While conversing in crowded places or surrounded by
people, we are able to filter out all the
random voices around us, and concentrate on only that of the person we are paying attention to. Even though the other speakers may be in the same
frequency range, and may even be closer to us (and thus
louder) in some situations, we can accurately attend to and understand the single person when we try. This ability is known as the
Cocktail Party Effect, a name coined by
E. C. Cherry in 1953, when cocktail parties were a bit more popular than they are now.
To find out more about the phenomenon, Dr. Cherry did a series of experiments involving shadowing, the process of repeating speech heard in headphones along with other sound. When two streams of speech were played at once, spoken by the same person and having the same sentence structure (i.e. adjacent paragraphs in a book), the task proved very difficult, with the listener often repeating phrases from both streams without regard to coherent meaning. The task only became a little bit easier if the two streams differed in content, but became much easier when the speakers were of different genders or were set at radically different volumes.
In further shadowing experiments, Cherry used the headphones to present the attended message in one ear, and the unattended one in the other, after telling the listener which ear to pay attention to. These experiments were reported to be “easy” by the participants, and the shadowing error rate was low -- none of the unshadowed message got mixed in, or could even be recalled when the participant was asked. Varying the unattended stream wasn't noticed either, whether switching it from English to German, from sentences to a single repeated word, or even from normal to reversed speech. A change in the speaker's gender was noticed by the participants, as was a change from speech to a continuous tone.
One of the cool variations on this experiment was playing the same message in both ears, but with a delay between the two. When the delay was less than five seconds or so, all of the participants noticed that the two streams had the exact same content, showing that they were processing the unattended information on some level. Another finding was that when the content in both ears was reversed, the participant would be confused for a tiny bit of time (about 170 ms) and unable to shadow anything. After that time, they would switch attention to the opposite ear, and continue shadowing the original stream, even though it was now in the ear they were supposed to be ignoring.
Interestingly, there's a different, but related phenomenon known in some circles as the Cocktail Party Effect, which involves our ability to hear our own name. Even when we are in a noisy or crowded place, and concentrating on speaking to another person, or reading a book, we can pick out our own name if it is spoken out loud. We notice it so quickly and accurately that we often turn our head toward whoever spoke it without even thinking about doing so.
This ability points to our minds doing some “bottom up” processing (that is, processing at a lower-than-conscious level), rather than the “top down” processing associated with paying attention to something. That is, while our consciousness may be completely involved with whatever we are doing, some part of our brain is actively parsing all of the audible noise around us, just waiting to give us a jolt when our name is spoken.