A recent study published in conjunction with Associate Professor of Computational Neuroscience Wilson Truccolo offers a new look at how neurons process speech and language.
The study originally set out to understand how phase transitions — shifts in neuronal communication — may lead to seizures in people with pharmacologically-resistant epilepsies, Truccolo wrote in an email to The Herald. But the study also allowed Trucculo and his team to understand the “neuronal dynamics underlying cognitive functions such as speech perception, comprehension and language processing,” he wrote.
The researchers specifically found that brains engage in a type of top-down processing as individuals hear sounds, which means that “information flows from meaning back to sounds, not only from sounds up to meaning,” Pavo Orepić, postdoctoral researcher at the University of Zurich and first author on the paper, wrote in an email to The Herald.
“As the brain begins to recognize a word, that emerging meaning changes how its individual sounds are represented,” he said. “So the brain is not just assembling a word from its sounds: once it has a sense of the word, it uses that information to reshape the processing of the sounds themselves.”
The study arose from an ongoing collaboration of almost 20 years between the Truccolo Lab and the Sydney Cash Lab at Massachusetts General Hospital.
Researchers looked at neuronal recordings from 624 different neurons in three participants. The study recorded neural activity in the anterior superior temporal gyrus, which is responsible for processing speech and word meaning, according to Orepić. The team studied these recordings alongside recordings from the brain’s cortical surface.
Minute effects were not visible when looking at cortical-surface recordings in isolation. “The result shows that some aspects of speech processing only become visible when we look at the coordinated activity of individual neurons,” Orepić wrote.
The team used microelectrodes inside the brain’s cortex — a rare technique applied only during specific surgeries —to record the activity of groups of nearby neurons, said Timothée Proix, an assistant professor of computational neuroscience at the Swiss Federal Institute of Technology Zürich and principal investigator. Viewing the activity of several neurons in parallel allowed the researchers to understand how neurons work together to encode a picture, said Proix.
Proix, a former postdoctoral researcher under Truccolo, said that while recording multiple neurons at once has been used as a strategy for characterizing motor movements and cognitive tasks in other animals, the recording technique is still very rare in humans.
According to Orepić, looking at these recordings in humans revealed that important mechanisms in speech comprehension can be found in the way that groups of neurons fire. “Their coordinated activity formed distinct population patterns: Vowels were organized according to properties related to tongue position, consonants according to how they are articulated and words according to aspects of their meaning,” Orepić wrote.
There are still gaps in researchers’ understanding of neuronal interactions in speech processing, which happens incredibly quickly and across several processing levels, Orepić wrote.
Moving forward, the researchers hope to expand the project to focus on how neurons communicate on different temporal scales to contribute to varying phonetic, semantic and syntactic dimensions of language, Trucculo wrote. With this continued work, the project could also pave the way for potentially restoring communication in patients with neurological disorders.
Alice Xie is a section editor for Science and Research from Los Angeles, California. She studies Applied Mathematics and Biology, and enjoys reading gut wrenching literature in her free time.
