Preliminary evidence for selective cortical responses to music in one-month-old infants

Heather L. Kosakowski, Sam Norman-Haignere, Anna Mynick, Atsushi Takahashi, Rebecca Saxe, Nancy Kanwisher

TL;DR

By the time an infant is just one month old, their brain already shows specialized cortical responses to music that cannot be explained by simple acoustic properties. Using fMRI on sleeping neonates, researchers found that music-selective regions in the non-primary auditory cortex emerge far earlier than previously thought, potentially independent of—or even preceding—the development of speech-selective areas.

What Problem Does This Paper Address?

Human adults have "category-selective" neural populations—specific patches of the brain that fire specifically for music or speech. A long-standing debate in neuroscience is whether these regions are "hard-wired" or if they are slowly sculpted by years of auditory experience.

The challenge in studying this in infants is that music and speech are acoustically very different from random noise. They have specific "spectrotemporal modulation statistics" (patterns of how pitch and volume change over time). If an infant's brain responds more to music than to white noise, is it because the brain recognizes "music," or just because it likes complex textures? This paper aims to unconfound these low-level acoustic features from high-level category recognition to see if the infant brain is truly "tuned" to music shortly after birth.

Methods

The researchers performed functional Magnetic Resonance Imaging (fMRI) on 45 sleeping infants (aged 2 to 12 weeks). To overcome the noise and movement challenges of infant fMRI, they used several clever techniques:

  • Stimulus Matching: They didn't just use any music. They recorded monophonic (single-instrument) lullabies played on instruments (flute, cello, violin) that matched the frequency range of the female voice.
  • Model-Matched Controls: This is the study's "secret sauce." For every music and speech clip, they synthesized a "model-matched" control—a synthetic sound that has the exact same mathematical "texture" (spectrotemporal statistics) as the original but sounds like unrecognizable noise to a human.
  • Functional ROI (fROI) Analysis: Rather than averaging all brains together (which is difficult given how much infant brain anatomy varies), they identified the top 5% of "sound-responsive" voxels in each individual infant and then tested those specific areas for selectivity.
  • Excitation-Matching: They used a novel algorithm to ensure the "cochleagrams" (how the inner ear processes the sound) were matched across music and speech categories.

Datasets

  • Primary Data: fMRI scans from 45 infants (36 provided usable data; 19 showed significant auditory activation).
  • Stimuli: 60 unique music recordings (lullabies) and 60 unique infant-directed speech recordings.
  • Open Science: The authors have made their full dataset and the custom-synthesized stimuli available on the Open Science Framework (OSF) at https://osf.io/8ty34/.

Key Insights and Ideas

  • Music Selectivity is Early-Onset: The most striking finding is that music-selective responses (responding more to music than to speech and more to music than to its synthetic control) were present in infants as young as one month.
  • Anatomical Specificity: Just like in adults, this selectivity was found in the non-primary auditory cortex (NPAC). The primary auditory cortex (Heschl’s Gyrus), which handles basic sound processing, did not show this category preference.
  • Music vs. Speech Trajectories: Surprisingly, music selectivity was more robust and easier to detect in the planned analysis than speech selectivity. This suggests that the neural machinery for music might emerge independently of, or perhaps even more rapidly than, the machinery for speech.
  • Beyond Physics: Because the brain responded more strongly to music than to the "model-matched" synthetic sounds, the researchers proved that the infant brain isn't just reacting to the statistics of the sound, but to the higher-order structure of music itself.

Key Results

  • Auditory Activation: 19 out of 36 infants showed significant brain activity in response to sounds compared to the background noise of the scanner.
  • Music Selectivity: In these 19 infants, voxels in the NPAC responded significantly more to music than to speech ($P=0.03$), model-matched music ($P=0.002$), and model-matched speech ($P=0.04$).
  • Speech Selectivity: While the primary planned analysis didn't find robust speech-selective voxels, an unplanned exploratory analysis of all 36 infants (including those with weaker overall auditory signals) did show voxels that preferred speech over music and synthetic controls.

Limitations and Future Work

  • The "Sleep" Factor: The infants were scanned while asleep. In adults, sleep is known to dampen cortical responses to speech and music. It is possible that selectivity would be even stronger or more widespread in awake infants.
  • Sample Size and Power: Only about half of the infants showed "significant" overall auditory activation. The authors acknowledge that infant fMRI is notoriously difficult and that larger datasets are needed to confirm the speech results.
  • Nature vs. Nurture: While one month is very early, it isn't "zero" experience. Infants hear muffled sounds in the womb. Future work could look at even younger neonates to further isolate the role of experience.
  • Genre Limitations: The study used only Western-style instrumental lullabies. Whether the infant NPAC responds the same way to a drum solo or non-Western tonal systems remains an open question.

Summary updated

Abstract

Prior studies have observed selective neural responses in adult human auditory cortex to music and speech that cannot be explained by the differing lower-level acoustic properties of these stimuli. Does infant cortex exhibit similarly selective responses to music and speech shortly after birth? To answer this question, we attempted to collect functional magnetic resonance imaging (fMRI) data from 45 sleeping infants (2.0- to 11.9-weeks-old) while they listened to monophonic instrumental lullabies and infant-directed speech produced by a mother. To match acoustic variation between music and speech sounds we (1) recorded music from instruments that had a similar spectral range as female infant-directed speech, (2) used a novel excitation-matching algorithm to match the cochleagrams of music and speech stimuli, and (3) synthesized “model-matched” stimuli that were matched in spectrotemporal modulation statistics to (yet perceptually distinct from) music or speech. Of the 36 infants we collected usable data from, 19 had significant activations to sounds overall compared to scanner noise. From these infants, we observed a set of voxels in non-primary auditory cortex (NPAC) but not in Heschl's Gyrus that responded significantly more to music than to each of the other three stimulus types (but not significantly more strongly than to background scanner noise). In contrast, our planned analyses did not reveal voxels in NPAC that responded more to speech than to model-matched speech, although other unplanned analyses did. These preliminary findings suggest that music selectivity arises within the first month of life.
BibTeX
@inproceedings{kosakowski2023preliminary,
  author = {Heather L. Kosakowski and Sam Norman-Haignere and Anna Mynick and Atsushi Takahashi and Rebecca Saxe and Nancy Kanwisher},
  title = {{Preliminary evidence for selective cortical responses to music in one-month-old infants}},
  booktitle = {Developmental},
  year = {2023},
  doi = {10.1111/desc.13387},
}
Figure