Distributed network flows generate localized category selectivity in human visual cortex

Area of Science:

• Neuroscience
• Cognitive Neuroscience
• Computational Neuroscience

Background:

• Understanding the neural basis of brain function is a core neuroscience goal.
• Visual cortex regions show category selectivity, responding more to specific image types (faces, places, etc.).
• The generation of these localized, function-relevant brain activations remains incompletely understood.

Purpose of the Study:

• To test the hypothesis that distributed brain network flows generate function-relevant brain activations.
• To investigate the role of intrinsic brain connectivity in visual category selectivity.
• To explore how stimulus-driven network interactions, starting in V1, lead to downstream category-specific responses.

Main Methods:

• Utilized functional magnetic resonance imaging (fMRI) data from 352 participants to identify category-specific responses in visual cortex.
• Employed a simulation approach modeling activity flow over intrinsic brain connections to generate task-evoked activations.
• Tested network architectures, including null models, to assess the contribution of connectivity patterns.

Main Results:

• Distributed network flows, originating from V1 (primary visual cortex), were sufficient to generate visual category selectivity.
• Globally distributed (whole-cortex) network flows further enhanced category selectivity.
• Each brain region's unique intrinsic "connectivity fingerprint" was critical for generating category selectivity.

Conclusions:

• The intrinsic network organization of the human brain significantly contributes to generating functionally relevant, localized responses.
• Distributed network dynamics play a primary role in creating category-specific brain activations.
• This study provides evidence for network flow models in explaining visual category selectivity in the human cortex.