Summary: Researchers reveal how the auditory cortex signals unexpected sounds and illuminate mechanisms of auditory memory and prediction.
Source: NYU
From a squeaky car door to a mistimed musical note, our brains constantly detect when sounds deviate from what we expect.
A team of neuroscientists at New York University has identified how the auditory cortex distinguishes expected from unexpected sounds, advancing our understanding of how the brain learns complex audio-motor behaviors like speaking, playing an instrument, or performing precise actions in sports.
“We listen to the sounds our movements produce to determine whether or not we made a mistake,” says David Schneider, assistant professor at NYU’s Center for Neural Science and senior author of the study published in Current Biology. “This process is obvious for musicians and speakers, but it operates continuously—for example, when a golfer listens for the contact between club and ball. In this work we show the brain can predict both the timing and the specific features of a sound tied to an action.”
The researchers set out to understand how the brain forms precise expectations for self-generated sounds and how it signals when those expectations are violated. Everyday examples make the point: after hundreds of car-door closings, you know the expected “thump.” If the seat belt gets caught, the sound changes to a “clank,” and that mismatch jumps out. Similarly, a batter who hits the ball squarely experiences a different acoustic consequence than a player who merely tips it, and a musician hears whether a note fits the intended melody.
To investigate the underlying neural mechanisms, Schneider and colleagues trained mice to push a lever with a forepaw, producing a tone when the lever reached a specific position—an experimental analog of closing a car door that produces a predictable sound. Over training, the animals came to expect the precise acoustic outcome at a given moment in the movement.
The team recorded neural activity in the auditory cortex while manipulating the sounds the mice heard. When the expected sound was presented at the expected time, auditory cortical neurons were largely suppressed—showing minimal activity. But when the sound’s frequency was changed, or when its timing was shifted even slightly, many neurons responded strongly. That pattern is consistent with a system tuned to signal prediction errors: activity increases when reality deviates from expectation.
Nicholas Audette, the study’s lead author and a postdoctoral fellow in the Schneider lab, explains: “The auditory cortex appears to encode not so much the precise content of every heard sound but whether that sound matches or violates a learned expectation.”
Supporting that view, the researchers observed that when the tone was omitted entirely, a distinct group of neurons fired at the precise time the sound was expected. Because many of these same neurons would have been active if the sound had actually occurred, their activity appears to reflect an internally generated memory or prediction of the missing sound.
The results point to three related signals present in auditory cortex circuits: movement signals associated with the action itself, expectation signals that peak at the anticipated moment of auditory feedback, and error or prediction signals that appear when expectations are violated. These signals were distributed across cortical layers, with particular concentrations in layers 2/3 and 5 where prediction-related suppression and prediction-error neurons were most evident.
Understanding these mechanisms in the healthy brain may also illuminate what goes wrong in psychiatric conditions. The same circuits that predict and suppress expected self-generated sounds are thought to malfunction in illnesses such as schizophrenia, where patients can perceive internally generated sensations—like “phantom voices”—as external. By mapping the normal circuitry for movement-based auditory prediction, the researchers hope to create a foundation for studying how these processes degrade in disease.
The study’s coauthors include Alessandro La Chioma, a postdoctoral fellow at the Center for Neural Science, and WenXi Zhou, an NYU doctoral student. Funding for the work came from the National Institutes of Health (T32-MH019524, 1R01-DC018802).
About this auditory neuroscience research news
Author: James Devitt
Source: NYU
Contact: James Devitt – NYU
Image: The image is in the public domain
Original Research: Closed access.
“Precise movement-based predictions in the mouse auditory cortex” by David Schneider et al. Current Biology
Abstract
Precise movement-based predictions in the mouse auditory cortex
Highlights
- Mice learn to expect the acoustic consequences of a specific forelimb movement
- Auditory cortex activity increases when a self-generated sound violates the learned expectation
- During silent movements, auditory cortex activity reflects precise temporal expectations
- Movement, expectation, and prediction-error signals are distributed across cortical layers
Summary
Many sensory experiences arise from an organism’s own actions. Accurately predicting both the timing and the features of self-generated stimuli supports fluent behaviors across domains, from speech and music to motor skills.
Previous work has shown frequency-specific suppression of neural responses to self-generated sounds in auditory cortex, suggesting early sensory implementation of movement-based predictions. Yet it was unclear whether this suppression reflected behaviorally specific, temporally precise predictions, and whether expectation signals were present locally within auditory cortex.
To address these open questions, researchers trained mice using a closed-loop lever that generated a tone at a precise movement position. Dense recordings from auditory cortex revealed suppression of responses to expected self-generated sounds that was specific to the sound’s spectral features, to a precise position within the movement, and to the trained movement itself.
Prediction-based suppression was concentrated in cortical layers L2/3 and L5, where deviations from expectation recruited a population of prediction-error neurons otherwise unresponsive under expected conditions. Recordings in the absence of sound revealed abundant movement-related signals in deep layers biased toward neurons tuned to the expected sound, along with expectation signals across layers that peaked at the time auditory feedback was anticipated.
Together, these findings identify distinct auditory cortical neuron populations that encode movement, expectation, and prediction-error signals consistent with a learned internal model linking a specific action to its expected acoustic outcome.