Cognitive network interactions through communication subspaces in large-scale models of the neocortex.

Area of Science:

• Neuroscience
• Computational Neuroscience
• Systems Neuroscience

Background:

• Neocortex-wide neural activity organizes into distinct networks for cognitive processes.
• Flexible reconfiguration of these networks is crucial for cognition but poorly understood.

Purpose of the Study:

• To elucidate the mechanisms of flexible network reconfiguration in the neocortex.
• To develop a computational model of whole-cortex dynamics constrained by anatomical connectivity and cognitive network activation.

Main Methods:

• Developed connectivity-constrained macaque and human whole-cortex models.
• Incorporated within-area connectivity motifs (symmetric, asymmetric, random) and sparse, low-rank plus random inter-areal connectivity.
• Constrained model connectivity by experimentally observed cognitive network activation maps.

Main Results:

• The model successfully captured key aspects of cognitive network dynamics and interactions, including the anti-correlation between the default mode network and dorsal attention network.
• Demonstrated that communication between networks is shaped by the alignment of long-range communication subspaces with local connectivity motifs.
• Showed that network communication is switchable via a bottom-up, salience-dependent routing mechanism.
• The frontoparietal multiple-demand network exhibited a coexistence of stable and dynamic coding, supporting top-down cognitive control.

Conclusions:

• The developed model provides a theoretical framework for understanding dynamic routing in cortical networks during cognition.
• Highlights the role of local connectivity motifs and inter-areal connectivity structure in shaping network dynamics and communication.
• Explains how bottom-up salience signals can modulate network interactions for flexible cognitive control.