Boebinger, D. | Search bioRxiv

bioRxiv2026

The language network does not support music perception across a range of music experience

Music and language share many similarities, yet music does not elicit a response in the language brain areas in adults with typical amounts of music experience, and patients with aphasia show preserved music ability. However, this dissociation might not generalize across development or expertise: it may not yet be present in children, who have limited experience with language and music, or it may be diminished in musicians with extensive music training. To evaluate these possibilities, we here used a precision fMRI approach to examine the responses in the language areas of children (aged 4-16 years, n=80) and adult musicians (n=10) to music. We found that the results obtained previously for non-musician adults replicate in two new control samples and, critically, generalize to both of the newly tested populations: the language areas are already selective for language relative to music in young children, showing no age-related change from early to late childhood, and remain selective despite extensive music training in highly experienced musicians. Thus, it appears that the amount of music exposure and practice have little effect on how the language brain areas respond to music.

neuroscience↗

bioRxiv2022

Many but not all deep neural network audio models capture brain responses and exhibit hierarchical region correspondence

Models that predict brain responses to stimuli provide one measure of understanding of a sensory system, and have many potential applications in science and engineering. Deep artificial neural networks have emerged as the leading such predictive models of the visual system, but are less explored in audition. Prior work provided examples of audio-trained neural networks that produced good predictions of auditory cortical fMRI responses and exhibited correspondence between model stages and brain regions, but left it unclear whether these results generalize to other neural network models, and thus how to further improve models in this domain. We evaluated model-brain correspondence for publicly available audio neural network models along with in-house models trained on four different tasks. Most tested models out-predicted previous filter-bank models of auditory cortex, and exhibited systematic model-brain correspondence: middle stages best predicted primary auditory cortex while deep stages best predicted non-primary cortex. However, some state-of-the-art models produced substantially worse brain predictions. Models trained to recognize speech in background noise produced better brain predictions than models trained to recognize speech in quiet, potentially because hearing in noise imposes constraints on biological auditory representations. The training task influenced the prediction quality for specific cortical tuning properties, with best overall predictions resulting from models trained on multiple tasks. The results generally support the promise of deep neural networks as models of audition, though they also indicate that current models do not explain auditory cortical responses in their entirety.

neuroscience↗