Imagine a staircase whose next step is always higher, yet one complete circuit leaves you exactly where you began. A drawing can suggest that contradiction. A carefully constructed sound can make it feel surprisingly plausible to the ear.
A Shepard-tone sequence is not a single frequency rising without limit. It combines several components and changes their balance so that an upward step can remain convincing while the overall pattern returns to its beginning. The interesting question is where the returning information hides.
Hear the stack, then hear the step
UC San Diego's explanation describes sinusoidal components separated by octaves beneath a fixed spectral envelope. Components near the low and high extremes are quiet. As the frequencies rise, high components fade away and quiet low components become available. After twelve semitone steps, the construction can return to its initial state.
Picture several identical name tags on different floors: C on one floor, C an octave above it, and another C higher still. All move upward together. Lighting is brightest on the middle floors and dim at the extremes. A tag entering below becomes visible gradually while one leaving above fades out. This is an analogy for amplitude weighting, not a literal picture of a cochlea.
Hear a loop conceal its reset
Click-start sine-wave sequence, finite duration, single-component comparison, visible loop counter and Stop button.
The demonstration starts only when you press a play button and stops automatically after a short sequence. Begin at a low device volume. You can choose the layered sequence or a single middle component. The display reports the twelve-step loop position, making the return available to the eye while the ear follows the apparent ascent.
What “higher” means in this experiment
In his 1964 paper, Roger Shepard reported a set of complex tones that could produce circular judgments of relative pitch. Neighboring tones around the constructed circle could be judged successively higher, while diametrically opposed comparisons were ambiguous. This is evidence about a special stimulus set and perception, not evidence that ordinary frequency has become circular.
The distinction is easier to feel with a familiar melody. Two versions an octave apart can sound related in note identity while differing in register. “Same note name” and “same overall height” are different claims. The Shepard construction deliberately makes the relationship between those aspects hard to settle.
Listen to one short run with your eyes away from the counter. Does it seem to ascend? Write your answer before revealing the step number. Then play another run while watching the reset. Your perception need not agree with someone else's description on every presentation. Hearing, playback equipment, the envelope, and attention can all affect an informal demonstration.
The quiet parts do essential work
Switch to the single-component version. Its return to a lower starting frequency is more exposed. Now restore the layered version. The contrast helps identify what the stack contributes: another component can preserve an apparent upward continuation while a quieter part changes role.
The local synthesis uses a small, finite set of sine waves and a smooth amplitude envelope. It is an educational implementation, not a replication of Shepard's experimental equipment or an individual hearing assessment. Its low-level start, fade-in, and automatic ending are presentation choices, not ingredients that prove the illusion.
A neighboring illusion asks a different question
Diana Deutsch's tritone-paradox work uses tones ambiguous in octave height to study how listeners judge pairs separated by half an octave. Her account distinguishes pitch class from height and reports systematic differences between listeners. That phenomenon should not be treated as merely another name for an endlessly rising scale.
Visit her examples after using the local twelve-step sequence. Ask what changed in the task. Following adjacent steps around a loop is different from deciding whether a widely separated pair rises or falls. A shared kind of ambiguous tone can support different experiments.
If the climb does not seem endless
That is an observation, not a failed test. Try the layered and single-component settings once each, at the same comfortable volume, and compare their resets. Do not turn the sound up to force an effect. You can still understand the construction by following the listed frequencies and amplitudes as they exchange roles.
At step zero, the component list begins at 55 hertz and includes octave multiples. Advancing the silent counter changes every frequency by the same semitone ratio. After step eleven, the display returns to its original list. The reset is exact in the recipe even if the sequence still feels directional when heard.
The satisfying part comes when explanation and experience coexist. You can know that the sequence repeats, watch the counter return to zero, and still hear a compelling next step. The contradiction belongs to the judgment your auditory system builds from the available cues. The frequencies remain finite, and the staircase has a carefully concealed landing.
Sources and further exploration: UC San Diego: Shepard-tone construction; Roger Shepard: circularity in relative-pitch judgments; Diana Deutsch: the tritone paradox.
Explore all 50 new articles · Browse the complete illustrated collection
Explore more art, ideas, and independent projects at Incoprea.
