Advertisement

Study shows how brain tunes out background noise, hears one voice

Partygoers at a 2003 event at Los Angeles' Museum of Contemporary Art. Researchers say they have discovered how people can hear a conversation over a noisy din.
(Ken Hively / Los Angeles Times)
Share

In some ways, parties seem like the worst possible places to socialize. A cacophony of voices -- not to mention a blaring stereo system -- make for a noisy environment in which to hear what a friend is saying.

Hence the term “the cocktail-party effect,” which refers to people’s ability to focus on one speaker and tune out another.

Now Nima Mesgarani and Edward F. Chang of UC San Francisco have figured out how the brain accomplishes this feat of selective hearing: The auditory cortex, which processes sounds, favors the voice that it needs or wants to hear. They reported their results Wednesday in the journal Nature (subscription required).

Advertisement

The scientists made their discovery during an experiment conducted on three patients who were preparing to undergo brain surgery for severe epilepsy. As part of their workup for the surgery, doctors mapped the patients’ brain activity through the use of a sheet of electrodes placed on the brain’s surface beneath the skull.

While the electrodes were in place, Mesgarani and Chang asked the patients to listen to speech samples from two speakers -- saying different things, but at the same time -- and identify what was said by one of the voices they heard.

Studying the brain activity readouts from the electrodes, the researchers found that the auditory cortex reacted only to the targeted speaker, “as if subjects were listening to that speaker alone,” they wrote.

They suggested that their discovery could help engineers devise better speech-recognition systems and could shed light on why some people have trouble hearing a voice among a crowd, perhaps because of aging, or a speech perception disorder.

UCSF has posted a website that lets you test your own ability to distinguish unique voices, with and without the benefit of knowing what you’re listening for.

Advertisement