1.1 A demonstration

Here is a pattern of three notes played in a triplet rhythm — high, low, high, rest, repeat — with two parameters you can adjust: the frequency separation between the high (A) and low (B) tones, and the rate at which the pattern repeats. Headphones or speakers are fine. Press play, and listen for a few seconds.

Now experiment. Start with a small frequency difference (say, two semitones) and a slow rate (around 4 triplets per second). What you hear, almost certainly, is a galloping rhythm: one perceptual object, a small melody that pulses three-and-a-rest, three-and-a-rest. The high and low tones group together.

Now increase the frequency difference to twelve semitones (an octave), and speed the rate up to 10 triplets per second. What you hear changes — discontinuously, in fact. The high tones unhitch from the low tones. You begin to hear two separate streams: a stream of high notes pulsing at half the rate of the original triplet pattern, and a stream of low notes pulsing slowly underneath. The gallop is gone. There are now two perceptual objects in your head, where moments ago there was one.

Move the sliders back and forth across the bistable region in the middle. The percept flips between the two organizations, in some places sticking on one side, in others on the other, in still others jumping back and forth involuntarily as you listen. The signal arriving at your eardrum is the same in every case (or differs only by frequency or rate, smoothly). What is changing in the bistable region is the listener.