Hear That? In A Din Of Voices, Our Brains Can Tune In To One

Mar 6, 2013
Originally published on March 7, 2013 6:49 am

Scientists are beginning to understand how people tune in to a single voice in a crowded, noisy room.

This ability, known as the "cocktail party effect," appears to rely on areas of the brain that have completely filtered out unwanted sounds, researchers report in the journal Neuron. So when a person decides to focus on a particular speaker, other speakers "have no representation in those [brain] areas," says Elana Zion Golumbic of Columbia University.

The ability to extract sense from auditory chaos has puzzled scientists since the 1950s, Golumbic says. "It's something we do all the time, not only in cocktail parties," she says. "You're on the street, you're in a restaurant, you're in your office. There are a lot of background sounds all the time, and you constantly need to filter them out and focus on the one thing that's important to you."

But until a few years ago, how the brain did this was a mystery. That's changing, Golumbic says, thanks to new technology that allows scientists to monitor many different areas of the brain as they listen to multiple voices.

The technology involves a grid of electrodes placed on the surface of the brain. Experiments have relied on volunteers who already had these electrodes in place: people in the hospital awaiting surgery for severe epilepsy.

"We bring in a cart with a computer and a screen and speakers," Golumbic says. "And we show them movies."

One movie, for example, shows a woman telling a brief story about a parrot. Another shows a man telling a story about how he never liked to clean up his room.

To simulate a cocktail party, though, participants watched a third movie in which the man and woman are both on screen telling those stories simultaneously. The researchers asked them to focus on just one of the speakers while they monitored what was going on in their brains.

And the brain monitoring revealed something remarkable. When a person's brain is in cocktail party mode, some areas, like those involved in hearing, continue to respond to both voices. But other parts of the brain, like those devoted to language, appear to respond only to the selected speaker.

Afterward, volunteers who focused on the man had no trouble remembering that he didn't like to clean his room. But they didn't recall anything about the woman's parrot.

The study also found that the brain areas responding only to the selected voice were constantly fine-tuning their reception, says Charles Schroeder, a neuroscientist at Columbia University and New York state's Nathan Kline Institute. "As the sentence unfolds, the brain's tracking of the signal becomes better and better and better," he says.

This suggests that the brain is separating one voice from the rest by identifying its unique characteristics, Schroeder says. It's also likely that the brain is using information from the first words in a sentence to predict which words are likely to come next.

A better understanding of the cocktail party effect could eventually help people who have trouble deciphering a single voice in a noisy environment, says Edward Chang, an assistant professor of neurological surgery and physiology at the University of California, San Francisco.

That's a problem for many people as they get older, he says. It's also a problem for people with attention deficit hyperactivity disorder, or ADHD, Chang says, adding that he saw this problem up close when a person with the disorder volunteered for an experiment that involved trying to focus on just one of two speakers.

"This person had significant problems with the ability to select the correct speaker," Chang says.

Understanding precisely how the brain solves the cocktail party problem could also allow machines to do a better job deciphering human speech, Chang says. That could mean better cellphones and less frustrating conversations with the computers that often answer phone calls to customer service hotlines.

Copyright 2013 NPR. To see more, visit http://www.npr.org/.

Transcript

MELISSA BLOCK, HOST:

Here's a question that's puzzled scientists since the 1950s: How can someone in a room filled with voices pay attention to just one voice? Well, now scientists are beginning to get answers, and that's thanks to new technologies that can measure electrical activity in many parts of the brain. NPR's Jon Hamilton explains this latest effort to understand the so-called cocktail party effect.

JON HAMILTON, BYLINE: At a cocktail party, the brain has to make sense out of auditory chaos. And Elana Zion Golumbic, a researcher at Columbia University, says that's something people do really well.

ELANA ZION GOLUMBIC: It's also something we do all the time, not only in cocktail parties. You're on the street, you're in a restaurant, you're in your office - there are a lot of background sounds all the time, and you constantly need to filter them out and focus on the one thing that's important to you.

HAMILTON: Golumbic says experiments show that the human brain is much better at doing this than even the most powerful computer.

GOLUMBIC: But in terms of how we can actually do it, what's the brain doing to enable us to do this online filtering all the time; that has been a mystery.

HAMILTON: So a few years ago, Golumbic and other researchers began using a new technology to monitor people's brains as they worked to separate one stream of sounds from another. The technology involves a grid of electrodes placed on the surface of the brain. Golumbic says the experiments relied on volunteers who already had these electrodes in place. They were people awaiting surgery for severe epilepsy.

GOLUMBIC: And we bring in a cart with a computer and a screen and speakers. And what we do is, we actually show them movies.

HAMILTON: One movie shows a woman telling a brief story.

(SOUNDBITE OF MOVIE IN EXPERIMENT)

UNIDENTIFIED WOMAN: My good friend Jonathan has a pet parrot called Gypsy, who can speak.

HAMILTON: Another shows a man.

(SOUNDBITE OF MOVIE IN EXPERIMENT)

UNIDENTIFIED MAN: As a child, I never liked to clean up my room.

HAMILTON: And a third movie shows a man and a woman talking at the same time.

(SOUNDBITE OF MOVIE IN EXPERIMENT)

UNIDENTIFIED WOMAN, MAN: (Speaking at the same time) My good friend Jonathan....as a child I never liked...has a pet parrot...to clean up my room...called Gypsy, who can speak...because I thought it was easier for me to find things when everything was out......he can say his own name and likes to call out...compared to when they were hiding inside a closet or a drawer...hello, come on in...

HAMILTON: Just hearing those voices, it's pretty hard to pay attention to a single speaker. But volunteers were able to focus their attention by looking at the speaker they wanted to understand. And the activity in their brains revealed something remarkable. When a person's brain is in cocktail party mode, some areas - like those involved in processing sound - continue to respond to both voices. But Golumbic says other areas of the brain, like those devoted to language, appear to respond only to the selected speaker - in this case, the man.

GOLUMBIC: The other speaker has no representation in those areas. Those areas have completely filtered her out.

HAMILTON: Afterward, the volunteers couldn't remember the woman's story about Gypsy the parrot, but they had no trouble remembering the man who didn't like to clean his room.

Charlie Schroeder is a co-author of the new study, which appears in the journal "Neuron." He's also a neuroscientist at Columbia University and New York state's Nathan Kline Institute. Schroeder says the study offers some hints about how the brain attends to one voice while ignoring others. He says these hints come from the so-called exclusive sites in the brain, those areas that seem to respond only to the selected voice.

DR. CHARLIE SCHROEDER: In these exclusive sites, as the sentence unfolds, the brain's tracking of the signal becomes better and better and better.

HAMILTON: Schroeder says these areas seem to be constantly fine-tuning themselves, to detect the unique characteristics of a particular voice. He says it's also likely that these areas are using information from the first words of a sentence to predict what's going to come next.

Edward Chang of U.C.-San Francisco says this sort of research may eventually help people who have trouble tracking a single voice at cocktail parties. That includes many people with ADHD. Chang says he saw this close up, when a person with the disorder took part in one of his studies.

DR. EDWARD CHANG: This person did have ADHD, and had significant problems with ability to select the correct speaker.

HAMILTON: Chang says understanding how the brain solves the cocktail party problem also could help machines - like cellphones - do a better job understanding what we're trying to tell them. Jon Hamilton, NPR News.

(SOUNDBITE OF MUSIC)

AUDIE CORNISH, HOST:

You're listening to ALL THINGS CONSIDERED, from NPR News. Transcript provided by NPR, Copyright NPR.