It's a classic cocktail party conundrum: How do our brains decide where we should train our attention when people are milling all about us chatting away—some to us, some to others?

In an attempt to find out, researchers at Stanford University and McGill University in Montreal scanned the brains of 18 subjects who were listening to classical music by 18th-century British composer William Boyce.

"You have to kind of segment these streams [of information] into chunks," says senior study author Vinod Menon, an associate professor of psychiatry and behavioral science at Stanford. The process of slicing the data, he continues, requires that you "identify something that's interesting and then you have to switch on the attentional network."

"Memory doesn't work like video recorder, it's more like DVD," in how it recalls events as discrete chapters, explains study co-author Daniel Levitin, a psychology professor at McGill.

But why music?

Simple, says Sridhar Devarajan, a Stanford neuroscience graduate student involved in the project. "Transitions between musical movements," he notes, "offer an ideal setting to study the dynamically changing landscape of activity in the brain during this segmentation process."

The team wove together different movements from several of Boyce's four- to five-minute symphonies and had volunteers listen to two nine-minute compositions through noise-canceling headphones while lying in an fMRI machine; each of the musical tapestries consisted of 10 orchestral movements. (Boyce's compositions were selected because he is a relatively obscure composer and subjects were less likely to be familiar with his work, canceling out brain activity that would be generated if they had recollections of his symphonies.)

The team conducted full brain scans that allowed them to focus on regions of the brain of particular interest, monitoring them from 10 seconds before a transition between movements to 10 seconds after. (A transition between movements is marked by a decline in the amplitude of sound followed by a brief period of silence that leads into a new section of music.)

According to Menon, during the transition periods the team not only observed activity in discrete brain regions, but they also noticed co-active networks (two areas responding simultaneously) reacting to the musical shifts. Activity began in areas in the forward section of the prefrontal cortex of the brain as well as in parts of the temporal cortex just above the brain stem. Menon speculates that it is this network that is "detecting a salient change" in the information stream. Next, areas toward the rear of the prefrontal cortex and parts of the parietal cortex (the outermost layer of the parietal lobe at the upper rear of the brain), began to respond. The regions, Menon notes, are linked to attention and working memory.

"We feel that it could be a very general brain mechanism…part of a core control signal that is generated in response to any attention-demanding task," Menon says. "I think any task that involves detecting a salient event and attending to it will command a similar type of response."

Levitin adds, "Here we had a peak of activation…associated with nothingness," when there was no sound at all. "[Clearly], there are neural processes that are responsible for signaling the beginning and ending of events."

Next up for Menon's team: trying to determine the next level of processing after recognizing and responding to a change in an information stream. It also plans to apply whatever new information it discovers to learn more about "what structure actually means in music."