Researchers Find Neuron Activity Underlying Human Speech
JAKARTA - The pattern of activity of hundreds of neurons in the brain can provide clues about grammar, word meaning, sentence structure, and the context of speech before a person speaks. These findings can be the basis for developing technology to help people with communication disorders.
The National Institutes of Health or NIH, through its official website, quoted Wednesday, July 15, said the study used a machine learning model to analyze recordings of single cell activity in the human brain while participants were talking.
The data comes from a set of microelectrodes implanted in eight patients to monitor epilepsy. The device was not specifically installed for language research.
Microelectrodes are very small devices used to record electrical signals from nerve cells or neurons.
Researchers from Massachusetts General Hospital in Boston then recorded natural conversations in English with each participant. The topics discussed varied.
Researchers matched the timing of conversations with the activity of hundreds of neurons in the frontotemporal cortex. The brain regions at the front and side of it were previously linked by the team with speech production.
They then used a natural language processing model to search for relationships between the two data sets. Natural language processing is a technology that computers use to analyze human language.
The result is that the activity of neurons just before the participant speaks can be used to predict a number of elements in the next speech, whatever the topic of the conversation.
The research also found a division of labor among neurons. Some nerve cells reflect basic information, such as the meaning and function of a particular word. Other neurons are associated with more complex processes, including grouping phrases into structured sentences.
The model is also able to distinguish similar words and phrases. Theseresults show that the activity of neurons captures the specific context in asentence.
"This kind of detailed observation is necessary so that we can better understand how the brain produces speech and, ultimately, develop technology to restore this ability in people with communication disorders," said Debara Tucci, director of the National Institute on Deafness and Other Communication Disorders.
The lead author of the study, Jing Cai, said the research was the first to describe the process of speech production down to the cellular level, not just in brain regions.
"Having found these basic elements, we now have a foundation to start answering a number of very interesting questions," said the researcher and teacher at Massachusetts General Hospital.
The study looked at the activity of neurons that arise before participants produce speech. The study involved eight patients undergoing epilepsy monitoring.
NIH said the results could potentially support the development of new technologies that translate neural activity into artificial speech. The technology could eventually help turn the neural activity of people with communication disorders into speech.