1.4 A cabinet of puzzles

Before we set out on the long journey of the essay, here are three more demonstrations of the same general principle as the bistable galloping triplet. Each one shows, in a different way, that the auditory percept is not a faithful copy of the acoustic input — that something at the listener’s end is choosing what the input means.

The three live in their own pages, with a canonical video, an interactive (where one can be built), and a short explanation of why it works. I am not going to fully explain any of them yet — every one is re-encountered, and solved, in its proper movement. They are here as a cabinet of curiosities, intended to be inspected briefly and remembered.

The three demonstrations

• 1.5 The McGurk effect. A multisensory illusion. You watch a face say one syllable, the audio plays a different one, and you hear a third syllable that is in neither track. Solved in movement 9.

• 1.6 Phonemic restoration. A speech sound is replaced by a cough or a burst of noise. Listeners hear the missing phoneme as if it had been there, restored from context. Solved in movement 9.

• 1.7 The Shepard tone. A tone that seems to rise forever, looping around like an Escher staircase in audio. Built from a stack of octaves with a fixed amplitude envelope. Solved in movement 8.

All three are demonstrations that the cochlea sends a signal and the brain interprets it. The two are not the same.