NEURAL NETWORK, when trained on the sounds of American speech, devotes lots of cells to distinguishing /r/ from /l/. But when trained on Japanese phonemes, the neurons organize so that a few cells are sensitive to the dominant frequencies (called F2 and F3) that differ between /r/ and /l/. Image: COURTESY OF FRANK H. GUENTHER Boston University
"Liver." The word rises from the voice box and passes the lips. It beats the air, enters an ear canal, sets nerve cells firing. Electrochemical impulses stream into the auditory cortex of a listener's brain. But then what? How does the brain's neural machinery filter that complex stream of auditory input to extract the uttered word: "liver"--or was it "river," or perhaps "lever"?
This article was originally published with the title From Mouth to Mind.