# Study Finds Music Drives Movement by 12 Months > Source: > Published: 2026-07-07 15:40:37+00:00 # Study Finds Music Drives Movement by 12 Months A bioRxiv preprint and July 7, 2026 coverage from Neuroscience News and PsyPost report that researchers recorded **79 infants** with simultaneous **EEG** and DeepLabCut pose tracking to study music responses across 3, 6, and 12 months. The ML-relevant result is a split between early perception and later movement: infants showed neural sensitivity to structured music by 3 months, while reliable extra movement to music versus shuffled audio appeared at **12 months**. For data scientists, the study is useful less as a music-development claim than as a worked multimodal pipeline, pairing time-aligned brain signals, video kinematics, PCA-derived movement components, and careful age cohorts to study perception-to-action development. The strongest AI/ML value in this study is methodological: it shows how to pair neural recordings with automated pose tracking to study when perception becomes action. For practitioners, the result is a compact multimodal-data case study, not just a developmental-neuroscience headline about babies and music. ### What happened A bioRxiv preprint, covered by Neuroscience News and PsyPost on July 7, 2026, reports simultaneous EEG and video-based motion tracking from 79 full-term infants aged 3, 6, and 12 months. PsyPost describes the cohort as 26 three-month-olds, 26 six-month-olds, and 27 twelve-month-olds, with a young-adult control group. The experiment compared structured children's-song refrains with shuffled and pitch-altered versions to separate responses to musical structure from general audio stimulation. ### Technical context Neuroscience News reports that the team used DeepLabCut pose estimation and PCA to extract 10 principal movement patterns, including rocking, side sway, proto-clapping, and limb movements. The reported split is important: neural responses to structured music appeared early, while reliable extra movement to structured music emerged only at 12 months, and the coverage says the infants did not synchronize movement to the beat. ### For practitioners The pipeline illustrates several reusable data-design choices: time-align EEG and video, convert noisy kinematics into compact behavioral components, and separate sensory encoding from motor output with age-stratified cohorts. Those choices matter for multimodal representation learning, developmental benchmarks, and neural-decoding experiments that need interpretable labels rather than raw motion streams alone. ### What to watch - •Whether the authors release raw EEG, video, or derived pose labels for reuse. - •Whether follow-up work replicates the finding across larger and more diverse cohorts. - •How future datasets handle infant-motion artifacts, lighting variation, and cross-modal clock synchronization. ## Key Points - 1The study pairs EEG and DeepLabCut pose tracking, giving practitioners an example of aligned neural and kinematic labels. - 2Structured music responses appeared neurally by 3 months, while reliable extra movement emerged only around 12 months. - 3The result is most useful as a multimodal data-design case, not as evidence that infants synchronize to beats. ## Scoring Rationale This is a notable methods-and-data item for multimodal learning and developmental benchmarking because it pairs EEG with pose estimation. It remains a preprint-level research result with limited immediate platform impact, so the score stays in the notable range but is slightly moderated. ## Sources Public references used for this report. Practice interview problems based on real data 1,625 SQL & Python problems across 15 industry datasets — the exact type of data you work with. [Try 250 free problems](/problems)