-term memory. The games are designed to engage the monkeys' ability to pay attention, recall information from long-term memory, learn new rules and make decisions.
Some of the games test their ability to hold a particular goal in mind and persist in that goal through distractions. This ability is missing in people with neural psychiatric disorders. Individuals with disorders involving the prefrontal cortex lose the capability to guide their behavior in accordance with their goals.
If Professor Miller can identify which neurons or groups of neurons are responsible for this ability, it may be possible to design drugs that can address its loss.
THE BINDING PROBLEM
Another central question in cognitive neural science is the binding problem. Researchers know that information about what an object looks like and where the object is are stored in two separate sections of the brain and analyzed in separate parts of the visual system. The question is, where and how does this information come together?
"To develop representations of rules, of little models of behavior and tasks, requires piecing together a whole bunch of different information," Professor Miller said.
"We did an experiment that required monkeys to put together 'what' and 'where' in their minds. What we found is that neurons in the prefrontal cortex had activity that exquisitely represented and integrated both 'what' and 'where.' This provides a first clue on how this information, and perhaps even more diverse information, comes together in the brain."
Professor Miller and colleagues are the first to address this question in detail. While early results seem to flag the prefrontal cortex as a prime binding location, it also is possible that the information could be coming together elsewhere in the brain first.
"We're starting to march back through the brain, doing the same test from the highest level of the prefrontal cortex and going back into the brain in lots of different places to figure out where this comes together for the first time," he said.
This knowledge would allow us to understand how normal perception and cognition work, which could lead to drug therapies designed to alleviate the disruption of normal perception and cognition by neuropsychiatric diseases, Professor Miller said.
This work is funded by the National Institute of Neurological Disorders and Stroke, the National Institute of Mental Health, the RIKEN-MIT Neuroscience Research Center, the Pew Charitable Trusts, the McKnight Foundation, the Whitehall Foundation, the John Merck Fund and the Alfred P. Sloan Foundation.
A version of this article appeared in MIT Tech Talk on May 31, 2000.
