The insight that neurological functions could be localized in the brain—that activities such as speech, vision and hearing take place in fixed locations, with the aid of specialized neural circuits—has served as one of the driving ideas in neuroscience. Less often appreciated is the companion notion that the power of the brain, the key to its flexibility and coordination, lies not just in the capacities of these dedicated processing centers, but also in the connections among them. It is not enough, as the phrenologists proposed centuries ago, to have islands of specialized function for each of the brain’s activities. For modern neuroscientists, the whole story must lie not just in the brain’s compartmentalization, but in its communication.

Nevertheless, modern neuroscience techniques often focus on localization at the expense of communication. Whole-brain imaging techniques such as functional MRI, for example, have allowed researchers to gain some insight into which regions of the brain are more active during a given behavior. But the most direct technique available for studying brain function during behavior—measuring the electrical activity of individual neurons—typically focuses on a specific location within the brain. This is not only because the technical challenges posed by simultaneous recordings in several brain areas are daunting, but also because many regions remain poorly understood, and others frequently share so many connections with the rest of the brain that they often appear to be involved in everything. Most of the time, the neural circuits involved are sufficiently complex that neuroscientists are simply trying to get a handle on what role, if any, a particular brain area plays in behavior; try to factor in communication among several of them, and most hypotheses become too complicated to test directly. In effect, studying information flow within the brain becomes a bit like tapping into a massive network switchbox: there’s a constant stream of information flowing past, but without a clever experiment, it’s nearly impossible to tell just where that information is going or how it’s being used. Unfortunately, these are the very questions neuroscientists suspect are most crucial for understanding one of the most complex of human behaviors: how we make decisions.

Listening to Cross Talk

Despite these obstacles, Bijan Pesaran of New York University and collaborators at the California Institute of Technology recently managed to pull off just this sort of coordinated eavesdropping, in an experiment designed to catch the cross talk between two specialized regions of the brain during decision making. Their work, published in the journal Nature, focused on two key areas involved in planning reach movements: the dorsal premotor area (PMd) in frontal cortex and the parietal reach region (PRR) in parietal cortex, which previous studies had indicated were directly connected by sets of neurons stretching back and forth across the brain. The plan was to record from both areas at once as an animal reached out to touch a computer screen, and to look for coordination in the electrical traffic between them as the decision context was varied.

The task they employed was deceptively simple: they trained two monkeys to select from among three cues simultaneously presented on a computer screen, searching through them for the one that produced reward. In one condition, all of the shapes were circles, and the monkeys were allowed to touch them in any order until they received a squirt of juice. In the other, each cue was a different shape, indicating the sequence in which the monkey had to select the targets (choosing the shapes out of sequence resulted in no reward). In both cases, the monkey was required to make a movement. But under the first condition, free searching, the scientists’ intuition is that the animals were more actively engaged in the decision process, and thus more in need of coordination between the two planning regions.

And in fact, this activity is just what Pesaran and collaborators found when they examined the correlation between neural firing in the two regions. When the monkeys were freely searching for reward, the firing in PMd and PRR was more coordinated than when they were following the fixed search pattern, suggesting that the two areas shared more information under the free choice than the mandated-order condition. Apparently, the "no-brainer" search pattern, executed by rote, resulted in less coordination between the two regions. What’s more, analysis of the relative timing of activity in the two areas seemed to suggest that information flowed first from PMd to PRR, as would be expected if PMd, a frontal area, played a more executive role in the decision. Nevertheless, firing in each region was influenced by activity in the other, arguing against the idea that PMd was simply "handing down" a motor plan for PRR to execute.

A Thought Network

So if decisions such as these are best thought of as distributed over networks, not localized to particular brain areas, just how close are we to disentangling the flow of information in the brain? In some respects, the problem goes all the way back to Camillo Golgi and Santiago Ramón y Cajal, who more than a century ago began the arduous process of tracing out the anatomy of its connections. On the other hand, studies such as the one by Pesaran and colleagues, which attempt to follow this information flow in real time, remind us that some of the most critical questions remain unanswered for basic systems: What information is being sent between brain areas? How is it encoded and processed? How do these interactions change in time?

Judged by such a standard, the neuroscience of decision making is arguably still in its infancy. Nevertheless, as research continues to suggest, the keys to some of the brain’s most intriguing secrets—its ability to learn, to imagine, to choose at will—will be found not only in its parts, but in their connections.

Mind Matters is edited by Jonah Lehrer, the science writer behind the blog The Frontal Cortex and the book Proust Was a Neuroscientist.