When people listen to spoken words at different speeds, neuroscientists wondered if the brain adjusts its processing pace to match the speech flow. The latest research published in Nature Neuroscience reveals that, at least in the auditory cortex, the brain follows a fixed temporal grid.
The study was coordinated by Sam Norman‑Haignere, Ph.D., an assistant professor of Biostatistics and Computational Biology, Biomedical Engineering, and Neuroscience at the Del Monte Institute for Neuroscience at the University of Rochester, in partnership with Columbia University teams led by Principal Investigator Nima Mesgarani, Ph.D., and Menoua Keshishian, Ph.D., Electrical Engineering.
"The finding was unexpected. When we slowed a word down, the auditory cortex kept its processing window the same. It appears to integrate sound over a rigid time span," said Norman‑Haignere, the study’s first author, who initiated the project as a postdoc at Columbia.
"A central goal of this work is to craft more accurate computational models of speech processing that will refine our scientific toolkit and eventually illuminate what goes wrong in listeners with speech comprehension difficulties," added Norman‑Haignere.
The auditory cortex, comprising multiple layers and subregions, handles the decoding of sounds. Researchers know that speech is processed by several brain areas: primary and secondary auditory cortex, as well as language regions beyond the hearing system, but how each layer functions and interacts remains poorly defined.
Advances in computational modeling have helped unravel these complexities. These computer models use mathematical algorithms to simulate how sound is interpreted by neurons and predict human perceptions.
The authors employed such models to distinguish between two possibilities: does the auditory cortex integrate information along linguistic units—words or syllables—or does it rely on a fixed temporal window? The simulations showed that while some models learned to group speech by structure, the auditory cortex consistently relied on a time‑based integration, validating the method used to probe structure versus time.
Traditional non‑invasive tools like EEG and fMRI provide indirect measures of brain activity; EEG records signals from the scalp, while fMRI tracks blood flow. Neither technique achieves the spatial or temporal fidelity needed to capture individual neurons’ firing patterns precisely.
To obtain high‑resolution neural recordings, the researchers collaborated with epilepsy monitoring patients at NYU Langone Medical Center, Columbia University Irving Medical Center, and the University of Rochester Medical Center. Implanted electrodes were temporarily positioned in the brain to identify seizure foci and simultaneously record electrical activity adjacent to active neurons.
Participants listened to a passage from an audiobook at its normal speed and then again at a slower rate. Researchers anticipated that reduced speech tempo would shift the neural processing window, but the observed changes were negligible, indicating that the brain’s fundamental processing unit is determined by physical time—about 100 milliseconds—rather than linguistic structure.
"Our results overturn the intuitive assumption that auditory processing is directly tied to speech units like syllables or words," said Mesgarani, a senior author and Associate Professor of Electrical Engineering at Columbia. "Instead, the auditory cortex operates on an internally fixed, time‑based scale; higher‑order regions must then decode linguistic meaning from this steady stream."
"Deepening our grasp of how speech is parsed will pave the way to understanding speech deficits," said Norman‑Haignere. "There are many researchers studying hearing and many studying language, but the brain must somehow transform raw acoustic input into words, phrases, and sentences. Defining and modeling that transformation is a fascinating frontier we’re pursuing."
Additional contributors include Guy McKhann and Catherine Schevon of Columbia University, and Orrin Devinsku, Werner Doyle, and Adeen Flinker from NYU Langone Medical Center.