The brain creates illusions of words and phrases.
Posted June 26, 2009
This appears as an eye, but in reality is a trillion-mile-long tunnel of glowing gases, known as the Helix Nebula. Just as we "see" visual objects that correspond to those that are familiar to us, we also 'hear' phantom words and phrases that relate to our experiences.
In October 2008, The Bolivar, Herald-Free Press ran the following story:
“When Gary Rofkahr from Ossowo heard there was a baby doll on local store shelves that said, “Islam is the light,” he didn’t believe it.
He bought the Little Mommy Real Loving Baby Cuddle and Coo doll to see for himself, and was shocked to hear “Islam is the light” among other baby gibberish including “Mama.”
“I have a 1-year-old granddaughter,” he said. It makes me mad that someone is trying to indoctrinate our children with an innocent toy.”
The Little Mama doll is only one of a long string of talking dolls that have shocked customers over the years. On April 12, 2000, APB News reported:
“Taylor, an unemployed office worker, told APBnews.com today that she bought a talking Tinky Winky doll a few weeks ago for her daughter's second birthday. When Taylor tugged at Tinky Winky's left hand to make him talk, the doll spouted words that shocked Taylor. "He was saying, 'I got a gun, I got a gun, run away, run away,’ recalled Taylor, 22. "My gut just kind of dropped.”
These incidents reflect a profound truth: When listening to speech, the words and phrases that we "hear" are strongly influenced, not only by the sounds that reach our ears, but also by our knowledge, beliefs, and expectations. This might seem surprising, as the process of speech communication generally appears effortless to us. So from our subjective experience, it would be reasonable to assume that the sound signal simply travels from our ears up to our brain where, in a few simple steps, it is chopped up into small units that are then reassembled correctly so that the original message is perceived.
But for half a century, engineers have struggled to develop computer software that can understand conversational speech as well as humans can and so far they haven’t even come close to achieving their goal.
There are monumental problems involved in designing speech recognition software that can handle conversational speech effectively. We can easily recognize speech when it occurs at a rate of, say, 15 or 20 phonemes per second. (Phonemes are basic units of sound that are strung together to produce morphemes, which are the smallest meaningful components of words.) Yet when we are presented with a sequence of clicks at a rate of 15 per second, rather than hearing them individually, we hear a single buzzing sound. So to understand speech, the phonemes must at some point be stored in the brain as overlapping in time, and then somehow disentangled so that the speech stream can be correctly perceived.
We don’t yet understand how this occurs. As a further complexity, we can recognize words and phrases when they are spoken by different speakers, including those who speak in different dialects, and also when they are spoken by the same speakers when they are in different emotional states. In addition, speech often occurs in the presence of other sounds, and those that we are trying to interpret the need to be separated out from the rest. As an added complication, a recording of the sound spectrum produced by a speech signal, when analyzed by computer, can be quite different depending on where in the room the recording is taken, or even depending on the position of the speaker’s head.
So how humans are able to understand normal conversational speech effectively is still a considerable mystery. But where we have an advantage over computers (and probably always will) is that we can draw on all the information we have stored in our memories—not only our knowledge of grammatical structure, but also our beliefs and expectations based on an enormous amount of personal experience—in making inspired guesses concerning what is probably being said.
However, this very process of guesswork can lead us to perceive phantom words and phrases that are not, in fact, being spoken. There are many examples here, such as the Mondegreens described by Jon Carroll in his column in the San Francisco Chronicle or the beliefs held by some people that they have received spoken messages from the spirit world, described by Michael Shermer in his column for Scientific American.
Some years ago I discovered a way to produce a large number of phantom words and phrases within a short period. To obtain the best effect, find a time when you will not be disturbed, and sit comfortably in front of two loudspeakers, with one to your left and the other to your right. (Headphones don’t work as well.) Make sure that your sound system is set for stereo, and that the two loudspeakers are reasonably balanced for loudness. In a phantom words demonstration, each track contains two words or a single word composed of two syllables, and these are repeated over and over again.
The same sequence is presented through both loudspeakers, but the tracks are offset in time so that when the first sound (word or syllable) is coming from the loudspeaker on the left the second sound is coming from the loudspeaker on the right; and vice versa. Because the sounds coming from the two loudspeakers are mixed in the air before they reach your ears, you are given a palette of sounds from which to chose, and so can create in your mind many different combinations of sounds.
It works well to have a pen and paper in front of you so that you can jot down the words and phrases that you hear. Often people initially hear a jumble of meaningless sounds, but after a while, distinct words and phrases suddenly emerge. It often seems that the left and right loudspeakers are producing different words, which sometimes appear to be spoken by different voices. So jot down separately the words that you hear as coming from the left, and those that you hear as coming from the right.
After a while, you will probably find that new words and phrases appear to be coming from one loudspeaker or the other. When this happens, jot these down also. In addition, it’s not unusual to hear a third stream of words or phrases, apparently coming from some location between the loudspeakers. Nonsense words, and musical sounds such as percussive sounds or tones, sometimes appear to be mixed in with the meaningful words. People often report hearing words and phrases spoken in strange or foreign accents, presumably they are fitting the sounds from the loudspeakers into words and phrases that are meaningful to them, even if this causes the words to appear distorted.
If English is your second language, you may find that you hear some words and phrases in your native language. I teach a course on illusions at the University of California, San Diego, and generally, I play some phantom words to my class. The students at our university are linguistically very diverse, and taken together I’ve received reports of phantom words in Spanish, Mandarin, Cantonese, Korean, Japanese, Vietnamese, Tagalog, French, German, Italian, Hebrew, and Russian, to name just a few. It’s not unusual for students in my class to feel strongly that such ‘foreign’ words have been inserted into the tracks, and sometimes they adamantly stick to this belief, despite my assurances to the contrary.
People appear to hear words and phrases that reflect what is on their minds, as in a Rorschach test, though it’s my impression that the present effect is stronger. I can bet who is likely to be on a diet, as they report words like I’m hungry, diet coke, or feel fat. And students who are stressed tend to report words that are related to stress; if I play these sounds close to exam time, some students may well hear phrases like I’m tired, no brain, or no time. Interestingly, female students often report the word ‘love’, while male students are more likely to report sexually explicit words and phrases.
Here are three phantom words. One is from my compact disc Musical Illusions and Paradoxes, and the other two are from my compact disc Phantom Words and Other Curiosities, which also contains four more examples.