Feature

When people think of illusions, they may imagine a magic trick or deceptive drawing, like the never- ending staircase popularized by artist M.C. Escher. But vision isn't the only sense that can be muddled by ambiguous information.

University of California, San Diego, psychologist Diana Deutsch, PhD, and her colleagues have created an auditory illusion called the tritone paradox, and have found that a person's linguistic history has a lot to do with how the illusion is heard.

Their research could have implications for discovering how the developing brain processes sound and speech.

"We are at the early stages of research," explains Magdalene Chalikia, PhD, of Minnesota State University, Moorhead, who's also researching the illusion, "but it's an exciting field, learning how music and language are connected."

A lesson in pitch

Deutsch's auditory illusion is rooted in the concept of pitch, which can be thought of in two dimensions--height and class. Pitch height is how high or low a note sounds. Pitch class is the name of a tone, such as A, B, C, etc., including sharps (#) and flats (b). There are 12 pitch classes (see right), and these notes are continually repeated as they ascend in pitch height. The 12 pitch classes can be thought of in a circle; one trip around the circle (from C to C) is an octave; halfway around (from C to F#) is an interval called a tritone.

In 1964, Roger N. Shepard, PhD, created an auditory illusion by using tones that consist of many octaves of the same pitch class note played at once--from very low to very high. Because each note contains many octaves of the same pitch class, it is neither high nor low--the tone's pitch height is ambiguous.

Playing two of these ambiguous tones in succession, Shepard gauged whether listeners heard the second note as ascending or descending, and found their judgments were related to the position of the notes on the pitch class circle. For example, the pair D#­E was always heard as ascending because the shorter distance is clockwise, but the pair A#­A was heard as descending because the shorter distance was counterclockwise.

Even more interesting was that when Shepard played tones as they repeatedly moved clockwise around the pitch class circle, the tones appeared to endlessly ascend; when moving counterclockwise through the 12 ambiguous notes, the pattern appeared to endlessly descend--the auditory equivalent of Escher's endless staircase.

The tritone paradox

Twenty years later, Deutsch added another dimension to Shepard's illusion by giving participants pairs of ambiguous tones, one played after the other, that were related by a tritone (for example, C followed by F#).

This half-octave interval takes away a listener's ability to make a judgment based on the shortest distance around the pitch class circle because the distance is equal in each direction. Technically, the two notes are perceptually equivalent in pitch height.

But when Deutsch presented subjects with these ambiguous tritones, they consistently heard a difference in pitch height between the notes. And there's the paradox--even though all ambiguous tones separated by a tritone interval are equivalent, most people hear the second as higher or lower than the first. Deutsch hypothesized that, even though most had no musical training, the listeners were tagging one region of the pitch class circle as higher in tone and the opposite region as lower.

To illustrate this idea, think of listeners who mentally place pitch class C in the highest position; they would hear the ambiguous tone C followed by F# as descending. But, if they placed F# at the top of their pitch class circle, then they would hear the exact same ambiguous tone pair as descending. (The perceptual illusion only occurs when the ambiguous tones are played, not when a single musical instrument--such as a piano--plays two notes.)

Deutsch plotted the percentage of times each participant heard a pattern as descending as a function of the pitch class of the first tone of the pair. As she had conjectured, the judgments of each participant depended systematically on the position of the tones along the pitch class circle. However, the note that was placed at the highest point in the pitch class circle varied among participants.

"It shows that everybody who shows the effect--and most do--has some form of perfect pitch," Deutsch explains. "Even though they're not explicitly labeling the tone, they are deciding which tone is higher or lower by the pitch class of the tone."

Music and language

What would cause people to hear the same tones in a different way?

Results of a 1990 experiment by Deutsch and colleagues Tom North, PhD, and Lee Ray, PhD, point toward language. They found a clear correspondence between the pitches a person uses in speech and their perception of the tritone paradox. And since most individuals speak in pitches within one octave range, it seems we define the orientation of our pitch class circle based on the range we speak in--whether it's from C to C or F# to F#.

Research has shown that people speak in the same pitches as others in their linguistic community, except females speak an octave higher than males. However, linguistic communities differ greatly from each other. More- over, physiological characteristics such as chest size or the length of the vocal tract have little effect on the pitch range of a person's speech.

Taking all this into account, both Deutsch and Chalikia are separately studying differences between linguistic communities. Findings include that:

  • Californians tend to have peak pitch classes in the range B, C, C#, D and D#, while people from southern England tend toward F#, G and G#, areas that are basically on opposite sides of the circle.

  • California children whose parents grew up out of state have pitch classes that are very similar to their mothers'--not to other California children.

  • In a study of Texans who spoke either English or Spanish and English, researchers found that regardless of which language was learned first, all of the bilinguals had similar peak pitch classes that were very different from the monolinguals.

  • In a study of Greeks who acquired English as a second language, researchers found the Greek pitch class circle was significantly different from the Texan, Californian and southern England distributions.

  • Peak pitch class distribution among the Swedish is more similar to the English and is significantly different from Texan and Californian distributions. The data also suggest a Swedish pitch class template that is exceptionally strong, since a tightly grouped peak pitch range was found.

In addition, Deutsch, Trevor Henthorn and Mark Dolson recently studied two groups living in California: younger participants who were born in Vietnam, raised in the United States and fluent in English but not in Vietnamese, and older participants who arrived in the United States as adults and spoke little English. There was a striking pattern: While native Californians centered on B to D#, both groups of Vietnamese centered on D# to F#, regardless of the language they spoke now. And when the peak pitch classes of the younger Vietnamese group were compared with native Californians of the same age, the differences remained highly significant.

What it could mean

These findings have led researchers to believe that perception of the tritone paradox is heavily influenced by the listener's first language, even when the listener no longer speaks this language fluently.

Odds are that pitch class templates are developed at a very young age--perhaps as infants listen to the intonation of voices around them. But no one's sure how the brain adapts its neural connections to the pitches heard in speech. To learn more, Chalikia emphasizes the need to conduct more systematic research on linguistic communities--such as testing Spanish-English bilinguals at the same time as Spanish monolinguals and English monolinguals.

And there are more questions. For example, is there a difference between how the brain processes English and other Western languages versus tonal languages, such as Vietnamese and Chinese? In tonal languages, the same syllable uttered at a different pitch means a completely different word. Deutsch found that Vietnamese and Chinese speakers showed perfect pitch when they spoke the same words on different days. Westerners are not nearly so precise.

Also, Chalikia's Greek data and Deutsch's Vietnamese experiment may indicate that bilinguals develop a pitch class template for each language they speak. "If so, does the brain get rewired in terms of pitch representation based on the first language, or does it maintain an ability to be rewired multiple times and store multiple pitch representations?" asks Chalikia. "We're just scratching the surface."