In 1976 researcher Harry McGurk and his assistant John MacDonald were running an infant study on how babies perceive language when a technician dubbed a video of a face saying one syllable over audio of another. When they played it back, they heard neither sound. Their brains produced a third. This episode explains the McGurk effect: hear ba ba while watching lips form ga ga and you perceive da da, a fusion the brain builds when the eyes and ears disagree. Researchers who have studied it for more than 20 years still fall for it every time.

The episode then follows the experiments that tried to break it. The vowel ee makes the illusion strongest while oo nearly destroys it, a female voice over a male face still fuses, and you can even trick yourself in a mirror. Brain damage, dyslexia, Alzheimer’s disease, schizophrenia and autism spectrum disorder all change how strongly people experience it, and Japanese and Chinese listeners show a much weaker effect than speakers of English, Dutch, Spanish or German. The hosts close on what it means for eyewitness testimony and for lip synced deepfakes.

• Why the brain compromises on da: ba is made with closed lips, ga at the back of the throat, and da sits between them
• Audio can lag the video by up to 180 milliseconds and still fuse, but arriving 60 milliseconds early breaks it, just as light outruns thunder
• The speaker’s mouth must fall within 60 degrees of your gaze, and the effect is stronger when you see the right side of their mouth
• A tactile task, like identifying a hidden object by touch, greatly reduces the illusion
• Babies show the effect at four to five months, and English speaking children grow more visually influenced after age six while Japanese children do not