Summary: Artificial intelligence reveals how music shapes our brains, bodies, and emotions. Specific auditory regions—such as Heschl’s gyrus and the superior temporal gyrus—respond strongly to pulse clarity and changes in dynamics, rhythm, timbre, and instrumentation. These contrasts drive physiological and emotional reactions, and the findings help identify song types that best support workout, study, and sleep routines, as well as therapeutic uses for mood and pain management.
Source: USC
Your heart may race, your palms may sweat, and a region of your brain called Heschl’s gyrus can light up. Most people don’t consider in detail what their brain and body do while they listen to music, yet these reactions are consistent and measurable.
Researchers at USC used artificial intelligence to examine how music affects listeners’ neural activity, physiological responses, and subjective emotions. The team measured heart rate, galvanic skin response (a proxy for sweat gland activity), brain activity, and listeners’ ratings of happiness and sadness while volunteers heard three unfamiliar, instrumental pieces of music.
Analyzing 74 distinct musical features, the researchers found that dynamics (changes in loudness), register, rhythm, and harmony were particularly informative in predicting how people respond to music. By combining multiple modalities—brain imaging, physiological signals, and self-reports—the study produced a more complete picture of music’s impact than prior work that focused on short segments or a single measurement type.
Contrast is crucial
The study highlighted the role of auditory cortex regions, especially Heschl’s gyrus and the superior temporal gyrus, which were sensitive to pulse clarity—the strength or prominence of the beat. In simple terms, clear strong beats reliably activate these areas of the brain.
Beyond beat clarity, the researchers observed that changes in dynamics, rhythmic shifts, timbral variation, or the entrance of new instruments all produce heightened neural and physiological responses. In other words, contrast within a track—variations in loudness, texture, and arrangement—amplifies listener engagement.
“If a song is loud throughout, there’s not a lot of dynamic variability, and the experience will not be as powerful as if the composer uses a change in loudness,” said Tim Greer, lead author and a computer science PhD student at USC’s Signal Analysis and Interpretation Laboratory (SAIL). Greer, also a composer and instrumentalist, added, “It’s the songwriter’s job to take you on a rollercoaster of emotions in under three minutes, and dynamic variability is one of the ways this is achieved.”
This explains why consistently loud genres like some black metal may produce less measurable fluctuation in response compared with songs that alternate between subtle verses and explosive choruses, such as Nirvana’s Smells Like Teen Spirit. The research also showed that galvanic skin response tends to spike when a new instrument enters or a musical crescendo begins.
“When each new instrument enters, you can see a spike in the collective response of the skin,” said Greer.
The most stimulating moments frequently followed a rise in musical complexity: as layers of instruments accumulate and texture thickens, listeners exhibit stronger physiological and emotional reactions. Classic examples include compositions that build by adding instruments and harmonic layers toward a climax.
On a more specific harmonic note, the study found an interesting correlation between sadness ratings and the raised seventh of a minor scale. For example, the F# raised seventh in a G minor context correlated with higher sadness ratings. This musical device appears in emotionally charged recordings such as The Animals’ House of the Rising Sun and can intensify the song’s emotive effect.
New territory
To isolate the musical effects from memory or lyrical associations, the research used instrumental excerpts that were unfamiliar to participants. Forty volunteers underwent MRI scanning at USC’s Brain and Creativity Institute while listening to happy and sad musical excerpts, an experiment led by Assal Habibi and colleagues. Separately, 60 participants listened via headphones while their heart activity and skin conductance were recorded, and they rated emotional intensity on a 1-to-10 scale.
Computer scientists processed these multimodal data with AI algorithms to identify the auditory features that consistently predicted neural, physiological, and subjective responses. This multimodal approach allowed the team to track responses over longer musical passages and to integrate brain imaging with peripheral physiological measures and self-reported emotion.
“Novel multimodal computing approaches help not just illuminate human affective experiences to music at the brain and body level, but in connecting them to how actually individuals feel and articulate their experiences,” said Professor Shrikanth (Shri) Narayanan, study co-author and a professor of electrical and computer engineering and computer science.
Feeling good
The practical applications of this research extend to designing playlists tailored for specific goals—energizing tracks for workouts, calm pieces for sleep, or focused selections for studying. Clinically, music tailored to individual responses could be used in therapies for depression, anxiety, and other mood disorders, or to help manage pain and support people with cognitive impairments.
“From a therapy perspective, music is a really good tool to induce emotion and engage a better mood,” said Assal Habibi. The study’s results can guide how musical stimuli are chosen and structured for therapeutic settings and deepen our understanding of how emotions are processed in the brain.
Future research may explore how different musical genres and compositional intentions influence emotional response, and whether listeners’ perceptions align with composers’ intent. By combining neuroscience, physiology, and AI, researchers can continue to map how sound shapes human feeling and behavior.
Source:
USC
Media Contacts:
Amy Blumenthal – USC
Image Source:
The image is in the public domain.
Original Research: The paper, titled “A Multimodal View into Music’s Effect on Human Neural, Physiological, and Emotional Experience,” was presented at ACM Multimedia on Oct. 22. The research team also includes Ben Ma, an undergraduate computer science student at USC Viterbi and member of USC SAIL, along with other collaborators who contributed to this multimodal investigation.