For us humans, having a conversation in a crowded room is challenging – it’s often called the cocktail party problem. The mix of sounds arriving at our eardrums needs to be analysed to pick out the sounds of individual talkers and that’s a difficult task.
The neural computations involved in this kind of analysis are daunting and easily disrupted. Even a relatively minor hearing impairment, almost undetectable by standard clinical tests, leads to notable defects in being able to pick out individual voices.
For many species of frogs, the auditory space is just as crowded as in our pre-pandemic cocktail parties. When male frogs call out, they are sharing space with thousands of other frogs – from their own and different species – each also calling at the same time. Within this cacophony of croaks, the female frogs must make the important choice of finding the right male for mating based on the quality of his call.
Our new paper shows frogs may have a simple yet elegant solution to this problem. The tree frog appears to use its lungs as a simple noise cancellation filter, enhancing the audibility of the species-specific calls and suppressing sounds made by other species.
Frogs have two eardrums that pick up sounds, very much like our own eardrums. But in frogs the two eardrums can interact acoustically through the mouth cavity. This allows sound to reach both faces of the eardrum, and makes the ear inherently directional. However, sound can also enter the mouth cavity from the frogs’ own lungs.
While she is listening, a female frog’s lungs resonate at a certain frequency range depending on the volume of her lungs. These sounds are efficiently transmitted through the glottis – the opening between the vocal chords – to the mouth cavity.
This kind of transmission from the lungs into eardrums has been found in several species of frogs, but has been puzzling up until now. The call frequencies of all species studied so far never matched the frequency range of lung resonance, so any functional implication of the lung to ear transmission was obscure.
We measured the transmission of sound through both eardrums and lungs in female green tree frogs using laser vibrometry – a technique to measure the vibration of lungs and eardrum and local sound stimulation. This enabled us to “dissect” the sound components.
Our measurements showed the sound component from the lung reduces eardrum motion – and hence sensitivity – in the frequency range where the lungs are most sensitive. In the green tree frog and many other species, this range is outside the frequencies of the species’ call.
This means the lung input is preventing “unwanted” sound from stimulating the ear – just like how a noise cancelling headphone quietens background noises by generating sound waves exactly out of phase with outside noises. This filtering is purely caused by passive acoustics, so it doesn’t require any neural processing.
We then investigated the ambient noise generated by other frog species to determine the reason behind the the sound filtering. Here, we used a large database of surveys of frog populations in the US, collected through citizen science. The surveys showed 42 other species of frogs were reported to live in the same ponds as the green tree frog, all making noises at the same time of day.
When we studied each of the individual frog frequencies, we found the calls of the two most common species matched the frequency of the noise made by the tree frog lungs. This means the tree frog lung noises cancel out these calls the most. By doing this, the green tree frog can hear its own species call more clearly.
This is how we concluded the tree frog uses its lungs to filter out the background noise. By suppressing the calls of other species calls, as noise cancelling headphones suppress the outside noise, it enhances those from its own species.
This solution to the frog’s cocktail party problem has been worked out by co-opting existing structures – the lungs and mouth cavity – through evolutionary history. We don’t know how general this mechanism is in frogs, but we have seen a similar mismatch between lung resonance frequency and call frequency in several other species and guess that these species could use a similar mechanism.