Summary: A neuroscience study shows that melodies people imagine can be decoded and reconstructed from surface electrode brain signals recorded during intracranial monitoring.
Researchers recorded invasive brain activity from ten epilepsy patients who were already undergoing clinical intracranial monitoring. Each participant listened to the opening bars of familiar children’s songs, silently imagined how the melody continued, and then hummed the melody aloud. The team analyzed the surface electrode signals to determine whether patterns of neural activity carried information about the relative pitches that make up imagined melodies.
By training a predictive model to target relative pitch classes (for example, do, re, mi) rather than absolute frequencies, the researchers were able to reconstruct the melodic contours of imagined music over time. These results provide direct neural evidence for internal musical representation and point toward potential clinical applications such as music-based brain-computer interfaces for people who cannot communicate verbally.
Key Facts
- Neural decoding of imagined pitch: Invasive surface electrode recordings carried both spectral and temporal information sufficient to decode sequences of relative pitches during silent musical imagery.
- Relative pitch class framework: The decoding model focused on relative pitch relationships (do, re, mi, fa, sol, la) rather than exact auditory frequencies, reflecting how humans recognize melodies across different keys.
- Sequential contour reconstruction: Aggregating note-level pitch predictions over time made it possible to reconstruct the melodic trajectory and contour of the imagined songs.
- Assistive technology potential: These findings serve as a proof-of-concept for future brain-computer interfaces that could synthesize internally imagined music or speech for individuals with severe motor or speech impairments.
Source: SfN
Study overview: In work led by Jii Kwon and Chun Kee Chung from Seoul National University, the researchers investigated whether melodies that subjects imagine after hearing a short musical cue can be reconstructed from intracranial brain signals. The team published their results in eNeuro.
Ten patients with medically implanted surface electrodes participated as part of their clinical monitoring for epilepsy. Each trial began with the participant listening to the first bars of a familiar children’s tune, followed by a period of silent imagery in which the participant mentally continued the melody. Participants then hummed what they had imagined, providing a behavioral confirmation of the internally generated sequence.
The researchers analyzed neural recordings during the silent imagery interval and applied a decoding model that classified neural patterns into relative pitch classes. Because the model is keyed to pitch relationships rather than absolute frequency, it captures the way listeners perceive melodies by intervals and contour, making the approach resilient if a melody is imagined in a different key.
Combining moment-by-moment pitch predictions enabled the team to reconstruct the melodic contours and sequential changes that reflected the structure of the imagined songs. These reconstructions show that ongoing brain activity during silent musical imagery contains information about the pitch relationships that define a melody.
Kwon explains, “The model decodes relative pitch classes—such as do, re, mi, fa, sol, and la—rather than exact absolute pitches. That matters because people often recognize melodies by the relationships between notes, even when the same melody is played in a different key.”
The authors note that this study supplies direct neural evidence that the brain encodes the pitch identities of notes in imagined music. The research opens several directions for further work, including attempts to decode additional musical attributes such as tempo or dynamics, testing the method with more complex or unfamiliar music, and refining algorithms to improve temporal resolution and robustness.
Kwon and colleagues hope the methods could one day support reconstruction of imagined melodies for people without musical training or for patient populations—such as those with Amyotrophic Lateral Sclerosis—who cannot verbally articulate what they imagine. Such neurotechnologies could expand communication options and enrich quality of life for non-verbal individuals.
Key Questions Answered:
A: The team recorded intracranial surface electrode signals from ten epilepsy patients who already had electrodes in place for clinical seizure monitoring. Participants listened to the opening phrase of familiar songs, silently imagined the continuation, and then hummed their imagined version as behavioral confirmation.
A: Human musical perception typically relies on the interval relationships between notes rather than absolute pitch. Training the decoder on relative pitch classes aligns the algorithm with perceptual invariances, making it effective even when the imagined melody is transposed to another key.
A: Beyond advancing fundamental auditory neuroscience, this approach could underpin communication neuroprosthetics. Future systems may enable people with severe motor neuron disease or other conditions that prevent speech to express musical ideas or internal content through a neural interface.
Editorial Notes:
- This article was edited by a Neuroscience News editor.
- The original journal paper was reviewed in full by the editorial team.
- Additional explanatory context was provided by staff editors to improve clarity for a general audience.
About this music and neuroscience research news
Author: SfN Media
Source: SfN
Contact: SfN Media
Image: The image is credited to Neuroscience News
Original Research: The findings are reported in eNeuro