The bionic ear: Graeme Clark's cochlear implant

The full story

This is the episode's narration, word for word. Headings jump to that point in the video.

The question 0:00

If the part of your ear that actually hears is gone, how can you hear? For hundreds of thousands of people, the answer is a thin wire coiled inside the inner ear, talking to the nerve in electricity. One of the most important versions came out of Melbourne, from a surgeon whose key idea, the story goes, involved a seashell and a blade of grass. To see why that mattered, start with how hearing works.

The shop in Camden 0:29

Graeme Clark grew up in Camden, in New South Wales, where his father was a pharmacist with a hearing loss. Customers had to repeat themselves, loudly, so the whole shop heard what they were buying. By Clark's own account, the boy was embarrassed, and decided then to become an ear doctor.

Sound becomes movement 0:49

Sound vibrates your eardrum, and three tiny bones in the middle ear pass that vibration on, amplifying it, into the cochlea. The cochlea is a coiled, fluid-filled tube about a centimetre across, shaped like a snail shell. Running along it is a flexible strip called the basilar membrane. Sound makes the fluid move, and the fluid makes that strip ripple.

A keyboard in a snail 1:16

Here's the clever part. The strip itself is narrow and stiff at the entrance, and wide and floppy at the far end. So high notes make it vibrate most near the entrance, and low notes travel further in before they peak. Unrolled, it's roughly three and a half centimetres long, and works like a piano keyboard, treble at one end and bass at the other. That map of pitch to place is called tonotopy. Hold on to it, because the whole implant depends on it.

The hair cell 1:49

Sitting on that strip are hair cells, the real sensors. Each has a bundle of tiny projections on top. When the strip moves, the bundle bends, and that pulls open little channels at the tips. Charged particles rush in, and the cell makes an electrical signal that sets off the hearing nerve. Each ear has about three and a half thousand of these inner hair cells, plus about twelve thousand outer ones that act as amplifiers.

Why they don't come back 2:19

Hair cells are fragile. Genes, infections like meningitis, loud noise, some drugs and plain ageing can all destroy them. And unlike birds and frogs, we can't grow new ones. In most people who are profoundly deaf, most of the hair cells are simply gone. But here's the hopeful bit. The nerve cells that the hair cells used to talk to are tougher, and at least some usually survive, even after years of deafness.

Why louder doesn't help 2:51

So why not just use a hearing aid? A hearing aid makes the sound louder, so a damaged ear can still pick it up. But if the hair cells are gone, there's nothing left to turn that vibration into a nerve signal. It's like turning up the volume on a microphone that's been unplugged. A cochlear implant takes the other route: it skips the hair cells and stimulates the surviving nerve directly with electricity.

Talking to the nerve 3:17

The idea is old. Around 1790, Alessandro Volta put rods wired to a battery in his ears, and heard something like boiling, thick soup. In Paris in 1957, André Djourno and Charles Eyriès implanted a wire that stimulated a deaf man's hearing nerve directly. In Los Angeles in 1961, William House implanted his first patients with simple wires. These patients could hear sounds around them, but nobody could understand speech.

The experts said no 3:54

In 1964, the hearing scientist Merle Lawrence wrote that understanding speech this way was not feasible. Two years later, the Stanford surgeon Blair Simmons, one of the pioneers himself, put the chances of it ever giving what he called uniquely useful communication at about five per cent. Clark had read Simmons's report of a deaf man who heard sounds but no speech, and, in his words, it lit a fire in the belly. By his own estimate, ninety-five to ninety-nine per cent of people said it wouldn't work.

Why one wire isn't enough 4:30

Why couldn't one wire do it? The rate of electrical pulses is heard as pitch, but only up to a point. Clark's animal experiments showed the nerve couldn't keep time with fast pulses, and his first patient later confirmed it: above about two hundred pulses a second, rates became hard to tell apart. And much of what makes speech understandable sits higher than that, from about seven hundred to two thousand three hundred hertz. The only way to send those higher pitches was by place: many electrodes, each at its own spot on the keyboard.

Getting inside 5:07

So how do you get a row of electrodes inside a tiny coiled tube? Drilling holes along the cochlea did serious damage in animal tests. Sliding a smooth, loose-fitting carrier in through the round window, a soft opening at the base, did only mild damage. But in trials on human ear bones and moulds, the array stalled about ten millimetres in, jammed against the outer wall of the spiral.

Grass in a shell 5:36

The fix came, the story goes, on a beach. As Clark told it, he had collected turban shells, which spiral like a cochlea, and found that grass blades, floppy at the tip and stiffer further back, slid neatly round inside. In 1977, his team worked out that an electrode bundle that grows stiffer towards its base, with a flexible tip, could go far enough round. The final array was thinner than a millimetre, with twenty platinum bands on its first fifteen millimetres.

Day one: silence 6:11

On the first of August 1978, at the Royal Victorian Eye and Ear Hospital in Melbourne, Clark and Brian Pyman implanted the University of Melbourne's prototype in Rod Saunders. He was an adult who had lost his hearing after a head injury. When they first switched it on, he heard nothing, and when he came back days later, nothing again. Then the engineers found a loose connection.

Electrodes that sound like vowels 6:39

With the connection fixed, Rod heard sound. Faster pulses sounded higher. Different electrodes sounded different, sharp near the entrance and dull further in, and Rod described some of them as vowels. But there were problems. The electrical loudness range was tiny, about five to ten decibels, against roughly a hundred and twenty for sound in general. And when several electrodes fired at once, their currents overlapped and the loudness jumped unpredictably.

The code that worked 7:13

Those vowels were the clue. Instead of copying the whole sound, the team sent just the key features of speech. One resonance of the voice, called the second formant, chose which electrode fired. The buzz of the voice itself set how fast the pulses came. And electrodes took turns, never firing at the same instant. With this code, Rod understood some running speech with no lip-reading. Clark later said he went into the next lab and burst into tears of joy.

How it works today 7:47

Modern implants follow the same plan. A microphone picks up sound, and a processor splits it into frequency bands. Each band's loudness is squeezed into that narrow electrical range. A coil on the skin sends both the data and the power by radio to a receiver under the skin. Then the electrodes stimulate the nerve, high pitches near the entrance and lower ones further in, just like the natural keyboard.

Twenty-two keys, a few notes 8:16

It isn't natural hearing. The commercial Nucleus array has twenty-two electrodes, but current from each one spreads onto its neighbours. So in practice, users get only about four to eight truly separate channels, against roughly twenty-eight in normal hearing across the speech range. Noisy rooms are hard, and so is music. The US National Institutes of Health puts it plainly: an implant does not restore normal hearing.

From a tin can to a company 8:47

Early money came from a service club, television telethons, and Clark shaking a collection tin on a Melbourne street corner. From 1979 the Commonwealth Government funded the research, and in the early 1980s the Australian medical-electronics company Nucleus set up a subsidiary called Cochlear to build it. In 1985, its twenty-two channel implant became the first multi-channel cochlear implant approved by the US Food and Drug Administration. In 1990, it became, by Clark's account, the first implant any regulator approved for children.

Who got there first 9:27

Now the honest part. Melbourne's was not the world's first multi-channel implant. In Paris, Claude-Henri Chouard's team implanted one in September 1976. In Vienna, an implant designed by Ingeborg and Erwin Hochmair went into its first patient in December 1977. Then in 1991, the American researcher Blake Wilson published a new way of interleaving the pulses that let most users understand sentences without lip-reading. In 2013, the Lasker Award for the modern cochlear implant went jointly to Clark, Ingeborg Hochmair and Wilson.

What was Australian 10:11

So what was distinctly Australian? A safe electrode that slides round the spiral, and a speech code built from the voice's key features, which let Rod Saunders follow some running speech by electricity alone. Then the first FDA approval of a multi-channel implant, an early approval for children, and an Australian company that by recent estimates holds about half the world market. Not everyone celebrated. Many in the signing Deaf community, who see sign language and Deaf culture as something to protect rather than a problem to fix, objected strongly, especially to implanting children, and some still do.

Next: the mould 10:55

Clark turned a decades-old idea into something that worked, was safe, and could be manufactured. Our next story has the same shape. Alexander Fleming noticed penicillin, but it took Howard Florey, born in Adelaide, and his team at Oxford to turn it into a medicine. Why did a mould that kills bacteria sit on the shelf for years, and how did Florey's team change that?

Sources

Every factual claim in the episode is tied to one of these. Spotted an error? Tell us.

1. NIH NIDCD: "Cochlear Implants" (vs hearing aids; does not restore normal hearing;… — nidcd.nih.gov
2. A. A. Eshraghi et al., "The cochlear implant: historical aspects and future… — pmc.ncbi.nlm.nih.gov
3. Lasker Foundation: "2013 Lasker~DeBakey Clinical Medical Research Award: Modern… — laskerfoundation.org
4. Wikipedia: "Graeme Clark (doctor)" (birth, Pyman, Rod Saunders, shell story, honours) — en.wikipedia.org
5. Australian Academy of Science: "Professor Graeme Clark, otolaryngologist", interview… — science.org.au
6. Lemelson-MIT Program: "Graeme Clark" — lemelson.mit.edu
7. NIH NIDCD: "How Do We Hear?" — nidcd.nih.gov
8. G. Ni, S. J. Elliott, M. Ayat, P. D. Teal, "Modelling cochlear mechanics", BioMed… — pmc.ncbi.nlm.nih.gov
9. Wikipedia: "Basilar membrane" (width base vs apex; stiffness; 3,500 inner hair cells) — en.wikipedia.org
10. HyperPhysics (R. Nave, Georgia State Univ.): "Place Theory" — hyperphysics.gsu.edu
11. A. Büchner et al., "The Greenwood function shows close alignment with pitch perceived… — pmc.ncbi.nlm.nih.gov
12. A. Dhanasingh, "Cochlear duct length along the outer wall vs organ of Corti: which one… — pmc.ncbi.nlm.nih.gov
13. HyperPhysics: "Organ of Corti" — hyperphysics.gsu.edu
14. Wikipedia: "Hair cell" (~3,500 inner, ~12,000 outer; mammals don't regrow them) — en.wikipedia.org
15. B. S. Wilson and M. F. Dorman, "Cochlear implants: a remarkable past and a brilliant… — pmc.ncbi.nlm.nih.gov
16. NIH NIDCD: "Noise-Induced Hearing Loss" ("human hair cells don't grow back") — nidcd.nih.gov
17. S. W. Choi, J. M. Abitbol, A. G. Cheng, "Hair cell regeneration: from animals to… — pmc.ncbi.nlm.nih.gov
18. G. M. Clark, "The multiple-channel cochlear implant: the interface between sound and… — pmc.ncbi.nlm.nih.gov
19. Wikipedia: "Cochlear Limited" (founding, Nucleus, FDA 1985/1990, ASX 1995, market… — en.wikipedia.org
20. Wikipedia: "Cochlear implant" (user counts; Deaf-community controversy) — en.wikipedia.org
21. Wikipedia: "Howard Florey" (born Adelaide 1898; 1945 Nobel with Chain and Fleming) — en.wikipedia.org

Image credits

• Cochlear Spectra 22 processor, 1994 · photo: Meg Lomax / Museums Victoria, CC BY 4.0, via Wikimedia Commons · licence: https://creativecommons.org/licenses/by/4.0/
• Graeme Clark, Ingeborg Hochmair, Claude-Henri Chouard, Toulouse 2015 · photo: Bruno Scala / EDP Santé, CC BY-SA 3.0, cropped, via Wikimedia Commons · licence: https://creativecommons.org/licenses/by-sa/3.0/

Researched and scripted with AI assistance, fact-checked claim by claim, with synthetic narration and diagrams drawn in code. How we make episodes.