The role of high-amplitude bursts of high-gamma activity in naturalistic speech and music listening.

A central challenge in systems neuroscience is to understand how distributed brain networks organize activity during naturalistic cognition. Here, we investigate whether transient, high-amplitude bursts of high-gamma activity, referred to as neuronal avalanches, provide a compact and functionally informative description of large-scale neural dynamics during auditory processing. Using intracranial stereotactic intracranial electroencephalography (sEEG) recordings from epileptic patients, we analyzed brain activity during speech listening, music listening, and rest. We show that high-amplitude bursts, defined as the top 1% of high-gamma activations, are strongly stimulus-driven: they synchronize across participants exposed to the same auditory input, exhibit condition-specific spatial topographies, and display distinct propagation patterns across cortical networks. Notably, although these events represent only a small fraction of the signal, they preserve substantial stimulus-related information. Temporal response function analyses reveal that avalanche activity reliably encodes both speech and music stimuli, even under extreme data sparsification. Together, these findings demonstrate that neuronal avalanches capture stimulus-relevant, large-scale neural dynamics and provide a computationally efficient framework for studying cognition in naturalistic settings.