Hemispheric asymmetries in the auditory cortex reflect discriminative responses to temporal details or summary statistics of stationary sounds

The processing of stationary sounds relies on both local features and compact representations. As local information is compressed into summary statistics, abstract representations emerge. Whether the brain is endowed with distinct neural architectures predisposed to such computations is unknown. In this magnetoencephalography (MEG) study, we employed a validated protocol to localize cortical correlates of local and summary auditory representations, exposing participants to sequences embedding triplets of synthetic sound textures systematically varying for either local details or summary statistics. Sounds varied for their duration and could be short (40 ms) or long (478 ms) to favor change detections based on local or summary statistics, respectively. Results clearly revealed distinct activation patterns for local features and summary auditory statistics. Neural activations diverged in magnitude, spatiotemporal distribution, and hemispheric lateralization. The right auditory cortex, comprising both primary and neighboring temporal and frontal regions were engaged to detect sound changes in both local features (for short sounds) and summary statistics (for long sounds). Conversely, the left auditory cortex was not selective to these auditory changes. However, the ventro-lateral portion of left frontal lobe, a region associated with sound recognition, was engaged in processing changes in summary statistics at a long sound duration. These findings highlight the involvement of distinct cortical pathways and hemispheric lateralization for the computation of local and summary acoustic information occurring at different temporal resolutions.