In a dimly lit hallway stand two guards. One, hearing a loud footstep, grabs the door so it won't slam. The other, sitting by the breaker panel, turns the dimmer knob when the noise becomes unbearable. From the outside you can't tell which one acted: the door is ajar, the light is dimmed, and the log has a single entry, "it was loud." That's roughly how our hearing is protected, and that's exactly why it's so hard to study.
The first guard is the middle ear muscle. When a sound exceeds a certain threshold, it contracts and pulls on the eardrum so it vibrates less. The second is the olivocochlear bundle, a thin nerve pathway that runs from the brain to the cochlea and reduces the sound's amplification there. Both work fast and almost simultaneously, but they do different jobs. The muscle protects against mechanical overload, while the nerve bundle, by all appearances, helps you pick out speech in noise. The problem is you can't switch them on separately: when the sound is loud, both fire.
The results were unexpected. The contraction strength of the middle ear muscle in older people with normal hearing was lower than in young people. But when the scientists adjusted for the state of the cochlea, the difference vanished. So the muscle weakens not from age but because the cochlea already has microscopic damage that a standard test can't hear. The second guard, the olivocochlear reflex, behaved differently. At high frequencies it weakened with age, but that link disappeared once audiometric thresholds, that is, obvious hearing loss, were taken into account. Hidden cochlear damage had no effect on it.
The most important discovery concerns not the reflexes themselves but the way they are measured. When the scientists played a noise signal to test the olivocochlear reflex, in two-thirds of the young participants and half of the older ones the middle ear muscle fired at the same time. This contaminated the results, and it was impossible to tell which reflex was doing what. The researchers found a way out: in the first two milliseconds of the response to a click, only the mechanical reaction of the muscle is visible. From that short burst you can tell that the muscle has kicked in and either account for it or repeat the measurement more quietly.
Now that the guards can be told apart by the first moments of their reaction, a path opens to more precise diagnostics. If the amplitude of the muscle's contraction reflects hidden cochlear damage, this test could detect hearing problems long before a person starts complaining about not catching speech in a cafe. And clean measurements of the nerve reflex will help explain why some people easily follow a conversation in a noisy room while others lose the thread. The guards at the ear's door, it turns out, don't just stand watch: they keep a detailed log of the whole building's condition.
The middle-ear reflex fires not only in response to external loud sounds but also to your own vocalizations: when a person speaks or sings, the muscle tenses to protect the cochlea from excessive stimulation.
Not published on arXiv. Here are its page, the PDF and the formalization: you can check it yourself, not only read our account.