Brice Bathellier has been awarded the CNRS Silver Medal for his work in neuroscience. As a CNRS Research Director at the reConnect - Hearing Institute IdA)¹, Brice Bathellier studies how the brain transforms sounds into meaningful objects. He has been awarded the CNRS Silver Medal.
Contrary to popular belief, auditory information is not merely transmitted to the brain; it is reshaped at every stage of its journey to construct an interpretation of the sound being perceived. At the Institute of Hearing, Brice Bathellier and his team are studying these complex processes at the heart of our brain.
“What fascinates me is understanding a function that we use every day, yet remains deeply mysterious.”
Using optical imaging and electrophysiological methods, they recorded from several thousand neurons simultaneously within the auditory system. In doing so, they uncovered a process of decorrelation.
“Perception isn’t just about transmitting sound information; it’s about understanding it. Sound waves are transformed into auditory objects that convey meaning,” explains the physicist.
Through this process, two auditory stimuli that are similar but have very distinct meanings activate very different neurons deep within the brain, even though they activate similar neurons near the ear. For example, a word spoken forward and the sound of the same word played backward are two distinct auditory objects composed of very similar information, but thanks to the decorrelation mechanism, we perceive them very differently.
To better understand this process, researchers have developed artificial neural networks capable of extracting auditory features from sound. By comparing the transformations performed by these models with those performed by the brain, they reveal a strong parallel: in both cases, the information becomes increasingly specific to the sequence of sound events as it progresses toward the cortex.
The team is now working on a cortical implant based on their algorithms to transmit sound directly to the brain, bypassing the ear. This solution could help people whose auditory nerve has been severed and who are not candidates for a cochlear implant. Having already been validated in mice, the project is expected to move forward with the launch of a startup.
As part of the ERC Synergy Chronology, a collaborative European project, Brice Bathellier and his partners also aim to develop artificial neural networks that replicate the encoding of sound sequences over time. The challenge: to understand not just how a word emerges from a sound, but how it fits into a sentence. This involves focusing on working memory, which allows us to retain the sounds we hear for a few seconds in order to reconstruct meaning.
Find the latest news on the CNRS website: Brice Bathellier | CNRS Biology
Note:
[1] CNRS/ Institut Pasteur Unit
Photo credit: Rémi Poulverel - Brain Institute
