Auditory Cortical Gradients Integrate Bottom-Up and Top-Down Structure During Natural Sound Categorisation

Area of Science:

• Neuroscience
• Auditory Perception
• Cognitive Science

Background:

• Understanding how the brain organizes natural categories is a key neuroscience challenge.
• Previous studies show categories are decodable from auditory cortex activity, but global organization and the interplay of acoustic and semantic factors remain unclear.

Purpose of the Study:

• To investigate the global organization of natural categories within the auditory cortex.
• To determine how low-level acoustic and higher-level semantic structures jointly shape this organization.

Main Methods:

• Derived low-dimensional functional gradients from high-depth functional magnetic resonance imaging (fMRI) data.
• Analyzed population activity patterns in the auditory cortex during a category-specific task.
• Modeled category structure using gradient-based approaches and compared them to region-of-interest and whole-brain methods.
• Projected acoustic and behavioral similarity spaces into a shared framework with fMRI functional axes.

Main Results:

• Gradient-based models provided more accurate explanations of category structure than traditional approaches.
• Category information is distributed across multiple continuous functional axes, not aligned with a single organizational dimension.
• Both acoustic and semantic structures contribute to auditory cortex category geometry, with acoustic structure having a stronger influence.
• Representational relationships varied across category pairs, reflecting acoustic similarity, semantic distinctions, or a combination.

Conclusions:

• Auditory cortex utilizes continuous functional gradients for organizing natural categories.
• The brain integrates multiple representational dimensions, including acoustic and semantic features, to define higher-level categories.
• This pairwise heterogeneity highlights the complex, multi-dimensional nature of auditory category representation.