Attention Selectively Reshapes the Geometry of Distributed Semantic Representation.

Area of Science:

• Neuroscience
• Cognitive Science
• Visual Perception

Background:

• Human attention selectively processes complex stimuli based on current goals.
• Neural representations of semantic features are distributed across brain networks.
• The interaction between attention and distributed neural coding remains incompletely understood.

Purpose of the Study:

• To investigate how attentional allocation influences the neural representation of semantic features (behavior and taxonomy) in naturalistic visual stimuli.
• To determine if attention selectively enhances behaviorally relevant information in distributed neural populations.

Main Methods:

• Participants viewed naturalistic animal video clips while attending to either behavior or taxonomy.
• Models of representational geometry were employed to analyze neural population codes.
• Brain activity was analyzed to assess the discriminability of neural representations.

Main Results:

• Attending to animal behavior increased the discriminability of action-related neural codes in anterior intraparietal, pericentral, and ventral temporal cortices.
• Attending to animal taxonomy enhanced the discriminability of category representations in the ventral temporal cortex.
• Attention selectively amplified neural response patterns relevant to the specific behavioral goals.

Conclusions:

• Behavioral goals dynamically modulate the extraction of semantic features from visual input.
• Attention sculpts representational geometry in perceptual areas, improving downstream information readout.
• This mechanism allows the brain to flexibly prioritize and process relevant semantic information.